text
stringlengths 87
880k
| pmid
stringlengths 1
8
| accession_id
stringlengths 9
10
| license
stringclasses 2
values | last_updated
stringlengths 19
19
| retracted
stringclasses 2
values | citation
stringlengths 22
94
| decoded_as
stringclasses 2
values | journal
stringlengths 3
48
| year
int32 1.95k
2.02k
| doi
stringlengths 3
61
| oa_subset
stringclasses 1
value |
|---|---|---|---|---|---|---|---|---|---|---|---|
==== Front
Breast Cancer ResBreast Cancer Research1465-54111465-542XBioMed Central London bcr13081616813210.1186/bcr1308Research ArticlePercentage density, Wolfe's and Tabár's mammographic patterns: agreement and association with risk factors for breast cancer Gram Inger T [email protected] Yngve [email protected] Giske [email protected] Gertraud [email protected] Nils [email protected] Eiliv [email protected] Institute of Community Medicine, University of Tromsø, Breivika, Norway2 Institute for Nutrition Research, University of Oslo, Norway3 Department of Preventive Medicine, University of Southern California, Keck School of Medicine, Los Angeles, CA, USA4 Cancer Research Center of Hawaii, University of Hawaii, Honolulu, HI, USA5 Department of Radiology, Center for Breast Imaging, University Hospital of North Norway, Tromsø, Norway2005 25 8 2005 7 5 R854 R861 15 3 2005 28 4 2005 26 6 2005 18 7 2005 Copyright © 2005 Gram et al.; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Introduction
The purpose of this report was to classify mammograms according to four methods and to examine their agreement and their relationship to selected risk factors for breast cancer.
Method
Mammograms and epidemiological data were collected from 987 women, aged 55 to 71 years, attending the Norwegian Breast Cancer Screening Program. Two readers each classified the mammograms according to a quantitative method (Cumulus or Madena software) and one reader according to two qualitative methods (Wolfe and Tabár patterns). Mammograms classified in the reader-specific upper quartile of percentage density, Wolfe's P2 and DY patterns, or Tabár's IV and V patterns, were categorized as high-risk density patterns and the remaining mammograms as low-risk density patterns. We calculated intra-reader and inter-reader agreement and estimated prevalence odds ratios of having high-risk mammographic density patterns according to selected risk factors for breast cancer.
Results
The Pearson correlation coefficient was 0.86 for the two quantitative density measurements. There was moderate agreement between the Wolfe and Tabár classifications (Kappa = 0.51; 95% confidence interval 0.46 to 0.56). Age at screening, number of children and body mass index (BMI) showed a statistically significant inverse relationship with high-risk density patterns for all four methods (all P < 0.05). After adjustment for percentage density, the Wolfe classification was not associated with any of the risk factors for breast cancer, whereas the association with number of children and BMI remained statistically significant for the Tabár classification. Adjustment for Wolfe or Tabár patterns did not alter the associations between these risk factors and percentage mammographic density.
Conclusion
The four assessments methods seem to capture the same overall associations with risk factors for breast cancer. Our results indicate that the quantitative methods convey additional information over the qualitative methods.
==== Body
Introduction
Different measures of mammographic images or density patterns [1-5] have been used as surrogate endpoints for breast cancer in etiologic research [6-9], in treatment [10-13] and in prevention trials [14-18]. Mammographic density is also important in the way in which it affects mammographic screening sensitivity [19]. It has been suggested that women with high-risk density patterns should be screened more frequently and/or with additional views per breast [20-22] or that they should be prescribed chemoprevention [23].
The first qualitative classification of mammographic density patterns was described by Wolfe in 1976 [1]. A modification of this method was proposed by Tabár [3]. Wolfe also measured the percentage of the breast containing radiographic densities on a continuous scale with the use of a polar planimeter [24]. A modification of the latter method is the BI-RADS system used in clinical radiology practice in the USA [2]. The computer-assisted technique of measuring percentage mammographic densities developed by Boyd and colleagues [4] as well as that developed at the University of Southern California [5] are other methods of quantitative assessment.
The percentage density methods and the Wolfe patterns have consistently been shown to be strongly related to breast cancer risk in different populations [6,7,15,25-29], whereas the Tabár classification has so far only been shown to be related to breast cancer risk among Chinese women [30]. Several risk factors for breast cancer have been shown to be associated with the different methods of density measurement [6-9,30-33].
The Mammography and Breast Cancer Study is a research project using mammographic density patterns as surrogate endpoints for breast cancer among postmenopausal women attending the Norwegian Breast Cancer Screening Program in Tromsø. The purpose of this study was to classify mammograms according to four methods and to examine their agreement and their relationship to selected risk factors for breast cancer.
Materials and methods
Study population
The study was conducted during 15 and 17 consecutive weeks in spring 2001 and 2002, respectively. In brief, women residing in the municipality of Tromsø, aged 55 to 71 years, attending the Norwegian Breast Cancer Screening Program at the University Hospital of North Norway were eligible. After giving informed consent, the study subjects were interviewed by a trained nurse about their current and previous postmenopausal hormone therapy use, reproductive and menstrual factors, previous history of cancer, and smoking status. The participants had their height measured to the nearest centimeter and weight to the nearest half kilogram. Waist and hip circumferences were measured in a standardized way to the nearest centimeter. In 2001, the women were asked to complete a four-page questionnaire at home. The questionnaire elicited information on demographic, menstrual and reproductive factors, as well as lifestyle and dietary factors. In 2002, another four pages with questions about diet were added. The National Data Inspection Board and the Regional Committee for Medical Research Ethics approved the study. Appropriate measures were taken to ensure confidentiality of the data. Altogether, 1,041 women entered the study. This accounted for 70.1% of the women attending the breast cancer screening program.
We excluded 22 women with a new or previous breast cancer diagnosis, 1 currently using chemotherapy, and 31 lacking classification from one of the three readers, leaving 987 women for analysis. Women were classified as postmenopausal if they were 56 years or older or reported having had no period during the last 12 months, or if the serum follicle-stimulating hormone level was above 20 IU/L. By these criteria three women were equivocal for menopausal status. Excluding these women did not alter the results, and they were included as postmenopausal. The readers were blinded to all characteristics of the study subjects. A total of 21 mammograms were marked as difficult to read by at least one of the readers. We reran all analyses without these mammograms, but the results were essentially the same. The 987 mammograms that were read four times were used in the analyses presented in this paper.
Mammographic classifications
For the computer-assisted density assessments, we digitized the left cranio-caudal using both a Kodak LS-85 X-ray digitizer (reader GM) with a pixel size of 260 μm (equal to a resolution of 98 pixels per inch) and an older version of the Cobrascan scanner (Cobrascan CX 312-T) from Radiographic Digital Imaging (Torrance, CA, USA) at a resolution of 150 pixels per inch (reader GU). Reader GM used the Cumulus software from Canada [34], and reader GU applied the Madena software developed at the University of Southern California [5]. Both readers had previously successfully used these techniques for other studies [5,27,35,36]. Both computer programs assign a pixel value of 0 to the darkest (black) shade in the image and a value of 255 to the lightest (white) shade; shades of grey are assigned intermediate values. The number of pixels in each area are measured before computing the ratio between the total and the dense areas; these are then multiplied by 100 to convert them to percentages. The mammograms were classified according to reader-specific quartiles of percentage density.
The experienced radiologist (NB) [37], classifying the mammograms according to the Wolfe [38] and the Tabár [3] methods, used the latter method for the first time. In brief, the Wolfe method assigns the mammograms to four parenchymal patterns (N1, P1, P2 and DY) according to the distribution of fat and the prominence of the ducts. These patterns were again dichotomized into low-risk (N1 and P1) and high-risk (P2 and DY) patterns [38]. The Tabár method classifies the mammograms in five patterns (I to V) based on an anatomic–mammographic correlation with a three-dimensional, subgross (thick-slice) technique. Patterns 1, 2 and 3 are considered low-risk and patterns 4 and 5 high-risk. These patterns are not considered to be on a continuous risk scale. The primary difference between the Wolfe and Tabár classification systems is Tabár's pattern I; this is described in more detail elsewhere [3]. To be able to compare all three methods, we dichotomized the mammograms in the upper reader-specific quartiles to high-risk and the remaining to low-risk. We also calculated an upper quartile for percentage density based on dense area readings by reader GU and the breast area outlined for reader GM.
Statistical analyses
Intra-rater and inter-rater agreement
Thirty-seven randomly selected mammograms were mixed and read blindly a second time by the three readers; the readings were 3 months apart for readers GM and NB and 18 months apart for reader GU. Subsequently, reader GU reread the dense area on a subset of 189 mammograms that had been scanned for reader GM. We calculated the percentage density from both the new and old dense area readings by reader GU.
We calculated the Pearson's correlation coefficient and the crude and weighted Kappa statistics to test the intra-reader and inter-reader reliabilities for the mammographic readings. The Kappa coefficient does not require any assumption about 'correct' categorization and includes a correction for the amount of agreement that would be expected by chance alone. A Kappa of 0% indicates that the agreement between two measurements is no greater than would be expected by chance. Kappa values of 50% or more indicate moderate agreement, 60 to 80% good agreement, those over 80% very good agreement, and those over 90% excellent agreement [39].
Relationship between the four methods and their associations with risk factors for breast cancer
We calculated the median percentage density for the four Wolfe and five Tabár pattern categories for each of the two readers. Similarly, we categorized the mammograms according to reader-specific quartiles with the corresponding proportion of high-risk versus low-risk Wolfe and Tabár patterns for the different quartiles. We also estimated prevalence odds ratios of having high-risk mammographic density patterns with 95% confidence intervals (CI) to express the degree of association between selected risk factors for breast cancer and the mammographic density patterns. Each of the following factors was evaluated as a potential confounder of the relation between the factor of interest and mammographic patterns: age (less than 60, 60 to 64, 65 or more), age at menarche (less than 13, 13 to 14, more than 14), age at menopause (less than 48, 48 to 50, 51 years or more), number of children (0, 1, 2, 3, 4 or more), age at first birth (less than 20, 20 to 24, 25 or more) and body mass index (BMI (defined as weight in kilograms divided by the square of the height in meters) less than 25.0, 25.0 to 29.9, 30.0 or more).
We performed multivariate analyses with models that included all the above listed variables as independent variables and the high-risk density patterns according to each of the four readings as the dependent variable. Subsequently, we reran the models with the Wolfe and Tabár patterns as outcome variables adjusting for percentage density from each of the two readers as a continuous variable. Similarly, we adjusted the models with percentage density as outcome variable for the two categories of Wolfe and Tabár patterns. Statistical trend tests were obtained by creating an ordinal exposure variable with equally spaced scores and including it in the logistic regression model. Results were considered as statistically significant if the two-sided P value was 0.05 or less. We performed data management and statistical analyses with the SAS statistical software package, version 8.2 (SAS Institute Inc., Cary, NC, USA) [40].
Results
Table 1 shows the basic characteristics of the total study population and the women with mammograms in the high-risk categories. Reader GM classified mammograms with 28.3% density or more to be in the upper quartile; the corresponding figure for reader GU was 19.0% density. The value for the upper quartile was in between these, namely 21% density, when the dense area from reader GU and breast area for reader GM was used. In all, 47% and 24% of the mammograms were classified as high-risk according to the Wolfe (P2, DY) and Tabár (IV, V) classifications, respectively. Among the mammograms in the upper reader-specific quartiles, more than 95% were also classified as high-risk according to the Wolfe method, whereas the corresponding proportion for the Tabár method was less than 70% (Table 1).
The Pearson correlation coefficient was 0.93 and 0.86 for the repeated quantitative readings conducted 3 months (GM) and 18 months (GU) apart, respectively. The intra-rater agreement for the upper reader-specific quartile versus the three lower quartiles was moderate for reader GM (Kappa = 0.59; 95% CI 0.29 to 0.90) and reader GU (Kappa = 0.59; 95% CI 0.29 to 0.90). For reader NB the intra-reader agreement was good for the Wolfe classification (Kappa = 0.61; 95% CI 0.34 to 0.89) and very good for the Tabár classification (Kappa = 0.89; 95% CI 0.69 to 1.00). The Pearson correlation coefficient was 0.86 for the two original percentage density readings and 0.93 for the subset of 189 mammograms. Both the agreements between the reader-specific upper quartiles (crude Kappa = 0.69; 95% CI 0.64 to 0.74) and between all four quartiles (weighted Kappa = 0.71; 95% CI 0.68 to 0.74) were good. The agreement between high-risk and low-risk Wolfe and Tabár patterns was moderate (crude Kappa = 0.51; 95% CI 0.46 to 0.56).
Table 2 shows the distribution of mammograms according to the four Wolfe and five Tabár patterns with the corresponding median percentage density assessment for the two readers. According to reader GM the median percentage density for Wolfe high-risk patterns was 28.6% and for Tabár 34.9%. The corresponding figures for reader GU were 19.4% and 25.2%.
Table 3 shows that there was a significant inverse relation between age at screening, number of children and BMI and unfavorable density patterns according to all four classifications. Ages at menarche and menopause showed no association with any of the four outcome variables. Women who had their first child after their 25th birthday were more likely to have high-risk density patterns according to all four readings. A trend test achieved borderline significance for two of the assessments (GU (P = 0.07) and the Wolfe classification (P = 0.05)).
None of the associations displayed in the table between risk factors and the two quantitative readings changed when adjusted for the Wolfe or Tabár classification.
After adjustment for percentage density as assessed by either reader, age at entry (P = 0.14), number of children (P = 0.54) and BMI (P = 0.57) were no longer associated with the Wolfe classification. The association between age (P = 1.0) and the Tabár classification disappeared, whereas the inverse association with number of children (P < 0.05) and BMI (P < 0.03) remained statistically significant after adjustment for percentage density.
Discussion
Our study finds that although the four methods vary depending on which mammograms are classified as unfavorable, all four methods seem to capture the same overall associations with risk factors for breast cancer. There was a high correlation for both inter-reader and intra-reader reliability between the two quantitative methods. Furthermore, mammograms classified as high-risk according to the Wolfe and Tabár classifications had the highest median percentage density. Age at screening, number of children and BMI were inversely associated with all four assessments of breast density. After adjustment for percentage density, there was no longer any association with these risk factors and the Wolfe pattern, whereas the association between number of children and BMI remained for the Tabár pattern. Controlling for the pattern classification did not change the associations between risk factors and percentage density.
All four methods of mammographic assessment have a subjective component. The percentage of dense area measurement is a ratio of two measures (dense area : total area). The mammograms in this study were rather dark, and differences in the appearance of the mammographic images due to the properties of the scanners did lead to variations in defining both areas. The two readers have both trained with Dr NF Boyd in Canada and have also worked together on previous studies [35,36]. Nevertheless, some of the discrepancy was due to differences in judgement as to what constitutes a dense area. However, the relative assessment between what constituted high-density or low-density mammograms was highly correlated. This supports the comparability of the readings.
The Kappa values for inter-rater agreement in our study are of the same magnitude as or better than those found in several other studies examining different kinds of mammographic reading. Venta and colleagues found a Kappa value of 0.46 for density measurements on X-ray and digital mammograms recorded by two radiologists [41]. In a previous study including more than 3,500 premenopausal and postmenopausal women, we found the overall agreement between high-risk and low-risk for the Wolfe and Tabár classifications to be poor (Kappa = 0.22) [3] in comparison with the moderate agreement (Kappa = 0.51) in the present study. The two classifications are not strictly independent because one reader performed both assessments. We attribute the latter higher Kappa value to the fact that all women were postmenopausal, resulting in more low-density patterns, which are easier to assess.
Our results are in agreement with a recent study describing that the Wolfe pattern classification was redundant when percentage density was available as a measure for breast cancer risk [42]. In contrast to this, the association between parity and BMI with the Tabár classification remained after adjustment for density, suggesting that this classification captures something more than just density assessments. However, we do not know whether this additional information from the Tabár classification is related to breast cancer risk. The association between age, parity, BMI and mammographic density is similar, as described in the literature [6,7]; this attests to the validity of the mammographic measurements. The results also underscore the importance of adjusting for these factors when other associations are being explored. The positive association between age at first birth and unfavorable patterns indicated in the present study has also been found in some previous reports [3,8,29,43-46], but not in others [47].
The present results are in agreement with studies finding no associations between ages at menarche [44] and menopause [47] and unfavorable density patterns, but not with others [8,45]. We found a positive association between age at menarche and unfavorable density patterns among premenopausal women and an inverse association among postmenopausal women [8]. In two other studies, a positive overall association was revealed [45,47]. In the study by El-Bastawissi and colleagues, age at menopause was positively related to unfavorable patterns [45]. In our previous study the same relationship was found to be of borderline significance [8].
Our study has several strengths. It was a part of a population-based screening project with a high attendance rate. The three readers were experienced and blinded to the risk factors. A study from the Finnish screening program using the Wolfe classification showed that at younger ages there was a greater probability of misclassification from low-risk to high-risk and at older ages there was a greater probability of misclassification from high-risk to low-risk [48]. Because the misclassification seems to be age dependent we also consider it a strength that all our women were postmenopausal. Our study is cross-sectional and does not have information on the temporal relationship between the risk factors examined and the mammographic density patterns. Several studies have shown that different types of hormone use will temporarily change mammographic density [49-51] but that there will be a reversal of the hormone-induced changes on cessation of treatment [50-52].
Conclusion
The high-risk classification of mammograms varied to some extent according to the four assessment methods commonly used previously. Our results indicate that the quantitative methods convey additional information over the qualitative methods. Quantitative measures should therefore be preferred when high-risk density patterns are used as surrogate endpoints in etiologic research, clinical or preventive trials. The difference between Tabár and Wolfe categories in relation to breast cancer risk and quantitative density assessment needs to be investigated further. Once breast cancer screening programs start to use digital mammograms, a quantitative method may become usable also in large-scale screening programs.
Abbreviations
BMI = body mass index; CI = confidence interval; OR = odds ratio.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
ITG conceived the study, had full access to all of the data in the study, and takes responsibility for the integrity of the data and the accuracy of the data analyses. Classification of mammograms was performed by NB, GM and GU. Analysis and interpretation of data were performed by ITG, YB, GU, GM and EL. ITG and YB drafted the manuscript. ITG, YB, NB, GU, GM and EL revised the manuscript critically for important intellectual content. All authors read and approved the final manuscript.
Acknowledgements
The study was supported by grants from the Norwegian Cancer Society, the Aakre Foundation, the Norwegian Women's Public Health Association (grant H3-02) and the Norwegian Research Council (Visiting Scientist Award grant 148365/300) for ITG's work at the Cancer Research Center of Hawaii, University of Hawaii, Honolulu, USA. Most importantly, we want to thank the women participating in the study.
Figures and Tables
Table 1 Characteristics of the study participants overall and with mammograms classified as high-risk density patterns
Characteristic All women Quantitative Reader GM Reader GU/GM Reader GU Pattern Reader NB
n = 987 Upper quartile ≥ 28.3% (n = 249) Upper quartile ≥ 21.0% (n = 246) Upper quartile ≥ 19.0% (n = 247) Wolfe P2, DY (n = 462) Tabár IV, V (n = 233)
Age (years) at screening, mean ± SD 61.4 ± 4.5 60.4 ± 4.4 60.3 ± 4.5 60.3 ± 4.5 60.8 ± 4.6 60.6 ± 4.5
Age (years) at menarche, mean ± SD 13.3 ± 1.4 13.4 ± 1.3 13.4 ± 1.3 13.4 ± 1.3 13.3 ± 1.3 13.4 ± 1.3
Age (years) at menopause, mean ± SD 48.5 ± 5.1 48.5 ± 4.9 48.8 ± 4.8 48.8 ± 4.7 48.8 ± 4.9 48.7 ± 4.8
Parous (%) 92.6 89.2 88.6 87.5 89.4 88.0
Age (years) at first birth/pregnancy, mean ± SD 22.9 ± 3.7 23.9 ± 4.3 24.1 ± 4.3 24.1 ± 4.3 23.4 ± 3.9 23.9 ± 4.4
Number of children, mean ± SD 2.6 ± 1.4 2.1 ± 1.2 2.1 ± 1.1 2.0 ± 1.2 2.3 ± 1.3 2.1 ± 1.2
Body mass index, mean ± SD 27.4 ± 4.8 24.9 ± 3.6 25.1 ± 3.7 25.1 ± 3.8 26.0 ± 4.0 25.9 ± 3.8
Wolfe high-risk patterns (P2, DY) (%) 46.8 95.6 96.3 96.4 - 99.1
Tabár high-risk patterns (IV, V) (%) 23.6 60.6 66.3 67.2 50 -
Reader GM
Median density, % (range) 14.5 (0–88.4) - 41.1 (9.4–88.4) 39.4 (7.6–88.4) 28.6 (1.9–88.4) 34.9 (1.9–88.4)
Reader GU
Median density, % (range) 9.6 (0–69.2) 26.6 (0.6–69.2) 27.7 (0.6–69.2) - 19.4 (0.6–69.2) 25.2 (3.5–69.2)
Readers GU and GM
Median density, % (range) 29.5 (3.9–73.8) - 30.5 (18.4–73.8) 21.7 (2.1–73.8) 28.1 (3.6–73.8)
For percentage mammographic densities, the upper quartile is defined as high-risk and the lower three quartiles as low-risk.
Table 2 Agreement between two qualitative (Wolfe, Tabár) and two quantitative (median percentage density measurements) classifications
Totals Wolfe
N1 P1 P2 DY
Tabár 190, 2.6, 0.4 335, 9.4, 5.5 228, 21.5, 15.5 234, 39.5, 26.5
I 470, 15.0, 9.4 0 255, 9.2, 5.1 165, 22.1, 13.9 50, 39.3, 19.8 Tabár low-risk: 754, 10.1, 6.0
II 194, 2.7, 0.4 188, 2.6, 0.4 6, 8.6, 2.9 0 0
III 90, 12.3, 8.6 1, 4.8, 1.8 73, 9.9, 7.6 14, 19.5, 15.8 2, 34.0, 14.6
IV 159, 30.3, 23.7 1, 41.8, 22.1 1, 9.2, 14.4 49, 20.5, 18.8 108, 36.4, 27.3 Tabár high-risk: 233, 34.9, 25.2
V 74, 46.4, 28.1 0 0 0 74, 46.4, 28.1
Wolfe low-risk: 525, 6.3, 2.8 Wolfe high-risk: 462, 28.6, 19.4
Measurements by reader GM (with Cumulus software) are shown in italics, and by reader GU (with Madena software) in bold. Total n = 987 women.
Table 3 Associations between selected risk factors for breast cancer andmammograms classified according to four methods
Characteristic Upper quartile High-risk patterns Reader NB
Reader GM ≥ 28.3% Reader GU ≥ 19.0% Wolfe Tabár
Age at screening
<60 years 1.00 1.00 1.00 1.00
60–64 years 0.65 0.74 0.64 0.79
65+ years 0.52 0.53 0.59 0.64
P for trend <0.01 <0.01 <0.01 <0.03
Age at menarche
<13 years 1.00 1.00 1.00 1.00
13–14 years 1.24 1.28 1.26 1.36
14+ years 0.90 1.04 0.81 0.76
P for trend 0.99 0.50 0.67 0.70
Age at menopause
<48 years 1.00 1.00 1.00 1.00
48–50 years 0.80 0.85 0.90 0.87
51+ years 0.95 1.14 1.38 1.11
P for trend 0.83 0.47 0.06 0.61
Age at first birth
<20 years 1.00 1.00 1.00 1.00
20–24 years 0.99 1.15 1.41 1.14
25+ years 1.30 1.66 1.67 1.40
P for trend 0.27 0.07 0.05 0.15
Number of children
0 1.00 1.00 1.00 1.00
1 1.14 1.54 1.61 1.16
2 0.52 0.87 0.87 0.40
3 0.51 0.60 0.90 0.37
4+ 0.18 0.23 0.41 0.21
P for trend – all <0.0001 <0.0001 <0.0001 <0.0001
Body mass index
<25 1.00 1.00 1.00 1.00
25–30 0.39 0.48 0.53 0.81
>30 0.15 0.19 0.24 0.36
P for trend <0.0001 <0.0001 <0.0001 <0.0001
Multivariate odds ratios have been adjusted for age at screening, age at menarche, age at menopause, number of children, age at first birth, and body mass index where applicable. Total numbers in the multivariate models may vary slightly owing to missing values for covariates.
==== Refs
Wolfe JN Risk for breast cancer development determined by mammographic parenchymal pattern Cancer 1976 37 2486 2492 1260729
American College of Radiology Breast Imaging Reporting and Data System (BI-RADS) 1995 2 Reston: American College of Radiology
Gram IT Funkhouser E Tabar L The Tabar classification of mammographic parenchymal patterns Eur J Radiol 1997 24 131 136 9097055 10.1016/S0720-048X(96)01138-2
Boyd NF Byng JW Jong RA Fishell EK Little LE Miller AB Lockwood GA Tritchler DL Yaffe MJ Quantitative classification of mammographic densities and breast cancer risk: results from the Canadian National Breast Screening Study J Natl Cancer Inst 1995 87 670 675 7752271
Ursin G Astrahan MA Salane M Parisky YR Pearce JG Daniels JR Pike MC Spicer DV The detection of changes in mammographic densities Cancer Epidemiol Biomarkers Prev 1998 7 43 47 9456242
Saftlas AF Szklo M Mammographic parenchymal patterns and breast cancer risk Epidemiol Rev 1987 9 146 174 3315715
Oza AM Boyd NF Mammographic parenchymal patterns: a marker of breast cancer risk Epidemiol Rev 1993 15 196 208 8405204 10.2188/jea.15.S196
Gram IT Funkhouser E Tabar L Reproductive and menstrual factors in relation to mammographic parenchymal patterns among perimenopausal women Br J Cancer 1995 71 647 650 7880753
Jakes RW Duffy SW Ng FC Gao F Ng EH Seow A Lee HP Yu MC Mammographic Parenchymal Patterns and Self-reported Soy Intake in Singapore Chinese Women Cancer Epidemiol Biomarkers Prev 2002 11 608 613 12101107
Greendale GA Reboussin BA Sie A Singh HR Olson LK Gatewood O Bassett LW Wasilauskas C Bush T Barrett-Connor E Effects of estrogen and estrogen-progestin on mammographic parenchymal density. Postmenopausal Estrogen/Progestin Interventions (PEPI) Investigators Ann Intern Med 1999 130 262 269 10068383
Freedman M San Martin J O'Gorman J Eckert S Lippman ME Lo SC Walls EL Zeng J Digitized mammography: a clinical trial of postmenopausal women randomly assigned to receive raloxifene, estrogen, or placebo J Natl Cancer Inst 2001 93 51 56 11136842 10.1093/jnci/93.1.51
Greendale GA Reboussin BA Slone S Wasilauskas C Pike MC Ursin G Postmenopausal hormone therapy and change in mammographic density J Natl Cancer Inst 2003 95 30 37 12509398
Chow CK Venzon D Jones EC Premkumar A O'Shaughnessy J Zujewski J Effect of tamoxifen on mammographic density Cancer Epidemiol Biomarkers Prev 2000 9 917 921 11008909
Spicer DV Ursin G Parisky YR Pearce JG Shoupe D Pike A Pike MC Changes in mammographic densities induced by a hormonal contraceptive designed to reduce breast cancer risk J Natl Cancer Inst 1994 86 431 436 8120917
Boyd NF Lockwood GA Martin LJ Byng JW Yaffe MJ Tritchler DL Mammographic density as a marker of susceptibility to breast cancer: a hypothesis IARC Sci Publ 2001 154 163 169 11220655
Maskarinec G Williams AE Carlin L Mammographic densities in a one-year isoflavone intervention Eur J Cancer Prev 2003 12 165 169 12671541 10.1097/00008469-200304000-00011
Atkinson C Warren RM Sala E Dowsett M Dunning AM Healey CS Runswick S Day NE Bingham SA Red-clover-derived isoflavones and mammographic breast density: a double-blind, randomized, placebo-controlled trial [ISRCTN42940165] Breast Cancer Res 2004 6 R170 R179 15084240 10.1186/bcr773
Cuzick J Warwick J Pinney E Warren RM Duffy SW Tamoxifen and breast density in women at increased risk of breast cancer J Natl Cancer Inst 2004 96 621 628 15100340
Couto E Harrison D Duffy S Myles J Sala E Warren R Day N Luben R Chen H Estimation of disease progression parameters from case-control data: application to mammographic patterns and breast cancer natural history J Epidemiol Biostat 2001 6 235 242 11434503 10.1080/135952201753173015
Ciatto S Visioli C Paci E Zappa M Breast density as a determinant of interval cancer at mammographic screening Br J Cancer 2004 90 393 396 14735182 10.1038/sj.bjc.6601548
Sala E Warren R McCann J Duffy S Day N Luben R Mammographic parenchymal patterns and mode of detection: implications for the breast screening programme J Med Screen 1998 5 207 212 9934652
Warren RM Young JR McLean L Lyons K Wilson AR Evans A Duffy SW Warsi IM Radiology review of the UKCCCR Breast Screening Frequency Trial: potential improvements in sensitivity and lead time of radiological signs Clin Radiol 2003 58 128 132 12623041 10.1053/crad.2002.1132
Harrison DA Duffy SW Sala E Warren RM Couto E Day NE Deterministic models for breast cancer progression: application to the association between mammographic parenchymal pattern and histologic grade of breast cancers J Clin Epidemiol 2002 55 1113 1118 12507675 10.1016/S0895-4356(02)00483-3
Wolfe JN Saftlas AF Salane M Mammographic parenchymal patterns and quantitative evaluation of mammographic densities: a case-control study AJR Am J Roentgenol 1987 148 1087 1092 3495132
Warner E Lockwood G Tritchler D Boyd NF The risk of breast cancer associated with mammographic parenchymal patterns: a meta-analysis of the published literature to examine the effect of method of classification Cancer Detect Prev 1992 16 67 72 1532349
Boyd NF Lockwood GA Byng JW Tritchler DL Yaffe MJ Mammographic densities and breast cancer risk Cancer Epidemiol Biomarkers Prev 1998 7 1133 1144 9865433
Ursin G Ma H Wu AH Bernstein L Salane M Parisky YR Astrahan M Siozon CC Pike MC Mammographic density and breast cancer in three ethnic groups Cancer Epidemiol Biomarkers Prev 2003 12 332 338 12692108
Maskarinec G Meng L A case-control study of mammographic densities in Hawaii Breast Cancer Res Treat 2000 63 153 161 11097091 10.1023/A:1006486319848
Vacek PM Geller BM A prospective study of breast cancer risk using routine mammographic breast density measurements Cancer Epidemiol Biomarkers Prev 2004 13 715 722 15159301
Jakes RW Duffy SW Ng FC Gao F Ng EH Mammographic parenchymal patterns and risk of breast cancer at and after a prevalence screen in Singaporean women Int J Epidemiol 2000 29 11 19 10750598 10.1093/ije/29.1.11
Byrne C Hankinson SE Pollak M Willett WC Colditz GA Speizer FE Insulin-like growth factors and mammographic density Growth Horm IGF Res 2000 10 Suppl A S24 S25 10984280 10.1016/S1096-6374(00)90011-X
Haiman CA Hankinson SE De V IGuillemette C Ishibe N Hunter DJ Byrne C Polymorphisms in steroid hormone pathway genes and mammographic density Breast Cancer Res Treat 2003 77 27 36 12602902 10.1023/A:1021112121782
van Gils CH Hendriks JH Otten JD Holland R Verbeek AL Parity and mammographic breast density in relation to breast cancer risk: indication of interaction Eur J Cancer Prev 2000 9 105 111 10830577 10.1097/00008469-200004000-00006
Byng JW Yaffe MJ Lockwood GA Little LE Tritchler DL Boyd NF Automated analysis of mammographic densities and breast carcinoma risk Cancer 1997 80 66 74 9210710 10.1002/(SICI)1097-0142(19970701)80:1<66::AID-CNCR9>3.0.CO;2-D
Maskarinec G Meng L Ursin G Ethnic differences in mammographic densities Int J Epidemiol 2001 30 959 965 11689504 10.1093/ije/30.5.959
Maskarinec G Lyu LC Meng L Theriault A Ursin G Determinants of mammographic densities among women of Asian, Native Hawaiian, and Caucasian ancestry Ethn Dis 2001 11 44 50 11289250
Bjurstam N Bjorneld L Warwick J Sala E Duffy SW Nystrom L Walker N Cahlin E Eriksson O Hafstrom LO The Gothenburg Breast Screening Trial Cancer 2003 97 2387 2396 12733136 10.1002/cncr.11361
Wolfe JN Breast parenchymal patterns and their changes with age Radiology 1976 121 545 552 981644
Fleiss JL Statistical methods for rates and Proportions 1981 2 New York: Wiley
SAS/STAT User's Guide 1999 Cary, NC: SAS Institute Inc
Venta LA Hendrick RE Adler YT DeLeon P Mengoni PM Scharl AM Comstock CE Hansen L Kay N Coveler A Rates and causes of disagreement in interpretation of full-field digital mammography and film-screen mammography in a diagnostic setting AJR Am J Roentgenol 2001 176 1241 1248 11312188
Brisson J Diorio C Masse B Wolfe's parenchymal pattern and percentage of the breast with mammographic densities: redundant or complementary classifications? Cancer Epidemiol Biomarkers Prev 2003 12 728 732 12917203
Funkhouser E Waterbor JW Cole P Rubin E Mammographic patterns and breast cancer risk factors among women having elective screening South Med J 1993 86 177 180 8434288
Vachon CM Kuni CC Anderson K Anderson VE Sellers TA Association of mammographically defined percent breast density with epidemiologic risk factors for breast cancer (United States) Cancer Causes Control 2000 11 653 662 10977110 10.1023/A:1008926607428
El Bastawissi AY White E Mandelson MT Taplin SH Reproductive and hormonal factors associated with mammographic breast density by age (United States) Cancer Causes Control 2000 11 955 963 11142530 10.1023/A:1026514032085
Andersson I Janzon L Pettersson H Radiographic patterns of the mammary parenchyma: variation with age at examination and age at first birth Radiology 1981 138 59 62 7455098
Sala E Warren R McCann J Duffy S Luben R Day N High-risk mammographic parenchymal patterns, hormone replacement therapy and other risk factors: a case-control study Int J Epidemiol 2000 29 629 636 10922338 10.1093/ije/29.4.629
Myles JP Salmininen T Duffy SW Prevost TC Day NE Hakama M Prospective estimation of rates of change in mammographic parenchymal patterns: influence of age and of hormone replacement therapy Breast 2004 13 56 60 14759717 10.1016/j.breast.2003.10.007
Warren R Hormones and mammographic breast density Maturitas 2004 49 67 78 15351098 10.1016/j.maturitas.2004.06.013
Ursin G Pike MC Spicer DV Porrath SA Reitherman RW Can mammographic densities predict effects of tamoxifen on the breast? J Natl Cancer Inst 1996 88 128 129 8537975
Gram IT Ursin G Spicer DV Pike MC Reversal of gonadotropin-releasing hormone agonist induced reductions in mammographic densities on stopping treatment Cancer Epidemiol Biomarkers Prev 2001 10 1117 1120 11700258
Rutter CM Mandelson MT Laya MB Seger DJ Taplin S Changes in breast density associated with initiation, discontinuation, and continuing use of hormone replacement therapy JAMA 2001 285 171 176 11176809 10.1001/jama.285.2.171
|
16168132
|
PMC1242160
|
CC BY
|
2021-01-04 16:54:51
|
no
|
Breast Cancer Res. 2005 Aug 25; 7(5):R854-R861
|
utf-8
|
Breast Cancer Res
| 2,005 |
10.1186/bcr1308
|
oa_comm
|
==== Front
Breast Cancer ResBreast Cancer Research1465-54111465-542XBioMed Central London bcr13101616812910.1186/bcr1310Research ArticleA new option for early breast cancer patients previously irradiated for Hodgkin's disease: intraoperative radiotherapy with electrons (ELIOT) Intra Mattia [email protected] Oreste 1Veronesi Paolo 12Ciocca Mario 3Luini Alberto 1Lazzari Roberta 4Soteldo Javier 1Farante Gabriel 1Orecchia Roberto 24Veronesi Umberto 11 Department of Breast Surgery, European Institute of Oncology, Milan, Italy2 University of Milan, School of Medicine, Milan, Italy3 Medical Physics Unit, European Institute of Oncology, Milan, Italy4 Department of Radiotherapy, European Institute of Oncology, Milan, Italy2005 16 8 2005 7 5 R828 R832 4 2 2005 18 3 2005 28 6 2005 21 7 2005 Copyright © 2005 Intra et al. licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Introduction
Patients who have undergone mantle radiotherapy for Hodgkin's disease (HD) are at increased risk of developing breast cancer. In such patients, breast conserving surgery (BCS) followed by breast irradiation is generally considered contraindicated owing to the high cumulative radiation dose. Mastectomy is therefore recommended as the first option treatment in these women.
Methods
Six patients affected by early breast cancer previously treated with mantle radiation for HD underwent BCS associated with full-dose intraoperative radiotherapy with electrons (ELIOT).
Results
A total dose of 21 Gy (prescribed at 90% isodose) in five cases and 17 Gy (at 100% isodose) in one case were delivered directly to the mammary gland without acute complications and with good cosmetic results. After an average of 30.8 months of follow up, no late sequelae were observed and the patients are free of disease.
Conclusion
In patients previously irradiated for HD, ELIOT can avoid repeat irradiation of the whole breast, permit BCS and decrease the number of avoidable mastectomies.
==== Body
Introduction
The risk of developing breast cancer after radiotherapy for Hodgkin's disease (HD) is well documented and is attributed to the incidental inclusion of portions of the breast in the portals used to irradiate the mediastinum with or without infraclavicular/axillary regions [1,2]. This increased risk depends on several patient and treatment factors, including radiation dose, radiation treatment field and age at treatment [3]. The estimated risk is also affected by the duration of follow up after radiotherapy and the method used to calculate risk [4]. Based on concern about possible severe sequelae arising after a high total cumulative dose to the breast, several authors [5,6] have suggested mastectomy as the treatment of choice for breast cancer arising after radiotherapy for HD.
Few studies have been published whereby alternative therapies to mastectomy have been employed; they describe local excision alone, or interstitial brachytherapy and local excision followed by local field external beam radiotherapy [7-10].
In order to avoid a dangerously high total cumulative dose of radiotherapy to the whole breast, soft tissues of the thoracic wall, lung and heart, without rejecting the possibility of breast conserving surgery (BCS), we evaluated the potential of performing full-dose intraoperative radiotherapy with electrons (ELIOT) in women with early breast cancer who had been previously irradiated for HD.
Materials and methods
Between July 1999 and April 2005, 756 breast cancer patients who were candidates for BCS received ELIOT as the sole radiation treatment in the European Institute of Oncology in Milan. Of these, 100 patients had been treated in the past within a preliminary validation phase I study [11,12] and 440 patients are currently in an institutional phase III ongoing randomized clinical trial, in which 21 Gy ELIOT (prescribed at 90% isodose) is prospectively compared with external postoperative fractionated radiotherapy (PFR) including elective treatment of the whole breast. Eligibility criteria for the ELIOT randomized trial included patients aged between 48 and 75 years affected by a unicentric breast invasive carcinoma with a maximum diameter of 2.5 cm. Locally advanced tumors (T3 and T4), the presence of a contralateral synchronous or metachronous tumor, non-invasive carcinomas (including Paget's disease), other breast malignancies different from carcinoma, multifocality or multicentricity of the disease, previous surgical biopsy and previous oncological history are considered exclusion criteria, while a breast thickness superior to 3 cm is considered a technical contraindication, it being impossible to deliver energy of more than 9 mega-electron volts (MeV) with the present ELIOT dedicated linac. No particular tumor breast locations are excluded from ELIOT, but the close proximity of the tumor to the skin and pectoral major muscle infiltration are considered relative contraindication for ELIOT. In fact, in these cases, the irradiation of the entire breast, including skin and pectoral major muscle, is mandatory. Finally, if the tumor is very close to the axilla, ELIOT is avoided due to the risk of irradiating the axillary nerves and vessels.
In a selected group of 216 invasive breast cancer patients who were candidates for BCS, and who had specifically asked for and signed informed consent for ELIOT or in whom PFR was not considered safe or feasible due to clinical reasons (e.g. severe cardiopathy, large scars from skin burns, vitiligo, geographic or social obstacles), ELIOT was performed outside the randomized trial. In this selected group of patients, only the tumor size (larger than 2.5 cm), locally advanced disease, multifocality or multicentricity were considered exclusion criteria. Of these 216 patients, 6 were previously treated for HD and received mantle radiation (Table 1).
The breast cancer presentation of these six patients is described in Table 2. Five patients underwent BCS and radioguided sentinel lymph node biopsy according to the standard technique adopted at the European Institute of Oncology [13]. In four out of the five cases, the biopsy showed that the sentinel lymph nodes were free of metastasis, so complete axillary dissection was not performed. In one patient, due to the presence of clinically metastatic axillary nodes, a complete axillary dissection was performed directly.
The radio-surgical technique of the six ELIOT procedures did not differ from the 'classic' technique, which has been described previously [14] in patients not previously irradiated for HD.
To evaluate acute and late radiation morbidity, the Radiation Therapy Oncology Group (RTOG) and European Organization for Research and Treatment of Cancer (EORTC) scoring scheme [15] was applied on the first and seventh day of the radio-surgical treatment and every six months during follow up.
Radiation therapy device
The dedicated ELIOT machine (Novac7, Hitesys, Italy) is a miniaturized linear accelerator that can be moved by a robotic arm, docked in the operating room and fulfils the statutory requirements with respect to radiological protection. Additional barriers (from 5 to 15 mm lead) are provided for positioning around the operating table. A lead shield (15 cm thick) is also placed under the surgical bed corresponding to the electron field (beam stopper). ELIOT delivers a single full-dose of radiation directly to the tumor bed after removal of the tumor. Based on the radiobiological models used to predict radiation effects (linear-quadratic surviving fraction or multitarget surviving fraction) [16], we can estimate that a dose of 60 Gy delivered at 2 Gy daily, which is the radiation dose required to control the microscopical residual disease after breast resection, is equivalent to a single fraction of 20 to 22 Gy when using an α/β ratio at 10 Gy, typical for tumors and acute reacting tissues. Using the same equation, but calculating the tolerance of late responding tissues (α/β ratio at 3 Gy), this equivalent value increases to at least 110 Gy. The linac does not produce photons, but only delivers electron beams at four different energy levels: 3, 5, 7 and 9 MeV as nominal energies, corresponding to 4.5, 5.2, 6.5 and 7.8 MeV effective energies to the phantom surface, respectively. The depth of 80% isodose ranged between 13 mm (3 MeV, not used in clinical practice) and 24 mm (9 MeV).
In one patient, a total dose of 17 Gy (prescribed at 100% isodose) using electron beams (7 MeV energy) was delivered. In five patients, 21 Gy (prescribed at 90% isodose, 7 or 9 MeV of energy) were administered.
Results
The biological characteristics of the tumors are shown in Table 3. Two patients received adjuvant chemotherapy with cyclophosphamide-methotrexate/5-fluorouracil, followed by tamoxifen 20 mg/day for 5 years associated in one case with LHRH analogue for 2 years. One patient received adriamicin/cyclophosphamide followed by tamoxifen 20 mg/day for 5 years associated with LHRH analogue for 2 years. Three patients received tamoxifen 20 mg/day for 5 years, associated in two cases with LHRH analogue for 2 years.
ELIOT was well tolerated in all patients without any unusual acute reactions (grade 0 according to the EORTC/RTOG acute radiation morbidity criteria). In the patients with longer follow up, no late sequelae were observed (grade 0 according to the EORTC/RTOG late radiation morbidity scoring scheme) and all women had an excellent cosmetic result. All the patients are free of disease after 53, 38, 29, 24, 23 and 17 months of follow up (average 30.8).
Discussion
Previous mantle radiotherapy for HD is considered a contraindication to BCS and radiotherapy [17]. This is based on concerns about possible severe sequelae arising from a high total cumulative dose, exceeding normal tissue tolerance, being delivered to the portions of the breast that have presumably already received radiation for the lymphoma, even though modifications of the breast gland over time render the exact calculation of the dose infeasible many years after radiation delivery. Most authors consider these patients at significant risk of complications (fibrosis, skin and soft tissue necrosis, rib fractures, potential lung and heart toxicities) [6] and do not candidate them for BCS and adjuvant radiotherapy. In contrast, other reports [18,19] support BCS followed by PFR when breast cancer develops many years after radiotherapy for HD. At present, no consensus exists regarding the correct management of breast cancer after mantle irradiation for HD and, given the discordant results and the small number of women treated with BCS, mastectomy continues to be recommended as the standard treatment. To avoid a high total cumulative dose to portions of the breast or soft tissues of the thoracic wall, one of the conservative options is to treat just the tumor bed: the irradiation of a small volume of the breast and adjacent structures could allow the risk of complications to be minimized [7-10]. Therefore, over the past decade there has been increasing interest in a variety of radiation techniques designed to treat only the portion of the breast deemed to be at high risk for local recurrence [20]: these include brachytherapy implants, MammoSite applicator, intraoperative orthovoltage device, and 3D conformal or intensity-modulated external radiotherapy (partial-breast irradiation (PBI)). All these techniques have similar indications but different applications [21,22]. In particular, they differ in the source of radiation (for example, X-ray, Iridium, photons) and the amount of breast volume treated. Although 5 to 7 year outcome data on patients treated with PBI are now becoming available, many issues remain unresolved, including the clinical and pathological selection criteria of patients, radiation dose and fractionation, and how these relate to the standard fractionation for whole breast irradiation, appropriate target volume, local control within the untreated ipsilateral breast tissue, and overall survival. With a view to furnishing guidelines and clarifying the issues of major controversy, a workshop on PBI was held in Bethesda, in December 2002. The workshop report emphasized the importance of education and training with regard to the results of PBI as it becomes an emerging clinical treatment [23]. In particular, ELIOT, a new radiotherapeutic technique that delivers a single dose of radiation directly to the tumor bed during the conservative surgical treatment of early breast cancer, has been proposed for evaluation in randomized clinical trials as a possible alternative to standard PFR.
When in July 1999 we focused our interest on the use of intraoperative radiotherapy as an exclusive treatment in small unifocal infiltrating breast carcinomas, we considered that ELIOT could be specifically applied in all those situations in which PFR was not considered safe or feasible for various reasons (for example, severe cardiopathy, large hypertrophic scarring from skin burns, vitiligo, geographic or social obstacles) [24] and, in particular, in patients irradiated for HD. After our first preliminary report [25], ELIOT was proposed in six patients previously irradiated for HD. In all cases, the dose delivered and the energy of electron beams from ELIOT did not differ from those in the other 549 patients submitted to ELIOT; in one patient a total dose of 17 Gy (prescribed at 100% isodose) using electron beams at 7 MeV energy and in five patients doses of 21 Gy (90% isodose, 7 and 9 MeV of energy) were delivered. The different electron energies administrated (7 or 9 MeV) are related to the different breast gland thickness.
The radio-surgical technique of the six ELIOT procedures did not differ from the 'classic' technique used for previously non-irradiated patients [14]. In particular, to ensure a good coverage of the target by the radiation dose and maximal protection of the normal tissues in the operative area, adequate preparation for irradiation of the portion of breast gland to undergo ELIOT is necessary. Protective devices are placed between the gland and the pectoral muscle; a dedicated lead disk 5 mm thick and an aluminium disk 4 mm thick, available in various diameters (4, 5, 6, 8, 10 cm) are commonly used. The wall protection is guaranteed both by the absorption properties of the lead and aluminium and the 9 mm outdistance created by the disks.
We scored the radiation morbidity according to the RTOG/EORTC criteria [15]. ELIOT was well tolerated without any unusual acute reactions despite previous breast irradiation. We did not observe any ischemic or necrotic problems of the skin flap due to the careful sparing of the subcutaneous vessels during the mobilization of the residual breast around the tumor bed. No increased post-operative complications (pain, seroma, emathoma, infection) were observed in these six patients when compared to the overall group of ELIOT patients. The length of hospital stay was therefore not prolonged. The cosmetic outcome was also very good in all patients: no skin erythema was observed as a result of the complete removal of the skin from the radiation beam.
The follow up is too short (average 30.8 months) to evaluate late sequelae in these six re-irradiated patients but no complications were observed in the four patients with more than 2 years of follow up.
Conclusion
ELIOT dramatically reduces the radiation exposure of the normal tissues and, in particular, of the previously irradiated breast, avoiding a high total cumulative dose to the gland and to the soft tissues of the thoracic wall. In patients previously treated for HD, ELIOT permits BCS independently of the interval between mantle radiotherapy and breast surgery, without acute local complications, decreasing the number of avoidable mastectomies. A longer follow up is necessary to evaluate the incidence of radiation-induced sequelae and the local recurrence rate.
Abbreviations
BCS = breast conserving surgery; ELIOT = intraoperative radiotherapy with electrons; EORTC = European Organization for Research and Treatment of Cancer; HD = Hodgkin's disease; MeV = mega-electron volts; PBI = partial-breast irradiation;PFR = postoperative fractionated radiotherapy; RTOG = Radiation Therapy Oncology Group.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MI conceived the study, and participated in its design and coordination and drafted the manuscript, OG helped to draft the manuscript, PV participated in the coordination of the study, MC participated in the coordination of the study, AL participated in the design of the study, JS carried out patient follow up, GF carried out patient follow up, and RO and UV participated in the design of the study. All authors read and approved the final manuscript.
Acknowledgements
The authors wish to acknowledge the AIRC (Italian Association for Cancer Research) and the AICF (American Italian Cancer Foundation), which partially funded this research project.
Figures and Tables
Table 1 Presentation of and therapy for Hodgkin's disease
Patient no. Age (years) Diagnosis Stage Therapy
1 29 HD, MC IIA MOPP × 6, Mantle 36 Gy
2 28 HD, NS IIB MOPP × 4, ABVD × 4, Mantle 20 Gy
3 42 Had IIB ABVD × 3, Mantle 40 Gy
4 38 HD, MC IIA MOPP × 6, Mantle 40 Gy
5 24 HD, NS IIB MOPP × 6, Mantle 40 Gy
6 30 HD, NS IIA MOPP × 4, ABVD × 4, Mantle 36 Gy
aNo histologic subtype determined. ABVD, adriamicine, bleomycin, vinblastine, dacarbazine; HD, Hodgkin's disease; MC, mixed cellularity; MOPP, nitrogen mustard, vincristine, procarbazine, prednisone; NS, nodular sclerosis.
Table 2 Presentation of breast cancer
Patient no. Age (years) Interval since radiation for HD (years) ELIOT dose
1 35 16 21 Gy (90% isodose), 7 MeV
2 42 14 17 Gy (100% isodose), 7 MeV
3 50 8 21 Gy (90% isodose), 9 MeV
4 50 12 21 Gy (90% isodose), 9 MeV
5 34 10 21 Gy (90% isodose), 7 MeV
6 39 9 21 Gy (90% isodose), 9 MeV
HD, Hodgkin's disease; MeV, mega electron volts.
Table 3 Characteristics of the breast cancer
Patient no. Tumor site T N G ER, PgR KI67 PVI
1 UO 1c 0 (sn) 2 15%, 25% 11% No
2 LI 1c 0 (sn) 3 0%, 0% 70% No
3 C 1c (is) 0 (sn) 1 90%, 90% 13% No
4 S 1c 1a 2 85%, 85% 25% No
5 UO 2 1a 2 30%, 0% 27% Yes
6 UO 1b 0(sn) 1 90%, 90% 8% No
T, tumor size; N, axillary lymph nodes; G, tumor grade; ER, estrogen receptors; PgR, progesterone receptors; KI67, proliferative index; C, central; is, extended intraductal component associated; LI, lower inner; PVI, perivascular invasion; S, superior; sn, sentinel node; UO, upper outer.
==== Refs
Boivin JF O'Brien K Solid cancer risk after treatment of Hodgkin's disease Cancer 1988 61 2541 2546 3284637
Tucker MA Coleman CN Cox RS Risk of second cancers after treatment for Hodgkin's disease N Engl J Med 1988 318 76 81 3336397
Tinger A Wasserman TH Klein EE Miller EA Roberts T Piephoff JV Kucik NA The incidence of breast cancer following mantle field radiation therapy as a function of dose and technique Int J Radiat Oncol Biol Phys 1997 37 865 870 9128963 10.1016/S0360-3016(97)00026-6
Wolden SL Lamborn KR Cleary SF Tate DJ Donaldson SS Second cancers following pediatric Hodgkin's disease J Clin Oncol 1998 16 536 544 9469338
American College of Radiotherapy, American College of Surgeons, College of American Pathologists, Society of Surgical Oncology Winchester DP Cox JD Standards for breast-conservation treatment CA Cancer J Clin 1992 42 134 162 1568135
Wolden SL Hancock SL Carlson RW Goffinet DR Jeffrey SS Hoppe RT Management of breast cancer after Hodgkin's disease J Clin Oncol 2000 18 765 772 10673517
Krishnan L Jewell WR Tawfik OW Krishnan EC Breast conservation therapy with tumor bed irradiation alone in a selected group of patients with stage I breast cancer Breast J 2001 7 91 96 11328314 10.1046/j.1524-4741.2001.007002091.x
Perera F Engel J Holliday R Scott L Girotti M Girvan D Chisela F Venkatesan V Local resection and brachytherapy confined to the lumpectomy site for early breast cancer: a pilot study J Surg Oncol 1997 65 263 267 9274791 10.1002/(SICI)1096-9098(199708)65:4<263::AID-JSO7>3.0.CO;2-3
Baglan KL Martinez AA Frazier RC Kini VR Kestin LL Chen PY Edmundson G Mele E Jaffray D Vicini FA The use of high-dose-rate brachytherapy alone after lumpectomy in patients with early-stage breast cancer treated with breast-conserving therapy Int J Radiat Oncol Biol Phys 2001 50 1003 1011 11429228 10.1016/S0360-3016(01)01547-4
Vicini F Kini VR Chen P Horwitz E Gustafson G Benitez P Edmundson G Goldstein N McCarthy K Martinez A Irradiation of the tumor bed alone after lumpectomy in selected patients with early-stage breast cancer treated with breast conserving therapy J Surg Oncol 1999 70 33 40 9989418 10.1002/(SICI)1096-9098(199901)70:1<33::AID-JSO6>3.0.CO;2-O
Veronesi U Orecchia R Luini A Gatti G Intra M Veronesi P Gatzemeier W Ciocca M Tosi G Ivaldi G Focalised intraoperative irradiation after conservative surgery for early stage breast cancer The Breast 2001 10 Suppl 3 84 89
Veronesi U Orecchia R Luini A Gatti G Intra M Zurrida S Ivaldi G Tosi G Ciocca M Tosoni A De Lucia F A preliminary report of intraoperative radiotherapy (IORT) in limited-stage breast cancers that are conservatively treated Eur J Cancer 2001 37 2178 2183 11677104 10.1016/S0959-8049(01)00285-4
Veronesi U Paganelli G Viale G Galimberti V Luini A Zurrida S Robertson C Sacchini V Veronesi P Orvieto E Sentinel lymph node biopsy and axillary dissection in breast cancer: results in a large series J Natl Cancer Inst 1999 91 368 373 10050871 10.1093/jnci/91.4.368
Intra M Gatti G Luini A Galimberti V Veronesi P Zurrida S Frasson A Ciocca M Orecchia R Veronesi U Surgical technique of intraoperative radiotherapy in conservative treatment of limited-stage breast cancer Arch Surg 2002 137 737 740 12049547 10.1001/archsurg.137.6.737
Cox JD Stetx J Pajak TF Toxicity criteria of the Radiation Therapy Oncology Group (RTOG) and the European Organization for Research and Treatment of Cancer (EORTC) Int J Radiat Oncol Biol Phys 1995 31 1341 1346 7713792 10.1016/0360-3016(95)00060-C
Tucker SS Thames HD Taylor JM How well is the probability of tumor cure after fractionated irradiation described by Poisson statistics Radiat Res 1990 124 273 282 2263727
The Steering Committee on Clinical Practice Guidelines for the Care and Treatment of Breast Cancer Breast radiotherapy after breast-conserving surgery CMAJ 1998 158 Suppl 3 S35 42 9484277
Aref I Cross P Conservative surgery and radiation therapy for early stage breast cancer after previous mantle radiation for Hodgkin's disease Br J Radiol 2000 73 905 906 11026871
Deutsch M Gerszten K Bloomer W Avisar E Lumpectomy and breast irradiation for breast cancer arising after previous radiotherapy for Hodgkin's disease Am J Clin Oncol 2001 24 33 34 11232946 10.1097/00000421-200102000-00005
Veronesi U Marubini E Mariani L Galimberti V Luini A Veronesi P Salvadori B Zucali R Radiotherapy after breast-conserving surgery in small breast carcinoma: long-term results of a randomized trial Ann Oncol 2001 12 997 1003 11521809 10.1023/A:1011136326943
Dickler A Kirk MC Chu J Nguyen C The MammoSitetrade mark breast brachytherapy applicator: A review of technique and outcomes Brachytherapy 2005 4 130 136 15893266 10.1016/j.brachy.2004.12.003
Weed DW Edmundson GK Vicini FA Chen PY Martinez AA Accelerated partial breast irradiation: A dosimetric comparison of three different techniques Brachytherapy 2005 4 121 129 15893265 10.1016/j.brachy.2004.12.005
Wallner P Arthur D Bartelink H Connolly J Edmundson G Giuliano A Goldstein N Hevezi J Julian T Kuske R Workshop on partial breast irradiation: state of the art and the science, Bethesda, MD, December 8–10, 2002 J Natl Cancer Inst 2004 96 175 184 14759984
Intra M Orecchia R Veronesi U Intraoperative radiotherapy: the debate continues Lancet Oncol 2004 5 340 15172353 10.1016/S1470-2045(04)01489-5
Intra M Leonardi MC Gatti G Vento AR Ciocca M Veronesi P Bassani G Dos Santos GR Rodriguez J Luini A Intraoperative radiotherapy during breast conserving surgery in patients previously treated with radiotherapy for Hodgkin's disease Tumori 2004 90 13 16 15143964
|
16168129
|
PMC1242162
|
CC BY
|
2021-01-04 16:54:50
|
no
|
Breast Cancer Res. 2005 Aug 16; 7(5):R828-R832
|
utf-8
|
Breast Cancer Res
| 2,005 |
10.1186/bcr1310
|
oa_comm
|
==== Front
Breast Cancer ResBreast Cancer Research1465-54111465-542XBioMed Central London bcr13121616813110.1186/bcr1312Research ArticleThe use of cystatin C to inhibit epithelial–mesenchymal transition and morphological transformation stimulated by transforming growth factor-β Sokol Jonathan P [email protected] Jason R [email protected] Barbara J [email protected] William P [email protected] Program in Cell Biology, Department of Pediatrics, National Jewish Medical and Research Center, Denver, CO, USA2005 23 8 2005 7 5 R844 R853 31 8 2004 29 9 2004 11 6 2005 26 7 2005 Copyright © 2005 Sokol et al.; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Introduction
Transforming growth factor-β (TGF-β) is a potent suppressor of mammary epithelial cell (MEC) proliferation and is thus an inhibitor of mammary tumor formation. Malignant MECs typically evolve resistance to TGF-β-mediated growth arrest, enhancing their proliferation, invasion, and metastasis when stimulated by TGF-β. Recent findings suggest that therapeutics designed to antagonize TGF-β signaling may alleviate breast cancer progression, thereby improving the prognosis and treatment of breast cancer patients. We identified the cysteine protease inhibitor cystatin C (CystC) as a novel TGF-β type II receptor antagonist that inhibits TGF-β binding and signaling in normal and cancer cells. We hypothesized that the oncogenic activities of TGF-β, particularly its stimulation of mammary epithelial–mesenchymal transition (EMT), can be prevented by CystC.
Method
Retroviral infection was used to constitutively express CystC or a CystC mutant impaired in its ability to inhibit cathepsin protease activity (namely Δ14CystC) in murine NMuMG MECs and in normal rat kidney (NRK) fibroblasts. The effect of recombinant CystC administration or CystC expression on TGF-β stimulation of NMuMG cell EMT in vitro was determined with immunofluorescence to monitor rearrangements of actin cytoskeletal architecture and E-cadherin expression. Soft-agar growth assays were performed to determine the effectiveness of CystC in preventing TGF-β stimulation of morphological transformation and anchorage-independent growth in NRK fibroblasts. Matrigel invasion assays were performed to determine the ability of CystC to inhibit NMuMG and NRK motility stimulated by TGF-β.
Results
CystC and Δ14CystC both inhibited NMuMG cell EMT and invasion stimulated by TGF-β by preventing actin cytoskeletal rearrangements and E-cadherin downregulation. Moreover, both CystC molecules completely antagonized TGF-β-mediated morphological transformation and anchorage-independent growth of NRK cells, and inhibited their invasion through synthetic basement membranes. Both CystC and Δ14CystC also inhibited TGF-β signaling in two tumorigenic human breast cancer cell lines.
Conclusion
Our findings show that TGF-β stimulation of initiating metastatic events, including decreased cell polarization, reduced cell–cell contact, and elevated cell invasion and migration, are prevented by CystC treatment. Our findings also suggest that the future development of CystC or its peptide mimetics hold the potential to improve the therapeutic response of human breast cancers regulated by TGF-β.
==== Body
Introduction
Oncogenic epithelial–mesenchymal transitions (EMTs) comprise a complex array of gene expression and repression that elicits tumor metastasis in localized carcinomas [1,2]. The acquisition of metastatic phenotypes by dedifferentiated tumors is the most lethal facet of cancer and is the leading cause of cancer-related death [3,4]. Transforming growth factor-β (TGF-β) is a powerful tumor suppressor that normally represses these processes by prohibiting epithelial cell proliferation, and by creating a cell microenvironment that inhibits epithelial cell motility, invasion, and metastasis [5,6]. Carcinogenesis often subverts the tumor-suppressing function of TGF-β, thereby endowing TGF-β with oncogenic activities that promote the growth and spread of developing tumors, including the initiation and stabilization of tumor EMT [1,2,5-7]. The duality of TGF-β in both suppressing and promoting cancer development was observed originally with transgenic TGF-β1 expression in mouse keratinocytes, which initially suppressed benign skin tumor formation before promoting malignant conversion and spindle cell carcinoma generation [8]. More recently, TGF-β signaling was shown to inhibit the tumorigenicity of normal, premalignant, and malignant breast epithelial cells, while stimulating that of highly invasive and metastatic breast cancer cells [9]. Fundamental gaps exist in our knowledge of how malignant cells overcome the cytostatic actions of TGF-β and of how TGF-β stimulates the progression of developing tumors. Indeed, these knowledge gaps have prevented science and medicine from developing treatments effective in antagonizing TGF-β oncogenicity in progressing cancers, particularly those of the breast.
The ability of TGF-β to induce cancer growth and metastasis suggests that developing therapeutics to antagonize and/or circumvent TGF-β signaling may prove effective in treating metastatic malignancies, perhaps by preventing the stimulation of EMT by TGF-β. Cystatin C (CystC) is a small, ubiquitously expressed cysteine protease inhibitor found in nearly all bodily fluids [10,11]. By inactivating cathepsin protease activities, CystC reduces bone resorption, neutrophil chemotaxis, and tissue inflammation, and also inhibits cancer cell invasion [10]. We showed recently that CystC not only inhibits cathepsin-mediated invasion but also antagonizes TGF-β signaling in normal and cancer cells by interacting physically with the TGF-β type II receptor (TβR-II), thereby preventing TGF-β binding [12]. Thus, CystC is a novel TβR-II antagonist that may prove useful in blocking the oncogenic activities of TGF-β, particularly its ability to stimulate EMT. In the present study we tested this hypothesis by measuring the ability of CystC to antagonize the oncogenic activities of TGF-β in two established in vitro models of cancer progression: first, EMT of normal murine NMuMG mammary epithelial cells (MECs), and second, morphological transformation and anchorage-independent growth of normal rat kidney (NRK) fibroblasts. We show that CystC significantly decreased TGF-β stimulation of EMT and morphological transformation in mammary epithelial cells and kidney fibroblasts, respectively. We show further that CystC significantly antagonized TGF-β signaling in two tumorigenic human breast cancer cell lines. Thus, by antagonizing TGF-β signaling and preventing its stimulation of EMT, CystC may represent a novel TGF-β chemopreventive agent effective in neutralizing the tumor-promoting activities of TGF-β and its stimulation of tumor metastasis.
Materials and methods
Recombinant CystC expression and purification
The synthesis of bacterial expression vectors encoding human CystC or Δ14CystC fused to the carboxy terminus of glutathione S-transferase (GST), and their purification from transformed Escherichia coli, were described previously [12].
Retroviral CystC expression
The creation of bicistronic retroviral vectors (namely pMSCV-IRES–GFP, where GFP stands for green fluorescent protein) encoding human CystC or Δ14CystC was described previously [12]. Mouse NMuMG, rat NRK kidney fibroblasts, and human MDA-MB-231 and MCF10A-CA1a breast cancer cells were infected overnight with control (pMSCV-IRES–GFP), CystC, or Δ14CystC retroviral supernatants produced by EcoPac2 retroviral packaging cells (Clontech) as described [12]. Cells expressing GFP were isolated and collected 48 hours later on a MoFlo cell sorter (Cytomation), and were subsequently expanded to yield stable polyclonal populations of control, CystC-expressing, or Δ14CystC-expressing cells. The expression and secretion of recombinant CystC proteins by individual infected cell lines were monitored by immunoblotting conditioned medium with anti-CystC antibodies as described [12].
Immunofluorescence studies
The ability of TGF-β to alter actin cytoskeletal architecture and E-cadherin expression was monitored essentially as described [13]. In brief, control, CystC-expressing, or Δ14CystC-expressing NMuMG cells were allowed to adhere overnight to glass coverslips in 24-well plates (50,000 cells per well). The cells were stimulated the following day with TGF-β1 (5 ng/ml) for 0 to 36 hours at 37°C. In some experiments, control NMuMG cells were stimulated with TGF-β1 in the absence or presence of 10 μg/ml recombinant GST, GST–CystC, or GST–Δ14CystC. On completion of agonist stimulation, the cells were washed in ice-cold PBS and immediately fixed in 3.7% formaldehyde. After extensive washing in PBS, the cells were blocked in PBS supplemented with 1.5% FBS, followed by incubation with rhodamine–phalloidin (0.25 μM). Alternatively, the cells were blocked in goat γ-globulin (200 μg/ml; Jackson Immunoresearch) before the detection of E-cadherin by sequential incubations with monoclonal anti-E-cadherin (1:50 dilution; BD Bioscience), followed by biotinylated goat anti-mouse antibody (5 μg/ml; Jackson Immunoresearch), and finally by Alexa-streptavidin (1.2 μg/ml; Molecular Probes). Images were captured on a Nikon Diaphot microscope.
The ability of TGF-β to alter E-cadherin expression was also monitored by immunoprecipitating E-cadherin from buffer H/Triton (1.5 mM EGTA, 50 mM β-glycerophosphate, 1 mM DTT, 0.2 mM Na3VO4, 1 mM benzamidine, 10 μg/ml Leupeptin, 10 μg/ml aprotinin, and 1% Triton X-100; pH 7.3) [14]. NMuMG whole cell extracts (250 μg per tube), followed by immunoblotting with anti-E-cadherin antibodies (1:1000 dilution). In other experiments, an aliquot of NMuMG whole cell extract was removed before immunoprecipitation of E-cadherin and used to control for differences in protein loading by immunoblotting for extracellular signal-related kinase 1/2. Both assay protocols yielded similar results.
Cell biological assays
The effect CystC or Δ14CystC on various MECs (such as NMuMG, MCF10A-CA1a, or MDA-MB-231) or NRK cell activities was determined as follows: first, cell proliferation with a [3H]thymidine incorporation assay as described [14,15]; second, cell invasion with a modified Boyden-chamber assay and Matrigel matrices as described [12,15]; and third, gene expression with pSBE-luciferase and p3TP-luciferase reporter gene assays as described [12,14].
In addition, the effect of recombinant GST, CystC, or Δ14CystC on Smad2 phosphorylation stimulated by TGF-β was determined by allowing MDA-MB-231 cells (100,000 cells per well) to adhere overnight to 24-well plates. The following morning, the cells were washed twice in ice-cold PBS and incubated in serum-free medium supplemented with 25 μg/ml recombinant GST, CystC, or Δ14CystC for 2 hours at 37°C. Afterward, the cells were stimulated with TGF-β1 (1 ng/ml) for 30 min at 37°C, and subsequently were subjected to phospho-Smad2 immunoblot analyses as described [14].
Soft agar assay
The growth of NRK cells in soft agar was performed as described [16]. In brief, duplicate cultures of control, CystC-expressing, or Δ14CystC-expressing NRK cells (10,000 cells per plate) were grown in 0.3% agar on a cushion of 0.6% agar in 35 mm plates. NRK cell growth in the absence or presence of 5 ng/ml TGF-β1 was allowed to proceed for 7 days, whereupon the number of colonies formed was quantified under a light microscope.
Results
CystC prevents TGF-β stimulation of EMT in NMuMG cells
The importance of EMT in promoting cancer progression and tumor metastasis is becoming increasingly apparent [1,2]. Although the formation and growth of early-stage tumors is normally suppressed by TGF-β, cancer progression typically enables TGF-β to stimulate the growth and metastasis of late-stage tumors, in part through its induction of EMT [2]. We recently identified the cysteine protease inhibitor CystC as a novel TβR-II antagonist [12]. Indeed, the physical association of CystC with TβR-II not only prevented TGF-β binding and, consequently, TGF-β signaling in normal and cancer cells, but it also inhibited their invasion through synthetic basement membranes [12]. Collectively, these findings led us to propose CystC as a novel chemopreventive agent capable of antagonizing the tumor-promoting activities of TGF-β in late-stage tumors, particularly the ability of TGF-β to induce EMT.
To test this hypothesis, we first overexpressed in NMuMG cells (Fig. 1a) either CystC or a CystC mutant impaired in its ability to inhibit cathepsin protease activity (namely Δ14CystC [12]) to determine whether they could antagonize TGF-β stimulation of EMT. As described previously [12], Δ14CystC protein migrates more slowly in SDS-PAGE than its wild-type counterpart, because of additional amino-terminal amino acids appended to Δ14CystC during the shuttling of its cDNA through the pSecTag vector. Unlike HT1080 and 3T3-L1 cells, the expression of Δ14CystC in NMuMG cells results in the production of two distinct Δ14CystC molecules (Fig. 1a). Although they are currently unknown, we suspect that the two recombinant Δ14CystC species observed in transduced NMuMG cells most probably result from differences in glycosylation [17,18]. Nonetheless, these stable populations of NMuMG cells were used to examine the effects of CystC or Δ14CystC on NMuMG cell EMT.
In contrast to the situation in developing tissues, inappropriate induction of EMT by TGF-β in adult tissues enhances tumorigenesis [2,19]. TGF-β stimulation of EMT is studied routinely in murine NMuMG MECs, which readily undergo EMT when treated with TGF-β [13,20,21]. Unstimulated NMuMG cells exhibited typical epithelial cuboidal morphology characterized by strong cortical and diffuse cytoplasmic actin staining (Fig. 1b). In response to TGF-β, NMuMG cells undergo transition into fibroblasts and exhibit distinct actin stress fibers emanating from focal adhesions (Fig. 1b). This NMuMG cell response to TGF-β was prevented by the overexpression of either CystC or Δ14CystC (Fig. 1b). Similarly, treatment of NMuMG cells with recombinant CystC or Δ14CystC prevented TGF-β stimulation of actin cytoskeletal reorganization at 8 hours, and significantly reduced this response at 24 hours (Fig. 1c).
TGF-β stimulation downregulates E-cadherin expression in MECs undergoing EMT [13]. We, too, find that TGF-β stimulation reduced NMuMG cell expression of E-cadherin and, more importantly, that CystC or Δ14CystC overexpression in or recombinant CystC or Δ14CystC treatment of NMuMG cells significantly attenuated the ability of TGF-β to downregulate E-cadherin expression (Fig. 2a, b). Figure 3a shows that TGF-β potently inhibited DNA synthesis in NMuMG cells. Quite surprisingly, CystC or Δ14CystC expression had little effect on the sensitivity of NMuMG cells to TGF-β-mediated growth arrest, despite the fact that both CystC molecules significantly inhibited tonic NMuMG cell proliferation (Fig. 3a). In addition, NMuMG cells undergoing EMT exhibit elevated invasion through synthetic basement membranes (Fig. 3b). We showed previously that, first, the invasion of 3T3-L1 cells through Matrigel matrices proceeds through pathways dependent on cathepsin and on TGF-β, and second, CystC abrogates both pathways, whereas Δ14CystC selectively blocks the TGF-β-dependent pathway [12]. Similarly to their actions on 3T3-L1 cells, the expression of CystC and Δ14CystC also inhibited and delineated cathepsin-dependent and TGF-β-dependent invasion in NMuMG cells (Fig. 3b). Collectively, these findings show that CystC and Δ14CystC do indeed effectively inhibit mammary cell EMT and invasion stimulated by TGF-β. Moreover, these findings suggest that when interacting with TβR-II, CystC alters the TGF-β receptor signaling complex in a manner that selectively inhibits some TGF-β functions (such as EMT) while leaving others (such as growth arrest) intact.
CystC inhibits TGF-β signaling in human breast cancer cells
Because NMuMG cells are non-tumorigenic and an incomplete model of breast cancer progression, we also determined whether CystC antagonizes TGF-β signaling in human breast cancer cells. To do so, we first stably expressed CystC in human MCF10A-CA1a breast cancer cells, whose metastatic activity is enhanced by TGF-β [9]. Given this fact, we hypothesized that TGF-β would stimulate the growth of MCF10A-CA1a cells in soft agar. In stark contrast, we found TGF-β stimulation to suppress the growth of MCF10A-CA1a cells in soft agar, a response that was unaffected by CystC expression (data not shown). However, CystC expression did significantly inhibit the anchorage-independent growth of MCF10A-CA1a cells (Fig. 4a), as well as significantly antagonized reporter gene expression stimulated by TGF-β (Fig. 4b).
We also investigated whether the inhibitory effects of CystC on TGF-β signaling were unique to MCF10A-CA1a cells or were instead a more generalized inhibitory mechanism of TGF-β signaling in human breast cancers. As shown in Fig. 5a, overexpression of CystC or Δ14CystC in human MDA-MB-231 breast cancer cells significantly reduced reporter gene expression stimulated by moderate TGF-β concentrations (such as 0.5 ng/ml), an effect that was overcome by stimulation with supraphysiological TGF-β concentrations (such as 5 ng/ml). Accordingly, treatment of MDA-MB-231 cells with recombinant CystC or Δ14CystC significantly antagonized reporter gene expression induced by TGF-β (Fig. 5b) and inhibited its ability to stimulate Smad2 phosphorylation (Fig. 5c). Taken together, these findings establish CystC and its derivative Δ14CystC as novel TGF-β antagonists in human breast cancer cells.
CystC prevents stimulation of morphological transformation in NRK cells by TGF-β
The loss of cell polarity and the ability of cancer cells to grow autonomously in an anchorage-independent manner are a hallmark of cancer [1,2]. TGF-β was originally described as a secreted factor that stimulates morphological transformation in rat NRK-49 kidney cells, leading to their acquisition of anchorage-independent growth in soft agar [22,23]. Thus, in addition to promoting EMT, TGF-β also enhances cancer progression by stimulating morphological transformation and anchorage-independent cell growth. Because CystC and Δ14CystC both eliminated EMT stimulated by TGF-β, we hypothesized that these TβR-II antagonists [12] would similarly inhibit NRK cell morphological transformation stimulated by TGF-β. We tested this hypothesis by infecting NRK cells with bicistronic retrovirus encoding either CystC or Δ14CystC (Fig. 6a) to determine their effects on NRK anchorage-independent growth stimulated by TGF-β. As expected, treatment with TGF-β enabled NRK cells to grow in an anchorage-independent manner when cultured in soft agar (Fig. 6b). Similar to their inhibitory activities in NMuMG and breast cancer cells, retroviral-mediated expression of CystC or Δ14CystC in NRK cells prevented their morphological transformation stimulated by TGF-β (Fig. 6b), as well as inhibiting Smad2/3-mediated reporter gene expression (Fig. 6c) and cell invasion induced by TGF-β (Fig. 6d). Thus, in addition to preventing TGF-β stimulation of EMT and inhibiting TGF-β signaling in breast cancer cells, CystC and Δ14CystC also antagonize morphological transformation and anchorage-independent growth stimulated by TGF-β.
Discussion
TGF-β is widely expressed during development to regulate the interactions between epithelial and mesenchymal cells, particularly those in the lung, kidney, and mammary gland. Inappropriate reactivation of EMT during tumorigenesis is now recognized as an important process necessary for the acquisition of invasive and metastatic phenotypes by tumors [1,2]. By cooperating with oncogenes and growth factors, TGF-β potently induces EMT and serves to stabilize this transition by means of autocrine signaling. Moreover, these events seem to underlie the oncogenic activities of TGF-β and its ability to promote cancer progression [20,24-26]. A comprehensive understanding of how TGF-β both suppresses and promotes tumorigenesis remains an unknown and fundamental question that directly affects our ability to effectively target the TGF-β signaling system during the treatment of human malignancies, particularly those of the breast. Indeed, solving this paradox remains the most important aspect of the biological and pathological actions of this multifunctional cytokine.
Despite recent advances in understanding the molecular mechanisms underlying EMT, the question of how to prevent this process effectively in response to TGF-β remains unanswered. We recently discovered that CystC antagonizes TGF-β signaling in normal and cancer cells by interacting physically with TβR-II, thereby preventing TGF-β binding [12]. Importantly, we demonstrated the effectiveness of CystC in inhibiting the invasion of cancer cells and the TGF-β-stimulated invasion of fibroblasts [12]. Because EMT is necessary for the acquisition of invasive and metastatic phenotypes by cancer cells, and because CystC inhibited TGF-β-stimulated invasion, we proposed CystC as a potential antagonist of EMT stimulated by TGF-β.
Accordingly, in this study we show that CystC and Δ14CystC both prevent EMT and its associated increase in MEC motility (Figs 1, 2, 3), as well as antagonizing TGF-β signaling in human breast cancer cells (Figs 4 and 5). We further show for the first time that these CystC molecules inhibit TGF-β signaling in NRK fibroblasts, thus preventing their morphological transformation and invasion through synthetic basement membranes (Fig. 6). Although our understanding of CystC function in regulating TGF-β signaling is in its infancy, our findings suggest that this protease inhibitor might provide an innovative model for the development of novel TβR-II antagonists designed to combat the stimulation of tumor progression and EMT by TGF-β. We further propose that the chemopreventive effectiveness of CystC will be potentiated by its inhibition of cathepsin B-mediated invasion and metastasis [27-30] and its inhibition of the cathepsin B-mediated activation of latent TGF-β [31-33], which co-localizes with cathepsin B to the invading face of malignant tumors [34-37]. Moreover, CystC-mediated cathepsin B inactivation will reduce the activity of the urokinase plasminogen system, which enhances tumor cell extracellular matrix degradation, as well as growth factor and latent TGF-β activation [38]. Cumulatively, the chemopreventive activities of CystC will antagonize cancer cell responses to TGF-β by inhibiting TGF-β binding [12] and by reducing TGF-β bioavailability within tumor microenvironments, thereby alleviating the stimulation of EMT and tumor metastasis in late-stage tumors by TGF-β.
Molecular dissection of TGF-β signaling systems necessary for its induction of EMT has clearly established a role for Smad2/3 in mediating EMT, particularly when coupled with signals emanating from oncogenic Ras [13,39,40]. However, Smad2/3-independent signaling has also been implicated in TGF-β stimulation of EMT. For instance, TGF-β stimulates EMT in cancers of the breast and other tissues by activating phosphoinositide 3-kinase, Akt, RhoA, p160(ROCK), and p38 mitogen-activated protein (MAP) kinase [40-44]. In addition, EMT in TGF-β-treated MECs is abrogated by measures that inhibit β1 integrin activity [42], thus establishing the necessity of β1 integrin expression for EMT stimulated by TGF-β. Finally, by repressing Id2 and Id3 expression [45], inducing Snail and SIP1 expression [46], and stimulating nuclear factor-κB activity [47], TGF-β regulates transcription factor activity operant in mediating the transition from epithelial to mesenchymal cell markers. We show that CystC inhibits the stimulation of Smad2 phosphorylation by TGF-β and the subsequent induction of reporter gene expression in normal and cancer MECs. Thus, reduced Smad2/3 signaling mediated by CystC probably underlies part of its ability to inhibit EMT stimulated by TGF-β. Future studies need to address the role of CystC in regulating Smad2/3-independent signaling stimulated by TGF-β, as well as determining their contribution in preventing the oncogenic activities of TGF-β in human breast cancer cells.
Finally, we were quite surprised to find that CystC, despite its ability to inhibit Smad2/3 signaling, failed to alter the growth-suppressing activities of TGF-β. Although the molecular mechanism(s) underlying this unexpected CystC activity remains to be elucidated, our findings suggest that CystC does not function to abrogate all TGF-β signaling, but may instead specifically alter and/or modulate certain aspects of TGF-β signaling when complexed to TβR-II. In support of this supposition, we find that activation of MAP kinases and Akt by TGF-β requires high cytokine concentrations, whereas that of Smad2/3 requires markedly lower cytokine concentrations (more than 10-fold lower; data not shown). Mechanistically, we propose that maximal stimulation of Smad2/3 by TGF-β requires minimal receptor occupancy (that is, large receptor reserves), whereas maximal stimulation of MAP kinases and AKT requires maximal receptor occupancy (that is, no receptor reserves). Thus, manipulations designed to antagonize the binding of TGF-β to its receptors might elicit disproportionate inhibition of TGF-β signaling systems, resulting in greater inhibition of Smad2/3-independent versus Smad2/3-dependent pathways. Future studies need to address this important question and determine which TGF-β signaling systems are preferentially inhibited by the formation of CystC–TβR-II complexes in normal and cancerous MECs.
Conclusion
Here we have shown the effectiveness of CystC in inhibiting MEC EMT and fibroblast morphological transformation stimulated by TGF-β, and in antagonizing TGF-β signaling in human breast cancer cells. The ability of CystC to inhibit TGF-β signaling in MECs occurs independently of its inactivation of cathepsin protease activity, presumably by means of CystC–TβR-II complex formation and the prevention of TGF-β binding [12]. We suggest that CystC or its peptide mimetics ultimately hold the potential to improve the therapeutic response of human malignancies regulated by TGF-β, particularly cancers of the breast. Experiments designed to test this hypothesis in animal models of TGF-β oncogenicity and to identify CystC determinants necessary in mediating TGF-β antagonism are currently underway.
Abbreviations
CystC = cystatin C; EMT = epithelial–mesenchymal transition; GFP = green fluorescent protein; GST = glutathione S-transferase; MAP = mitogen-activated protein; MEC = mammary epithelial cell; NRK = normal rat kidney; TGF-β = transforming growth factor-β; TβR-II = TGF-β type II receptor.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
JPS performed the EMT and morphological transformation studies, as well as acquired and analyzed the data. JRN performed functional analyses of human breast cancer cells, which were generated in part by BJS. WPS conceived of the study, participated in its design, coordination, and data analysis, and drafted the manuscript. All authors read and approved the final manuscript.
Acknowledgements
TGF-β1 was generously provided by R&D Systems Inc. Members of the Schiemann Laboratory are thanked for critical reading of the manuscript. We also thank William Townend, Shirley Sobus, and Joshua Loomis for expertise and help provided on studies performed in the Cytometry Core Facility at the National Jewish Medical and Research Center. Support was provided in part by the National Institutes of Health (CA095519 and CA114039), the Elsa U. Pardee Foundation, and the Cancer League of Colorado to WPS.
Figures and Tables
Figure 1 CystC and Δ14CystC inhibit actin cytoskeletal rearrangements stimulated by TGF-β in NMuMG cells. (a) NMuMG cells were infected with ecotropic retrovirus encoding either green fluorescent protein (GFP; control), CystC, or Δ14CystC, and subsequently were isolated by fluorescence-activated cell sorting for GFP expression to yield stable polyclonal populations of control, CystC-expressing, and Δ14CystC-expressing cells as indicated (upper panel). The expression and secretion of recombinant CystC proteins by infected NMuMG cells were monitored by immunoblotting conditioned medium with anti-CystC antibodies (lower panel). (b) Control, CystC-expressing, and Δ14CystC-expressing NMuMG cells were incubated in the absence or presence of transforming growth factor-β1 (TGF-β1; 5 ng/ml) for 8 hours, whereupon alterations in actin cytoskeletal architecture were revealed by direct rhodamine–phalloidin immunofluorescence. Shown are representative images from a single experiment that was repeated twice with identical results. (c) NMuMG cells were stimulated with TGF-β1 (5 ng/ml) for 0 to 24 hours in the presence of either glutathione S-transferase (GST), GST–CystC, or GST–Δ14CystC (each at 10 μg/ml) as indicated. Afterward, altered actin cytoskeletal architecture was revealed by direct rhodamine–phalloidin immunofluorescence. Shown are representative images from a single experiment that was repeated once with identical results.
Figure 2 CystC and Δ14CystC antagonize E-cadherin downregulation stimulated by TGF-β in NMuMG cells. (a) Control, CystC-expressing, and Δ14CystC-expressing NMuMG cells were incubated in the absence or presence of transforming growth factor-β1 (TGF-β1; 5 ng/ml) for 36 hours, whereupon alterations in E-cadherin expression was monitored by immunoprecipitation and subsequent immunoblotting with anti-E-cadherin antibodies. The representative immunoblot depicts the downregulation of E-cadherin expression induced by TGF-β in NMuMG cells. The lower panel shows the alterations in E-cadherin expression (means ± SEM) induced by TGF-β relative to their untreated counterparts observed in three independent experiments. TGF-β significantly downregulated E-cadherin expression in NMuMG cells (*P < 0.05; Student's t-test). (b) NMuMG cells were stimulated with TGF-β1 (5 ng/ml) for 36 hours in the presence of glutathione S-transferase (GST) fusion proteins (10 μg/ml) as indicated. E-cadherin expression was monitored by indirect immunofluorescence with anti-E-cadherin antibodies. Shown are representative images from a single experiment that was repeated once with identical results.
Figure 3 CystC and Δ14CystC inhibit NMuMG cell proliferation and invasion. (a) Control, CystC-expressing, or Δ14CystC-expressing NMuMG cells were incubated with increasing concentrations of transforming growth factor-β1 (TGF-β1; from 0 to 5 ng/ml) for 48 hours. Cellular DNA was radiolabeled with [3H]thymidine and quantified by scintillation counting. Data are means ± SEM for two independent experiments presented as the percentage [3H]thymidine incorporation normalized to untreated control cells. CystC and Δ14CystC both significantly decreased DNA synthesis in resting NMuMG cells (*P < 0.05; Student's t-test). (b) Control, CystC-expressing, and Δ14CystC-expressing NMuMG cells were allowed to invade through Matrigel matrices in the absence or presence of TGF-β1 (5 ng/ml) for 48 hours. Values are means ± SEM for three independent experiments presented as the percentage invasion relative to GFP-expressing NMuMG cells. TGF-β1 significantly enhanced NMuMG cell invasion (*P < 0.05; Student's t-test), a response that was inhibited significantly by CystC and Δ14CystC expression (#P < 0.05; Student's t-test). CystC expression also significantly inhibited tonic NMuMG cell invasion (*P < 0.05; Student's t-test).
Figure 4 CystC antagonizes TGF-β signaling in human MCF10A-CA1a breast cancer cells. (a) Control and CystC-expressing MCF10A-CA1a cells were cultured in soft agar for 14 days, whereupon MCF10A-CA1a colony formation was quantified by light microscopy. Values are colony formation per microscope field (means ± SEM) observed in two independent experiments. CystC expression significantly reduced anchorage-independent growth of MCF10A-CA1a cells (*P < 0.05; Student's t-test). (b) Control and CystC-expressing MCF10A-CA1a cells were transiently transfected with p3TP-luciferase and pCMV-β-Gal cDNAs, and were subsequently stimulated with transforming growth factor-β1 (TGF-β1; 0.5 ng/ml) for 24 hours. Afterward, luciferase and β-Gal activities contained in detergent-solubilized cell extracts were measured. Values are luciferase activities (means ± SEM) observed in two independent experiments normalized to the maximal reporter gene expression stimulated by TGF-β in cells expressing green fluorescent protein. CystC expression significantly inhibited luciferase expression induced by TGF-β (*P < 0.05; Student's t-test).
Figure 5 CystC and Δ14CystC inhibit TGF-β signaling in human MDA-MB-231 breast cancer cells. (a) Control, CystC-expressing, or Δ14CystC-expressing MDA-MB-231 cells were transiently transfected with p3TP-luciferase and pCMV-β-Gal cDNAs, and were subsequently stimulated with increasing concentrations of transforming growth factor-β1 (TGF-β1; from 0 to 5 ng/ml) for 24 hours. Afterward, luciferase and β-Gal activities contained in detergent-solubilized cell extracts were measured. Values are luciferase activities (means ± SEM) observed in two independent experiments normalized to maximal reporter gene expression in TGF-β-stimulated cells expressing green fluorescent protein. CystC and Δ14CystC both significantly inhibit reporter gene expression stimulated by TGF-β (*P < 0.05; Student's t-test). (b) MDA-MB-231 cells were transiently transfected with p3TP-luciferase and pCMV-β-Gal cDNAs. Afterward, the transfectants were treated with 25 μg/ml recombinant glutathione S-transferase (GST; G), CystC (C), or Δ14CystC (D) and immediately stimulated with TGF-β1 (1 ng/ml) for 24 hours before determination of the luciferase and β-Gal activities contained in detergent-solubilized cell extracts. Values are luciferase activities (means ± SEM) observed in four independent experiments normalized to maximal reporter gene expression stimulated by TGF-β in GST-treated cells. (c) MDA-MB-231 cells were treated with 25 μg/ml recombinant GST (G), CystC (C), or Δ14CystC (D) for 2 hours before their stimulation with TGF-β (1 ng/ml) for 30 min. Afterward, the activation status of Smad2 was determined by immunoblot analysis with phospho-specific Smad2 antibodies. Differences in protein loading were monitored by reprobing stripped membranes with antibodies against extracellular signal-related kinase 1. Data are from a representative experiment that was repeated once with similar results.
Figure 6 CystC and Δ14CystC inhibit NRK cell morphological transformation and anchorage-independent growth stimulated by TGF-β. (a) Normal rat kidney (NRK) fibroblasts were infected with control (namely pMSCV-IRES–GFP, where GFP stands for green fluorescent protein), CystC, or Δ14CystC retroviral supernatants as in Fig. 1a, and the resulting infected cells were isolated by GFP fluorescence on a MoFlo cell sorter 48 hours later. Shown in the upper panel are the GFP expression profiles of the resulting polyclonal populations. Also shown, in the lower panel, are recombinant CystC (CC) and Δ14CystC (Δ14) proteins present in conditioned medium of individual NRK populations and revealed by anti-CystC immunoblotting. (b) Control, CystC-expressing, and Δ14CystC-expressing NRK cells were cultured in soft agar in the absence or presence of transforming growth factor-β1 (TGF-β1; 5 ng/ml) for 7 days, whereupon NRK colony formation was quantified by light microscopy (upper panel). Middle panel, magnification of boxed regions. Lower panel, colony formation per microscope field (means ± SEM) observed in five independent experiments. (c) Control, CystC-expressing, or Δ14CystC-expressing NRK cells were transiently transfected with pSBE-luciferase and pCMV-β-Gal cDNAs, and were subsequently stimulated with TGF-β1 (5 ng/ml) for 24 hours. Afterward, luciferase and β-Gal activities contained in detergent-solubilized cell extracts were measured. Values are luciferase activities (means ± SEM) observed in three independent experiments normalized to maximal reporter gene expression induced by TGF-β in GFP-expressing cells. CystC expression significantly inhibited luciferase expression induced by TGF-β (*P < 0.05; Student's t-test). (d) Control, CystC-expressing, and Δ14CystC-expressing NRK cells were allowed to invade through Matrigel matrices in the absence or presence of TGF-β1 (5 ng/ml) for 48 hours. Values are means ± SEM for three independent experiments presented as the percentage invasion relative to GFP-expressing NRK cells. TGF-β1 significantly enhanced NRK cell invasion (*P < 0.05; Student's t-test). This TGF-β response was inhibited significantly by CystC and Δ14CystC expression (#P < 0.05; Student's t-test), whereas tonic NRK cell invasion was significantly inhibited only by CystC expression (*P < 0.05; Student's t-test). TGF-β1 significantly enhanced NMuMG cell invasion (*P < 0.05; Student's t-test), a response that was inhibited significantly by CystC and Δ14CystC expression (#P < 0.05; Student's t-test).
==== Refs
Thiery JP Epithelial–mesenchymal transitions in tumour progression Nat Rev Cancer 2002 2 442 454 12189386 10.1038/nrc822
Grunert S Jechlinger M Beug H Diverse cellular and molecular mechanisms contribute to epithelial plasticity and metastasis Nat Rev Mol Cell Biol 2003 4 657 665 12923528 10.1038/nrm1175
Yoshida BA Sokoloff MM Welch DR Rinker-Schaeffer CW Metastasis-suppressor genes: a review and perspective on an emerging field J Natl Cancer Inst 2000 92 1717 1730 11058615 10.1093/jnci/92.21.1717
Fidler IJ The organ microenvironment and cancer metastasis Differentiation 2002 70 498 505 12492492 10.1046/j.1432-0436.2002.700904.x
Blobe GC Schiemann WP Lodish HF Role of transforming growth factor β in human disease N Engl J Med 2000 342 1350 1358 10793168 10.1056/NEJM200005043421807
Siegel PM Massague J Cytostatic and apoptotic actions of TGF-β in homeostasis and cancer Nat Rev Cancer 2003 3 807 821 14557817 10.1038/nrc1208
Wakefield LM Roberts AB TGF-β signaling: positive and negative effects on tumorigenesis Curr Opin Genet Dev 2002 12 22 29 11790550 10.1016/S0959-437X(01)00259-3
Cui W Fowlis DJ Bryson S Duffie E Ireland H Balmain A Akhurst RJ TGFβ1 inhibits the formation of benign skin tumors, but enhances progression to invasive spindle carcinomas in transgenic mice Cell 1996 86 531 542 8752208 10.1016/S0092-8674(00)80127-0
Tang B Vu M Booker T Santner SJ Miller FR Anver MR Wakefield LM TGF-β switches from tumor suppressor to prometastatic factor in a model of breast cancer progression J Clin Invest 2003 112 1116 1124 14523048 10.1172/JCI200318899
Vray B Hartmann S Hoebeke J Immunomodulatory properties of cystatins Cell Mol Life Sci 2002 59 1503 1512 12440772 10.1007/s00018-002-8525-4
Brown WM Dziegielewska KM Friends and relations of the cystatin superfamily – new members and their evolution Protein Sci 1997 6 5 12 9007972
Sokol JP Schiemann WP Cystatin C antagonizes transforming growth factor-β signaling in normal and cancer cells Mol Cancer Res 2004 2 183 195 15037657
Piek E Moustakas A Kurisaki A Heldin CH ten Dijke P TGF-β type I receptor/ALK-5 and Smad proteins mediate epithelial to mesenchymal transdifferentiation in NMuMG breast epithelial cells J Cell Sci 1999 112 4557 4568 10574705
Schiemann BJ Neil JR Schiemann WP SPARC inhibits epithelial cell proliferation in part through stimulation of the TGF-β-signaling system Mol Biol Cell 2003 14 3977 3988 14517312 10.1091/mbc.E03-01-0001
Schiemann WP Blobe GC Kalume DE Pandey A Lodish HF Context-specific effects of fibulin-5 (DANCE/ EVEC) on cell proliferation, motility, and invasion: fibulin-5 is induced by TGF-β and affects protein kinase cascades J Biol Chem 2002 277 27367 27377 12021267 10.1074/jbc.M200148200
Moustakas A Stournaras C Regulation of actin organisation by TGF-β in H-ras-transformed fibroblasts J Cell Sci 1999 112 1169 1179 10085252
Cole T Dickson PW Esnard F Averill S Risbridger GP Gauthier F Schreiber G The cDNA structure and expression analysis of the genes for the cysteine proteinase inhibitor cystatin C and for beta 2-microglobulin in rat brain Eur J Biochem 1989 186 35 42 2689174 10.1111/j.1432-1033.1989.tb15174.x
Esnard F Esnard A Faucher D Capony JP Derancourt J Brillard M Gauthier F Rat cystatin C: the complete amino acid sequence reveals a site for N-glycosylation Biol Chem Hoppe Seyler 1990 371 Suppl 161 166 2400577
Jechlinger M Grunert S Beug H Mechanisms in epithelial plasticity and metastasis: insights from 3D cultures and expression profiling J Mammary Gland Biol Neoplasia 2002 7 415 432 12882526 10.1023/A:1024090116451
Miettinen PJ Ebner R Lopez AR Derynck R TGF-β induced transdifferentiation of mammary epithelial cells to mesenchymal cells: involvement of type I receptors J Cell Biol 1994 127 2021 2036 7806579 10.1083/jcb.127.6.2021
Gotzmann J Mikula M Eger A Schulte-Hermann R Foisner R Beug H Mikulits W Molecular aspects of epithelial cell plasticity: implications for local tumor invasion and metastasis Mutat Res 2004 566 9 20 14706509 10.1016/S1383-5742(03)00033-4
Roberts AB Anzano MA Lamb LC Smith JM Sporn MB New class of transforming growth factors potentiated by epidermal growth factor: isolation from non-neoplastic tissues Proc Natl Acad Sci USA 1981 78 5339 5343 6975480
Assoian RK Komoriya A Meyers CA Miller DM Sporn MB Transforming growth factor-β in human platelets. Identification of a major storage site, purification, and characterization J Biol Chem 1983 258 7155 7160 6602130
Oft M Heider KH Beug H TGFβ signaling is necessary for carcinoma cell invasiveness and metastasis Curr Biol 1998 8 1243 1252 9822576 10.1016/S0960-9822(07)00533-7
Oft M Peli J Rudaz C Schwarz H Beug H Reichmann E TGF-β1 and Ha-Ras collaborate in modulating the phenotypic plasticity and invasiveness of epithelial tumor cells Genes Dev 1996 10 2462 2477 8843198
Portella G Cumming SA Liddell J Cui W Ireland H Akhurst RJ Balmain A Transforming growth factor β is essential for spindle cell conversion of mouse skin carcinoma in vivo: implications for tumor invasion Cell Growth Differ 1998 9 393 404 9607560
Turk V Turk B Guncar G Turk D Kos J Lysosomal cathepsins: structure, role in antigen processing and presentation, and cancer Adv Enzyme Regul 2002 42 285 303 12123721 10.1016/S0065-2571(01)00034-6
Yan S Sloane BF Molecular regulation of human cathepsin B: implication in pathologies Biol Chem 2003 384 845 854 12887051 10.1515/BC.2003.095
Turk B Turk D Turk V Lysosomal cysteine proteases: more than scavengers Biochim Biophys Acta 2000 1477 98 111 10708852
Roshy S Sloane BF Moin K Pericellular cathepsin B and malignant progression Cancer Metastasis Rev 2003 22 271 286 12785001 10.1023/A:1023007717757
Somanna A Mundodi V Gedamu L Functional analysis of cathepsin B-like cysteine proteases from Leishmania donovani complex. Evidence for the activation of latent transforming growth factor β J Biol Chem 2002 277 25305 25312 12000761 10.1074/jbc.M203034200
Guo M Mathieu PA Linebaugh B Sloane BF Reiners JJ Jr Phorbol ester activation of a proteolytic cascade capable of activating latent transforming growth factor-β L a process initiated by the exocytosis of cathepsin B J Biol Chem 2002 277 14829 14837 11815600 10.1074/jbc.M108180200
Gantt KR Schultz-Cherry S Rodriguez N Jeronimo SM Nascimento ET Goldman TL Recker TJ Miller MA Wilson ME Activation of TGF-β by Leishmania chagasi : importance for parasite survival in macrophages J Immunol 2003 170 2613 2620 12594289
Wakefield LM Piek E Bottinger EP TGF-β signaling in mammary gland development and tumorigenesis J Mammary Gland Biol Neoplasia 2001 6 67 82 11467453 10.1023/A:1009568532177
Calkins CC Sameni M Koblinski J Sloane BF Moin K Differential localization of cysteine protease inhibitors and a target cysteine protease, cathepsin B, by immuno-confocal microscopy J Histochem Cytochem 1998 46 745 751 9603786
Sinha AA Wilson MJ Gleason DF Reddy PK Sameni M Sloane BF Immunohistochemical localization of cathepsin B in neoplastic human prostate Prostate 1995 26 171 178 7716081
Sameni M Moin K Sloane BF Imaging proteolysis by living human breast cancer cells Neoplasia 2000 2 496 504 11228542 10.1038/sj.neo.7900116
Choong PF Nadesapillai AP Urokinase plasminogen activator system: a multifunctional role in tumor progression and metastasis Clin Orthop Relat Res 2003 S46 S58 14600592 10.1097/01.blo0000093845.72468.bd
Oft M Akhurst RJ Balmain A Metastasis is driven by sequential elevation of H-ras and Smad2 levels Nat Cell Biol 2002 4 415 Suppl 487 494 12105419 10.1038/ncb807
Janda E Lehmann K Killisch I Jechlinger M Herzig M Downward J Beug H Grunert S Ras and TGFβ cooperatively regulate epithelial cell plasticity and metastasis: dissection of Ras signaling pathways J Cell Biol 2002 156 299 313 11790801 10.1083/jcb.200109037
Bhowmick NA Ghiassi M Bakin A Aakre M Lundquist CA Engel ME Arteaga CL Moses HL Transforming growth factor-β1 mediates epithelial to mesenchymal transdifferentiation through a RhoA-dependent mechanism Mol Biol Cell 2001 12 27 36 11160820
Bhowmick NA Zent R Ghiassi M McDonnell M Moses HL Integrin β1 signaling is necessary for transforming growth factor-β activation of p38MAPK and epithelial plasticity J Biol Chem 2001 276 46707 46713 11590169 10.1074/jbc.M106176200
Bakin AV Tomlinson AK Bhowmick NA Moses HL Arteaga CL Phosphatidylinositol 3-kinase function is required for transforming growth factor β-mediated epithelial to mesenchymal transition and cell migration J Biol Chem 2000 275 36803 36810 10969078 10.1074/jbc.M005912200
Yu L Hebert MC Zhang YE TGF-β receptor-activated p38 MAP kinase mediates Smad-independent TGF-β responses EMBO J 2002 21 3749 3759 12110587 10.1093/emboj/cdf366
Kowanetz M Valcourt U Bergstrom R Heldin CH Moustakas A Id2 and Id3 define the potency of cell proliferation and differentiation responses to transforming growth factor β and bone morphogenetic protein Mol Cell Biol 2004 24 4241 4254 15121845 10.1128/MCB.24.10.4241-4254.2004
Kang Y Massague J Epithelial–mesenchymal transitions: twist in development and metastasis Cell 2004 118 277 279 15294153 10.1016/j.cell.2004.07.011
Huber MA Azoitei N Baumann B Grunert S Sommer A Pehamberger H Kraut N Beug H Wirth T NF-κB is essential for epithelial–mesenchymal transition and metastasis in a model of breast cancer progression J Clin Invest 2004 114 569 581 15314694 10.1172/JCI200421358
|
16168131
|
PMC1242164
|
CC BY
|
2021-01-04 16:04:34
|
no
|
Breast Cancer Res. 2005 Aug 23; 7(5):R844-R853
|
utf-8
|
Breast Cancer Res
| 2,005 |
10.1186/bcr1312
|
oa_comm
|
==== Front
Breast Cancer ResBreast Cancer Research1465-54111465-542XBioMed Central London bcr13131616813310.1186/bcr1313Research ArticleRelationship between the expression of cyclooxygenase 2 and MDR1/P-glycoprotein in invasive breast cancers and their prognostic significance Surowiak Pawel 123Materna Verena 1Matkowski Rafal 4Szczuraszek Katarzyna 2Kornafel Jan 4Wojnar Andrzej 3Pudelko Marek 3Dietel Manfred 1Denkert Carsten 1Zabel Maciej 25Lage Hermann [email protected] Institute of Pathology, Charité Campus Mitte, Berlin, Germany2 Chair and Department of Histology and Embryology, University School of Medicine, Wrocław, Poland3 Lower Silesian Centre of Oncology, Wrocław, Poland4 Chair and Department of Oncology, University School of Medicine, Wrocław, Poland5 Chair and Department of Histology and Embryology, University School of Medicine, Poznań, Poland2005 25 8 2005 7 5 R862 R870 8 4 2005 26 4 2005 9 6 2005 2 8 2005 Copyright © 2005 Surowiak et al.; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Introduction
Recent reports suggest that expression of the cyclooxygenase 2 (COX-2) enzyme may up-regulate expression of MDR1/P-glycoprotein (MDR1/P-gp), an exponent of resistance to cytostatic drugs. The present study aimed at examining the relationship between the expression of COX-2 and of MDR1/P-gp in a group of breast cancer cases.
Methods
Immunohistochemical reactions were performed using monoclonal antibodies against COX-2 and MDR1/P-gp on samples originating from 104 cases of primary invasive breast cancer.
Results
COX-2-positive cases were shown to demonstrate higher expression of MDR1/P-gp (P < 0.0001). The studies also demonstrate that COX-2 expression was typical for cases of a higher grade (P = 0.01), a shorter overall survival time (P < 0.0001) and a shorter progression-free time (P < 0.0001). In the case of MDR1/P-gp, its higher expression characterised cases of a higher grade (P < 0001), with lymph node involvement (P < 0001), and shorter overall survival (P < 0.0001) and progression-free time (P < 0.0001).
Conclusion
Our studies confirmed the unfavourable prognostic significance of COX-2 and MDR1/P-gp. We also document a relationship between COX-2 and MDR1/P-gp, which suggests that COX-2 inhibitors should be investigated in trials as a treatment supplementary to chemotherapy of breast cancers.
==== Body
Introduction
Breast cancer is the most common malignant tumour of females in the western world [1]. The incidence of breast cancer remains high, and its clinical courses are highly variable. It is of general importance to predict the biology of the tumour and, thus, the course of the disease in the individual patient to ensure adequate therapy and patient surveillance [2]. The principal therapeutic approach in breast cancer involves surgery. In advanced cases supplementary therapy is needed, involving pharmacotherapy and/or radiotherapy. Among the pharmacological means, tamoxifen used to be applied most frequently, as well as various chemotherapeutic regimes, including CMF (cyclophosphamide, methothrexate and 5-fluorouracil), anthracyclines and paclitaxel [3,4]. The main reason for therapeutic failure in cases of invasive breast cancers involves resistance to anti-estrogenic treatment and to chemotherapy [5,6]. Identification of the factors that characterise the resistant cases would permit immediate treatment of the patients with alternative therapeutic approaches. These factors could also provide potential targets for studies on novel therapeutic procedures.
Cycloxygenases (COXs) comprise a group of enzymes that participate in the conversion of arachidonic acid to prostaglandins [7]. COX-2 has been characterised as an unfavourable prognostic factor in numerous solid tumours [8-10]. We demonstrated previously in breast cancer patients that expression of COX-2 represents an independent, unfavourable prognostic factor [11]. Numerous in vivo and in vitro studies indicate that COX-2 inhibitors (coxibs) enhance the efficacy of various anticancer therapy methods [7]. The effect of coxibs on the biology of the tumour has been explained by induction of apoptosis, inhibition of angiogenesis and by a decreased invasive potential of tumour cells [7]. COX-2 has also been shown to up-regulate expression of aromatase [12,13]. In cases of hormone-dependent tumours, such as breast cancer, coxibs might slow down development of the neoplastic disease by decreasing aromatase expression and, therefore, decreasing estrogen secretion. The in vitro studies have demonstrated also that COX-2 up-regulates expression of MDR1/P-glycoprotein (MDR1/P-gp) [14], the energy-dependent pump that participates in the phenomenon of multidrug resistance (MDR) [5]. MDR1/P-gp efficiently removes drugs and many commonly used pharmaceuticals from the lipid bilayer. Confirmation of the relationship between COX-2 and MDR1/P-gp in a clinical material may open novel perspectives in the therapy of tumours. Coxibs could be employed as a chemotherapy-supporting treatment, aimed at the inhibition or prevention of the development of the MDR phenomenon.
The present study aimed to examine the relationship between the expression of COX-2 and of MDR1/P-gp in primary invasive breast cancers as well as the definition of their prognostic and predictive values.
Materials and methods
Patients
Immunohistochemical analysis was performed retrospectively on tissue samples that were taken for routine diagnostic purposes. The cases were selected based on availability of tissue and were not stratified for known preoperative or pathological prognostic factors. The study was approved by an Institutional Review Board (University School of Medicine, Wrocław, Poland) and the patients gave their informed consent before their inclusion into the study. A total of 104 patients with primary invasive breast cancer who were diagnosed in the years 1993 to 1994 in the Lower Silesian Centre of Oncology in Wrocław, Poland, qualified for the studies. All the patients were subjected to mastectomy and, subsequently treated with radiotherapy and/or chemotherapy and/or hormonotherapy (Table 1). Compliance was monitored by the doctors in charge. The patients were monitored by periodic medical check-ups and ultrasonographic and radiological examinations. During the follow-up period, 23 patients (22%) had recurrent disease and 25 patients (24%) died of the disease. The mean (median) progression-free survival time was 76 months (range 8 to 103 months), while the mean (median) overall survival time was 81 months (range 8 to 103 months).
Fragments sampled from studied tumours were fixed in 10% buffered formaline and embedded in paraffin. In every case, hematoxylin and eosin stained preparations were subjected to histopathological evaluation by two pathologists. The stage of the tumours was assessed according to the TNM classification system [15]. Tumour grade was estimated according to Bloom-Richardson and the modification of Elston and Ellis [16] (Table 1).
Immunohistochemistry
Freshly cut sections (4 μm) of the formalin-fixed, paraffin embedded tissue were mounted on Superfrost slides (Menzel Gläser, Braunschweig, Germany), dewaxed with xylene, and gradually hydrated. Activity of endogenous peroxidase was blocked by 5 minute exposure to 3% H2O2. All the studied sections were boiled in Antigen Retrieval Solution (DakoCytomation, Glostrup, Denmark), in the case of COX-2 for 10 minutes and in the case of MDR1/P-gp for 15 minutes. Immunohistochemical reactions were performed using monoclonal mouse antibodies against COX-2 (Cayman Chemical Company, Ann Arbor, MI, USA) at a dilution of 1:2000, monoclonal mouse antibodies (clone C219) against MDR1/P-gp (Alexis Biochemicals, Grünberg, Germany) at a dilution of 1:100 and monoclonal mouse antibodies (clone JSB-1) against MDR1/P-gp (Roche Diagnostics, Mannheim, Germany) at a dilution of 1:100. The antibodies were diluted in Antibody Diluent with background reducing component (DakoCytomation). Tested sections were incubated with antibodies for 1 h at room temperature. Subsequent incubations involved biotinylated antibodies (15 minutes, room temperature) and streptavidin-biotinylated peroxidase complex (15 minutes, room temperature) using a LSAB+ HRP system DakoCytomation). NovaRed (Vector Laboratories, Peterborough, UK) was used as a chromogen (10 minutes, room temperature). All the sections were counterstained with Meyer's hematoxylin.
Evaluation of reaction intensity
The intensity of immunohistochemical reactions with COX-2 was appraised using a simplified scale. A case was diagnosed as COX-2 positive (1) when expression was observed in all tumour cells or in numerous cell clumps, or as COX-2 negative (0) when no reaction was noted or the reaction was present in only individual tumour cells (<10%). The intensity of immunohistochemical reactions with MDR1/P-gp was appraised using the semi-quantitative immunoreactive score (IRS) scale [17], in which the score reflected both the intensity of the reaction and the proportion of positive cells (Table 2). The final score represented the product of points given for individual characteristics and ranged between 0 and 12. The intensity of immunohistochemical reactions was appraised independently by two pathologists; in doubtful cases, a re-evaluation was performed using a double-headed microscope.
Control reactions
In each case, control reactions were included, in which specific antibody was substituted by Primary Mouse Negative Control (DakoCytomation). Control reactions were also performed for each of the examined antigens. For MDR1/P-gp, positive controls involved sections of six formalin-fixed and paraffin-embedded human liver samples (from the archive of the Chair and Department of Histology and Embryology, University School of Medicine in Poznañ, Poland) for each antibody. To evaluate specificity of the COX-2 antibody (Cayman Chemical Company) we [18] and other investigators [8] performed blocking experiments using the COX-2 blocking peptide (Cayman Chemical Company) according to the manufacturer's instructions.
Statistical analysis
Statistical analysis of the results took advantage of Statistica 98 PL software (Statsoft, Krakow, Poland). The employed tests included chi2 test, Spearman's rank correlation and ANOVA rank Kruskal-Wallis test. Kaplan-Meier's statistics and log-rank tests were performed using SPSS software (release 10.0; SPSS Inc., Chicago, IL, USA) to estimate the significance of differences in survival times. The length of survival was defined as the time between the primary surgical treatment and diagnosis of a recurrent tumour or death due to the neoplastic disease. Because the univariate analysis failed to disclose any significant relationships between studied clinicopathological parameters (age, menopausal status, grade and stage) and overall survival and progression free time in studied patients (P > 0.05), no multivariate analysis was conducted. The absence of a relationship between the until now most recognised and most effective prognostic factors [19] and patient survival time resulted most probably from the highly uniform character of the examined group of patients. As many as 96% of the examined patients represented the stage II (UICC) [15]. The uniform character of the group allowed a more unequivocal evaluation of the effect of the intensity of expression of the studied protein on the survival time of the patients.
Results
Immunostaining in control preparations and in breast cancers
Immunostaining for COX-2 occurred in a cytoplasmic localization and was of varying intensity among individual cases (Fig. 1). In six COX-2 positive cases, we performed blocking experiments using the COX-2 blocking peptide, as described previously [18]. We found no staining in control preparations (Fig. 1). Expression of COX-2 was noted in 46 cases (44%).
Immunostaining for MDR1/P-gp occurred in membranes in samples of healthy human liver and in the cytoplasm and membranes in breast cancers, and was of varying intensity among individual cases (Fig. 2a,b). The mean overall immunoreactivity score for MDR1/P-gp expression detected with antibody C219 was 3.45 ± 3.49 standard deviation (range: 0 IRS to 12 IRS) and with antibody JSB-1 was 3.34 ± 3.49 standard deviation (range: 0 IRS to 12 IRS). We found a strict positive correlation between expression with C219 and JSB-1 (Spearman's rank correlation, R = 0.99, P < 0.001).
Using the ANOVA rank Kruskal-Wallis test, we examined the relationship between COX-2 expression and the overall immunoreactivity score of MDR1/P-gp expression. We found that the overall immunoreactivity score of MDR1/P-gp expression with C219 and JSB-1 was significantly higher in cases showing expression of COX-2 (P < 0.0001) (Fig. 3a,b; Table 3).
Chi2 tests were also used to analyse the relationships between the intensity of COX-2 and MDR1/P-gp expression on the one hand and the grade, stage, pT (UICC) [15], pN (UICC) [15] and menopausal status of studied patients on the other. We found that at a grade of G3, a significantly higher proportion of cases manifested COX-2 expression compared to patients with a G2 grade (Table 4). The pN1 cases were also shown to exhibit a higher overall immunoreactivity score for C219 and JSB-1 compared to pN0 cases (Table 4); similarly, cases with a G3 grade showed a higher overall immunoreactivity score for C219 and JSB-1 compared to those with a G2 grade (Table 4).
COX-2 and MDR1/P-gp expression and patient survival
Kaplan-Meier statistics were used to analyse overall survival time and progression-free survival. In the entire study group, the COX-2 positive cases manifested a significantly shorter overall survival time and progression-free survival compared to COX-2 negative cases (Fig. 4c,f). Cases with an overall C219 and JSB-1 immunoreactivity score between 0 and 3 were also shown to manifest a significantly extended overall survival time and progression-free survival compared to cases with an overall C219 and JSB-1 immunoreactivity score between 4 and 12 (Fig. 4a,b,d,e). The same results were obtained in the subgroup of patients that, following surgery, were treated only with chemotherapy (Fig. 5a–f).
Discussion
We have described the expression of COX-2 and MDR1/P-gp proteins detected by immunohistochemistry in primary invasive breast cancers. Following the recommendations of the St Judge MDR Workshop on 'Methods to Detect MDR1/P-gp-associated Multidrug Resistance' [20], we have used two different monoclonal antibodies (C219 and JSB-1) directed against MDR1/P-gp. We found a strict positive correlation between MDR1/P-gp expression detected with C219 and JSB-1. We have confirmed that the studied proteins are expressed in a subset of breast cancers [11,21-23]. No relationship was discovered between COX-2 expression and such clinicopathological traits as pT, pN, stage or menopausal status. A higher proportion of COX-2 positive cases was noted in G3 compared to G2 patients. Previously, we noted that COX-2 expression was significantly associated with higher grade, lymph node status and larger tumour size [11]. Ristimäki et al. [21] described COX-2 expression using a tissue microarray in 1,576 cases of invasive breast cancer. They demonstrated that elevated COX-2 expression was associated with a large tumour size and high histological grade. In the present study, we have corroborated the positive correlation between COX-2 expression and the unfavourable clinicopathological prognostic indices in another group of patients. We have shown that a higher overall MDR1/P-gp immunoreactivity score is associated with lymph node involvement and a higher histological grade. Numerous authors have demonstrated a positive correlation between MDR1/P-gp expression and tumour stage [24,25]. To our knowledge, this study demonstrates for the first time a significant positive correlation between overall MDR1/P-gp immunoreactivity score and grade of breast cancer tumours. Thus, expression of MDR1/P-gp is typical for the less differentiated cases of breast cancer.
Expression of COX-2 in tumour cells represents an unfavourable prognostic factor in numerous tumours [9,10]. We [11] and other authors [21] have previously demonstrated that COX-2 represents an independent unfavourable prognostic factor in breast cancers. In this study, we have shown that COX-2 positive cases exhibit a significantly shorter overall survival and progression-free time in the entire study group and in a group of patients treated postoperatively with cytostatic drugs. Thus, we have confirmed the previously described observations that COX-2 expression in breast cancer is a negative prognostic factor.
The unfavourable prognostic significance of MDR1/P-gp expression has been documented in several tumours, including breast cancer [24,26-29]. Most of the studies have described the negative prognostic significance in breast cancer cases treated with chemotherapy. Few of the studies have suggested that MDR1/P-gp may also participate in the resistance to hormonal therapy [28]. In our study, we have shown, both in the entire group of patients (in whom surgery was followed by hormonal therapy and/or radiotherapy) and in the group of patients postoperatively treated with chemotherapy, that patients with a higher overall MDR1/P-gp immunoreactivity score exhibited a significantly shorter overall survival and progression-free time. As shown by our data, expression of MDR1/P-gp may not only be linked to resistance to cytostatic drugs but, considering that the higher overall MDR1/P-gp immunoreactivity score is associated with lymph node involvement and higher histological grade, may also represent an unfavourable prognostic factor independent of the applied therapy.
Several reports have suggested that elevated expression of COX-2 may stimulate expression of MDR1/P-gp. Such a phenomenon has been demonstrated, for example, in mesangial cells of rat kidneys, in which transfection with the COX-2 gene was followed by an increase in MDR1/P-gp expression [14], and in cells of the gastric mucosa during infection with Helicobacter pylori [30]. Ratnasinghe et al. [31] have described a positive correlation between COX-2 and MDR1/P-gp expression in breast cancer cases and cell lines. In this study, we confirm that a higher overall MDR1/P-gp immunoreactivity score is typical for COX-2 positive breast cancer cases. To our knowledge, this study demonstrates for the first time a negative prognostic significance of COX-2 and MDR1/P-gp coexpression in breast cancers. The data suggest that clinical studies should be performed on coxibs in chemotherapy-supporting treatment. Apart from their anti-tumour effects, these drugs might prevent the development of, or decrease the intensity of, the already existing MDR phenomenon.
Conclusion
We have shown the predictive significance of the immunohistochemical estimation of COX-2 and MDR1/P-gp expression in human breast cancers. We found a positive correlation between COX-2 and MDR1/P-gp expression and demonstrated that COX-2 and MDR1/P-gp are unfavourable prognostic factors in breast cancers and unfavourable predictive factors in chemotherapy-treated breast cancer cases. Clinical studies should be performed on coxibs as a supporting treatment in breast cancer chemotherapy.
Abbreviations
COX = cyclooxygenase; coxibs = cyclooxygenase 2 inhibitors; IRS = immunoreactive score; MDR = multidrug resistance; MDR1/P-gp = MDR1/P-glycoprotein.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
PS contributed to the conception and design of the study, performed immunohistochemistry experiments, statistics and interpretation of data. VM contributed to the conception and design of the study, statistics and interpretation of data. RM performed clinical data analysis and interpretation of data. KS, AW and MP performed immunohistochemistry experiments and contributed to statistical analysis and interpretation of data. JK was involved in clinical data analysis and interpretation of data. MD, CD, HL and MZ contributed to the conception and design of the study and interpretation of data. All the authors read and approved the final manuscript.
Acknowledgements
The study was supported by the Berliner Krebsgesellschaft e.V., Germany.
Figures and Tables
Figure 1 Immunohistochemical localization of cyclooxygenase-2 (red). The inset shows the control reaction with blocking peptide under the same conditions (hematoxylin, ×400).
Figure 2 Immunohistochemical localization of MDR1/P-glycoprotein (red). (a) With antibody C219 in breast cancer (showing cytoplasmic and membrane localizations in cancer cells) and healthy human liver (inset) (hematoxylin, ×400). (b) With antibody JSB-1 in breast cancer (showing cytoplasmic and membrane localizations in cancer cells) and healthy human liver (inset) (hematoxylin, ×400).
Figure 3 Correlation between cyclooxygenase-2 (COX-2) and MDR1/P-glycoprotein (MDR1/P-gp) expression. (a) Correlation between COX-2 and MDR1/P-gp expression detected with antibody C219. (b) Correlation between COX-2 and MDR1/P-gp expression detected with antibody JSB-1 in breast cancers. Cases with COX-2 expression show higher MDR1/P-gp expression (ANOVA Kruskall-Wallis rank test, P < 0.001).
Figure 4 Kaplan-Meier analysis of the complete group of 104 breast cancer patients. (a) Patients with a lower overall immunoreactivity score for MDR1/P-glycoprotein (MDR1/P-gp) expression detected with antibody C219 exhibit significantly longer overall survival. (b) Patients with a lower overall immunoreactivity score for MDR1/P-gp expression detected with antibody JSB-1 exhibit significantly longer overall survival. (c) Cyclooxygenase-2 (COX-2) negative cases exhibit significantly longer overall survival. (d) Patients with a lower overall immunoreactivity score for MDR1/P-gp expression detected with antibody C219 exhibit significantly longer progression-free survival. (e) Patients with a lower overall immunoreactivity score for MDR1/P-gp expression detected with antibody JSB-1 exhibit significantly longer progression-free survival. (f) COX-2 negative cases exhibit significantly longer progression-free survival.
Figure 5 Kaplan-Meier analysis of 28 breast cancer patients treated postoperatively with chemotherapy. (a) Patients with a lower overall immunoreactivity score for MDR1/P-glycoprotein (MDR1/P-gp) expression detected with antibody C219 exhibit significantly longer overall survival. (b) Patients with a lower overall immunoreactivity score for MDR1/P-gp expression detected with antibody JSB-1 exhibit significantly longer overall survival. (c) Cyclooxygenase-2 (COX-2) negative cases exhibit significantly longer overall survival. (d) Patients with lower overall immunoreactivity score for MDR1/P-gp expression detected with antibody C219 exhibit significantly longer progression-free survival. (e) Patients with lower overall immunoreactivity score for MDR1/P-gp expression detected with antibody JSB-1 exhibit significantly longer progression-free survival. (f) COX-2 negative cases exhibit significantly longer progression-free survival.
Table 1 Patient and tumour characteristics
Characteristics No. (%)
All patients 104 (100)
Age (years; mean 56.2)
≤ 50 33 (32)
51–60 29 (28)
>60 42 (40)
Menopause
Premenopausal 30 (29)
Postmenopausal 74 (71)
Grade
2 71 (68)
3 33 (32)
pTb
1 17 (16)
2 86 (83)
4 1 (1)
pNb
0 29 (28)
1 75 (72)
pMb
0 104 (100)
Stageb
I 3 (3)
IIa 40 (38)
IIb 60 (58)
IIIb 1 (1)
Histology
Ductal 103 (99)
Scirrhous 1 (1)
Therapya
Tamoxifen 70 (67)
Radiotherapy 51 (49)
Cyclophosphamide/Methotrexate/5-Fluorouracil 28 (27)
Cyclophosphamide/Adriamycin/5-Fluorouracil 1 (1)
Cyclophosphamide/Adriamycin 1 (1)
Progesterone 1 (1)
Letrozol 1 (1)
aSome patients received more than one special treatment.
bAccording to [15]
Table 2 Evaluation criteria of MDR1/P-gp expression using the immunoreactive score [17]
Percentage of positive cells Points Intensity of reaction Points
No positive cells 0 No reaction 0
<10% 1 Weak reaction 1
10–50% 2 Moderate reaction 2
51–80% 3 Intense reaction 3
>80% 4
MDR1/P-gp, MDR1/P-glycoprotein.
Table 3 Relationship between COX-2 and MDR1/P-gp (detected with C219 or JSB-1 antibodies) expression (Chi2-Test, P < 0.001)
COX-2 positive cases COX-2 negative cases Sum
Entire study group (n = 104)
C219 low [IRS 0–3] 56 5 61
C219 high [IRS 4–12] 2 41 43
Sum 58 46 104
JSB-1 low [IRS 0–3] 55 4 59
JSB-1 high [IRS 4–12] 3 42 45
Sum 58 46 104
Patients postoperatively treated only with chemotherapy (n = 28)
C219 low [IRS 0–3] 13 1 14
C219 high [IRS 4–12] 1 13 14
Sum 14 14 28
JSB-1 low [IRS 0–3] 12 1 13
JSB-1 high [IRS 4–12] 2 13 15
Sum 14 14 28
COX-2, cyclooxygenase-2; MDR1/P-gp, MDR1/P-glycoprotein; IRS, immunoreactive score [17].
Table 4 Relationship between COX-2 and MDR1/P-gp (detected with C219 or JSB-1 antibodies) expression and clinicopathologic factors
Characteristics No. of patients (%)
COX-2 positive COX-2 negative P value Chi2 test C219 high (IRS 4–12) C219 low (IRS 0–3) P value Chi2 test JSB-1 high (IRS 4–12) JSB-1 low (IRS 0–3) P value Chi2 test
pT
1 8 (8) 9 (9) 7 (7) 10 (10) 8 (8) 9 (9)
2 37 (36) 49 (47) 35 (34) 51 (49) 36 (35) 50 (48)
4 1 (1) 0 (0) 0.4954 1 (1) 0 (0) 0.7869 1 (1) 0 (0) 0.6306
pN
0 10 (10) 19 (18) 11 (11) 18 (17) 11 (11) 18 (17)
1 36 (35) 39 (37) 0.2117 32 (31) 43 (41) <0.001 34 (33) 41 (39) <0.001
Stage
I 1 (1) 2 (2) 1 (1) 2 (2) 1 (1) 2 (2)
IIa 16 (15) 24 (23) 16 (15) 24 (23) 17 (16) 23 (22)
IIb 28 (27) 32 (31) 25 (24) 35 (34) 29 (28) 31 (30)
IIIb 1 (1) 0 (0) 0.6628 1 (1) 0 (0) 0.5915 1 (1) 0 (0) 0.5874
Grade
2 26 (25) 45 (43) 25 (24) 46 (44) 26 (25) 45 (43)
3 20 (19) 13 (13) 0.0135 18 (17) 15 (14) <0.001 19 (18) 14 (13) < 0.001
Menopause
Praemenopausal 12 (12) 18 (17) 13 (13) 17 (16) 14 (13) 16 (15)
Postmenopausal 34 (33) 40 (38) 0.4553 30 (30) 44 (42) 0.8307 31 (30) 43 (41) 0.9703
COX-2, cyclooxygenase-2; MDR1/P-gp, MDR1/P-glycoprotein; IRS, immunoreactive score.
==== Refs
Jemal A Thomas A Murray T Thun M Cancer statistics CA Cancer J Clin 2002 52 23 47 11814064
Mori I Yang Q Kakudo K Predictive and prognostic markers for invasive breast cancer Pathol Int 2002 52 186 194 11972862 10.1046/j.1440-1827.2002.01335.x
Eifel P Axelson JA Costa J Crowley J Curran WJ JnrDreshler A Fulton S Hendricks CB Kemeny M Kornblith AB Louis TA Markman M Mayer R Roter D National Institutes of Health Consensus Development Conference Statement: adjuvant therapy for breast cancer, November 1-3, 2000 J Natl Cancer Inst 2001 90 13 979 989
Goldhirsch A Glick JH Gelber RD Meeting highlights: international consensus panel on the treatment of primary breast cancer J Natl Cancer Inst 1998 90 1601 1608 9811309 10.1093/jnci/90.21.1601
Lage H ABC-transporters: implications on drug resistance from microorganisms to human cancer Int J Antimicrob Agents 2003 22 188 199 13678820 10.1016/S0924-8579(03)00203-6
Lage H Drug resistance in breast cancer Cancer Therapy 2003 1 81 91
Gasparini G Longo R Sarmiento R Morabito A Inhibitors of cyclo-oxygenase 2: a new class of anticancer agents? Lancet Oncol 2003 4 605 615 14554238 10.1016/S1470-2045(03)01220-8
Ristimäki A Nieminen O Saukkonen K Hotakainen K Nordling S Haglund C Expression of cyclooxygenase-2 in human transitional cell carcinoma of the urinary bladder Am J Pathol 2001 158 849 853 11238034
Denkert C Kobel M Pest S Koch I Licht D Kobel M Reles A Sehouli J Dietel M Hauptmann S Expression of cyclooxygenase 2 is an independent prognostic factor in human ovarian carcinoma Am J Pathol 2002 160 893 903 11891188
Denkert C Kobel M Berger S Kobel M Siegert A Hauptmann S Expression of cyclooxygenase 2 in human malignant melanoma Cancer Res 2001 61 303 308 11196178
Denkert C Winzer KJ Muller BM Weichert W Pest S Kobel M Kristiansen G Reles A Siegert A Guski H Hauptmann S Elevated expression of cyclooxygenase-2 is a negative prognostic factor for disease free survival and overall survival in patients with breast carcinoma Cancer 2003 97 2978 2987 12784332 10.1002/cncr.11437
Brueggemeier RW Richards JA Petrel TA Aromatase and cyclooxygenase in breast cancer J Steroid Biochem Mol Biol 2003 86 501 507 14623550 10.1016/S0960-0760(03)00380-7
Davies G Martin LA Sacks N Dowsett M Cyclooxygenase-2 (COX-2), aromatase and breast cancer: a possible role for COX-2 inhibitors in breast cancer chemoprevention Ann Oncol 2002 13 669 678 12075734 10.1093/annonc/mdf125
Patel VA Dunn MJ Sorokin A Regulation of MDR1 (P-glycoprotein) by cyclooxygenase-2 J Biol Chem 2002 277 38915 38920 12138126 10.1074/jbc.M206855200
Sobin LH Wittekind C TNM classification of malignant tumours 2002 5 New York: Wiley-Liss Inc
Elston CW Ellis IO Pathological prognostic factors in breast cancer. The value of histological grade in breast cancer: experience from a large study with long-term follow-up Histopathology 1991 19 403 410 1757079
Remmele W Stegner HE Recommendation for uniform definition of an immunoreactive score (IRS) for immunohistochemical estrogen receptor detection (ER-ICA) in breast cancer tissue Pathologe 1987 8 138 140 3303008
Denkert C Weichert W Pest S Koch I Licht D Kobel M Reles A Sehouli J Dietel M Hauptmann S Overexpression of the embryonic-lethal abnormal vision-like protein HuR in ovarian carcinoma is a prognostic factor and is associated with increased cyclooxygenase 2 expression Cancer Res 2004 64 189 195 14729623
Fitzgibbons PL Page DL Weaver D Thor AD Allred DC Clark GM Ruby SG O'Malley F Simpson JF Connoly JL Prognostic factors in breast cancer. College of American Pathologists Consensus Statement 1999 Arch Pathol Lab Med 2000 124 966 978 10888772
Beck WT Grogan TM Willman CL Cordon-Cardo C Parham DM Kuttesch JF Andreeff M Bates SE Berard CW Boyett JM Methods to detect P-glycoprotein-associated multidrug resistance in patients' tumours: consensus recommendations Cancer Res 1996 56 3010 3020 8674056
Ristimäki A Sivula A Lundin J Lundin M Salminen T Haglund C Joensuu H Isola J Prognostic significance of elevated cyclooxygenase-2 expression in breast cancer Cancer Res 2002 62 632 635 11830510
Kanzaki A Toi M Nakayama K Bando H Mutoh M Uchida T Fukumoto M Takebayashi Y Expression of multidrug resistance-related transporters in human breast carcinoma Jpn J Cancer Res 2001 92 452 458 11346468
Hennequin E Delvincourt C Pournz C Jardillier JC Expression of mdr1 gene in human breast primary tumours and metastases Breast Cancer Res Treat 1993 26 267 274 8251651 10.1007/BF00665804
Decker DA Morris LW Levine AJ Pettinga JE Grudzien JL Farkas DH Immunohistochemical analysis of P-glycoprotein expression in breast cancer: clinical correlations Ann Clin Lab Sci 1995 25 52 59 7762969
Sun SS Hsieh JF Tsai SC Ho YJ Lee JK Kao CH Expression of mediated P-glycoprotein multidrug resistance related to Tc-99m MIBI scintimammography results Cancer Lett 2000 153 95 100 10779636 10.1016/S0304-3835(00)00356-6
Gottesman MM Fojo T Bates SE Multidrug resistance in cancer: role of ATP-dependent transporters Nat Rev Cancer 2002 2 48 58 11902585 10.1038/nrc706
Materna V Pleger J Hoffmann U Lage H RNA expression of MDR1/P-glyocprotein, DNA-topoisomerase, and MRP2 in ovarian carcinoma patients: correlation with chemotherapeutic response Gynecol Oncol 2004 94 152 160 15262134 10.1016/j.ygyno.2004.03.035
Linn SC Giaccone G van Diest PJ Blokhuis WM van der Valk P van Kalken CK Kuiper CM Pinedo HM Baak JP Prognostic relevance of P-glycoprotein expression in breast cancer Ann Oncol 1995 6 679 685 8664189
Nooter K Brutel DLR Look MP van Wingerden KE Henzen-Logmans SC Scheper RJ Flens MJ Klijn JG Stoter G Foekens JA The prognostic significance of expression of the multidrug resistance-associated protein (MRP) in primary breast cancer Br J Cancer 1997 76 486 493 9275026
Nardone G Rocco A Vaira D Staibano S Budillon A Tatangelo F Sciulli MG Perna F Salvatore G Di Benedetto M Expression of COX-2, mPGE-synthase 1, MDR-1 (P-gp), and Bcl-xL: a molecular pathway of H pylori-related gastric carcinogenesis J Pathol 2004 202 305 312 14991895 10.1002/path.1512
Ratnasinghe D Daschner PJ Anver MR Kasprzak BH Taylor PR Yeh GC Tangera JA Cyclooxygenase-2, P-glycoprotein-170 and drug resistance; is chemoprevention against multidrug resistance possible? Anticancer Res 2001 21 2141 2147 11501838
|
16168133
|
PMC1242165
|
CC BY
|
2021-01-04 16:04:34
|
no
|
Breast Cancer Res. 2005 Aug 25; 7(5):R862-R870
|
utf-8
|
Breast Cancer Res
| 2,005 |
10.1186/bcr1313
|
oa_comm
|
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-9-r731616808010.1186/gb-2005-6-9-r73ResearchIdentification of cyanobacterial non-coding RNAs by comparative genome analysis Axmann Ilka M [email protected] Philip [email protected] Jörg [email protected] Stefan [email protected] Hanspeter [email protected] Wolfgang R [email protected] Humboldt-University, Department of Biology/Genetics, Chausseestrasse, D-Berlin, Germany2 Humboldt-University, Institute for Theoretical Biology, Invalidenstrasse, Berlin, Germany3 Max Planck Institute for Infection Biology, Schumannstrasse, Berlin, Germany4 University Freiburg, Institute of Biology II/Experimental Bioinformatics, Schänzlestrasse, Freiburg, Germany2005 17 8 2005 6 9 R73 R73 30 3 2005 1 6 2005 20 7 2005 Copyright © 2005 Axmann et al.; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
The first genome-wide and systematic screen for non-coding RNAs (ncRNAs) in cyanobacteria. Several ncRNAs were computationally predicted and their presence was biochemically verified. These ncRNAs may have regulatory functions, and each shows a distinct phylogenetic distribution.
Background
Whole genome sequencing of marine cyanobacteria has revealed an unprecedented degree of genomic variation and streamlining. With a size of 1.66 megabase-pairs, Prochlorococcus sp. MED4 has the most compact of these genomes and it is enigmatic how the few identified regulatory proteins efficiently sustain the lifestyle of an ecologically successful marine microorganism. Small non-coding RNAs (ncRNAs) control a plethora of processes in eukaryotes as well as in bacteria; however, systematic searches for ncRNAs are still lacking for most eubacterial phyla outside the enterobacteria.
Results
Based on a computational prediction we show the presence of several ncRNAs (cyanobacterial functional RNA or Yfr) in several different cyanobacteria of the Prochlorococcus-Synechococcus lineage. Some ncRNA genes are present only in two or three of the four strains investigated, whereas the RNAs Yfr2 through Yfr5 are structurally highly related and are encoded by a rapidly evolving gene family as their genes exist in different copy numbers and at different sites in the four investigated genomes. One ncRNA, Yfr7, is present in at least seven other cyanobacteria. In addition, control elements for several ribosomal operons were predicted as well as riboswitches for thiamine pyrophosphate and cobalamin.
Conclusion
This is the first genome-wide and systematic screen for ncRNAs in cyanobacteria. Several ncRNAs were both computationally predicted and their presence was biochemically verified. These RNAs may have regulatory functions and each shows a distinct phylogenetic distribution. Our approach can be applied to any group of microorganisms for which more than one total genome sequence is available for comparative analysis.
==== Body
Background
Cyanobacteria constitute a huge and diverse group of photoautotrophic bacteria that perform oxygenic photosynthesis and populate widely diverse environments such as freshwater, the oceans, the surface of rocks, desert soil or the Antarctic. Their existence can be traced back by fossil records possibly up to 3.5 billion years [1].
Because of its small cell size of less than one micron and its requirement for special isolation and cultivation procedures, the marine cyanobacterium Prochlorococcus marinus had escaped discovery until just a decade ago [2,3]. In contrast to the majority of cyanobacteria, Prochlorococcus shares with Prochlorothrix hollandica and Prochloron sp. the presence of a protein-chlorophyll b complex for photosynthetic light harvesting [4,5]. The presence of chlorophyll b had previously been taken as evidence for a separate phylum, the prochlorophyta, to join these three taxa. Molecular evidence has shown, however, that Prochlorococcus, Prochlorothrix and Prochloron are not closely related to each other [6].
Cyanobacteria of the genera Prochlorococcus and Synechococcus constitute the most important primary producers within the oceans [7]. Of these, the four marine cyanobacteria, Prochlorococcus marinus MED4, MIT 9313, SS120 and Synechococcus sp. WH 8102 share a 16S ribosomal RNA identity of more than 97%. In the natural environment, Prochlorococcus exists in two distinct 'ecotypes' that thrive at different light optima and constitute distinct phylogenetic clades [8,9]. Thus, the genomes of the low-light-adapted isolates Prochlorococcus MIT 9313 and SS120, and of the high-light-adapted MED4 differ by hundreds of genes, facilitating their specialization to different niches within the marine ecosystem [10-12].
An extreme genome minimization occurred in MED4 and SS120 [13], which is thought to be an adaptation to the very oligotrophic and stable environment from which these two strains originated [10,12]. The MED4 strain was isolated from a depth of 5 m in the Mediterranean Sea; its genome of 1.66 megabase pairs (Mbp) encodes 1,716 open reading frames, among them only four histidine kinases, six response regulators and five sigma factors [12]. Prochlorococcus SS120 originated from 120 m in the Sargasso Sea [3], and 1,884 predicted protein-coding genes, including five histidine kinases, six response regulators and five sigma factors, have been annotated for its 1.7 Mbp genome [10]. These data indicate a drastically reduced number of systems for signal transduction and environmental stress response (e.g. two-component systems) compared to the larger and more complex genomes of cyanobacteria such as Synechocystis sp. PCC 6803 and Anabaena sp. PCC 7120, which each harbour 42 and 126 histidine kinases, respectively [14,15]. The small number of regulatory genes in marine Synechococcus and Prochlorococcus may reflect a more stable environment, in which reactive regulatory responses are less relevant.
It is now becoming increasingly clear that aside from regulatory proteins, bacteria also possess a significant number of regulatory non-coding RNAs (ncRNAs). These are a heterogeneous group of functional RNA molecules normally without a protein-coding function. They are frequently smaller than 200 nucleotides (nt) in size, and act to regulate mRNA translation/decay but can also bind to proteins and thereby modify protein function (for a recent review see [16]). It is well established that such RNAs control plasmid and viral replication [17], transposition of transposable elements [18], bacterial virulence [19], quorum sensing [20] and are important factors in bacterial regulatory networks that respond to environmental changes [21,22]. As a result of recent systematic searches, more than 60 ncRNAs are now known in Escherichia coli, most of which had been overlooked by traditional genome analysis [23-28]. Many of these versatile bacterial riboregulators use base pairing interactions to regulate the translation of target mRNAs. Because most of these antisense-acting ncRNAs have only incomplete target complementarity, duplex formation frequently depends on the activity of Hfq, an RNA chaperone, which is structurally and functionally somewhat similar to eukaryotic Sm proteins [29]. Only very recently, an hfq homologue was predicted in cyanobacterial genomes, including two of the strains used in this study (Synechococcus WH 8102 and Prochlorococcus MIT 9313) [29]. This lends support to the idea that riboregulatory processes similar to those of enterobacteria should exist in cyanobacteria.
There is currently no information about the presence of regulatory RNAs and their genes in marine cyanobacteria. Apart from rRNA and tRNA genes, only three other well-characterized RNA genes have been annotated by sequence similarity in each of the four genomes used in this study. These encode the RNA components of RNAse P (M1 RNA), the signal recognition particle (scRNA) and tmRNA (rnpB, ffs and ssrA, respectively). Although the Prochlorococcus tmRNA has not been analyzed experimentally so far, it was subject to several in silico analyses, predicting it would consist of two separate molecules derived from a common precursor [30,31]. Such a permuted gene structure producing a two-piece mature tmRNA results in a dramatically reduced number of secondary structure elements: only two pairings were predicted in the tRNA-like domain, and a single transient pseudoknot and three other stem-loops were computed for the molecule containing the tag reading frame, whereas the pseudoknot number alone is five in one-piece cyanobacterial tmRNA [30]. It remains unclear, however, what, if any, selective advantage such a simplification in the structural elements of this RNA species would bring. This prompts the question of whether number and complexity of ncRNAs in these organisms is generally reduced as seen with tmRNA and regulatory proteins. And if so, what kind of ncRNAs might have escaped such an elimination and simplification process?
Systematic searches for ncRNAs are still lacking for most eubacterial phyla outside the enterobacteria. Recently, an effective approach to score multiple alignments in terms of secondary structure conservation was suggested [32,33]. Using a comparative genomics approach based on the recently published genome sequences, we have predicted candidates for ncRNAs in four marine cyanobacteria. The expression of these candidate sequences was tested under various growth and stress conditions that are encountered in the natural environment. This resulted in the identification of seven new ncRNAs in MED4, and several homologues in the other three strains.
Results
Small RNAs in marine cyanobacteria
Total RNA samples from the four marine cyanobacteria Prochlorococcus MED4, MIT 9313, SS120 and Synechococcus WH 8102 were separated on high-resolution polyacrylamide gels to get an overview of the presence of small RNAs. This analysis showed abundant RNA molecules with sizes in the range 50 to 250 nt (Figure 1). A particularly abundant class of RNAs in the 70 to 90 nt size range indicates the location of tRNAs in this gel, which was confirmed by hybridization to the tRNASer [GCU]. The hybridization signal for this tRNA was located at the upper end of this abundant cluster of bands, consistent with the fact that it is the largest annotated tRNA in these genomes. Several small RNAs migrated above the tRNA cluster and very few below it (indicated by the weakly visible bands below the tRNAs). These bands collectively indicated the occurrence of abundant small mRNAs, ncRNAs and precursors to tRNAs and rRNAs.
Eubacterial RNA species, however, very rarely reach a concentration that allows direct identification in a gel. For known RNA species and their possible precursors or degradation products, information on their expression can be gained from hybridization. Here we used oligonucleotide probes for the scRNA and tmRNA and, as controls, the 5S rRNA and tRNASer [GCU], which was predicted to be the tRNA with the highest molecular mass. The lengths of the scRNAs in the four strains vary between 90 and 100 nt, in keeping with the varying lengths of the respective annotated ffs genes. The 5S rRNA was detected as a very abundant RNA species together with two precursors. Furthermore, the results of these Northern hybridizations confirmed that Prochlorococcus tmRNA is indeed composed of two separate molecules [30].
Several additional bands in the investigated size range indicate the presence of additional abundant small mRNAs or ncRNAs. The lack of specific oligonucleotide probes for hybridisation, however, makes it difficult to get information about these. We thus used a computational prediction to identify candidates for further testing.
Computational screening and experimental testing identifies novel RNA species
An overview of the computational screening is displayed in Figure 2 and a summary of the highest scoring clusters is given in Table 1. The analysis was basically focused on sequence and structure similarities. Detailed information on all clusters predicted by our method, including the positions of all sequences, is available online [34].
Although the sequence similarities between the predicted RNA elements in cyanobacteria and other organisms were weak, for many of the clusters, clues for their possible function could be obtained from the literature. These included elements that, according to location or structure, might be functionally related to enterobacterial mRNA leader regions mediating the autogenous control of r-protein and rRNA expression (clusters 5, 92, 227, 228) [35,36], the rpoBC leader (cluster 245) [37] and the likely terminator (cluster 226). We decided against direct experimental analysis of these elements, which are less likely to be novel types of ncRNAs. Additionally, two possible riboswitches for thiamine pyrophosphate (cluster 2) [38] and cobalamin (cluster 101) [39] were excluded from further experimental investigations.
In the remaining clusters, all candidate sequences from MED4 were tested by Northern hybridization. This restriction was introduced in order to focus the experimental analysis on one particular strain. Each of these seven candidate regions was probed for transcripts from both strands. Three distinct ncRNAs and a group of four related ones yielded strong signals with RNA preparations from MED4. Because some of these ncRNAs have a phylogenetic distribution beyond Prochlorococcus (see below), we introduced a more general gene designation, yfr (for cyanobacterial functional RNA-coding gene), and Yfr for the respective RNAs. Each of these genes is discussed in detail in the following sections.
Yfr1: a small RNA encoded between guaB and trxA
The yfr1 gene was detected in three of the four cyanobacteria in the intergenic region separating guaB and trxA (Figure 3). In the computational screening only the Yfr1 RNAs from MIT 9313 and WH 8102 were detected with a reasonable Z-score of -3.97 and the MED4 sequence was identified with relaxed BLASTN parameters manually. Although the two adjacent genes guaB and trxA are located in a similar genomic arrangement in SS120, a yfr1 gene was not found at this or any other genomic position nor indicated by a Northern hybridization signal. This result is in agreement with the high sequence divergence of the guaB-trxA intergenic spacer in SS120 compared to MED4, MIT 9313 and WH 8102.
The direction of yfr1 is conserved between MED4, MIT 9313 and WH 8102. It is transcribed in the same direction as the mRNAs from two close-by neighbouring genes, indicating the possibility of cotranscription. Therefore, we searched for the presence of specific transcriptional initiation sites (TIS) for yfr1 and for trxA by rapid amplification of cDNA ends (RACE). A conserved TIS was mapped for yfr1, indicating that this transcript originates from a specific promoter (Figure 3a) and reducing the likelihood that it is cotranscribed with guaB. Transcription of the adjacent trxA gene, encoding the redox regulator thioredoxin, was found to initiate approximately 100 bp downstream of the 3' end of the yfr1 gene (Figure 3a); cotranscription of yfr1 with trxA is thus unlikely. In SS120, the lack of the yfr1 TATA box, and the fact that the trxA TIS and TATA box are shifted upstream by about 20 nt compared to the other three strains (Figure 3a), lends additional support for the absence of a yfr1 gene.
Compared to other eubacterial ncRNAs [25,40], Yfr1 is one of the shortest bacterial ncRNAs, with a length of only 54, 56 or 57 nt (in strains MED4, MIT 9313 and WH 8102, respectively; Figure 3b). Although direct information on cyanobacterial RNAs is scarce [41,42] and not a single study exists for marine cyanobacteria, the half-lifes of eubacterial mRNAs are frequently in the range of a few minutes. In contrast, Yfr1 is extremely stable as a half-life of more than 60 minutes was measured after transcriptional arrest was induced by rifampicin (see Additional data file 1). No peptide reading frame within yfr1 is conserved between any of the three strains, although, as expected for a stable RNA, the three strains that express yfr1 share extensive structural conservation. They contain two terminal tetranucleotide loops separated by a 16 to 19 nt unpaired region that contains a CA dinucleotide repeat. Consistently, the 3' located stem-loop element is formed by at least five GC pairs, and is followed by a short stretch of U residues, indicative of a Rho-independent transcription terminator (Figure 3c).
The expression of many bacterial regulatory RNAs is stimulated by varying environmental cues, and often so by the stress response in which these RNAs then play a role. Therefore, a variety of stress conditions and their possible impact on the accumulation of ncRNAs were tested. Figure 4 shows a series of Northern hybridizations with RNA samples from cells that had been depleted of nitrogen, phosphate or iron, exposed to higher intensities of white or of blue light, or treated with 2 μM 3-(3,4-dichlorophenyl)-1, 1-N-N'-dimethylurea (DCMU) to induce oxidative stress or grown at elevated or lowered temperatures (30°C and 15°C). Normalization of loaded RNA used 5S rRNA as an internal standard to compensate for small RNA sample loading differences; however, Yfr1 levels were unaffected by any of these conditions.
A new family of related short RNAs
In top scoring cluster 194, a family of structurally highly similar RNAs (Yfr2, Yfr3 and Yfr4) was predicted (Table 1). Subsequent local alignments identified yet another similar sequence in MED4, and at least one homologue each in SS120, MIT 9313 and WH 8102.
Northern hybridizations with oligonucleotide probes specific for each of these candidate genes in MED4 yielded distinct bands of 89 to 95 nt. RACE mapping of 5' ends further confirmed that all four loci are transcribed in this organism (Figure 5). The RNAs Yfr2 through Yfr5 in MED4 and their homologues in the other genomes are each encoded by distant genomic loci and the position of their genes is not fixed within the four investigated genomes with respect to adjacent genes (Table 2). The sequence comparison shows that for MED4, Yfr2 and Yfr5 on one hand and Yfr3 and Yfr4 on the other are more similar to each other (Figure 5a). The predicted secondary structures of the Yfr2-5 ncRNA family in MED4 are highly conserved with a GGAAACA repeat within the loop of the predicted 5' hairpin (Figure 5c). Among the different tested environmental conditions, the amount of Yfr2-5 was affected by temperature (up at 15°C and down at 30°C) as well as by nitrogen limitation and incubation in blue light (Figure 4).
A long RNA in MED4 and SS120
The yfr6 gene was predicted in cluster 53 (Table 1). This cluster included nine different sequences (see Additional data file 1, Figure S10), among which only yfr6 in MED4 and SS120 may code for a functional RNA. The seven other sequences each have only about 40 nucleotide positions from their respective 5' untranslated region in common with Yfr6. That was sufficient to cluster all nine sequences together, but these other seven sequences included mRNAs for two previously unannotated open reading frames in MED4 and MIT 9313 (PMM3822n and PMT3904n [13]), the three annotated genes Pro0415 (in SS120), SYNW1950 and SYNW2450 (in WH 8102) as well as two more possible open reading frames in WH 8102, (27_W1i1019 and 6_W1i283), which possibly code for peptides with similarity to the first five gene products (see also Figure S10B in Additional data file 1). In contrast, Yfr6 from the two strains each have an extended sequence and structural similarity to each other.
In MED4, yfr6 is located between the hypothetical PMM0660 gene and PMM0659, the latter encoding 322 amino terminal residues of a DNA ligase. The region is framed by trnS and nrdJ (encoding a B12-dependent ribonucleotide reductase). In SS120, the nrdJ-trnS region lacks the yfr6 gene, which instead is located 448 nt downstream of another ncRNA gene, yfr7. Despite the different genomic locations, Yfr6 sequences from the two strains show a nucleotide identity of approximately 70% to each other (Figure 6a; Additional data file 1, Figure S10). A Northern blot signal for Yfr6 is restricted to MED4 and SS120 and no signal was found in WH 8102 and MIT 9313 (Figure 6b). This 244 nt RNA had a half-life of approximately 2 minutes in MED4. In MED4, blue light and incubation in the cold elevated the expression of Yfr6 compared to white light or darkness. In addition, expression was reduced upon nitrogen depletion and under high light conditions (Figure 4). The yfr6 locus could also code for a 33 amino acid peptide as there is a possible reading frame that is conserved between MED4 and SS120 that begins at nucleotide 97 of the Yfr6 transcript in MED4. This situation, a relatively long transcript with strong structural potential (Figure 6c) and a very short centrally located reading frame, resembles the RNAIII from Staphylococcus aureus, a riboregulator from which the 26 amino acid δ-hemolysin peptide is also translated [43]. In the hyperthermophilic archaeon Sulfolobus solfataricus, recently as many as 13 sense strand RNA sequences have been found that were encoded either within, or overlapping, annotated open reading frames [44].
Yfr7 exists in 11 different marine cyanobacteria
The yfr7 gene is located downstream of purK (encoding phosphoribosylaminoimidazole carboxylase) in all four strains analyzed here (Table 2). At first, our search strategy identified this gene only in MED4 and SS120 (Table 1), due to the fact that in MIT 9313 and WH 8102 this corresponding region is located within annotated mRNA genes. These hypothetical genes, PMT0670 in MIT 9313 and SYNW1307 in WH 8102, are annotated on the forward strand. We did not detect their expression, but found strong signals for Yfr7, which is transcribed from the complementary strand. The sequence of Yfr7 is highly conserved between the four strains (Figure 7a). Rifampicin tests showed this RNA to be stable (half-life >1 h). In MED4, expression of Yfr7 was not affected by conditions employed in Figure 4.
Its high sequence conservation enabled us also to define oligonucleotides that hybridized to this RNA in four additional, unsequenced strains of Prochlorococcus and in three additional Synechococcus strains (Figure 7b). The signal pattern is very distinct as all three Prochlorococcus strains adapted to high light (MED4, MIT 9312, MIT 9215) have two signals in hybridization, one at approximately 200 nt and one at approximately 300 nt, whereas RNA from the four low-light-adapted Prochlorococcus (SS120, MIT 9313, NATL2A and MIT 9211) and four Synechococcus (WH 8102, WH 7803, WH 8020, RS9906) strains gave a single signal at approximately 175 to 185 nt (Figure 7b). These strains represent a large genetic diversity within the marine cyanobacterial radiation [45], thus the presence of orthologues of yfr7 in additional and even more distant cyanobacteria appeared likely. Indeed, in the freshwater cyanobacteria Synechococcus PCC 6301 and Synechocystis PCC 6803, a 6Sa (or SsaA) RNA has also been described, which is located directly downstream of purK [46]. There is some structural similarity between Yfr7 and the 6Sa RNA, which leads us to assume that these RNAs are homologues of each other. In addition, a recent publication provided comparative structural information suggesting that the ncRNA Yfr7 we describe here and SsaA or 6Sa RNA from the latter cyanobacteria have structural elements in common with the 6S RNA of γ-proteobacteria, in particular a large internal loop (the central bubble in Figure 7c), a typical closing stem and terminal loop [47]. This possibly indicates that the here described Yfr7s are the orthologues of γ-proteobacterial 6S RNA and may have a similar role throughout the whole eubacterial radiation.
Discussion
The genomes of Prochlorococcus marinus SS120, MIT 9313, MED4 and Synechococcus WH 8102 provide a unique dataset for cyanobacterial genome analysis. These genomes differ by several hundred genes from each other, yet most of the operons and gene clusters present in more than a single genome are co-linear [10-12]. Furthermore, the Synechococcus/Prochlorococcus group is very well investigated with regard to their global significance in the marine ecosystem, and there is clear evidence for speciation processes in terms of specific ecological niches, the position in phylogenetic trees, and the presence of more or less derived features (for a review, see [7]). Although there is no well established genetic system for Prochlorococcus to test gene functions directly, these features collectively make these cyanobacteria emerging model organisms for marine photoautotroph bacteria.
In certain other eubacteria such as E. coli and Vibrio cholerae, several ncRNAs were demonstrated to be essential regulatory factors mediating rapid responses to environmental changes. The underlying regulatory mechanisms range from antisense binding to mRNAs to direct sensing of metabolites, as it is the case with riboswitches. For free-living marine phototrophs such as the cyanobacteria investigated here, regulatory circuits involving ncRNAs can be expected too. However, except for RNase P RNA, scRNA and tmRNA, the three ncRNAs that are easiest to identify, little had been known about ncRNAs genes in these marine cyanobacteria. In a broader context, information has remained scarce on riboregulators and RNA-coding genes even for the group of cyanobacteria as a whole.
Using an elaborate biochemical protocol, a single ncRNA was previously identified in the freshwater cyanobacteria Synechococcus PCC 6301 and Synechocystis PCC 6803 [46]. In addition, mapping of transcriptional units within the gas vesicle operon of Calothrix identified a single antisense transcript [48]. Here, we report the presence of new non-coding RNAs in the group of marine unicellular cyanobacteria with a focus on Prochlorococcus marinus MED4. Several more ncRNA candidate genes were predicted in the two relatively larger genomes of WH 8102 and MIT 9313 but still await experimental testing. An overview of the candidate regions identified by our screen is presented in Table 1 and a summary of the experimentally confirmed new ncRNAs is presented in Table 2. In addition to the identification of ncRNAs, the computational results indicated the presence of conserved secondary structure elements relating to the upstream untranslated regions of several r-protein operons. Thus, autogenous control mechanisms over the expression of these operons, similar to those in enterobacteria [35,36] may exist in these cyanobacteria.
The percentage of true RNA elements and ncRNAs found in our screen is very high, whereas the number of predicted ncRNA genes above the Z-score cut-off was low in MED4. It is likely that additional candidate ncRNAs have escaped detection. The performance of the computational algorithm is sensitive to the number of sequences. One important limitation in this context relates to the focus on RNA structures that are additionally conserved by primary sequence. Furthermore, because of the restriction to intergenic regions, ncRNAs that reside within annotated regions will be missed. This affects the whole class of antisense RNAs that are encoded complementary to their target. Also, misannotations may reduce the number of sequences in a cluster, like in the case of yfr7, which is in a region in which a reading frame was annotated on the complementary strand in two of the genomes investigated here. Indeed, in a test using an alignment of Yfr7 from all four species, an improved Z-score of -7.8 was detected.
Our analysis did reveal an interesting set of structural elements. Especially for MED4 and SS120, which underwent a strong genome reduction, the ncRNAs found may be of considerable importance. Both WH 8102 and MIT 9313 contain a hfq gene, whose product has been shown to be intimately linked to the activity of small regulatory RNAs in enterobacteria [29]. Intriguingly, there is no hfq gene in SS120 or MED4, although the genomic region flanking hfq is otherwise conserved among the four species (Figure 8). It is likely that, together with hfq, several ncRNA genes have been deleted during the evolution of the Prochlorococcus group towards the minimal genome. Thus, those ncRNAs still remaining in an organism such as MED4 must have been subject to strong positive selection and may act independently of Hfq. (It is worthwhile noting that in E. coli, only 30% of investigated ncRNAs have been shown to be bound by Hfq [28].)
The functions of these ncRNAs are currently unclear. The mode of action of ncRNAs supposed to act through an antisense mechanism can be studied by transferring the ncRNA as well as the putative target(s) to an appropriate host or model organism. For unicellular marine cyanobacteria, Synechococcus WH 7803 might become such a model as its genome analysis has almost completely been finished [49] and because there is a genetic system. Those ncRNAs with orthologues over a wider phylogenetic distance could be functionally analyzed also directly in cyanobacteria for which well-established genetic tools exist, such as Synechococcus PCC 7942 or Synechocystis PCC 6803. A very good candidate is Yfr7, which is likely to be present in all cyanobacteria and could be the orthologue of γ-proteobacterial 6S RNA [47]. 6S RNA is required for the repression of σ70-dependent promoters under nutrient limitation and concomitant activation of certain σS-dependent promoters [50]. Cyanobacteria do not harbor an obvious orthologue of the enterobacterial stationary phase sigma factor σS. Therefore, it remains to be shown if Yfr7/SsaA/6Sa RNA is also functionally related to γ-proteobacterial 6S RNA. But the widespread occurrence of this ncRNA opens exciting opportunities to test the function of Yfr7 directly in cyanobacteria.
Evidence of function may further come from the comparison of expression patterns, structures as well as genomic location, and from the presence or absence of a given ncRNA gene in the different strains. For instance, yfr1 might be dispensable for growth at greater depths; this gene is clearly absent from the ultra low-light-adapted SS120 but is present in the other three cyanobacteria, whereas Yfr2 through Yfr5 are in length and the degree of mutual identity similar to four ncRNAs implicated in quorum sensing in Vibrio species [20]. Consequently, the ncRNAs identified here may constitute important regulatory or structural components of a free-living marine cyanobacterium.
Conclusion
The first genome-wide and systematical screen for ncRNAs in cyanobacteria is provided. Genes encoding functional RNAs are notoriously difficult to predict during standard annotation of microbial genomes. Here, we took a comparative computational approach that was based on sequence and structure conservation as was recently introduced for the identification of eukaryotic ncRNAs [33]. In view of the rapidly growing number of microbial genome sequences, such screens that are based on comparative analysis will become increasingly possible. We have analyzed the highest scoring candidates of the prediction further and detected several previously unknown ncRNAs as well as other elements that function at the RNA level. The list of high scoring candidates contained a very low rate of true negatives. This indicates two points: first, the employed method is very efficient in finding microbial ncRNAs and other RNA elements. Although we already used a soft cut-off value, however, an even lower limit might be used for microbial genomes such as those analyzed here. Second, the 17 ncRNAs detected here in MED4, SS120, MIT 9313 and WH8102 are only a part of the total ncRNA population present in these species. Thus, our data indicate that it is very likely that ncRNAs play an important regulatory and structural role in cyanobacteria. Consequently, they deserve more attention in view of the important function these microbes play in the global ecosystem.
Materials and methods
Cultivation of cyanobacteria
Cultures of Prochlorococcus and Synechococcus were grown in artificial sea water medium (Prochlorococcus MED4, NATL2A-MIT and Synechococcus WH 7803, RS9906, WH 8102) [51], or based on Atlantic seawater in PRO99 media (SS120, MIT 9313, MIT 9312, MIT 9211, MIT 9215) [52] under 18 (MED4, MIT 9312, MIT 9215, WH 7803, RS9906 and WH 8102) or 10 (all other strains) μmol quanta m-2s-1 white light at 23°C in a 12 h day-12 h night cycle.
Prochlorococcus MED4 was subjected to various environmental perturbations by depletion of nitrate, phosphate, iron in artificial seawater; a shift from approximately 10 μmol quanta m-2s-1 white light into darkness or into 10 μmol quanta m-2s-1 blue light or into 50 μmol quanta m-2s-1 daylight as high light condition, or the addition of DCMU to a final concentration of 2 μM for the inhibition of photosynthetic electron transport and the induction of severe oxidative stress; as well as temperature shifts to 15°C or 30°C. MED4 cultures were concentrated ten-fold by centrifugation for 10 minutes at 9,000 rpm at 15°C to 20°C and cell pellets were washed once with the corresponding depleted media if necessary. The concentrated cultures were incubated for 3 h at the respective condition.
RNA analysis
Total RNA was isolated as previously described [53] but with modified lysis conditions for MIT 9313 and WH 8102 as these strains gave poor RNA yields using the standard procedure. The resuspended cells from these strains were homogenized in Z6 buffer [54] by several freeze-thaw cycles using liquid nitrogen over a time of 30 minutes, followed by the addition of one volume of acidic phenol and incubation at 60°C for another 30 minutes. Total RNA was separated in 10% polyacrylamide-urea gels. Polyacrylamide gels were stained with ethidium bromide (0.3 μg/l) in 1 × TBE buffer [55], rinsed with water and analyzed with a Lumi-Imager F1 system (Roche, Mannheim, Germany). Transcript sizes were determined by correlation to MspI-digested DNA of plasmid puc19. Mapping of RNA 5' ends was performed by rapid amplification of cDNA ends as described [24]. We verified in three different ways that the same amounts of RNA samples were loaded in Northern blots: first, by measurement of RNA concentrations; second, by direct comparison of rRNA band intensities after staining by ethidium bromide; and third, by control hybridizations using the 5S rRNA as an internal standard.
To determine RNA stability, cells were treated with rifampicin (200 μg/ml; SIGMA, Munich, Germany) and filtered rapidly (within 60 s) through Supor 0.45 μm membrane filters (PALL, Dreieich, Germany) at different time points after treatment, transferred in resuspension buffer (10 mM NaOAc, pH 4.5, 200 mM sucrose, 5 mM EDTA) and frozen in liquid nitrogen. RNA was isolated by dissolving the filter in acidic phenol at 60°C followed by standard phenol-chloroform extraction as described above. Gel-separated RNAs were electroblotted to Hybond-N+ membranes (Amersham, Freiburg, Germany). Following prehybridization for at least 10 minutes in 50% deionized formamide, 7% SDS, 250 mM NaCl and 120 mM Na(PO4) pH 7.2 at 45°C, oligonucleotide probes labelled by polynucleotide kinase with 30 μCi γ32P-ATP were added and hybridized at 52°C for at least 4 h (except for the probes designed for Yfr2 through Yfr5, which were hybridized at 45°C and washed at 40°C). All DNA oligonucleotides are listed in the Additional data Table S3. The membranes were washed in 2 × SSC (3 M NaCl, 0.3 M sodium citrate, pH 7.0) [55], 1% SDS at 45°C for 10 minutes; 1 × SSC, 0.5% SDS at 45°C for 5 min; and briefly in 0.1 × SSC, 0.1% SDS at ambient temperature. Signals were detected and analyzed on a Personal Molecular Imager FX system with Quantity One software (BIO-RAD, Munich, Germany).
Computational methods
To identify candidates for our experimental investigations, we took a comparative computational approach that was based on sequence and structure conservation and used the program ALIFOLDZ [33]. The genome sequences of Prochlorococcus SS120, MED4, MIT 9313 and Synechococcus WH 8102 were used in the versions given in Additional data file 2 (Table S4). A summary of the computational screening is given in Figure 2 and a complete list of parameters is available in Additional data file 2 (Table S5).
We assumed that homologous RNA structures would show a reasonable degree of conservation on the sequence level for the given set of genomes. BLASTN (Version 2.2.8 [56]) was used to screen for local sequence conservations within intergenic spacer regions (IGRs) longer than 49 bp. These were defined as those regions not overlapping any annotated CDS, rRNA, tRNA or misc_RNA feature (primary tags according to EMBL feature table definition [57]) on either strand. An overview of some characteristics of the intergenic sequences is given in Additional data file 2 (Table S4). Because sequence conservation concerns both DNA strands and because the local alignment was done asymmetrically (e.g. MED4 IGRs were aligned versus MIT 9313 IGRs, but not vice versa; Figure 2B), all hit sequences were reverse complemented.
ALIFOLDZ shows increased sensitivity with the number of aligned sequences [33]. Thus, to take advantage of a multi-genome comparison, we transformed the pairwise sequence alignments into multi-sequence clusters via single-linkage clustering. Before proceeding to single-linkage clustering, redundancy was reduced by unifying those hits from each genome that showed a maximum reciprocal overlap of 85% or greater. The reduced sequence set was used as both query and subject set in another local alignment step (BLASTN considering only the query strand as possible subject strand). Sequences that produced a significant blast hit (E-value ≤ 10-10) for a given query were collected into initial clusters. These were unified if they contained at least one common sequence. The procedure produced a total of 310 clusters plus 310 clusters with the reverse complement of these sequences. Candidate sequences that overlap less than the previous coverage cut-off of 85% but are long enough to produce significant BLASTN hits can result in duplicate sequences within clusters. These may negatively affect the alignment and scoring. Therefore, these sequences were merged within each individual cluster using a less restrictive reciprocal coverage cut-off (≥10%).
Finally, each cluster was aligned using CLUSTALW (Version 1.81, default parameters) [58] and the resulting alignments were scored by ALIFOLDZ. Also, single sequence clusters were scored by ALIFOLDZ (by normalized minimum fold energy). As the scoring method is, besides any biological limitations, sensitive to the number of sequences in the alignment, we considered the Z-score cut-off of -4 used by Washietl and Hofacker [33] as a soft cut-off for both alignments and single sequences. For all structure computations, folding temperatures were set to 24°C, which is the approximate habitat temperature of the marine cyanobacteria studied here [7].
Despite any structural conservation, any RNA in principle may encode for a peptide. The necessary reading frame as defined in this analysis consisted of at least ten consecutive codons starting with either of the possible start codons ATG, GTG, TTG or ATT and finishing with TAA, TAG or TGA. If a reading frame was present, the possible conservation of the encoded peptide sequence amongst other cyanobacteria was evaluated by alignments. Only in the case of a conserved open reading frame did we consider the RNA to be coding.
If not indicated otherwise, all individual secondary structure predictions were done using MFOLD [59].
Additional data files
The following additional data are available with the online version of this article. Additional data file 1 includes figures showing the determination of half-lifes for several ncRNAs (Figure S9) and the composition of cluster 53 (Figure S10). Additional data file 2 includes tables listing the genome versions used in this study and details of intergenic regions (Table S4), the parameters for the initial local alignment of intergenic spacer regions, the clustering step and ALIFOLDZ (Table S5) and the sequences of oligonucleotides used in this study (Table S3). Furthermore, detailed information on all clusters predicted by our method including the positions of all sequences is available online [34].
Supplementary Material
Additional data file 1
Figures showing the determination of half-lifes for several ncRNAs (Figure S9) and the composition of cluster 53 (Figure S10).
Click here for file
Additional data file 2
Tables listing the genome versions used in this study and details of intergenic regions (Table S4), the parameters for the initial local alignment of intergenic spacer regions, the clustering step and ALIFOLDZ (Table S5) and the sequences of oligonucleotides used in this study (Table S3).
Click here for file
Acknowledgements
Supported by grants from the European Union (MARGENES, QLRT-2001-01226; Marine Genomics Europe, GOCE-CT-2004-505403) to W.R.H. and by an EMBO long-term fellowship to J.V.. We thank Carolin Adams for careful technical assistance, Alice Boit for discussion of RNA structural motifs and Martin Meixner of Molecular Biology Systems for sequencing a long chain of RACE fragments.
Figures and Tables
Figure 1 Small RNAs in marine Cyanobacteria. About 10 μg of total RNA from Prochlorococcus strains MIT 9313 (MIT), SS120 (SS1) and MED4 (MED) and from Synechococcus sp. WH 8102 (WH8) was analyzed by staining a 10% polyacrylamide gel with ethidium bromide (center) and by Northern blot hybridization with DNA-oligonucleotides directed against known RNA molecules such as scRNA (ffs gene product), the separate 5' and 3' ends of tmRNA and, as controls, tRNASerin and 5S rRNA. Two distinct precursors of the 5S rRNA were detected. Selected bands have been labeled by arrows in the hybridization and in the gel picture and their sizes (nt, nucleotides) are indicated.
Figure 2 Pipeline for comparative prediction of non-coding RNAs. (a) Intergenic sequences (IGRs) longer than 49 base-pairs were gathered from four Prochlorococcus and Synechococcus genomes and locally aligned using BLASTN. An overview of the intergenic sequences is given in Additional data file 2 (Table S4). Because of the initial asymmetric local alignment using BLASTN (see Figure 2b for a summary of significant BLASTN hits between the strains Prochlorococcus MED4 (MED), MIT 9313 (MIT), SS120 (SS) and Synechococcus WH 8102 (WH)), all candidate sequences were reverse-complemented. Redundancy in this data set was reduced by unifying those hits from each genome that showed a reciprocal overlap of 85% or greater. This candidate set was used as both query and subject in another local alignment step (BLASTN considering only the query strand as possible subject strand). Sequences that directly produced a significant blast hit (E-value ≤ 10-10), or were connected by a chain of such hits, were gathered into clusters ('single-linkage clustering'). Both genome strands were screened; thus, the pipeline produced 310 pairs of clusters in both forward and reverse complementary orientation. After an additional unification step of overlapping sequences within each cluster, the resulting clusters and their complement clusters were scored using ALIFOLDZ [33]. (b) The number of BLASTN high-scoring segment pairs for each query and subject combination of intergenic regions is given for a BLASTN E-value cut-off of 10-5 and after import of high-scoring segment pairs with an E-value of 10-10 or lower (in parentheses). MIT, Prochlorococcus strain MIT 9313; SS, Prochlorococcus strain SS120; WH, Synechococcus sp. WH 8102, MED, Prochlorococcus strain MED4.
Figure 3 Experimental screen for the presence of an RNA-coding gene in the guaB-trxA intergenic region. (a) Sequence alignment of the guaB-trxA (guaB: sequence not shown, located upstream of yfr1) intergenic region visualises the conserved yfr1 gene labeled by the bar above the alignment and its transcriptional initiation site in three of the analyzed strains (MED, MED4; MIT, Prochlorococcus strain MIT 9313; WH8, Synechococcus sp. WH 8102) but not in Prochlorococcus strain SS120 (SS1). Transcriptional initiation sites (TIS) and the deduced -10 elements are indicated. (b) Northern blots show a signal for Yfr1 at a size of 54, 56 and 57 nucleotides (nt) for MED4, WH 8102 and MIT 9313, respectively. No signal with RNA from SS120 confirms the absence of this gene in this strain, as was predicted from the sequence data. (c) Predicted secondary structures of Yfr1 in MED4, MIT 9313 and WH 8102 by MFOLD [59].
Figure 4 Test of transcript accumulation of Yfr1-7 from MED4 (MED) under different conditions. The left side shows the Northern hybridizations for which the following conditions were used: nutrient depletion (phosphate (P-), nitrogen (N-), iron (Fe-)); blue light for three hours (3 h); controls under blue (Blue), white (White) and no light (Dark); oxidative stress mediated by the application of 3-(3,4-dichlorophenyl)-1,1-N-N'-dimethylurea (DCMU); low (15°C) and high (30°C) temperatures; and high light intensity (50 μE). For comparison, 5S rRNA was hybridized as an internal standard and the mRNA of gene PMM3822n which, with a length of approximately 250 nucleotides, was taken as an example for a small mRNA. Additional controls by quantitative RT-PCR for the genes isiB (Fe), glnA (N), pstS (P) and hli8 (high light) [data not shown] were carried out to confirm the effects of nutrient depletion or high light. The amounts of these mRNAs were enhanced by a factor of 79.7 (isiB), 5.8 (glnA), 2.8 (hli8) and 4.0 (pstS) under the respective treatment compared to standard conditions (data not shown). Yfr6 shows an inconstant signal; for example, at cold, blue/white light, N-, Yfr2 to Yfr5 were hybridized with the consensus oligonucleotide y_gen (Figure 5). The band intensities were quantified and normalized to the amount of 5S rRNA as an internal standard (right).
Figure 5 Comparison of Yfr2, Yfr3, Yfr4 and Yfr5 from MED4. (a) Sequence comparison of the yfr2 through yfr5 coding regions of MED4. Transcriptional initiation sites (TIS) and the deduced -10 elements are indicated. The location of specific oligonucleotide probes y2aM, y3aM, y4aM and y5aM used in Figure 5b and in 5' RACE and of the y_gen consensus probe used in Figure 4 is indicated by the lines with black diamonds on the ends on top of the alignment. (b) Signals for the four individual non-coding RNAs (ncRNAs) were detected in Northern blots using probes y2aM, y3aM, y4aM and y5aM. These probes have a minimum of five mismatches to their non-target ncRNAs, making cross-hybridizations impossible. The numbers indicate transcript lengths in nucleotides. (c) Prediction of secondary structure of MED4 Yfr2 by MFOLD [59].
Figure 6 Characterization of a gene encoding Yfr6. (a) Sequence alignment of the region containing the transcriptional initiation site (TIS) and the first 97 transcribed nucleotides of Yfr6 from MED4 and SS120. The alignment begins with the TATA element (in red) preceding the mapped first transcribed guanidine (labelled by an arrow). (b) In Northern blots, a signal for the predicted Yfr6 was detected for 244 and 239 nucleotides in total RNA from MED4 (MED) and SS120 (SS1), respectively, but not in RNA from WH 8102 (WH8) and MIT 9313 (MIT). (c) Comparison of Yfr6 secondary structures using ConStruct version 3.0a [60] The base pairing probability is colour-coded from light yellow (low) to red (high). Missing positions are indicated by a dash. The predicted RNA structures were obtained by RNAfold at 24°C. Both sequences were equally weighted (1.0). The consensus was calculated based on the predicted optimum structures. Default parameters were used for all other options.
Figure 7 Characterization of Yfr7. (a) Sequence alignment of Yfr7 from four marine cyanobacteria. The 5' end transcriptional initiation site (TIS) was mapped for Yfr7 from MED4 and SS120. (b) In Northern blots, a signal for the predicted Yfr7 was detected with RNA from all four strains and seven additional strains from the marine cyanobacterial radiation: Prochlorococcus NATL2A (N2), MED4 (MED), SS120 (SS1), MIT 9313 (MIT), MIT 9312 (9312), MIT 9211 (9211), MIT 9215 (9215) and Synechococcus WH 8102 (WH8), WH 7803 (7803), WH 8020 (8020), RS9906 (9906). The high-light-adapted Prochlorococcus strains are labelled in red, low-light-adapted strains in blue, and Synechococcus strains are colour-coded in green; M indicates the marker lane. Numbers indicate lengths of RNA markers in nucleotides. (c) Prediction of secondary structure of the Synechococcus WH 8102 ncRNA Yfr7.
Figure 8 A putative gene encoding the RNA chaperone Hfq can be predicted in two of the four marine cyanobacteria investigated here. (a) The dapF-leuS intergenic region in Synechococcus WH 8102 (WH8) and Prochlorococcus MIT 9313 (MIT) is, at 298 and 297 nucleotides, respectively, relatively long and contains a short reading frame for a putative hfq gene. In Prochlorococcus SS120 (SS1) and MED4 (MED), this region is only 123 and 108 nucleotides, respectively. (b) Sequence comparison of putative Hfq proteins from the three cyanobacteria Synechocystis PCC 6803 (ssr3341 gene), Synechococcus WH 8102 (WH8) and Prochlorococcus MIT 9313 (MIT). Hydrophobic residues within the Sm1 and Sm2 motifs [29] are indicated by an H.
Table 1 List of high scoring clusters
CLID Sequence number Strain Alignment length Z Z rev Exp Comment Reference
MED SS1 MIT WH8
194 3 3 - - - 201 -6.28 -12.94 + yfr2, yfr3, yfr4 This paper
5 3 - 1 1 1 345 -7.58 -10.18 NT rplCD operon leader, corresponds to Escherichia coli S10 r-operon [61, 62]
92 5 1 1 1 2 756 -4.47 -9.9 NT rrn operon leader [63, 64]
112 2 2 - - - 1129 -8.15 -9.15 NT Reciprocal coverage of 7.9%, artifact due to low-complexity sequences -
229 2 - - 1 1 161 -7.98 -4.90 NT Hghly similar sequences, putative ncRNAs This paper
227 2 - - 1 1 229 -7.38 -7.32 NT rplJ operon leader, corresponds to E. coli β r-operon [65]
84 1 - 1 - - 122 -6.27 -5.54 + Yfr2 This paper
101 3 - 1 1 1 152 -5.77 -5.61 NT Putative Cobalamin riboswitch [39, 66]
226 2 - - 1 1 142 -5.29 -5.28 NT Possible bi-directional terminator of the rplKAJL operon Predicted by TransTerm [67]
9 6 - 1 1 4 397 -4.38 -4.95 NT No conserved position, no significant BLASTN hit to MED4 This paper
51 2 1 1 - - 146 -0.84 -4.92 + yfr7 This paper
53 9 2 2 1 4 697 -3.26 -4.59 + yfr6 in MED4 and SS120 and a subgroup of 5' UTR regions to annotated genes and putative unannotated genes in all four strains This paper
245 2 - - 1 1 259 -3.7 -4.53 - rpoBC operon leader, corresponds to E. coli attenuator separating the rpl genes from rpoBC in the rplKAJLrpoBC gene cluster [37, 68]
217 1 1 - - - 153 -1.63 -4.28 - Located between genes for a two-component sensor histidine kinase and a conserved hypothetical protein This paper
87 2 - - 1 1 336 -4.24 -3.64 NT Region upstream of the rbcLS cluster containing conserved promoter [51]
228 2 - - 1 1 106 -0.67 -4.00 NT Rpl11 operon leader, corresponds to E. coli L11 r-operon [69, 70]
257 2 - - 1 1 176 -3.42 -3.97 + Yfr1 This paper
2 3 - 1 1 1 197 -3.93 -2.94 NT Putative TPP riboswitch in front of thiC [38]
RNA elements were predicted according to the scheme shown in Figure 2. The total number of sequences in each cluster and the distribution within the four compared genomes plus the total alignment length are given. The elements are ordered according to the lowest score in either forward (Z) or reverse (Z rev) orientation (in bold letters). The lower the Z-score the higher the support for structural conservation. Exp (experimental testing): +, tested positively by Northern hybridisation; NT, not tested. The cluster identities (CLID) were also used in Table 2. For further details and exact positions of sequences see Table 2 and [34].
Table 2 Summary of identified ncRNA genes in Prochlorococcus MED4 and their orthologues in three related strains of marine cyanobacteria
Strain RNA gene name CLID Coordinates of RNA gene Length of RNA in nucleotides Adjacent protein-coding genes Orientation
MED4 yfr1 257 Complement (1000744..1000797) 54 trxA and guaB ← ← ←
yfr2 194 346828..346921 94 PMM0363 and PMM0364 → → →
yfr3 194 654511..654604 95 PMM0686 and PMM0687 → → ←
yfr4 194 383389..383483 94 PMM0404 and phdC ← → →
yfr5 NP Complement (972088..972176) 89 PMM1027 and PMM1028 → ← →
yfr6 53 Complement (627729..627972 244 PMM0659 and PMM0660 → ← ←
yfr7 51 652625..652844 220 purK and PMM0684 → → →
SS120 yfr2 84 Complement (556612..556701) 90 rpsU and Pro0591 ← ← ←
yfr6 53 Complement (923780..924018) 239 Pro1007 and purK → ← ←
yfr7 51 Complement (924466..924640) 175 Pro1007 and purK → ← ←
MIT 9313 yfr1 257 1220973..1221029 57 guaB and trxA → → →
yfr2 NP Complement (1667304..1667390) 87 PMT1567 and PMT1568 → ← ←
yfr7 NP 727045..727219 (complementary to PMT0670) 175 purK and PMT0671 → → ←
WH 8102 yfr1 257 Complement (706826..706881) 56 trxA and guaB ← ← ←
yfr2 NP 1127972..1128056 NT Overlapping SYNW1139
yfr3 NP 1131773..1131856 NT SYNW1140 and SYNW1141 → → ←
yfr7 NP Complement (1302885..1303058) (complementary to SYNW1307) 174 SYNW1306 and purK → ← ←
The genome positions and names of protein coding genes refer to the genome versions indicated in the Additional data file 2 (Table S4). The cluster identifier (CLID) is identical to that used in Table 1. NP, not directly predicted by the pipeline; NT, not experimentally tested.
==== Refs
Schopf JW Microfossils of the early Archean Apex chert: new evidence of the antiquity of life. Science 1993 260 640 646 11539831
Chisholm SW Olson RJ Zettler ER Waterbury JB Goericke R Welschmeyer N A novel free-living prochlorophyte abundant in the oceanic euphotic zone. Nature 1988 334 340 343 10.1038/334340a0
Chisholm SW Frankel SL Goericke R Olson RJ Palenik B Waterbury JB West-Johnsrud L Zettler ER Prochlorococcus marinus nov. gen. nov. sp.: an oxyphototrophic marine prokaryote containing divinyl chlorophyll a and b. Arch Microbiol 1992 157 297 300 10.1007/BF00245165
LaRoche J van der Staay GW Partensky F Ducret A Aebersold R Li R Golden SS Hiller RG Wrench PM Larkum AW Green BR Independent evolution of the prochlorophyte and green plant chlorophyll a/b light-harvesting proteins. Proc Natl Acad Sci USA 1996 93 15244 15248 8986795 10.1073/pnas.93.26.15244
Chen M Hiller RG Howe CJ Larkum AW Unique origin and lateral transfer of prokaryotic chlorophyll-b and chlorophyll-d light-harvesting systems. Mol Biol Evol 2005 22 21 28 15356274 10.1093/molbev/msh250
Urbach E Robertson DL Chisholm SW Multiple evolutionary origins of prochlorophytes within the cyanobacterial radiation. Nature 1992 355 267 270 1731225 10.1038/355267a0
Partensky F Hess WR Vaulot D Prochlorococcus, a marine photosynthetic prokaryote of global significance. Microbiol Mol Biol Rev 1999 63 106 127 10066832
Moore LR Rocap G Chisholm SW Physiology and molecular phylogeny of coexisting Prochlorococcus ecotypes. Nature 1998 393 464 467 9624000 10.1038/30965
Moore LR Chisholm SW Photophysiology of the marine cyanobacterium Prochlorococcus: Ecotypic differences among cultured isolates. Limnol Oceanogr 1999 44 628 638
Dufresne A Salanoubat M Partensky F Artiguenave F Axmann I Barbe V Duprat S Galperin M Koonin EV Le Gall F Genome sequence of the cyanobacterium Prochlorococcus marinus SS120, a nearly minimal oxyphototrophic genome. Proc Natl Acad Sci USA 2003 100 10020 10025 12917486 10.1073/pnas.1733211100
Palenik B Brahamsha B Larimer FW Land M Hauser L Chain P Lamerdin J Regala R Allen RE McCarren J The genome of a motile marine Synechococcus. Nature 2003 424 1037 1042 12917641 10.1038/nature01943
Rocap G Larimer FW Lamerdin J Malfatti S Chain P Ahlgren NA Arellano A Coleman M Hauser L Hess WR Genome divergence in two Prochlorococcus ecotypes reflects oceanic niche differentiation. Nature 2003 424 1042 1047 12917642 10.1038/nature01947
Dufresne A Garczarek L Partensky F Accelerated evolution associated with genome reduction in a free-living prokaryote. Genome Biol 2005 6 R14 15693943 10.1186/gb-2005-6-2-r14
Ohmori M Ikeuchi M Sato N Wolk P Kaneko T Ogawa T Kanehisa M Goto S Kawashima S Okamoto S Characterization of genes encoding multi-domain proteins in the genome of the filamentous nitrogen-fixing Cyanobacterium Anabaena sp. strain PCC 7120. DNA Res 2001 8 271 284 11858227
Mizuno T Kaneko T Tabata S Compilation of all genes encoding bacterial two-component signal transducers in the genome of the cyanobacterium, Synechocystis sp. strain PCC 6803. DNA Res 1996 3 407 414 9097043
Gottesman S The small RNA regulators of Escherichia coli: roles and mechanisms. Annu Rev Microbiol 2004 58 303 328 15487940 10.1146/annurev.micro.58.030603.123841
Wagner EG Simons RW Antisense RNA control in bacteria, phages, and plasmids. Annu Rev Microbiol 1994 48 713 742 7826024 10.1146/annurev.mi.48.100194.003433
Lankenau S Corces VG Lankenau DH The Drosophila micropia retrotransposon encodes a testis-specific antisense RNA complementary to reverse transcriptase. Mol Cell Biol 1994 14 1764 1775 7509447
Morfeldt E Taylor D von Gabain A Arvidson S Activation of alpha-toxin translation in Staphylococcus aureus by the trans-encoded antisense RNA, RNAIII. EMBO J 1995 14 4569 4577 7556100
Lenz DH Mok KC Lilley BN Kulkarni RV Wingreen NS Bassler BL The small RNA chaperone Hfq and multiple small RNAs control quorum sensing in Vibrio harveyi and Vibrio cholerae. Cell 2004 118 69 82 15242645 10.1016/j.cell.2004.06.009
Sledjeski DD Gupta A Gottesman S The small RNA, DsrA, is essential for the low temperature expression of RpoS during exponential growth in Escherichia coli. EMBO J 1996 15 3993 4000 8670904
Altuvia S Weinstein-Fischer D Zhang A Postow L Storz G A small, stable RNA induced by oxidative stress: role as a pleiotropic regulator and antimutator. Cell 1997 90 43 53 9230301 10.1016/S0092-8674(00)80312-8
Wassarman KM Repoila F Rosenow C Storz G Gottesman S Identification of novel small RNAs using comparative genomics and microarrays. Genes Dev 2001 15 1637 1651 11445539 10.1101/gad.901001
Argaman L Hershberg R Vogel J Bejerano G Wagner EG Margalit H Altuvia S Novel small RNA-encoding genes in the intergenic regions of Escherichia coli. Curr Biol 2001 11 941 950 11448770 10.1016/S0960-9822(01)00270-6
Rivas E Klein RJ Jones TA Eddy SR Computational identification of noncoding RNAs in E. coli by comparative genomics. Curr Biol 2001 11 1369 1373 11553332 10.1016/S0960-9822(01)00401-8
Eddy SR Non-coding RNA genes and the modern RNA world. Nat Rev Genet 2001 2 919 929 11733745 10.1038/35103511
Vogel J Bartels V Tang TH Churakov G Slagter-Jäger JG Hüttenhofer A Wagner EGH RNomics in Escherichia coli detects new sRNA species and indicates parallel transcriptional output in bacteria. Nucleic Acids Res 2003 31 6435 6443 14602901 10.1093/nar/gkg867
Zhang A Wassarman KM Rosenow C Tjaden BC Storz G Gottesman S Global analysis of small RNA and mRNA targets of Hfq. Mol Microbiol 2003 50 1111 1124 14622403 10.1046/j.1365-2958.2003.03734.x
Valentin-Hansen P Eriksen M Udesen C The bacterial Sm-like protein Hfq: a key player in RNA transactions. Mol Microbiol 2004 51 1525 1533 15009882 10.1111/j.1365-2958.2003.03935.x
Gaudin C Zhou X Williams KP Felden B Two-piece tmRNA in cyanobacteria and its structural analysis. Nucleic Acids Res 2002 30 2018 2024 11972341 10.1093/nar/30.9.2018
Keiler KC Shapiro L Williams KP tmRNAs that encode proteolysis-inducing tags are found in all known bacterial genomes: A two-piece tmRNA functions in Caulobacter. Proc Natl Acad Sci USA 2000 97 7778 7783 10884408 10.1073/pnas.97.14.7778
Washietl S Hofacker IL Stadler PF Fast and reliable prediction of noncoding RNAs. Proc Natl Acad Sci USA 2005 102 2454 2459 15665081 10.1073/pnas.0409169102
Washietl S Hofacker IL Consensus folding of aligned sequences as a new measure for the detection of functional RNAs by comparative genomics. J Mol Biol 2004 342 19 30 15313604 10.1016/j.jmb.2004.07.018
Complete Results for Non-coding RNA Screening of Cyanobacteria
Zengel JM Lindahl L Diverse mechanisms for regulating ribosomal protein synthesis in Escherichia coli. Prog Nucleic Acid Res Mol Biol 1994 47 331 370 7517053
Lindahl L Zengel JM Ribosomal genes in Escherichia coli. Annu Rev Genet 1986 20 297 326 2434021 10.1146/annurev.ge.20.120186.001501
Barry G Squires C Squires CL Attenuation and processing of RNA from the rplJL-rpoBC transcription unit of Escherichia coli. Proc Natl Acad Sci USA 1980 77 3331 3335 6158044
Winkler W Nahvi A Breaker RR Thiamine derivatives bind messenger RNAs directly to regulate bacterial gene expression. Nature 2002 419 952 956 12410317 10.1038/nature01145
Nahvi A Barrick JE Breaker RR Coenzyme B12 riboswitches are widespread genetic control elements in prokaryotes. Nucleic Acids Res 2004 32 143 150 14704351 10.1093/nar/gkh167
Tjaden B Saxena RM Stolyar S Haynor DR Kolker E Rosenow C Transcriptome analysis of Escherichia coli using high-density oligonucleotide probe arrays. Nucleic Acids Res 2002 30 3732 3738 12202758 10.1093/nar/gkf505
Reyes JC Muro-Pastor MI Florencio FJ Transcription of glutamine synthetase genes (glnA and glnN) from the cyanobacterium Synechocystis sp. strain PCC 6803 is differently regulated in response to nitrogen availability. J Bacteriol 1997 179 2678 2689 9098067
Golden SS Brusslan J Haselkorn R Expression of a family of psbA genes encoding a photosystem II polypeptide in the cyanobacterium Anacystis nidulans R2. EMBO J 1986 5 2789 2798 3098559
Tegmark K Morfeldt E Arvidson S Regulation of agr-dependent virulence genes in Staphylococcus aureus by RNAIII from coagulase-negative staphylococci. J Bacteriol 1998 180 3181 3186 9620969
Zago MA Dennis PP Omer AD The expanding world of small RNAs in the hyperthermophilic archaeon Sulfolobus solfataricus. Mol Microbiol 2005 55 1812 1828 15752202 10.1111/j.1365-2958.2005.04505.x
Fuller NJ Marie D Partensky F Vaulot D Post AF Scanlan DJ Clade-specific 16S ribosomal DNA oligonucleotides reveal the predominance of a single marine Synechococcus clade throughout a stratified water column in the Red Sea. Appl Environ Microbiol 2003 69 2430 2443 12732508 10.1128/AEM.69.5.2430-2443.2003
Watanabe T Sugiura M Sugita M A novel small stable RNA, 6Sa RNA, from the cyanobacterium Synechococcus sp. strain PCC6301. FEBS Lett 1997 416 302 306 9373174 10.1016/S0014-5793(97)01237-4
Barrick JE Sudarsan N Weinberg Z Ruzzo WL Breaker RR 6S RNA is a widespread regulator of eubacterial RNA polymerase that resembles an open promoter. RNA 2005 11 774 784 15811922 10.1261/rna.7286705
Damerval T Houmard J Guglielmi G Csiszar K Tandeau de Marsac N A developmentally regulated gvpABC operon is involved in the formation of gas vesicles in the cyanobacterium Calothrix 7601. Gene 1987 54 83 92 3111941 10.1016/0378-1119(87)90350-7
Synechococcus sp. Genome Analysis
Trotochaud AE Wassarman KM 6S RNA function enhances long-term cell survival. J Bacteriol 2004 186 4978 4985 15262935 10.1128/JB.186.15.4978-4985.2004
Vogel J Axmann IM Herzel H Hess WR Experimental and computational analysis of transcriptional start sites in the cyanobacterium Prochlorococcus MED4. Nucleic Acids Res 2003 31 2890 2899 12771216 10.1093/nar/gkg398
Chisholm lab protocols
Garcia-Fernandez JM Hess WR Houmard J Partensky F Expression of the psbA gene in the marine oxyphotobacteria Prochlorococcus spp. Arch Biochem Biophys 1998 359 17 23 9799555 10.1006/abbi.1998.0862
Logemann J Schell J Willmitzer L Improved method for the isolation of RNA from plant tissues. Anal Biochem 1987 163 16 20 2441623 10.1016/0003-2697(87)90086-8
Sambrook J Fritsch EF Maniatis T Molecular Cloning A Laboratory Manual 1989 Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press
Altschul SF Gish W Miller W Myers EW Lipman DJ Basic local alignment search tool. J Mol Biol 1990 215 403 410 2231712 10.1006/jmbi.1990.9999
The DDBJ/EMBL/GenBank Feature Table: Definition
Thompson JD Higgins DG Gibson TS ClustalW: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res 1994 22 4673 4680 7984417
Zuker M Mfold web server for nucleic acid folding and hybridization prediction. Nucleic Acids Res 2003 31 3406 3415 12824337 10.1093/nar/gkg595
Luck R Graf S Steger G ConStruct: a tool for thermodynamic controlled prediction of conserved secondary structure. Nucleic Acids Res 1999 27 4208 4217 10518612 10.1093/nar/27.21.4208
Sanangelantoni AM Tiboni O The chromosomal location of genes for elongation factor Tu and ribosomal protein S10 in the cyanobacterium Spirulina platensis provides clues to the ancestral organization of the str and S10 operons in prokaryotes. J Gen Microbiol 1993 139 2579 2584 8277243
Zengel JM Mueckl D Lindahl L Protein L4 of the E. coli ribosome regulates an eleven gene r protein operon. Cell 1980 21 523 535 6157482 10.1016/0092-8674(80)90490-0
Li SC Squires CL Squires C Antitermination of E. coli rRNA transcription is caused by a control region segment containing lambda nut-like sequences. Cell 1984 38 851 860 6091902 10.1016/0092-8674(84)90280-0
Condon C Squires C Squires CL Control of rRNA transcription in Escherichia coli. Microbiol Rev 1995 59 623 645 8531889
Christensen T Johnsen M Fiil NP Friesen JD RNA secondary structure and translation inhibition: analysis of mutants in the rplJ leader. EMBO J 1984 3 1609 1612 6378628
Vitreschak AG Rodionov DA Mironov AA Gelfand MS Regulation of the vitamin B12 metabolism and transport in bacteria by a conserved RNA structural element. RNA 2003 9 1084 1097 12923257 10.1261/rna.5710303
Ermolaeva MD Khalak HG White O Smith HO Salzberg SL Prediction of transcription terminators in bacterial genomes. J Mol Biol 2000 301 27 33 10926490 10.1006/jmbi.2000.3836
Steward KL Linn T Transcription frequency modulates the efficiency of an attenuator preceding the rpoBC RNA polymerase genes of Escherichia coli: possible autogenous control. Nucleic Acids Res 1992 20 4773 4779 1408790
Branlant C Krol A Machatt A Ebel JP The secondary structure of the protein L1 binding region of ribosomal 23S RNA. Homologies with putative secondary structures of the L11 mRNA and of a region of mitochondrial 16S rRNA. Nucleic Acids Res 1981 9 293 307 7010313
Huynen MA Bork P Measuring genome evolution. Proc Natl Acad Sci USA 1998 95 5849 5856 9600883 10.1073/pnas.95.11.5849
|
16168080
|
PMC1242208
|
CC BY
|
2021-01-04 16:05:42
|
no
|
Genome Biol. 2005 Aug 17; 6(9):R73
|
utf-8
|
Genome Biol
| 2,005 |
10.1186/gb-2005-6-9-r73
|
oa_comm
|
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-9-r741616808110.1186/gb-2005-6-9-r74ResearchDifferential gene expression in anatomical compartments of the human eye Diehn Jennifer J [email protected] Maximilian [email protected] Michael F [email protected] Patrick O [email protected] Department of Ophthalmology, Stanford University School of Medicine, Stanford, CA 94305, USA2 Department of Biochemistry, Stanford University School of Medicine, Stanford, CA 94305, USA3 Howard Hughes Medical Institute, Stanford University School of Medicine, Stanford, CA 94305, USA4 Department of Ophthalmology, University of California, San Francisco, San Francisco, CA 94143, USA2005 17 8 2005 6 9 R74 R74 10 5 2005 5 7 2005 15 7 2005 Copyright © 2005 Diehn et al.; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
DNA microarrays (representing approximately 30,000 human genes) were used to analyze gene expression in six different human eye compartments, revealing candidate genes for diseases affecting the cornea, lens and retina.
Background
The human eye is composed of multiple compartments, diverse in form, function, and embryologic origin, that work in concert to provide us with our sense of sight. We set out to systematically characterize the global gene expression patterns that specify the distinctive characteristics of the various eye compartments.
Results
We used DNA microarrays representing approximately 30,000 human genes to analyze gene expression in the cornea, lens, iris, ciliary body, retina, and optic nerve. The distinctive patterns of expression in each compartment could be interpreted in relation to the physiology and cellular composition of each tissue. Notably, the sets of genes selectively expressed in the retina and in the lens were particularly large and diverse. Genes with roles in immune defense, particularly complement components, were expressed at especially high levels in the anterior segment tissues. We also found consistent differences between the gene expression patterns of the macula and peripheral retina, paralleling the differences in cell layer densities between these regions. Based on the hypothesis that genes responsible for diseases that affect a particular eye compartment are likely to be selectively expressed in that compartment, we compared our gene expression signatures with genetic mapping studies to identify candidate genes for diseases affecting the cornea, lens, and retina.
Conclusion
Through genome-scale gene expression profiling, we were able to discover distinct gene expression 'signatures' for each eye compartment and identified candidate disease genes that can serve as a reference database for investigating the physiology and pathophysiology of the eye.
==== Body
Background
The human eye is composed of multiple substructures of diverse form, function, and even embryologic origin that work in concert to provide us with our sense of sight. Identifying the global patterns of gene expression that specify the distinctive characteristics of each of the various compartments of the eye is an important step towards understanding how these complex normal tissues function, and how dysfunction leads to disease. The Human Genome sequence [1,2] provides a basis for examining gene expression on a genomic scale, and cDNA microarrays provide an efficient method for analyzing the expression of thousands of genes in parallel. Previous studies have used microarrays to investigate gene expression within normal eye tissues, including cornea [3] and retina [4], as well as within pathological tissues such as glaucomatous optic nerve heads [5], uveal melanomas [6], and aging retina [7].
Analysis of gene expression in the eye has been notoriously difficult because of the technical obstacles associated with extracting sufficient quantities of high quality RNA from the tissues. This is especially true for the lens and cornea, which have relatively few RNA-producing cells when compared to a highly cellular tissue such as retina. Furthermore, pigmented ocular tissues contain melanin, which often co-purifies with RNA and inhibits subsequent enzymatic reactions [8]. Any delay between the patient's death and the harvesting of ocular tissues can also compromise RNA quality and yield. To date, many experiments examining the gene expression profile of particular eye compartments have relied on pooled samples or cell culture in order to obtain adequate amounts of RNA. In contrast to these studies, the experiments described in this paper were performed using a linear amplification procedure [9], which made it possible to examine individual specimens using DNA microarrays, thereby eliminating the potentially confounding effects of pooling multiple donor samples or culturing cells, which can elicit dramatic changes in gene expression based on the cell culture media [10]. We chose an in vitro transcription-based, linear amplification approach because this has previously been shown to reproducibly generate microarray gene expression results that are extremely similar to data generated using unamplified RNA [9,11,12]. Additionally, the amplification process has been shown to selectively and reproducibly 'over-amplify' some low-copy number transcripts, resulting in a larger fraction of the expressed genome that can be reliably measured on DNA microarrays. Importantly, by analyzing individual donor samples on arrays, we can detect variation in the eye compartments of different donors, which will be critical for future studies that examine how gene expression varies between individuals at baseline and also in disease states.
A major goal of this study was to discover how the various eye compartments differ from one another on a molecular level by identifying clusters of differentially expressed genes, or 'gene signatures', characteristic of each eye compartment. We also wanted to investigate how gene expression varies between geographical regions of the retina. Because certain retinal diseases such as retinitis pigmentosa (RP) and age-related macular degeneration (ARMD) preferentially affect a specific retinal region, identification of genes that are differentially expressed in the macula versus peripheral retina may provide valuable clues to the molecular mechanisms underlying these diseases. Recent work using serial analysis of gene expression (SAGE), a method that involves sequencing thousands of transcripts from a given RNA sample, identified several genes that were significantly enriched in either the macula or the periphery [13]. Our cDNA microarray studies confirmed some of these genes, but also significantly added to the catalog of macula-enriched genes. Lastly, because many ophthalmologic diseases preferentially affect a particular eye compartment, our study demonstrates that gene signatures can be combined with gene linkage studies in order to identify candidate disease genes.
Results
To explore relationships among the different eye compartments and among genes expressed in these compartments, we performed hierarchical cluster analysis of both genes and samples [14] using genes that met our selection criteria (see Materials and methods). The display generated through hierarchical clustering analysis is shown in Figure 1a. In this display, relatively high expression levels are indicated by a red color, and relatively low expression levels are represented by a green color; each column represents data from a single tissue sample, and each row represents the series of measurements for a single gene. Tissue samples with similar gene expression patterns are clustered adjacent to one another, and genes with similar expression patterns are clustered together. In our experiments, samples of the same eye compartment from different donors clustered in discrete groups (for example, cornea with cornea, retina with retina), with the only exception being an intermingling of the ciliary body and iris specimens (Figure 1a). The lack of a clear distinction between the expression patterns of the ciliary body and iris may be due to both their shared embryological origin and their close anatomical approximation, resulting in sub-optimal separation during dissection. The division between the retinal samples and all other samples was the most striking. Furthermore, there was a distinct grouping of the various macula specimens, which formed a tightly clustered subgroup among the retinal samples. The expression patterns of the optic nerve samples were most similar to those of the three brain specimens.
Each anatomical compartment of the eye expressed a distinct set of genes that were not expressed, or expressed at much lower levels, in the other eye compartments (Figure 1b). The repertoire of genes specifically expressed in the retina was especially large and diverse (3,727 genes), but we also found a surprisingly large number of transcripts (1,777 genes) expressed predominantly in the lens. To explore the connections between these compartment-enriched genes and phenotypic features of the compartments in which they were expressed, we considered each group of compartment-enriched genes in detail.
Corneal signature
The cornea is a multi-layered structure consisting of an epithelium of stratified squamous cells, a thick stroma of layered collagen fibrils, and an underlying endothelial layer. To provide an effective physical barrier to the outside world, the corneal epithelial cells bind to one another and to the underlying connective tissue through a series of linked structures known collectively as the 'adhesion complex'. As shown in Figure 2a, many genes enriched in the corneal signature encoded proteins that stabilize epithelial sheets and promote cell-cell adhesion, including keratins (KRT5, KRT6B, KRT13, KRT15, KRT16, KRT17, KRT19), laminins (LAMB3, LAMC2), and desmosomal components (DSG1, DSC3, BPAG1).
Other genes highly expressed in the cornea signature encoded proteins that help maintain the shape, transparency, or integrity of the cornea, which serves as the primary refractive element in the eye. Some of the genes encoded proteins specifically expressed by either squamous epithelial cells or fibroblasts, reflecting the histological composition of corneal tissue. For example, the signature included numerous genes that encode collagens (COL5A2, COL6A3, COL12A1, COL17A1), along with the gene for lysyl oxidase (LOX), an enzyme that promotes collagen cross-linking. The gene encoding keratocan (KERA), a proteoglycan involved in maintaining corneal shape in mice knock-out studies [15], and linked to abnormal corneal morphology (keratoconus and cornea plana) in humans, was selectively expressed in corneal tissue, as were the genes encoding lumican (LUM), a keratan sulfate-containing proteoglycan that has been shown to be important for mouse corneal transparency [16], and aquaporin 3 (AQP3), which encodes a water/small solute-transporting molecule. Immunolabeling studies performed on corneas with pseudophakic bullous keratopathy demonstrated increased AQP3 in the superficial epithelial cells, suggesting that AQP3 may be associated with increased fluid accumulation, resulting in the decrease in corneal transparency seen in pseudophakic bullous keratopathy corneas [17]. Modulating genes or proteins involved in corneal shape and transparency could potentially lead to non-invasive treatments for some corneal diseases, which are often only remediable through corneal transplantation.
An intriguing subset of genes in the cornea signature has been studied in tumor metastasis models because these genes encode proteins that regulate cell-cell or cell-matrix interactions (TWIST, MMP10, SERPINB5, THBS1, CEACAM1, C4.4A). For example, TWIST encodes a transcription factor shown to promote metastasis in a murine breast tumor model through the loss of cadherin-mediated cell-cell adhesion [18]. Another corneal signature gene encodes matrix metalloproteinase 10 (MMP10), a protein capable of degrading extracellular matrix components. Overexpression of MMP10 in transfected lymphoma cells has been shown to stimulate invasive activity in vitro and promote thymic lymphoma growth in an in vivo murine model [19]. Various matrix metalloproteinases have been examined for their roles in corneal wound healing (reviewed in [20]), including MMP10, which was identified in migrating epithelial cells in cultured human cornea tissues that were experimentally wounded [21], which may suggest that the process of corneal wound healing may mimic some aspects of tumor biology. Certainly, in both wound healing and cancer, cells undergo rapid proliferation, invade and remodel the extracellular matrix, and migrate to other areas.
Recent microarray investigations identified a gene expression signature related to a wound response in the expression profiles of several common carcinomas, and the presence of this wound healing gene signature predicted an increased risk of metastasis and death in breast, lung, and gastric carcinomas [22,23]. Further research into corneal wound healing may also provide us with a model for better understanding the pathophysiology underlying tumor metastasis because the cornea is exceptionally efficient among human tissues at degrading and remodeling its extracellular matrix, allowing it to heal superficial wounds within hours.
Ciliary body/iris signature
The ciliary body and iris are components of the eye's highly pigmented and vascular layer known as the uveal tract. As might be expected, genes related to pigmentation were a feature of the distinctive expression pattern of these tissues (Figure 2b). These genes encoded enzymes involved in melanogenesis, including tyrosinase (TYR), tyrosinase-related protein 1 (TYRP1), and dopachrome tautomerase (DCT), as well as melanosomal matrix proteins such as SILV and MLANA. Several of the ciliary body/iris signature genes were noteworthy in that their mutation can lead to albinism or hypopigmentation phenotypes, including OA1 (ocular albinism type 1), TYR and TYRP1 (oculocutaneous albinism 1A and 3, respectively), and MLPH (Griscelli syndrome). Investigation of the numerous uncharacterized genes with similar expression patterns to those of pigmentation genes may expand our knowledge about the pigmentation process in eyes and the molecular mechanisms behind hypopigmentation syndromes.
The ciliary body is also responsible for aqueous humor formation and lens accommodation, while the contiguous iris filters light entering the eye by constricting and dilating the muscles around the pupillary opening. Histologically, the ciliary body consists predominately of smooth muscle, but also contains striated muscle (reviewed in [24]). Previous work has demonstrated that contractility of both the ciliary body and the trabecular meshwork is critical in modulating aqueous humor outflow (reviewed in [25]), one of the key determinants of intraocular pressure, along with aqueous humor production and episcleral venous pressure. Muscle-related proteins encoded by genes in the ciliary body/iris cluster included smooth muscle actin (ACTG2), and actin cross-linking proteins such as filamin (FLNC), tropomyosin (TPM2), and tensin (TNS). Other iris/ciliary body signature genes have known roles in myosin phosphorylation (PPP1R12B), sarcolemmal calcium homeostasis (CASQ2), and ATP availability (CKMT2), all of which may contribute to ciliary body/trabecular meshwork contractility.
Both ciliary body and trabecular meshwork contractility, as well as aqueous humor production, have been linked to changes in membrane potential, and membrane channels have been studied extensively in the ciliary body [25-27]. Of note, transcripts encoding an inward-rectifying potassium channel (KCNJ8), not previously identified in the ciliary body, were highly enriched in the ciliary body/iris signature and may warrant further study. The signature also included the gene for adrenergic receptor 2α (ADRA2A), a regulator of aqueous humor production and outflow, and the molecular target of the ocular hypotensive agent brimonidine. Identification of other genes that facilitate aqueous production and outflow may provide additional molecular targets for future glaucoma therapeutics aimed at lowering intraocular pressure, the only modifiable risk factor for the development and progression of glaucoma.
Immune system genes expressed within anterior segment tissues
Genes related to immune defense mechanisms were prominent among the large set of genes selectively expressed in both the ciliary body/iris and corneal tissues. These included genes encoding proteins involved in intracellular antigen processing and transport for eventual surface presentation to immune cells (PSMB8, TAP1), antigen presentation proteins, including HLA class I molecules (HLA-A, HLA-C, HLA-F, and HLA-G) and HLA class II molecules (HLA-DRB1, DRB4, DRB5, DPA1, and DPB1), cytokines involved in the recruitment of monocytes (SCYA3, SCYA4, CD14), and cytokine receptors (IL1R2, IL4R, and IL6R). Several anterior segment-enriched genes encoded proteins with intrinsic antibiotic activity, including defensin (DEFB1) and lysozyme (LYZ), which may protect epithelial surfaces from microbial colonization.
Genes encoding components of the complement cascade, a major arm of the innate immune system, were a particularly prominent feature of the anterior segment signature. Most of the early classical pathway complement genes, including C1 components (C1S, C1QA, C1QG, C1R), C2, and C4b, as well as a component of the late complement cascade (C7), were selectively expressed in both the corneal and ciliary body/iris tissues. In addition, the gene encoding the trigger for the alternative complement pathway, properdin (BF), was highly expressed in these tissues.
To prevent the destructive reactions that could ensue from the daily bombardment of the eye with potentially antigenic stimuli, regulatory mechanisms must counteract the multitude of pro-inflammatory mediators found in the eye. A study by Sohn et al. [28] that examined a number of complement and complement-regulating components in rat eyes suggested that the complement system is continuously active at a low level in the normal eye and is kept in check by regulatory proteins. Indeed, we found that the anterior segment selectively expressed many critical negative regulators of the immune system, especially of the complement cascade. These included SERPING1 and DAF, two genes that encode proteins that limit the production of early complement components, and CD59, which encodes a protein that inhibits the assembly of complement subunits into the membrane attack complex.
The presence of complement activation products in the human eye during infection or inflammation has been previously described [29]. Studies have suggested that the complement pathway contributes to the pathophysiology of uveitis, an inflammatory disease of the uveal tract that is often idiopathic in etiology [30]. In support of this theory, Bardenstein et al. [31] showed that blocking the complement regulator CD59 in the rat eye precipitated massive inflammation in the anterior eye, including intense conjunctival inflammation and iritis. Our evidence that complement pathway components and regulators are highly expressed in anterior segment tissues provides further impetus for investigating their links to ocular disease.
A caution to bear in mind in interpreting these results is that all of our ocular specimens were obtained post-mortem. The expression of the inflammatory genes could therefore reflect, at least in part, changes in the eye that occur after death. Future studies examining gene expression in fresh tissue samples obtained at surgery, such as peripheral iridectomy specimens, should help to further address this issue.
Lens signature
The distinctive features of the lens are its transparency, precisely crafted shape, and deformability, all of which are critical for proper light refraction. Elucidating the molecular mechanisms that maintain or disrupt lens transparency is fundamental in preventing cataract, the leading cause of world blindness. Our studies showed that lens gene expression is very distinct from the other eye compartments (Figure 2c), perhaps reflecting the extraordinary specialization of the lens as an isolated, avascular structure within the eye. We found more than a thousand genes selectively expressed in the lens; clearly, diverse RNA populations are still present in the adult lens, even though its population of active epithelial cells is outnumbered by the mature fiber cells that have lost their organelles, including nuclei.
Genes encoding the subunits of crystallins, the predominant structural proteins in the lens, were prominent in the lens signature, including subunits for crystallin alpha (CRYAA), beta (CRYBA1, CRYBA4), and gamma (CRYGA, CRYGC). Work by Horwitz and colleagues [32,33] on alpha-crystallins, which are structurally similar to small heat shock proteins, showed these crystallins may preserve lens transparency by serving as molecular chaperones that protect other lens proteins from irreversible denaturation and aggregation. Of the other heat shock proteins highly enriched in the lens signature (HSPA6, HSPA8, HSPB1), HSPB1 may be of particular interest because it is a protein with an alpha-crystallin domain that may have a role in lens differentiation [34]. The lens signature also included genes encoding subunits of the proteasome complex (PSMA6, PSMA7, PSMB6, PSMB7, PSMB9, PSMD13), a multicatalytic proteinase structure that is responsible for degrading intracellular proteins. Previous studies have demonstrated the significance of the proteasome pathway in removing oxidatively damaged proteins within the lens [35].
Besides the crystallin genes, other genes encoding previously described structural components of the lens, including lens intrinsic membrane (LIM2), beaded filament structural protein (BFSP2), spectrin (SPTBN2), and actin binding protein (ABLIM) were included in the lens signature. More interestingly, the signature also contained intermediate filament genes, such as those encoding erythrocyte membrane band 4.9 and 4.1 (EPB49 and EPB41L1, EPB41L4), that are characteristically expressed in erythrocytes, another cell whose highly stereotyped shape is critical to its function. Previous studies have shown that protein 4.1 helps stabilize the spectrin-actin cytoskeleton, which is present in both erythrocytes and lenticular tissue [36]. Further investigations comparing erythroid and lens cells may reveal other similarities in their cytoskeletons, both of which define a distinctive and stereotyped cell shape that must endure substantial amounts of mechanical stress.
Another notable feature of the lens signature was the enrichment of genes encoding proteins involved in endocytosis, including clathrin (CLTCL1, PICALM) and caveolin (CAV1). Currently, intercellular transport within the lens is thought to occur predominately by diffusion through gap junctions, but several investigators have proposed the uptake of nutrients must be supplemented by mechanisms other than gap junctions because of the paucity of gap junctions identified in microscopy studies and the confirmed presence of clathrin-coated vesicles in freeze-fracture studies [37,38].
Oxidative stress mediated by free radical production has been associated with cataract formation (reviewed in [39]). Therefore, we looked for genes involved in scavenging free radicals in the lens signature. Two of these genes encode enzymes, glutathione synthetase (GSS) and glutathione reductase (GSR), that facilitate the production of glutathione, a potent anti-oxidant and essential cofactor for redox enzymes. Superoxide dismutase (SOD1) and anti-oxidant protein 2 (AOP2), two proteins responsible for reducing free oxygen radicals and hydrogen peroxide species, respectively, were also selectively expressed in lens tissue. Drugs or environmental agents that modulate the expression or activity of these proteins could have a significant impact on cataract progression or prevention.
Optic nerve signature
The gene expression pattern in the optic nerve was overall quite similar to that seen in brain tissue (Figure 2d), very likely reflecting the preponderance of glial cells present in both tissues. Both signatures included a number of genes (MBP, MOBP, MAG, OLIG1, and OLIG2) previously found in glial cells, several of which have been linked to neurological diseases. For example, myelin-associated oligodendrocyte basic protein (MOBP) is implicated as an antigen stimulus for multiple sclerosis, a disease that also can present with optic neuritis (reviewed in [40]). Interestingly, the optic nerves in MOBP knock-out mice lacked the radial component of myelination [41]. In another study, transgenic mice with T-cell receptors specific to myelin associated glycoprotein (MAG) spontaneously presented with optic neuritis [42]. The majority of the genes in the brain and optic nerve signatures encoded proteins of unknown function; our results, showing that these genes may have specialized roles in these tissues, may be a step toward discovering the biological role(s) for these uncharacterized proteins.
Retina signature
The retina, a complex tissue composed of neuronal and glial elements, is essentially an extension of the central nervous system, and the genes found in the retina signature appear to reflect its distinctive histology and embryology (Figure 3a). For example, the signature included the receptors for known retinal neurotransmitters, including gamma-aminobutyric acid (GABRA1, GABRG2, GABRB3), glutamate (GRIA1, GRIN2D), glycine (GLRB), and dopamine (DRD2). Retinal neurotransmitters are packaged into small vesicles in the pre-synaptic regions of photoreceptors. Many retinal signature genes encoded proteins associated with synaptic vesicle docking and fusion (SNAP25, VAMP2, SYP, SNPH), as well as vesicle exocytosis and neurotransmitter release (SYN2, SYT4). One of the retinal signature genes with a role in synaptic transmission, human retinal gene 4 (HRG4/UNC119), has been linked to late-onset cone-rod dystrophy in humans and marked synaptic degeneration in a transgenic mouse model [43].
The retina protects the integrity of its neuronal layers by regulating its extracellular environment through a blood-retina barrier consisting of vessel tight junctions and cell basement membranes. The exchange of nutrients and metabolites across these barriers likely requires diverse, specialized transporters. Indeed, over 30 different genes encoding small molecule transporters were found within the retina signature, including carriers of glucose (SLC2A1, SLC2A3), glutamate (SLC1A7), and other amino acids (SLC7A5, SLC38A1, SLC6A6). Of note, severe retinal degeneration was observed in mice mutated in SLC6A6, a gene encoding a transporter of the amino acid taurine [44]. Several genes encoding ABC transporters (ABCA3, ABCA4, ABCA5, ABCA7), which use ATP energy to transport various molecules across cell membranes, were contained in the retinal signature. The most notable of these, ABCA4, is involved in vitamin A transport in photoreceptor cells; mutations in the gene encoding ABCA4 can result in a spectrum of retinopathies, including retinitis pigmentosa, Stargardt's disease, cone-rod dystrophy, and ARMD.
The retinal signature was also enriched in transcripts encoding vitamin and mineral transporters. The inclusion of a vitamin C transporter (SLC23A1) and a zinc transporter (SLC39A3) within the signature was of particular interest, in light of the Age-Related Eye Disease Research Group study that demonstrated supplementation with zinc and anti-oxidants, including vitamin C, lowered the probability of developing neovascular ARMD in some high-risk patient subgroups [45]. The presence of transferrin (TF), an iron transport molecule, and its receptor (TFRC), in the retina signature may also be noteworthy because a higher accumulation of iron has been observed in some ARMD-affected maculas [46].
Somewhat unexpectedly, the retina signature contained the gene encoding thyroid releasing hormone (TRH) and numerous thyroid hormone receptor-related genes (THRA, TRIP8, TRIP15, TRAP100). TRH expression was previously observed in the retinal amacrine cells of amphibians [47]. Previous work has demonstrated the importance of thyroid hormone in the developing rat retina [48], and thyroid hormone receptors are required for green cone photoreceptor development in rodents [49]. Further studies of these genes may uncover additional roles of thyroid hormone and its receptors in the human retina.
The retina is ultimately responsible for executing the visual cycle, the process by which a photon signal is translated into an electrical impulse. This complex cycle is initiated when photoreceptor pigments activate G-proteins. G-proteins in turn activate phosphodiesterases to break down cyclic GMP (cGMP) to GMP, thereby influencing cell polarization via the downstream modulation of ion channel efflux. The retina signature incorporated many genes encoding known visual cycle elements, including the photopigment rhodopsin (RHO), G-proteins from rods and cones (GNAT1, GNAT2, GNB5), subunits of rod and cone phosphodiesterases (PDE6A, PDE6B, PDE6G, PDE6H), and cGMP-sensitive channels (CNGB1, CNGA1). Genes responsible for visual cycle recovery, such as arrestins (SAG, ARR3), were also present. Intriguingly, transcripts encoding other G-proteins (GNB1, GNAZ) and several phosphodiesterases (PDE8B, PDE7A, PDE4A) with no established roles in the visual cycle were enriched in the retinal signature. Additionally, the signature contained CDS1, which, though it has no clear function in humans, is homologous to the phototransduction gene CDS that has been linked to light-induced retinal degeneration in Drosophila mutants [50]. Perhaps further in-depth study of the many uncharacterized genes in the retinal signature will reveal roles in phototransduction for these genes, which may expand our current concept of the visual cycle pathway.
Macula signature
We used the statistical analysis of microarrays (SAM) algorithm to select genes whose expression differed significantly between the central and peripheral retinal tissues (Figure 3b). The large set of genes that we identified as selectively expressed in macula tissues included a subset of genes involved in lipid biosynthesis. The majority of these genes are regulated by sterol response element-binding protein (SREBP), a transcription factor that has emerged as a master regulator of cholesterol and fatty acid metabolic pathways [51]. Previous studies by Fliesler et al. [52] have provided evidence for rapid de novo synthesis of cholesterol in the rat retina in vivo, and our findings strongly suggested the human retina also contains the enzymes needed for cholesterol biogenesis. Transcripts encoding the enzymes that catalyze multiple steps in cholesterol synthesis were enriched in the macula, including stearoyl-CoA desaturase (SCD), mevalonate decarboxylase (MVD), hydroxy-3-methylglutaryl-coenzyme A synthase 1 (HMGCS1), and HMG-coenzyme A reductase (HMGCR), the rate-limiting enzyme in cholesterol synthesis and the target of the 'statin' class of drugs for patients with dyslipidemia. Other macula signature genes encoded enzymes that act later in cholesterol biosynthesis, such as lanosterol synthase (LSS) and squalene epoxide (SQLE). In addition, the macula-enriched cluster included the gene for low-density lipoprotein receptor (LDLR), known for its role in binding low-density lipoprotein (LDL), the major cholesterol-carrying lipoprotein of plasma. LDL receptors and LDL-like receptors have been previously identified in retinal pigment epithelium and retinal muller cells [53,54], but their function in cholesterol transport within the retina has been minimally explored.
The genes represented in the macula cluster at least partially reflect cell types present in a higher density in the macula than in the peripheral retina, such as ganglion cells and photoreceptors. For example, a substantial number of genes in the macula signature have previously been characterized in ganglion cells (THY1, POU4F1, L1CAML1, NRN1). Interestingly, cholesterol is involved in the physiology of both retinal ganglion cells and photoreceptors. Cholesterol has been identified in rod outer segments in a wide variety of animal species (reviewed in [55]), as well as in oil droplets isolated from chicken cone photoreceptors [56]. In vitro, cholesterol has the capacity to modulate phototransduction in rods by altering the rod outer segment membrane structure [57], as well as by directly binding to rhodopsin itself [58]. Histological studies on retinas from patients with abetalipoproteinemia and familial hypobetalipoproteinemia, (serum LDL-cholesterol levels <5% of normal) demonstrated a profound absence of photoreceptors throughout most of the posterior retina [59,60]. In addition, patients with Smith-Lemli-Opitz Syndrome, a disease of abnormal cholesterol metabolism caused by a defect in 7-dehydrocholesterol reductase (DHCR7), another enzyme encoded by a gene selectively expressed in macula tissues, exhibited slower activation and recovery kinetics of their rod photoreceptors [61].
In vitro studies by Mauch et al. [62] have demonstrated that retinal ganglion cells require cholesterol in order to form mature, functioning synapses. The retinal ganglion cells in their experiments produced enough cholesterol to survive and grow, but effective synaptogenesis demanded additional cholesterol supplied by glial cells. Other work by Hayashi et al. [63] showed that exposure to lipoproteins containing cholesterol and apolipoprotein E stimulated retinal ganglion cell axons to extend, and that this effect was mediated by receptors of the LDL receptor family present on distal axons. Studying the role of cholesterol in synaptogenesis may lead to insights useful in the development of protective or restorative therapeutics for neurodegenerative disease, as well as for ocular diseases that affect ganglion cells.
In view of epidemiological studies that have suggested connections among atherosclerosis, serum cholesterol levels, and ARMD [64-66], the enrichment of cholesterol biosynthesis genes within the macula warrants further investigation. The presence of cholesterol in drusen, the extracellular deposits of ARMD, has been confirmed [67,68], although the origin of this cholesterol remains unclear. Disregulation of lipid metabolism and transport, either on a local and/or systemic level, may contribute to macular diseases, such as ARMD. Studies have associated statin use with a decreased rate of ARMD [69,70], but randomized, prospective studies have yet to be completed.
Identifying candidate disease genes
One direct application of the gene expression patterns we have defined is the identification of candidate genes for genetic diseases that differentially affect the various eye compartments. This strategy relies on the hypothesis that if mutations in a gene cause physiological aberrations specifically in a particular tissue, the gene is more likely to be selectively expressed in that tissue. We therefore used the literature, RetNet [71], and the Online Mendelian Inheritance in Man [72] databases, to collate lists of genetic diseases affecting the lens, cornea, and retina, along with the genetic intervals to which the disease loci have been mapped. Next, we identified genes that were relatively selectively expressed in each of the three compartments. Briefly, we standardized the Cy5 intensity data for each array and calculated the average intensity for every gene across all samples from each compartment. We then empirically identified an intensity cut-off that resulted in selection of greater than 85% of genes included in the retinal compartment signature from Figure 1, but also included highly expressed genes that were expressed in more than one compartment. Using this cut-off, we identified separate compartment gene lists for the three compartments and identified the subset of these genes that were located in the appropriate cytogenetic intervals for each compartment-specific disease (see Additional data files 4, 5, 6 and Materials and methods).
To assess the potential of this approach, we analyzed the subset of diseases for which candidate intervals were listed in our sources but for which the causative gene is now known. The density of affected-tissue-expressed genes located in the candidate intervals was similar to that for the unknown diseases, and thus this subset served as a reliable positive control. The disease gene for a remarkable 50% to 70% of the diseases of known genetic cause was selectively expressed in the cognate compartment (Table 1). We tested the statistical significance of this result by comparing the number of disease genes identified by the compartment gene expression lists with the aggregate list of all genes detectably expressed in any of the samples shown in Figure 1. We found that for all three groups of diseases, the compartment signatures were significantly enriched for candidate disease genes (lens, p < 0.002; cornea, p < 0.005; retina, p < 0.0004, by the hypergeometric distribution). By focusing on the genes expressed within the compartment displaying the disease phenotype, we could enrich for potential candidate genes by an average of 2 to 2.5-fold.
As an example of this approach, we more closely examined Retinitis Pigmentosa 29 (RP 29), an autosomal recessive form of RP that was mapped to chromosomal region 4q32-q34 in a consanguineous Pakistani family [73]. At least two genes within this interval (WDR17, GPM6A), and one gene near the interval (CCN3), were previously examined by sequencing and were excluded as candidates [74]. In our data, only one gene, KIAA1712, was both located within the mapped interval and selectively expressed in our retinal samples. Little is currently known about this gene, except that it appears in expressed sequence tags (EST) and SAGE libraries from several tissues, including brain. Our analysis suggests that KIAA1712 is a strong candidate gene for RP 29, and deserves further study. We expect our candidate gene lists to be highly enriched for the causative genes for a large fraction of the diseases we analyzed, and thus should prove useful in accelerating identification of genes important in various aspects of ocular pathology.
Discussion
Our microarray studies identified distinct molecular signatures for each compartment of the human eye. As we predicted, many of the genes differentially expressed in each tissue could be related to the histology and embryology of the cognate structure in the eye; more usefully, each signature uncovered numerous genes whose expression or function in the eye had not been previously characterized and for which their expression pattern now provides a new clue to their roles. Through a comparative analysis of gene expression among eye compartments, we can also gain insight into the pathophysiology of diseases that afflict specific eye tissues. Furthermore, our data may help anticipate or understand drug effects and side-effects, when the molecular targets of the drugs are preferentially expressed in particular ocular tissues.
The extensive set of genes selectively expressed in the macula demonstrates that there is significant regional variation in gene expression programs in the human retina. The macula-enriched expression pattern may provide clues to the pathogenesis of retinal diseases that preferentially affect the macula, such as ARMD. Because no ophthalmologic clinical data accompanied the autopsy globe samples used in our experiments and because of our limited sample sizes, we were unable to correlate our gene expression data with clinical exam findings or disease course. The techniques used in these experiments did, however, allow us to examine tissues from individual donors rather than requiring us to rely on either pooled tissue samples or cultured cells. Thus, our results show that future experiments examining individual diseased samples will be possible.
By analyzing our global gene expression data together with previous genetic mapping data, we were able to greatly refine sets of candidate genes for many corneal, lenticular, and retinal diseases whose genetic basis is still undefined. When we used a control set of diseases with known causative genes, the candidate gene lists we generated included 50% to 70% of the causative genes for this control set. One explanation for why we did not identify all the causative genes for the control disease set was that some causative genes did not meet our intensity threshold, and thus were not included in the compartment expression lists. Furthermore, we could not have identified those causative genes that are only expressed in the diseased state (but not in normal tissues), because we limited our microarray analyses to tissues with no known ocular pathology. Other reasons why our approach may have missed causative genes include expression of causative genes only at certain points in development and not in adult tissues, technical problems with the array element(s) representing these genes, and possible loss of transcripts in the RNA isolation or amplification process. Future investigation of these potential problems and comparison of our candidate gene lists with genome-scale gene expression data from diseased tissues will result in further refinement of the approach presented here.
Finally, our studies were designed to provide an open resource for all investigators interested in ocular physiology and disease. The tissue signature data, as well as the diseases, genetic intervals, and candidate genes for all the diseases we examined, and the complete set of data from our studies is freely available without restriction from the Authors' Web Supplement accompanying this manuscript [75].
Materials and methods
Tissue specimens
Eight whole globes (G1 to G8) were harvested from autopsy donors (age range 30 to 85 years old) within 24 h of death, and the tissues were immediately stored at 4°C in RNAlater (Ambion, Austin, TX, USA). Four of the globes were from female donors (G3, G6 to G8) and four were from male donors (G1, G2, G4, G5). Globes 4 and 5 were harvested as a set from a single donor, as were globes 6 and 7. No ophthalmologic clinical records were available for any of the globes at the time of harvest. Seven of the globes (G1 to G7) were dissected into the following components: cornea, lens, iris, ciliary body, retina, and optic nerve, while only retinal tissue was available from G8. The maculas and the peripheral retinal tissues were further dissected from several of the retinal samples. The macula was defined as the visible xanthophyll-containing tissue temporal to the optic nerve, which encompassed an approximate area of 4 mm2. For comparison purposes, three post-mortem brain specimens were analyzed.
RNA extraction and amplification
Specimens were disrupted in TRIZOL (Gibco, Carlsbad, CA, USA) solution using a tissue homogenizer. Samples were processed according to the manufacturer's protocol until the aqueous supernatant was retrieved. The supernatant was mixed with 1 volume of 70% ethanol, applied to an RNeasy column (Qiagen, Valencia, CA, USA), and purified according to the manufacturer's protocol. RNA quality and quantity were assessed by gel electrophoresis and spectrophotometer measurements. Total RNA was amplified using a single round, linear amplification method [9] (also see Additional data files 1 and 2). Tissue samples that yielded inadequate amounts of RNA were excluded from any further analysis. A reference mixture of mRNAs derived from 10 different cell lines (Universal Human Reference RNA, Stratagene, La Jolla, CA, USA) was used in all experiments as an internal standard for comparative two-color fluorescence hybridization.
Microarray procedures
Human cDNA microarray construction and hybridization were as previously described [76]. The microarrays contained 43,198 elements, representing approximately 30,000 genes (estimated by UniGene clusters) and were manufactured by the Stanford Functional Genomics Facility [77]. In each analysis, amplified RNA from an eye tissue sample was labeled with Cy5, and amplified reference RNA was labeled with Cy3. The two labeled samples were combined, and the mixture was hybridized to a microarray. Arrays were scanned using a GenePix 4000B scanner (Axon Instruments Inc., Sunnyvale, CA, USA). The array images were processed using GenePix Pro 3.0, and the resulting data were indexed in the Stanford Microarray Database and normalized using their default total intensity normalization algorithm (more detailed methods are available in Additional data file 3). Searchable figures and all raw microarray data can be found at [75]. The complete microarray dataset is also accessible through the Gene Expression Omnibus [78] (accession number GSE3023).
Bioinformatic analyses
For the data shown in Figures 1 and 2, only elements for which at least 50% of the measurements across all samples had fluorescence intensity in either channel at least 3.25-fold over background intensity were included. The logarithm of the ratio of background-subtracted Cy5 fluorescence to background-subtracted Cy3 fluorescence was calculated. Then values for each array and each gene were median centered, and only cDNA array elements for which at least two measurements differed by more than 2.5-fold from the median were included in subsequent analyses. For the data in Figures 3, we employed the Statistical Analysis of Microarrays (SAM) package [79]. Only elements for which the intensity to background ratio was at least 3.25 in at least 35% of the retina samples were considered. Only genes whose expression significantly differed between the macula and peripheral retina (false discovery rate <0.05 with 500 permutations) were selected. Finally, to focus on genes with the largest absolute difference in expression between the two regions, we selected genes whose expression differed by at least four-fold from the median in at least two samples.
Candidate disease gene analysis
To identify the gene sets expressed in each compartment, background-subtracted Cy5 intensities from each microarray were standardized to an array-median of 1,500, and genes exhibiting an average intensity of at least 2,500 in a compartment were identified (see Additional data file 4). This threshold was chosen empirically because it resulted in greater than 85% of the retinal signature from Figures 1 to be included in the retina set, while less than 5% of these genes were contained in any of the other compartment sets. Genetic diseases affecting the lens, cornea, or retina were collated from the Online Mendelian Inheritance in Man database [72] and the Retinal Information Network [71], along with the genetic intervals to which they have been mapped (see Additional data file 5). Using Perl scripts, we mapped every sequence on our arrays to the human genome using data from the UCSC genome browser [80]. Genes in the corresponding compartment expression set that were located in the genetic interval associated with each compartment-specific disease were identified. For the benchmark analysis of diseases that were associated with known genes, we also identified all genes in the human genome that fell into the genetic interval associated with each disease. The compartment expression sets and our lists of candidate genes for the 147 diseases we analyzed can be found in Additional data file 6.
Additional data files
The following additional data are available with the online version of this paper. Additional data file 1 contains the step-by-step amplification protocol used in this work. Additional data file 2 is a table detailing RNA isolation and amplification yields. Additional data file 3 contains more detailed supplemental materials and methods. Additional data file 4 contains the compartment gene lists used in the disease gene analysis. Additional data file 5 contains the list of diseases for each compartment with their mapped genetic intervals. Additional data file 6 contains the results of the disease gene analysis, including the list of candidate genes for each disease.
Supplementary Material
Additional data file 1
Step-by-step amplification protocol.
Click here for file
Additional data file 2
A table detailing RNA isolation and amplification yields.
Click here for file
Additional data file 3
Detailed supplemental materials and methods.
Click here for file
Additional data file 4
Compartment gene lists used in the disease gene analysis.
Click here for file
Additional data file 5
The list of diseases for each compartment with their mapped genetic intervals.
Click here for file
Additional data file 6
Results of the disease gene analysis, including the list of candidate genes for each disease.
Click here for file
Acknowledgements
We wish to thank members of the Brown laboratory for helpful advice and discussions, M van de Rijn and Stanford pathology for help with tissue acquisition, and T Hernandez-Boussard for computational assistance. This work was supported by the Howard Hughes Medical Institute, NCI grant CA77097, the Stanford Medical Scholars Program (J.D.), and by NIGMS training grant GM07365 (MD). P.O.B. is an investigator of the HHMI.
Figures and Tables
Figure 1 Gene expression programs in the human eye. Unsupervised hierarchical clustering of 38 samples from human cadaver eyes and normal brain. Array elements that varied at least 2.5-fold from the median on at least two microarrays were included (9,634 cDNA elements representing approximately 6,600 genes). (a) Array dendrogram. G1 to G8 indicate the globes from which each compartment sample was dissected (see Materials and methods). Inf., inferior; Sup., superior; Temp., temporal. (b) Cluster image. Data are displayed as a hierarchical cluster where rows represent genes (unique cDNA elements) and columns represent experimental samples. Colored pixels capture the magnitude of the response for any gene, where shades of red and green represent induction and repression, respectively, relative to the median for each gene. Black pixels reflect no change from the median and gray pixels represent missing data. Compartment-specific gene signatures are indicated. See our website for a searchable version of this cluster [75].
Figure 2 Expanded view of compartment-specific gene expression signatures in the human eye. Data were extracted from Figure 1 and are displayed similarly. Individual clusters depict genes associated with (a) cornea, (b) ciliary body and iris, (c) lens and (d) optic nerve. Many of the array elements encode uncharacterized genes and only a subset of named genes is shown.
Figure 3 Retinal gene expression. (a) The retina-specific gene expression signature was extracted from Figure 1 and is displayed similarly. Many of the array elements encode uncharacterized genes and only a subset of named genes is shown. (b) Macula versus peripheral retina gene expression. Using the statistical analysis of microarrays algorithm as described in Materials and methods, we selected genes that differed significantly between the central and peripheral retinal arrays at a false discovery rate <0.05.
Table 1 Compartment gene sets are enriched for candidate genes of ocular diseases
Disease-associated genes on array Disease-associated genes expressed in affected compartment Percentage of known disease-associated genes identified Average fold enrichment compared to total number of genes in interval P-value
Lens 15 8 53 2.4 0.002
Cornea 13 9 69 2.0 0.005
Retina 42 23 55 2.3 0.0004
Arrays were standardized to the same median intensity and genes exhibiting minimum intensities of 2,500 in any compartment were identified. Genetic diseases affecting the lens, cornea, or retina were collated from the RetNet [71] and Online Mendelian Inheritance in Man [72] databases, along with their cytogenetic map positions. The table indicates the number of cloned disease genes on the arrays, the number contained in a given compartment gene set, the percentage of known disease genes included in the signatures, the average fold enrichment compared to the total number of genes in each cytogenetic interval, and the statistical significance of this enrichment (using the hypergeometric distribution).
==== Refs
Lander ES Linton LM Birren B Nusbaum C Zody MC Baldwin J Devon K Dewar K Doyle M FitzHugh W Initial sequencing and analysis of the human genome. Nature 2001 409 860 921 11237011 10.1038/35057062
Venter JC Adams MD Myers EW Li PW Mural RJ Sutton GG Smith HO Yandell M Evans CA Holt RA The sequence of the human genome. Science 2001 291 1304 1351 11181995 10.1126/science.1058040
Jun AS Liu SH Koo EH Do DV Stark WJ Gottsch JD Microarray analysis of gene expression in human donor corneas. Arch Ophthalmol 2001 119 1629 1634 11709013
Chowers I Liu D Farkas RH Gunatilaka TL Hackam AS Bernstein SL Campochiaro PA Parmigiani G Zack DJ Gene expression variation in the adult human retina. Hum Mol Genet 2003 12 2881 2893 14519682 10.1093/hmg/ddg326
Hernandez MR Agapova OA Yang P Salvador-Silva M Ricard CS Aoi S Differential gene expression in astrocytes from human normal and glaucomatous optic nerve head analyzed by cDNA microarray. Glia 2002 38 45 64 11921203 10.1002/glia.10051
Seftor EA Meltzer PS Kirschmann DA Pe'er J Maniotis AJ Trent JM Folberg R Hendrix MJ Molecular determinants of human uveal melanoma invasion and metastasis. Clin Exp Metastasis 2002 19 233 246 12067204 10.1023/A:1015591624171
Yoshida S Yashar BM Hiriyanna S Swaroop A Microarray analysis of gene expression in the aging human retina. Invest Ophthalmol Vis Sci 2002 43 2554 2560 12147584
Wang WH McNatt LG Shepard AR Jacobson N Nishimura DY Stone EM Sheffield VC Clark AF Optimal procedure for extracting RNA from human ocular tissues and expression profiling of the congenital glaucoma gene FOXC1 using quantitative RT-PCR. Mol Vis 2001 7 89 94 11320352
Wang E Miller LD Ohnmacht GA Liu ET Marincola FM High-fidelity mRNA amplification for gene profiling. Nat Biotechnol 2000 18 457 459 10748532 10.1038/74546
Iyer VR Eisen MB Ross DT Schuler G Moore T Lee JC Trent JM Staudt LM Hudson J JrBoguski MS The transcriptional program in the response of human fibroblasts to serum. Science 1999 283 83 87 9872747 10.1126/science.283.5398.83
Hu L Wang J Baggerly K Wang H Fuller GN Hamilton SR Coombes KR Zhang W Obtaining reliable information from minute amounts of RNA using cDNA microarrays. BMC Genomics 2002 3 16 12086591 10.1186/1471-2164-3-16
Zhao H Hastie T Whitfield ML Borresen-Dale AL Jeffrey SS Optimization and evaluation of T7 based RNA linear amplification protocols for cDNA microarray analysis. BMC Genomics 2002 3 31 12445333 10.1186/1471-2164-3-31
Sharon D Blackshaw S Cepko CL Dryja TP Profile of the genes expressed in the human peripheral retina, macula, and retinal pigment epithelium determined through serial analysis of gene expression (SAGE). Proc Natl Acad Sci USA 2002 99 315 320 11756676 10.1073/pnas.012582799
Eisen MB Spellman PT Brown PO Botstein D Cluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci USA 1998 95 14863 14868 9843981 10.1073/pnas.95.25.14863
Liu CY Birk DE Hassell JR Kane B Kao WW Keratocan-deficient mice display alterations in corneal structure. J Biol Chem 2003 278 21672 21677 12665512 10.1074/jbc.M301169200
Quantock AJ Meek KM Chakravarti S An x-ray diffraction investigation of corneal structure in lumican-deficient mice. Invest Ophthalmol Vis Sci 2001 42 1750 1756 11431438
Kenney MC Atilano SR Zorapapel N Holguin B Gaster RN Ljubimov AV Altered expression of aquaporins in bullous keratopathy and Fuchs' dystrophy corneas. J Histochem Cytochem 2004 52 1341 1350 15385580 10.1369/jhc.4A6385.2004
Yang J Mani SA Donaher JL Ramaswamy S Itzykson RA Come C Savagner P Gitelman I Richardson A Weinberg RA Twist, a master regulator of morphogenesis, plays an essential role in tumor metastasis. Cell 2004 117 927 939 15210113 10.1016/j.cell.2004.06.006
Van Themsche C Alain T Kossakowska AE Urbanski S Potworowski EF St-Pierre Y Stromelysin-2 (matrix metalloproteinase 10) is inducible in lymphoma cells and accelerates the growth of lymphoid tumors in vivo. J Immunol 2004 173 3605 3611 15356104
Sivak JM Fini ME MMPs in the eye: emerging roles for matrix metalloproteinases in ocular physiology. Prog Retin Eye Res 2002 21 1 14 11906808 10.1016/S1350-9462(01)00015-5
Daniels JT Geerling G Alexander RA Murphy G Khaw PT Saarialho-Kere U Temporal and spatial expression of matrix metalloproteinases during wound healing of human corneal tissue. Exp Eye Res 2003 77 653 664 14609553 10.1016/j.exer.2003.08.010
Chang HY Sneddon JB Alizadeh AA Sood R West RB Montgomery K Chi JT van de Rijn M Botstein D Brown PO Gene expression signature of fibroblast serum response predicts human cancer progression: similarities between tumors and wounds. PLoS Biol 2004 2 E7 14737219 10.1371/journal.pbio.0020007
Chang HY Nuyten DS Sneddon JB Hastie T Tibshirani R Sorlie T Dai H He YD van't Veer LJ Bartelink H Robustness, scalability, and integration of a wound-response gene expression signature in predicting breast cancer survival. Proc Natl Acad Sci USA 2005 102 3738 3743 15701700 10.1073/pnas.0409462102
Stahl F Lepple-Wienhues A Kuppinger M Schneider U Wiederholt M Measurements of intracellular calcium and contractility in human ciliary muscle. Pflugers Arch 1991 418 531 537 1945746 10.1007/BF00370567
Stumpff F Wiederholt M Regulation of trabecular meshwork contractility. Ophthalmologica 2000 214 33 53 10657743 10.1159/000027471
Wiederholt M Dorschner N Groth J Effect of diuretics, channel modulators and signal interceptors on contractility of the trabecular meshwork. Ophthalmologica 1997 211 153 160 9176896
Wiederholt M Thieme H Stumpff F The regulation of trabecular meshwork and ciliary muscle contractility. Prog Retin Eye Res 2000 19 271 295 10749378 10.1016/S1350-9462(99)00015-4
Sohn JH Kaplan HJ Suk HJ Bora PS Bora NS Chronic low level complement activation within the eye is controlled by intraocular complement regulatory proteins. Invest Ophthalmol Vis Sci 2000 41 3492 3502 11006244
Streilein JW Regulation of ocular immune responses. Eye 1997 11 171 175 9349408
Rogoz S Avramescu C Silosi I Drackoulogona O Badea P State A [Humoral and cellular immunity investigation in idiopathic uveitis.] Oftalmologia 2002 54 56 60 12723200
Bardenstein DS Cheyer CJ Lee C Cocuzzi E Mizuno M Okada N Medof ME Blockage of complement regulators in the conjunctiva and within the eye leads to massive inflammation and iritis. Immunology 2001 104 423 430 11899428 10.1046/j.1365-2567.2001.01316.x
Horwitz J Alpha-crystallin can function as a molecular chaperone. Proc Natl Acad Sci USA 1992 89 10449 10453 1438232
Horwitz J Emmons T Takemoto L The ability of lens alpha crystallin to protect against heat-induced aggregation is age-dependent. Curr Eye Res 1992 11 817 822 1424725
Chiesa R Noguera I Sredy J Phosphorylation of HSP25 during lens cell differentiation. Exp Eye Res 1997 65 223 229 9268590 10.1006/exer.1997.0324
Shang F Nowell TR JrTaylor A Removal of oxidatively damaged proteins from lens cells by the ubiquitin-proteasome pathway. Exp Eye Res 2001 73 229 238 11446773 10.1006/exer.2001.1029
Granger BL Lazarides E Appearance of new variants of membrane skeletal protein 4.1 during terminal differentiation of avian erythroid and lenticular cells. Nature 1985 313 238 241 3855501 10.1038/313238a0
Brown HG Pappas GD Ireland ME Kuszak JR Ultrastructural, biochemical, and immunologic evidence of receptor-mediated endocytosis in the crystalline lens. Invest Ophthalmol Vis Sci 1990 31 2579 2592 2176185
Lo WK Mills A Zhang W Zhu H Polarized distribution of coated pits and coated vesicles in the rat lens: an electron microscopy and WGA-HRP tracer study. Curr Eye Res 1991 10 1151 1163 1724956
Ottonello S Foroni C Carta A Petrucco S Maraini G Oxidative stress and age-related cataract. Ophthalmologica 2000 214 78 85 10657746 10.1159/000027474
Dharmasaroja P Specificity of autoantibodies to epitopes of myelin proteins in multiple sclerosis. J Neurol Sci 2003 206 7 16 12480078 10.1016/S0022-510X(02)00349-0
Yoshikawa H Myelin-associated oligodendrocytic basic protein modulates the arrangement of radial growth of the axon and the radial component of myelin. Med Electron Microsc 2001 34 160 164 11793190 10.1007/s007950100009
Bettelli E Pagany M Weiner HL Linington C Sobel RA Kuchroo VK Myelin oligodendrocyte glycoprotein-specific T cell receptor transgenic mice develop spontaneous autoimmune optic neuritis. J Exp Med 2003 197 1073 1081 12732654 10.1084/jem.20021603
Kobayashi A Higashide T Hamasaki D Kubota S Sakuma H An W Fujimaki T McLaren MJ Weleber RG Inana G HRG4 (UNC119) mutation found in cone-rod dystrophy causes retinal degeneration in a transgenic model. Invest Ophthalmol Vis Sci 2000 41 3268 3277 11006213
Heller-Stilb B van Roeyen C Rascher K Hartwig HG Huth A Seeliger MW Warskulat U Haussinger D Disruption of the taurine transporter gene (taut) leads to retinal degeneration in mice. FASEB J 2002 16 231 233 11772953
Age-Related Eye Disease Research Group A randomized, placebo-controlled, clinical trial of high-dose supplementation with vitamins C and E, beta carotene, and zinc for age-related macular degeneration and vision loss: AREDS report no. 8. Arch Ophthalmol 2001 119 1417 1436 11594942
Hahn P Milam AH Dunaief JL Maculas affected by age-related macular degeneration contain increased chelatable iron in the retinal pigment epithelium and Bruch's membrane. Arch Ophthalmol 2003 121 1099 1105 12912686 10.1001/archopht.121.8.1099
Anadon R Luz Diaz M Becerra M Jesus Manso M Presence of thyrotropin-releasing-hormone-immunoreactive (TRHir) amacrine cells in the retina of anuran and urodele amphibians. Brain Res 2002 926 86 93 11814410 10.1016/S0006-8993(01)03307-8
Sevilla-Romero E Munoz A Pinazo-Duran MD Low thyroid hormone levels impair the perinatal development of the rat retina. Ophthalmic Res 2002 34 181 191 12297689 10.1159/000063885
Ng L Hurley JB Dierks B Srinivas M Salto C Vennstrom B Reh TA Forrest D A thyroid hormone receptor that is required for the development of green cone photoreceptors. Nat Genet 2001 27 94 98 11138006
Wu L Niemeyer B Colley N Socolich M Zuker CS Regulation of PLC-mediated signalling in vivo by CDP-diacylglycerol synthase. Nature 1995 373 216 222 7816135 10.1038/373216a0
Horton JD Shah NA Warrington JA Anderson NN Park SW Brown MS Goldstein JL Combined analysis of oligonucleotide microarray data from transgenic and knockout mice identifies direct SREBP target genes. Proc Natl Acad Sci USA 2003 100 12027 12032 14512514 10.1073/pnas.1534923100
Fliesler SJ Florman R Rapp LM Pittler SJ Keller RK In vivo biosynthesis of cholesterol in the rat retina. FEBS Lett 1993 335 234 238 8253203 10.1016/0014-5793(93)80736-E
Hayes KC Lindsey S Stephan ZF Brecker D Retinal pigment epithelium possesses both LDL and scavenger receptor activity. Invest Ophthalmol Vis Sci 1989 30 225 232 2536645
Birkenmeier G Grosche J Reichenbach A Immunocytochemical demonstration of alpha 2-M-R/LRP on Muller (glial) cells isolated from rabbit and human retina. Neuroreport 1996 8 149 151 9051770
Fliesler SJ Anderson RE Chemistry and metabolism of lipids in the vertebrate retina. Prog Lipid Res 1983 22 79 131 6348799 10.1016/0163-7827(83)90004-8
Johnston D Hudson RA Isolation and composition of the carotenoid-containing oil droplets from cone photoreceptors. Biochim Biophys Acta 1976 424 235 245 1252489
Albert AD Boesze-Battaglia K Paw Z Watts A Epand RM Effect of cholesterol on rhodopsin stability in disk membranes. Biochim Biophys Acta 1996 1297 77 82 8841383
Albert AD Young JE Yeagle PL Rhodopsin-cholesterol interactions in bovine rod outer segment disk membranes. Biochim Biophys Acta 1996 1285 47 55 8948474
Brosnahan DM Kennedy SM Converse CA Lee WR Hammer HM Pathology of hereditary retinal degeneration associated with hypobetalipoproteinemia. Ophthalmology 1994 101 38 45 8302562
Cogan DG Rodrigues M Chu FC Schaefer EJ Ocular abnormalities in abetalipoproteinemia. A clinicopathologic correlation. Ophthalmology 1984 91 991 998 6493710
Elias ER Hansen RM Irons M Quinn NB Fulton AB Rod photoreceptor responses in children with Smith-Lemli-Opitz syndrome. Arch Ophthalmol 2003 121 1738 1743 14662594 10.1001/archopht.121.12.1738
Mauch DH Nagler K Schumacher S Goritz C Muller EC Otto A Pfrieger FW CNS synaptogenesis promoted by glia-derived cholesterol. Science 2001 294 1354 1357 11701931 10.1126/science.294.5545.1354
Hayashi H Campenot RB Vance DE Vance JE Glial lipoproteins stimulate axon growth of central nervous system neurons in compartmented cultures. J Biol Chem 2004 279 14009 14015 14709547 10.1074/jbc.M313828200
Abalain JH Carre JL Leglise D Robinet A Legall F Meskar A Floch HH Colin J Is age-related macular degeneration associated with serum lipoprotein and lipoparticle levels? Clin Chim Acta 2002 326 97 104 12417100 10.1016/S0009-8981(02)00288-7
van Leeuwen R Ikram MK Vingerling JR Witteman JC Hofman A de Jong PT Blood pressure, atherosclerosis, and the incidence of age-related maculopathy: the Rotterdam Study. Invest Ophthalmol Vis Sci 2003 44 3771 3777 12939290 10.1167/iovs.03-0121
van Leeuwen R Klaver CC Vingerling JR Hofman A van Duijn CM Stricker BH de Jong PT Cholesterol and age-related macular degeneration: is there a link? Am J Ophthalmol 2004 137 750 752 15059717 10.1016/j.ajo.2003.09.015
Mullins RF Russell SR Anderson DH Hageman GS Drusen associated with aging and age-related macular degeneration contain proteins common to extracellular deposits associated with atherosclerosis, elastosis, amyloidosis, and dense deposit disease. FASEB J 2000 14 835 846 10783137
Malek G Li CM Guidry C Medeiros NE Curcio CA Apolipoprotein B in cholesterol-containing drusen and basal deposits of human eyes with age-related maculopathy. Am J Pathol 2003 162 413 425 12547700
Wilson HL Schwartz DM Bhatt HR McCulloch CE Duncan JL Statin and aspirin therapy are associated with decreased rates of choroidal neovascularization among patients with age-related macular degeneration. Am J Ophthalmol 2004 137 615 624 15059698 10.1016/j.ajo.2003.10.025
McGwin G JrOwsley C Curcio CA Crain RJ The association between statin use and age related maculopathy. Br J Ophthalmol 2003 87 1121 1125 12928279 10.1136/bjo.87.9.1121
RetNet
Hamosh A Scott AF Amberger J Valle D McKusick VA Online Mendelian Inheritance in Man (OMIM). Hum Mutat 2000 15 57 61 10612823 10.1002/(SICI)1098-1004(200001)15:1<57::AID-HUMU12>3.0.CO;2-G
Hameed A Khaliq S Ismail M Anwar K Mehdi SQ Bessant D Payne AM Bhattacharya SS A new locus for autosomal recessive RP (RP29) mapping to chromosome 4q32-q34 in a Pakistani family. Invest Ophthalmol Vis Sci 2001 42 1436 1438 11381043
Stohr H Mohr N Frohlich S Mehdi SQ Bhattacharya SS Weber BH Cloning and characterization of WDR17, a novel WD repeat-containing gene on chromosome 4q34. Biochim Biophys Acta 2002 1579 18 25 12401215
Authors' Web Supplement
Eisen MB Brown PO DNA arrays for analysis of gene expression. Methods Enzymol 1999 303 179 205 10349646
Stanford Functional Genomics Facility
Gene Expression Omnibus (GEO)
Tusher VG Tibshirani R Chu G Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci USA 2001 98 5116 5121 11309499 10.1073/pnas.091062498
UCSC Genome Browser
|
16168081
|
PMC1242209
|
CC BY
|
2021-01-04 16:05:41
|
no
|
Genome Biol. 2005 Aug 17; 6(9):R74
|
utf-8
|
Genome Biol
| 2,005 |
10.1186/gb-2005-6-9-r74
|
oa_comm
|
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-9-r751616808210.1186/gb-2005-6-9-r75ResearchEvidence for selection on synonymous mutations affecting stability of mRNA secondary structure in mammals Chamary JV [email protected] Laurence D [email protected] Department of Biology and Biochemistry, University of Bath, Bath BA2 7AY, UK2005 16 8 2005 6 9 R75 R75 27 4 2005 8 6 2005 20 7 2005 Copyright © 2005 Chamary and Hurst; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Simulating evolution and reallocating the substitutions observed in mouse genes revealed that in mammals synonymous sites do not evolve neutrally and synonymous mutations may be under selection because of their effects on the thermodynamic stability of mRNA.
Background
In mammals, contrary to what is usually assumed, recent evidence suggests that synonymous mutations may not be selectively neutral. This position has proven contentious, not least because of the absence of a viable mechanism. Here we test whether synonymous mutations might be under selection owing to their effects on the thermodynamic stability of mRNA, mediated by changes in secondary structure.
Results
We provide numerous lines of evidence that are all consistent with the above hypothesis. Most notably, by simulating evolution and reallocating the substitutions observed in the mouse lineage, we show that the location of synonymous mutations is non-random with respect to stability. Importantly, the preference for cytosine at 4-fold degenerate sites, diagnostic of selection, can be explained by its effect on mRNA stability. Likewise, by interchanging synonymous codons, we find naturally occurring mRNAs to be more stable than simulant transcripts. Housekeeping genes, whose proteins are under strong purifying selection, are also under the greatest pressure to maintain stability.
Conclusion
Taken together, our results provide evidence that, in mammals, synonymous sites do not evolve neutrally, at least in part owing to selection on mRNA stability. This has implications for the application of synonymous divergence in estimating the mutation rate.
==== Body
Background
At least in mammals, it is typically assumed that selection does not affect the fate of synonymous (silent) mutations, those nucleotide changes occurring within a gene that affect the coding sequence but not the protein [1,2]. This presumption is in no small part based on the understanding that effective population sizes (Ne) in mammals are small. According to the nearly neutral theory [3], if s is the strength of selection against weakly deleterious mutations, then selection is expected to oppose their fixation when s > 1/2Ne [4]. Consequently, when s is small, species with low Ne are less likely to prevent the fixation of weakly deleterious mutations [5]. Indeed, for species with large effective population sizes, there is little doubt that selection is a strong enough force to determine the fate of synonymous mutations (for example, see [6]). Conversely, in mammals, analyses of codon usage have failed to detect clear signatures of selection (reviewed in [7]).
That synonymous mutations are effectively free of selection is important, not least because, if they really are neutral, their rate of evolution should be equal to the mutation rate. The rate of synonymous evolution could hence be used to provide a simple and convenient measure of the mutation rate [8,9]. More recently, however, the assumption of neutrality at synonymous sites has been called into question [10-16]. This view has proven contentious, not least because of the absence of a functional role for supposedly silent sites.
Here we examine one hypothesis, that synonymous mutations in mammals are under selection because they affect the thermodynamic stability of mRNA secondary structures [17,18], possibly to prolong cellular half-lives [19,20]. Unlike many non-coding RNAs [21-23], for which a stable secondary structure is selectively favored [24-28], the evolution of a stable structure for mRNA would be constrained by the need to encode a functional protein [17-19,29-31]. Consequently, were selection to operate on mRNA stability, synonymous mutations might be especially important (but see also [32,33]).
The hypothesis is supported by findings that synonymous mutations not only alter mRNA stem-loop structure [34,35], but also affect decay rates, and may lead to disease [35-37]. One possibility is that stem (base-paired) structures protect [38,39] against passive degradation by endoribonucleases [36,40,41]. Similarly, stable structures would be less likely to fall apart and thus expose vulnerable loop (single-stranded) regions to cleavage. Notably, analysis of computationally predicted mRNA stability across a wide taxonomic range revealed that real transcripts are more stable than comparable sequences in which synonymous codons were shuffled while the protein sequence remained unaltered [42,43].
Unfortunately, broad scale empirical analysis of mRNA stability is currently intractable because the structure of sequences much longer than tRNAs cannot be directly observed [20,44]. Consequently, mRNA folding is typically predicted computationally, by one of a variety of methods (see Materials and methods). Importantly, however, no in silico method can completely predict how cellular conditions might affect secondary structure [45]. For instance, proteins bound to mature transcripts [46] may have an effect, while chaperones are probably required to guide folding and/or prevent RNAs becoming kinetically trapped in unfavorable conformations [47,48]. Programs that attempt to incorporate the kinetics of the folding process that results from the directionality of transcription [49-51] are still under development [51]. Additionally, although a structure predicted in silico might be designated 'correct' because it forms in vitro, folding may be somewhat different in vivo [48,50].
The premise of this paper is not then to suppose that the prediction method and assumptions are flawless. Rather, we suppose that, if the method is telling us nothing about selection on mRNA stability, there is no reason why multiple independent tests should all point towards the same conclusion. In particular we ask: whether the nucleotides at synonymous sites are non-random with respect to stability; whether the excess of cytosine at synonymous sites in rodents [15] might be accounted for in terms of selection on mRNA stability; whether the location of substitutions in the mouse lineage are non-random with respect to stability; and whether genes under stronger purifying selection also have higher relative stability.
Although the hypothesis predicts that high mRNA stability should be favored, note that we do not expect stability to be extremely high, as ultra-stable structures would impose kinetic barriers that could hinder ribosome translocation [36,52]. While we presume that the transcripts of most genes will be relatively stable, in some cases mRNAs may actually need to be particularly unstable [43]. For example, selection might not act to promote stability because the mRNA is protein-bound and control of expression occurs at the translational level. Alternatively, some genes may only need to be transiently expressed, such as those encoding transcription factors [53,54]. As it is difficult to identify a priori which genes these might be, we cannot filter the dataset. This does, however, render our results conservative.
Results
For 70 mouse mRNAs (Additional data file 1), we predict a single optimal putative secondary structure and its thermodynamic stability (ΔG, kcal/mol, the difference in free energy between the folded and unfolded states). Prior studies providing evidence of selection on mRNA structure have employed a randomization protocol that shuffles synonymous codons to generate numerous simulants [42,43,55,56]. Based on the idea that 'interesting' RNAs should be more stable than expected by chance [57], one can then ask whether the stability of a real (wild-type) transcript is, on average, greater than that of its simulants. Seffens and Digby [42], for example, did this for a range of taxa (from bacteria to human). To determine if there is a prima facie case to answer, we first performed an analysis similar to that done previously, but specifically restricted to mammalian sequences.
Nucleotide content at synonymous sites is non-random with respect to mRNA stability
If selection acts on synonymous sites, by comparing a real mouse mRNA to simulants differing only at synonymous sites, we should find that, on average, the real transcript is more stable. For each gene we generated 1,000 random mRNAs identical in all regards to the real sequence, but with the bases at 4-fold degenerate (synonymous) sites in the coding sequence (CDS) randomly shuffled between the 4-fold degenerate positions. For each mRNA we determined Z(ΔG), the number of standard deviations the real mRNA is away from the mean stability of the simulants. Z(ΔG) is thus a measure of 'relative stability', the stability of a given mRNA relative to what one would expect by chance alone. Relative stability can also be considered as a measure of the strength of selection for stability, with a negative Z-score implying higher than expected stability. As Table 1 shows, real mRNAs are, on average, highly significantly more stable than 'Sh.4-fold' simulants (Figure 1; Additional data files 2, 3). Note, however, that on an individual basis, the effect (if any) is weak, with only 26 (37%) of genes having significantly high relative stability at the 5% level (Additional data file 4). Moreover, were we to apply Bonferonni correction for multiple testing on the by-gene P-values, no more than four genes would be significant at the 5% level. Inspection of the genes in our dataset (Additional data files 5, 6) did not reveal an obvious pattern that relates relative stability to their function.
In organisms from large effective populations, bias in codon usage is usually attributed to translational selection, favoring efficient (fast and/or accurate) protein synthesis as a consequence of skews in iso-acceptor tRNA abundance (reviewed in [7,58,59]). Whether this occurs in mammals, however, remains a contentious issue. While some have suggested that preferred sets of codons do exist to match the most abundant tRNAs [60], others maintain that codon usage does not reflect tRNAs skews [7,61] and that translational selection does not occur [62]. To be cautious, however, we also employed a protocol ('Sh.codon') that preserves the relative frequency of codons within a given set by shuffling codons within synonymous sets. This protocol gave very similar results to the previous ('Sh.4-fold') randomization (Table 1; Additional data files 2, 3).
Cytosine preference at synonymous sites, diagnostic of selection, can be explained by selection on mRNA stability
While the above results suggest that the identity of the nucleotide at any given synonymous site is non-random, this need not reflect maintenance of mRNA stability. Selection could instead be acting on a thermodynamic property of DNA, such as bendability [63]. As more G:C pairings make helices more bendable and gene-dense regions are GC-rich (for example, see [64]), the putative selection on GC content we observe at the mRNA level might actually function to provide the transcriptional machinery with easier access to the most gene-dense regions of DNA. To address this issue, we asked about the strand-specific preference for cytosine at 4-fold degenerate sites observed in rodent exons [15].
Cytosine preference is indicated by two related features: a higher C content at 4-fold sites than in flanking introns (not observed for guanine) and an excess of C over G at 4-fold degenerate sites [15]. Correspondingly, we found C4 > G4 in 87% of our mouse genes and a mean skew in GC4 (G - C/G + C) of -0.1506 (P = 1e-11 for expected mean (μ) < 0 by one-sample t-test on GC4 skew). Importantly, the skew towards C is specific to exons and, therefore, cannot be accounted for by effects at the DNA level (for example, mutational biases such as transcription-coupled repair, or selection on transcription). Note also that the sign of the skew is the opposite of that derived from transcription-coupled repair, which yields a G excess [65,66]. Significantly, introducing synonymous changes that increase C|G dinucleotide content (where | is the codon boundary) extends mRNA half-life in vitro while increasing A|U enhances degradation [36]. If selection is acting on mRNA stability, then this could be explained by a high C content at third sites increasing the number of potential G:C base-pairs, which are stronger than A:U interactions (triple and double hydrogen bonds, respectively). Consistent with this, we find that genes with the highest relative stability also have a greater excess of C over G (Spearman rank correlation coefficient (ρ) = 0.27, P = 0.0225 for GC4 skew versus Z(ΔGSh.4-fold); Additional data file 7).
To further examine the possibility that the C preference is explained by selection on RNA structure, we also asked whether replacing C residues with G decreases stability. We found that real mRNAs are more stable than modified transcripts in which, at 4-fold sites, we swapped all Cs for Gs and vice versa (Table 1; Additional data files 2, 3). 'Swap G4C4' mRNAs, however, possess a higher percentage of base-pairs than real transcripts (62.11 ± 0.33% and 60.96 ± 0.28% in CDS, respectively, P = 0.0003 by paired t-test; 60.84 ± 0.26% and 61.61 ± 0.26% in mRNA, P = 0.0007). That 'Swap G4C4' mRNAs have more base-pairs but lower stability can be explained by the existence of G:U base-pairs within stems, as G:Us are weaker than Watson-Crick interactions (A:U and G:C). An increased G content increases the amount of G:U pairs (real 10.50 ± 0.26% and 'Swap G4C4' 11.64 ± 0.21% in mRNA, P = 3e-07) and thus the proportion of base-paired mRNA, but their stems are less stable (there is no difference in the proportion of A:U pairs: real 36.60 ± 0.70%, 'Swap G4C4' 36.39 ± 0.71%, P = 0.2449). These results further underline the importance of nucleotide content for mRNA rather than DNA stability, not least because the location of bases that can potentially form Watson-Crick base-pairs in DNA is preserved in the modified transcripts.
Biased amino acid content and RNA stability may together drive C preference at third sites
The results above suggest that, given the nucleotide content at non-synonymous sites, C enrichment at synonymous sites is adaptive in regards to mRNA stability. Is there something about non-synonymous sites that causes C in particular to be enriched at synonymous sites? Fitch [17] proposed that, if genetic code degeneracy is exploited to optimize base-pairing in mRNA, third sites within codons (usually synonymous) should be preferentially paired with first and second sites (few and no synonymous sites, respectively). This would also provide a buffer for mRNA structure against non-synonymous substitutions via compensatory changes. Cytosine preference at third sites might, therefore, be driven by selection on amino acid content and mRNA stability [19].
In stems, we expect that, to permit base-pairing, a high G content at first and second sites should be matched by a high C content at third sites (and vice versa), that is, selection on non-synonymous sites would, at least in part, dictate nucleotide content at synonymous sites. At base-paired sites in mRNAs, there is a strong negative correlation in GC skew between first/second sites and third sites (for example, Pearson correlation coefficient (R) = -0.65, P = 1e-09 for GC12 skew versus GC3 skew) that is not observed at unpaired sites (R = 0.70, P = 1e-11; Additional data file 7; note that a positive correlation is expected from isochore structure [67]).
Given the potential inaccuracies of minimum free energy prediction methods (see Materials and methods), we also asked whether the above relationship is robust to the exclusion of sites at which one is less confident that base-pairing occurs (either with a particular site in the optimal structure or with any other site). GC skew is then only calculated for those sites where the probability of pairing is greater than some minimum threshold. We found that the significant negative correlation in GC skew between first/second and third sites is strikingly insensitive to different threshold values (Additional data file 8).
Jia et al. [68] recently observed that α-helices and β-sheets of protein secondary structures are preferentially 'coded' by mRNA stems. Using data on the amino acid preferences for protein conformations [69], we found G to be more abundant than C at first and second sites in both α-helices and β-sheets (GC12 skews of 0.001 and 0.0420, respectively). Similarly, there is a bias towards G in these regions within the proteins from our dataset (α-helix GC12 skew of 0.0608 ± 0.0143, P = 8e-05 for μ = 0 by one-sample t-test; β-sheet skew of 0.0879 ± 0.0312, P = 0.0102). The C preference at third sites may, therefore, reflect selection to maintain stable stems in these regions enriched for G at largely non-synonymous sites.
The location of observed synonymous substitutions is non-random with respect to mRNA stability
While randomization protocols that shuffle or swap nucleotides provide insights into how putative selection for mRNA stability and nucleotide content interact, these processes do not occur in nature. The most direct evidence that we can consider is to examine the locations of observed synonymous mutations. Reallocating point mutations is a more realistic form of analysis as it mimics the process of selection following mutation (nucleotide substitutions that are not the result of single point mutations are very rare in mammals, for example, see [70-73]). This minimizes potential biases. For example, randomization protocols that shuffle nucleotides or codons (for example, see [42]) might be problematic [74] as they generate a large number of variants in which there will be a profound effect on dinucleotide relative abundances [75-77]. Simulating the process of evolution, however, only introduces 7 to 8 synonymous changes per 100 sites, hence only about 1 to 2 per 100 nucleotides in the coding sequence. This will have negligible impact on dinucleotide distribution.
Parenthetically, as recent evidence suggests that dinucleotide content in rodent exons is the result of selection [15] and not of biased mutation and/or repair [56,75], the desirability of controlling for dinucleotide distribution is highly questionable. Put differently, if a real mRNA is on average more stable than expected when compared to simulants in which the observed point substitutions have been reallocated, biased dinucleotide distribution is more likely to be a consequence of selection for favorable base-stacking interactions rather than mutational/repair biases.
If certain mutations really were under selection because they diminished mRNA stability, relocating those substitutions actually seen to random locations ('Re-sub.') should lower stability. We used parsimony to determine the substitutions that have arisen in the mouse lineage, inferring the CDS of the rat-mouse common ancestor using hamster as the outgroup to maximize reliability and the number of informative sites (Additional data file 9). We reverted all synonymous changes back to the ancestral state and then simulated mutation by randomly reallocating substitutions at synonymous sites, maintaining the number of observed changes and the encoded protein.
Note that the application of parsimony, while a common practice in the mouse-rat comparison (for example, see [78,79]), can sometimes provide biased ancestral state reconstructions (for example, see [80]). We therefore also reconstructed rat-mouse common ancestor sequences using a maximum likelihood approach. At only 3 of 86,334 reconstructed sites did the parsimony and maximum likelihood methods disagree (excluding sites differing in all three species, see Materials and methods). All three discrepancies occurred in the same gene (Gadd45a). Exclusion of this one gene makes no difference to our results (Additional data file 2).
As nucleotide content is influenced by genomic location (isochores; for example, see [67]), the re-introduced nucleotides were selected at random, but in proportion to base composition at third sites in the appropriate mouse gene. This also further minimizes the negligible effect on dinucleotide distributions. From this randomization ('Re-sub.N3') we again find that real mRNAs are, on average, more stable than expected by chance (Table 1; Additional data files 2, 3). Ignoring the effect of isochores and changing the profile of permitted substitutions does not qualitatively alter this result. For example, allowing all mutations to occur with equal likelihood ('Re-sub.K') also shows that the locations of observed substitutions have had minimal impact on stability (Table 1; Additional data files 2, 3). Simulants and real transcripts possess a similar amount of base-pairs (P > 0.15 by one-sample t-tests on Z(%base-pairs), μ = 0; Table 1).
Signals of selection or methodological artifact?
While the above results indicate that the location at which certain synonymous mutations are observed is in part determined by constraints on mRNA stability, could the above results be artifacts of an inaccurate methodology? We have attempted to minimize such problems by considering those sequences in which a priori we expect the method to be more accurate and by considering only those sites that have a high probability of being base-paired. We can, however, consider additional tests. If selection for mRNA stability occurs, we also expect that substitution rates should be related to predicted stem-loop structure and that genes known to be under strong purifying selection should possess mRNAs with high relative stability. We examine these two predictions in turn.
Genes with a high proportion of base-pairs may have fast-evolving stems: evidence for compensatory substitutions?
Testing the first prediction, that evolutionary rates should be linked to mRNA secondary structure, is not straightforward, even if structure prediction were perfect. Although one expects that the majority of compensatory changes will occur to restore substructures, the thermodynamic hypothesis posits that some will also act to restore the overall stability of the molecule. Even if a precise secondary structure were conserved, the difficulty lies in the fact that a given substitution can only be assigned to having occurred within a stem or loop before or after it potentially affects base-pairing, for example, a transversion at a base-paired site in the ancestral mRNA will create a bulge/loop. Consequently, the only substitutions that can be observed within the same (conserved) structure of the descendant sequence are those that arise within loops with little stem-forming potential or within stems in which a compensatory substitution has restored complementary base-pairing. With this caveat in mind, we examined observed substitutions with respect to the predicted secondary structure in mouse.
We first asked whether substitution rates correlate with the percentage of sequence involved in base-pairing interactions. We found that both the number of synonymous substitutions per synonymous site (Ks) and the non-synonymous substitution rate (Ka) for the whole CDS are higher in genes with more base-pairs (Ka ρ = 0.31, P = 0.0091, N = 70; Ks ρ = 0.31, P = 0.0101, N = 69), although the result for non-synonymous mutations is sensitive to restricting analysis to the subset of small mRNAs (Additional data file 10). These effects seem to be a consequence of substitutional processes within stems. While there is a positive correlation between %base-pairs and rates within putative stems (Ka ρ = 0.31, P = 0.0090, N = 69; Ks ρ = 0.37, P = 0.0020, N = 68), no such relationship exists in loops (Ka ρ = -0.03, P = 0.7941, N = 69; Ks ρ = -0.03, P = 0.8264, N = 69; Additional data file 10).
Note that these latter correlations do not mean that stems evolve faster per se (one would predict the opposite), only that they may evolve faster when a lot of the sequence is base-paired. Indeed, consistent with stems being under purifying selection to maintain secondary structure, while non-synonymous rates are the same between codons in putative stems and those in loops (P = 0.6233, N = 69 by paired t-test, stem = 0.0110 ± 0.0018, loop = 0.0095 ± 0.0012), synonymous sites in loops evolve 37% faster than those in stems (P = 0.0045, N = 68, stem = 0.0833 ± 0.0071, loop = 0.0608 ± 0.0034; Additional data file 10).
Why might a high proportion of base-pairing be associated with rapid substitution rates within stems? One possibility is that an abundance of base-pairs ensures that no single mutation can grossly destabilize an mRNA. While one might then predict a negative correlation between %base-pairs and Z(ΔG) (that is, changes to mRNAs with little secondary structure will have a large impact on stability), this may not be observed because when substitutions are randomly reallocated the majority will not fall within stems. Alternatively, the relationship between %base-pairs and substitution rates within stems may indicate a high rate of compensatory changes restoring stem structures. Consider a mutation that arises within a stem that destabilizes the mRNA secondary structure. If selection maintains transcript stability, the substitution will only be tolerated if it is adaptive at the protein level or has such a negligible impact on stability as to be effectively neutral. In the latter case, further changes could accumulate that in combination might significantly alter structure. Under both scenarios, subsequent compensatory mutations restoring stability would thus be under positive selection. The effect of one mutation arising within a stem that has the knock-on effect of increasing substitution rates within stems would be most pronounced in genes with a high proportion of base-pairing. Consequently, compensations would be most favored when there is high pressure to maintain stability. Indeed, we find that in those genes under the strongest selective pressure for high stability, putative stems are fast-evolving (ρ = -0.37, P = 0.0020 for Z(ΔGRe-sub.N3) versus Ks, N = 68).
Housekeeping genes have high relative stability
To test the second prediction, it is necessary to define a priori a set of genes likely to be under stronger purifying selection. Prior evidence indicates that genes expressed in most tissues, housekeeping genes, may be good candidates for two reasons. First, housekeeping proteins evolve slower than tissue-specific ones [73,81-83]. Second, experimental assays of half-life have demonstrated that mRNAs of housekeeping genes degrade relatively slowly [53,54].
Here we identify housekeeping genes by calculating the breadth of expression, the proportion of tissues in which a given gene is expressed. We call a gene 'expressed' in a particular tissue if the average hybridization intensity on microarrays ('average difference' (AD)) for the transcript is greater than 100 or 200 (approximately 2 or 4 copies per cell, respectively, [84]). Housekeeping genes are those expressed in a large proportion of tissues. As described previously (for example, see [73]), we found that protein evolution is slowest in housekeeping genes (%tissues versus Ka: ρ = -0.39, P = 0.0008 for AD > 200; ρ = -0.32, P = 0.0065 for AD > 100).
Significantly, consistent with the prediction, we found that genes subject to strong purifying selection (housekeeping genes) also have the highest relative stability, with the inferred intensity of selection on mRNA stability being correlated with breadth of expression in the expected direction (ρ = -0.25, P = 0.0335 for %tissues versus Z(ΔGRe-sub.N3) at AD > 200). Using a less conservative cut-off to define a gene as expressed (AD > 100) increases the strength and significance of the correlation (ρ = -0.29, P = 0.0159). The relationship becomes more pronounced after controlling for sequence length (partial ρ = -0.25, P = 0.0179 for AD > 200; partial ρ = -0.30, P = 0.0069 for AD > 100; significance determined by 10,000 randomizations). Expression breadth is not associated with the proportion of the sequence that is base-paired (ρ = -0.01, P > 0.9 for %tissues versus %base-pairs in CDS), nor does the amount of base-pairing predict relative stability (R = -0.14, P = 0.2630 for %base-pairs versus Z(ΔGRe-sub.N3)). As suggested from the 'Swap G4C4' modification protocol, this supports the importance of overall stability over the amount of secondary structure.
Discussion
We have provided numerous lines of evidence that support the hypothesis that selection on synonymous mutations can be mediated by effects on mRNA stability in mammals. Importantly, the signature of selection in rodents, the C preference at 4-fold degenerate sites [15], can potentially be explained by selection on synonymous mutations affecting mRNA stability. That it should be C in particular (rather than A, G or T), is further explained by skews in nucleotide usage at largely non-synonymous sites: G enrichment at the first and second sites in codons is matched by C enrichment at third sites, so as to ensure, we argue, strong G:C pairs in the mRNA. Moreover, through a randomization that simulates evolution in the mouse lineage, we show that, had the observed substitutions occurred elsewhere within a sequence, they would have had a greater impact on mRNA stability. Additionally, not only do housekeeping genes have unusually low rates of protein evolution, their mRNAs have unusually high relative stability, both features being consistent with stronger selection on this class of genes. Although the structure prediction tool is by no means perfect, it is not obvious how it could be biased in such a way as to cause all our results to point towards the same conclusion.
Synonymous mutations can also be under selection for other functions. Can we be confident that these effects are independent? Recent evidence also suggests that a preference for exonic splicing enhancers (ESEs) affects codon choice [85,86] and that ESEs are under selection [87]. It is likely, however, that the results presented here and selection on ESEs are independent, as ESE hexamers are rich in G compared with C (24% and 14%, respectively, see [86] for dataset), while mRNA stability appears to explain high C content. Moreover, ESEs define relatively little sequence, being short and predominantly located within 20 nucleotides of splice junctions [87].
Experimental predictions for selection on mRNA stability
One might suppose that in silico simulations could explain variation in decay rates between genes. Z(ΔG) is not a measure of absolute stability, however, but rather of stability relative to what might have been observed given the underlying parameters of a gene, such as length and coding capacity. Only if all such parameters were equal between genes would one expect relative stability to predict decay rate. However, all else is not equal; for example, we find that Z(ΔG) and nucleotide content covary. Therefore, looking for a correlation between Z(ΔG) and half-life [56] is a weak test because an absence of a relationship would not be strong evidence against the hypothesis unless other variables could be controlled. Indeed, results are ambiguous. Mammalian housekeeping genes have longer half-lives [53,54] and we find that they also have high relative stability. In contrast, Katz and Burge [56] found no correlation between decay rate and local Z(ΔG) in yeast. The interpretation of the yeast result is made even less clear due to uncertainty over when mRNAs should be folded globally. The issue might be easier to resolve once high-quality non-human sequence from primates becomes available, as one could then compare available large-scale surveys of human mRNA decay rates (for example, see [54]) with relative stability. As hominid Ne is around an order of magnitude lower than in murids [88], however, it is also conceivable that selection may not be strong enough to act on mRNA stability.
On the other hand, simulations should predict relative decay rates of mutant versions of a given gene. In at least one case, the dopamine receptor D2 gene, it has been demonstrated that only single nucleotide polymorphisms that induce a conspicuous change in structure predicted in silico affect mRNA half-life in vitro [35]. A much larger sample set is required to determine whether this is more generally true. We predict that, for those genes with the highest relative stability, the real mRNA should have a longer half-life than the majority of mutants in which one has randomly reallocated synonymous mutations.
Implications for understanding codon usage and mutation rates
That selection maintains mRNA stability contradicts the accepted wisdom that synonymous mutations evolve neutrally [1,2], not only because changes do not alter protein sequence, but also because mammalian effective population sizes (Ne) are thought to be too small to permit selection on mutations of small effect on fitness [6]. Moreover, nucleotide content at silent sites in mammals is best predicted by genomic location (isochores; for example, see [67]). Our observations, however, nonetheless tally with recent evidence that selection acts on synonymous mutations [10-16].
Selection favoring accurate or fast protein synthesis, the classically cited functional role for biased usage of synonymous codons, is not well supported in mammals [7,61,62]. Translational selection predicts that highly expressed genes should exhibit the greatest bias in codon usage [7], but the effect is only weak [13,60,89] and a bias is also observed in lowly expressed genes [89]. On the other hand, selection for mRNA stability need not correlate with expression level (indeed, we find no relationship between Z(ΔG) and mean or peak expression level; P > 0.1 in all cases).
When translational selection is known to occur, it can be at odds with selection for mRNA secondary structure (fly, [20]) and stability (yeast, [90]), leading to a trade-off between the two forces [20,90]. Given the difficulties involved in detecting codon usage bias in mammals [7] and our results above, we infer that selection on mRNA stability must be strong relative to translational selection (if the latter occurs at all). This has two repercussions. First, selection for mRNA stability could, in principle, weaken any signal of a preferred set of codons for translational efficiency. Second, in terms of detecting selection at synonymous sites in mammals, asking whether a given amino acid always prefers a certain codon is not necessarily asking the right question. Indeed, it is quite possible that there exist no preferred codon within a gene while at the same time synonymous mutations are under selection. More generally, a complex set of trade-offs between different forms of selection and mutational biases may render interpretation of patterns of codon usage very difficult.
The evidence for selection on synonymous mutations also has implications for our understanding of both the mutation rate and the mutational load. The substitution rate at synonymous sites in exons is often used as a measure of the mutation rate [8,9]; however, this assumes neutral evolution of synonymous mutations [1,2]. By providing a parsimonious mechanism by which selection could act on synonymous sites, we can ignore the objection that prior evidence is indirect. Nevertheless, it is presently unclear to what degree synonymous mutations are favored or opposed by selection due to their effects on mRNA stability. Without being able to quantify the latter, as well as the net effect of other biases (for example, splice-associated), it will not be possible to directly estimate the extent to which use of the synonymous substitution rate leads to underestimates of the mutation rate and the mutational load.
Conclusion
Recent evidence has suggested that, despite assumptions to the contrary, synonymous mutations in mammalian exons can be under selection. Here we have provided several independent lines of evidence to support the notion that this effect may in part be mediated by selection for mRNA stability. Notably, the preference for cytosine at synonymous sites can be accounted for by such a process. Importantly, the observed substitutions appear to be present at particular sites so as to avoid affecting mRNA stability. Our results have implications for the manner in which codon usage bias should be analyzed to detect selection and for attempts to estimate the mutation rate.
Materials and methods
Orthologous rodent genes
We identified gene families from HOVERGEN (Release 44) [91,92] with complete CDSs for Mus musculus, Rattus norvegicus and hamster. Orthology was defined as the topology (((mouse, rat), hamster), non-rodent outgroup) within the phylogenetic tree for a given gene, without intervening non-rodent branches between the rodents. Seventy well-described genes matched these criteria and had a <5% size difference between the longest and shortest CDS. Non-redundancy and orthology were supported by syntenic comparisons [93]. Unless otherwise stated, N = 70 for all statistical tests.
Mouse mRNA sequences
Accession numbers from HOVERGEN were used to extract mRNAs from the EnsEMBL genome assembly (Build 30) [94]. When alternative transcripts existed, we used the rat and hamster sequences to identify the desired exons. The untranslated region (UTR) database (Release 15) [95,96] was used for six genes because the UTRs in the EnsEMBL files were unreliably annotated. If present, poly(A) tails were removed as they are coated with binding proteins and so are unlikely to be involved in base-pairing [97].
Coding sequence alignments
Each CDS was extracted using GBPARSE [98] and translated. We aligned amino acid sequences as previously described [15] then reconstructed the three-way nucleotide alignment using AA2NUC (available from L.D.H.).
Reconstruction of rat-mouse common ancestor sequence
Parsimony and maximum likelihood were used to reconstruct ancestral sequence. At 0.3% of sites in the rodent alignment, the rat-mouse ancestral state could not be determined (for example, a different base was present in each species). In these cases, we used the mouse sequence to be conservative for the number of substitutions that have occurred in the mouse lineage. Ancestral states derived from maximum likelihood were determined using codeml in the PAML package [99,100].
RNA secondary structure prediction
There are two main computational approaches to predicting RNA secondary structure. The first is a thermodynamic method, which assumes that a given sequence will fold into the structure with the minimum free energy [101]. The second approach compares multiple orthologous sequences to identify patterns of co-evolution between sites that could be indicative of compensatory mutations [102] to maintain complementary base-pairing within stems [103-108].
In the context of our analysis, the choice is highly constrained and comparative methods may not be applicable to the hypothesis we test. Comparative methods require all input sequences to be of high quality and for the alignment to be accurate. Here we are particularly interested in knowing where substitutions have occurred in a given mammalian lineage and, therefore, need sequence from three species, with mouse-rat-hamster being the obvious choice. Currently, however, rat genomic sequence is not of sufficiently high quality and annotation of UTRs is unreliable. UTRs from hamster are largely unavailable.
Although a moot point under the above circumstances, it may also be undesirable to apply a comparative method in the current context, not least because the logic would be circular: the method requires us to assume that selection is strong enough to maintain secondary structure, while at the same time we are testing for selection. More importantly, based on the evolution of non-coding RNAs, comparative methods are geared towards detecting secondary structure that has been conserved despite sequence divergence [49], that is, well-conserved substructures exist which tend to have specific functions (for example, the anti-codon within a tRNA must always be within a loop). For mRNA, however, a more realistic model is that selection favors the stability of the mRNA conformation as a whole [17,18]. Highly conserved substructures are not expected a priori [109], in part because such conservation may not always be possible, as protein-coding function should outweigh any RNA structure considerations. Essentially, the model assumes that the mRNA will adopt the optimal structure given the available sequence.
Structure and stability were predicted using RNAfold from the Vienna package (Version 1.4) [110,111] under default settings (folding at 37°C, tolerating non-Watson-Crick G:U pairs). Thermodynamic parameters were derived experimentally [112]. RNAfold implements an algorithm that, for a given RNA, finds the conformation with the minimum free energy by maximizing favorable base-pairing interactions [101].
Global versus local mRNA stability
A second methodological issue concerns whether selection might act on stability at the local or global scale. There are two critical issues when choosing which to assess. First, if opposite ends of a molecule are able to pair with one another, RNAs may adopt a conformation closer to a global optimal structure. In eukaryotes, unlike bacteria (where transcription and translation are simultaneous and co-localized), long-range interactions between opposite ends of mRNA molecules can occur [113-116]. This suggests that global [20] rather than local stability is more important to analysis of mammalian sequence.
Second, one must also ask whether the genes contain introns. Generation of a globally stable structure would require the action of spliceosome-associated helicases (for example, [117-119]) to maximize the amount of available sequence. Indeed, it is significant that intronless genes in yeast are less biased for structure than those with introns [56]. All genes in our dataset contain introns, further suggesting global stability to be the more relevant measure. Nonetheless, our assumption of global maximum stability, while an appropriate functional hypothesis, may at best only be a good approximation, as in some cases (for example, short transcripts) there may not be enough time for an mRNA to discover the most optimal structure.
Controling for sequence length
While minimum free energy predictions often agree with laboratory-based methods (for example, stem-loops are avoided at the AUG initiation codon, [120-123]), they are less reliable for long sequences (for example, [112]). The mean length of transcripts in our dataset is 2,101.41 ± 139.84 nucleotides (nt). Consequently, where relevant, we endeavored to control for length effects. In most cases, we carried out the same analyses for mRNAs shorter than 2,000 nt (N = 36, mean mRNA length of 1219.38 ± 77.32 nt), this being the cut-off defining two halves of the dataset. Through Mantell simulations, we found that, when testing for selection on stability, in no instance is the P-value for the smaller dataset both not significant and higher than that expected if one were to randomly sample half the dataset, where the full data set analysis suggested significance (Additional data file 3). Consequently we conclude that the results are not obviously biased by the inclusion of long sequence.
Protein function and secondary structure prediction
The attributes of mouse gene products were obtained from the Gene Ontology database (June 2004) [124].
Amino acid sequence was designated as occurring in α-helix, β-sheet (strand) and coil regions using PSIPRED (Version 2.3) [125,126] under default parameters (masking low complexity regions).
Rates of evolution
The number of non-synonymous substitutions per non-synonymous site (Ka) and the synonymous (Ks) distance were estimated with the Li method [127] using the Kimura 2-parameter model. We excluded one fast-evolving gene (Ka = 0.5; Ks = 0.17) in our analyses of evolutionary rates, although inclusion of the outlier gave similar results.
Coding sequence randomization protocols and statistical significance
Simulant mRNAs are identical to their real counterparts in their 5' and 3' untranslated regions and the encoded protein.
On a single-gene basis, the significance of whether its mRNA is more stable than expected by chance is given by:
R is the number of artificial mRNAs that are more stable than the real transcript, N is the number (1,000) of randomizations (see Box 1 in [128]).
The Z-score for stability is given by:
The Z-scores derived from all randomization protocols are normally distributed.
Expression
Cellular mRNA levels from normalized microarray data on Affymetrix chips were obtained from SymAtlas [129]. We identified the expression profile for each gene by BLASTing mRNA sequences against the probes for the GNF1M chip [130], which has measurements from 61 non-redundant tissue types (the five 'embryo' tissues were ignored). We used the 45-tissue dataset [84] from the U74A chip for six genes where the suggested BLAST hit from GNF1M were not syntenically feasible. For each tissue we took the mean level across replicate hybridizations. Breadth was set to 0 if AD < 50 in all tissues.
Mantell simulations
To determine whether the incorporation of long genes substantially biased our results, for each modification/randomization protocol, we considered the effect of removing the half of the dataset containing the longest genes. Given that this subset of small mRNAs is by necessity half the size of the full dataset, it is inevitable that P-values will be increased. The issue is whether they have increased more than would be expected had we randomly sampled half the dataset. To this end, we randomly extracted 36 genes and re-calculated the significance from t-tests. This was repeated 10,000 times per modification/randomization protocol, yielding the underlying distribution in P-values that would be expected were sequence length unimportant. The observed P-value (for the shortest genes) was then compared to this expected distribution (see Additional data file 3).
Additional data files
Additional data are available with the online version of this paper. Additional data file 1 contains sequences for all 70 mouse mRNAs in FASTA format. Additional data file 2 is equivalent to Table 1, but excludes the one gene (Gadd45a/HBG000516) where the rat-mouse common ancestor sequence differed slightly using the parsimony and maximum likelihood reconstructions. Additional data file 3 is equivalent to Table 1, but only considers mRNAs shorter than 2,000 nucleotides. Additional data file 4 provides the stabilities, relative stabilities and significance values for each modification/randomization on a by-gene basis. Additional data file 5 contains various sequence identifiers (for example, accession numbers) for each mouse gene. Additional data file 6 features gene ontology information, including a description of the function of each mouse gene product. Additional data file 7 contains various correlations for short genes, including GC4 skew versus Z(ΔGSh.4-fold), GC12 skew versus GC3 skew (separately for base-paired and unpaired sites) and Z(ΔGRe-sub.N3) versus Ks at base-paired sites. Additional data file 8 is a table of correlations between GC skew at first/second sites versus skew at third sites, provided for a series of thresholds where the sites analyzed must have a minimum probability of base-pairing. Additional data file 9 is a FASTA file containing three-way alignments of coding sequences from hamster, rat and mouse orthologous genes. Additional data file 10 is a table of correlations for short genes, between the proportion of base-paired sites and non-synonymous or synonymous substitution rates within the coding sequence, base-paired sites and unpaired sites.
Supplementary Material
Additional data file 1
Sequences for all 70 mouse mRNAs in FASTA format.
Click here for file
Additional data file 2
A table of the stability of mRNA secondary structures excluding the Gadd45a gene, where the rat-mouse common ancestor sequence differed slightly using the parsimony and maximum likelihood reconstructions.
Click here for file
Additional data file 3
A table of the stability of mRNA secondary structures for short genes (only considers mRNAs shorter than 2,000 nucleotides).
Click here for file
Additional data file 4
Stability and relative stability values for individual genes.
Click here for file
Additional data file 5
Sequence identifiers for each mouse gene.
Click here for file
Additional data file 6
Ontology information for each mouse gene.
Click here for file
Additional data file 7
Miscellaneous correlations for short genes, including GC4 skew versus Z(ΔGSh.4-fold), GC12 skew versus GC3 skew (separately for base-paired and unpaired sites) and Z(ΔGRe-sub.N3) versus Ks at base-paired sites.
Click here for file
Additional data file 8
A table of correlations between GC skew at first/second sites versus skew at third sites, provided for a series of thresholds where the sites analyzed must have a minimum probability of base-pairing.
Click here for file
Additional data file 9
A FASTA file containing three-way alignments of coding sequences from hamster, rat and mouse orthologous genes.
Click here for file
Additional data file 10
A table of correlations for short genes, between the proportion of base-paired sites and non-synonymous or synonymous substitution rates within the coding sequence, base-paired sites and unpaired sites.
Click here for file
Acknowledgements
We thank Csaba Pál for suggesting RNAfold, Fyodor Kondrashov and several anonymous referees for comments. We are also thankful for additional information from the various authors of the programs and databases that were used in this study. J.V.C. is funded by the UK Biotechnology and Biological Sciences Research Council.
Figures and Tables
Figure 1 Stability of mRNA secondary structures for 'Sh.4-fold' simulants relative to real transcripts. Histogram of Z-scores for ΔG, the number of standard deviations the real mRNA is away from the mean stability of the simulants, following randomizations shuffling nucleotides at 4-fold degenerate sites (1,000 randomizations per gene, N = 70). The line shows the null normal distribution.
Table 1 Stability of mRNA secondary structures
Protocol Mean ΔG P Mean Z(ΔG) Mean %pairs
Real (mouse) -737.98 ± 55.52 60.96 ± 0.28
Modification Swap G4C4 -734.10 ± 55.08 0.0169 62.11 ± 0.33
Randomization Sh.4-fold -725.76 ± 54.71 9e-15 -1.41 ± 0.14 60.77 ± 0.23
Sh.codon -728.49 ± 55.01 6e-10 -1.04 ± 0.14 60.61 ± 0.23
Re-sub.K -733.28 ± 55.15 4e-05 -0.64 ± 0.15 61.06 ± 0.24
Re-sub.N3 -734.14 ± 55.20 4e-04 -0.51 ± 0.14 61.09 ± 0.24
Means ± SEM are shown, N = 70. P-values for modifications were determined by paired t-tests (μ = Real < Modification) on ΔG. P-values for randomizations were by one-sample t-tests (expected mean (μ) = 0) on Z(ΔG). %Pairs is the proportion of the coding sequence involved in base-pairing interactions. Artificial sequences generated by the first five protocols encode the same protein as the mouse sequence. A brief description of each protocol follows (see Results for details). 'Sh.4-fold': nucleotides at all 4-fold degenerate sites are shuffled. 'Sh.codon': for each amino acid, the synonymous codons are permuted. 'Re-sub.K': synonymous substitutions are reverted back to the rat-mouse common ancestor (rat-mouse common ancestor) state, followed by reallocation of the same number of synonymous point mutations. 'Re-sub.N3': like the previous protocol, except that the nucleotide replacement is also selected at random from the nucleotide distribution at third sites observed in the mouse sequence. 'Swap G4C4': all guanine bases at 4-fold sites are replaced by cytosine, and vice versa.
==== Refs
King JL Jukes TH Non-Darwinian evolution. Science 1969 164 788 798 5767777
Kimura M Preponderance of synonymous changes as evidence for the neutral theory of molecular evolution. Nature 1977 267 275 276 865622 10.1038/267275a0
Ohta T Gillespie JH Development of Neutral and Nearly Neutral Theories. Theor Popul Biol 1996 49 128 142 8813019 10.1006/tpbi.1996.0007
Kreitman M The neutral theory is dead. Long live the neutral theory. Bioessays 1996 18 678 683 8760341 10.1002/bies.950180812
Sharp PM Averof M Lloyd AT Matassi G Peden JF DNA sequence evolution: the sounds of silence. Philos Trans R Soc Lond B Biol Sci 1995 349 241 247 8577834
Shields DC Sharp PM Higgins DG Wright F "Silent" sites in Drosophila genes are not neutral: Evidence of selection among synonymous codons. Mol Biol Evol 1988 5 704 716 3146682
Duret L Evolution of synonymous codon usage in metazoans. Curr Opin Genet Dev 2002 12 640 649 12433576 10.1016/S0959-437X(02)00353-2
Eyre-Walker A Keightley PD High genomic deleterious mutation rates in hominids. Nature 1999 397 344 347 9950425 10.1038/16915
Keightley PD Eyre-Walker A Deleterious mutations and the evolution of sex. Science 2000 290 331 333 11030650 10.1126/science.290.5490.331
Iida K Akashi H A test of translational selection at 'silent' sites in the human genome: base composition comparisons in alternatively spliced genes. Gene 2000 261 93 105 11164041 10.1016/S0378-1119(00)00482-0
Bustamante CD Nielsen R Hartl DL A maximum likelihood method for analyzing pseudogene evolution: Implications for silent site evolution in humans and rodents. Mol Biol Evol 2002 19 110 117 11752196
Hellmann I Zollner S Enard W Ebersberger I Nickel B Paabo S Selection on human genes as revealed by comparisons to chimpanzee cDNA. Genome Res 2003 13 831 837 12727903 10.1101/gr.944903
Urrutia AO Hurst LD The signature of selection mediated by expression on human genes. Genome Res 2003 13 2260 2264 12975314 10.1101/gr.641103
Keightley PD Gaffney DJ Functional constraints and frequency of deleterious mutations in noncoding DNA of rodents. Proc Natl Acad Sci USA 2003 100 13402 13406 14597721 10.1073/pnas.2233252100
Chamary JV Hurst LD Similar rates but different modes of sequence evolution in introns and at exonic silent sites in rodents: evidence for selectively-driven codon usage. Mol Biol Evol 2004 21 1014 1023 15014158 10.1093/molbev/msh087
Lu J Wu CI Weak selection revealed by the whole-genome comparison of the X chromosome and autosomes of human and chimpanzee. Proc Natl Acad Sci USA 2005 102 4063 4067 15728731 10.1073/pnas.0500436102
Fitch WM The large extent of putative secondary nucleic acid structure in random nucleotide sequences or amino acid derived messenger-RNA. J Mol Evol 1974 3 279 291 4416342 10.1007/BF01796043
Klambt D A model for messenger RNA sequences maximizing secondary structure due to code degeneracy. J Theor Biol 1975 52 57 65 1152488 10.1016/0022-5193(75)90039-9
Huynen MA Konings DA Hogeweg P Equal G and C contents in histone genes indicate selection pressures on mRNA secondary structure. J Mol Evol 1992 34 280 291 1569583 10.1007/BF00160235
Carlini DB Chen Y Stephan W The relationship between third-codon position nucleotide content, codon bias, mRNA secondary structure and gene expression in the drosophilid alcohol dehydrogenase genes Adh and Adhr. Genetics 2001 159 623 633 11606539
Okazaki Y Furuno M Kasukawa T Adachi J Bono H Kondo S Nikaido I Osato N Saito R Suzuki H Analysis of the mouse transcriptome based on functional annotation of 60,770 full-length cDNAs. Nature 2002 420 563 573 12466851 10.1038/nature01266
Kampa D Cheng J Kapranov P Yamanaka M Brubaker S Cawley S Drenkow J Piccolboni A Bekiranov S Helt G Novel RNAs identified from an in-depth analysis of the transcriptome of human chromosomes 21 and 22. Genome Res 2004 14 331 342 14993201 10.1101/gr.2094104
Mattick JS RNA regulation: a new genetics? Nat Rev Genet 2004 5 316 323 15131654 10.1038/nrg1321
Rivas E Eddy SR Secondary structure alone is generally not statistically significant for the detection of noncoding RNAs. Bioinformatics 2000 16 583 605 11038329 10.1093/bioinformatics/16.7.583
Wang HC Hickey DA Evidence for strong selective constraint acting on the nucleotide composition of 16S ribosomal RNA genes. Nucleic Acids Res 2002 30 2501 2507 12034839 10.1093/nar/30.11.2501
Bonnet E Wuyts J Rouze P Van De Peer Y Evidence that microRNA precursors, unlike other non-coding RNAs, have lower folding free energies than random sequences. Bioinformatics 2004 20 2911 2917 15217813 10.1093/bioinformatics/bth374
Meyers LA Lee JF Cowperthwaite M Ellington AD The robustness of naturally and artificially selected nucleic acid secondary structures. J Mol Evol 2004 58 681 691 15461425 10.1007/s00239-004-2590-2
Clote P Ferre F Kranakis E Krizanc D Structural RNA has lower folding energy than random RNA of the same dinucleotide frequency. RNA 2005 11 578 591 15840812 10.1261/rna.7220505
White HB 3rdLaux BE Dennis D Messenger RNA structure: compatibility of hairpin loops with protein sequence. Science 1972 175 1264 1266 5061249
Ball LA Secondary structure and coding potential of the coat protein gene of bacteriophage MS2. Nat New Biol 1973 242 44 45 4512008
Hasegawa M Yasunaga T Miyata T Secondary structure of MS2 phage RNA and bias in code word usage. Nucleic Acids Res 1979 7 2073 2079 537920
Konecny J Schoniger M Hofacker I Weitze MD Hofacker GL Concurrent neutral evolution of mRNA secondary structures and encoded proteins. J Mol Evol 2000 50 238 242 10754066
Pedersen JS Forsberg R Meyer IM Hein J An evolutionary model for protein-coding regions with conserved RNA structure. Mol Biol Evol 2004 21 1913 1922 15229291 10.1093/molbev/msh199
Shen LX Basilion JP Stanton VP Jr Single-nucleotide polymorphisms can cause different structural folds of mRNA. Proc Natl Acad Sci USA 1999 96 7871 7876 10393914 10.1073/pnas.96.14.7871
Duan J Wainwright MS Comeron JM Saitou N Sanders AR Gelernter J Gejman PV Synonymous mutations in the human dopamine receptor D2 (DRD2) affect mRNA stability and synthesis of the receptor. Hum Mol Genet 2003 12 205 216 12554675 10.1093/hmg/ddg055
Duan J Antezana MA Mammalian mutation pressure, synonymous codon choice, and mRNA degradation. J Mol Evol 2003 57 694 701 14745538 10.1007/s00239-003-2519-1
Capon F Allen MH Ameen M Burden AD Tillman D Barker JN Trembath RC A synonymous SNP of the corneodesmosin gene leads to increased mRNA stability and demonstrates association with psoriasis across diverse ethnic groups. Hum Mol Genet 2004 13 2361 2368 15333584 10.1093/hmg/ddh273
Eichler DC Eales SJ The effect of RNA secondary structure on the action of a nucleolar endoribonuclease. J Biol Chem 1983 258 10049 10053 6193106
Hambraeus G Karhumaa K Rutberg B A 5' stem-loop and ribosome binding but not translation are important for the stability of Bacillus subtilis aprE leader mRNA. Microbiology 2002 148 1795 1803 12055299
Beutler E Gelbart T Han JH Koziol JA Beutler B Evolution of the genome and the genetic code: selection at the dinucleotide level by methylation and polyribonucleotide cleavage. Proc Natl Acad Sci USA 1989 86 192 196 2463621
Qiu L Moreira A Kaplan G Levitz R Wang JY Xu C Drlica K Degradation of hammerhead ribozymes by human ribonucleases. Mol Gen Genet 1998 258 352 362 9648739 10.1007/s004380050741
Seffens W Digby D mRNAs have greater negative folding free energies than shuffled or codon choice randomized sequences. Nucleic Acids Res 1999 27 1578 1584 10075987 10.1093/nar/27.7.1578
Cohen B Skiena S Natural selection and algorithmic design of mRNA. J Comput Biol 2003 10 419 432 12935336 10.1089/10665270360688101
Furtig B Richter C Wohnert J Schwalbe H NMR spectroscopy of RNA. Chembiochem 2003 4 936 962 14523911 10.1002/cbic.200300700
Gardner PP Giegerich R A comprehensive comparison of comparative RNA structure prediction approaches. BMC Bioinformatics 2004 5 140 15458580 10.1186/1471-2105-5-140
Dreyfuss G Kim VN Kataoka N Messenger-RNA-binding proteins and the messages they carry. Nat Rev Mol Cell Biol 2002 3 195 205 11994740 10.1038/nrm760
Herschlag D RNA chaperones and the RNA folding problem. J Biol Chem 1995 270 20871 20874 7545662
Schroeder R Barta A Semrad K Strategies for RNA folding and assembly. Nat Rev Mol Cell Biol 2004 5 908 919 15520810 10.1038/nrm1497
Morgan SR Higgs PG Evidence for kinetic effects in the folding of large RNA molecules. J Chem Physics 1996 105 7152 7157 10.1063/1.472517
Schroeder R Grossberger R Pichler A Waldsich C RNA folding in vivo. Curr Opin Struct Biol 2002 12 296 300 12127447 10.1016/S0959-440X(02)00325-1
Meyer IM Miklos I Co-transcriptional folding is encoded within RNA genes. BMC Mol Biol 2004 5 10 15298702 10.1186/1471-2199-5-10
Kozak M Pushing the limits of the scanning mechanism for initiation of translation. Gene 2002 299 1 34 12459250 10.1016/S0378-1119(02)01056-9
Hollams EM Giles KM Thomson AM Leedman PJ mRNA stability and the control of gene expression: implications for human disease. Neurochem Res 2002 27 957 980 12462398 10.1023/A:1020992418511
Yang E van Nimwegen E Zavolan M Rajewsky N Schroeder M Magnasco M Darnell JE Jr Decay rates of human mRNAs: correlation with functional characteristics and sequence attributes. Genome Res 2003 13 1863 1872 12902380 10.1101/gr.997703
Shpaer EG The secondary structure of mRNAs from Escherichia coli: its possible role in increasing the accuracy of translation. Nucleic Acids Res 1985 13 275 288 3889832
Katz L Burge CB Widespread selection for local RNA secondary structure in coding regions of bacterial genes. Genome Res 2003 13 2042 2051 12952875 10.1101/gr.1257503
Le SV Chen JH Currey KM Maizel JV Jr A program for predicting significant RNA secondary structures. Comput Appl Biosci 1988 4 153 159 2454711
Ikemura T Codon usage and tRNA content in unicellular and multicellular organisms. Mol Biol Evol 1985 2 13 34 3916708
Akashi H Eyre-Walker A Translational selection and molecular evolution. Curr Opin Genet Dev 1998 8 688 693 9914211 10.1016/S0959-437X(98)80038-5
Comeron JM Selective and mutational patterns associated with gene expression in humans: Influences on synonymous composition and intron presence. Genetics 2004 167 1293 1304 15280243 10.1534/genetics.104.026351
Kanaya S Yamada Y Kinouchi M Kudo Y Ikemura T Codon usage and tRNA genes in eukaryotes: correlation of codon usage diversity with translation efficiency and with CG-dinucleotide usage as assessed by multivariate analysis. J Mol Evol 2001 53 290 298 11675589 10.1007/s002390010219
dos Reis M Savva R Wernisch L Solving the riddle of codon usage preferences: a test for translational selection. Nucleic Acids Res 2004 32 5036 5044 15448185 10.1093/nar/gkh834
Vinogradov AE Bendable genes of warm-blooded vertebrates. Mol Biol Evol 2001 18 2195 2200 11719569
Versteeg R van Schaik BD van Batenburg MF Roos M Monajemi R Caron H Bussemaker HJ van Kampen AH The human transcriptome map reveals extremes in gene density, intron length, GC content, and repeat pattern for domains of highly and weakly expressed genes. Genome Res 2003 13 1998 2004 12915492 10.1101/gr.1649303
Green P Ewing B Miller W Thomas PJ Green ED Transcription-associated mutational asymmetry in mammalian evolution. Nat Genet 2003 33 514 517 12612582 10.1038/ng1103
Majewski J Dependence of mutational asymmetry on gene-expression levels in the human genome. Am J Hum Genet 2003 73 688 692 12881777 10.1086/378134
Eyre-Walker A Hurst LD The evolution of isochores. Nat Rev Genet 2001 2 549 555 11433361 10.1038/35080577
Jia M Luo L Liu C Statistical correlation between protein secondary structure and messenger RNA stem-loop structure. Biopolymers 2004 73 16 26 14691936 10.1002/bip.10496
Creighton TE Proteins: Structure and Molecular Properties 1993 2 New York: WH Freeman
Silva JC Kondrashov AS Patterns in spontaneous mutation revealed by human-baboon sequence comparison. Trends Genet 2002 18 544 547 12414177 10.1016/S0168-9525(02)02757-9
Kondrashov AS Direct estimates of human per nucleotide mutation rates at 20 loci causing Mendelian diseases. Hum Mutat 2003 21 12 27 12497628 10.1002/humu.10147
Smith NGC Webster MT Ellegren H A low rate of simultaneous double-nucleotide mutations in primates. Mol Biol Evol 2003 20 47 53 12519905 10.1093/molbev/msg003
Lercher MJ Chamary JV Hurst LD Genomic regionality in rates of evolution is not explained by clustering of genes of comparable expression profile. Genome Res 2004 14 1002 1013 15173108 10.1101/gr.1597404
Workman C Krogh A No evidence that mRNAs have lower folding free energies than random sequences with the same dinucleotide distribution. Nucleic Acids Res 1999 27 4816 4822 10572183 10.1093/nar/27.24.4816
Karlin S Mrazek J What drives codon choices in human genes? J Mol Biol 1996 262 459 472 8893856 10.1006/jmbi.1996.0528
Karlin S Global dinucleotide signatures and analysis of genomic heterogeneity. Curr Opin Microbiol 1998 1 598 610 10066522 10.1016/S1369-5274(98)80095-7
Gentles AJ Karlin S Genome-scale compositional comparisons in eukaryotes. Genome Res 2001 11 540 546 11282969 10.1101/gr.163101
Bazykin GA Kondrashov FA Ogurtsov AY Sunyaev S Kondrashov AS Positive selection at sites of multiple amino acid replacements since the mouse-rat divergence. Nature 2004 429 558 562 15175752 10.1038/nature02601
Jordan IK Kondrashov FA Adzhubei IA Wolf YI Koonin EV Kondrashov AS Sunyaev S A universal trend of amino acid gain and loss in protein evolution. Nature 2005 433 633 638 15660107 10.1038/nature03306
Eyre-Walker A Problems with parsimony in sequences of biased base composition. J Mol Evol 1998 47 686 690 9847410
Duret L Mouchiroud D Determinants of substitution rates in mammalian genes: Expression pattern affects selection intensity but not mutation rate. Mol Biol Evol 2000 17 68 74 10666707
Williams EJB Hurst LD The proteins of linked genes evolve at similar rates. Nature 2000 407 900 903 11057667 10.1038/35038066
Zhang L Li WH Mammalian housekeeping genes evolve more slowly than tissue-specific genes. Mol Biol Evol 2004 21 236 239 14595094 10.1093/molbev/msh010
Su AI Cooke MP Ching KA Hakak Y Walker JR Wiltshire T Orth AP Vega RG Sapinoso LM Moqrich A Large-scale analysis of the human and mouse transcriptomes. Proc Natl Acad Sci USA 2002 99 4465 4470 11904358 10.1073/pnas.012025199
Willie E Majewski J Evidence for codon bias selection at the pre-mRNA level in eukaryotes. Trends Genet 2004 20 534 538 15475111 10.1016/j.tig.2004.08.014
Chamary JV Hurst LD Biased codon usage near intron-exon junctions: selection on splicing enhancers, splice-site recognition or something else? Trends Genet 2005 21 256 259 15851058 10.1016/j.tig.2005.03.001
Fairbrother WG Holste D Burge CB Sharp PA Single nucleotide polymorphism-based validation of exonic splicing enhancers. PLoS Biol 2004 2 E268 15340491 10.1371/journal.pbio.0020268
Keightley PD Lercher MJ Eyre-Walker A Evidence for widespread degradation of gene control regions in hominid genomes. PLoS Biol 2005 3 e42 15678168 10.1371/journal.pbio.0030042
Lavner Y Kotlar D Codon bias as a factor in regulating expression via translation rate in the human genome. Gene 2005 345 127 138 15716084 10.1016/j.gene.2004.11.035
Carlini DB Context-dependent codon bias and mRNA longevity in the yeast transcriptome. Mol Biol Evol 2005 22 1403 1411 15772378 10.1093/molbev/msi135
Duret L Mouchiroud D Gouy M HOVERGEN - a database of homologous vertebrate genes. Nucleic Acids Res 1994 22 2360 2365 8036164
HOVERGEN
LocusLink
EnsEMBL Mouse Genome Server
Pesole G Liuni S Grillo G Licciulli F Mignone F Gissi C Saccone C UTRdb and UTRsite: specialized databases of sequences and functional elements of 5' and 3' untranslated regions of eukaryotic mRNAs. Update 2002. Nucleic Acids Res 2002 30 335 340 11752330 10.1093/nar/30.1.335
The UTR Database
Keller RW Kuhn U Aragon M Bornikova L Wahle E Bear DG The nuclear poly(A) binding protein, PABP2, forms an oligomeric particle covering the length of the poly(A) tail. J Mol Biol 2000 297 569 583 10731412 10.1006/jmbi.2000.3572
GBPARSE
Yang Z PAML: a program package for phylogenetic analysis by maximum likelihood. Comput Appl Biosci 1997 13 555 556 9367129
Phylogenetic Analysis by Maximum Likelihood
Zuker M Stiegler P Optimal computer folding of large RNA sequences using thermodynamics and auxiliary information. Nucleic Acids Res 1981 9 133 148 6163133
Kimura M The role of compensatory neutral mutations in molecular evolution. J Genet 1985 64 7 19
Dixon MT Hillis DM Ribosomal RNA secondary structure: compensatory mutations and implications for phylogenetic analysis. Mol Biol Evol 1993 10 256 267 8450759
Stephan W The rate of compensatory evolution. Genetics 1996 144 419 426 8878705
Higgs PG Compensatory neutral mutations and the evolution of RNA. Genetica 1998 102-103 91 101 9720274 10.1023/A:1017059530664
Chen Y Carlini DB Baines JF Parsch J Braverman JM Tanda S Stephan W RNA secondary structure and compensatory evolution. Genes Genet Sys 1999 74 271 286 10.1266/ggs.74.271
Innan H Stephan W Selection intensity against deleterious mutations in RNA secondary structures and rate of compensatory nucleotide substitutions. Genetics 2001 159 389 399 11560913
Savill NJ Hoyle DC Higgs PG RNA sequence evolution with secondary structure constraints: comparison of substitution rate models using maximum-likelihood methods. Genetics 2001 157 399 411 11139520
Buratti E Baralle FE Influence of RNA secondary structure on the pre-mRNA splicing process. Mol Cell Biol 2004 24 10505 10514 15572659 10.1128/MCB.24.24.10505-10514.2004
Hofacker IL Fontana W Stadler PF Bonhoeffer LS Tacker M Schuster P Fast folding and comparison of RNA secondary structures. Monatshefte fur Chimie 1994 125 167 188 10.1007/BF00818163
The Vienna RNA Package
Mathews DH Sabina J Zuker M Turner DH Expanded sequence dependence of thermodynamic parameters improves prediction of RNA secondary structure. J Mol Biol 1999 288 911 940 10329189 10.1006/jmbi.1999.2700
Konings DA van Duijn LP Voorma HO Hogeweg P Minimal energy foldings of eukaryotic mRNAs form a separate leader domain. J Theor Biol 1987 127 63 78 3669684
Doktycz MJ Larimer FW Pastrnak M Stevens A Comparative analyses of the secondary structures of synthetic and intracellular yeast MFA2 mRNAs. Proc Natl Acad Sci USA 1998 95 14614 14621 9843938 10.1073/pnas.95.25.14614
Parsch J Tanda S Stephan W Site-directed mutations reveal long-range compensatory interactions in the Adh gene of Drosophila melanogaster. Proc Natl Acad Sci USA 1997 94 928 933 9023359 10.1073/pnas.94.3.928
Parsch J Stephan W Tanda S Long-range base pairing in Drosophila and human mRNA sequences. Mol Biol Evol 1998 15 820 826 9656483
Hamm J Lamond AI Spliceosome assembly: the unwinding role of DEAD-box proteins. Curr Biol 1998 8 R532 R534 9705927 10.1016/S0960-9822(07)00340-5
Wang Y Wagner JD Guthrie C The DEAH-box splicing factor Prp16 unwinds RNA duplexes in vitro. Curr Biol 1998 8 441 451 9550699 10.1016/S0960-9822(98)70178-2
Rocak S Linder P DEAD-box proteins: the driving forces behind RNA metabolism. Nat Rev Mol Cell Biol 2004 5 232 241 14991003 10.1038/nrm1335
Kozak M Influences of mRNA secondary structure on initiation by eukaryotic ribosomes. Proc Natl Acad Sci USA 1986 83 2850 2854 3458245
de Smit MH van Duin J Secondary structure of the ribosome binding site determines translational efficiency: a quantitative analysis. Proc Natl Acad Sci USA 1990 87 7668 7672 2217199
Ganoza MC Louis BG Potential secondary structure at the translational start domain of eukaryotic and prokaryotic mRNAs. Biochimie 1994 76 428 439 7849110 10.1016/0300-9084(94)90120-1
Rocha EP Danchin A Viari A Translation in Bacillus subtilis: roles and trends of initiation and termination, insights from a genome analysis. Nucleic Acids Res 1999 27 3567 3576 10446248 10.1093/nar/27.17.3567
The Gene Ontology Database
Jones DT Protein secondary structure prediction based on position-specific scoring matrices. J Mol Biol 1999 292 195 202 10493868 10.1006/jmbi.1999.3091
The PSIPRED Protein Structure Prediction Server
Li WH Unbiased estimation of the rates of synonymous and nonsynonymous substitution. J Mol Evol 1993 36 96 99 8433381
Hurst LD Pal C Lercher MJ The evolutionary dynamics of eukaryotic gene order. Nat Rev Genet 2004 5 299 310 15131653 10.1038/nrg1319
SymAtlas
Su AI Wiltshire T Batalov S Lapp H Ching KA Block D Zhang J Soden R Hayakawa M Kreiman G A gene atlas of the mouse and human protein-encoding transcriptomes. Proc Natl Acad Sci USA 2004 101 6062 6067 15075390 10.1073/pnas.0400782101
|
16168082
|
PMC1242210
|
CC BY
|
2021-01-04 16:05:41
|
no
|
Genome Biol. 2005 Aug 16; 6(9):R75
|
utf-8
|
Genome Biol
| 2,005 |
10.1186/gb-2005-6-9-r75
|
oa_comm
|
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-9-r761616808310.1186/gb-2005-6-9-r76ResearchA molecular map of mesenchymal tumors Henderson Stephen R [email protected] David [email protected] Nadege [email protected] Sean [email protected] Richard [email protected] Sonja [email protected] John [email protected] Neil [email protected] Jeremy [email protected] Nick [email protected] Adrienne M [email protected] Chris [email protected] Cancer Research UK, Viral Oncology Group, Wolfson Institute for Biomedical Research, Gower Street, University College London, London, WC1E 6BT, UK2 Division of Cell and Molecular Biology, Biochemistry Building, Faculty of Life Sciences, Imperial College London, South Kensington Campus, London, SW7 2AZ, UK3 Institute of Orthopaedics and Department of Pathology, Royal National Orthopaedic Hospital, Stanmore, Middlesex, HA7 4LP, UK4 Unit of Molecular Haematology and Cancer Biology, Institute of Child Health and Great Ormond Street Hospital, Guildford Street, London, WC1N 1EH, UK5 Department of Pathology, Great Ormond Street Hospital for Children, London, WC1N 3JH, UK6 London Bone and Soft Tissue Tumour Service, University College London Hospitals, London, UK7 Department of Pathology, Nuffield Department of Orthopaedic Surgery, Nuffield Orthopaedic Centre, Headington, Oxford, OX3 7LD, UK2005 26 8 2005 6 9 R76 R76 31 3 2005 7 6 2005 26 7 2005 Copyright © 2005 Henderson et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
A comprehensive study of the gene expression profile of 96 mesenchymal tumors identifies molecular fingerprints for most tumors in this group.
Background
Bone and soft tissue tumors represent a diverse group of neoplasms thought to derive from cells of the mesenchyme or neural crest. Histological diagnosis is challenging due to the poor or heterogenous differentiation of many tumors, resulting in uncertainty over prognosis and appropriate therapy.
Results
We have undertaken a broad and comprehensive study of the gene expression profile of 96 tumors with representatives of all mesenchymal tissues, including several problem diagnostic groups. Using machine learning methods adapted to this problem we identify molecular fingerprints for most tumors, which are pathognomonic (decisive) and biologically revealing.
Conclusion
We demonstrate the utility of gene expression profiles and machine learning for a complex clinical problem, and identify putative origins for certain mesenchymal tumors.
==== Body
Background
Tumors of bone and soft tissue are a wide spectrum of benign and malignant neoplasms (sarcoma) derived from mesenchymal precursor cells (hereafter referred to as mesenchymal tumors) [1,2]. Many show heterogeneous patterns of differentiation or exhibit little similarity to differentiated mesenchymal tissues, while others have diverse cellular morphology (pleomorphism). Thus, specialist expertise is required for diagnosis as the histopathology of mesenchymal tumors is often overlapping or indistinct. With the introduction of neo-adjuvant cytotoxic therapies, diagnosis has become even more challenging as pathologists must rely increasingly upon needle core biopsies that produce only small quantities of tissue for immunohistochemistry and histopathological diagnosis. Furthermore, molecular therapies have been developed targeting oncogenic pathways that may transcend the current histopathological categories.
The discovery of definitive oncogenic gene fusions for certain mesenchymal tumors has aided pathologists greatly. These include the EWS-ERG or EWS-FLI1 fusion transcripts for Ewing's sarcoma (EWS) [3-5], or the SYT-SSX fusion transcript for synovial sarcoma [6,7]. Also, reverse transcriptase polymerase chain reaction (RT-PCR) has become the gold standard for diagnosis. These 'simple sarcomas' are ideal candidates for targeted therapy, with relatively stable karyotypes and stereotypical molecular pathology [8]. Nonetheless, chromosomal translocations are confirmatory for only a fraction of mesenchymal tumors while those with complex karyotypes remain diagnostically challenging.
There are many gene expression microarray (GEM) studies covering a range of mesenchymal tumors [9-28]. These studies have proved the general applicability of GEM for the diagnosis of mesenchymal tumors. Yet each study compares a fraction of the mesenchymal tumors in isolation. They do not address the spectra of disease nor the challenges of diagnostic pathology, which deals with many confounding diagnoses. Here we present a more comprehensive study encompassing representative tumors derived from all mesenchymal tissue types, as well as more poorly differentiated tumors.
Using a machine learning model, we assess the overall utility of GEM for the diagnosis of 19 types of mesenchymal tumors. We find both expected and unexpected relationships between certain mesenchymal tumors, and ascertain expression fingerprints for decisive diagnostic or pathognomonic features for most tumor classes. These fingerprints give clues to the etiology of several mesenchymal tumors.
Our machine learning model fits the data in two steps. The first step incorporates biological assumptions by merging related tumors into broader groups. Step two splits the broader groups into specific tumors. This latter process is derived by expert decision, rather than automated model selection. We consciously mirror the clinical diagnostic method of progressively resolving general differentiation types into specific tumors in a branch-wise (or tree-like) method. The use of a decision process structured by prior knowledge (together with a well-suited feature selection algorithm) gives our model an estimated error of about 0.1. Although further study and validation will be required to bring it to clinical use, we believe this method is a prototype for the extension of GEM and machine learning to other complex diagnostic problems. The GEM data from this study are available from the European Bioinformatics Institute (EBI) public repository ArrayExpress (accession no. E-MEXP-353) [29].
Results
We determined the gene expression profile of 96 mesenchymal tumors, representing 19 different sub-types, using the Affymetrix HG-U133Av2 oligonucleotide GeneChip®. A multi-dimensional scaling (MDS) of tumor samples using all genes reveals much structure in the data (Figure 1). We chose MDS rather than hierarchical clustering for a more compact and comprehensible summary. An average linkage hierarchical clustering using the same distance matrix is given in Additional data file 1 online.
The differentiated tumors largely cluster in groups reflecting common tissue types. Examples are the neurofibroma (NFB) and schwannoma (SWN), the alveolar (ARMS) and embryonal rhabdomyosarcoma (ERMS), and the well-differentiated liposarcoma (WLS) and lipoma (LMA). Similarities in the GEM profiles of malignant peripheral nerve sheath tumors (MPNST) and monophasic synovial sarcoma (MSS) were previously reported and this relationship is supported in our data [18]. Likewise, the relation between the small round blue cell tumors: EWS, ARMS and ERMS is clear from this plot. An interesting finding was the similarity of the chordoma (CMA), a rare tumor arising along the midline to the chondrosarcoma (CHS). Reflecting the experience of histopathological diagnosis, the osteosarcoma (OS), pleomorphic sarcoma (PMS) and leiomyosarcoma (LMS) show the greatest diversity being dispersed throughout the other samples.
The MDS is based on distances calculated from the whole molecular signature (all genes or probesets) (Figure 1). Thus, confounding factors such as intercalating tissue, tumor site or patient age and sex may affect the resultant pattern.
Supervised learning
The key to the success of supervised learning is to focus on a small set of strongly distinguishing features, thus ignoring confounding factors. Strong feature selection is appropriate in this study as we have sound histological evidence that there are underlying patterns to be uncovered (at least for most tumors). This reduced set of features (or genes) gives a molecular fingerprint of the tumors that is both diagnostic and descriptive. Our feature selection method is suited to this application as it selects ten genes evenly for each class. Using a simple k = nearest neighbors (k-NN) machine learning algorithm and the extended feature selection algorithm we attempted initially to create a model for all tumors simultaneously (see Materials and methods section for more detail on feature selection and k-NN algorithm). The overall cross-validation error was 0.33 compared with a random guessing error of 0.96 +/- S.E. 0.01 (see Materials and methods: classification algorithm). Therefore, even with this simplest approach, the machine learning model is fairly successful.
The true error, and thus the generalization to other datasets, is likely to be underestimated (see Discussion). However, an important control shows that our method does not over-fit the data excessively. In simple terms, the method does not erroneously fit this particular dataset using random patterns. After randomly permuting all the labels to create a semi-random dataset we tried to cross-validate these false data (see Materials and methods: classification algorithm). The error was 0.94 +/- S.E. 0.01 (n = 10 permutations) so was therefore not significantly better than random guessing (Table 1).
Two-step model
The majority of errors from the simple model above were as expected between the adipocytic tumors LMA, WLS and myxoid liposarcoma (MLS), between MPNST and MSS, and between LMS, PMS and de-differentiated chondrosarcoma (DCS). By considering unsupervised learning methods (MDS and hierarchical clustering) plus biological assumptions and trial-and-error, we arrived at an improved two-step method. This was designed, in part, to mirror the thought processes involved during clinical histopathological diagnosis. This two-step model is, thus, expert derived and not an automated procedure.
In step one, we grouped tumors into composite classes assumed to reflect similar pathways or levels of differentiation; MPNST/MSS, PMS/LMS/DCS, ARMS/ERMS, NFB/SWN and LMA/WLS/MLS. In step two, we decomposed the composite classes into their sub-types. This decision process is illustrated in color code in Figure 2.
Using the same feature selection and learning algorithm as the initial method, the new model is remarkably effective in the first step with ~0.09 errors (9/96 errors). There were important limitations, however, as we could not decompose the adipocytic tumors fully in step two, yet recognized MLS as distinct. Moreover, after compositing the spindle-like tumors PMS/LMS/DCS in step one, we could not decompose this artificial grouping. Accepting these limitations, the sum error over both steps was 0.14 (n = 9 errors in step one, and n = 4 errors in step two). Note that the samples wrongly classified in step two were classified correctly in step one thus having no carry-over error in this particular dataset. The types of error in step one are summarized in Table 2, and the error for each component of step two is shown in Table 3.
Molecular fingerprints
Feature selection of genes was carried out for each fold of our model, and while 237 distinct genes were encountered during cross-validation only 61 were selected in every fold (96 folds for leave-one-out cross-validation, see Materials and methods: feature selection). This robust 61-gene signature is illustrated in the gene-clustered heatmap of Figure 3. The adipocytic tumors (LMA/WLS/MLS), rhabdomyosarcoma, NFB/SWN, EWS, CMA, desmoid fibromatosis (FMT), and to a lesser extent MSS and MPNST have clear signatures. The CHS, chondromyxoid fibroma (CMF) and OS have less distinct patterns. However, the poorly differentiated spindle-like tumors have no distinct pattern. Yet this does not prevent our classification algorithm from recognizing the spindle-like tumor group as a whole. Negative expression of all other markers by the spindle-like tumors is informative to the k-NN algorithm just as it would be to a histopathologist. Additional data file 2 (sheet a) provides the gene information for Figure 3, including probeset identifications, full names, accession numbers and references describing their relation to respective pathways of differentiation and/or involvement in cancer.
From the molecular fingerprint used in step two we selected genes used in the majority of cross-validation folds for display in Figure 4. Again these genes are fully described in Additional data file 2 (sheet b) online and some are discussed below.
Discussion
We demonstrate here the feasibility of using GEM and machine learning to aid the diagnosis of mesenchymal tumors. In contrast to previous studies, our work encompasses a wide range of mesenchymal neoplasms and reflects the open-ended nature of histopathological diagnosis. It is clear from unsupervised methods such as MDS that there is information within the expression profile that can be further refined (Figure 1). This complex problem is surprisingly tractable firstly because of the large number of definitive gene markers associated with many of the tumors, especially the well-differentiated groups. In machine learning terminology we may state that these genes (or features) have no overlap in their class distributions, while in histopathological parlance their expression is pathognomonic. This would be rare in prognostic studies, for instance, as certain patients with good molecular signatures may still have poor outcomes due to unobserved factors. Secondly, we have used a novel machine learning strategy breaking this complex problem into soluble steps. We have implicitly admitted limitations in our model by not attempting to further decompose groups such as LMS/PMS/DCS (spindle-like tumors) and WLS/LMA. Thirdly, we have used a feature selection strategy that captures information evenly on all the tumor groups. Further validation on prospective samples is required before GEM studies are of clinical use for the diagnosis of mesenchymal tumors.
Cross-validation of the first step of our model gives 0.09 errors. This is a useful guide but its accuracy should not be overstated. A reliable estimate of error usually requires hundreds of samples per class, and a leave-one-out estimation is likely to underestimate error [30]. Moreover, there is the problem of statistically uncontrollable bias in our wide but shallowly sampled dataset [31]. However, our model is not grossly over-fitted (see Table 2) and more importantly the generalization to other datasets is clear from the biology of the molecular fingerprints (arising from feature selection). These fingerprints may be useful as the basis of a custom diagnostic chip and also provide insight into the etiology and cell of origin of certain tumors (see Additional data file 2 (sheets a and b online).
Several of the fingerprints are clearly related to the metabolism or function of differentiated mesenchymal tissues. For instance, adipocytic tumors are identifiable by their lipid-associated genes such as perilipin (PLIN), lipoprotein lipase (LPL), and glycerol-3-phosphate dehydrogenase 1 (GPD1) (Figure 3a). Similarly, rhabdomyosarcoma are characterized by genes such as cholinergic receptor alpha (CHRNA1) and receptor-associated protein of the synapse (RAPSN) associated with the musculoskeletal synaptic junction (see Figure 3b and Additional data file 2 (sheet a) online for supporting references).
When resolving these classes further in a second classification step some of the fingerprints reflect degrees of differentiation. Thus, the ARMS is defined by its higher expression of tropomyosin-2 (TPM2), skeletal muscle actin (ACTA1) and myosin-light polypeptide-4 (MYL4), indicative of more maturity than the ERMS (see Figure 4a and 4b and supporting references in Additional data file 2 (sheet b) online).
Within these highly focused signatures there are clues to the origins of some of the poorly differentiated tumors. One striking similarity of gene expression is between the MPNST and the MSS (see Additional data file 1). The MSS have a simple karyotype consisting of an aberrant SYT-SSX fusion. The MPNST have a complex karyotype and are believed to arise from Schwann cell precursors (they are more malignant than SWN and frequently associated with NF1 mutations [32]) derived from the neural crest rather than the mesenchyme [12]. Both MPNST and MSS are aggressive and poorly differentiated. Their composite signature (Figure 3g) contains the gene Endothelin-3 (EDN3), a key molecule in the development of the neural crest. EDN3 promotes self-renewal of multi-potent neural crest precursors or de-differentiation of matured cell types including Schwann cells [33-35].
Despite their similarity, the MPNST and MSS have distinct molecular signatures (Figure 4c and 4d) as do the NFB and SWN (Figure 4e and 4f), which are typically difficult to distinguish immunohistochemically. For instance, SWN but not NFB express the secreted glycoprotein WNT5A recently linked to metastases of OS [36] and melanoma [37]. Yet SWN is a benign tumor that never metastasizes.
Some genes could make useful monoclonal antibodies for immunohistochemistry. Genes such as the myxoid liposarcoma-associated transcript-4 (MLAT4), which as its name suggests distinguishes MLS from WLS/LMA (Figure 4g and 4h), appear to have been identified in molecular screens but not pursued as markers thus far. The overall expression of MLS is intriguing as it shows similarities to the putative neural crest tumors MPNST and MSS and the small round blue cell tumors ERMS, ARMS and EWS. Similarly, the fingerprint itself contains early mesenchymal and neural development genes (EMX2, SOX11), and neural restricted genes (SHANK2; a post-synaptic molecule). The profile also confirms previous reports of the expression of the immunotherapeutic targets PRAME and CTAG-1 (also known as NY-ESO-1) [38]. These are so-called 'germ cell antigens' because of their restriction to the testes of healthy males.
There is another clue to cellular origin within the CMA fingerprint. The overall expression pattern of CMA is closely related to the chondroblastic tumors CHS and chondroblastoma (CHB) (Figure 1). Likewise, the CMA are known to commonly contain focal regions of chondroid differentiation or more rarely chondrosarcomatous elements. As the tumor occurs solely along the midline it is proposed to originate from remnant notochord, an embryonic structure that is known to persist at least into infancy [39,40]. The expression of the brachyury (T) which is highly expressed in developing notochord strongly supports this theory [41,42].
Not all tumors had distinctive markers. The LMS, PMS and DCS were not distinguishable from each other. Cross-validation of our model in step one was fairly successful as our algorithm collectively recognized these tumors through a lack of specific markers, perhaps analogous to the way in which histopathologists recognize them (Figure 3l). At least part of the success of our model in step one is that it incorporated this uncertainty implicitly by compositing these tumors into a single 'spindle-like' group.
There are a number of extensions to the current study that may bring GEM technology closer to clinical use. Firstly, a larger study focusing on the poorly modeled cases (PMS/LMS/DCS) may help to elucidate categories supplementing current histological guidelines. Yet the pleomorphism and complex karyotypes of some mesenchymal tumors may confer a continuum of molecular pathology irreducible to simple categories. Such an analysis is unlikely to find adherents if it does not correspond with an improved interpretation of the histopathology or with appreciable clinical differences such as prognosis.
Secondly, by sampling all groups more thoroughly we could manage uncertainty. Our model gives a simple prediction of tumor type appropriate for the broad base and low sample numbers in our dataset. A more nuanced approach would be to calculate a probability of tumor type. This could be achieved most simply using Bayesian classifiers that implicitly calculate class membership probability. Thus in a clinical setting a sarcoma could be identified to a histopathologist as a central or classical example of its type, or be highlighted as an uncertain outlier for further immunohistochemical examination.
Thirdly, there is a large quantity of relevant molecular information that may transcend the diagnostic categories investigated here, yet may have prognostic significance. There is great interest, for instance, in both drug-resistance genes [43] and signatures associated with metastasis [44], which are common to a number of different tumor groups. Finally, a custom chip and user-friendly software that incorporates both a predictive model and such additional knowledge needs to be developed. This package might incorporate a prediction algorithm, visualization tools, plus additional housekeeping genes to aid normalization and quality controls. The molecular fingerprints shown here and fully listed in the additional data files (Figure 3 and 4 and Additional data file 2 online) could form the core of such a custom chip.
Materials and methods
Tumor specimens
Tumor biopsies were obtained from 96 participants presenting at the London Bone and Soft Tissue Tumour Service (Royal National Orthopaedic Hospital, Stanmore and University College London Hospitals, London), Great Ormond Street Hospital, London, or the Nuffield Orthopaedic Center, Headington, Oxford, in the UK. The diagnosis was determined by pathological examination using criteria established by the World Health Organization (WHO) tumor classification of soft tissue and bone tumors [45]. Where necessary, RT-PCR was performed to confirm common translocations (such as EWS, ARMS, MLS and synovial sarcoma). Ethical committee approval was obtained from all three treatment centers for the collection of fresh samples for this study. Clinical data such as diagnosis, site, grade and stage, age and sex are summarized in Additional data file 3 online. Patient biopsies were snap frozen with liquid nitrogen prior to RNA extraction and stored at -70°C.
Nucleic acid extraction
Frozen sections from each tumor sample (needle core or resection) were examined microscopically prior to RNA extraction to confirm that the sample was representative and contained more than 80% tumor cells. Total RNA was extracted from adjacent tissue sections with >80% tumor cell content, using TRIzol™ reagent (Invitrogen Ltd, Paisley, UK) followed by purification with RNeasy columns (Qiagen Ltd, Sussex, UK) according to the manufacturer's instructions. RNA quality and quantity was assessed using RNA 6000 Nano chips on Agilent 2100 Bioanalyser according to manufacturer's instructions (Agilent Technologies UK Ltd, Cheshire, UK). Our RNA quality-control threshold for the rRNA peak ratio was 28s/18s ≤ 2.
Microarray processing
The biotinylated hybridization target (biotin cRNA) was prepared from 10 μg of total RNA as previously described [46,47]. The quality and quantity of the biotinylated cRNA was checked prior to hybridization using the RNA 6000 Nano chips on Agilent 2100 Bioanalyser according to manufacturer's instructions. A total of 20 μg of the biotinylated probe was hybridized to Affymetrix HG-U133A Human GeneChips (Affymetrix®, Santa Clara, CA, USA) according to manufacturer's recommendations. In cases where smaller amounts of total RNA were obtained (EWS and rhabdomyosarcoma <5 μg total RNA), 150 ng of total RNA was subjected to two rounds of amplification to obtain a biotinylated cRNA yield of 20 μg according to Affymetrix recommendations (Genechip® Eukaryotic small sample target labeling assay version II). Parallel hybridizations of probe synthesized from 10 μg and 150 ng (via single and double amplification respectively) for the same tumor sample were found to have similar quality-control and expression profiles (see ARMS, EWS samples in Figure 1). Hybridization of the synthesized biotinylated probes and scanning of the images were performed as previously reported according to Affymetrix recommendations [46,47].
Data analysis
Data analysis was carried out using the R statistical environment and programming language [48]. We extensively used R software packages from Bioconductor [49], an open source bioinformatics resource. We used the 'affy' package written to handle Affymetrix data, and specifically the 'rma' algorithm for pre-processing, normalizing and calculation of expression values [50,51]. A modified version of the 'ipred' package was used for cross-validation and machine learning [52] together with the 'limma' package which was used for feature selection [53] (both described more fully below).
Hierarchical clustering and MDS
The Cluster and Treeview software packages were used to produce the average linkage hierarchical clustering shown in Figure 3 and 4 (correlation distance) [54]. MDS was chosen for Figure 1 instead of hierarchical clustering as this captured the complexity of the data optimally within the space available. We used the classical MDS method (or principle coordinates method part of the R 'base' package) [48]. The stress metric measures distortion required to plot high-dimensional data [55].
Classification algorithm
We chose simple k-NN pattern classification to model the data. This is a non-parametric algorithm, here based upon a table of inter-sample Euclidean distances [56]. The identity of an unknown (or test) sample is attributed by majority vote to that of the k = 3 nearest neighbors. We chose k = 3 due to the limiting minimum of three samples in our smallest class (DCS). The k-NN method was comparably effective to other methods tested (linear models, support vector machine, Bayes classifier; results not shown). We used leave-one-out cross-validation to estimate the error of our model. This method iteratively builds a predictive model from all combinations (or folds hereafter) of the data excepting one sample. The excepted sample is used to test the model error, the error being the proportion of inaccuracies from all folds. The multiple tumor classes and the limited number of samples in each class are unsuitable for accurate estimation of the classification error. This is therefore only a useful guide as the fingerprints are more important than the error. As a guide only, we used a permutation test to calculate the baseline error that might be expected from random guessing. Simply, this involves randomly permuting the class labels of the data ten times and counting the fraction of correctly corresponding labels with the original labels. To assess potential over-fitting we repeated this random permuting of the data class labels ten times but asked the computer to model the data and cross-validate, each time counting the fraction of errors. This does not, however, demonstrate a lack of over-fitting but merely a lack of gross over-fitting.
Feature selection
As most expression data are uninformative, or even confounding of pathological categories, feature selection is incorporated into each fold of cross-validation. For multi-class models we used a well-suited feature selection method to capture information from each of the 19 tumor classes (not to split classes in step two). The algorithm selects 10 genes for each class that are different in expression from all others. We tested 5, 10, 20 and 30 gene sets finding 10 to be sufficient (better than 5 or 30 and as good as 20). We used the 'limma' method for each comparison [53]. Limma uses a variant of linear models with an empirically moderated estimate of the standard error effectively borrowing information from the ensemble variance of genes to aid inference about individual genes. This gives improved statistical power for small sample sizes.
Additional data files
The following additional data are included with the online version of this article: an average linkage hierarchical clustering of the tumor samples using the same distance matrix as in Figure 1 (additional data file 1), a table providing the gene information for Figures 3 and 4 (additional data file 2) and a table containing clinical and pathological details of all samples used in this study (additional data file 3).
Supplementary Material
Additional data file 1
Average hierarchical clustering of tumor samples. The same set of inter-sample distances were used as in Figure 1. The cophenetic correlation of this clustering, a measure of the summary quality, was 0.71.
Click here for file
Additional data file 2
Supplementary information on the genes shown in Figures 3 and 4 in Worksheet S2A and S2B respectively. This includes references to the function of genes that lend support to our model feature selection, plus references the previously reported role of many genes in cancer.
Click here for file
Additional data file 3
Clinical and pathological details of all samples used in this study.
Click here for file
Acknowledgements
This work was supported by Cancer Research UK, The Wellcome Trust, the Adam Dealey Fund and the Royal National Orthopaedic Hospital NHS Trust Research and Development Fund. We thank Torsten Hothorn for discussion and assistance with adaptation of the 'ipred' software package.
Figures and Tables
Figure 1 Multi-dimensional scaling of all 96 mesenchymal tumor samples. There are 19 types of tumor shown; the color coding of which is used consistently for all figures. All gene expression values were used to calculate the inter-sample Euclidean distance matrix. The distances are translated here onto a two-dimensional plane using the classical cmd-scale algorithm of R. The stress of the plot was 0.34; an index of the goodness of fit between the original distance matrix and the MDS distance (see Materials and methods). WLS, well-differentiated liposarcoma; LMA, lipoma; MLS, myxoid liposarcoma; EWS, Ewing's sarcoma; FMT, desmoid fibromatosis; CHS, chondrosarcoma; CHB, chondroblastoma; ARMS, alveolar rhabdomyosarcoma; ERMS, embryonal rhabdomyosarcoma; DCS, de-differentiated chondrosarcoma; MSS, monophasic synovial sarcoma; MPNST, malignant peripheral nerve sheath tumors; CMF, chondromyxoid fibroma; PMS, pleomorphic sarcoma; LMS, leiomyosarcoma; OS, osteosarcoma; CMA, chordoma; NFB, neurofibroma; SWN, schwannoma.
Figure 2 Schematic of two-step model. In order to successfully classify the sarcoma with the minimum of errors a two-step approach was used. A mixture of single sarcoma and composite classes were used for prediction in step one. Then, in step two, composite classes were separated into their constituent tumors. * The SPIN (spindle-like) group comprising PMS, LMS, and DCS could not be separated by our model. ** The WLS and LMA could not be separated by our model but were distinct from MLS. ADIP, adipocytic tumors; RHAB, rhabdomyosarcoma.
Figure 3 Pathognomonic fingerprints for many tumor types. In step one of our model, 61 genes were used in all folds of cross-validation. Average linkage clustering of this geneset reveals strong sets of distinct genes for many single mesenchymal tumors or composite groups. The sample types are color coded as before. A, adipocytic tumors; B, rhabdomyosarcoma; C, NFB/SWN; D, EWS; E, CMA; F, FMT; G, MSS/MPNST; H, CHB; I, CHS; J, CMF; K, OS; L, spindle-like tumors.
Figure 4 Pathognomonic fingerprints step two. Molecular fingerprints of genes for A and B: ARMS and ERMS; C and D: MSS and MPNST; E and F: NFB and SWN; G and H: WLS/LMA and MLS. We have selected genes based upon their inclusion in the majority of folds of cross-validation then clustered them by average linkage.
Table 1 Errors of the simplest machine learning model
Model (n = 96) Error +/- S.E.
Cross-validation 0.33
Random guessing (n = 10) 0.93 +/- 0.01
Re-permuted and cross-validated (n = 10) 0.94 +/- 0.01
We used leave-one-out cross-validation of all samples simultaneously. With random guessing we compared the matching of the true order of classification labels against a randomly generated set, which produces a baseline for cross-validation comparison. For re-permutation, the true labels were randomized to create a semi-random false dataset.
Table 2 Frequency table of errors
ADIP CHB CHS CMA CMF EWS FMT MSS MPNST NFB SWN OS RHAB SPIN
ADIP 12
CHB 3
CHS 5 1 2
CMA 4
CMF 4
EWS 5
FMT 5
MSS MPNST 11
NFB SWN 8
OS 1 1 9
RHAB 7
SPIN 2 1 1 13
The frequency table shows the agreement between the sarcoma classes and the cross-validation prediction along the diagonal. The true numbers of tumors in each group can be calculated by summing the numbers in columns and the predictions of our model from summing the rows. ADIP, adipocytic tumors; RHAB, rhabdomyosarcoma; SPIN, spindle-like tumors.
Table 3 Errors of two-step machine learning model
Step one Error (absolute number)
All sarcoma (n = 96) 0.09 (9)
Step two
MSS and MPNST (n = 14) 0.14 (2)
NFB and SWN (n = 8) 0.125 (1)
ARMS and ERMS (n = 7) 0.14 (1)
LMA/WLS and MLS (n = 12) 0
Cross-validation errors for step one and two of our prediction model are shown. The algorithm focuses on these samples of each row in isolation. Note that all the samples wrongly classified in step two were correctly classified in step one.
==== Refs
Helman LJ Meltzer P Mechanisms of sarcoma development. Nat Rev Cancer 2003 3 685 694 12951587 10.1038/nrc1168
Mackall CL Meltzer PS Helman LJ Focus on sarcomas. Cancer Cell 2002 2 175 178 12242149 10.1016/S1535-6108(02)00132-0
Aurias A Rimbaut C Buffe D Zucker JM Mazabraud A Translocation involving chromosome 22 in Ewing's sarcoma. A cytogenetic study of four fresh tumors. Cancer Genet Cytogenet 1984 12 21 25 6713357 10.1016/0165-4608(84)90003-7
Turc-Carel C Philip I Berger MP Philip T Lenoir G [Chromosomal translocation (11; 22) in cell lines of Ewing's sarcoma]. C R Seances Acad Sci III 1983 296 1101 1103 6416622
Turc-Carel C Philip I Berger MP Philip T Lenoir GM Chromosome study of Ewing's sarcoma (ES) cell lines. Consistency of a reciprocal translocation t(11;22)(q24;q12). Cancer Genet Cytogenet 1984 12 1 19 6713356 10.1016/0165-4608(84)90002-5
Turc-Carel C Dal Cin P Limon J Li F Sandberg AA Translocation X;18 in synovial sarcoma. Cancer Genet Cytogenet 1986 23 93 3017544 10.1016/0165-4608(86)90153-6
Turc-Carel C Dal Cin P Limon J Rao U Li FP Corson JM Zimmerman R Parry DM Cowan JM Sandberg AA Involvement of chromosome X in primary cytogenetic change in human neoplasia: nonrandom translocation in synovial sarcoma. Proc Natl Acad Sci USA 1987 84 1981 1985 3031659
Tomescu O Barr FG Chromosomal translocations in sarcomas: prospects for therapy. Trends Mol Med 2001 7 554 559 11733218 10.1016/S1471-4914(01)02244-4
Allander SV Illei PB Chen Y Antonescu CR Bittner M Ladanyi M Meltzer PS Expression profiling of synovial sarcoma by cDNA microarrays: association of ERBB2, IGFBP2, and ELF3 with epithelial differentiation. Am J Pathol 2002 161 1587 1595 12414507
Baer C Nees M Breit S Selle B Kulozik AE Schaefer KL Braun Y Wai D Poremba C Profiling and functional annotation of mRNA gene expression in pediatric rhabdomyosarcoma and Ewing's sarcoma. Int J Cancer 2004 110 687 694 15146558 10.1002/ijc.20171
Fritz B Schubert F Wrobel G Schwaenen C Wessendorf S Nessling M Korz C Rieker RJ Montgomery K Kucherlapati R Microarray-based copy number and expression profiling in dedifferentiated and pleomorphic liposarcoma. Cancer Res 2002 62 2993 2998 12036902
Holtkamp N Reuss DE Atallah I Kuban RJ Hartmann C Mautner VF Frahm S Friedrich RE Algermissen B Pham VA Subclassification of nerve sheath tumors by gene expression profiling. Brain Pathol 2004 14 258 264 15446580
Khan J Simon R Bittner M Chen Y Leighton SB Pohida T Smith PD Jiang Y Gooden GC Trent JM Meltzer PS Gene expression profiling of alveolar rhabdomyosarcoma with cDNA microarrays. Cancer Res 1998 58 5009 5013 9823299
Lee YF John M Edwards S Clark J Flohr P Maillard K Edema M Baker L Mangham DC Grimer R Molecular classification of synovial sarcomas, leiomyosarcomas and malignant fibrous histiocytomas by gene expression profiling. Br J Cancer 2003 88 510 515 12592363 10.1038/sj.bjc.6600766
Lee YF John M Falconer A Edwards S Clark J Flohr P Roe T Wang R Shipley J Grimer RJ A gene expression signature associated with metastatic outcome in human leiomyosarcomas. Cancer Res 2004 64 7201 7204 15492233
Leonard P Sharp T Henderson S Hewitt D Pringle J Sandison A Goodship A Whelan J Boshoff C Gene expression array profile of human osteosarcoma. Br J Cancer 2003 89 2284 2288 14676807 10.1038/sj.bjc.6601389
Linn SC West RB Pollack JR Zhu S Hernandez-Boussard T Nielsen TO Rubin BP Patel R Goldblum JR Siegmund D Gene expression patterns and gene copy number changes in dermatofibrosarcoma protuberans. Am J Pathol 2003 163 2383 2395 14633610
Nagayama S Katagiri T Tsunoda T Hosaka T Nakashima Y Araki N Kusuzaki K Nakayama T Tsuboyama T Nakamura T Genome-wide analysis of gene expression in synovial sarcomas using a cDNA microarray. Cancer Res 2002 62 5859 5866 12384549
Nilbert M Meza-Zepeda LA Francis P Berner JM Namlos HM Fernebro J Myklebost O Lessons from genetic profiling in soft tissue sarcomas. Acta Orthop Scand Suppl 2004 75 35 50 15188664
Nilbert M Engellau J Experiences from tissue microarray in soft tissue sarcomas. Acta Orthop Scand Suppl 2004 75 29 34 15188663
Ochi K Daigo Y Katagiri T Nagayama S Tsunoda T Myoui A Naka N Araki N Kudawara I Ieguchi M Prediction of response to neoadjuvant chemotherapy for osteosarcoma by gene-expression profiles. Int J Oncol 2004 24 647 655 14767549
Ohali A Avigad S Zaizov R Ophir R Horn-Saban S Cohen IJ Meller I Kollender Y Issakov J Yaniv I Prediction of high risk Ewing's sarcoma by gene expression profiling. Oncogene 2004 23 8997 9006 15467746 10.1038/sj.onc.1208060
Ren B Yu YP Jing L Liu L Michalopoulos GK Luo JH Rao UN Gene expression analysis of human soft tissue leiomyosarcomas. Hum Pathol 2003 34 549 558 12827608 10.1016/S0046-8177(03)00014-5
Segal NH Pavlidis P Noble WS Antonescu CR Viale A Wesley UV Busam K Gallardo H DeSantis D Brennan MF Classification of clear-cell sarcoma as a subtype of melanoma by genomic profiling. J Clin Oncol 2003 21 1775 1781 12721254 10.1200/JCO.2003.10.108
Shmulevich I Hunt K El Naggar A Taylor E Ramdas L Laborde P Hess KR Pollock R Zhang W Tumor specific gene expression profiles in human leiomyosarcoma: an evaluation of intratumor heterogeneity. Cancer 2002 94 2069 2075 11932911 10.1002/cncr.10425
Skubitz KM Skubitz AP Differential gene expression in leiomyosarcoma. Cancer 2003 98 1029 1038 12942572 10.1002/cncr.11586
Skubitz KM Skubitz AP Gene expression in aggressive fibromatosis. J Lab Clin Med 2004 143 89 98 14966464 10.1016/j.lab.2003.10.002
Nielsen TO West RB Linn SC Alter O Knowling MA O'Connell JX Zhu S Fero M Sherlock G Pollack JR Molecular characterisation of soft tissue tumors: a gene expression study. Lancet 2002 359 1301 1307 11965276 10.1016/S0140-6736(02)08270-3
Brazma A Parkinson H Sarkans U Shojatalab M Vilo J Abeygunawardena N Holloway E Kapushesky M Kemmeren P Lara GG ArrayExpress: a public repository for microarray gene expression data at the EBI. Nucleic Acids Res 2003 31 68 71 12519949 10.1093/nar/gkg091
Ambroise C McLachlan GJ Selection bias in gene extraction on the basis of microarray gene-expression data. Proc Natl Acad Sci USA 2002 99 6562 6566 11983868 10.1073/pnas.102102699
Ransohoff DF Bias as a threat to the validity of cancer molecular-marker research. Nat Rev Cancer 2005 5 142 149 15685197 10.1038/nrc1550
Carroll SL Stonecypher MS Tumor suppressor mutations and growth factor signaling in the pathogenesis of NF1-associated peripheral nerve sheath tumors. I. The role of tumor suppressor mutations. J Neuropathol Exp Neurol 2004 63 1115 1123 15581179
Dupin E Glavieux C Vaigot P Le Douarin NM Endothelin 3 induces the reversion of melanocytes to glia through a neural crest-derived glial-melanocytic progenitor. Proc Natl Acad Sci USA 2000 97 7882 7887 10884419 10.1073/pnas.97.14.7882
Le Douarin NM Creuzet S Couly G Dupin E Neural crest cell plasticity and its limits. Development 2004 131 4637 4650 15358668 10.1242/dev.01350
Trentin A Glavieux-Pardanaud C Le Douarin NM Dupin E Self-renewal capacity is a widespread property of various types of neural crest precursor cells. Proc Natl Acad Sci USA 2004 101 4495 4500 15070746 10.1073/pnas.0400629101
Nakano T Tani M Ishibashi Y Kimura K Park YB Imaizumi N Tsuda H Aoyagi K Sasaki H Ohwada S Yokota J Biological properties and gene expression associated with metastatic potential of human osteosarcoma. Clin Exp Metastasis 2003 20 665 674 14669798 10.1023/A:1027355610603
Weeraratna AT Jiang Y Hostetter G Rosenblatt K Duray P Bittner M Trent JM Wnt5a signaling directly affects cell motility and invasion of metastatic melanoma. Cancer Cell 2002 1 279 288 12086864 10.1016/S1535-6108(02)00045-4
Segal NH Blachere NE Guevara-Patino JA Gallardo HF Shiu HY Viale A Antonescu CR Wolchok JD Houghton AN Identification of cancer-testis genes expressed by melanoma and soft tissue sarcoma using bioinformatics. Cancer Immun 2005 5 2 15683221
Salisbury JR The pathology of the human notochord. J Pathol 1993 171 253 255 8158454 10.1002/path.1711710404
Salisbury JR Deverell MH Cookson MJ Whimster WF Three-dimensional reconstruction of human embryonic notochords: clue to the pathogenesis of chordoma. J Pathol 1993 171 59 62 8229458 10.1002/path.1711710112
Edwards YH Putt W Lekoape KM Stott D Fox M Hopkinson DA Sowden J The human homolog T of the mouse T(Brachyury) gene; gene structure, cDNA sequence, and assignment to chromosome 6q27. Genome Res 1996 6 226 233 8963900
Tagawa K Humphreys T Satoh N T-Brain expression in the apical organ of hemichordate tornaria larvae suggests its evolutionary link to the vertebrate forebrain. J Exp Zool 2000 288 23 31 10750050 10.1002/(SICI)1097-010X(20000415)288:1<23::AID-JEZ3>3.0.CO;2-H
Longley DB Johnston PG Molecular mechanisms of drug resistance. J Pathol 2005 205 275 292 15641020 10.1002/path.1706
Ramaswamy S Ross KN Lander ES Golub TR A molecular signature of metastasis in primary solid tumors. Nat Genet 2003 33 49 54 12469122 10.1038/ng1060
Fletcher CD Unni KK Mertens F World Health Organization Classification of Tumors Pathology and Genetics of Tumors of Soft Tissue and Bone 2002 Lyon: ARC Press
Wang HW Trotter MW Lagos D Bourboulia D Henderson S Makinen T Elliman S Flanagan AM Alitalo K Boshoff C Kaposi sarcoma herpesvirus-induced cellular reprogramming contributes to the lymphatic endothelial gene expression in Kaposi sarcoma. Nat Genet 2004 36 687 693 15220918 10.1038/ng1384
Adib TR Henderson S Perrett C Hewitt D Bourmpoulia D Ledermann J Boshoff C Predicting biomarkers for ovarian cancer using gene-expression microarrays. Br J Cancer 2004 90 686 692 14760385 10.1038/sj.bjc.6601603
Venables WN Ripley BD Modern Applied Statistics with S 2002 New York: Springer-Verlag
Gentleman RC Carey VJ Bates DM Bolstad B Dettling M Dudoit S Ellis B Gautier L Ge Y Gentry J Bioconductor: open software development for computational biology and bioinformatics. Genome Biol 2004 5 R80 15461798 10.1186/gb-2004-5-10-r80
Irizarry RA Hobbs B Collin F Beazer-Barclay YD Antonellis KJ Scherf U Speed TP Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 2003 4 249 264 12925520 10.1093/biostatistics/4.2.249
Irizarry RA Bolstad BM Collin F Cope LM Hobbs B Speed TP Summaries of Affymetrix GeneChip probe level data. Nucleic Acids Res 2003 31 e15 12582260 10.1093/nar/gng015
Peters A Hothorn T Lausen B Ipred: improved predictors. R News 2002 2 33 36
Smyth GK Linear models and empirical Bayes methods for assessing differential expression in microarray experiments. Stat Appl Genet Mol Biol 2004 3
Eisen MB Spellman PT Brown PO Botstein D Cluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci USA 1998 95 14863 14868 9843981 10.1073/pnas.95.25.14863
Hand DJ Mannila H Smyth P Principles of Data Mining (Adaptive Computation & Machine Learning) 2001 Cambridge, MA: MIT Press
Hastie T Tibshirani R Friedman J The Elements of Statistical Learning: Data Mining, Inference and Prediction 2001 New York: Springer-Verlag Inc
|
16168083
|
PMC1242211
|
CC BY
|
2021-01-04 16:05:42
|
no
|
Genome Biol. 2005 Aug 26; 6(9):R76
|
utf-8
|
Genome Biol
| 2,005 |
10.1186/gb-2005-6-9-r76
|
oa_comm
|
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-9-r771616808410.1186/gb-2005-6-9-r77ResearchCharacterization of the yeast ionome: a genome-wide analysis of nutrient mineral and trace element homeostasis in Saccharomyces cerevisiae Eide David J [email protected] Suzanne 1Nair T Murlidharan 2Gehl Mathias 3Gribskov Michael 2Guerinot Mary Lou 4Harper Jeffrey F 31 Department of Nutritional Sciences, University of Wisconsin-Madison, Madison, WI 53706, USA2 San Diego Supercomputer Center, University of California-San Diego, La Jolla, CA 92903, USA3 Biochemistry Department, University of Nevada, Reno, Nevada 89557, USA4 Department of Biological Sciences, Dartmouth College, Hanover, NH 03755, USA2005 30 8 2005 6 9 R77 R77 29 3 2005 21 6 2005 18 7 2005 Copyright © 2005 Eide et al.; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
The accumulation of thirteen minerals was assayed in 4,385 yeast mutant strains, identifying 212 strains that showed altered ionome (mineral accumulation) profiles.
Background
Nutrient minerals are essential yet potentially toxic, and homeostatic mechanisms are required to regulate their intracellular levels. We describe here a genome-wide screen for genes involved in the homeostasis of minerals in Saccharomyces cerevisiae. Using inductively coupled plasma-atomic emission spectroscopy (ICP-AES), we assayed 4,385 mutant strains for the accumulation of 13 elements (calcium, cobalt, copper, iron, potassium, magnesium, manganese, nickel, phosphorus, selenium, sodium, sulfur, and zinc). We refer to the resulting accumulation profile as the yeast 'ionome'.
Results
We identified 212 strains that showed altered ionome profiles when grown on a rich growth medium. Surprisingly few of these mutants (four strains) were affected for only one element. Rather, levels of multiple elements were altered in most mutants. It was also remarkable that only six genes previously shown to be involved in the uptake and utilization of minerals were identified here, indicating that homeostasis is robust under these replete conditions. Many mutants identified affected either mitochondrial or vacuolar function and these groups showed similar effects on the accumulation of many different elements. In addition, intriguing positive and negative correlations among different elements were observed. Finally, ionome profile data allowed us to correctly predict a function for a previously uncharacterized gene, YDR065W. We show that this gene is required for vacuolar acidification.
Conclusion
Our results indicate the power of ionomics to identify new aspects of mineral homeostasis and how these data can be used to develop hypotheses regarding the functions of previously uncharacterized genes.
==== Body
Background
Living cells are composed of a large variety of chemical elements. In addition to carbon, nitrogen, and oxygen, cells require other elements either as additional components of macromolecules (for example, phosphorus, sulfur, and selenium), as cofactors required for the structural integrity (such as zinc) or enzymatic activity (such as copper and iron) of proteins, or as second messengers in cellular signal transduction (such as calcium). Because of the many important roles these elements play in cellular biochemistry, efficient mechanisms are required to obtain these nutrients from the environment, utilize or store them within intracellular organelles, and regulate their intracellular abundance to prevent overaccumulation and resultant toxicity. Identifying the molecular components of these mechanisms is a critical step toward a complete understanding of the nutritional aspects and toxicity of these elements. In addition, such information will be important as we attempt to genetically engineer plants and other organisms that are capable of removing toxic elements from the environment to remediate polluted sites (bioremediation).
The yeast Saccharomyces cerevisiae has been a useful model organism for the study of many different fundamental cellular processes, including the uptake, metabolism, and homeostatic control of mineral nutrients and trace elements. The usefulness of yeast for genome-wide studies of nutrient homeostasis has markedly increased with the recent completion of the Saccharomyces Genome Deletion Project [1]. This effort resulted in a collection of mutant strains disrupted in most of the approximately 6,000 genes in the yeast genome. This strain collection provides a unique resource for the analysis of gene function in a model eukaryotic cell.
Many studies of yeast have focused on the molecular mechanisms relevant to the utilization of nutrients [2-5]. The great majority of these studies have focused on the metabolism of specific nutrients without considering the effects of these systems on other elements. Thus, despite our growing understanding of the mechanisms controlling specific nutrients, the individual genes and gene networks that influence the acquisition and utilization of multiple elements remain largely unknown. To address this question, we have combined the genomic technologies provided by the Saccharomyces Genome Deletion collection with spectroscopic methods for the simultaneous analysis of multiple mineral nutrients accumulated by cells. The method used here, inductively coupled plasma-atomic emission spectroscopy (ICP-AES), can detect a broad range of elements simultaneously in a single assay [6]. The high sensitivity and dynamic range of this technology allows for the accurate quantitative measurement of element levels in small sample volumes.
Using ICP-AES, we have defined the elemental profile of wild-type yeast cells grown under standardized laboratory conditions. We refer to this profile as the yeast 'ionome', which expands on the previous concept of the 'metallome' to include several nonmetals [7-9]. The levels of 13 elements were assayed: calcium, cobalt, copper, iron, magnesium, manganese, nickel, phosphorus, potassium, selenium, sodium, sulfur, and zinc. We then determined the ionome profiles for a collection of over 4,000 different yeast mutants. The results of this study provide insights into the cellular systems controlling the homeostasis of multiple nutrients and provide new data for the functional characterization of as yet unstudied yeast genes.
Results and discussion
Characterizing the ionome of wild-type yeast cells
In this study, we used ICP-AES to simultaneously determine the levels of 13 different elements accumulated in yeast cells. Rich yeast extract-peptone-dextrose (YPD) medium was supplemented with several elements (calcium, cobalt, copper, manganese, nickel, selenium, zinc) to levels sufficient to facilitate their detection in cell extracts by ICP-AES (see Materials and methods). Boron and molybdenum were also added to the medium but these elements did not accumulate to sufficient levels to allow their detection by our methods. Furthermore, while neither nickel nor selenium is known to be required for yeast cell growth, many organisms use these elements for a variety of roles. Therefore, they were included in this analysis in the hope of better understanding the factors affecting their accumulation. In no case did the supplemented concentration of these elements exceed 10% of the minimal growth inhibitory concentration determined for this wild-type strain of yeast (data not shown).
Cells were grown to the post-diauxic-shift phase before harvesting. The cells were then collected by filtration and thoroughly washed to remove extracellular elements, and the organic material was then digested by overnight incubation in concentrated nitric acid before ICP-AES analysis. The 13-element ionome profile determined for wild-type cells is shown in Figure 1a. The minerals detected in our analysis accumulated to levels spanning almost four orders of magnitude, demonstrating the broad range over which these elements are found in living cells. Those elements that accumulated to the lowest levels were the trace elements manganese, cobalt, and copper (0.5 to 3 × 106 atoms per cell). Those accumulating to the highest levels were the macronutrients potassium and phosphorus (1.6 to 2.8 × 109 atoms per cell). The level of accumulation for many of these elements was very different from that observed previously for Escherichia coli grown in rich LB (Luria-Bertani) medium [7]. When converted to molar concentrations to adjust for the differences between bacterial and yeast cell volume and assuming homogeneous intracellular distributions, the accumulated levels of copper, potassium, magnesium, and manganese were similar to the levels in E. coli, whereas others, such as calcium, iron, and zinc, accumulated in yeast to 10-fold higher levels. Some of these differences may reflect the ability of eukaryotic cells to accumulate high levels of these elements within intracellular organelles that are not present in prokaryotes. Previous studies have indicated that yeast cells store many mineral nutrients within intracellular organelles [10-13].
We also compared the levels of these elements within cells with the corresponding levels in the growth medium (Figure 1b). Elements such as calcium, copper, and manganese accumulated to similar molar concentrations relative to the medium used for this study. As expected, sodium was largely excluded, with cells showing only 30% of media levels. In contrast, cobalt, iron, potassium, magnesium, phosphorus, sulfur, selenium, and zinc accumulated in cells to 3 to 30 times the level in the external environment, an observation consistent with the ability of cells to concentrate these elements intracellularly.
Analysis of yeast deletion mutants for effects on the ionome profile
To identify yeast genes critical to the homeostatic control of these elements, we determined the ionome profile of mutants generated by the Saccharomyces Genome Deletion Project. Approximately 25% of the total number of yeast genes (approximately 6,000) are either essential for viability under our growth conditions or had not yet been generated by the deletion project at the inception of this project. Therefore, we did not assay these strains. As a result, we analyzed a total of 4,385 different yeast mutants for their effects on the yeast ionome. To facilitate a genome-wide analysis, all of these strains were subjected to a high-throughput 'first-pass' ionome profile determination in which cells from a single culture of each mutant strain were assayed (see Additional data file 1 for a complete list of all strains tested). Of those 4,385 yeast mutants, 773 (18%) were identified as showing a twofold or greater difference for at least one element relative to triplicate wild-type controls prepared alongside each set of mutant samples. These 773 strains were then subjected to a 'second-pass' analysis of three independent cultures for each strain. A total of 233 strains were then identified that showed differences exceeding 3 standard deviations from the wild-type mean for at least one element in their respective profiles. These 233 strains were then analyzed in a 'third-pass' analysis of six independent cultures for each. Through this process, a total of 212 strains were identified as having mutations that cause reproducible effects on the yeast ionome, judged here as mean values increasing or decreasing by more than 2.5 standard deviations of the wild-type mean. The high ratio of strains showing reproducible effects in the second- and third-pass experiments (212/233 or 91%) indicates that few false positives are likely to be present in the final list of mutants identified as having ionome changes. Including cultures of wild-type cells assayed as controls, our results are based on the ICP-AES analysis of over 10,000 independent cultures.
The specific mutations leading to alterations in the level of one or more element are listed in Additional data file 2. An analysis of the effects of these mutations on the accumulation of specific elements revealed remarkable differences among them (Figure 2a). First, sodium and zinc showed the fewest number of mutants with alterations in their levels (69 and 70 of 212 total mutants, respectively). In contrast, nickel levels were altered in the most strains (162 of 212). When these effects were examined in more detail, the elements could be divided into three distinct groups. First, for elements such as cobalt, iron, and potassium, approximately equal numbers of mutants showed increases and decreases in element accumulation. In marked contrast, the results for calcium, copper, manganese, sulfur, and zinc were dominated by mutants showing increased mineral levels, whereas decreased levels were most frequently observed for magnesium, nickel, and selenium.
The numbers of mutants affected for each element represented in Figure 2a add up to considerably more than the 212 total strains identified in the analysis. This observation demonstrates an additional important point arising from these data. Most of these mutations are very pleiotropic in their effects on the ionome profiles; that is, more than one element was frequently altered for a given mutant. This pleiotropy is also clear when the number of elements affected per mutant is plotted versus the number of mutants (Figure 2b). The number of elements altered per strain ranged from as few as 1 element (4 mutants) to as many as 12 of the 13 elements we measured (12 mutants). A peak distribution observed in our experiments was around 7 to 10 elements affected per mutant.
Functional classification of mutations that alter the yeast ionome
The genes altered in the 212 mutant strains were grouped into 25 broad functional classes. An analysis of the distribution of the mutant strains among these functional classes is shown in Figure 3 and the specific genes in all groups are listed in Additional data file 3. The largest class had mutations in genes encoding proteins of unknown function, representing approximately 25% (59) of the mutants identified. This percentage reflects the relative frequency of genes in the entire yeast genome that remain uncharacterized. The two largest classes of mutations affecting proteins of known function were those with effects on vacuole biogenesis and function (27 mutants) (Table 1) and those involved in mitochondrial function (30 mutants) (Table 2). Classes containing fewer mutants included those affecting proteins involved in secretory pathway function (8). Thus, the largest percentage of genes identified (65 of 212 genes or 31%) are involved in the biogenesis or function of intracellular organelles. This result emphasizes the importance of these compartments in ion homeostasis. Other functional classes include genes involved in mRNA processing and protein synthesis (13) and transcription/chromatin structure (9). These classes of mutants are likely to cause changes in mineral content through indirect effects on gene expression and/or protein abundance. Surprisingly, genes known to be specifically involved in ion homeostasis accounted for only 3% (6/212) of the genes identified.
Effects of mutations disrupting organellar function on the ionome profile
The major role of organelles in controlling the ionome profiles warranted closer examination. As shown in Figure 4a, many mutants defective for vacuolar biogenesis and/or function caused increased accumulation of manganese, calcium, sulfur, and copper as well as decreases in cobalt, phosphorus, selenium, magnesium, and nickel. These were among the most pleiotropic mutations identified. The 27 vacuole-related mutants identified affected genes involved in many aspects of vacuolar function. First, six of these mutants were altered in genes encoding subunits of the vacuolar H+-ATPase (for example, CUP5, TFP1) (Table 1). These mutants have normal vacuole morphologies but lack the ability to acidify the organelle [14]. It was initially surprising that only a subset of V-ATPase mutants were identified in our screen, given that mutations in these genes are very likely to cause the same phenotypes. An examination of the ionomics dataset indicated that about half of the V-ATPase subunit mutants failed to meet the twofold cutoff criterion used in our first-pass analysis to identify strains for reanalysis. This observation suggested that the high stringency of this cutoff value was the main reason these genes were not included in our final list of mutants. Confirming this hypothesis, we reassayed eleven V-ATPase subunit mutants (n = 6) and found good accord among them. For example, 9 of the 11 mutants showed increased manganese and 8 of the 11 strains had significantly increased copper and decreased selenium (Additional data file 4).
Several mutants affecting vacuolar biogenesis were also identified. Previous studies of vacuolar protein sorting in yeast resulted in the identification of six classes, designated A through F, of mutants affecting this process [15,16]. These mutant classes exhibit a number of different vacuolar morphologies. For example, mutants of class A have normal-appearing vacuoles but show defects in protein sorting. Class B mutants have fragmented vacuolar morphologies, while class C mutants lack any recognizable vacuolar structure. Class D mutants have defects in vacuolar inheritance, resulting in daughter cells with a class C appearance, while class E mutants accumulate vacuolar proteins in the prevacuolar compartment because of defects in membrane trafficking from this compartment to the vacuole or the Golgi apparatus. Mutants of the final group, class F, have both normal-appearing vacuoles and fragmented vacuoles similar to those of class B mutants. Vacuolar mutants of four of these six classes were found to affect the ionome (Table 1). No mutants of either class A or F were identified, suggesting that the normal-appearing vacuoles in mutants of these classes are capable of maintaining the wild-type ionome profile. In addition to the 27 vacuolar mutants, several of the mutants with altered secretory pathway function (for example, RIC1, YPT6, COG7, COG8) showed similar profiles to the vacuolar mutants, suggesting that the effects of these mutations are due to indirect disruption of vacuolar function.
Thirty genes required for mitochondrial function were also identified (Table 2). These include genes required for mitochondrial transcription and protein synthesis (for example, MTF1, MRPL20, MRPL35), mitochondrial mRNA processing (for example, CBP1, CBP2), electron transport chain function (for example, COX10, CYT2, COQ4, COQ5), and oxidative phosphorylation (for example, ATP10). These mutants share common disruption of selenium and nickel accumulation, with the levels of both decreasing (Figure 4b). These effects were clearly distinguishable from the effects of vacuolar mutants that showed changes in other minerals in addition to nickel and selenium.
Finally, mutants disrupted for five genes involved in endocytosis (CLC1, SAC6, RVS161, RVS167, and YPK1) were also isolated. All five mutants showed increases in both calcium and copper accumulation. This result is consistent with the likely contribution of endocytosis to downregulating yeast copper uptake transporters [17] and suggests that calcium accumulation may be regulated in a similar fashion. To our knowledge, this potential mechanism of calcium homeostasis has not been tested experimentally.
The interrelationships between different elements in the yeast ionome
The similar effects of mutations in particular functional categories suggests that the homeostatic mechanisms that control the levels of different elements are interconnected. For example, mutants defective for vacuolar function show similar effects on several elements. This point was further emphasized when the entire third-pass ionome dataset was analyzed by principal-component analysis [18]. Both positive and negative correlations among elements are readily detected by this analysis and the results are presented as a biplot graph in Figure 5. The length of each eigenvector arrow reflects the variance in the data for each element. Thus, considering the configuration of the 13 elements depicted in Figure 5, it is evident that the largest variance is seen for potassium, while cobalt and nickel show the smallest variances. The biplot representation also displays the relationships among elements. The angles between positively correlated eigenvectors approach 0° while those between negative correlations approach 180° on the biplot representation. Elements showing no correlation have 90° eigenvector angles. Significantly, several of the elements cluster into one of three positively correlated groupings. In group I, magnesium, phosphorus, cobalt, and nickel are found to correlate in a large number of mutant strains. In addition, the elements in group I show a strong negative correlation with the effects of these mutations on sulfur levels. In group II, calcium, manganese, copper, and zinc show a strong correlation with each other, while group III includes iron and selenium. Group III elements also show a strong negative correlation with potassium. Some of the possible molecular explanations underlying these relationships will be considered below.
To our knowledge, this genomic analysis of ionome profiles is only the second of its kind, the first being the analysis of random mutants in Arabidopsis [8]. This yeast study has the added benefit of using a collection of already defined mutant strains. The results from yeast differ from the plant study in two significant ways. First, a greater degree of pleiotropy was observed among the yeast mutants than in plants. As shown in Figure 2b, the number of elements affected per strain peaked at around 7 to 10. In contrast, the peak among the plant mutants was at three elements altered per plant line. The second major difference is in the effects of mutations on particular elements. As shown in Figure 2a, the results for some elements are dominated by mutants showing either increases or decreases in their accumulation. While similar trends were observed among the plant results for some elements (such as copper), others differed markedly. For example, while most mutants in yeast affecting calcium caused increased accumulation, the majority of plant mutants had the opposite effect. Magnesium, phosphorus, nickel, and selenium show similarly divergent results. The dissimilar results obtained with yeast and plants may reflect fundamental differences in the cellular metabolism of these elements or, more likely, differences in element homeostatic mechanisms at work in single-celled versus multicellular organisms.
We found that mutations in 3% to 4% of the total genes in the yeast genome caused reproducible effects on the ionome under the growth conditions we used in this study. A similar recovery rate was obtained in the Arabidopsis study [8]. It was initially surprising that only 6 of the 212 yeast genes identified were previously determined to play specific roles in mineral homeostasis either as transporters or as transcription factors controlling expression of transporters and other genes. These genes are SMF3, CCC1, GEF1, SPF1, RCS1 (AFT1), and ROX1. SMF3 and CCC1 encode metal ion transporters in the vacuolar membrane [11,19]. GEF1 encodes a chloride channel in the Golgi apparatus that is involved in assembly of a functional iron uptake system [20], and SPF1 encodes a P-type ATPase in the secretory pathway whose substrate is unknown but likely to be an inorganic ion, perhaps Ca2+ [21]. Aft1 controls genes involved in iron uptake and metabolism, while Rox1 represses genes under aerobic conditions. At least one Rox1 target gene, FET4, is involved in metal ion uptake [22,23]. Mutants affecting many genes known to play roles in the homeostasis of these elements under certain conditions were included in our analysis. These included transporters involved in calcium (Cch1, Pmr1, Vcx1, Pmc1), cobalt (Cot1), copper (Ctr1, Ccc2), iron (Fet3, Ftr1, Smf3), magnesium (Alr1, Mrs2, Lpe10), manganese (Smf1, Pmr1, Atx2), phosphorus (Pho87, Pho88, Pho89), potassium (Trk1, Trk2, Tok1), sodium (Nhx1, Nha1), sulfur (Sul1), and zinc (Zrt1, Zrt3, Zrc1). The remarkably small number of such genes in our final list of mutants probably represents the redundancy of systems involved in the uptake and intracellular distribution of minerals. The cells in these cultures were grown under nutrient-rich conditions where multiple systems are likely to mediate these processes. For example, at least four different zinc uptake systems (Zrt1, Zrt2, Fet4, and one unknown system) are present in yeast [23-25] and loss of any one system fails to exert a major effect on the overall zinc accumulation under these conditions because of the compensatory control of the other pathways. As a further example, vacuolar storage of zinc requires the Zrc1 and Cot1 transporters, but mutation of either single gene has very little effect on zinc accumulation in the vacuole [12].
By far the largest groups of previously characterized genes that we identified were those involved in the function of the vacuole or the mitochondria. This observation highlights the importance of these compartments in maintaining mineral homeostasis. Vacuolar mutants were found to frequently show increases in manganese, calcium, sulfur, and copper as well as decreases in cobalt, phosphorus, selenium, magnesium, and nickel accumulation. The effects of these mutants on nickel and selenium accumulation may be explained by the current hypotheses that both of these elements are detoxified in the vacuole [26,27]. Failure to accumulate nickel and selenium in the vacuole may increase their cytosolic concentrations and thereby inhibit further uptake. The vacuole has also been previously implicated in the intracellular storage of phosphorus and magnesium, and our results support those of previous studies. Phosphorus is stored in great abundance in the vacuole as polyphosphate: long chains of phosphate groups linked by phosphoanhydride bonds. This material, which can accumulate to ≥10% of the dry weight of a yeast cell, has been proposed to bind Mg2+ to facilitate its storage in the vacuole [28]. This scenario provides a plausible explanation for the effects of vacuolar mutants on both phosphorus and magnesium accumulation; that is, the decrease in polyphosphate accumulation decreases the binding capacity for Mg2+ in the vacuole lumen. Consistent with this role, we found here that mutations that disrupt polyphosphate accumulation [29], namely vtc1/phm4 and vtc4/phm3, also reduce magnesium accumulation.
Given the ability of vacuolar polyphosphate to bind other metal ions such as Zn2+, it was predicted that mutations in vacuolar function would also disrupt accumulation of other metals [30]. The vacuole has been implicated as a major storage site for excess intracellular zinc [12,31]. However, no strong correlation was observed between mutants affecting vacuolar biogenesis and/or function and zinc levels in this study, indicating that polyphosphate may not be required for zinc storage. Alternatively, while disruption of the vacuole may indeed reduce vacuolar zinc storage, other compartments (for example, mitochondria) may then accumulate the excess zinc and maintain a consistent total cellular content [32]. It is also intriguing to note that disruption of the vacuole does not consistently alter the accumulation of iron, which has recently been proposed to be stored there [11]. Accumulation in other sites as proposed above for zinc may be involved here as well. Mitochondria have been found to be a site for iron accumulation under certain conditions [33,34]. Thus, impairment of vacuolar iron storage may lead to increases in mitochondrial iron under our culture conditions.
In addition to these other elements, calcium and manganese are also thought to accumulate in the yeast vacuole and are probably bound by polyphosphate [13,35,36]. However, mutants with altered vacuolar function showed consistent increases in the accumulation of these metals. While surprising, this observation is not without precedent. Miseta et al. noted previously that mutations in VPS33, a class C vacuolar protein-sorting gene, caused elevated cellular calcium accumulation [36]. These authors attributed this increase to an activation of the Pmr1 Ca2+/Mn2+-transporting ATPase located in the Golgi apparatus and subsequent calcium hyperaccumulation in that compartment. Given the ability of Pmr1 to transport both Ca2+ and Mn2+, this scenario may also explain the effects of vacuole disruption on total manganese accumulation observed here. Our results extend those previous observations by demonstrating that mutations that disrupt vacuolar acidification without disrupting vacuolar morphology also have this effect. Therefore, it is likely that hyperaccumulation of calcium and manganese in these mutants arises from the downstream effects of failing to store these ions in the vacuole. In a previous study, Ramsay and Gadd observed that mutants disrupted for vacuolar acidification had reduced manganese accumulation, whereas the levels of manganese increased in our experiments [13]. The treatment conditions were very different in these two studies. The previous study used 1 mM manganese while our medium contained only 11 μM manganese, and this difference may explain the opposite results. Thus, the vacuole may play a greater role in manganese storage under extremely high manganese conditions.
Another intriguing observation of this study is the strong negative correlation between the elements in group I (magnesium, phosphorus, nickel, cobalt) and sulfur (Fig. 5). One major driving force for this inverse correlation may be the role of the vacuole in sulfur homeostasis as well as for magnesium and phosphorus, for example, as noted above. Strains carrying mutations in 21 of the 27 genes identified in our study that are involved in vacuolar biogenesis and function showed marked increases in sulfur levels. The underlying molecular mechanism for this increase is currently unclear. One possible explanation is the potential role of the vacuole in sulfur storage. S-adenosylmethionine (AdoMet) is one of the major organic sulfur compounds in cells. Intracellular levels of AdoMet are approximately 1 mM [37], with about 70% of the total accumulating in the vacuole [38]. Given the effects of vacuolar mutations on other elements such as magnesium and phosphorus, we would have predicted a priori that the total levels of sulfur would decrease in these vacuolar mutant cells. Surprisingly, the effects are just the opposite: vacuolar mutants accumulate more sulfur than do wild-type cells. This effect was observed in a previous report where vps33 mutants were isolated because they hyperaccumulated AdoMet [39]. Because the transcriptional control of methionine biosynthetic genes are responsive to intracellular AdoMet levels [40], this hyperaccumulation led to the inappropriate repression of methionine biosynthesis and, therefore, methionine auxotrophy. Based on the analysis of the methionine auxotrophy phenotype, it was concluded that the disruption in sulfur homeostasis was limited to vacuolar mutants that eliminated the vacuolar structure (that is, class C mutants) and did not occur in mutants with lesser defects in vacuolar function [39]. In contrast, our results indicate that sulfur homeostasis is disrupted even in mutants that retain vacuolar structure but are simply unable to acidify the organelle (vma5, vma7, tfp1). The question still remains how disruption of vacuolar function leads to increased sulfur accumulation. It is conceivable that sulfur homeostasis is mediated in part by a signal of AdoMet storage in the vacuole. Loss of that signal due to vacuolar disruption might then lead to increased sulfur accumulation elsewhere in the cell.
Several other novel relationships between elements were also observed in this study. For example, iron and selenium show a strong positive correlation with each other and also a strong negative correlation with potassium accumulation. Genes showing this profile include those functioning in vacuolar function (TFP1, AVT5), secretary pathway function (COG7, COG8, RIC1), protein synthesis (RPL22A, RPL23A, RPL27A), and ion homeostasis (SPF1, ROX1). Given the diverse processes represented in this group of genes, future studies will be required to discover the mechanism(s) underlying this correlation.
The data obtained in this study are likely to be useful in assigning function to genes that have not yet been characterized. Among the 212 genes identified are 59 of unknown function. Many of these mutants show ionome profile patterns consistent with other profiles observed in the dataset. For example, mutants disrupted for 11 genes (YGL260W, YGR122W, YGR206W, YHL005C, YHL029C, YHR033W, YIR024C, YKL075C, YKR035C, YMR066W, and YMR098C) showed increased accumulation of nickel and selenium without the broader effects observed in vacuolar mutants. This profile is similar to that observed among mutants with disrupted mitochondrial function. Therefore, these genes may perform some role in mitochondria. Consistent with this prediction, three of their protein products have been tentatively localized to mitochondria by a genome-wide protein localization project (YIR024C, YMR066W, YMR098C) [41]. In addition, mutants disrupted for these three and a fourth gene (YHL005C) in this group not yet localized grow poorly on carbon sources requiring respiration [42].
In addition, ionome profiles similar to vacuole-defective mutants are also displayed by mutants disrupted in six uncharacterized genes (YDR065W, YDR220C, YGL220W, YGL226W, YKL171W, YOR331C). Thus, the encoded proteins are likely to be involved in vacuolar biogenesis or function. To test this hypothesis, the ability of the Δydr065w mutant to acidify its vacuole was assayed using LysoSensor Green DND-189 (Molecular Probes, Eugene, OR, USA). Accumulation of this fluorophore in the vacuolar membrane is dependent on the lumenal acidity of the compartment. As shown in Figure 6, the Δydr065w mutant failed to accumulate LysoSensor Green DND-189, indicating a severe disruption of vacuolar acidification. Similar results were also obtained with quinacrine (data not shown), another marker of vacuolar acidification. These results clearly demonstrate that the ionomics data provide important clues about the function of uncharacterized genes.
Conclusion
In this study, we used a genome-wide approach to identify genes that control the yeast ionome. With the application of ICP-AES, we determined the elemental profile of mutants defective in over 4,000 different yeast genes. Of these, 212 mutant strains were identified that showed reproducible changes in their ionome profiles. The majority of these mutants had pleiotropic effects with changes in the levels of multiple elements. Both positive and negative correlations were observed among groups of elements, thereby highlighting previously unsuspected relationships between elements. Mutants in certain functional categories, such as those with disrupted vacuolar or mitochondrial function, showed related ionome profile changes. We show that these results can then be used to develop hypotheses regarding the functions of previously uncharacterized genes. It is noteworthy that our ionomics analysis used post-diauxic-shift cells grown in a rich medium. Different results would most likely be obtained using exponential-phase cells and/or cells grown in minimal media or with other carbon sources. This ionomics approach provided new information about the mechanisms controlling mineral accumulation in yeast. Given that S. cerevisiae has served as such a useful model for the study of many different processes, including mineral homeostasis, we predict that insights ultimately gained from this type of analysis will also aid in our understanding of how plant and animal cells control these processes at the cellular and perhaps even organismal levels.
Materials and methods
Yeast strains analyzed
The mutants analyzed were prepared by the Saccharomyces Genome Deletion Project [1] and were purchased from Open Biosystems (Huntsville, AL, USA). The method used to generate this collection was a polymerase chain reaction based deletion strategy to generate a complete deletion of each of the open reading frames in the yeast genome. As part of the deletion process, each gene was replaced with the KanMX module, which confers resistance to G418. We analyzed the homozygous diploid collection generated in strain BY4743, whose full genotype is MATa/MATα his3/his3 leu2/leu2 ura3/ura3 lys2/+ met15/+. Nearly all open reading frames larger than 100 codons were disrupted in this collection. Many yeast genes are essential for growth in rich medium, so the corresponding strains were not included in our analysis. In addition, closely related genes (for example, ENA1-5) with ≥97% similarity were also not tested, because mutants in these genes were not generated by the deletion project consortium.
Culture conditions
Cells were recovered from frozen stocks and streaked for colonies on agar plates containing YPD (1% yeast extract, 2% peptone, 2% glucose) + 200 μg/ml G418 (Sigma, St Louis, MO, USA). A single colony from each plate was then inoculated into 5 ml of YPD + 1/100 volume of a 100 × mineral supplement stock (Table 3). The effects of metal supplementation on accumulation by wild-type cells is presented in Additional data file 5. In later experiments with multiple replicates, either three or six separate colonies from each strain were used for inoculations. The cells were grown with aeration at 30°C to post-diauxic-shift phase (≥7.5 × 107 cells/ml). For most strains, this phase was reached after 2 days of culturing. Slower-growing strains were harvested at similar cell densities after longer periods of incubation. In preliminary studies, we found that exponential-phase cells and post-diauxic-shift cells accumulate different levels of some minerals. For example, accumulation of iron, manganese, and zinc doubles in post-diauxic shift phase cells, while copper, nickel, and selenium levels increase more than 10-fold (data not shown). Post-diauxic-shift cells were used for this analysis because large numbers of cells could be obtained in smaller and more manageable culture volumes. Therefore, when considering the effects of mutations on the yeast ionome, it should be noted that different results may be obtained with cells harvested in exponential phase. No mutants assayed showed ionome profiles similar to exponential-phase cells. Three wild-type control cultures were included in each first-pass (n = 1) and second-pass (n = 3) experiment for use as references. Six wild-type cultures were included in each third-pass (n = 6) experiment.
Sample processing and ICP-AES analysis
The same lots of all medium components (such as yeast extract, peptone) were used throughout this study, to maintain consistent growth conditions. Culture volumes of 2.5 ml were collected by vacuum filtration using Isopore membrane filters (1.2 μm pore size) (Fisher Scientific, Pittsburgh, PA, USA). Cells were then washed three times with 5 ml of 1 μM ethylenediaminetetraacetic acid disodium salt solution, pH 8.0, by vacuum filtration followed by three washes with 5 ml each of distilled, deionized H2O. Pilot studies indicated that these conditions efficiently remove unbound elements (data not shown). The filters were then placed in screw-top microcentrifuge tubes, and 500 μl 30% HNO3 was added. The samples were digested overnight in a 65°C water bath. Afterward, 500 μl of distilled, deionized H2O was added, the samples were vortexed briefly, and the filters were removed. The cell digests were then centrifuged for 10 min at 12,000 × g and the supernatants were transferred to new tubes. ICP-AES analysis was performed with a Varian Vista ICP-AES (Varian Inc, Palo Alto, CA, USA) with a three-channel peristaltic pump.
The genes altered in the 212 mutant strains were grouped into 25 broad functional classes based on information available in the literature, the Saccharomyces Genome Database [43] and the Comprehensive Yeast Genome Database [44].
Data scaling and normalization
The scaling and normalization process was based on the concept that the wild-type samples differ only in total cell mass. They can therefore be brought to a common scale using a scale factor equal to the median value of the ratio of each element to a common standard. This scale factor can then be used to normalize the mutant data in the same experiment. The specific subset of elements used for scaling is given by E = {K, Ca, Mg, Mn, P, S, Zn}. The common standard used for normalization is the average concentration of each element over the set of wild-type samples within the experiment (n = 6). These standard values a, for each element j are given by:
where xij is the concentration of metal j in replicate i. The scale factor, wi, for a particular replicate sample i is then given by:
The scaled average concentrations, , of each element j in the wild-type samples are given by:
with standard deviation .
The scaled concentration of the elements in the mutant samples, , relative to wild type, is then calculated in a similar way, with scale factors for each replicate, , calculated with respect to the average values of the wild-type samples in the same experimental set, and z scores (number of standard deviations from the wild-type means) were calculated relative to these same wild-type samples.
with standard deviation
The z scores so obtained were then transformed to (1, 0, -1) depending on whether the values were greater than 2.5, from 2.5 to -2.5, or less than -2.5. This was then used for higher level analyses.
Principal-component analysis and biplot PCA
Principal-component analysis (PCA) and biplot PCA were used to characterize the structure of correlations within the ionome data and biplot [18] to present the results of the PCA. Briefly, a biplot is a graphical display of a matrix M = (mij) of n rows and m columns, using markers r1, r2,..., rn for its rows and markers c1, c2,..., cm for its columns. These markers are chosen in such a way that the inner product riT ci represents mij, the i, jth element of M. 'bi' in the biplot indicates a joint display of row and columns of the matrix M. The rows for the ionome matrix correspond to yeast knockouts and the columns are the ions. The dimensionality of the matrix for the ionome is 212 × 13. The row markers correspond to genes knocked out (not shown in the figure) and the arrows or column markers represent ions. The length of the arrow represents the variances of the different ions and the angle represents their correlation.
LysoSensor Green DND-189 labeling
Assessment of vacuolar acidification was performed with LysoSensor Green DND-189 as previously described [45].
Additional data files
The following additional data are available with the online version of this paper. Additional data file 1 is an Excel file summarizing the mutants analyzed. Additional data file 2 is an Excel file showing the effects of mutations in yeast genes on their respective ionome profiles. Additional data file 3 is an Excel table showing the functional classifications of the 212 strains showing reproducible effects. Additional data file 4 is an Excel file showing ionome analysis of mutants disrupted for V-ATPase subunit genes. Additional data file 5 is an Excel file showing the effects of metal supplementation on accumulation by wild-type cells.
Supplementary Material
Additional data file 1
The mutant strains assayed in the first-pass (n = 1), second-pass (n = 3), and third-pass (n = 6) inductively coupled plasma-atomic emission spectroscopy (ICP-AES) analyses are listed. Genes identified by genome sequencing or other studies that were not analyzed are also listed.
Click here for file
Additional data file 2
Shown are the results of the third-pass (n = 6) inductively coupled plasma-atomic emission spectroscopy (ICP-AES) analysis. Sheet 1: Normalized concentrations in parts per million (ppm) for each of the 13 elements assayed. Sheet 2: z scores are reported for each mutant and each element. These values are the number of standard deviations that the mutant value differed from the mean of the wild-type samples included in that experiment. Sheet 3: The z score data from Sheet 2 were converted into trinary data. A value of 1 was assigned if the z score was ≥2.5 and -1 if the z score was ≤-2.5. Values of 0 were assigned for values that fell between those boundaries.
Click here for file
Additional data file 3
The descriptions of gene function were obtained from the Saccharomyces Genome Database [43].
Click here for file
Additional data file 4
Shown are the results of an inductively coupled plasma-atomic emission spectroscopy (ICP-AES) analysis (n = 6) of V-ATPase subunit mutants grown in yeast extract-peptone-dextrose (YPD) medium with metal supplements. Normalized accumulation is reported in parts per million (ppm) and z scores are reported for each element. Calcium values from this experiment are not reported due to a high background of this element in these particular samples.
Click here for file
Additional data file 5
Shown are the results of an inductively coupled plasma-atomic emission spectroscopy (ICP-AES) analysis (n = 6) of wild-type cells grown in yeast extract-peptone-dextrose (YPD) medium with or without the metal supplements. Normalized accumulation is reported in parts per million (ppm) and z scores are reported for each element.
Click here for file
Acknowledgements
This study was supported by a National Science Foundation Plant Functional Genome program grant (DBI-0077378) awarded to M.L.G., J.F.H., D.J.E., M.G., Julian Schroeder, David Salt, and John Ward. We thank Papiya Ray, Jonathan Heidt, Ashkan Mojdehi, and Ann-Marie Woelbel for preparing the yeast samples. We also thank Jerry Kaplan, Sandy Davis-Kaplan, and Diane McVey Ward for conducting the LysoSensor Green analysis and David Salt for his helpful advice throughout this project.
Figures and Tables
Figure 1 Characterization of the wild-type yeast ionome. (a) Wild-type BY4743 cells were grown in rich yeast extract-peptone-dextrose (YPD) + mineral supplements to post-diauxic-shift phase, harvested, digested with HNO3, and then analyzed for the levels of the indicated elements. Mean values are shown and the error bars indicate 1 standard deviation (n = 40). (b) The element content of the supplemented growth medium was also assayed (n = 6). The ratio of cell concentration, calculated from the data in panel (a) and assuming homogeneous distribution in the cell, to medium concentration is plotted.
Figure 2 Overview of the effects of mutations on element content. (a) Number of mutants showing increases (open bars) and decreases (filled bars) for each element. (b) Number of mutants showing one or more changes in their ionome profiles.
Figure 3 Functional classes of genes identified by ionome profiling of their corresponding mutants. The number of genes identified in each functional class is represented. See Additional data file 3 for a complete list of the specific genes in each functional category.
Figure 4 Mutants within functional categories show similar ionome phenotypes. The effects of mutations altering (a) vacuolar or (b) mitochondrial function on the ionome profile are shown. Elements are listed along the horizontal axis and the genes affected are listed along the vertical axis. Increases greater than 2.5 standard deviations of the wild-type means are shown in red and decreases greater than 2.5 standard deviations are shown in green. The bars at the top represent the consensus for each group of genes. This figure was generated using TreeView software.
Figure 5 Biplot representation of the ionome results. The length of each eigenvector is proportional to the variance in the data for that element. The angle between eigenvectors represents correlations among different elements. Three groups of elements (circled, and denoted I, II, and III) show strong positive correlations.
Figure 6 Δydr065w mutants are defective for vacuolar acidification. Wild-type (BY4743) and BY4743 Δydr065w cells were harvested in exponential phase, incubated with LysoSensor Green DND-189, and then examined by differential interference contrast (DIC) (left panel) and fluorescence (right panel) microscopy. Failure to accumulate the fluorophore indicates defective vacuolar acidification. Intact vacuoles in the mutant cells are apparent in the DIC image.
Table 1 Genes identified involved in vacuolar function
Gene Functiona
CUP5 Vacuolar H+-ATPase subunit
TFP1 Vacuolar H+-ATPase subunit
TFP3 Vacuolar H+-ATPase subunit
VMA5 Vacuolar H+-ATPase subunit
VMA7 Vacuolar H+-ATPase subunit
VMA8 Vacuolar H+-ATPase subunit
VMA21 Vacuolar H+-ATPase assembly
VAM10 Vacuole fusion (B)
VPS41 Golgi-to-vacuole vesicular transport (B)
VPS16 Golgi-to-vacuole vesicular transport (C)
VPS33 Golgi-to-vacuole vesicular transport (C)
VPS9 Golgi-to-vacuole vesicular transport (D)
PEP12 Golgi-to-vacuole vesicular transport (D)
VPS45 Golgi-to-vacuole vesicular transport (D)
PEP7 Golgi-to-vacuole vesicular transport (D)
SNF8 Vacuolar protein targeting (E)
BRO1 Vacuolar protein targeting (E)
VPS36 Vacuolar protein targeting (E)
VPS4 Endosome-to-vacuole vesicular transport (E)
VAC14 Vacuolar protein targeting
VAM3 Golgi-to-vacuole vesicular transport
VPS53 Endosome-to-Golgi vesicular transport
VPS63 Vacuolar protein targeting
VPS64 Vacuolar protein targeting
VPS65 Vacuolar protein targeting
VPS66 Vacuolar protein targeting
AVT5 Potential Vacuolar amino acid transporter
aThe letter in parentheses indicates the assigned class of vacuolar biogenesis defect to which each strain belongs, if known.
Table 2 Genes identified involved in mitochondrial function
Gene Function
BCS1 Cytochrome bc(1) complex biogenesis
PDB1 Pyruvate dehydrogenase activity
DIA4 Serine-tRNA ligase activity
MTF1 Mitochondrial RNA polymerase specificity factor
PPA2 Inorganic phosphatase
MRPL35 Mitochondrial ribosome subunit
PET117 Cytochrome oxidase assembly
MRPL20 Mitochondrial ribosome subunit
NUC1 DNA/RNA nuclease
PTH1 Peptidyl-tRNA hydrolase
QCR2 Ubiquinol cytochrome c reductase subunit
MRP17 Mitochondrial ribosome subunit
RRF1 Mitochondrial ribosome recycling factor
RSM7 Mitochondrial ribosome subunit
COX10 Heme a biosynthesis
CBP3 Ubiquinol cytochrome c reductase assembly
CBP2 Mitochondrial RNA splicing
MSM1 Methionyl-tRNA synthetase
ATP10 ATP synthase assembly
CBP1 Mitochondrial RNA processing
MDL2 ABC transporter
YTA12 Protein turnover
IMP1 Inner membrane protease subunit
CYT2 Cytochrome c1 heme lyase
CBP4 Ubiquinol-cytochrome c reductase assembly
COQ5 Ubiquinone biosynthesis
COQ4 Ubiquinone biosynthesis
CTP1 Inner membrane citrate transporter
MRP13 Mitochondrial ribosome subunit
FIS1 Mitochondrial division
Table 3 Final concentration of elements in growth medium
Element Form added Supplemented concentration Final concentration
Calcium CaCl2 1 mM 1.2 mM
Cobalt CoCl2 5 μM 5.1 μM
Copper CuCl2 100 μM 110 μM
Iron - - 19 μM
Magnesium - - 490 μM
Manganese MnCl2 10 μM 11 μM
Nickel NiCl2 125 μM 160 μM
Potassium - - 12 mM
Phosphorus - - 5.7 mM
Selenium Na2SeO3 75 μM 75 μM
Sodium - - 16 mM
Sulfur - - 5.4 mM
Zinc ZnCl2 100 μM 160 μM
Calcium, cobalt, copper, manganese, nickel, selenium and zinc were added to rich YPD medium (1% yeast extract, 2% peptone, 2% glucose) to facilitate their detection in cells by inductively coupled plasma-atomic emission spectroscopy (ICP-AES). These supplemented levels did not exceed 10% of the minimal growth inhibitory concentration determined for this strain (data not shown). The final concentration of these elements in the growth medium measured by ICP-AES is also shown and represents the supplemented levels plus those present in the YPD medium alone. The same lots of all medium components (such as yeast extract, peptone) were used throughout this study to maintain consistent growth conditions. Boron was added as H2BO3 at 181 μM and molybdenum was added as NaMoO4 at 10 μM. Despite this supplementation, levels of these two minerals accumulated by cells remained below the level of detection. -, not supplemented.
==== Refs
Winzeler EA Shoemaker DD Astromoff A Liang H Anderson K Andre B Bangham R Benito R Boeke JD Bussey H Functional characterization of the S. cerevisiae genome by gene deletion and parallel analysis. Science 1999 285 901 906 10436161 10.1126/science.285.5429.901
Van Ho A Ward DM Kaplan J Transition metal transport in yeast. Annu Rev Microbiol 2002 56 237 261 12142483 10.1146/annurev.micro.56.012302.160847
Gaither LA Eide DJ Eukaryotic zinc transporters and their regulation. Biometals 2001 14 251 270 11831460 10.1023/A:1012988914300
Puig S Thiele DJ Molecular mechanisms of copper uptake and distribution. Curr Opin Chem Biol 2002 6 171 180 12039001 10.1016/S1367-5931(02)00298-3
Culotta VC Manganese transport in microorganisms. Met Ions Biol Syst 2000 37 35 56 10693130
Komaromy-Hiller G Flame, flameless, and plasma spectroscopy. Anal Chem 1999 71 338R 342R 10384790 10.1021/a1999907z
Outten CE O'Halloran TV Femtomolar sensitivity of metalloregulatory proteins controlling zinc homeostasis. Science 2001 292 2488 2492 11397910 10.1126/science.1060331
Lahner B Gong J Mahmoudian M Smith EL Abid KB Rogers EE Guerinot ML Harper JF Ward JM McIntyre L Genomic scale profiling of nutrient and trace elements in Arabidopsis thaliana. Nat Biotechnol 2003 21 1215 1221 12949535 10.1038/nbt865
Salt DE Update on plant ionomics. Plant Physiol 2004 136 2451 2456 15375201 10.1104/pp.104.047753
Cunningham KW Fink GR Ca2+ transport in Saccharomyces cerevisiae. J Exp Biol 1994 196 157 166 7823019
Li L Chen OS McVey Ward D Kaplan J CCC1 is a transporter that mediates vacuolar iron storage in yeast. J Biol Chem 2001 276 29515 29519 11390404 10.1074/jbc.M103944200
MacDiarmid CW Gaither LA Eide DJ Zinc transporters that regulate vacuolar zinc storage in Saccharomyces cerevisiae. EMBO J 2000 19 2845 2855 10856230 10.1093/emboj/19.12.2845
Ramsay LM Gadd GM Mutants of Saccharomyces cerevisiae defective in vacuolar function confirm a role for the vacuole in toxic metal ion detoxification. FEMS Microbiol Lett 1997 152 293 298 9231423 10.1016/S0378-1097(97)00213-9
Nelson N Perzov N Cohen A Hagai K Padler V Nelson H The cellular biology of proton-motive force generation by V-ATPases. J Exp Biol 2000 203 89 95 10600677
Banta LM Robinson JS Klionsky DJ Emr SD Organelle assembly in yeast: characterization of yeast mutants defective in vacuolar biogenesis and protein sorting. J Cell Biol 1988 107 1369 1383 3049619 10.1083/jcb.107.4.1369
Raymond CK Roberts CJ Moore KE Howald I Stevens TH Biogenesis of the vacuole in Saccharomyces cerevisiae. Int Rev Cytol 1992 139 59 120 1428679
Ooi CE Rabinovich E Dancis A Bonifacino JS Klausner RD Copper-dependent degradation of the Saccharomyces cerevisiae plasma membrane copper transporter Ctr1p in the apparent absence of endocytosis. EMBO J 1996 15 3515 3523 8670854
Gabriel KR The biplot graphics display of matrices with application to principal component analysis. Biometrika 1971 58 453 467
Portnoy ME Liu XF Culotta VC Saccharomyces cerevisiae expresses three functionally distinct homologues of the Nramp family of metal transporters. Mol Cell Biol 2000 20 7893 7902 11027260 10.1128/MCB.20.21.7893-7902.2000
Gaxiola RA Yuan DS Klausner RD Fink GR The yeast CLC chloride channel functions in cation homeostasis. Proc Natl Acad Sci USA 1998 95 4046 4050 9520490 10.1073/pnas.95.7.4046
Cronin SR Rao R Hampton RY Cod1p/Spf1p is a P-type ATPase involved in ER function and Ca2+ homeostasis. J Cell Biol 2002 157 1017 1028 12058017 10.1083/jcb.200203052
Jensen LT Culotta VC Regulation of Saccharomyces cerevisiae FET4 by oxygen and iron. J Mol Biol 2002 318 251 260 12051835 10.1016/S0022-2836(02)00093-1
Waters BM Eide DJ Combinatorial control of yeast FET4 gene expression in response to iron, zinc, and oxygen. J Biol Chem 2002 277 33749 33757 12095998 10.1074/jbc.M206214200
Zhao H Eide D The ZRT2 gene encodes the low affinity zinc transporter in Saccharomyces cerevisiae. J Biol Chem 1996 271 23203 23210 8798516 10.1074/jbc.271.38.23203
Zhao H Eide D The yeast ZRT1 gene encodes the zinc transporter of a high affinity uptake system induced by zinc limitation. Proc Natl Acad Sci USA 1996 93 2454 2458 8637895 10.1073/pnas.93.6.2454
Joho M Inouhe M Tohoyama H Murayama T Nickel resistance mechanisms in yeasts and other fungi. J Ind Microbiol 1995 14 164 168 7766209 10.1007/BF01569899
Gharieb MM Gadd GM Evidence for the involvement of vacuolar activity in metal(loid) tolerance: vacuolar-lacking and -defective mutants of Saccharomyces cerevisiae display higher sensitivity to chromate, tellurite and selenite. Biometals 1998 11 101 106 9542063 10.1023/A:1009221810760
Beeler T Bruce K Dunn T Regulation of cellular Mg2+ by Saccharomyces cerevisiae. Biochim Biophys Acta 1997 1323 310 318 9042353
Ogawa N DeRisi J Brown PO New components of a system for phosphate accumulation and polyphosphate metabolism in Saccharomyces cerevisiae revealed by genomic expression analysis. Mol Biol Cell 2000 11 4309 4321 11102525
Kornberg A Rao NN Ault-Riche D Inorganic polyphosphate: a molecule of many functions. Ann Rev Biochem 1999 68 89 125 10872445 10.1146/annurev.biochem.68.1.89
Miyabe S Izawa S Inoue Y Zrc1 is involved in zinc transport system between vacuole and cytosol in Saccharomyces cerevisiae. Biochem Biophys Res Commun 2001 282 79 83 11263974 10.1006/bbrc.2001.4522
Sensi SL Ton-That D Weiss JH Mitochondrial sequestration and Ca(2+)-dependent release of cytosolic Zn(2+) loads in cortical neurons. Neurobiol Dis 2002 10 100 108 12127148 10.1006/nbdi.2002.0493
Cobine PA Ojeda LD Rigby KM Winge DR Yeast contain a non-proteinaceous pool of copper in the mitochondrial matrix. J Biol Chem 2004 279 14447 14455 14729672 10.1074/jbc.M312693200
Babcock M de Silva D Oaks R Davis-Kaplan S Jiralerspong S Montermini L Pandolfo M Kaplan J Regulation of mitochondrial iron accumulation by Yfh1p, a putative homolog of frataxin. Science 1997 276 1709 1712 9180083 10.1126/science.276.5319.1709
Dunn T Gable K Beeler T Regulation of cellular Ca2+ by yeast vacuoles. J Biol Chem 1994 269 7273 7278 8125940
Miseta A Fu L Kellermayer R Buckley J Bedwell DM The Golgi apparatus plays a significant role in the maintenance of Ca2+ homeostasis in the vps33Delta vacuolar biogenesis mutant of Saccharomyces cerevisiae. J Biol Chem 1999 274 5939 5947 10026219 10.1074/jbc.274.9.5939
Chan SY Appling DR Regulation of S-adenosylmethionine levels in Saccharomyces cerevisiae. J Biol Chem 2003 278 43051 43059 12937179 10.1074/jbc.M308696200
Farooqui JZ Lee HW Kim S Paik WK Studies on compartmentation of S-adenosyl-L-methionine in Saccharomyces cerevisiae and isolated rat hepatocytes. Biochim Biophys Acta 1983 757 342 351 6342684
Jacquemin-Faure I Thomas D Laporte J Cibert C Surdin-Kerjan Y The vacuolar compartment is required for sulfur amino acid homeostasis in Saccharomyces cerevisiae. Mol Gen Genet 1994 244 519 529 8078479 10.1007/BF00583903
Thomas D Surdin-Kerjan Y Metabolism of sulfur amino acids in Saccharomyces cerevisiae. Microbiol Mol Biol Rev 1997 61 503 532 9409150
Huh WK Falvo JV Gerke LC Carroll AS Howson RW Weissman JS O'Shea EK Global analysis of protein localization in budding yeast. Nature 2003 425 686 691 14562095 10.1038/nature02026
Steinmetz LM Scharfe C Deutschbauer AM Kokranjac D Herman ZS Jones T Chu AM Giaever G Prokisch H Oefner PJ Davis RW Systematic screen for human disease genes in yeast. Nat Genet 2002 31 400 404 12134146
Saccharomyces Genome Database
Comprehensive Yeast Genome Database
Perzov N Padler-Karavani V Nelson H Nelson N Characterization of yeast V-ATPase mutants lacking Vph1p or Stv1p and the effect on endocytosis. J Exp Biol 2002 205 1209 1219 11948198
|
16168084
|
PMC1242212
|
CC BY
|
2021-01-04 16:05:42
|
no
|
Genome Biol. 2005 Aug 30; 6(9):R77
|
utf-8
|
Genome Biol
| 2,005 |
10.1186/gb-2005-6-9-r77
|
oa_comm
|
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-9-r781616808510.1186/gb-2005-6-9-r78MethodE-Predict: a computational strategy for species identification based on observed DNA microarray hybridization patterns Urisman Anatoly [email protected] Kael F 1Chiu Charles Y 13Kistler Amy L 1Beck Shoshannah 1Wang David 4DeRisi Joseph L [email protected] Department of Biochemistry and Biophysics, University of California San Francisco, San Francisco, CA 94143, USA2 Biomedical Sciences Graduate Program, University of California San Francisco, San Francisco, CA 94143, USA3 Department of Infectious Diseases, University of California San Francisco, San Francisco, CA 94143, USA4 Departments of Molecular Microbiology and Pathology and Immunology, Washington University School of Medicine, Saint Louis, MO 63110, USA2005 30 8 2005 6 9 R78 R78 26 4 2005 23 6 2005 26 7 2005 Copyright © 2005 Urisman et al.; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
An algorithm, E-Predict, for microarray-based species identification is presented. E-Predict compares an observed hybridization pattern with a set of theoretical energy profiles. Each profile represents a species that may be identified.
DNA microarrays may be used to identify microbial species present in environmental and clinical samples. However, automated tools for reliable species identification based on observed microarray hybridization patterns are lacking. We present an algorithm, E-Predict, for microarray-based species identification. E-Predict compares observed hybridization patterns with theoretical energy profiles representing different species. We demonstrate the application of the algorithm to viral detection in a set of clinical samples and discuss its relevance to other metagenomic applications.
==== Body
Background
Metagenomics, an emerging field of biology, utilizes DNA sequence data to study unculturable microorganisms found in the natural environment. Metagenomic applications include studies of diversity and ecology in microbial communities, detection and identification of representative species in environmental and clinically relevant samples, and discovery of genes or organisms with novel or useful functional properties (for recent reviews, see [1-4]).
Common to all of these applications is the task of identifying (and often quantifying the abundance of) individual genes, species, or even groups of species from the large and often complex sequence space being explored. In the most general approach, shotgun sequencing is used to both identify and quantify individual sequences in a sample of interest [5-8]. In a more targeted approach, polymerase chain reaction (PCR) is used to amplify a particular subset of sequences, which can then be cloned and analyzed. For example, 16S rRNA sequences are frequently used to identify bacterial and archaeal species [9-12]. Another approach is based on functional screening of shotgun expression libraries to identify DNA fragments that encode proteins with desirable activities [13-15].
DNA microarrays are also emerging as an important tool in metagenomics [2,16-18]. Particularly in applications concerned with real-time identification of known or related species, microarrays provide a practical high-throughput alternative to costly and time-consuming cloning and repetitive sequencing. For example, as previously reported, DNA microarrays have successfully been used to detect known viruses [19-22] and to discover a novel human viral pathogen [23]. Other metagenomic applications in which microarrays have great potential include monitoring food and water quality [24], tracking bioremediation progress [2,25], and assessment of biologic threat [26].
Use of DNA microarrays in metagenomics introduces a series of analytical challenges. First, the sequence space to explore may be very large, especially in the case of environmental samples. Given the technologic constraints on the total number of probes that can be placed on a microarray, improved algorithms are required for optimal probe selection to maximize coverage. Second, microarray data generated in metagenomic studies can be very complex. In the case of viral diagnostics, nucleic acid extracted from clinical specimens usually contains host and bacterial contaminants in addition to viral RNA and DNA. As a result, hybridization patterns are complicated by substantial amounts of noise introduced by specific and nonspecific cross-hybridization that cannot be anticipated or controlled. Third, multiple and potentially closely related species may be present in a single sample, resulting in complex or even overlapping hybridization patterns. Finally, a species identification strategy based on the use of experimentally derived patterns alone is not feasible, because such empirical controls can be obtained only for a limited number of species available as pure cultures or genomic clones. New analytical tools capable of overcoming these challenges are acutely needed.
We have previously reported the development of a DNA microarray-based platform for viral detection and discovery [23] (NCBI GEO [27], accession GPL366). Briefly, the platform employs a spotted 70-mer oligonucleotide microarray containing approximately 11,000 oligonucleotides, which represent the most conserved sequences from 954 distinct viruses corresponding to every NCBI reference viral genome available at the time of design. Nucleic acids are extracted from a sample of interest, typically a clinical specimen, and are amplified and labeled using random-primed reverse transcription, second strand synthesis, and PCR. The labeled DNA is then hybridized to the microarray, and hybridization patterns are analyzed to identify particular viruses that are present in the sample.
Here we report a computational strategy, called E-Predict, for species identification based on observed microarray hybridization patterns (Figure 1a). Using this strategy, an observed pattern of intensities is compared with a set of theoretical hybridization energy profiles, representing species with known genomic sequence. We illustrate the use of E-Predict on data obtained with our viral detection microarray and demonstrate its effectiveness in identifying viral species in a variety of clinical specimens. Based on these results, we argue that E-Predict is relevant for a broad range of microarray-based metagenomic applications.
Results
The E-Predict algorithm
Theoretical hybridization energy profiles were computed for every completely sequenced reference viral genome available in GenBank as of July 2004 (1,229 distinct viruses). This set of profiles included all viruses represented on the microarray and many viruses whose genomes became available after the array design had been completed. All microarray oligonucleotides expected to hybridize to a given viral genome were identified using nucleotide BLAST (basic local alignment search tool) alignment [28]. Free energy of hybridization (ΔG) was then computed for each alignment using the nearest neighbor method [29,30]. Oligonucleotides that failed to produce a BLAST alignment were assumed to have hybridization energies equal to zero. Thus, a given theoretical energy profile consists of the non-zero hybridization energies calculated for the subset of oligonucleotides producing a BLAST alignment to the corresponding genome. Collectively, the energy profiles of all the viruses constitute a sparsely populated energy matrix, in which each row corresponds to a viral species and each column corresponds to an oligonucleotide from the microarray (Figure 1b).
The general E-Predict algorithm for interpreting observed hybridization patterns is shown in Figure 1b. A vector of oligonucleotide intensities is normalized and compared with every normalized profile in the energy matrix using a simple similarity metric, resulting in a vector of raw similarity scores. Each element in this vector denotes the similarity between the observed pattern and one of the predicted profiles for a species represented in the energy matrix. The statistical significance of the raw similarity scores is estimated using a set of experimentally obtained null probability distributions. Profiles associated with statistically significant similarity scores suggest the presence of the corresponding viral species in the sample.
Normalization and similarity metric choice
In order to optimize the ability of E-Predict to discriminate between true positive and true negative predictions, we first evaluated the performance of several commonly used normalizations and similarity metrics. For this purpose we constructed a training dataset of 32 microarrays obtained from samples known to be infected by specific viruses. Fifteen microarrays represented independent hybridizations of RNA extracted from HeLa cells - a human cell line that is permanently infected with human papillomavirus (HPV) type 18. The remaining microarrays were obtained from 17 independent clinical specimens from children with respiratory tract infections. Ten specimens contained respiratory syncytial virus (RSV) and seven contained influenza A virus (FluA), as determined by direct fluorescent antibody (DFA) test.
Intensity and energy vectors were independently normalized using sum, quadratic, unit-vector, or no normalization (Table 1). Similarity scores between the vectors were computed using dot product, Pearson correlation, uncentered Pearson correlation, Spearman rank correlation, or similarity based on Euclidean distance (Table 2). All nonequivalent combinations of intensity vector normalization, energy vector normalization, and similarity metrics were evaluated. For each combination, similarity scores were obtained by comparing every microarray in the training dataset with every virus profile in the energy matrix. The performance of each combination was then evaluated by calculating the separation between the score obtained for the correct (match) virus profile and the best scoring nonmatch profile from either the same or a different virus family (Figure 2a and Figure 2b, respectively). We defined separation as the difference between the similarity scores of a match and the appropriate nonmatch profiles, divided by the range of all similarity scores on a given microarray. Using this statistic, a value of one corresponds to the best possible separation, a value of zero corresponds to no separation, and negative values represent cases in which a match profile is assigned a score lower than a nonmatch profile.
With the exception of Spearman rank correlation, all considered metrics assigned the highest similarity scores to the match profiles on all 32 microarrays, independent of normalization choice. Not surprisingly, separation between interfamily profiles was greater than that between intrafamily profiles. In addition, changes in normalization and similarity metric had greater impact on intrafamily than on interfamily separation. The best overall separation was determined by calculating the product of the means of the intrafamily and interfamily separations divided by the corresponding standard deviations. Sum normalization of the intensity vectors, quadratic normalization of the energy vectors, and uncentered Pearson correlation as the similarity metric achieved the highest overall separation, producing a mean intrafamily separation of 0.69 (standard deviation 0.17) and a mean interfamily separation of 0.93 (standard deviation 0.08). Therefore, we settled on this combination of normalization and similarity metric parameters as our method of choice.
Significance estimation
Raw similarity scores, as described above, provide an effective means of ranking viral energy profiles based on similarity to an observed hybridization pattern. However, such ranking provides no explicit information regarding the likelihood that viruses corresponding to the best scoring profiles are actually present in a sample under investigation. For example, two profiles may have identical high scores, but one of the scores may reflect a true positive whereas the other may be the result of over-representation of cross-hybridizing oligonucleotides in a profile.
To facilitate the interpretation of individual raw similarity scores, we sought to develop a test of their statistical significance. For this purpose, we obtained empirical distributions of the scores for every virus profile in the energy matrix. The distributions were based on 1,009 independent microarray experiments collected from a wide range of clinical and nonclinical samples representing different tissues, cell types, and nucleic acid complexities. Given such sample diversity, we assumed that any given virus was present in only a small fraction of all samples. Therefore, the empirical distributions are essentially distributions of true negative scores. The loge-transformed similarity scores were approximately normally distributed. Outliers on the right tails of the distributions, assumed to be true positives, were removed (see Materials and methods, below), and parameters of the null distributions were estimated as the mean and standard deviation of the remaining observations. These parameters were used to calculate the probability associated with any observed similarity score. Probabilities obtained this way should be interpreted as one-tail P values for the null hypothesis, that the virus represented by the profile is not present in the sample.
As shown in Figure 3, the most significant similarity scores for all 32 microarrays in the training dataset were correctly matched to the virus known to be present in the input sample: HPV18 for HeLa samples, RSV for RSV-positive samples, and FluA for FluA-positive samples. Corresponding P values ranged between 8.7 × 10-3 and 7.7 × 10-7 (median 2.1 × 10-5), between 4.0 × 10-4 and 1.4 × 10-8 (median 5.1 × 10-8), and between 1.8 × 10-6 and 1.4 × 10-7 (median 4.7 × 10-7), respectively (Figure 3; red circles). Energy profiles of unrelated viruses from six representative families (black circles) as well as profiles of divergent members belonging to the same families as the match viruses (blue circles) had similarity scores of essentially background significance (P values > 0.14). Even P values of the most closely related intrafamily virus profiles (purple circles) were separated from those of the match viruses by more than 1.1 (HPV45), 2.1 (human metapneumovirus), and 3.4 (influenza B virus) logs. Although the P values obtained for these profiles are more significant than background, their similarity scores are entirely based on oligonucleotides that also belong to the match virus profiles. P values resulting from such profile overlaps can be easily recognized and masked if desired (see Example 3, below).
Examples
Our laboratory is conducting a series of studies focused on human diseases suspected of having viral etiologies. The E-Predict algorithm was developed to assist in the analysis of samples obtained as part of these investigations. As an illustration of its versatility we present four example applications of E-Predict, as it is used in our laboratory.
Example 1
In this example, E-predict was used to interpret a hybridization pattern complicated by a low signal-to-noise ratio (Tables 3 and 4). The microarray result was obtained as part of our ongoing study of viral agents associated with acute hepatitis. Total nucleic acid from a serum sample was amplified, labeled, and hybridized to the microarray using our standard protocol (see Materials and methods, below). Despite the fact that very few oligonucleotides had intensity higher than background (Table 4), E-Predict assigned highly significant scores to hepatitis B virus (P = 0.002) and several closely related hepadnaviruses (Table 3). Specifically, no hepadnavirus oligonucleotide had intensity greater than 500 (for reference, background intensities are around 100, and the possible range is between 0 and 65,536). PCR with hepatitis B specific primers confirmed the presence of the virus in the sample. Complete E-Predict output for this example is available as Additional data file 1. The microarray data have been submitted to the NCBI GEO database [27] (accession GSE2228).
Example 2
In this example, E-Predict was used to identify the presence of two distinct viral species in the same sample (Table 5). The microarray result was obtained from a nasopharyngeal aspirate sample, which was collected as part of our ongoing investigation of childhood respiratory tract infections. On this microarray, E-Predict assigned highest significance to two unrelated viruses, namely FluA (P < 10-6) and RSV (P = 0.008), suggesting a double infection. The sample was independently confirmed to contain FluA and RSV, by DFA and specific PCR, respectively. Complete E-Predict output for this example is available as Additional data file 2. The microarray data have been submitted to the NCBI GEO database [27] (accession GSE2228).
Example 3
This example illustrates the ability of E-Predict to identify a virus that was not included in the microarray design. Table 6 shows E-Predict results for a microarray used to identify a novel coronavirus (severe acute respiratory syndrome (SARS) coronavirus (CoV)) during the 2003 outbreak of SARS, as reported previously [23,31]. Because our microarray was designed before 2003, it did not contain oligonucleotides derived from the SARS CoV genome. However, after the entire genome sequence of the virus became available [32], its theoretical energy profile was added to the E-Predict energy matrix. Reanalysis of the original SARS microarray data (NCBI GEO [27], accession GSM8528) using E-Predict revealed that the SARS CoV energy profile attained the highest similarity score and a highly significant P value (P = 1 × 10-6), despite the fact that the microarray, and therefore the profile, did not contain any oligonucleotides derived from the SARS CoV genome.
In addition to the SARS CoV prediction mentioned above, several astrovirus and picornavirus profiles had similarity scores with significant P values. However, these predictions were based on oligonucleotides corresponding to a conserved 3'-untranslated region shared by these viruses with the SARS CoV [23,33]. To identify incorrect predictions, such as these, resulting from partial profile overlaps with a match virus, we implemented an iterative version of E-Predict in which oligonucleotide intensities corresponding to the top scoring profile from one iteration are set to zero before running the next iteration. As a consequence, misleading predictions resulting from oligonucleotides shared with the top scoring profile fail to attain significant similarity scores in subsequent iterations. Conversely, only those predictions that are based on alternative oligonucleotides, namely predictions representing distinct species, remain. When iterative E-Predict was used on the SARS microarray, no astrovirus or picornavirus profile attained a statistically significant score (P > 0.04) in the second iteration, effectively removing these profiles from consideration. Complete E-Predict output for this example is available as Additional data file 3.
Example 4
This example illustrates the use of E-Predict to discriminate between closely related viral species such as human rhinovirus (HRV) serotypes (Figure 4). Rhinoviruses are a genus in the picornavirus family, which also includes enterovirus, aphthovirus, cardiovirus, hepatovirus, and parechovirus genera. Partial sequence analysis [34-36] indicates that HRV serotypes can be divided into two major groups (A and B), with the exception of HRV87, which is more closely related to enteroviruses. Only two complete rhinovirus reference genomes are available, one for each group: HRV89 (group A) and HRV14 (group B). Energy profiles of both viruses are included in our energy profile matrix as well as profiles of several enteroviruses and other more distant members of the picornavirus family. RNA samples from cultures of 22 representative serotypes were individually hybridized to the microarray, and the results were analyzed by E-Predict. In the absence of complete genome sequence data and corresponding energy profiles for each of the 22 serotypes, the E-Predict results revealed whether a particular serotype was most similar to HRV89, HRV14, or one of the enterovirus genomes in the energy matrix. To further refine our analysis, we clustered the E-Predict similarity scores from all 22 microarrays across all picornavirus profiles (Figure 4a). The resulting cluster dendrogram of the serotypes exhibited striking similarity to a phylogenetic tree based on nucleotide sequences of VP1 capsid protein (Figure 4b; also see Ledford and coworkers [34]). Serotypes 4, 26, 27, 70, and 83 were correctly grouped together on the basis of their similarity to the profile of HRV14 (group B); HRV87 formed a separate node, and the remaining serotypes were grouped together on the basis of their similarity to the profile of HRV89 (group A). Complete E-Predict output for this example is available as Additional data file 4. The microarray data have been submitted to the NCBI GEO database [27] (accession GSE2228).
Discussion
Identifying individual species present in a complex environmental or clinical sample is an essential component of many current and proposed metagenomic applications. Given a foundation of genomic sequence information, DNA microarrays are a high-throughput and cost-effective methodology for detecting species in an unbiased and highly parallel manner. Metagenomic applications employing DNA microarrays include characterization of microbial communities from environmental samples such as soil and water [2,17], pathogen detection in clinical specimens and field isolates [16], monitoring of bacterial contamination of food and water [24], and detection of agents involved in potential cases of bioterrorism [26].
Despite the increasing use of DNA microarrays for species detection and identification, bioinformatics tools for interpreting hybridization patterns associated with complex clinical and environmental samples are lacking. Existing methods have utilized direct visual inspection of hybridizing oligonucleotides [23,37] or inspection following clustering [19,38]. Such methods are intractable for interpreting complex hybridization patterns, are time consuming, and suffer from user bias. Improved data interpretation tools must address several challenges. First, hybridization patterns may represent signal from dozens or even hundreds of species. Also, several closely related species may be present in a sample, giving rise to overlapping hybridization signals. A likely additional source of noise is unanticipated cross-hybridization, because many of the genomes present in a complex sample may be uncharacterized. Finally, obtaining pure samples of each possible species for the purpose of generating reference hybridization patterns is impractical or impossible in most cases.
When challenged with each of these problems, E-Predict proved to be a useful tool for interpreting hybridization patterns, correctly identifying viruses from diverse viral families present in a variety of clinical samples. In particular, E-Predict does not rely on the use of empirically generated reference hybridization patterns, because species identification is based instead on theoretical hybridization energy profiles. The energy profile matrix currently represents over 1,200 distinct viruses whose complete genomic sequences are known. As new viral genomes are sequenced, profiles are added to the matrix to broaden the range of species detection. For example, addition of the SARS CoV profile enabled accurate identification of the virus, even though no oligonucleotides derived from its genome were present on the microarray. Conversely, even when a perfectly matching profile is not available because of limited sequence coverage, E-Predict will identify the closest related species, as long as such species are represented on the microarray. This feature is particularly useful for detecting novel viruses as well as for discriminating between closely related viruses such as HRV serotypes. Naturally, maximum range and precision of detection is achieved through addition of new profiles and periodic microarray updates to include specific oligonucleotides from newly sequenced species.
E-Predict is also useful in overcoming problems related to nucleic acid complexity frequently encountered in clinical samples. For example, E-Predict correctly identified hepatitis B virus in a serum sample, despite the fact that the hybridization pattern was complicated by a low signal-to-noise ratio. In another example, E-Predict deconvoluted a complex hybridization pattern, correctly suggesting the presence of two viruses (FluA and RSV) in a nasopharyngeal aspirate sample. In yet another example, iterative application of E-Predict (see Materials and methods, below) to a hybridization pattern involving oligonucleotides derived from seemingly unrelated families (coronaviridae and astroviridae) premitted objective recognition that the pattern represented the presence of only one virus (SARS CoV).
Using a training dataset of 32 microarrays derived from samples known to contain specific viral species, we identified a set of normalization and similarity metric parameters, which yielded the best discrimination between true positive and true negative species predictions. The combination of sum normalization of the intensity vectors, quadratic normalization of the energy vectors, and uncentered Pearson correlation as the similarity metric was the optimal choice for our data. However, a different set of parameters may be required for applications that use a different nucleic acid amplification or detection strategy. An independent evaluation of potentially useful normalization and similarity metric parameters is therefore recommended for each specific application of the algorithm.
Using our best combination of normalization and similarity metric parameters, we obtained a set of null distributions representing true negative scores. These distributions were based on over 1,000 independent hybridizations and the assumption that the majority of samples were negative for the presence of any given virus. Although valid for our data, this assumption will not hold for all cases. For example, in applications concerned with bacterial species detection, some species may be present in most or even all samples and others encountered only rarely. In this case, a more complicated model will be required to assess whether a specific distribution represents negative, positive, or both negative and positive scores. For example, in cases in which distributions appear bimodal, one mode may represent true negatives and the other true positives. In some cases, targeted experimental verification of a subset of representative scores may be necessary. If both positive and negative score distributions are available, then P values can be calculated for each distribution.
Several modifications to the algorithm may potentially result in improved prediction accuracy. First, in the current implementation oligonucleotides exhibiting nonspecific cross-hybridization are filtered and the remaining oligonucleotides are weighted equally. Because oligonucleotides exhibit a continuous range of nonspecific hybridization [20,30], a more sophisticated system of oligonucleotide weights may result in better performance. For example, using a procedure similar to that used to generate null distributions for the virus profile scores, empirical distributions can be obtained for individual oligonucleotide intensities, and individual oligonucleotide contributions may be weighted by the probabilities associated with the corresponding observed intensities. Such weighting may allow a more accurate assessment of significance.
Second, no attempt was made to normalize nucleic acid abundances of individual species, which may vary widely in different samples depending on factors such as target-to-background ratio, number of species present, and efficiency of nucleic acid extraction and amplification. Although individual nucleic acid abundances are difficult or impossible to estimate in most metagenomic applications, particularly before the corresponding species have been identified, in applications in which such estimates can be made, either experimentally or on theoretical grounds, the use of correction factors for calculating similarity scores or stratification of P value estimation may be needed. In addition, for highly abundant species, care should be taken to avoid saturation of individual oligonucleotides, because E-predict performance drops sharply after 20-25% of oligonucleotides in a given profile are saturated (data not shown).
Third, even though viral genomes were used as the basis for calculating energy profiles, the concept can easily be extended to other taxonomy nodes such as genera or families of viruses. This requires every sequence element to be classified at the appropriate node in the taxonomy hierarchy.
Finally, iterative use of E-Predict was intended for identification of multiple species that may be present in a sample. In this setting, it is important to distinguish between true predictions representing unique species present in the sample and misleading predictions arising from partially overlapping profiles. In each iteration it is assumed that the profile attaining the highest score corresponds to the species most likely to be among those present in the sample. When a novel species is present, this assumption may not hold because of limited oligonucleotide coverage. For instance, in the SARS CoV example, although SARS CoV attained a higher similarity score than mink astrovirus, the corresponding P values were comparable. However, even if mink astrovirus were the top prediction in the first iteration, SARS CoV would be the top prediction in the second iteration (P = 2 × 10-6; data not shown) and therefore would not be missed as a true positive. In our current studies P values in all iterations are estimated using the same set of null probability distributions. In addition, we use two iterations as our default, and essentially never need to run more than three iterations, because detection of more than two or three viruses is rare. However, iterative resolution of hundreds or thousands of species present in a sample may necessitate other normalization methods or adjustments to the null distributions for P value estimation. As an alternative, noniterative algorithms for analyzing overlapping profile signatures are also being explored.
In conclusion, E-Predict is a novel computational approach for species identification, which is generally applicable to a wide range of metagenomic applications using DNA microarrays. In particular, as more sequencing efforts are being directed at natural microbial communities, DNA microarrays are bound to become a central tool for various downstream applications such as identification of microbial species or detection of genes and biochemical pathways in such communities. E-Predict addresses an acute need for computational tools that are capable of interpreting the highly complex microarray data obtained through such studies. E-Predict was developed for viral species identification and therefore has immediate implications for medical diagnostics and viral discovery. In addition, the concept of theoretical energy profiles can be extended to represent other microorganisms, particular genes, or biochemical pathways.
Materials and methods
Sample preparation and hybridization to microarrays
All patient samples were collected according to protocols approved by the University of California San Francisco Committee on Human Research.
HeLa cells were grown to confluence in a T150 tissue culture flask in Dulbeco's modified Eagle medium supplemented with 10% fetal bovine serum and antibiotics. The cells were harvested by adding 10 ml Trizol reagent (Invitrogen, Carlsbad, CA, USA), and total RNA was isolated according to the manufacturer's protocol. A quantity of HeLa total RNA (50 ng) was used for each amplification and hybridization.
With respect to pediatric respiratory samples, frozen nasopharyngeal aspirate samples were thawed and 200 μl aliquots were used to extract RNA using RNeasy Mini Kit (Qiagen USA, Valencia, CA,USA) as follows. RLT buffer (750 μl) containing 1% 2-mercaptoethanol was added to each sample and mixed. Then, 1 ml of 100% ethanol was added, and the resulting mixture was applied to the columns in three 650 μl aliquots. The remaining steps were carried out in accordance with the manufacturer's protocol, including on-column DNase digest. RNA was eluted from the columns with 30 μl nuclease-free water, and 9 μl was used for amplification and hybridization. For the hepatitis sample, frozen serum sample was thawed and a 150 μl aliquot was used to extract total nucleic acid using MagNA Pure LC Total Nucleic Acid Isolation Kit (Roche Molecular Systems, Alameda, CA, USA), in accordance with the manufacturer's protocol. RNA was eluted in 50 μl nuclease-free water, and 9 μl was used for amplification and hybridization.
For HRV serotypes, frozen samples of low passage viral culture supernatants were thawed on ice and pre-filtered with a 0.2 μm syringe filter. Aliquots (200 μl) of the pre-filtered supernatants were treated with 600 U micrococcal nuclease (Fermentas USA, Hanover, MD, USA) in the presence of 10 mmol/l CaCl2 for 3 hours at 37°C. RNA was then extracted using Trizol reagent (Invitrogen), in accordance with the manufacturer's protocol. Linearized polyacrylamide (20 μg; Ambion, Austin, TX, USA) was used as the carrier during the 2-propanol precipitation. RNA was resuspended in 30 μl nuclease-free water, and 9 μl was used for amplification and hybridization.
Microarrays used in the study were essentially identical to those previously described [23]. Detailed description of the microarray platform, including oligonucleotide sequences, can be found in the NCBI GEO database [27] (accession GPL 1834). Briefly, 70-mer oligonucleotides representing the most conserved viral genomic elements were selected as 70-mers having sequence similarity (determined by nucleotide alignment) to the highest number of viral genomes [19]. Oligonucleotides were resuspended in 3 × SSC (0.45 M sodium chloride, 0.045 M sodium citrate, pH 7.0) at 50 μmol/l concentration and spotted onto poly-lysine coated glass slides [39]. Each spot on the microarray also contained a unique 'alien' sequence 70-mer (Spike70: 5'-ACC TCG CTA ACC TCT GTA TTG CTT GCC GGA CGC GAG ACA AAC CTG AAC ATT GAG AGT CAC CCT CGT TGT T-3'), spotted at a 1:50 ratio with the viral oligonucleotide to facilitate gridding of the microarrays (see below).
RNA extracted from the samples was amplified using a modified Round A-B random PCR method [40], as previously described (protocol S1 in [23]). Briefly, random-primed reverse transcription and second strand synthesis were carried out using primer A (5'-GTT TCC CAG TCA CGA TCN NNN NNN NN-3'). The resulting material was then amplified with 40 cycles of PCR using primer B (5'-GTT TCC CAG TCA CGA TC-3'). This was followed by an additional 20 cycles of PCR with primer B to incorporate aminoallyl-dUTP. The amplified material was then labeled with Cy5, and 0.1-1.0 pmol Probe70 (an oligonucleotide complementary to Spike70 containing five amino-modified bases for dye coupling: 5'-AAC AAC GAG GG[AmC6-dT] GAC TCT CAA [AmC6-dT]GT TCA GGT TTG TC[AmC6-dT] CGC GTC CGG CAA GCA A[AmC6-dT]A CAG AGG T[AmC6-dT]A GCG AGG T-3', Operon Biotechnologies, Huntsville, AL, USA) was labeled with Cy3. The Cy5 and Cy3 probes were pooled and hybridized to the microarray in 3 × SSC at 65°C overnight [39]. The Cy3 channel was used to facilitate gridding but otherwise was ignored in the data analysis. Microarrays were scanned with an Axon 4000B scanner (Axon Instruments, Union City, CA, USA) and gridded using the bundled GenePix 3.0 software.
Microarray data have been submitted to the NCBI GEO database [27] (accession GSE2228). The SARS microarray data are also available in NCBI GEO (accession GSM8528), as previously reported [23].
Training dataset
Fifteen HeLa microarrays were chosen randomly from a set of 43 HeLa hybridizations having at least five papillomavirus oligonucleotides with sum-normalized intensities greater or equal to 0.005. Ten RSV microarrays were chosen randomly from a set of 22 clinical hybridizations having at least five paramyxovirus oligonucleotides with sum-normalized intensities greater than or equal to 0.005 and confirmed to be RSV-positive by DFA. Seven FluA microarrays were chosen from eight available clinical hybridizations having at least five orthomyxovirus oligonucleotides with sum-normalized intensities greater than or equal to 0.005 and confirmed to be FluA-positive by DFA. The eighth FluA microarray was excluded because it was also positive for RSV by visual inspection.
Theoretical energy profiles
The energy profile matrix used in this study included all NCBI reference viral genomes (1,229) available as of July 2004 [41]. Nucleotide BLAST (blastall version 2.2.8 [42] with the default settings) was used to align microarray oligonucleotides with the viral genomes. Energies of hybridization were computed from the alignments using a program distributed with ArrayOligoSelector [30,43]. In cases in which an oligonucleotide had multiple alignments to the same genome, energy calculations were based on the highest scoring alignment. The energy profile matrix is available as Additional data file 5.
Similarity scores
Control oligonucleotides and oligonucleotides known to result in nonspecific hybridization were removed from consideration by setting their intensities and energies to zero. The list of these oligonucleotides (Additional data file 6) was obtained by including 129 oligonucleotides with unnormalized median intensity greater than 500, calculated from 1,009 independent hybridizations described below. The list also included 137 oligonucleotides obtained by clustering of distributions of sum-normalized intensity, based on the same set of 1,009 hybridizations, and visual identification of an outlier cluster with median sum-normalized intensities significantly higher than those observed for most oligonucleotides. Energy vectors were further filtered to exclude terms with energy predictions higher than -30 kcal/mol (again by setting their values to zero), because such predictions on our platform do not correspond to detectable array intensities [30]. A profile was considered only if it had at least three oligonucleotides with non-zero energy predictions. The resulting intensity and energy vectors were normalized using appropriate normalization methods (no normalization, sum, quadratic, and unit-vector). Similarity scores were computed using an appropriate similarity metric (dot product, Pearson correlation, uncentered Pearson correlation, Spearman rank correlation, and similarity based on Euclidean distance).
Probability estimation
Null distributions of similarity scores were obtained using a set of 1,009 hybridizations, which included all hybridizations performed on our platform to date. Similarity scores were calculated as described above using uncentered Pearson correlation as the similarity metric, and sum and quadratic normalizations for intensity and energy vectors, respectively. Scores were log-transformed. Right tail outliers corresponding to positive cases were excluded by iterative trimming of the top scores in 1% increments until the best normality fit was obtained, as judged by the Shapiro-Wilk normality test [44] (implemented in R [45]). Trimming was allowed to involve 0-25% of all scores. Over one-third of virus profiles required no trimming at all. Only a small number of profiles (34) required trimming beyond 10%, all of which corresponded to viruses frequently present in our samples. No profile required trimming of more than 17% of the scores. The resulting trimmed distributions were assumed to be normal, and their parameters were estimated as the mean and standard deviation of the included scores (Additional data file 7). Obtained parameters were used to estimate significance of individual scores as probabilities associated with observing values equal or greater than the scores. For this purpose, only profiles with at least three oligonucleotides with raw intensity greater than 100 (about two to four times background) were considered.
Iterative E-Predict
The first iteration was carried out as described above. For each additional iteration, oligonucleotide intensities of the profile attaining the highest similarity score in the previous iteration were set to zero. The resulting intensity vector was normalized, and similarity scores and P values were calculated using the same normalization method, similarity metric, and null distributions as in the initial iteration.
Clustering of human rhinovirus serotypes
Similarity scores were calculated as described above using uncentered Pearson correlation as the similarity metric, and sum and quadratic normalizations for intensity and energy vectors, respectively. Scores corresponding to picornavirus profiles were clustered using Cluster (version 2.0) [46,47] by hierarchical average linkage clustering with Pearson correlation as the similarity metric. Cluster images were obtained using Java TreeView (version 1.0.8) [48,49].
The phylogenetic tree based on nucleotide sequences of VP1 capsid protein was constructed using data from the report by Jonassen and coworkers [34]. Sequence alignment of relevant serotypes and the resulting tree were obtained using ClustalX (version 1.81 for Windows [50,51]) with default settings.
Polymerase chain reaction
The presence of hepatitis B virus in the hepatitis sample was confirmed using primers Hep_1F (5'-GAC TCG TGG TGG ACT TCT CTC AA-3') and Hep_4R (5'-GAA AGC CCT GCG AAC CAC TGA A-3') with amplified cDNA (Round B material; see [19] for amplification details) as the template. The presence of RSV in the FluA/RSV double-infected sample was confirmed by PCR using primers AU_041 (5'-GAT GAA AAA TTA AGT GAA ATA TTA GG-3') and AU_042 (5'-GTT CAC GTA TGT TTC CAT ATT TG-3') with cDNA (Round A material; see [19] for amplification details) as the template. In both cases, amplified PCR fragments were sequenced and had at least 99% nucleotide identity to the genomes of Hepatitis B virus (GenBank: NC_003977) and RSV (GenBank: NC_001803).
E-Predict software
The E-Predict software is available for download by any interested party [52].
Additional data files
The following additional data are available with the online version of this paper: a text file of E-Predict output for the hepatitis example (example 1) (Additional data file 1); a text file of E-Predict output for the FluA/RSV double infection example (example 2) (Additional data file 2); a text file of E-Predict output for the SARS CoV example (example 3) (Additional data file 3); a text file of E-Predict output for the HRV serotypes example (example 4) (Additional data file 4); a tab delimited text file containing the energy profile matrix (Additional data file 5); a text file containing the list of nonspecific oligonucleotides ignored during E-Predict (Additional data file 6); a tab delimited text file containing the list of profile parameters used to estimate P values (Additional data file 7). A text file of E-Predict output used to evaluate normalization and similarity metric parameters (Additional data file 8).
Supplementary Material
Additional data file 1
A text file of E-Predict output for the hepatitis example
Click here for file
Additional data file 2
A text file of E-Predict output for the FluA/RSV double infection example
Click here for file
Additional data file 3
A text file of E-Predict output for the SARS CoV example
Click here for file
Additional data file 4
A text file of E-Predict output for the HRV serotypes example
Click here for file
Additional data file 5
A tab delimited text file containing the energy profile matrix
Click here for file
Additional data file 6
A text file containing the list of nonspecific oligonucleotides ignored during E-Predict
Click here for file
Additional data file 7
A tab delimited text file containing the list of profile parameters used to estimate P values
Click here for file
Additional data file 8
A text file of E-Predict output used to evaluate normalization and similarity metric parameters
Click here for file
Acknowledgements
We thank Dr Hao Li, Christina Chaivorapol, and Amir Najmi for helpful discussions. We thank Dr Yu-Tsueng Liu for performing microarray hybridization and PCR follow up of the hepatitis sample. The hepatitis sample was graciously provided as part of an ongoing study by Dr Tim Davern (UCSF). Pediatric respiratory samples were graciously provided as part of an ongoing study by Dr Tara Greenhow, Dr Peggy Weintrub, Dr Lawrence Drew, and Carolyn Wright (UCSF). Cultures of HRV serotypes were graciously provided as part of an ongoing study by Dr David Schnurr and Dr Shigeo Yagi (California Viral and Rickettsial Disease Laboratory, Richmond, CA, USA). This work was supported by a Genentech Graduate Fellowship (A.U.) and grants from the Sandler Program for Asthma Research, and Doris Duke Charitable Foundation (J.D.).
Figures and Tables
Figure 1 E-Predict algorithm. (a) Nucleic acid from an environmental or clinical sample is labeled and hybridized to a species detection microarray. The resulting hybridization pattern is compared with a set of theoretical hybridization energy profiles computed for every species of interest. Energy profiles attaining statistically significant comparison scores suggest the presence of the corresponding species in the sample. (b) Observed hybridization intensities are represented by a row vector x, where each intensity value corresponds to an oligonucleotide on the microarray. Theoretical hybridization energy profiles form a matrix of energy values, Y, where each row represents a profile, and each column corresponds to an oligonucleotide in x. A suitable similarity metric function compares x with each row of Y to produce a column vector of similarity scores, s. Statistical significance of the individual scores in s is estimated to produce the output column vector of probabilities, P, where each probability value corresponds to a profile in Y.
Figure 2 Evaluation of normalization and similarity metric parameters. A training set of 32 microarrays was used to evaluate all nonequivalent combinations of intensity and energy vector normalization (N, none; Q, quadratic; S, sum; U, unit-vector) and similarity metric (DP, dot product; ED, similarity based on Euclidean distance; PC, Pearson correlation; SR, Spearman rank correlation; UP, uncentered Pearson correlation) parameters. For each combination of parameters, intrafamily and interfamily separations were calculated for each microarray as the score of the virus profile matching the virus present in the sample minus the score of the best scoring nonmatch profile from the same or a different virus family (top and bottom panels, respectively), normalized by the range of all scores on that microarray. Bars represent the mean, and error bars represent the standard deviation (±) of separation values from all microarrays. The best performing combinations are shown in order of increasing performance (calculated as the product of the intrafamily and interfamily separation means divided by the corresponding standard deviations).
Figure 3 Estimation of significance of individual similarity scores. Probabilities associated with the similarity scores of nine representative virus profiles obtained for the 15 HeLa, 10 respiratory syncytial virus (RSV), and seven influenza A virus (FluA) microarrays from the training dataset are shown in the top, center, and bottom panels, respectively. Each circle represents one microarray, and vertical 'jitter' is used to resolve individual circles. Probabilities for virus profiles from seven diverse virus families are included with each microarray set: herpes simplex virus (HSV)1; human T-lymphotropic virus (HTLV)1; severe acute respiratory syndrome coronavirus (SARS CoV); human rhinovirus B (HRV)B; FluA; human RSV; and three human papillomaviruses (HPV)18. Red circles represent match and black circles nonmatch interfamily profiles. Two intrafamily nonmatch profiles are also included and are different for the three microarray sets. The most closely related intrafamily profiles are represented by purple circles: HPV45, human metapneumovirus (HMPV), and influenza B virus (FluB). More distant intrafamily profiles are shown in blue: HPV37, mumps virus (MuV), and influenza C virus (FluC). The inset in each panel shows a normalized histogram (density) of the empirical distribution of log-transformed similarity scores for a match profile (black curve) and the corresponding normal fit representing true negative scores (green curve). Inset red bars depict observed log-transformed similarity scores corresponding to the match profile probabilities (red circles).
Figure 4 Human rhinovirus (HRV) serotype discrimination using E-Predict similarity scores. (a) Culture samples of 22 distinct HRV serotypes were separately hybridized to the microarray. E-Predict similarity scores were obtained for all virus profiles in the energy matrix and clustered using average linkage hierarchical clustering and Pearson correlation as the similarity metric. Virus profiles for which similarity scores could be calculated in all 22 experiments were included in the clustering. Both microarrays (rows) and virus profiles (columns) were clustered. (b) Published nucleotide sequences of VP1 capsid protein from the 22 HRV serotypes were aligned using ClustalX. Phylogenetic tree based on the resulting alignment is shown.
Table 1 Normalization methods
Normalization Formula Abbreviation
None N
Sum S
Quadratic Q
Unit vector U
Table 2 Similarity metrics
Similarity metric Formula Abbreviation
Dot product DP
Pearson correlation PC
Uncentered Pearson correlation UP
Spearman rank correlation SR
Similarity based on Euclidean distance ED
Table 3 Example 1: Hepatitis microarray - predicted virus profiles
Taxonomy ID Virus profile Virus family Similarity score Probability
10407 Hepatitis B virus Hepadnaviridae 0.145209 0.002451*
113194 Orangutan hepadnavirus Hepadnaviridae 0.143754 0.002482*
68416 Woolly monkey hepatitis B virus Hepadnaviridae 0.123794 0.003111*
35269 Woodchuck hepatitis B virus Hepadnaviridae 0.106576 0.002896*
41952 Arctic ground squirrel hepatitis B virus Hepadnaviridae 0.098908 0.003555*
10406 Ground squirrel hepatitis virus Hepadnaviridae 0.093975 0.003475*
10372 Human herpesvirus 7 Herpesviridae 0.027847 0.115068
All virus profiles for which a score could be calculated (see Materials and methods) are shown sorted by similarity score. *Statistically significant probabilities (P < 0.01).
Table 4 Example 1: hepatitis microarray - oligonucleotides contributing to hepatitis B virus profile prediction
Oligonucleotide Parental virus genome Virus family Raw intensity Raw energy
21326584_16 Hepatitis B virus Hepadnaviridae 403 102.9
9628700_11_rc Hepatitis B virus Hepadnaviridae 316 102.9
9634216_16 Orangutan hepadnavirus Hepadnaviridae 357 96.6
21326584_25 Hepatitis B virus Hepadnaviridae 262 109.6
9634216_11_rc Orangutan hepadnavirus Hepadnaviridae 308 99.1
9634216_11 Orangutan hepadnavirus Hepadnaviridae 288 99.1
9630370_16 Woolly monkey hepatitis B virus Hepadnaviridae 464 72.2
9628700_20_rc Hepatitis B virus Hepadnaviridae 160 120
21326584_9 Hepatitis B virus Hepadnaviridae 175 104.7
9628700_4 Hepatitis B virus Hepadnaviridae 153 104.7
Ten oligonucleotides contributing most to the hepatitis B virus similarity score are shown sorted by their relative contribution (product of normalized intensity and normalized energy values).
Table 5 Example 2 - FluA, RSV double infection
Taxonomy ID Virus profile Virus family Similarity score Probability
11320 Influenza A virus Orthomyxoviridae 0.504133 0.000000*
183764 Influenza A virus Orthomyxoviridae 0.486601 0.000000*
130760 Influenza A virus Orthomyxoviridae 0.105047 0.000151*
11250 Human respiratory syncytial virus Paramyxoviridae 0.033523 0.007895*
12814 Respiratory syncytial virus Paramyxoviridae 0.022144 0.007512*
11246 Bovine respiratory Syncytial virus Paramyxoviridae 0.009983 0.029254
162145 Human metapneumovirus Paramyxoviridae 0.001604 0.467995
All virus profiles for which a score could be calculated (see Materials and methods) are shown sorted by similarity score. *Statistically significant probabilities (P < 0.01).
Table 6 Example 3: SARS microarray
Taxonomy ID Virus profile Virus family Similarity score Probability
Iteration 1
227859 SARS coronavirus Coronaviridae 0.415354 0.000001*
219688 Mink astrovirus Astroviridae 0.335302 0.000000*
70793 Turkey astrovirus Astroviridae 0.217455 0.000000*
11120 Avian infectious bronchitis virus Coronaviridae 0.175788 0.000004*
70794 Ovine astrovirus Astroviridae 0.153207 0.000031*
107033 Avian nephritis virus Astroviridae 0.057325 0.000020*
47001 Equine rhinitis B virus Picornaviridae 0.048009 0.000054*
12702 Human astrovirus Astroviridae 0.044928 0.002118*
11852 Simian type D virus 1 Retroviridae 0.034479 0.016202
31631 Human coronavirus OC43 Coronaviridae 0.029834 0.002178
Iteration 2
11852 Simian type D virus 1 Retroviridae 0.053705 0.007108*
39068 Mason-Pfizer monkey virus Retroviridae 0.031347 0.026931
10359 Human herpesvirus 5 Herpesviridae 0.024634 0.167435
147712 Human rhinovirus B Picornaviridae 0.022551 0.048232
208177 Tomato leaf curl Vietnam virus Geminiviridae 0.022090 0.149573
85752 Tomato yellow leaf curl Thailand virus Geminiviridae 0.021844 0.080110
223334 Tobacco leaf curl Kochi virus Geminiviridae 0.021469 0.108687
188763 Chimpanzee cytomegalovirus Herpesviridae 0.021088 0.132918
32610 Tomato geminivirus Geminiviridae 0.021055 0.081960
83839 Pepper leaf curl virus Geminiviridae 0.020882 0.082562
For each iteration, ten profiles with highest similarity scores are shown sorted by score. *Statistically significant probabilities (P < 0.01). SARS, severe acute respiratory syndrome.
==== Refs
Riesenfeld CS Schloss PD Handelsman J Metagenomics: genomic analysis of microbial communities. Annu Rev Genet 2004 38 525 552 15568985 10.1146/annurev.genet.38.072902.091216
Eyers L George I Schuler L Stenuit B Agathos SN El Fantroussi S Environmental genomics: exploring the unmined richness of microbes to degrade xenobiotics. Appl Microbiol Biotechnol 2004 66 123 130 15316685 10.1007/s00253-004-1703-6
Rodriguez-Valera F Environmental genomics, the big picture? FEMS Microbiol Lett 2004 231 153 158 15027428 10.1016/S0378-1097(04)00006-0
Schloss PD Handelsman J Biotechnological prospects from metagenomics. Curr Opin Biotechnol 2003 14 303 310 12849784 10.1016/S0958-1669(03)00067-3
Venter JC Remington K Heidelberg JF Halpern AL Rusch D Eisen JA Wu D Paulsen I Nelson KE Nelson W Environmental genome shotgun sequencing of the Sargasso Sea. Science 2004 304 66 74 15001713 10.1126/science.1093857
Tyson GW Chapman J Hugenholtz P Allen EE Ram RJ Richardson PM Solovyev VV Rubin EM Rokhsar DS Banfield JF Community structure and metabolism through reconstruction of microbial genomes from the environment. Nature 2004 428 37 43 14961025 10.1038/nature02340
Breitbart M Hewson I Felts B Mahaffy JM Nulton J Salamon P Rohwer F Metagenomic analyses of an uncultured viral community from human feces. J Bacteriol 2003 185 6220 6223 14526037 10.1128/JB.185.20.6220-6223.2003
Breitbart M Felts B Kelley S Mahaffy JM Nulton J Salamon P Rohwer F Diversity and population structure of a near-shore marine-sediment viral community. Proc Biol Sci 2004 271 565 574 15156913 10.1098/rspb.2003.2628
Acinas SG Klepac-Ceraj V Hunt DE Pharino C Ceraj I Distel DL Polz MF Fine-scale phylogenetic architecture of a complex bacterial community. Nature 2004 430 551 554 15282603 10.1038/nature02649
van der Wielen PW Bolhuis H Borin S Daffonchio D Corselli C Giuliano L D'Auria G de Lange GJ Huebner A Varnavas SP The enigma of prokaryotic life in deep hypersaline anoxic basins. Science 2005 307 121 123 15637281 10.1126/science.1103569
Liles MR Manske BF Bintrim SB Handelsman J Goodman RM A census of rRNA genes and linked genomic sequences within a soil metagenomic library. Appl Environ Microbiol 2003 69 2684 2691 12732537 10.1128/AEM.69.5.2684-2691.2003
Woese CR Bacterial evolution. Microbiol Rev 1987 51 221 271 2439888
Brady SF Chao CJ Clardy J New natural product families from an environmental DNA (eDNA) gene cluster. J Am Chem Soc 2002 124 9968 9969 12188643 10.1021/ja0268985
Rondon MR August PR Bettermann AD Brady SF Grossman TH Liles MR Loiacono KA Lynch BA MacNeil IA Minor C Cloning the soil metagenome: a strategy for accessing the genetic and functional diversity of uncultured microorganisms. Appl Environ Microbiol 2000 66 2541 2547 10831436 10.1128/AEM.66.6.2541-2547.2000
Henne A Schmitz RA Bomeke M Gottschalk G Daniel R Screening of environmental DNA libraries for the presence of genes conferring lipolytic activity on Escherichia coli. Appl Environ Microbiol 2000 66 3113 3116 10877816 10.1128/AEM.66.7.3113-3116.2000
Bodrossy L Sessitsch A Oligonucleotide microarrays in microbial diagnostics. Curr Opin Microbiol 2004 7 245 254 15196491 10.1016/j.mib.2004.04.005
Zhou J Microarrays for bacterial detection and microbial community analysis. Curr Opin Microbiol 2003 6 288 294 12831906 10.1016/S1369-5274(03)00052-3
Cook KL Sayler GS Environmental application of array technology: promise, problems and practicalities. Curr Opin Biotechnol 2003 14 311 318 12849785 10.1016/S0958-1669(03)00057-0
Wang D Coscoy L Zylberberg M Avila PC Boushey HA Ganem D DeRisi JL Microarray-based detection and genotyping of viral pathogens. Proc Natl Acad Sci USA 2002 99 15687 15692 12429852 10.1073/pnas.242579699
Sengupta S Onodera K Lai A Melcher U Molecular detection and identification of influenza viruses by oligonucleotide microarray hybridization. J Clin Microbiol 2003 41 4542 4550 14532180 10.1128/JCM.41.10.4542-4550.2003
Klaassen CH Prinsen CF de Valk HA Horrevorts AM Jeunink MA Thunnissen FB DNA microarray format for detection and subtyping of human papillomavirus. J Clin Microbiol 2004 42 2152 2160 15131183 10.1128/JCM.42.5.2152-2160.2004
Lin B Vora GJ Thach D Walter E Metzgar D Tibbetts C Stenger DA Use of oligonucleotide microarrays for rapid detection and serotyping of acute respiratory disease-associated adenoviruses. J Clin Microbiol 2004 42 3232 3239 15243087 10.1128/JCM.42.7.3232-3239.2004
Wang D Urisman A Liu YT Springer M Ksiazek TG Erdman DD Mardis ER Hickenbotham M Magrini V Eldred J Viral discovery and sequence recovery using DNA microarrays. PLoS Biol 2003 1 E2 14624234 10.1371/journal.pbio.0000002
Lemarchand K Masson L Brousseau R Molecular biology and DNA microarray technology for microbial quality monitoring of water. Crit Rev Microbiol 2004 30 145 172 15490968 10.1080/10408410490435142
Rhee SK Liu X Wu L Chong SC Wan X Zhou J Detection of genes involved in biodegradation and biotransformation in microbial communities by using 50-mer oligonucleotide microarrays. Appl Environ Microbiol 2004 70 4303 4317 15240314 10.1128/AEM.70.7.4303-4317.2004
Ivnitski D O'Neil DJ Gattuso A Schlicht R Calidonna M Fisher R Nucleic acid approaches for detection and identification of biological warfare and infectious disease agents. Biotechniques 2003 35 862 869 14579752
NCBI Gene Expression Omnibus
Altschul SF Gish W Miller W Myers EW Lipman DJ Basic local alignment search tool. J Mol Biol 1990 215 403 410 2231712 10.1006/jmbi.1990.9999
Peyret N Seneviratne PA Allawi HT SantaLucia J Jr Nearest-neighbor thermodynamics and NMR of DNA sequences with internal A.A, C.C, G.G, and T.T mismatches. Biochemistry 1999 38 3468 3477 10090733 10.1021/bi9825091
Bozdech Z Zhu J Joachimiak MP Cohen FE Pulliam B DeRisi JL Expression profiling of the schizont and trophozoite stages of Plasmodium falciparum with a long-oligonucleotide microarray. Genome Biol 2003 4 R9 12620119 10.1186/gb-2003-4-2-r9
Ksiazek TG Erdman D Goldsmith CS Zaki SR Peret T Emery S Tong S Urbani C Comer JA Lim W A novel coronavirus associated with severe acute respiratory syndrome. N Engl J Med 2003 348 1953 1966 12690092 10.1056/NEJMoa030781
Rota PA Oberste MS Monroe SS Nix WA Campagnoli R Icenogle JP Penaranda S Bankamp B Maher K Chen MH Characterization of a novel coronavirus associated with severe acute respiratory syndrome. Science 2003 300 1394 1399 12730500 10.1126/science.1085952
Jonassen CM Jonassen TO Grinde B A common RNA motif in the 3' end of the genomes of astroviruses, avian infectious bronchitis virus and an equine rhinovirus. J Gen Virol 1998 79 715 718 9568965
Ledford RM Patel NR Demenczuk TM Watanyar A Herbertz T Collett MS Pevear DC VP1 sequencing of all human rhinovirus serotypes: insights into genus phylogeny and susceptibility to antiviral capsid-binding compounds. J Virol 2004 78 3663 3674 15016887 10.1128/JVI.78.7.3663-3674.2004
Blomqvist S Savolainen C Raman L Roivainen M Hovi T Human rhinovirus 87 and enterovirus 68 represent a unique serotype with rhinovirus and enterovirus features. J Clin Microbiol 2002 40 4218 4223 12409401 10.1128/JCM.40.11.4218-4223.2002
Savolainen C Blomqvist S Mulders MN Hovi T Genetic clustering of all 102 human rhinovirus prototype strains: serotype 87 is close to human enterovirus 70. J Gen Virol 2002 83 333 340 11807226
Loy A Lehner A Lee N Adamczyk J Meier H Ernst J Schleifer KH Wagner M Oligonucleotide microarray for 16S rRNA gene-based detection of all recognized lineages of sulfate-reducing prokaryotes in the environment. Appl Environ Microbiol 2002 68 5064 5081 12324358 10.1128/AEM.68.10.5064-5081.2002
Cho JC Tiedje JM Bacterial species determination from DNA-DNA hybridization by using genome fragments and DNA microarrays. Appl Environ Microbiol 2001 67 3677 3682 11472947 10.1128/AEM.67.8.3677-3682.2001
Eisen MB Brown PO DNA arrays for analysis of gene expression. Methods Enzymol 1999 303 179 205 10349646
Bohlander SK Espinosa R 3rdLe Beau MM Rowley JD Diaz MO A method for the rapid sequence-independent amplification of microdissected chromosomal material. Genomics 1992 13 1322 1324 1505965 10.1016/0888-7543(92)90057-Y
NCBI Reference Viral Genomes
NCBI BLAST
ArrayOligoSelector
Shapiro SS Wilk MB An analysis of variance test for normality (complete samples). Biometrika 1965 52 591 611
The R Project
Eisen MB Spellman PT Brown PO Botstein D Cluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci USA 1998 95 14863 14868 9843981 10.1073/pnas.95.25.14863
Eisen Lab, Software
Saldanha AJ Java Treeview: extensible visualization of microarray data. Bioinformatics 2004 20 3246 3248 15180930 10.1093/bioinformatics/bth349
Java TreeView
Thompson JD Gibson TJ Plewniak F Jeanmougin F Higgins DG The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res 1997 25 4876 4882 9396791 10.1093/nar/25.24.4876
ClustalX
DeRisi Lab: E-Predict Download
|
16168085
|
PMC1242213
|
CC BY
|
2021-01-04 16:05:42
|
no
|
Genome Biol. 2005 Aug 30; 6(9):R78
|
utf-8
|
Genome Biol
| 2,005 |
10.1186/gb-2005-6-9-r78
|
oa_comm
|
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-9-r791616808610.1186/gb-2005-6-9-r79MethodA computational method to predict genetically encoded rare amino acids in proteins Chaudhuri Barnali N [email protected] Todd O [email protected] UCLA-DOE Institute for Genomics and Proteomics and Department of Chemistry and Biochemistry, University of California, Los Angeles, USA2005 31 8 2005 6 9 R79 R79 8 3 2005 20 6 2005 27 7 2005 Copyright © 2005 Chaudhuri and Yeates; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
A new method for predicting recoding by rare amino acids such as selenocysteine and pyrrolysine was used to survey a set of microbial genomes.
In several natural settings, the standard genetic code is expanded to incorporate two additional amino acids with distinct functionality, selenocysteine and pyrrolysine. These rare amino acids can be overlooked inadvertently, however, as they arise by recoding at certain stop codons. We report a method for such recoding prediction from genomic data, using read-through similarity evaluation. A survey across a set of microbial genomes identifies almost all the known cases as well as a number of novel candidate proteins.
==== Body
Background
Codon redefinitions that expand upon the standard genetic code beyond the 20 canonical amino acids are reported in all three domains of life [1,2]. Two known genetically encoded rare amino acids (RAAs) are selenocysteine and pyrrolysine, the proposed 21st and the 22nd amino acids, respectively [3-7]. Selenocysteine, a selenium-analog of cysteine, is a potent nucleophile [5] and has been reported in organisms as diverse as Escherichia coli and human beings [4,5]. Selenium plays a dual role in nature as an essential micronutrient in human health, and as an environmental hazard to humans, livestock and wildlife [8] when it is present in high amounts. Thus, selenium is a target for both molecular biology and bioremediation research [8,9]. The distribution of selenium in the form of selenocysteine residues [5,10] in specific proteins is not completely understood. Pyrrolysine is a recently discovered amino acid in the methanogenic archaeon Methanosarcina barkeri, where it supposedly plays a critical role in methyltransferase chemistry as an electrophile [6,7]. Traditional genomic sequence analyses tend to overlook these RAAs, leading to mis-annotation in the sequence databases. Systematic bioinformatic investigations of the genomic data offer the possibility of understanding which organisms utilize RAAs, and which proteins in particular incorporate them into their structures.
Predicting which natural proteins contain the RAA selenocysteine or pyrrolysine on the basis of genomic sequence data is a difficult problem [2]. The difficulty arises from the distinction that, unlike other amino acids, RAAs are not coded for by dedicated codons. Instead, they are incorporated in special circumstances by the UGA (opal; selenocysteine) and the UAG (amber; pyrrolysine) codons [3-7], which are ordinarily interpreted as stop signals to terminate translation (Figure 1a). From a genomics point of view, the problem is how to discriminate between all the true stop signals in genomic sequence data, and those cases that signal for incorporation of a RAA. At the mRNA level, one feature referred to as the selenocysteine insertion sequence (SECIS) hairpin motif is understood to signal for selenocysteine insertion. The situation is greatly complicated, however, by the divergence of the signal between different proteins and between different organisms with respect to the sequence and position of the signaling element, situated in either the 3' or 5' untranslated region of a recoded open reading frame (ORF; archaea/eukaryotes) or downstream of the recoded UGA (bacteria). Much less is understood about the newly discovered pyrrolysine incorporation machinery. The presence of a PYLIS (SECIS-equivalent) cis-acting element [2], and competition between translational termination and read-through, have been anticipated [11].
A number of earlier studies by Gladyshev and coworkers [12-16] have addressed the problem of predicting selenoproteomes, producing sets of selenoproteins encoded in various genomes. Systematic selenocysteine predictions in prokaryotes have been based on two criteria: alignment of the 'UGA' codon in the mRNA sequence with cysteine in homologous proteins in a pair-wise sequence alignment (henceforth, the cysteine alignment criterion), and the detection of a consensus SECIS signal in the nucleotide sequences (henceforth, the SECIS criterion). Both methods performed very well with near-zero false negatives [13,16]. Nevertheless, certain aspects of these approaches make them less suitable for generalized applications. For example, they cannot be applied to selenoproteins that fail to fit the cysteine alignment criterion (those selenoproteins that do not have a homolog in the database with a cysteine residue taking the place of the selenocysteines). The SECIS criterion also presents some limitations. High numbers of false positives arise with the genome-wide prediction of short, local RNA folding motifs, such as the SECIS element [17]. The observation that different organisms have divergent signals for selenocysteine insertion complicates the problem further [13,16]. Other models that do not rely on the identification of specific recoding signals, such as evaluation of the coding potential of the nucleotide sequence beyond the UGA termini, have been developed for eukaryotes [14]. To overcome the various difficulties associated with the detection of rare selenoproteins from genomic data, a combination of strategies is shown to be advantageous [2,14]. A database homology search using the entire lengths of candidate genes with an in-frame UAG codon has been employed recently for analyzing the nature of pyrrolysine decoding in methanogens [11].
Here we expand upon ideas developed by Gladyshev and colleagues [12-16], and introduce a new, multi-component scheme for microbial selenocysteine and pyrrolysine prediction. Several criteria are combined in series, including a new predictive element, 'read-through similarity analysis' (RSA; Figure 1b). The RSA criterion is applied in the early stage of the procedure to evaluate the read-through potential of an ORF based on an analysis of sequence similarity involving the hypothetical amino acid sequence translated beyond the candidate stop codon. This scheme is model-free, in the sense that it does not rely on any special RNA context, read-through mechanism, or incorporation of any particular amino acid residue at the recoding site. Following the RSA analysis, subsequent criteria (for example, cysteine alignment and SECIS) can be enforced, or overridden in special cases where the other criteria provide compelling evidence for a bona-fide read-through situation. Success of this predictive approach is not, therefore, strictly contingent on the presence of a protein homolog containing a cysteine substitution in the database or on a canonical SECIS motif in the case of selenoproteins. In addition to almost all of the known cases of UGA-encoded selenocysteines (Table 1), the present method successfully identifies several proteins with UAG-encoded pyrrolysine (Table 2), including novel candidates, as well as instances of genome-wide redefinition of UGA as a particular amino acid, such as tryptophan in Mycoplasma spp. The generality and wide applicability of the present approach makes it well suited to the critical problem of analyzing the rapidly growing number of new genomes.
Results and discussion
The selenoprotein prediction scheme
Our selenoproteome prediction scheme was developed based on the expectation that a putative selenoprotein will satisfy the following, specific conditions. It should show: a significant 'read-through similarity' (see below); an alignment of the selenocysteine residue with semi-invariant cysteine residue(s) in a set of aligned homologs; and a hairpin motif (putative SECIS) near the candidate ORF, which is consistent with the hairpin motifs near the other selenoproteins found in the same organism. The components of the predictive approach are combined as shown in Figure 1c. The RSA method incorporates an analysis of the protein sequences following the presumptive stop codons in a genome (Figure 1b). Due to the recoding of UGA as a selenocysteine, the sequence following the UGA codon would be translated as the carboxy-terminal part of an extended protein. This makes it possible to identify candidate selenoproteins in situations where the putative protein sequence immediately following a UGA codon is statistically similar to the aligned region of another homologous protein in a protein sequence database. The statistical detection of sequence homology in relatively short regions following the presumptive stop codon is achieved using a modified interpretation of standard dynamic alignment methods [18,19] (see Materials and methods section).
A search for selenoproteins was restricted to those organisms that contain at least one of the genes that are required for synthesizing selenoproteins [3,4]. A set of 35 microbial genomes that have one or more of the three essential components of the selenocysteine insertion device (SID; SelA, the seryl tRNA selenium transferase; SelB, the elongation factor; and SelC, the sec-tRNA gene) were used (see Additional data file 1 for a list). The labile selenium donor selenophosphate synthetase (SelD) was not included as part of the SID because it can be a selenoprotein itself.
The RSA method was applied to all the predicted theoretical ORFs (length ≥ 90 residues) that contain an in-frame UGA stop codon. Out of a total 203,339 ORFs analyzed, 3,594 satisfied the test for likely similarity in the read-through region. These were subjected to further analysis.
Multiple sequence alignments (MSAs) were used as a subsequent step in analyzing the candidate selenoproteins, following the cysteine alignment criterion [13]. Cysteine residues often play special functional roles in proteins, such as in nucleophilic attack, or in metal coordination. A selenocysteine residue can substitute for a cysteine residue in these functional roles [10]. Functionally important residues usually form the most conserved features in a MSA. Therefore, we expect selenocysteine to align with conserved or semi-conserved residues (cysteines and selenocysteines) in homologous proteins. The MSA analysis step detected 109 candidate ORFs for further scrutiny.
As a final test, candidate selenoprotein genes were subjected to SECIS-element detection. Unlike archaea or eukaryotes, bacterial SECIS sequences are less conserved, thus complicating a search for a canonical SECIS profile [13], although a consensus bacterial SECIS model has been recently reported [16]. We used a fast, heuristic-based search [20] for a short hairpin motif common to a set of short, un-aligned mRNA segments downstream of the 'UGA' codon of the candidate selenoprotein ORFs in each bacterial organism (see Materials and methods section). The underlying assumption is that the SECIS elements in all the candidate mRNA strings within a given organism will have somewhat conserved primary (sequence) and secondary (base-paired) structures, so they can be recognized by the SID machinery in that organism. Thus, non-SECIS sequences should be distinguishable from well-aligned SECIS elements within an organism. This step was very useful in rejecting false positives when two or more bona fide selenoproteins were detected in an organism. In archaeal microbes, SECIS motif detection was not performed by the above method, as the SECISearch [12,13] program described earlier was sufficient.
The predicted selenoproteins
The multi-step selenoprotein prediction scheme was highly successful in detecting a large number of known selenoproteins in a range of organisms (Table 1; Figure 2a). A comparison of the number of selenoproteins detected by our method versus the existing selenoprotein entries in the database of recoded proteins for those organisms (RECODE [21]) is shown in Figure 2a. About 96% (estimated sensitivity) of the RECODE entries (53 out of 55) were successfully predicted. Approximately 90% (estimated specificity) of the selenoproteins predicted here belong to previously known families. Amongst the proteins identified, it was noteworthy that a remarkably high number (approximately 48%) of selenoproteins fall within the formate dehydrogenase (FDH) protein family (Figure 2b). FDH is a member of the molybdopterin-dependant FDH/DMSO reductase superfamily of homologous enzymes in the SCOP classification [22]. Several ORFs showed the presence of -CxxC- or -CxxCxxC- motifs typical of a special subset of redox proteins in which one of the cysteines is replaced with a selenocysteine. Consistent with earlier reports [13,23], a set of selenoproteins was identified in a group of methanogenic archaea (Table 1), including Methanococcus jannaschii, Methanopyrus kandleri and Methanococcus maripaludis. Apart from an almost complete coverage of all the known selenoproteins, our method identifies seven additional likely selenoproteins (Table 1) for further experimental validation.
Although our method was highly successful in detecting almost all of the selenoproteins in the known database, it could not detect two known selenoproteins. The first one was a SelD gene in Campylobacter jejuni that could not be identified due to a sequence error in the genomic data [16]. The second one was the radical S-adenosylmethionine (SAM) domain protein in Geobacter sulfurreducens. Here, the selenocysteine residue is situated too close to the carboxyl terminus, thus causing a very low RSA Z-value (1.8). This is a true false negative and illustrates a shortcoming of relying on read-through similarity.
One advantage of the generalized RSA approach over the existing SECIS search-based methods is its ability to detect selenoproteins with non-standard SECIS motifs. This requires overlooking the SECIS criterion, which is made possible in the present approach by the power and selectivity of the other two criteria (RSA and cysteine alignment). We were able to detect all four known selenoproteins in the piezophile Photobacterium profundum [24], two of which could not be detected by the SECIS criterion [16] due to the presence of a divergent SECIS element. In addition, a fifth candidate selenoprotein is identified here (Figure 2c), which had a divergent SECIS element and whose predicted selenocysteine residues line up with cysteine in all four homologous proteins identified. Putative SECIS motifs for these four selenoproteins and the additional candidate in P. profundum are presented in Figure 3a.
A second advantage of the RSA-based approach is the potential ability to detect selenoproteins that are not represented in the database by a homologous protein with a cysteine in the position corresponding to the presumptive stop codon. A close look at the multiple sequence alignments of certain selenoprotein homologs in the Conserved Domain database [25] indicated that nucleophilic serine, aspartate and glutamate residues sometimes replace the catalytic cysteine functionality. Unlike the previously described cysteine alignment criterion [13], the RSA-based approach does not analyze cysteine/selenocysteine alignment in an early stage. The presence of these conserved, non-cysteine residues aligned with putative selenocysteine can, therefore, be analyzed while inspecting the MSA, followed by an analysis of the SECIS feature. The protein formylmethanofuran dehydrogenase in M. maripaludis provides an example of a verified selenoprotein that is detected by our method without invoking the cysteine/selenocysteine alignment criterion (Figure 2d). The subject selenocysteine aligns with a set of aspartate residues in the MSA. However, glycine reductase A (GrdA), a selenoprotein whose homologs do not have cysteine in place of selenocysteine [13], could not be identified using our method on a test run. This failure resulted from a crucial lack of significant read-through similarity between GrdA and the other proteins homologous to GrdA.
A small number of ORFs (see Additional data file 2) were found with the translated UGA codon (U) aligned with strictly invariant nucleophilic residues (aspartate, glutamate or serine) in the MSA. None of these ORFS belong to previously known selenoprotein families or had a convincing SECIS motif adjacent to the UGA codon. Because of the lack of any additional evidence, it is not possible to further separate the true read-through events from the false-positives that might arise from statistical uncertainty or sequencing error. Nevertheless, some of these ORFs could be genuine read-through cases.
Putative pyrrolysine recoding in archaea
The RSA method was also used to search for proteins potentially containing the pyrrolysine residue, the so-called 22nd amino acid (Table 2). The pyrrolysine amino acid residue was recently discovered to be encoded by the UAG (amber) codon in the monomethylamine methyltransferase enzyme in Methanosarcina barkeri, where it serves as an electrophile to methylate the cobalt-corrinoid cofactor [6,7,26]. First, a search for homologs of the PylS gene (which codes for the pyrrolysine-specific aminoacyl tRNA synthetase [6,7]) in the available genomic data identified several methanogenic archaea as organisms likely to encode pyrrolysine containing proteins. These organisms include: Methanosarcina barkeri fusaro, Methanosarcina acetivorans, Methanosarcina mazei and Methanococcoides burtonii. Putative pyrrolysine-containing methylamine methyltransferses from methanogenesis pathways have been reported in this same set of organisms [11,26]. Within these four organisms, a total of 34 ORFs containing putative pyrrolysine residues were found to exhibit significant read-through similarity to homologous methyltransferases (Table 2). Out of 2,086 and 3,611 theoretical ORFs (see Materials and methods section) analyzed in the complete genomes of M. mazei and M. acetivorans, 87 and 97, respectively, showed significant read-through similarity. We have listed all those ORFs with an in-frame UAG codon that exhibit high RSA similarity, as well as well-aligned MSA for M. acetivorans and M. mazei (Additional data file 2). Apart from previously described transposases [11], the list contains several other candidate proteins, including a novel homolog of the cobalamin biosynthesis protein CobN (Figure 4c).
Overall distribution of the recoded proteins
A marked tendency was noted for the selenoproteins to occur in certain pathways and functional categories (Figure 2b). The majority of detected selenoproteins in bacteria were FDHs that convert formate to carbon dioxide in anaerobic environments [27]. Other known selenoproteins include SelD, GrdA and GrdB (from the anaerobic glycine reduction pathway), HesB (associated with the nitrogen fixation genes), and several oxidoreductases (for example, thioredoxin and peroxiredoxin). In archaea, selenocysteine usage appears to be confined to a small group of enzymes in the anaerobic methanogenesis pathway [23] (such as FDH and formylmethanofuran dehydrogenase from the FDH family, and heterodisulfide reductase) that have conceivably co-evolved under similar evolutionary constraints in a number of methanogens. Pyrrolysine-encoding is found in methyltransferases [26] from a pathway that converts methylamines to methane in Methanosarcina sp. and in the Antarctic archaeon M. burtonii. A high incidence of unusual stop codon reassignments, both selenocysteines and pyrrolysines, in methanogenesis enzymes in ancient archaea is intriguing.
Relative merit of the RSA-based approach
The selenoprotein identification scheme presented herein (an 'RSA-first, SECIS-later' approach) differs from the previously reported methods in several ways. Earlier studies (based on the SECIS search approach) provided an estimated rate of false SECIS hits to be 3 to 15 per 10 Mb [12,17] in eukaryotes, greatly surpassing the number of true selenoproteins. An improved result has been obtained by using a statistical profile computed from a training dataset of aligned known SECIS elements in metazoa [17]. A recent bacterial SECIS-search method analyzed 48,472 SECIS hits in a set of 29 organisms (representing 1.5% of all the UGA codons analyzed), out of which 28,974 (approximately 60%) were selected for further analysis of protein sequence conservation in the UGA flanking regions [16]. Still, difficulties remain for approaches that rely on detecting small RNA signal sequences as an early step in analysis, especially in situations such as new genomes, where the nature of the signal may not be understood in advance. Examining presumptive protein sequences as a prior step mitigates these difficulties. In the present study, although RSA was applied to fairly small segments of the protein sequences following the UGA codon, it was quite efficient in identifying candidates representing read-through events (Figure 1c). This ability of RSA to limit the predicted set to a relatively small, manageable number of likely candidates (3,594 out of 203,339, approximately 1.7%) facilitated further detailed calculations in genome-wide analyses. Of the small set of 109 ORFs selected by the subsequent MSA analysis, 92 (approximately 84%) were selected afterward as putative selenoproteins. Thus, an analysis of protein sequences is able to filter out most of the false-positives, without using any mRNA context information. Our combined 'RSA-first, SECIS-later' method is, therefore, applicable to cases (for example, P. profundum) where a divergent signal makes a SECIS-based search unsuitable [16]. In the present approach, it becomes possible to scrutinize putative non-canonical SECIS signals. In addition, our method provides a useful way to search for selenoproteins lacking homologs containing corresponding cysteine residues [13] (Figure 2d).
The RSA approach was likewise successful in predicting putative pyrrolysine-proteins in archaea. Out of the 9,515 theoretical ORFs analyzed for putative pyrrolysine residues in four methanogens, 321 ORFs (3.4%) displayed significant read-through similarity. Unlike the case for selenoproteins, a reliable benchmarking of pyrrolysine-protein predictions against a known dataset was not possible. The predicted result encompasses the previously reported methylamine methyltransferases [26], however, and includes a number of likely candidates for further experiments. Intriguingly, the putative pyrrolysine residues do not align so exclusively with a particular, conserved amino acid in homologous proteins (Figure 4a-c) [11]. The RSA method appears, therefore, to be generally useful as an initial predictor for pyrrolysine proteins. In addition, the RSA approach offers wider utility for identifying cases of genome-wide stop codon redefinition (for example, in Mycoplasma spp.; see Materials and methods section) or special instances of stop codon read-through (for example, UAG read-through in a pilus biosynthesis gene in E. coli [28] (data not shown)).
Conclusion
To summarize, we have developed a novel computational scheme for predicting selenocysteine and pyrrolysine residues in proteins and have applied the method to microbes with complete genomes. In addition to confirming well-known examples, our method predicts new prospective candidates for further experimental validation. A worldwide web site has been developed for the interested user community [29]. The method should be a useful tool for predicting rare amino acids, as well as other read-through events, and for correcting gene annotations in the growing genomic databases.
Materials and methods
All the complete genomes were obtained from the National Center for Biotechnology Information (NCBI) [30]. Unfinished M. barkeri and M. burtonii genomes were obtained from The Institute for Genomic Research [31]. A list of accession numbers is provided in the Additional data files. A Perl script was written to perform all the computations (available upon request). All computations were performed in a local cluster of Linux computers. tRNA genes were computationally identified using the tRNASCAN-SE program [32]. Genes encoding SelA and SelB were detected directly from annotated genomes from NCBI.
All theoretical ORFs (≥ 90 residues) that begin with a start codon (ATG, TTG or GTG) and end with a stop codon (TAA, TAG or TGA) and contain one in-frame TGA (for selenocysteine) or TAG (for pyrrolysine) codon were extracted from the genomic data for analysis. In order to detect two short SelW proteins in G. sulfurreducens and C. jejuni, a reduced (80 residue) length constraint was used.
Read-through similarity analysis (RSA)
For each of the predicted ORFs, the BLAST program [33,34] was used to search for homologous proteins in a customized sequence database. The BLAST search space was restricted to a window of a maximum 100 residue length, pivoting at the stop codon. The BLOSUM62 matrix was used throughout and the selenocysteine residue was treated as 'any amino acid' (X). The BLAST database contained a maximum of 650,870 protein sequences from all the annotated complete microbial genomes from NCBI (dated 4 December 2005). A self-excluding BLAST database was used for the homology search in each organism. Top hits (E-value ≤ 10-1) that encompassed either side of the stop codon were identified. For each of the selected, truncated ORF sequences ({x1, x2,..., xi,..., xu,..., xn} where n = min{u + 60, u + t}; u = position of the stop codon; t = position of the subsequent stop codon) and the corresponding top hit sequence from the BLAST search ({y1, y2..., yj,..., ym}), a (n + 1) by (m + 1) dynamic alignment matrix was calculated with an affine gap penalty function [18,19]. N-terminal overhangs for both the sequences were not penalized; the 0th row and the 0th column were initialized with zero values.
For each cell (i,j) in the matrix:
a(i,j) = s(i,j) + max { a(I - 1,j - 1),
b(i-1,j-1),
c(i-1,j-1) }
b(i,j) = max { -(h + g) + a(i,j - 1),
-g + b(i,j - 1),
-(h + g) + c(i,j - 1) }
c(i,j) = max { -(h + g) + a(I - 1,j),
-(h + g) + b(I - 1,j),
-g + c(I - 1,j) }
score (i,j) = max {a(i,j), b(i,j), c(i,j)}; s(i,j) → BLOSUM62 matrix; h = 12, g = 2
Best_scoreORF = max{score(n,j), j = 1,..,m} (1)
Because we were exclusively interested in the significance of the alignment at the carboxy-terminal extension region beyond the stop position, the highest score from the nth column (that is the alignment of the terminal residue xn of the truncated ORF with the {y1,..., ym} residues) was taken as the maximal score (Best_scoreORF) instead of the usual Smith-Waterman score. A Z-value was computed by shuffling the terminal extension region 100 times, re-computing the scores in the terminal block of the matrix ({xstop,..,xn} and {y1,..,ym}) and averaging the maximal score (<Best_scorerand>). A test calculation with 10,000 times shuffling for one genome produced similar results. The values from randomized sequences were used to calculate a Z-value:
ZORF = (Best_scoreORF - <Best_scorerand>)/standard_deviation (2)
A generally weak dependence on length and amino acid composition makes the Z-values (which follow an extreme value distribution) useful for evaluating the significance of alignment scores [35]. We have used a fairly conservative Z-value cutoff (Zc = 8.0) [35] to decide the statistical significance of a C-terminal alignment. Selection criteria had to be relaxed for two legitimate selenoproteins, a sulfur transferase in G. sulfurreducens (Z-value 7.9) and a coenzyme F420-reducing hydrogenase subunit in M. maripaludis (Z-value 4.6).
Multiple sequence alignment
For each of the selected candidate ORFs (Zc ≥ 8), a sensitive, iterative PSI-BLAST search was performed using position-specific scoring matrices. The top 10 hits (E ≤ 10-3) were used to construct a MSA with ClustalW [36]. Amino acids lining up with the putative selenocysteine residue were examined. Selenocysteines that aligned with two or more cysteine residues were selected for further analysis.
SECIS element analysis
In accordance with a recent analysis of bacterial SECIS elements [16], a 111 nucleotide long mRNA stretch surrounding the UGA codon position (-10 to +100) was extracted from each of the selected ORFs passing the previous tests in each bacterial organism. The extracted set of RNA sequences for each organism was used to detect a common, single hairpin motif using the rapid, heuristic-based RNAPROFILE program [20]. A test calculation predicted the known SECIS element of the gene encoding FDH from E. coli correctly [37] (Figure 3b). The putative SECIS hairpin motifs were manually inspected for consistency.
Control analysis
To evaluate the performance of the RSA step, we analyzed the Mycoplasma genitalium organism that utilizes UGA to code for tryptophan throughout its genome. M. genitalium is a small genome with 470 genes [38], the majority of which have a homolog in our database, thus minimizing database effects in our calculation. We applied the RSA method to all the theoretical ORFs with one in-frame TAA (313) or TAG (137) or TGA (780) codon. A self-excluding BLAST database of microbial proteins was used. In M. genitalium, over 78% of the TGA cases (91% when a self-included database was used) were identified by the RSA method as recoding events with a Z-value of 8 or higher. These cases aligned overwhelmingly with tryptophan residues in homologs. In contrast, only about 2% to 3% of the ORFs contatining a TAA or TAG stop codon passed the same RSA test.
We also applied the selenoprotein detection scheme to the Aeropyrum pernix (BA000002) genome, which does not contain any selenocysteine insertion genes. Out of 26 of 1,288 ORFs with in-frame 'UGA' that were selected by RSA (approximately 2%), none were selected in the subsequent MSA test.
A web-server for RSA analysis
A web-based service is available for RSA analysis of submitted DNA sequences [29]. The server was designed to analyze an ORF with one in-frame stop codon (UAA, UAG or UGA). A larger, non-redundant BLAST database (to be updated regularly) is used by the web server. The Z-value score and the MSA for the ORF are returned to the user.
Sensitivity and specificity
Sensitivity = true positive/(true positive + false negative)
Specificity = true positive/(true positive + false positive)
Estimates of true positives, false negatives and false positives were based on predictions performed on the set of organisms whose selenoproteins have been described in the RECODE [21] database (Figure 2a). The number of true positives is taken to be the number of predictions that are already known selenoproteins in the RECODE database. False negatives are those known selenoproteins not predicted by our method. False positives are difficult to estimate. As an extreme estimate, we have taken as an upper bound all those predictions that are not in the known database. The actual false positive rate is probably considerably lower than this estimate.
Additional data files
The following additional data are available with the online version of this paper. Additional data file 1 is a list of all the genomes analyzed together with the NCBI accession number. Additional data file 2 contains all the predicted recoded proteins from the complete genomes analyzed in this study in FASTA format.
Supplementary Material
Additional File 1
A list of all the genomes analyzed together with the NCBI accession number
Click here for file
Additional File 2
All the predicted recoded proteins from the complete genomes analyzed in this study in FASTA format
Click here for file
Acknowledgements
This work was supported by the DOE office of Biological and Environmental Research. The authors thank T Holton for assistance with the web-based server preparation.
Figures and Tables
Figure 1 Schematic representation of the selenocysteine insertion machinery and the selenoprotein detection scheme. (a) A cartoon diagram of selenocysteine incorporation during protein translation inside the cell. The selenocysteine-specific elongation factor (SelB; pink) is shown interacting with the selenocysteine insertion sequence (SECIS) hairpin element in the mRNA and tRNA-sec (SelC). The anticodon of SelC tRNA interacts with and recognizes the 'UGA' codon. The ribosome and other components of the translational machinery are omitted for clarity. (b) Schematic representation of the 'read-through similarity analysis' approach. The top BLAST hit is shown in blue. The window lengths used for the BLAST search and read-through similarity evaluation are marked in the drawing. (c) A flow chart describing how the different components of the predictive scheme are combined for selenoprotein prediction. ORF, open reading frame.
Figure 2 An overview of the predicted selenoproteome. (a) A Venn diagram representation of the overlap between the known selenoproteins in the RECODE database (bold line) and the results of our prediction method (plain line) over the same set of organisms as included in RECODE. (b) A pie chart illustrating the types of selenoproteins in our predicted dataset. The dataset was divided into the following groups: formate dehydrogenase (FDH) family enzymes; archaeal methanogenesis selenoproteins (excluding the FDH family); selenophosphate synthetase (SelD); other known selenoproteins (for example, thioredoxin, hesB); glycine reductase genes (GRD); and new candidate selenoproteins. (c) A section of the multiple sequence alignments (MSA) of the newly predicted candidate selenoprotein from P. profundum with its four homologs found in our database. Note the alignment of putative selenocysteine (U denotes selenocysteine) with cysteine residues in the MSA. (d) The MSA of a selenoprotein formylmethanofuran dehydrogenase from M. maripaludis in which the recoded selenocysteine aligns with a set of conserved aspartate residues rather than the cysteine residues. The MSA illustrations were prepared using ALSCRIPT [39].
Figure 3 Representatives of the putative selenocysteine insertion sequence (SECIS) hairpin elements in various genomes as identified by the present study. (a) The SECIS elements from the genes coding for the following proteins from P. profundum: 1, glycine reductase GrdA; 2, glycine reductase GrdB2; 3, glycine reductase GrdA; 4, selenophosphate synthetase (SelD); 5, a hypothetical protein. (b) The SECIS elements from the genes coding for the following proteins from E. coli: 1, formate dehydrogenase; 2, formate dehydrogenase-N; 3, formate dehydrogenase-O.
Figure 4 Sections of the multiple sequence alignments of the putative pyrrolysine-containing proteins. (a) A protein known to use UAG read-through, methylamine methyltransferase from M. acetivorans. (b) A putative methyltransferase from M. burtonii. (c) A predicted read-through ORF homologous to a cobalamin biosynthesis protein CobN (gi|20906100|gb|AAM31298.1|, Methanosarcina mazei Goe1) from M. acetivorans. Note the alignment of presumed pyrrolysine residues (denoted as X) with various amino acids.
Table 1 A list of predicted selenoproteins encoded by UGA read-through
Accession ID Organism Computationally identified selenoproteins* annotated by their homologs
AE000657 Aquifex aeolicus 1. gi|12515210|gb|AAG56295.1|AE005358_3 formate dehydrogenase-N, nitrate-inducible, alpha subunit [Escherichia coli]
2. gi|51589698|emb|CAH21328.1| selenide, water dikinase [Yersinia pseudotuberculosis IP 32953]
AE017125 Helicobacter hepaticus 1.gi|27362035|gb|AAO10941.1|AE016805_198 formate dehydrogenase, alpha subunit [Vibrio vulnificus CMCP6]
2. gi|46914191|emb|CAG20971.1| putative selenophosphate synthase [Photobacterium profundum]
AE017143 Haemophilus ducreyi 35000HP 1. gi|26108424|gb|AAN80626.1|AE016761_201 selenide, water dikinase [Escherichia coli CFT073]
AE004439 Pasteurella multocida 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
2. gi|5103639|dbj|BAA79160.1| 194 amino acid long hypothetical protein [Aeropyrum pernix K1]
AE005674 Shigella flexneri 2a 1. gi|12515215|gb|AAG56300.1|AE005358_8 orf; unknown function [Escherichia coli O157:H7 EDL933]
2. gi|1788928|gb|AAC75627.1| quinolinate synthetase, B protein; quinolinate synthetase, B protein, catalytic and NAD/flavoprotein subunit [Escherichia coli >K12]
3. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
4. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
5. gi|3868721|gb|AAD13462.1| selenopolypeptide subunit of formate dehydrogenase H; formate dehydrogenase H, selenopolypeptide subunit [Escherichia coli K12]
AE014073 Shigella flexneri 2a 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
2. gi|1788928|gb|AAC75627.1| quinolinate synthetase, B protein; quinolinate synthetase, B protein, catalytic and NAD/flavoprotein subunit [Escherichia coli K12]
3. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
4. gi|3868721|gb|AAD13462.1| selenopolypeptide subunit of formate dehydrogenase H; formate dehydrogenase H, selenopolypeptide subunit [Escherichia coli K12]
AE006469 Sinorhizobium meliloti 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
AE008691 Thermoanaerobacter tengcongensis 1. gi|41816370|gb|AAS11237.1| glycine reductase complex selenoprotein GrdA [Treponema denticola ATCC 35405]
2. gi|51857693|dbj|BAD41851.1| glycine reductase complex selenoprotein B [Symbiobacterium thermophilum IAM 14863]
3. gi|46914191|emb|CAG20971.1| putative selenophosphate synthase [Photobacterium profundum]
AE014075 Escherichia coli CFT073 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
2. gi|56130341|gb|AAV79847.1| formate dehydrogenase H [Salmonella enterica subsp. enterica serovar Paratyphi A str. ATCC 9150]
3. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
BA000007 Escherichia coli O157H7 1. gi|56130341|gb|AAV79847.1| formate dehydrogenase H [Salmonella enterica subsp. enterica serovar Paratyphi A str. ATCC 9150]
2. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
3. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
U00096 Escherichia coli K12 1. gi|5105267|dbj|BAA80580.1| 114 amino acid long hypothetical protein [Aeropyrum pernix K1]
2. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
3. gi|56130341|gb|AAV79847.1| formate dehydrogenase H [Salmonella enterica subsp. enterica serovar Paratyphi A str. ATCC 9150]
4. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
AE014299 Shewanella oneidensis 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
AE015451 Pseudomonas putida KT2440 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
AE004091 Pseudomonas aeruginosa 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
AE016958 Mycobacterium avium paratuberculosis 1. gi|13880045|gb|AAK44759.1| hypothetical protein MT0536 [Mycobacterium tuberculosis CDC1551]
2. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
AE017042 Yersinia pestis biovar Mediaevalis 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
AE009952 Yersinia pestis KIM 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
AL590842 Yersinia pestis CO92 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
AE017180 Geobacter sulfurreducens 1. gi|19918170|gb|AAM07420.1| 4-carboxymuconolactone decarboxylase [Methanosarcina acetivorans str. C2A]
2. gi|21956737|gb|AAM83670.1|AE013608_5 glutaredoxin 3 [Yersinia pestis KIM]
3. gi|37201109|dbj|BAC96933.1| thiol-disulfide isomerase and thioredoxins [Vibrio vulnificus YJ016]
4. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
5. gi|34105000|gb|AAQ61356.1| conserved hypothetical protein [Chromobacterium violaceum ATCC 12472]; gi|53758707|gb|AAU92998.1| HesB/YadR/YfhF family protein [Methylococcus capsulatus str. Bath];
6. gi|46914191|emb|CAG20971.1| Putative selenophosphate synthase [Photobacterium profundum]
7. gi|32448022|emb|CAD77542.1| peroxiredoxin [Pirellula sp.]
8. gi|29605647|dbj|BAC69712.1 hypothetical protein [Streptomyces avermitilis MA-4680] (SelW)
9. gi|34482757|emb|CAE09757.1| sulfur transferase precursor [Wolinella succinogenes]
AE017226 Treponema denticola ATCC 35405 1. gi|51857694|dbj|BAD41852.1| glycine reductase complex selenoprotein A [Symbiobacterium thermophilum IAM 14863]
2. gi|51857693|dbj|BAD41851.1| glycine reductase complex selenoprotein B [Symbiobacterium thermophilum IAM 14863]
3. gi|56380162|dbj|BAD76070.1| glutathione peroxidase [Geobacillus kaustophilus HTA426]
4. gi|51857693|dbj|BAD41851.1| glycine reductase complex selenoprotein B [Symbiobacterium thermophilum IAM 14863]
5. gi|26108424|gb|AAN80626.1|AE016761_201 selenide, water dikinase [Escherichia coli CFT073]
6. gi|52209545|emb|CAH35498.1| thioredoxin 1 [Burkholderia pseudomallei K96243]
AL111168 Campylobacter jejuni 1. gi|27362035|gb|AAO10941.1|AE016805_198 formate dehydrogenase, alpha subunit [Vibrio vulnificus CMCP6]
2. gi|54018125|dbj|BAD59495.1| hypothetical protein [Nocardia farcinica IFM 10152]; (SelW)
AL513382 Salmonella typhi 1. gi|3868721|gb|AAD13462.1| selenopolypeptide subunit of formate dehydrogenase H; formate dehydrogenase H, selenopolypeptide subunit [Escherichia coli K12]
2. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
AE006468 Salmonella typhimurium LT2 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
2. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
3. gi|3868721|gb|AAD13462.1| selenopolypeptide subunit of formate dehydrogenase H; formate dehydrogenase H, selenopolypeptide subunit [Escherichia coli K12]
BA000016 Clostridium perfringens 1. gi|28202985|gb|AAO35429.1| conserved protein [Clostridium tetani E88]; gi|20906561|gb|AAM31712.1| HesB protein [Methanosarcina mazei Goe1]
2. gi|46914191|emb|CAG20971.1| putative selenophosphate synthase [Photobacterium profundum]
BX470251 Photorhabdus luminescens 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase alpha subunit [Aquifex aeolicus VF5]
BX571656 Wolinella succinogenes 1. gi|27362035|gb|AAO10941.1|AE016805_198 formate dehydrogenase, alpha subunit [Vibrio vulnificus CMCP6]
L42023 Haemophilus influenzae 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
2. gi|26108424|gb|AAN80626.1|AE016761_201 selenide, water dikinase [Escherichia coli CFT073]
CR354531 Photobacterium profundum 1. gi|58428447|gb|AAW77484.1| conserved hypothetical protein [Xanthomonas oryzae pv. oryzae KACC10331]
CR354532 Photobacterium profundum 1. gi|41816370|gb|AAS11237.1| glycine reductase complex selenoprotein GrdA [Treponema denticola ATCC 35405]
2. gi|51589698|emb|CAH21328.1| selenide, water dikinase [Yersinia pseudotuberculosis IP 32953]
3. gi|41816370|gb|AAS11237.1| glycine reductase complex selenoprotein GrdA [Treponema denticola ATCC 35405]
4. gi|41818450|gb|AAS12639.1| glycine reductase complex selenoprotein GrdB2 [Treponema denticola ATCC 35405]
AE009439 Methanopyrus kandleri (archaea) 1. gi|2622673|gb|AAB86026.1| formate dehydrogenase, alpha subunit homolog [Methanothermobacter thermautotrophicus]; gi|2622681|gb|AAB86033.1| tungsten formylmethanofuran dehydrogenase, subunit B [Methanothermobacter thermautotrophicus]
2. gi|57160335|dbj|BAD86265.1| probable formate dehydrogenase, alpha subunit [Thermococcus kodakaraensis KOD1]
3. gi|33566318|emb|CAE37231.1| putative iron-sulfur binding protein [Bordetella parapertussis]
4. gi|44921146|emb|CAF30381.1| heterodisulfide reductase, subunit A [Methanococcus maripaludis]
5. gi|44921142|emb|CAF30377.1| coenzyme F420-non-reducing hydrogenase, subunit delta [Methanococcus maripaludis]; gi|2622243|gb|AAB85627.1| methyl viologen-reducing hydrogenase, delta subunit homolog FlpD [Methanothermobacter thermautotrophicus]; gi|20904385|gb|AAM29752.1| heterodisulfate reductase, subunit A [Methanosarcina mazei Goe1]
6. gi|45047811|emb|CAF30938.1| coenzyme F420-reducing hydrogenase subunit alpha [Methanococcus maripaludis]
7. gi|39576202|emb|CAE80367.1| selenide, water dikinase [Bdellovibrio bacteriovorus HD100]
L77117 Methanococcus jannaschii (archaea) 1. gi|44921146|emb|CAF30381.1| heterodisulfide reductase subunit A [Methanococcus maripaludis]
2. gi|45047811|emb|CAF30938.1| coenzyme F420-reducing hydrogenase subunit alpha [Methanococcus maripaludis]
3. gi|50875900|emb|CAG35740.2| methyl-viologen-reducing hydrogenase, delta subunit [Desulfotalea psychrophila LSv54]
4. gi|2622240|gb|AAB85625.1| methyl viologen-reducing hydrogenase, delta subunit [Methanothermobacter thermautotrophicus]; gi|44921142|emb|CAF30377.1| coenzyme F420-non-reducing hydrogenase subunit delta [Methanococcus maripaludis]
5. gi|2622673|gb|AAB86026.1| formate dehydrogenase, alpha subunit homolog [Methanothermobacter thermautotrophicus]; gi|45048129|emb|CAF31247.1| tungsten containing formylmethanofuran dehydrogenase, subunit B [Methanococcus maripaludis] (overlaps with #4)
6. gi|26108424|gb|AAN80626.1|AE016761_201 selenide, water dikinase [Escherichia coli CFT073]
7. gi|53758707|gb|AAU92998.1| HesB/YadR/YfhF family protein [Methylococcus capsulatus str. Bath]
8. gi|45047727|emb|CAF30854.1| formate dehydrogenase, alpha subunit [Methanococcus maripaludis]
BX950229 Methanococcus maripaludis (archaea) 1. gi|2622673|gb|AAB86026.1| formate dehydrogenase, alpha subunit homolog [Methanothermobacter thermautotrophicus]; gi|19886584|gb|AAM01476.1| Formylmethanofuran dehydrogenase subunit B [Methanopyrus kandleri AV19]
2. gi|2622673|gb|AAB86026.1| formate dehydrogenase, alpha subunit homolog [Methanothermobacter thermautotrophicus]
3. gi|2622240|gb|AAB85625.1| methyl viologen-reducing hydrogenase, delta subunit [Methanothermobacter thermautotrophicus]; gi|39981962|gb|AAR33424.1| heterodisulfide reductase subunit [Geobacter sulfurreducens PCA]
4. gi|2622673|gb|AAB86026.1| formate dehydrogenase, alpha subunit homolog [Methanothermobacter thermautotrophicus]
5. gi|2622673|gb|AAB86026.1| formate dehydrogenase, alpha subunit homolog [Methanothermobacter thermautotrophicus]; gi|19918286|gb|AAM07526.1| formylmethanofuran dehydrogenase, subunit B [Methanosarcina acetivorans str. C2A]
6. gi|19886593|gb|AAM01482.1| Heterodisulfide reductase, subunit A, polyferredoxin [Methanopyrus kandleri AV19]
Organism names, National Center for Biotechnology Information accession numbers for the genomes and the top PSI-BLAST hit(s) from our database are shown. Seven novel candidate selenoproteins are shown in bold type. *Each entry corresponds to a computationally identified read-through protein in the organism indicated to the left. FASTA files for these recoded protein sequences are provided in the Additional file 2. For each recoded protein, the GI number and the functional annotation for a homologous protein are given.
Table 2 Methyltransferases predicted to encode pyrrolysine by UAG read-through in a set of methanogenic archaea
Organism Computationally identified pyrrolysine-proteins* annotated by their homologs
Methanosarcina acetivorans (AE010299) 1. gi|56678713|gb|AAV95379.1| trimethylamine methyltransferase family protein [Silicibacter pomeroyi DSS-3]
2. gi|14247242|dbj|BAB57633.1| menaquinone biosynthesis methyltransferase [Staphylococcus aureus subsp. Aureus Mu50]
3. gi|36785418|emb|CAE14364.1| protein methyltranferase [Photorhabdus luminescens subsp. laumondii TTO1]
4. gi|56679325|gb|AAV95991.1| trimethylamine methyltransferase family protein [Silicibacter pomeroyi DSS-3]
5. i|20904823|gb|AAM30145.1| SAM-dependent methyltransferases [Methanosarcina mazei Goe1]
6. gi|56312282|emb|CAI06927.1| predicted methyltransferase [Azoarcus sp. EbN1]
7. gi|45047608|emb|CAF30735.1| generic methyltransferase [Methanococcus maripaludis]
8. gi|20905508|gb|AAM30766.1| methylcobalamin: Coenzyme M methyltransferase [Methanosarcina mazei Goe1]
9. Predicted ORF monomethylamine methyltransferase [Methanosarcina mazei Goe1]†
10. Predicted ORF monomethylamine methyltransferase [Methanosarcina mazei Goe1]†
11. Predicted ORF dimethylamine methyltransferase [Methanosarcina mazei Goe1]†
12. Predicted ORF dimethylamine methyltransferase [Methanosarcina mazei Goe1]†
13. Predicted ORF dimethylamine methyltransferase [Methanosarcina mazei Goe1]†
Methanosarcina mazei (AE008384) 1. gi|19914316|gb|AAM03972.1| trimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
2. gi|19914320|gb|AAM03976.1| dimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
3. gi|19914753|gb|AAM04365.1| trimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
4. gi|19913899|gb|AAM03597.1| monomethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
5. gi|19914755|gb|AAM04366.1| dimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
6. gi|19914320|gb|AAM03976.1| dimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
7. gi|19913899|gb|AAM03597.1| monomethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
Methanosarcina barkeri (draft genome) 1. gi|19914320|gb|AAM03976.1| dimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
2. gi|19913899|gb|AAM03597.1| monomethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
3. gi|19914316|gb|AAM03972.1| trimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
4. gi|19914320|gb|AAM03976.1| dimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
5. gi|19914334|gb|AAM03988.1| protein-L-isoaspartate (D-aspartate) O-methyltransferase [Methanosarcina acetivorans str. C2A]
6. gi|19913899|gb|AAM03597.1| monomethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
7. gi|19913899|gb|AAM03597.1| monomethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
Methanococcoides burtonii (draft genome) 1. gi|19914320|gb|AAM03976.1| dimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
2. gi|19914753|gb|AAM04365.1| trimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
3. gi|5458504|emb|CAB49992.1| methlytransferase, putative [Pyrococcus abyssi]
4. gi|5458504|emb|CAB49992.1| methlytransferase, putative [Pyrococcus abyssi] (overlaps with #3)
5. gi|19914320|gb|AAM03976.1| dimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
6. gi|19914753|gb|AAM04365.1| trimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
7. gi|19913899|gb|AAM03597.1| monomethylamine methyltransferase [Methanosarcina acetivorans str. C2A
*Each entry corresponds to a computationally identified read-through protein in the organism indicated to the left. FASTA files for these recoded protein sequences are provided in the Additional data files. For each recoded protein, the GI number and the functional annotation for a homologous protein are given. †These open reading frames (ORFs) in M. acitovorans were predicted during a repeat search using a BLAST database containing putative methylamine methyltransferase ORFs in M. mazei as identified by our method. Although the M. acitovorans genome was annotated for several pyrrolysine-containing methylamine methyltranferases, this was not the case with the M. mazei genome. Thus, several methyltransferases that are specific to these methanosarcina species could not be detected in our original calculation due to the lack of read-through homologs. Such repeat searches were not performed for the two unfinished genomes.
==== Refs
Gesteland RF Atkins JF Recoding: dynamic reprogramming of translation. Annu Rev Biochem 1996 65 741 768 8811194 10.1146/annurev.bi.65.070196.003521
Namy O Rousset JP Napthine S Brierley I Reprogrammed genetic decoding in cellular gene expression. Mol Cell 2004 13 157 168 14759362 10.1016/S1097-2765(04)00031-0
Bock A Biosynthesis of selenoproteins: an overview. Biofactors 2000 11 77 78 10705967
Stadtman TC Selenocysteine. Annu Rev Biochem 1996 65 83 100 8811175 10.1146/annurev.bi.65.070196.000503
Hatfield DL Gladyshev VN How selenium has altered our understanding of the genetic code. Mol Cell Biol 2002 22 3565 3576 11997494 10.1128/MCB.22.11.3565-3576.2002
Srinivasan G James CM Krzycki JA Pyrrolysine encoded by UAG in Archaea: charging of a UAG-decoding specialized tRNA. Science 2002 296 1459 1462 12029131 10.1126/science.1069588
Hao B Gong W Ferguson TK James CM Krzycki JA Chan MK A new UAG-encoded residue in the structure of a methanogen methyltransferase. Science 2002 296 1462 1466 12029132 10.1126/science.1069556
Rayman MP The importance of selenium to human health. Lancet 2000 356 233 241 10963212 10.1016/S0140-6736(00)02490-9
Frankenberger WT JrArshad M Bioremediation of selenium-contaminated sediments and water. Biofactors 2001 14 241 254 11568461
Jacob C Giles GI Giles NM Sies H Sulfur and selenium: the role of oxidation state in protein structure and function. Angew Chem Int Ed Engl 2003 42 4742 4758 14562341 10.1002/anie.200300573
Zhang Y Baranov PV Atkins JF Gladyshev VN Pyrrolysine and selenocysteine use dissimilar decoding strategies. J Biol Chem 2005 280 20740 20751 15788401 10.1074/jbc.M501458200
Kryukov GV Castellano S Novoselov SV Lobanov AV Zehtab O Guigo R Gladyshev VN Characterization of mammalian selenoproteomes. Science 2003 300 1439 1443 12775843 10.1126/science.1083516
Kryukov GV Gladyshev VN The prokaryotic selenoproteome. EMBO Rep 2004 5 538 543 15105824 10.1038/sj.embor.7400126
Castellano S Novoselov SV Kryukov GV Lescure A Blanco E Krol A Gladyshev VN Guigo R Reconsidering the evolution of eukaryotic selenoproteins: a novel nonmammalian family with scattered phylogenetic distribution. EMBO Rep 2004 5 71 77 14710190 10.1038/sj.embor.7400036
Zhang Y Fomenko DE Gladyshev VN The microbial selenoproteome of the Sargasso Sea. Genome Biol 2005 6 R37 15833124 10.1186/gb-2005-6-4-r37
Zhang Y Gladyshev VN An algorithm for identification of bacterial selenocysteine insertion sequence elements and selenoprotein genes. Bioinformatics 2005 21 2580 2589 15797911 10.1093/bioinformatics/bti400
Lambert A Lescure A Gautheret D A survey of metazoan selenocysteine insertion sequences. Biochimie 2002 84 953 959 12458087 10.1016/S0300-9084(02)01441-4
Gotoh O An improved algorithm for matching biological sequences. J Mol Biol 1982 162 705 708 7166760 10.1016/0022-2836(82)90398-9
Setubal C Meidanis J Introduction to Computational Molecular Biology 1997 Boston: PWS Publishing Company
Pavesi G Mauri G Stefani M Pesole G RNAProfile: an algorithm for finding conserved secondary structure motifs in unaligned RNA sequences. Nucleic Acids Res 2004 32 3258 3269 15199174 10.1093/nar/gkh650
Baranov PV Gurvich OL Hammer AW Gesteland RF Atkins JF RECODE 2003. Nucleic Acids Res 2003 31 87 89 12519954 10.1093/nar/gkg024
Murzin AG Brenner SE Hubbard T Chothia C SCOP: a structural classification of proteins database for the investigation of sequences and structures. J Mol Biol 1995 247 536 540 7723011 10.1006/jmbi.1995.0159
Cobucci-Ponzano B Rossi M Moracci M Recoding in archaea. Mol Microbiol 2005 55 339 348 15659155 10.1111/j.1365-2958.2004.04400.x
Vezzi A Campanaro S D'Angelo M Simonato F Vitulo N Lauro FM Cestaro A Malacrida G Simionati B Cannata N Life at depth: Photobacterium profundum genome sequence and expression analysis. Science 2005 307 1459 1461 15746425 10.1126/science.1103341
Marchler-Bauer A Anderson JB DeWeese-Scott C Fedorova ND Geer LY He S Hurwitz DI Jackson JD Jacobs AR Lanczycki CJ CDD: a curated Entrez database of conserved domain alignments. Nucleic Acids Res 2003 31 383 387 12520028 10.1093/nar/gkg087
Krzycki JA Function of genetically encoded pyrrolysine in corrinoid-dependent methylamine methyltransferases. Curr Opin Chem Biol 2004 8 484 491 15450490 10.1016/j.cbpa.2004.08.012
Jormakka M Byrne B Iwata S Formate dehydrogenase: a versatile enzyme in changing environments. Curr Opin Struct Biol 2003 13 418 423 12948771 10.1016/S0959-440X(03)00098-8
Jalajakumari MB Thomas CJ Halter R Manning PA Genes for biosynthesis and assembly of CS3 pili of CFA/II enterotoxigenic Escherichia coli: novel regulation of pilus production by bypassing an amber codon. Mol Microbiol 1989 3 1685 1695 2576094
Read-through Similarity Analysis
National Center for Biotechnology Information
The Institute for Genomic Research
Lowe TM Eddy SR tRNAscan-SE: a program for improved detection of transfer RNA genes in genomic sequence. Nucleic Acids Res 1997 25 955 964 9023104 10.1093/nar/25.5.955
Altschul SF Gish W Miller W Myers EW Lipman DJ Basic local alignment search tool. J Mol Biol 1990 215 403 410 2231712 10.1006/jmbi.1990.9999
Altschul SF Madden TL Schaffer AA Zhang J Zhang Z Miller W Lipman DJ Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res 1997 25 3389 3402 9254694 10.1093/nar/25.17.3389
Comet JP Aude JC Glemet E Risler JL Henaut A Slonimski PP Codani JJ Significance of Z-value statistics of Smith-Waterman scores for protein alignments. Comput Chem 1999 23 317 331 10627144 10.1016/S0097-8485(99)00008-X
Thompson JD Higgins DG Gibson TJ CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res 1994 22 4673 4680 7984417
Liu Z Reches M Groisman I Engelberg-Kulka H The nature of the minimal 'selenocysteine insertion sequence' (SECIS) in Escherichia coli. Nucleic Acids Res 1998 26 896 902 9461445 10.1093/nar/26.4.896
Fraser CM Gocayne JD White O Adams MD Clayton RA Fleischmann RD Bult CJ Kerlavage AR Sutton G Kelley JM The minimal gene complement of Mycoplasma genitalium. Science 1995 270 397 403 7569993
Barton GJ ALSCRIPT: a tool to format multiple sequence alignments. Protein Eng 1993 6 37 40 8433969
|
16168086
|
PMC1242214
|
CC BY
|
2021-01-04 16:05:42
|
no
|
Genome Biol. 2005 Aug 31; 6(9):R79
|
utf-8
|
Genome Biol
| 2,005 |
10.1186/gb-2005-6-9-r79
|
oa_comm
|
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-9-r791616808610.1186/gb-2005-6-9-r79MethodA computational method to predict genetically encoded rare amino acids in proteins Chaudhuri Barnali N [email protected] Todd O [email protected] UCLA-DOE Institute for Genomics and Proteomics and Department of Chemistry and Biochemistry, University of California, Los Angeles, USA2005 31 8 2005 6 9 R79 R79 8 3 2005 20 6 2005 27 7 2005 Copyright © 2005 Chaudhuri and Yeates; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
A new method for predicting recoding by rare amino acids such as selenocysteine and pyrrolysine was used to survey a set of microbial genomes.
In several natural settings, the standard genetic code is expanded to incorporate two additional amino acids with distinct functionality, selenocysteine and pyrrolysine. These rare amino acids can be overlooked inadvertently, however, as they arise by recoding at certain stop codons. We report a method for such recoding prediction from genomic data, using read-through similarity evaluation. A survey across a set of microbial genomes identifies almost all the known cases as well as a number of novel candidate proteins.
==== Body
Background
Codon redefinitions that expand upon the standard genetic code beyond the 20 canonical amino acids are reported in all three domains of life [1,2]. Two known genetically encoded rare amino acids (RAAs) are selenocysteine and pyrrolysine, the proposed 21st and the 22nd amino acids, respectively [3-7]. Selenocysteine, a selenium-analog of cysteine, is a potent nucleophile [5] and has been reported in organisms as diverse as Escherichia coli and human beings [4,5]. Selenium plays a dual role in nature as an essential micronutrient in human health, and as an environmental hazard to humans, livestock and wildlife [8] when it is present in high amounts. Thus, selenium is a target for both molecular biology and bioremediation research [8,9]. The distribution of selenium in the form of selenocysteine residues [5,10] in specific proteins is not completely understood. Pyrrolysine is a recently discovered amino acid in the methanogenic archaeon Methanosarcina barkeri, where it supposedly plays a critical role in methyltransferase chemistry as an electrophile [6,7]. Traditional genomic sequence analyses tend to overlook these RAAs, leading to mis-annotation in the sequence databases. Systematic bioinformatic investigations of the genomic data offer the possibility of understanding which organisms utilize RAAs, and which proteins in particular incorporate them into their structures.
Predicting which natural proteins contain the RAA selenocysteine or pyrrolysine on the basis of genomic sequence data is a difficult problem [2]. The difficulty arises from the distinction that, unlike other amino acids, RAAs are not coded for by dedicated codons. Instead, they are incorporated in special circumstances by the UGA (opal; selenocysteine) and the UAG (amber; pyrrolysine) codons [3-7], which are ordinarily interpreted as stop signals to terminate translation (Figure 1a). From a genomics point of view, the problem is how to discriminate between all the true stop signals in genomic sequence data, and those cases that signal for incorporation of a RAA. At the mRNA level, one feature referred to as the selenocysteine insertion sequence (SECIS) hairpin motif is understood to signal for selenocysteine insertion. The situation is greatly complicated, however, by the divergence of the signal between different proteins and between different organisms with respect to the sequence and position of the signaling element, situated in either the 3' or 5' untranslated region of a recoded open reading frame (ORF; archaea/eukaryotes) or downstream of the recoded UGA (bacteria). Much less is understood about the newly discovered pyrrolysine incorporation machinery. The presence of a PYLIS (SECIS-equivalent) cis-acting element [2], and competition between translational termination and read-through, have been anticipated [11].
A number of earlier studies by Gladyshev and coworkers [12-16] have addressed the problem of predicting selenoproteomes, producing sets of selenoproteins encoded in various genomes. Systematic selenocysteine predictions in prokaryotes have been based on two criteria: alignment of the 'UGA' codon in the mRNA sequence with cysteine in homologous proteins in a pair-wise sequence alignment (henceforth, the cysteine alignment criterion), and the detection of a consensus SECIS signal in the nucleotide sequences (henceforth, the SECIS criterion). Both methods performed very well with near-zero false negatives [13,16]. Nevertheless, certain aspects of these approaches make them less suitable for generalized applications. For example, they cannot be applied to selenoproteins that fail to fit the cysteine alignment criterion (those selenoproteins that do not have a homolog in the database with a cysteine residue taking the place of the selenocysteines). The SECIS criterion also presents some limitations. High numbers of false positives arise with the genome-wide prediction of short, local RNA folding motifs, such as the SECIS element [17]. The observation that different organisms have divergent signals for selenocysteine insertion complicates the problem further [13,16]. Other models that do not rely on the identification of specific recoding signals, such as evaluation of the coding potential of the nucleotide sequence beyond the UGA termini, have been developed for eukaryotes [14]. To overcome the various difficulties associated with the detection of rare selenoproteins from genomic data, a combination of strategies is shown to be advantageous [2,14]. A database homology search using the entire lengths of candidate genes with an in-frame UAG codon has been employed recently for analyzing the nature of pyrrolysine decoding in methanogens [11].
Here we expand upon ideas developed by Gladyshev and colleagues [12-16], and introduce a new, multi-component scheme for microbial selenocysteine and pyrrolysine prediction. Several criteria are combined in series, including a new predictive element, 'read-through similarity analysis' (RSA; Figure 1b). The RSA criterion is applied in the early stage of the procedure to evaluate the read-through potential of an ORF based on an analysis of sequence similarity involving the hypothetical amino acid sequence translated beyond the candidate stop codon. This scheme is model-free, in the sense that it does not rely on any special RNA context, read-through mechanism, or incorporation of any particular amino acid residue at the recoding site. Following the RSA analysis, subsequent criteria (for example, cysteine alignment and SECIS) can be enforced, or overridden in special cases where the other criteria provide compelling evidence for a bona-fide read-through situation. Success of this predictive approach is not, therefore, strictly contingent on the presence of a protein homolog containing a cysteine substitution in the database or on a canonical SECIS motif in the case of selenoproteins. In addition to almost all of the known cases of UGA-encoded selenocysteines (Table 1), the present method successfully identifies several proteins with UAG-encoded pyrrolysine (Table 2), including novel candidates, as well as instances of genome-wide redefinition of UGA as a particular amino acid, such as tryptophan in Mycoplasma spp. The generality and wide applicability of the present approach makes it well suited to the critical problem of analyzing the rapidly growing number of new genomes.
Results and discussion
The selenoprotein prediction scheme
Our selenoproteome prediction scheme was developed based on the expectation that a putative selenoprotein will satisfy the following, specific conditions. It should show: a significant 'read-through similarity' (see below); an alignment of the selenocysteine residue with semi-invariant cysteine residue(s) in a set of aligned homologs; and a hairpin motif (putative SECIS) near the candidate ORF, which is consistent with the hairpin motifs near the other selenoproteins found in the same organism. The components of the predictive approach are combined as shown in Figure 1c. The RSA method incorporates an analysis of the protein sequences following the presumptive stop codons in a genome (Figure 1b). Due to the recoding of UGA as a selenocysteine, the sequence following the UGA codon would be translated as the carboxy-terminal part of an extended protein. This makes it possible to identify candidate selenoproteins in situations where the putative protein sequence immediately following a UGA codon is statistically similar to the aligned region of another homologous protein in a protein sequence database. The statistical detection of sequence homology in relatively short regions following the presumptive stop codon is achieved using a modified interpretation of standard dynamic alignment methods [18,19] (see Materials and methods section).
A search for selenoproteins was restricted to those organisms that contain at least one of the genes that are required for synthesizing selenoproteins [3,4]. A set of 35 microbial genomes that have one or more of the three essential components of the selenocysteine insertion device (SID; SelA, the seryl tRNA selenium transferase; SelB, the elongation factor; and SelC, the sec-tRNA gene) were used (see Additional data file 1 for a list). The labile selenium donor selenophosphate synthetase (SelD) was not included as part of the SID because it can be a selenoprotein itself.
The RSA method was applied to all the predicted theoretical ORFs (length ≥ 90 residues) that contain an in-frame UGA stop codon. Out of a total 203,339 ORFs analyzed, 3,594 satisfied the test for likely similarity in the read-through region. These were subjected to further analysis.
Multiple sequence alignments (MSAs) were used as a subsequent step in analyzing the candidate selenoproteins, following the cysteine alignment criterion [13]. Cysteine residues often play special functional roles in proteins, such as in nucleophilic attack, or in metal coordination. A selenocysteine residue can substitute for a cysteine residue in these functional roles [10]. Functionally important residues usually form the most conserved features in a MSA. Therefore, we expect selenocysteine to align with conserved or semi-conserved residues (cysteines and selenocysteines) in homologous proteins. The MSA analysis step detected 109 candidate ORFs for further scrutiny.
As a final test, candidate selenoprotein genes were subjected to SECIS-element detection. Unlike archaea or eukaryotes, bacterial SECIS sequences are less conserved, thus complicating a search for a canonical SECIS profile [13], although a consensus bacterial SECIS model has been recently reported [16]. We used a fast, heuristic-based search [20] for a short hairpin motif common to a set of short, un-aligned mRNA segments downstream of the 'UGA' codon of the candidate selenoprotein ORFs in each bacterial organism (see Materials and methods section). The underlying assumption is that the SECIS elements in all the candidate mRNA strings within a given organism will have somewhat conserved primary (sequence) and secondary (base-paired) structures, so they can be recognized by the SID machinery in that organism. Thus, non-SECIS sequences should be distinguishable from well-aligned SECIS elements within an organism. This step was very useful in rejecting false positives when two or more bona fide selenoproteins were detected in an organism. In archaeal microbes, SECIS motif detection was not performed by the above method, as the SECISearch [12,13] program described earlier was sufficient.
The predicted selenoproteins
The multi-step selenoprotein prediction scheme was highly successful in detecting a large number of known selenoproteins in a range of organisms (Table 1; Figure 2a). A comparison of the number of selenoproteins detected by our method versus the existing selenoprotein entries in the database of recoded proteins for those organisms (RECODE [21]) is shown in Figure 2a. About 96% (estimated sensitivity) of the RECODE entries (53 out of 55) were successfully predicted. Approximately 90% (estimated specificity) of the selenoproteins predicted here belong to previously known families. Amongst the proteins identified, it was noteworthy that a remarkably high number (approximately 48%) of selenoproteins fall within the formate dehydrogenase (FDH) protein family (Figure 2b). FDH is a member of the molybdopterin-dependant FDH/DMSO reductase superfamily of homologous enzymes in the SCOP classification [22]. Several ORFs showed the presence of -CxxC- or -CxxCxxC- motifs typical of a special subset of redox proteins in which one of the cysteines is replaced with a selenocysteine. Consistent with earlier reports [13,23], a set of selenoproteins was identified in a group of methanogenic archaea (Table 1), including Methanococcus jannaschii, Methanopyrus kandleri and Methanococcus maripaludis. Apart from an almost complete coverage of all the known selenoproteins, our method identifies seven additional likely selenoproteins (Table 1) for further experimental validation.
Although our method was highly successful in detecting almost all of the selenoproteins in the known database, it could not detect two known selenoproteins. The first one was a SelD gene in Campylobacter jejuni that could not be identified due to a sequence error in the genomic data [16]. The second one was the radical S-adenosylmethionine (SAM) domain protein in Geobacter sulfurreducens. Here, the selenocysteine residue is situated too close to the carboxyl terminus, thus causing a very low RSA Z-value (1.8). This is a true false negative and illustrates a shortcoming of relying on read-through similarity.
One advantage of the generalized RSA approach over the existing SECIS search-based methods is its ability to detect selenoproteins with non-standard SECIS motifs. This requires overlooking the SECIS criterion, which is made possible in the present approach by the power and selectivity of the other two criteria (RSA and cysteine alignment). We were able to detect all four known selenoproteins in the piezophile Photobacterium profundum [24], two of which could not be detected by the SECIS criterion [16] due to the presence of a divergent SECIS element. In addition, a fifth candidate selenoprotein is identified here (Figure 2c), which had a divergent SECIS element and whose predicted selenocysteine residues line up with cysteine in all four homologous proteins identified. Putative SECIS motifs for these four selenoproteins and the additional candidate in P. profundum are presented in Figure 3a.
A second advantage of the RSA-based approach is the potential ability to detect selenoproteins that are not represented in the database by a homologous protein with a cysteine in the position corresponding to the presumptive stop codon. A close look at the multiple sequence alignments of certain selenoprotein homologs in the Conserved Domain database [25] indicated that nucleophilic serine, aspartate and glutamate residues sometimes replace the catalytic cysteine functionality. Unlike the previously described cysteine alignment criterion [13], the RSA-based approach does not analyze cysteine/selenocysteine alignment in an early stage. The presence of these conserved, non-cysteine residues aligned with putative selenocysteine can, therefore, be analyzed while inspecting the MSA, followed by an analysis of the SECIS feature. The protein formylmethanofuran dehydrogenase in M. maripaludis provides an example of a verified selenoprotein that is detected by our method without invoking the cysteine/selenocysteine alignment criterion (Figure 2d). The subject selenocysteine aligns with a set of aspartate residues in the MSA. However, glycine reductase A (GrdA), a selenoprotein whose homologs do not have cysteine in place of selenocysteine [13], could not be identified using our method on a test run. This failure resulted from a crucial lack of significant read-through similarity between GrdA and the other proteins homologous to GrdA.
A small number of ORFs (see Additional data file 2) were found with the translated UGA codon (U) aligned with strictly invariant nucleophilic residues (aspartate, glutamate or serine) in the MSA. None of these ORFS belong to previously known selenoprotein families or had a convincing SECIS motif adjacent to the UGA codon. Because of the lack of any additional evidence, it is not possible to further separate the true read-through events from the false-positives that might arise from statistical uncertainty or sequencing error. Nevertheless, some of these ORFs could be genuine read-through cases.
Putative pyrrolysine recoding in archaea
The RSA method was also used to search for proteins potentially containing the pyrrolysine residue, the so-called 22nd amino acid (Table 2). The pyrrolysine amino acid residue was recently discovered to be encoded by the UAG (amber) codon in the monomethylamine methyltransferase enzyme in Methanosarcina barkeri, where it serves as an electrophile to methylate the cobalt-corrinoid cofactor [6,7,26]. First, a search for homologs of the PylS gene (which codes for the pyrrolysine-specific aminoacyl tRNA synthetase [6,7]) in the available genomic data identified several methanogenic archaea as organisms likely to encode pyrrolysine containing proteins. These organisms include: Methanosarcina barkeri fusaro, Methanosarcina acetivorans, Methanosarcina mazei and Methanococcoides burtonii. Putative pyrrolysine-containing methylamine methyltransferses from methanogenesis pathways have been reported in this same set of organisms [11,26]. Within these four organisms, a total of 34 ORFs containing putative pyrrolysine residues were found to exhibit significant read-through similarity to homologous methyltransferases (Table 2). Out of 2,086 and 3,611 theoretical ORFs (see Materials and methods section) analyzed in the complete genomes of M. mazei and M. acetivorans, 87 and 97, respectively, showed significant read-through similarity. We have listed all those ORFs with an in-frame UAG codon that exhibit high RSA similarity, as well as well-aligned MSA for M. acetivorans and M. mazei (Additional data file 2). Apart from previously described transposases [11], the list contains several other candidate proteins, including a novel homolog of the cobalamin biosynthesis protein CobN (Figure 4c).
Overall distribution of the recoded proteins
A marked tendency was noted for the selenoproteins to occur in certain pathways and functional categories (Figure 2b). The majority of detected selenoproteins in bacteria were FDHs that convert formate to carbon dioxide in anaerobic environments [27]. Other known selenoproteins include SelD, GrdA and GrdB (from the anaerobic glycine reduction pathway), HesB (associated with the nitrogen fixation genes), and several oxidoreductases (for example, thioredoxin and peroxiredoxin). In archaea, selenocysteine usage appears to be confined to a small group of enzymes in the anaerobic methanogenesis pathway [23] (such as FDH and formylmethanofuran dehydrogenase from the FDH family, and heterodisulfide reductase) that have conceivably co-evolved under similar evolutionary constraints in a number of methanogens. Pyrrolysine-encoding is found in methyltransferases [26] from a pathway that converts methylamines to methane in Methanosarcina sp. and in the Antarctic archaeon M. burtonii. A high incidence of unusual stop codon reassignments, both selenocysteines and pyrrolysines, in methanogenesis enzymes in ancient archaea is intriguing.
Relative merit of the RSA-based approach
The selenoprotein identification scheme presented herein (an 'RSA-first, SECIS-later' approach) differs from the previously reported methods in several ways. Earlier studies (based on the SECIS search approach) provided an estimated rate of false SECIS hits to be 3 to 15 per 10 Mb [12,17] in eukaryotes, greatly surpassing the number of true selenoproteins. An improved result has been obtained by using a statistical profile computed from a training dataset of aligned known SECIS elements in metazoa [17]. A recent bacterial SECIS-search method analyzed 48,472 SECIS hits in a set of 29 organisms (representing 1.5% of all the UGA codons analyzed), out of which 28,974 (approximately 60%) were selected for further analysis of protein sequence conservation in the UGA flanking regions [16]. Still, difficulties remain for approaches that rely on detecting small RNA signal sequences as an early step in analysis, especially in situations such as new genomes, where the nature of the signal may not be understood in advance. Examining presumptive protein sequences as a prior step mitigates these difficulties. In the present study, although RSA was applied to fairly small segments of the protein sequences following the UGA codon, it was quite efficient in identifying candidates representing read-through events (Figure 1c). This ability of RSA to limit the predicted set to a relatively small, manageable number of likely candidates (3,594 out of 203,339, approximately 1.7%) facilitated further detailed calculations in genome-wide analyses. Of the small set of 109 ORFs selected by the subsequent MSA analysis, 92 (approximately 84%) were selected afterward as putative selenoproteins. Thus, an analysis of protein sequences is able to filter out most of the false-positives, without using any mRNA context information. Our combined 'RSA-first, SECIS-later' method is, therefore, applicable to cases (for example, P. profundum) where a divergent signal makes a SECIS-based search unsuitable [16]. In the present approach, it becomes possible to scrutinize putative non-canonical SECIS signals. In addition, our method provides a useful way to search for selenoproteins lacking homologs containing corresponding cysteine residues [13] (Figure 2d).
The RSA approach was likewise successful in predicting putative pyrrolysine-proteins in archaea. Out of the 9,515 theoretical ORFs analyzed for putative pyrrolysine residues in four methanogens, 321 ORFs (3.4%) displayed significant read-through similarity. Unlike the case for selenoproteins, a reliable benchmarking of pyrrolysine-protein predictions against a known dataset was not possible. The predicted result encompasses the previously reported methylamine methyltransferases [26], however, and includes a number of likely candidates for further experiments. Intriguingly, the putative pyrrolysine residues do not align so exclusively with a particular, conserved amino acid in homologous proteins (Figure 4a-c) [11]. The RSA method appears, therefore, to be generally useful as an initial predictor for pyrrolysine proteins. In addition, the RSA approach offers wider utility for identifying cases of genome-wide stop codon redefinition (for example, in Mycoplasma spp.; see Materials and methods section) or special instances of stop codon read-through (for example, UAG read-through in a pilus biosynthesis gene in E. coli [28] (data not shown)).
Conclusion
To summarize, we have developed a novel computational scheme for predicting selenocysteine and pyrrolysine residues in proteins and have applied the method to microbes with complete genomes. In addition to confirming well-known examples, our method predicts new prospective candidates for further experimental validation. A worldwide web site has been developed for the interested user community [29]. The method should be a useful tool for predicting rare amino acids, as well as other read-through events, and for correcting gene annotations in the growing genomic databases.
Materials and methods
All the complete genomes were obtained from the National Center for Biotechnology Information (NCBI) [30]. Unfinished M. barkeri and M. burtonii genomes were obtained from The Institute for Genomic Research [31]. A list of accession numbers is provided in the Additional data files. A Perl script was written to perform all the computations (available upon request). All computations were performed in a local cluster of Linux computers. tRNA genes were computationally identified using the tRNASCAN-SE program [32]. Genes encoding SelA and SelB were detected directly from annotated genomes from NCBI.
All theoretical ORFs (≥ 90 residues) that begin with a start codon (ATG, TTG or GTG) and end with a stop codon (TAA, TAG or TGA) and contain one in-frame TGA (for selenocysteine) or TAG (for pyrrolysine) codon were extracted from the genomic data for analysis. In order to detect two short SelW proteins in G. sulfurreducens and C. jejuni, a reduced (80 residue) length constraint was used.
Read-through similarity analysis (RSA)
For each of the predicted ORFs, the BLAST program [33,34] was used to search for homologous proteins in a customized sequence database. The BLAST search space was restricted to a window of a maximum 100 residue length, pivoting at the stop codon. The BLOSUM62 matrix was used throughout and the selenocysteine residue was treated as 'any amino acid' (X). The BLAST database contained a maximum of 650,870 protein sequences from all the annotated complete microbial genomes from NCBI (dated 4 December 2005). A self-excluding BLAST database was used for the homology search in each organism. Top hits (E-value ≤ 10-1) that encompassed either side of the stop codon were identified. For each of the selected, truncated ORF sequences ({x1, x2,..., xi,..., xu,..., xn} where n = min{u + 60, u + t}; u = position of the stop codon; t = position of the subsequent stop codon) and the corresponding top hit sequence from the BLAST search ({y1, y2..., yj,..., ym}), a (n + 1) by (m + 1) dynamic alignment matrix was calculated with an affine gap penalty function [18,19]. N-terminal overhangs for both the sequences were not penalized; the 0th row and the 0th column were initialized with zero values.
For each cell (i,j) in the matrix:
a(i,j) = s(i,j) + max { a(I - 1,j - 1),
b(i-1,j-1),
c(i-1,j-1) }
b(i,j) = max { -(h + g) + a(i,j - 1),
-g + b(i,j - 1),
-(h + g) + c(i,j - 1) }
c(i,j) = max { -(h + g) + a(I - 1,j),
-(h + g) + b(I - 1,j),
-g + c(I - 1,j) }
score (i,j) = max {a(i,j), b(i,j), c(i,j)}; s(i,j) → BLOSUM62 matrix; h = 12, g = 2
Best_scoreORF = max{score(n,j), j = 1,..,m} (1)
Because we were exclusively interested in the significance of the alignment at the carboxy-terminal extension region beyond the stop position, the highest score from the nth column (that is the alignment of the terminal residue xn of the truncated ORF with the {y1,..., ym} residues) was taken as the maximal score (Best_scoreORF) instead of the usual Smith-Waterman score. A Z-value was computed by shuffling the terminal extension region 100 times, re-computing the scores in the terminal block of the matrix ({xstop,..,xn} and {y1,..,ym}) and averaging the maximal score (<Best_scorerand>). A test calculation with 10,000 times shuffling for one genome produced similar results. The values from randomized sequences were used to calculate a Z-value:
ZORF = (Best_scoreORF - <Best_scorerand>)/standard_deviation (2)
A generally weak dependence on length and amino acid composition makes the Z-values (which follow an extreme value distribution) useful for evaluating the significance of alignment scores [35]. We have used a fairly conservative Z-value cutoff (Zc = 8.0) [35] to decide the statistical significance of a C-terminal alignment. Selection criteria had to be relaxed for two legitimate selenoproteins, a sulfur transferase in G. sulfurreducens (Z-value 7.9) and a coenzyme F420-reducing hydrogenase subunit in M. maripaludis (Z-value 4.6).
Multiple sequence alignment
For each of the selected candidate ORFs (Zc ≥ 8), a sensitive, iterative PSI-BLAST search was performed using position-specific scoring matrices. The top 10 hits (E ≤ 10-3) were used to construct a MSA with ClustalW [36]. Amino acids lining up with the putative selenocysteine residue were examined. Selenocysteines that aligned with two or more cysteine residues were selected for further analysis.
SECIS element analysis
In accordance with a recent analysis of bacterial SECIS elements [16], a 111 nucleotide long mRNA stretch surrounding the UGA codon position (-10 to +100) was extracted from each of the selected ORFs passing the previous tests in each bacterial organism. The extracted set of RNA sequences for each organism was used to detect a common, single hairpin motif using the rapid, heuristic-based RNAPROFILE program [20]. A test calculation predicted the known SECIS element of the gene encoding FDH from E. coli correctly [37] (Figure 3b). The putative SECIS hairpin motifs were manually inspected for consistency.
Control analysis
To evaluate the performance of the RSA step, we analyzed the Mycoplasma genitalium organism that utilizes UGA to code for tryptophan throughout its genome. M. genitalium is a small genome with 470 genes [38], the majority of which have a homolog in our database, thus minimizing database effects in our calculation. We applied the RSA method to all the theoretical ORFs with one in-frame TAA (313) or TAG (137) or TGA (780) codon. A self-excluding BLAST database of microbial proteins was used. In M. genitalium, over 78% of the TGA cases (91% when a self-included database was used) were identified by the RSA method as recoding events with a Z-value of 8 or higher. These cases aligned overwhelmingly with tryptophan residues in homologs. In contrast, only about 2% to 3% of the ORFs contatining a TAA or TAG stop codon passed the same RSA test.
We also applied the selenoprotein detection scheme to the Aeropyrum pernix (BA000002) genome, which does not contain any selenocysteine insertion genes. Out of 26 of 1,288 ORFs with in-frame 'UGA' that were selected by RSA (approximately 2%), none were selected in the subsequent MSA test.
A web-server for RSA analysis
A web-based service is available for RSA analysis of submitted DNA sequences [29]. The server was designed to analyze an ORF with one in-frame stop codon (UAA, UAG or UGA). A larger, non-redundant BLAST database (to be updated regularly) is used by the web server. The Z-value score and the MSA for the ORF are returned to the user.
Sensitivity and specificity
Sensitivity = true positive/(true positive + false negative)
Specificity = true positive/(true positive + false positive)
Estimates of true positives, false negatives and false positives were based on predictions performed on the set of organisms whose selenoproteins have been described in the RECODE [21] database (Figure 2a). The number of true positives is taken to be the number of predictions that are already known selenoproteins in the RECODE database. False negatives are those known selenoproteins not predicted by our method. False positives are difficult to estimate. As an extreme estimate, we have taken as an upper bound all those predictions that are not in the known database. The actual false positive rate is probably considerably lower than this estimate.
Additional data files
The following additional data are available with the online version of this paper. Additional data file 1 is a list of all the genomes analyzed together with the NCBI accession number. Additional data file 2 contains all the predicted recoded proteins from the complete genomes analyzed in this study in FASTA format.
Supplementary Material
Additional File 1
A list of all the genomes analyzed together with the NCBI accession number
Click here for file
Additional File 2
All the predicted recoded proteins from the complete genomes analyzed in this study in FASTA format
Click here for file
Acknowledgements
This work was supported by the DOE office of Biological and Environmental Research. The authors thank T Holton for assistance with the web-based server preparation.
Figures and Tables
Figure 1 Schematic representation of the selenocysteine insertion machinery and the selenoprotein detection scheme. (a) A cartoon diagram of selenocysteine incorporation during protein translation inside the cell. The selenocysteine-specific elongation factor (SelB; pink) is shown interacting with the selenocysteine insertion sequence (SECIS) hairpin element in the mRNA and tRNA-sec (SelC). The anticodon of SelC tRNA interacts with and recognizes the 'UGA' codon. The ribosome and other components of the translational machinery are omitted for clarity. (b) Schematic representation of the 'read-through similarity analysis' approach. The top BLAST hit is shown in blue. The window lengths used for the BLAST search and read-through similarity evaluation are marked in the drawing. (c) A flow chart describing how the different components of the predictive scheme are combined for selenoprotein prediction. ORF, open reading frame.
Figure 2 An overview of the predicted selenoproteome. (a) A Venn diagram representation of the overlap between the known selenoproteins in the RECODE database (bold line) and the results of our prediction method (plain line) over the same set of organisms as included in RECODE. (b) A pie chart illustrating the types of selenoproteins in our predicted dataset. The dataset was divided into the following groups: formate dehydrogenase (FDH) family enzymes; archaeal methanogenesis selenoproteins (excluding the FDH family); selenophosphate synthetase (SelD); other known selenoproteins (for example, thioredoxin, hesB); glycine reductase genes (GRD); and new candidate selenoproteins. (c) A section of the multiple sequence alignments (MSA) of the newly predicted candidate selenoprotein from P. profundum with its four homologs found in our database. Note the alignment of putative selenocysteine (U denotes selenocysteine) with cysteine residues in the MSA. (d) The MSA of a selenoprotein formylmethanofuran dehydrogenase from M. maripaludis in which the recoded selenocysteine aligns with a set of conserved aspartate residues rather than the cysteine residues. The MSA illustrations were prepared using ALSCRIPT [39].
Figure 3 Representatives of the putative selenocysteine insertion sequence (SECIS) hairpin elements in various genomes as identified by the present study. (a) The SECIS elements from the genes coding for the following proteins from P. profundum: 1, glycine reductase GrdA; 2, glycine reductase GrdB2; 3, glycine reductase GrdA; 4, selenophosphate synthetase (SelD); 5, a hypothetical protein. (b) The SECIS elements from the genes coding for the following proteins from E. coli: 1, formate dehydrogenase; 2, formate dehydrogenase-N; 3, formate dehydrogenase-O.
Figure 4 Sections of the multiple sequence alignments of the putative pyrrolysine-containing proteins. (a) A protein known to use UAG read-through, methylamine methyltransferase from M. acetivorans. (b) A putative methyltransferase from M. burtonii. (c) A predicted read-through ORF homologous to a cobalamin biosynthesis protein CobN (gi|20906100|gb|AAM31298.1|, Methanosarcina mazei Goe1) from M. acetivorans. Note the alignment of presumed pyrrolysine residues (denoted as X) with various amino acids.
Table 1 A list of predicted selenoproteins encoded by UGA read-through
Accession ID Organism Computationally identified selenoproteins* annotated by their homologs
AE000657 Aquifex aeolicus 1. gi|12515210|gb|AAG56295.1|AE005358_3 formate dehydrogenase-N, nitrate-inducible, alpha subunit [Escherichia coli]
2. gi|51589698|emb|CAH21328.1| selenide, water dikinase [Yersinia pseudotuberculosis IP 32953]
AE017125 Helicobacter hepaticus 1.gi|27362035|gb|AAO10941.1|AE016805_198 formate dehydrogenase, alpha subunit [Vibrio vulnificus CMCP6]
2. gi|46914191|emb|CAG20971.1| putative selenophosphate synthase [Photobacterium profundum]
AE017143 Haemophilus ducreyi 35000HP 1. gi|26108424|gb|AAN80626.1|AE016761_201 selenide, water dikinase [Escherichia coli CFT073]
AE004439 Pasteurella multocida 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
2. gi|5103639|dbj|BAA79160.1| 194 amino acid long hypothetical protein [Aeropyrum pernix K1]
AE005674 Shigella flexneri 2a 1. gi|12515215|gb|AAG56300.1|AE005358_8 orf; unknown function [Escherichia coli O157:H7 EDL933]
2. gi|1788928|gb|AAC75627.1| quinolinate synthetase, B protein; quinolinate synthetase, B protein, catalytic and NAD/flavoprotein subunit [Escherichia coli >K12]
3. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
4. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
5. gi|3868721|gb|AAD13462.1| selenopolypeptide subunit of formate dehydrogenase H; formate dehydrogenase H, selenopolypeptide subunit [Escherichia coli K12]
AE014073 Shigella flexneri 2a 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
2. gi|1788928|gb|AAC75627.1| quinolinate synthetase, B protein; quinolinate synthetase, B protein, catalytic and NAD/flavoprotein subunit [Escherichia coli K12]
3. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
4. gi|3868721|gb|AAD13462.1| selenopolypeptide subunit of formate dehydrogenase H; formate dehydrogenase H, selenopolypeptide subunit [Escherichia coli K12]
AE006469 Sinorhizobium meliloti 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
AE008691 Thermoanaerobacter tengcongensis 1. gi|41816370|gb|AAS11237.1| glycine reductase complex selenoprotein GrdA [Treponema denticola ATCC 35405]
2. gi|51857693|dbj|BAD41851.1| glycine reductase complex selenoprotein B [Symbiobacterium thermophilum IAM 14863]
3. gi|46914191|emb|CAG20971.1| putative selenophosphate synthase [Photobacterium profundum]
AE014075 Escherichia coli CFT073 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
2. gi|56130341|gb|AAV79847.1| formate dehydrogenase H [Salmonella enterica subsp. enterica serovar Paratyphi A str. ATCC 9150]
3. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
BA000007 Escherichia coli O157H7 1. gi|56130341|gb|AAV79847.1| formate dehydrogenase H [Salmonella enterica subsp. enterica serovar Paratyphi A str. ATCC 9150]
2. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
3. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
U00096 Escherichia coli K12 1. gi|5105267|dbj|BAA80580.1| 114 amino acid long hypothetical protein [Aeropyrum pernix K1]
2. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
3. gi|56130341|gb|AAV79847.1| formate dehydrogenase H [Salmonella enterica subsp. enterica serovar Paratyphi A str. ATCC 9150]
4. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
AE014299 Shewanella oneidensis 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
AE015451 Pseudomonas putida KT2440 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
AE004091 Pseudomonas aeruginosa 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
AE016958 Mycobacterium avium paratuberculosis 1. gi|13880045|gb|AAK44759.1| hypothetical protein MT0536 [Mycobacterium tuberculosis CDC1551]
2. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
AE017042 Yersinia pestis biovar Mediaevalis 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
AE009952 Yersinia pestis KIM 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
AL590842 Yersinia pestis CO92 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
AE017180 Geobacter sulfurreducens 1. gi|19918170|gb|AAM07420.1| 4-carboxymuconolactone decarboxylase [Methanosarcina acetivorans str. C2A]
2. gi|21956737|gb|AAM83670.1|AE013608_5 glutaredoxin 3 [Yersinia pestis KIM]
3. gi|37201109|dbj|BAC96933.1| thiol-disulfide isomerase and thioredoxins [Vibrio vulnificus YJ016]
4. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
5. gi|34105000|gb|AAQ61356.1| conserved hypothetical protein [Chromobacterium violaceum ATCC 12472]; gi|53758707|gb|AAU92998.1| HesB/YadR/YfhF family protein [Methylococcus capsulatus str. Bath];
6. gi|46914191|emb|CAG20971.1| Putative selenophosphate synthase [Photobacterium profundum]
7. gi|32448022|emb|CAD77542.1| peroxiredoxin [Pirellula sp.]
8. gi|29605647|dbj|BAC69712.1 hypothetical protein [Streptomyces avermitilis MA-4680] (SelW)
9. gi|34482757|emb|CAE09757.1| sulfur transferase precursor [Wolinella succinogenes]
AE017226 Treponema denticola ATCC 35405 1. gi|51857694|dbj|BAD41852.1| glycine reductase complex selenoprotein A [Symbiobacterium thermophilum IAM 14863]
2. gi|51857693|dbj|BAD41851.1| glycine reductase complex selenoprotein B [Symbiobacterium thermophilum IAM 14863]
3. gi|56380162|dbj|BAD76070.1| glutathione peroxidase [Geobacillus kaustophilus HTA426]
4. gi|51857693|dbj|BAD41851.1| glycine reductase complex selenoprotein B [Symbiobacterium thermophilum IAM 14863]
5. gi|26108424|gb|AAN80626.1|AE016761_201 selenide, water dikinase [Escherichia coli CFT073]
6. gi|52209545|emb|CAH35498.1| thioredoxin 1 [Burkholderia pseudomallei K96243]
AL111168 Campylobacter jejuni 1. gi|27362035|gb|AAO10941.1|AE016805_198 formate dehydrogenase, alpha subunit [Vibrio vulnificus CMCP6]
2. gi|54018125|dbj|BAD59495.1| hypothetical protein [Nocardia farcinica IFM 10152]; (SelW)
AL513382 Salmonella typhi 1. gi|3868721|gb|AAD13462.1| selenopolypeptide subunit of formate dehydrogenase H; formate dehydrogenase H, selenopolypeptide subunit [Escherichia coli K12]
2. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
AE006468 Salmonella typhimurium LT2 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
2. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
3. gi|3868721|gb|AAD13462.1| selenopolypeptide subunit of formate dehydrogenase H; formate dehydrogenase H, selenopolypeptide subunit [Escherichia coli K12]
BA000016 Clostridium perfringens 1. gi|28202985|gb|AAO35429.1| conserved protein [Clostridium tetani E88]; gi|20906561|gb|AAM31712.1| HesB protein [Methanosarcina mazei Goe1]
2. gi|46914191|emb|CAG20971.1| putative selenophosphate synthase [Photobacterium profundum]
BX470251 Photorhabdus luminescens 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase alpha subunit [Aquifex aeolicus VF5]
BX571656 Wolinella succinogenes 1. gi|27362035|gb|AAO10941.1|AE016805_198 formate dehydrogenase, alpha subunit [Vibrio vulnificus CMCP6]
L42023 Haemophilus influenzae 1. gi|2983532|gb|AAC07107.1| formate dehydrogenase, alpha subunit [Aquifex aeolicus VF5]
2. gi|26108424|gb|AAN80626.1|AE016761_201 selenide, water dikinase [Escherichia coli CFT073]
CR354531 Photobacterium profundum 1. gi|58428447|gb|AAW77484.1| conserved hypothetical protein [Xanthomonas oryzae pv. oryzae KACC10331]
CR354532 Photobacterium profundum 1. gi|41816370|gb|AAS11237.1| glycine reductase complex selenoprotein GrdA [Treponema denticola ATCC 35405]
2. gi|51589698|emb|CAH21328.1| selenide, water dikinase [Yersinia pseudotuberculosis IP 32953]
3. gi|41816370|gb|AAS11237.1| glycine reductase complex selenoprotein GrdA [Treponema denticola ATCC 35405]
4. gi|41818450|gb|AAS12639.1| glycine reductase complex selenoprotein GrdB2 [Treponema denticola ATCC 35405]
AE009439 Methanopyrus kandleri (archaea) 1. gi|2622673|gb|AAB86026.1| formate dehydrogenase, alpha subunit homolog [Methanothermobacter thermautotrophicus]; gi|2622681|gb|AAB86033.1| tungsten formylmethanofuran dehydrogenase, subunit B [Methanothermobacter thermautotrophicus]
2. gi|57160335|dbj|BAD86265.1| probable formate dehydrogenase, alpha subunit [Thermococcus kodakaraensis KOD1]
3. gi|33566318|emb|CAE37231.1| putative iron-sulfur binding protein [Bordetella parapertussis]
4. gi|44921146|emb|CAF30381.1| heterodisulfide reductase, subunit A [Methanococcus maripaludis]
5. gi|44921142|emb|CAF30377.1| coenzyme F420-non-reducing hydrogenase, subunit delta [Methanococcus maripaludis]; gi|2622243|gb|AAB85627.1| methyl viologen-reducing hydrogenase, delta subunit homolog FlpD [Methanothermobacter thermautotrophicus]; gi|20904385|gb|AAM29752.1| heterodisulfate reductase, subunit A [Methanosarcina mazei Goe1]
6. gi|45047811|emb|CAF30938.1| coenzyme F420-reducing hydrogenase subunit alpha [Methanococcus maripaludis]
7. gi|39576202|emb|CAE80367.1| selenide, water dikinase [Bdellovibrio bacteriovorus HD100]
L77117 Methanococcus jannaschii (archaea) 1. gi|44921146|emb|CAF30381.1| heterodisulfide reductase subunit A [Methanococcus maripaludis]
2. gi|45047811|emb|CAF30938.1| coenzyme F420-reducing hydrogenase subunit alpha [Methanococcus maripaludis]
3. gi|50875900|emb|CAG35740.2| methyl-viologen-reducing hydrogenase, delta subunit [Desulfotalea psychrophila LSv54]
4. gi|2622240|gb|AAB85625.1| methyl viologen-reducing hydrogenase, delta subunit [Methanothermobacter thermautotrophicus]; gi|44921142|emb|CAF30377.1| coenzyme F420-non-reducing hydrogenase subunit delta [Methanococcus maripaludis]
5. gi|2622673|gb|AAB86026.1| formate dehydrogenase, alpha subunit homolog [Methanothermobacter thermautotrophicus]; gi|45048129|emb|CAF31247.1| tungsten containing formylmethanofuran dehydrogenase, subunit B [Methanococcus maripaludis] (overlaps with #4)
6. gi|26108424|gb|AAN80626.1|AE016761_201 selenide, water dikinase [Escherichia coli CFT073]
7. gi|53758707|gb|AAU92998.1| HesB/YadR/YfhF family protein [Methylococcus capsulatus str. Bath]
8. gi|45047727|emb|CAF30854.1| formate dehydrogenase, alpha subunit [Methanococcus maripaludis]
BX950229 Methanococcus maripaludis (archaea) 1. gi|2622673|gb|AAB86026.1| formate dehydrogenase, alpha subunit homolog [Methanothermobacter thermautotrophicus]; gi|19886584|gb|AAM01476.1| Formylmethanofuran dehydrogenase subunit B [Methanopyrus kandleri AV19]
2. gi|2622673|gb|AAB86026.1| formate dehydrogenase, alpha subunit homolog [Methanothermobacter thermautotrophicus]
3. gi|2622240|gb|AAB85625.1| methyl viologen-reducing hydrogenase, delta subunit [Methanothermobacter thermautotrophicus]; gi|39981962|gb|AAR33424.1| heterodisulfide reductase subunit [Geobacter sulfurreducens PCA]
4. gi|2622673|gb|AAB86026.1| formate dehydrogenase, alpha subunit homolog [Methanothermobacter thermautotrophicus]
5. gi|2622673|gb|AAB86026.1| formate dehydrogenase, alpha subunit homolog [Methanothermobacter thermautotrophicus]; gi|19918286|gb|AAM07526.1| formylmethanofuran dehydrogenase, subunit B [Methanosarcina acetivorans str. C2A]
6. gi|19886593|gb|AAM01482.1| Heterodisulfide reductase, subunit A, polyferredoxin [Methanopyrus kandleri AV19]
Organism names, National Center for Biotechnology Information accession numbers for the genomes and the top PSI-BLAST hit(s) from our database are shown. Seven novel candidate selenoproteins are shown in bold type. *Each entry corresponds to a computationally identified read-through protein in the organism indicated to the left. FASTA files for these recoded protein sequences are provided in the Additional file 2. For each recoded protein, the GI number and the functional annotation for a homologous protein are given.
Table 2 Methyltransferases predicted to encode pyrrolysine by UAG read-through in a set of methanogenic archaea
Organism Computationally identified pyrrolysine-proteins* annotated by their homologs
Methanosarcina acetivorans (AE010299) 1. gi|56678713|gb|AAV95379.1| trimethylamine methyltransferase family protein [Silicibacter pomeroyi DSS-3]
2. gi|14247242|dbj|BAB57633.1| menaquinone biosynthesis methyltransferase [Staphylococcus aureus subsp. Aureus Mu50]
3. gi|36785418|emb|CAE14364.1| protein methyltranferase [Photorhabdus luminescens subsp. laumondii TTO1]
4. gi|56679325|gb|AAV95991.1| trimethylamine methyltransferase family protein [Silicibacter pomeroyi DSS-3]
5. i|20904823|gb|AAM30145.1| SAM-dependent methyltransferases [Methanosarcina mazei Goe1]
6. gi|56312282|emb|CAI06927.1| predicted methyltransferase [Azoarcus sp. EbN1]
7. gi|45047608|emb|CAF30735.1| generic methyltransferase [Methanococcus maripaludis]
8. gi|20905508|gb|AAM30766.1| methylcobalamin: Coenzyme M methyltransferase [Methanosarcina mazei Goe1]
9. Predicted ORF monomethylamine methyltransferase [Methanosarcina mazei Goe1]†
10. Predicted ORF monomethylamine methyltransferase [Methanosarcina mazei Goe1]†
11. Predicted ORF dimethylamine methyltransferase [Methanosarcina mazei Goe1]†
12. Predicted ORF dimethylamine methyltransferase [Methanosarcina mazei Goe1]†
13. Predicted ORF dimethylamine methyltransferase [Methanosarcina mazei Goe1]†
Methanosarcina mazei (AE008384) 1. gi|19914316|gb|AAM03972.1| trimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
2. gi|19914320|gb|AAM03976.1| dimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
3. gi|19914753|gb|AAM04365.1| trimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
4. gi|19913899|gb|AAM03597.1| monomethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
5. gi|19914755|gb|AAM04366.1| dimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
6. gi|19914320|gb|AAM03976.1| dimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
7. gi|19913899|gb|AAM03597.1| monomethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
Methanosarcina barkeri (draft genome) 1. gi|19914320|gb|AAM03976.1| dimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
2. gi|19913899|gb|AAM03597.1| monomethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
3. gi|19914316|gb|AAM03972.1| trimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
4. gi|19914320|gb|AAM03976.1| dimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
5. gi|19914334|gb|AAM03988.1| protein-L-isoaspartate (D-aspartate) O-methyltransferase [Methanosarcina acetivorans str. C2A]
6. gi|19913899|gb|AAM03597.1| monomethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
7. gi|19913899|gb|AAM03597.1| monomethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
Methanococcoides burtonii (draft genome) 1. gi|19914320|gb|AAM03976.1| dimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
2. gi|19914753|gb|AAM04365.1| trimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
3. gi|5458504|emb|CAB49992.1| methlytransferase, putative [Pyrococcus abyssi]
4. gi|5458504|emb|CAB49992.1| methlytransferase, putative [Pyrococcus abyssi] (overlaps with #3)
5. gi|19914320|gb|AAM03976.1| dimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
6. gi|19914753|gb|AAM04365.1| trimethylamine methyltransferase [Methanosarcina acetivorans str. C2A]
7. gi|19913899|gb|AAM03597.1| monomethylamine methyltransferase [Methanosarcina acetivorans str. C2A
*Each entry corresponds to a computationally identified read-through protein in the organism indicated to the left. FASTA files for these recoded protein sequences are provided in the Additional data files. For each recoded protein, the GI number and the functional annotation for a homologous protein are given. †These open reading frames (ORFs) in M. acitovorans were predicted during a repeat search using a BLAST database containing putative methylamine methyltransferase ORFs in M. mazei as identified by our method. Although the M. acitovorans genome was annotated for several pyrrolysine-containing methylamine methyltranferases, this was not the case with the M. mazei genome. Thus, several methyltransferases that are specific to these methanosarcina species could not be detected in our original calculation due to the lack of read-through homologs. Such repeat searches were not performed for the two unfinished genomes.
==== Refs
Gesteland RF Atkins JF Recoding: dynamic reprogramming of translation. Annu Rev Biochem 1996 65 741 768 8811194 10.1146/annurev.bi.65.070196.003521
Namy O Rousset JP Napthine S Brierley I Reprogrammed genetic decoding in cellular gene expression. Mol Cell 2004 13 157 168 14759362 10.1016/S1097-2765(04)00031-0
Bock A Biosynthesis of selenoproteins: an overview. Biofactors 2000 11 77 78 10705967
Stadtman TC Selenocysteine. Annu Rev Biochem 1996 65 83 100 8811175 10.1146/annurev.bi.65.070196.000503
Hatfield DL Gladyshev VN How selenium has altered our understanding of the genetic code. Mol Cell Biol 2002 22 3565 3576 11997494 10.1128/MCB.22.11.3565-3576.2002
Srinivasan G James CM Krzycki JA Pyrrolysine encoded by UAG in Archaea: charging of a UAG-decoding specialized tRNA. Science 2002 296 1459 1462 12029131 10.1126/science.1069588
Hao B Gong W Ferguson TK James CM Krzycki JA Chan MK A new UAG-encoded residue in the structure of a methanogen methyltransferase. Science 2002 296 1462 1466 12029132 10.1126/science.1069556
Rayman MP The importance of selenium to human health. Lancet 2000 356 233 241 10963212 10.1016/S0140-6736(00)02490-9
Frankenberger WT JrArshad M Bioremediation of selenium-contaminated sediments and water. Biofactors 2001 14 241 254 11568461
Jacob C Giles GI Giles NM Sies H Sulfur and selenium: the role of oxidation state in protein structure and function. Angew Chem Int Ed Engl 2003 42 4742 4758 14562341 10.1002/anie.200300573
Zhang Y Baranov PV Atkins JF Gladyshev VN Pyrrolysine and selenocysteine use dissimilar decoding strategies. J Biol Chem 2005 280 20740 20751 15788401 10.1074/jbc.M501458200
Kryukov GV Castellano S Novoselov SV Lobanov AV Zehtab O Guigo R Gladyshev VN Characterization of mammalian selenoproteomes. Science 2003 300 1439 1443 12775843 10.1126/science.1083516
Kryukov GV Gladyshev VN The prokaryotic selenoproteome. EMBO Rep 2004 5 538 543 15105824 10.1038/sj.embor.7400126
Castellano S Novoselov SV Kryukov GV Lescure A Blanco E Krol A Gladyshev VN Guigo R Reconsidering the evolution of eukaryotic selenoproteins: a novel nonmammalian family with scattered phylogenetic distribution. EMBO Rep 2004 5 71 77 14710190 10.1038/sj.embor.7400036
Zhang Y Fomenko DE Gladyshev VN The microbial selenoproteome of the Sargasso Sea. Genome Biol 2005 6 R37 15833124 10.1186/gb-2005-6-4-r37
Zhang Y Gladyshev VN An algorithm for identification of bacterial selenocysteine insertion sequence elements and selenoprotein genes. Bioinformatics 2005 21 2580 2589 15797911 10.1093/bioinformatics/bti400
Lambert A Lescure A Gautheret D A survey of metazoan selenocysteine insertion sequences. Biochimie 2002 84 953 959 12458087 10.1016/S0300-9084(02)01441-4
Gotoh O An improved algorithm for matching biological sequences. J Mol Biol 1982 162 705 708 7166760 10.1016/0022-2836(82)90398-9
Setubal C Meidanis J Introduction to Computational Molecular Biology 1997 Boston: PWS Publishing Company
Pavesi G Mauri G Stefani M Pesole G RNAProfile: an algorithm for finding conserved secondary structure motifs in unaligned RNA sequences. Nucleic Acids Res 2004 32 3258 3269 15199174 10.1093/nar/gkh650
Baranov PV Gurvich OL Hammer AW Gesteland RF Atkins JF RECODE 2003. Nucleic Acids Res 2003 31 87 89 12519954 10.1093/nar/gkg024
Murzin AG Brenner SE Hubbard T Chothia C SCOP: a structural classification of proteins database for the investigation of sequences and structures. J Mol Biol 1995 247 536 540 7723011 10.1006/jmbi.1995.0159
Cobucci-Ponzano B Rossi M Moracci M Recoding in archaea. Mol Microbiol 2005 55 339 348 15659155 10.1111/j.1365-2958.2004.04400.x
Vezzi A Campanaro S D'Angelo M Simonato F Vitulo N Lauro FM Cestaro A Malacrida G Simionati B Cannata N Life at depth: Photobacterium profundum genome sequence and expression analysis. Science 2005 307 1459 1461 15746425 10.1126/science.1103341
Marchler-Bauer A Anderson JB DeWeese-Scott C Fedorova ND Geer LY He S Hurwitz DI Jackson JD Jacobs AR Lanczycki CJ CDD: a curated Entrez database of conserved domain alignments. Nucleic Acids Res 2003 31 383 387 12520028 10.1093/nar/gkg087
Krzycki JA Function of genetically encoded pyrrolysine in corrinoid-dependent methylamine methyltransferases. Curr Opin Chem Biol 2004 8 484 491 15450490 10.1016/j.cbpa.2004.08.012
Jormakka M Byrne B Iwata S Formate dehydrogenase: a versatile enzyme in changing environments. Curr Opin Struct Biol 2003 13 418 423 12948771 10.1016/S0959-440X(03)00098-8
Jalajakumari MB Thomas CJ Halter R Manning PA Genes for biosynthesis and assembly of CS3 pili of CFA/II enterotoxigenic Escherichia coli: novel regulation of pilus production by bypassing an amber codon. Mol Microbiol 1989 3 1685 1695 2576094
Read-through Similarity Analysis
National Center for Biotechnology Information
The Institute for Genomic Research
Lowe TM Eddy SR tRNAscan-SE: a program for improved detection of transfer RNA genes in genomic sequence. Nucleic Acids Res 1997 25 955 964 9023104 10.1093/nar/25.5.955
Altschul SF Gish W Miller W Myers EW Lipman DJ Basic local alignment search tool. J Mol Biol 1990 215 403 410 2231712 10.1006/jmbi.1990.9999
Altschul SF Madden TL Schaffer AA Zhang J Zhang Z Miller W Lipman DJ Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res 1997 25 3389 3402 9254694 10.1093/nar/25.17.3389
Comet JP Aude JC Glemet E Risler JL Henaut A Slonimski PP Codani JJ Significance of Z-value statistics of Smith-Waterman scores for protein alignments. Comput Chem 1999 23 317 331 10627144 10.1016/S0097-8485(99)00008-X
Thompson JD Higgins DG Gibson TJ CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res 1994 22 4673 4680 7984417
Liu Z Reches M Groisman I Engelberg-Kulka H The nature of the minimal 'selenocysteine insertion sequence' (SECIS) in Escherichia coli. Nucleic Acids Res 1998 26 896 902 9461445 10.1093/nar/26.4.896
Fraser CM Gocayne JD White O Adams MD Clayton RA Fleischmann RD Bult CJ Kerlavage AR Sutton G Kelley JM The minimal gene complement of Mycoplasma genitalium. Science 1995 270 397 403 7569993
Barton GJ ALSCRIPT: a tool to format multiple sequence alignments. Protein Eng 1993 6 37 40 8433969
|
16168087
|
PMC1242215
|
CC BY
|
2021-01-04 16:05:42
|
no
|
Genome Biol. 2005 Aug 16; 6(9):R80
|
latin-1
|
Genome Biol
| 2,005 |
10.1186/gb-2005-6-9-r80
|
oa_comm
|
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-9-r811616808810.1186/gb-2005-6-9-r81SoftwareL2L: a simple tool for discovering the hidden significance in microarray expression data Newman John C [email protected] Alan M [email protected] Department of Biochemistry, University of Washington, Seattle, WA 98115, USA2005 31 8 2005 6 9 R81 R81 5 4 2005 16 6 2005 26 7 2005 Copyright © 2005 Newman and Weiner; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
A database with lists of differentially expressed genes from published microarray studies is presented together with an application for mining the database with the user’s own microarray data, allowing the identification of novel biological patterns in microarray data.
L2L is a database consisting of lists of differentially expressed genes compiled from published mammalian microarray studies, along with an easy-to-use application for mining the database with the user's own microarray data. As illustrated by re-analysis of a recent study of diabetic nephropathy, L2L identifies novel biological patterns in microarray data, providing insights into the underlying nature of biological processes and disease. L2L is available online at the authors' website [].
==== Body
Rationale
In only a few years since their development, high-throughput, whole-genome DNA microarrays have become an invaluable tool throughout biology. The appeal of microarrays seems most irresistible when the biological problem is most intractable; microarrays have become perhaps the most popular contemporary tool for hypothesis generation. Yet interpreting the mountain of data produced by a microarray experiment can be a frustrating chore. The most common outcome of such an experiment is a list of genes, or many such lists: genes that are induced or repressed under one condition or another, at one time point or another, in one cluster or another. The daunting task is to extract some meaning from these lists, either by identifying 'critical genes' which might single-handedly produce a biological effect, or by finding patterns in the list that point to an underlying biological process. The latter universally involves annotating each gene on the list and looking for groups of genes that share a particular characteristic. Until recently, this was done entirely by hand. Each gene was assigned, after a laborious literature search, to an arbitrary functional category like 'DNA repair' or 'metabolism'. A hypothesis might be based on which arbitrary categories appeared most often. Like any non-systematic approach, this one is vulnerable to our very human knack of seeing whatever pattern we wish in a noisy field. The Gene Ontology (GO) consortium [1] has brought systematic order to the field of gene annotation by pre-categorizing genes by biological process, molecular function, and cell component - thus eliminating the pattern-creating risk of post hoc annotation. A number of software tools now exist to automate the process of annotating a list of genes with GO categories. Several of these, including EASE [2], GOMiner [3], Onto-Express [4] and GO::TermFinder [5], also calculate the over-abundance of each category in the list, along with its statistical significance. However, even after functional annotation of the list of genes, uncertainty remains as to whether the results advance understanding of the biology at work in the system, and, if the system is a complex disease, whether the results help explain why the gene expression changes occurred. An alternative approach to interpreting gene expression data is to compare it with other related (or potentially related) gene expression data. The motivation is that microarray experiments exhibiting common changes in gene expression are likely to share one or more underlying molecular mechanisms. Furthermore, in some experiments, the underlying cause of the gene expression changes is well-defined: a specific gene deletion, for example, or treatment with a single receptor ligand. In such cases, the ability to connect the user's experiment with gene expression changes caused by a well-defined perturbation may lead immediately to a hypothesis regarding the underlying mechanism in the system under study.
L2L is a database and associated software tool (Figure 1a) that systematically compares the user's own list of differentially expressed genes with a database of lists of differentially expressed genes that were derived from published microarray data, with the goal of finding common expression patterns that can help generate new hypotheses. The L2L Microarray Database was culled from 111 selected publications, and contains 357 lists of genes that were found to be either upregulated or downregulated under a particular experimental condition. The conditions represented in the database range from normal ageing to space flight, and from interferon treatment to histone deacetylase inhibition (Figure 1b). The L2L Microarray Analysis Tool compares each list in the database with a list of genes supplied by the user, and reports the statistical significance of any overlap between them. It also annotates each gene on the user's list with all the lists in the database on which it is found. The results are presented as a set of hyperlinked HTML documents, which can be conveniently explored by surfing from list to list and gene to gene. L2L is available as an easy-to-use online tool [6], and as a downloadable, command-line application released under the GNU General Public License.
L2L Microarray Database
The need for a standardized format for presenting and storing microarray data from disparate platforms has been recognized for several years. A consortium of researchers [7] has detailed a standardized format for presenting microarray data (MAIME) [8] as well as a markup language in which to encode those now-standardized data (MAGE-ML) [9]. The data can be deposited in any of a number of large public repositories, including CIBEX, ArrayExpress, Oncomine and the NIH's Gene Expression Omnibus (GEO) [10-13]. All of these include web-accessible data-mining tools for browsing experiments and searching for the expression results associated with a particular gene. The sheer volume of deposited data is staggering, and represents a gold mine for bioinformaticians. Yet it all remains remarkably inaccessible to lay biologists. Although we can search GEO, for example, for microarray-identified genes one-by-one, there is no simple way to compare our data en masse with any other data in the repository, much less against all the data in the repository. Furthermore, repositories can make it difficult to extract the original results from the mass of deposited data; an interested user is often required to essentially re-analyze the data, with little knowledge of the original data analysis protocol or, in some cases, without access to all of the relevant data (for instance, GEO submissions do not usually include Affymetrix test-statistic data, a qualitative 'change call' which can be more accurate than the quantitative fold-change for detecting differential expression [14]).
The L2L Microarray Database collects an interesting subset of this public data in its most essential and accessible form - simple, well-annotated lists of genes, using a universal identifier, which were found to be either upregulated or downregulated under a particular condition. It is not intended to be an alternative to the public repositories, but an accessible and utilitarian supplement. The database can be easily applied to the global analysis of any gene expression experiment, producing insights that go well beyond gene-by-gene annotation. The development of L2L was inspired by our efforts to extract meaning from our own microarray analysis of the progeroid Cockayne syndrome (Newman JC, Bailey AD, Weiner AM, unpublished data), so the publications included in the database initially reflected topics thought to be related to this disease - ageing, cancer and DNA damage. Since then, the scope of the publications we included has expanded considerably to include chromatin structure, immune and inflammatory mediators, the hypoxic response, adipogenesis, growth factors, cell cycle regulators, and others. In spite of the parochial origins of the database, the wide range of topics now covered will make L2L of general interest to any investigator using microarrays to study human (and more generally, mammalian) biology. We demonstrate the breadth of L2L's utility below, by re-analyzing a published microarray dataset from a study of diabetic nephropathy - a subject completely unrelated to our original interests. Newman JC, Bailey AD, Weiner AM: manuscript in preparation.
A good list is hard to find
We faced two major challenges in the creation of L2L, one philosophical and one practical. The philosophical problem, which has prevented any significant effort in this direction to date, is that no two microarray experiments are ever perfectly comparable. There is an almost infinite combinatorial complexity of organism, tissue type or cell line, RNA isolation technique, microarray platform, scanning instrument, experimental design, and data analysis technique - even if the question being asked is identical. To make a tool like L2L even possible, it is essential to exclude any incomparable information from each experiment, and convert the remainder to a common language that can be shared by all included experiments. We therefore removed all references to platform-specific probe identifiers, primarily because these would limit L2L to comparing experiments performed on identical platforms, but also because many manuscripts do not report probe IDs. Instead, we converted the probe IDs to the HUGO-approved symbols [15] of the genes they each represent, according the manufacturer's annotations, and ignored those that have no gene association because these cannot be reliably compared across platforms. We also excluded the reported magnitude of expression changes, because fold-changes are often not comparable across platforms [16]. Furthermore, fold-change can be a misleading indicator of the significance of expression changes, especially for platforms like Affymetrix GeneChips that use an independent, and more robust, change call calculation [14]. Finally, ignoring fold-changes vastly simplifies the computational task of comparing hundreds or thousands of lists.
The practical challenge was the extraction of published data and conversion to HUGO gene symbols. This was by far the most time-consuming of the tasks required to create L2L, despite the liberal use of automated tools. The first hurdle was the difficulty of extracting data from published papers in a usable form. Many tables of genes are published as graphical figures rather than textual tables. Supplemental data is often in the form of HTML tables, rather than text files. In both cases, the data are easy to view, but difficult to extract for other uses. More willful is the use of digital-rights management by certain journals to frustrate copying of any information from the electronic (PDF) version of the paper. In all of these situations, laborious manual transcription was required, instead of simple keystrokes to cut-and-paste the data. Repositories like GEO are only a partial solution to this presentation problem; the repositories contain all the raw data, but often lack information about the data analysis used to define a robust change, as well as the actual lists of robustly changed genes.
The second hurdle was actually identifying the genes on published lists. Many publications do not provide an unambiguous reference for each gene - only a common name and/or description. Those that do provide unambiguous references do so in a variety of forms - a HUGO name, LocusLink ID, GenBank accession, or (rarely) commercial probe ID. Online tools exist to interconvert many of these [17,18] and were used whenever possible to convert each list to HUGO names. Ambiguous references were hand-converted by finding the proper match in LocusLink or EntrezGene. Some lists in the L2L Microarray Database are derived from mouse experiments; these were first converted to standard mouse gene names, then mapped to the corresponding HUGO gene name using the HomoloGene database [19] with an ad hoc tool. Any genes without HomoloGene entries were matched by hand in EntrezGene to the proper human homolog. Any gene reference, mouse or human, which could not be unambiguously mapped to a HUGO name was ignored. Duplicates within a list were also ignored. The fraction of the original data that could eventually be mapped to a HUGO name varied with the quality of the gene reference, the proportion of expressed sequence tags (ESTs), and whether mouse-human conversion was required. Most datasets with unambiguous human references have greater than 90% of non-EST, non-duplicate gene references represented in the L2L list of HUGO names. Mouse-human conversion reduced this proportion somewhat (largely due to immunity-related genes), as did descriptive gene references (due to ambiguity). Each list in the database is annotated with a meaningful short name, a longer description, the platform used to generate the list (for example, Affymetrix U95Av2), one or more keywords, and the PubMed ID of the source publication.
More than just microarray data
In addition to the L2L Microarray Database, L2L includes a set of lists for each of the three organizing principles of Gene Ontology - biological process, molecular function and cell component. These lists were compiled from the July 2004 GO association tables, which include associations between UNIPROT names and GO terms. UNIPROT's flat-files associate many human UNIPROT entries with a HUGO alias; an ad hoc tool was used to extract these relationships and convert the UNIPROT GO term assignments to unique HUGO GO term assignments. Another ad hoc tool then created a list for each GO term that contained every HUGO name associated with either that term or any of its descendants. Any lists with fewer than five genes were discarded because comparison to such a small list is unlikely to be informative. In all, there remained 2,169 GO-derived lists with a total of about 240,000 annotations, divided among the three organizing principles. A more detailed description of how the GO lists were compiled, along with downloadable versions of the ad hoc tools, is available on the L2L website [6].
Finally, L2L is not limited to using the four included sets of lists: L2L Microarray Database, GO: Biological Process, GO: Molecular Function, and GO: Cell Component. The modular nature of the tool means that new sets of lists can be created from any source of gene annotations. Some examples include protein-protein interaction databases like DIP, BRITE or BIND [20-22]; pathway annotations from KEGG, BioCarta or GenMAPP [23,24]; experimental gene expression modules [25]; or the associations of gene names with literature keywords that can be compiled using tools like PubGene and TXTGate [26,27]. Any source of gene annotation that can be represented as a set of lists, each specifying a group of genes that share some characteristic, can be easily used with L2L. We hope that the simple and open file formats will encourage others to contribute their own sets of lists to augment L2L or to create similar platform-independent resources.
Although we designed L2L for the lay biologist, we hope that the L2L Microarray Database will prove to be a valuable resource for the bioinformatician as well. For example, many investigators are interested in mapping networks of gene coexpression relationships with the goal of inferring previously unknown functional relationships, or even physical interactions, from shared expression profiles [28-30]. The L2L database is a significant source of primary data for such coexpression analyses. It currently contains 28,026 data points derived from microarray experiments, each of which represents a significant gene expression change. These data points encompass 10,151 gene names - a substantial fraction of the 33,000 HUGO names that had been assigned at the time of writing - and 6,009 of these genes occur at least twice in the database. Among these genes, there are 258,461 unique positive coexpression relationships (a pair of genes found together on different lists) that are found on at least two, and in some cases as many as 16, different lists. There are 20,338 negative coexpression relationships (pairs of genes that are inversely regulated, that is, one appearing on the 'up' and the other on the 'down' list for the same condition) that are found in at least two, and as many as ten, different conditions. We believe the L2L database's catalog of co-expression relationships is one of the largest yet available for human genes, and is based on more robust expression changes and a broader set of experimental conditions than other, albeit more sophisticated, efforts [31].
L2L microarray analysis tool
Compiling the L2L Microarray Database took a large investment of effort that we are eager to share with the community. The open file format of the L2L lists can be easily adapted for use in existing list-comparison tools, like EASE [2] and VennMapper [32]. We saw a need, however, for a similar general-purpose tool that was as straight-forward to use as, for example, PubMed Entrez, and which could be optimized for presenting the unique sort of relationship data contained in the database. Therefore, we created the L2L Microarray Analysis Tool - simple to use for the lay biologist, while powerful and customizable for the technically inclined. Upon entering the L2L website [6], the user follows four steps - step 1: enters a name for the analysis, step 2: uploads a data file, step 3: selects the microarray platform from a menu, and step 4: chooses which set of lists will be used to analyze the data (the database or one of the GO sets) (Figure 2a). After L2L has finished comparing the user's data with all the selected lists, it creates a set of easy-to-navigate HTML pages to visualize the results. These are of three types: the Results Summary page, Listmatch pages and Probematch pages. The Results Summary (Figure 2b) displays all of the lists that have a statistically significant overlap with the user's data, along with all relevant statistics. Each list has a unique Listmatch page (Figure 2c), which displays all the probes in the data that matched that list, along with a variety of annotations for each probe. Similarly, each probe in the data has a Probematch page (Figure 2d), which displays all the lists on which that probe was found. The pages are interconnected by hyperlinks, making it easy to surf, for example, from the Results Summary to a list, to a gene found on that list, to a different list on which that gene is found. Lists and genes are described briefly on each page, but are also hyperlinked to external annotations: for the database lists, this is usually the PubMed abstract of the source publication; for GO categories it is the AmiGO browser page [33] for that category; for genes it is the GeneCards [34] and EntrezGene [35] entries. From the Results Summary page, all of the output files can be downloaded by the user, and viewed later with any web browser.
The analytic engine of L2L is the L2L application, written in Perl (Figure 3). This program receives user input from the web interface and performs the actual data processing tasks, along with the creation of the output HTML pages. The program requires three inputs: the data to be analyzed, in the form of a list of microarray probe identifiers; a translator library that pairs each probe on the microarray with its corresponding HUGO gene name; and a folder of lists with which the data will be compared. As described above, these lists are in the form of HUGO gene names. The program works sequentially through all the lists, first using the translator to map each gene name in the list to all the probes on the microarray that represent that gene (Figure 3a). Each of these translated probe IDs is then queried against the data. Thus, a given gene on a list may be represented by several microarray probes, or none at all. This name-to-probe translation - the reverse of the process by which the database lists were originally generated - allows L2L to retain the greatest possible amount of the user's data, by performing comparisons based on the probe IDs of the user's microarray, rather than the gene names those probes represent. The loss of this probe ID information from the database lists was an unfortunate necessity, since relatively few studies from which the database was compiled even reported probe IDs. The retention of probe IDs from the user's data allows some expression of the subtleties that multiple probes per gene can afford. If only one splice form of a gene is upregulated in the user's data, only that one probe will be scored as a match to a database list the gene is on; all other probes for that gene will be queried and counted as non-matches. The program records the number of probes derived from the list that match the data, the total number of probes on the microarray that represent the gene names on the list, and the fraction of probes on the microarray that are found in the data (Figure 3b). From these three numbers, the program first calculates the number of expected matches for that list, then the relative enrichment of actual matches, and finally a p value for the significance of the overlap. The p value represents the cumulative probability of finding at least as many matches between the data and the list, given the fraction of all microarray probes that are found in the data, as calculated with a cumulative binomial distribution (see below for a more detailed discussion of the statistics of L2L). The results are logged and written to a raw output file. In addition, for each list, the program records the IDs of all the probes from the data that matched that list. Similarly, for each probe in the data, the program records the names of all the lists on which it was found. All of this information is then used to create the output HTML pages (Figure 3c).
The modular design of L2L means that there are a variety of ways to interact with the L2L application, depending on the user's needs. The simplest is through the web interface. In addition to the four-step form described above, there is a 'More Options' page that allows the user to upload a custom translator library for microarray platforms that are not on the menu. Thus, while L2L is intended primarily for use with whole-genome expression microarrays, it can be used with data from any genomic or proteomic analysis. Alternatively, the L2L application itself can be downloaded and run from the command line on any computer with Perl and a UNIX-like command shell. This is ideal for users who want to use a custom set of lists or who need to rapidly process many different data files in a batch mode. L2L includes a basic textual interface that prompts the user for the location of the three necessary inputs: data file, translator library and set of lists. A batch mode bypasses the interface and allows the processing of any number of data files, each from a different microarray platform, against any or all sets of lists with a single command. Users are also free to download the entire L2L website and run it on their own web server.
L2L is remarkably fast because all of the potentially billions of search-for-match operations are implemented as hash-table lookups in Perl. Since relatively few data are stored in memory at any one time, performance is processor-bound on modern machines, and scales linearly only with the combined size of the lists - not with the size of the data file. A comparison of virtually any size data file to all 357 lists in the database, along with the creation of all output files, takes only about 15 seconds on a 1.4 GHz PowerPC. All files associated with L2L, including data, translator library and list, are in a simple tab-delimited, flat-file format. A detailed description of each file type is available on the L2L website [6]; users can create their own files from any text editor.
L2L in the real world: diabetic nephropathy
The ultimate test of a utility like L2L is whether it can produce novel biological insights from real-world microarray data. With this objective in mind, we downloaded several publicly available datasets and analyzed their lists of gene expression changes with L2L (the sample datasets and all results are available at the L2L website [6]). Diabetic nephropathy (DN) is one of the most common, and most devastating, complications of type 2 diabetes mellitus (T2DM) but its molecular etiology remains poorly understood. To generate new hypotheses, Baelde and colleagues examined gene expression patterns in human kidney glomeruli isolated either from normal kidneys or from kidneys afflicted with DN [36]. Several hundred genes were found to be significantly changed in DN, and these were then classified according to GO category using MAPPFinder [37]. The primary hypothesis that ultimately emerged from the experiment, however, relied entirely on an analysis of 'critical genes' - a handful of genes with biological functions that seemed likely to be relevant. Specifically, dysregulation of several tissue repair genes and repression of the growth factor VEGF led the authors to suggest diminished repair capacity in capillary endothelium as a possible etiology for DN. They also suggested, based on MAPPfinder's list of overabundant GO categories, that DN kidneys suffer from reduced nucleotide metabolism and disturbed cytoskeleton formation.
Analysis of the same data with L2L not only quickly confirmed some of the authors' conclusions (Figure 4a), but also detected the fingerprints of the underlying disease process (Figure 4b). Using L2L with Gene Ontology lists, we confirmed the finding of disturbed cytoskeletal formation within moments. We also found that genes repressed in DN are enriched for genes that function in apoptotic pathways involving JAK-STAT, IκK-NFκB and caspases, as well as IGF-binding proteins. Although the latter evidence for a reduced insulin-like growth factor response appears to support the authors' central hypothesis, comparison of the DN data with the L2L Microarray Database produced contrary evidence. We found a correlation between genes upregulated in DN and the response to serum, EGF and VEGF. The observation that glomerular cells express higher levels of growth factor target genes in DN than in normal kidneys suggests that DN kidneys may be coping adequately with lower VEGF expression. The molecular etiology of DN may, therefore, lie elsewhere.
Three novel themes emerged from the comparison with the L2L Microarray Database of genes downregulated in DN. Firstly, many of these genes are induced by interferon - nine lists related to interferon and the viral response overlap very significantly with the list of genes repressed by DN (p values from 2e-4 to 2e-14). Perhaps related to this, genes downregulated in DN also significantly overlap with genes induced by tumor necrosis factor (TNF)α (p = 5e-5). Secondly, hypoxia-induced genes are repressed in DN - five lists have p values from 8e-3 to 8e-6. Thirdly, and most surprisingly, five lists of genes upregulated in adipocyte differentiation and function overlap with genes repressed by DN (p values from 2e-3 to 2e-7), whereas two lists of genes downregulated during adipocyte differentiation correlate with genes upregulated in DN (p = 0.002 and 0.0008).
The relationship between genes repressed in DN and genes induced by interferon (IFN) illustrates an important caveat regarding tissue-based microarray experiments: the complexity of the tissue itself makes it difficult to determine whether the results reflect changes in expression within glomerular cells, a different degree of leukocyte contamination, or even changing gene expression within those leukocytes. The latter two scenarios are consistent with previous findings of dysfunctional cell-mediated immunity in diabetes [38-41]. The association of genes repressed by DN with those induced by TNFα may be interpreted in this context as well, because at least one study suggested poor response to TNFα as one reason for the immune deficiency in T2DM [39]. Since no cytokines appear on the list of differentially expressed genes, these data suggest - supposing the gene expression changes reflect contaminating leukocytes - that a poor transcriptional response of leukocytes to cytokines may cause the immune deficiency in T2DM.
The most widely accepted theory of pancreatic β-islet cell dysfunction in T2DM is that a variety of inflammatory signals from diet, adipocytes and the immune system combine to trigger apoptosis in those cells [42,43]. Two of the most important signals are thought to be TNFα from adipocytes and IFNγ from leukocytes. It is intriguing, therefore, that while the L2L analysis found downregulation of IFNγ- and TNFα-induced genes in DN, the GO:Biological Process analysis specifically identified the downstream apoptotic effectors of these two cytokines (JAK/STAT for IFNγ, IκK/NFκB for TNFα) as also downregulated in DN. So rather than being an artifact of leukocyte contamination, these results could reflect reduced sensitivity to the blood-borne inflammatory signals that, in sensitive pancreatic islets, trigger β-islet cell apoptosis - the hallmark of the underlying disease.
The second theme - a poor hypoxic response - suggests a transcriptional defect more specific to glomerular cells. At first glance, the direction of this correlation is surprising: DN kidneys should already be under hypoxic stress if poor angiogenesis and endothelial dysfunction are partially responsible for DN. However, this effect is apparently swamped by the ischemia experienced by all kidneys following extraction, before RNA is harvested. Although all kidneys were handled identically, hypoxia-response genes were more strongly induced in the normal controls. This could suggest that DN glomeruli are already stressed, and unable to respond fully to further stress. The result could be a downward spiral of increasing damage and reduced function.
Adipogenesis, the third theme, also seems puzzling at first. Why would adipocyte differentiation genes be differentially regulated in kidney glomeruli? Another hallmark of diabetes is deranged adipocyte function - adipocytes are insulin-resistant, have diminished capacity to store fat, and secrete excessive amounts of inflammatory cytokines and free fatty acids [44]. Such dysfunctional adipocytes may be primarily responsible for creating the chronic inflammatory state that brings about overt disease [45]. Adipocytes are also one of the primary targets of the most widely used class of antidiabetic drugs. Thiazolidinediones (TZDs) are agonists of PPARγ, a transcription factor required for early adipocyte differentiation. TZDs can help restore normal adipocyte function in diabetics [46]. The dysregulation of adipocyte differentiation genes, therefore, may be another fingerprint of the underlying disease, indicating either the dysfunction of contaminating adipocytes in the glomeruli preparations, or a surprising sensitivity of glomerular cells to the same dyslipidemic signals that perturb adipocyte function in diabetics. Interestingly, a microarray analysis of a mouse model of DN, contemporary with this human study, found deregulation of a number of lipid homeostasis genes [47].
Taken together, the L2L results demonstrate the importance of considering T2DM and its complications as part of a single, integrated disease process. The fingerprints of the underlying disease - inflammatory factors and adipocyte dysfunction - are readily detectable in kidney glomeruli, and suggest that the same factors that cause β-islet cell and adipocyte dysfunction are responsible for glomerular dysfunction as well. In fact, PPARγ is expressed in rodent glomeruli [48,49] and treatment with a TZD enhances renal function in both rats and humans [50-52]. It would be interesting to determine which dyslipidemic signals affect DN glomeruli; how those signals are transduced in glomerular cells; and whether the result is abnormal intracellular lipid accumulation [47], or direct inhibition of glomerular function by activation of specific intracellular signaling pathways [50] - either of which might prevent glomerular cells from responding to normal growth and stress signals.
L2L and the genomics of ageing
Deregulation of gene expression is now thought to underlie many of the effects of ageing in a variety of organisms, including humans. There is a well-defined link between human ageing and disruption of normal DNA methylation patterns [53-55]. A 'unified theory of ageing' has even been proposed, which asserts that 'the progressive and patterned alteration of chromosome structure is the primary cause of ageing' [56]. Other investigators have suggested that such transcriptional deregulation is a programmed response to stresses that increase with age [57], the stochastic result of failed genome maintenance [58], or the specific result of the disruption of some critical (but unknown) cellular function [59,60].
We analyzed two recent gene expression studies of the ageing human brain, to see if there were common patterns in the transcriptional deregulation. Lu and colleagues [61] found significant gene expression changes in the frontal cortex of individuals from 26 to 106 years of age. Genes involved in synaptic plasticity, vesicular transport and mitochondrial function were downregulated, while stress-response, antioxidant and DNA repair genes were upregulated. They found increased DNA damage at the promoters of downregulated genes, leading them to suggest that 'DNA damage may reduce the expression of selectively vulnerable genes involved in learning, memory and neuronal survival, initiating a programme of brain ageing that starts early in adult life'. Blalock and colleagues [62] correlated hippocampal gene expression with histological and clinical markers of Alzheimer's disease (AD). They found a large number of genes whose expression changes correlate with either or both incipient and overt disease, and suggest that the pathogenesis of AD is 'genomically orchestrated'. EASE analysis [2] showed that growth, differentiation and tumor suppressor pathways are upregulated early in the disease process, while protein-processing pathways are downregulated.
Using Gene Ontology lists, L2L quickly replicated the EASE results of Blalock et al. (the complete analysis is available on the L2L website [6]). Using the L2L Microarray Database, L2L also revealed a novel link between AD and the hypoxia response. Genes upregulated with overt AD overlapped significantly with two lists of genes upregulated in myocardium during heart failure (p values 2e-5 and 8e-10) and three lists of genes specifically induced by hypoxic stress (p values 0.002 to 0.005). Moreover, genes downregulated with overt AD overlapped with two lists of genes downregulated in heart failure (p values 0.004 and 5e-5).
Most intriguing, though, was that analysis of these two datasets with the L2L Microarray Database showed a surprisingly consistent overlap in gene expression with each other and with a variety of other ageing-related studies, suggesting that models of mammalian ageing exhibit a common transcriptional pattern (Figure 5). The database currently contains a total of 39 ageing-related lists, including the six lists derived from these two studies. Querying those six lists produced 29 instances of significant overlap (p < 0.01) with other ageing-related lists (encompassing 17 of the 39). Furthermore, 24 of the 29 overlaps were in the expected direction (up-up or down-down). In particular, the degree of overlap between these two datasets was dramatic. When the dataset of Lu et al. was compared with the database, genes upregulated in the ageing human brain overlapped very significantly with genes upregulated in incipient (p = 1e-11) and overt (p = 3e-11) AD. Conversely, genes downregulated in the ageing human brain overlapped genes downregulated in incipient (p = 7e-6) and overt (p = 1e-23) AD. Querying the database with the data of Blalock et al. produced similar results (p values ranging from 5e-5 to 8e-27), as well as demonstrating the enormous overlap between the incipient AD and overt AD datasets (p values from 2e-89 to 2e-211). Other significant overlaps were found with the progeroid Werner syndrome [60], caloric restriction in mice [63] and rhesus monkeys [64], ageing monkey muscle [64] and ageing mouse brain [65,66].
Although patterns of related gene expression changes were easily found in a variety of ageing models, we could not clearly define a set of age-regulated genes. A small group of genes was commonly regulated in the two human studies we examined, but none was also consistently regulated in studies of mouse or monkey models, or even in human studies of other tissue types. Indeed, when only those genes that are commonly regulated in human brain were queried against the L2L Microarray Database, no significant overlaps were found except with the studies from which they were derived. Taken together, these data suggest that while transcriptional deregulation is a fundamental feature of cellular ageing phenotypes, the detailed transcriptional profiles are tissue-specific and perhaps, to some degree, stochastic. Thus, ageing-related gene expression changes in different tissues and models are sufficiently similar to suggest a common underlying mechanism, perhaps DNA damage to sensitive promoters [61] or failure to maintain chromatin structure [67]; however, differences between the profiles suggest that the specific genes deregulated in each situation must be drawn from a larger pool of genes exhibiting varying degrees of vulnerability to deregulation. This illustrates both the danger of relying too heavily on a 'critical genes' approach to explain ageing phenotypes, as well as the hope that there may well be a common underlying mechanism of transcriptional dysregulation waiting to be elucidated.
Reliability of L2L results
The question remains as to whether the results of an L2L analysis can be trusted. These concerns fall into two major categories, which might be described as qualitative and quantitative. The qualitative concern is whether the lists of differentially expressed genes in the database are trustworthy, and if comparison to a user's data can be meaningful. The quantitative concern is whether the statistics we use to judge the significance of the overlaps between a user's data and lists from the database provide a useful metric of biological meaning.
Could a small amount of poorly analyzed or biased data in the L2L database poison the well for all who drink? Much like the scientific process as a whole, L2L takes a distributed-competence approach, augmented by independent replication and careful statistical analysis, to mitigate this concern. Our working assumption is that investigators themselves are best qualified to judge the quality of their own data, and that published lists usually include only those genes for which a change call can be assigned with a reasonable probability. We augment this assumption by including in the database, whenever possible, microarray datasets generated by independent groups that have addressed the same or a closely related question. Given the noise inherent in any microarray experiment, a user can feel much more secure interpreting results which reflect overlap with several related database lists from different sources, rather than idiosyncratic overlap with just one list. Finally, L2L calculates a p value for each comparison that provides a quantitative assessment of the significance of an overlap. If an experiment is contaminated with random data due to experimental error or systematic bias, the likelihood of the L2L list derived from that experiment overlapping significantly with any other experimental data would be purely stochastic - unless both experiments suffer from a common systematic bias. For example, we performed a 10,891-trial simulation with randomized data to help validate our sample analysis of diabetic nephropathy. The odds of achieving a p value below 0.05 with random data was no greater than 0.05 for any list in the database, and as low as 0.001 (see supplemental data on the L2L website [6]). In the absence of common systematic bias, therefore, random data are very unlikely to produce spuriously significant L2L results.
There are two major potential sources of systematic bias: genes that are considered a priori to be 'interesting' or 'critical' based on previous data or theory, and platform-specific bias. Certain often-studied, well-understood genes - the very kind that lend themselves to 'critical gene' hypotheses - are represented on virtually all microarray platforms, and thus could be more likely to be found in random data acquired with any platform. Certain genes may also be more likely to be flagged as differentially expressed on a particular type of chip, perhaps because the chip is more sensitive to small variations at particular expression levels or because of probe-specific effects. If any systematic bias exists, it could only represent a higher likelihood of a random change in signal for that gene or probe - the chip does not know whether the control or experimental RNA is washed onto it, or with which dye color. So proper experimental design and data analysis should eliminate these false-positives before a user turns to L2L. The same applies to the published data from which the L2L lists are derived. If any false-positive genes do persist on database lists, the fact that L2L separately analyzes 'up' and 'down' lists mitigates their impact, because they will be randomly distributed between the two lists. These separate lists also provide great potential assurance for the user, if the 'up' and 'down' lists in the user's data both correlate significantly and respectively with the 'up' and 'down' lists (or vice versa) for a particular condition in the database (see Figure 4b, diabetic nephropathy and adipogenesis). The inclusion of data from independent groups can provide further assurance, because the same set of randomly changing genes is unlikely to be found in independent datasets from different platforms. Still, both sources of systematic bias can be directly addressed in a future release of L2L by more sophisticated statistical analysis algorithms. Each list in the database is annotated with the platform that produced it, so the frequency of occurrence of genes among lists from a given platform (platform-specific bias) as well as the overall occurrence of genes in the database (bias toward 'interesting genes') could be used to weight the contribution of each gene match to the overall significance of the overlap between two lists.
Statistical considerations
If we accept in principle that measuring the overlap between a user's list and the various lists in the L2L database can produce biological insights, we still must resolve how to quantify that overlap with a meaningful statistic that provides at least a relative gauge of which overlaps deserve the most attention. Three major considerations are the choice of statistic, the multiple-hypothesis problem, and the issue of p value inflation. We performed a variety of analyses on our sample dataset of genes downregulated in diabetic nephropathy in order to determine how well the relatively simple binomial distribution calculation performs under real-world circumstances. The results for a selection of 22 lists, those upon which we based our conclusions about diabetic nephropathy, are presented in Table 1. Complete results for all lists, and more detailed information about how these analyses were performed, are available on the L2L website [6].
The essential task for a statistical test in over-abundance analysis is to quantify how surprised we should be to see a particular degree of overlap or, conversely, how likely it is that the overlap occurred by chance. If the likelihood of success in a trial is p , and we perform n trials, what are the odds that we will see m or more successes? In the case of L2L, n is the number of probes that map to a list in the database, and p is the likelihood that any one of them will be found in the data by chance - the proportion of probes in the user's data out of all the probes on the microarray. A 'trial' tests whether one of the n probes derived from a database list is found in the user's data; success is a match. The binomial distribution permits the exact calculation of the odds of achieving a particular number of matches out of n trials. The cumulative probability of achieving m or more matches is found as follows:
L2L uses the Double Precision Cumulative Distribution Function Library (DCDFLIB) [68], implemented in the Math::CDF Perl module [69], to compute binomial probabilities. The binomial distribution performs trials with replacement - the odds of scoring a success remain constant for all trials. In reality, a probe can only be selected once, so the hypergeometric distribution, which calculates probabilities without replacement, is more accurate. However, it is more difficult to calculate than the binomial distribution, and in any event approaches the binomial distribution at large values of n and m, where replacement has little impact on the odds of the next trial. Alternatively, the Poisson distribution is easier to calculate than the binomial distribution, and approaches it where values of n are large and p small (as in most L2L analyses) [70]. In our sample dataset of genes upregulated in diabetic nephropathy, the p values calculated from the hypergeometric distribution or Poisson distribution closely followed those calculated from the binomial distribution (Table 1; compare columns 5, 6 and 7). We therefore chose to use the binomial distribution as a reasonable compromise between accuracy and computational requirements.
The multiple-hypothesis problem is that when testing a large number of hypotheses simultaneously - here, that each of the hundreds of lists in the L2L database might overlap significantly with the user's data - the odds of producing a low p value by chance become substantial [71]. For example, with 357 lists in the L2L database, we might expect purely random data to produce about 18 'significant' overlaps with p values <0.05 (357 * 0.05). There are two common approaches that either reduce the odds of seeing any such false-positive p values, or mitigate their effect. The former approach is to control the family-wise error rate, usually by applying some adjustment to the calculated p values. This adjustment can be the same for all p values (termed 'single-step') or can vary as we evaluate each p value in order ('step-down' or 'step-up'). The single-step Bonferroni is the most common adjustment, and is simply the multiplication of the p value by the number of hypotheses (p * n, n being 357 in this case). We found the Bonferroni adjustment to be excessively conservative, based on the simulation-adjusted p values and false discovery rate (see below, and Table 1). The single-step Sidák, which uses the adjustment (1 - (1 - p )^n ), produced near-identical results to the Bonferroni for low p values. Since n has a large initial value, step-down procedures for these two adjustments - where n is decremented by 1 as we adjust each p value in ascending order - did not produce substantially different adjusted p values.
An attractive alternative to simple adjustments based on the number of hypotheses is to perform simulations with random data, and adjust p values based on their frequency of occurrence among the random results. We therefore undertook a 10,891-trial simulation using datasets of the same size as our diabetic nephropathy sample (513 probes), drawn randomly from all the probes on the U95Av2 microarray (10,877 probes). We used true random numbers from Random.org [72] for all simulations. As expected, the median binomial p value calculated from these random data was not significant for any list (Table 1, column 9). We compared each p value from the actual sample data to the simulation-generated p values for that specific list, and for all lists together. In both cases, the frequency of occurrence of a p value equal to or less than the actual p value (that is, the simulation-adjusted p value) was generally lower than the actual p value (Table 1). This shows that, at least for the diabetic nephropathy dataset on the U95Av2 platform, a simple calculation of p values based on the binomial distribution gives a good approximation of the actual likelihood of seeing an overlap by chance. The capability to perform a simulation analysis will be included in a future release of the downloadable L2L application. However, the utility of a simulation analysis is proportional to the number of trials run, because an adjusted p value cannot be lower than (1/number of trials). Each 'trial' is a full-fledged L2L analysis, so a 10,000-trial simulation takes four orders of magnitude longer to run than a single analysis, not considering the time required to create random datasets. The computational requirements are therefore daunting, and preclude it from being practical in a web-based tool.
All such p value adjustments, however they are made, aim to reduce the chances of seeing any false positives. They can therefore be too conservative if, as in most biological questions, permitting a few false-positives is a reasonable trade-off for seeing more true data. The false-discovery rate (FDR) is an increasingly popular approach to the multiple-hypothesis problem that mitigates the effect of false-positives by estimating how many there are at a given level of significance, rather than trying to eradicate them [73]. It can therefore be substantially more powerful than controlling the family-wise error rate. We used our random-data simulation to calculate the FDR at all levels of significance by dividing the average number of random occurrences of a p value less than or equal to a given number by the number of occurrences in the actual data of a p value less than or equal to that number. Column 12 of Table 1 shows that if we use the least significant binomial p value of our 22 sample lists (0.0075) as a cutoff, only 2% of the lists with equal or lower p values are expected to be false positives. Overall, a binomial p value of 0.05 corresponded to an FDR of about 10%, and 0.01 to 2.5%. The capability to calculate FDR from simulation data will be included in a future version of the downloadable L2L application, but these sample data suggest that the simple and economical binomial calculation of L2L, with a rough p value threshold of 0.05-0.01, strikes a reasonable balance between stringency and power.
Finally, we must address the issue of p value inflation: the generation of p values that, while genuinely statistically significant, are devoid of biological meaning. One way this can occur is through the statistics of small numbers - the anthropic principle of over-representation analysis. When only very few genes in the universe being tested possess a given characteristic, even one occurring in the data may be calculated as highly significant. Unlike a Fisher's exact test, the binomial distribution makes no explicit accommodation for small numbers. However, in creating L2L we assumed that comparisons with very short database lists would not be meaningful, and excluded lists (including those generated from GO annotations) with fewer than five genes. For a moderately sized dataset like our sample (513 genes), two out of five probes must match the data for a significant p value (0.01) to be generated. For much smaller datasets, only a single matching probe could produce a significant p value (for 50 genes, 0.02). However, the goal of L2L is to tease out complex patterns of gene expression that might be produced by a kaleidoscope of pathways. There is simply too small a signal among a few dozen genes to identify meaningful patterns, unless the investigator is certain that only a single pathway is at work - in which case L2L is unlikely to be helpful anyway. We therefore intend L2L to be used with relatively large database lists and relatively larger datasets, and in such circumstances the dangers of small numbers should be minor. We quantitatively tested the robustness of L2L's results by performing a 10,891-trial permutation simulation. In each trial, 52 probes from the sample data (10%) were thrown out and replaced with 52 different probes drawn randomly from the universe of the U95Av2 microarray. We found that the median p values generated by the permutations were only slightly reduced from the actual values. In no case was the actual p value a significant outlier among the permuted data: all had list-specific p values of >0.05, most with FDRs of 70-80% (Table 2).
The second potential source of p value inflation arises from the universal nature of the database. The common language, HUGO symbols, must be translated to platform-specific probe identifiers for the user's microarray. If only a handful of genes in a database list are represented on the microarray, and one of those genes happens to be represented by several probes, all of which are differentially expressed in the user's experiment, the list will generate a highly significant p value on the questionably narrow basis of that single gene. A user can see on a Listmatch page exactly which genes or probes created a small but significant overlap, and judge if it appears to be an artifact of translation. Users should be particularly wary of genes used as hybridization controls. We re-analyzed our diabetic nephropathy sample data without probe translation, using only gene symbols (Table 2). Several of the 22 sample lists dropped out of statistical significance; most of these were due to STAT1, an Affymetrix hybridization control, being represented by six probes in the data. Users may wish to remove control probes from their data before analyzing it with L2L. A future release of L2L will incorporate a directed-permutation algorithm into the statistical analysis to ensure that a reported p value is not overly reliant on a single gene.
L2L is a unique microarray analysis tool
The idea of finding the overlap between two lists of differentially expressed genes, like the idea of a central repository of microarray data, dates to the earliest microarray experiments. One of its earliest expressions was through Venn diagrams that compare differentially expressed genes within a single series of experiments. Global clustering of microarrays is a more sophisticated, and more popular, example of this sort of comparative analysis [74], and has proven its worth for class discovery - for example, defining new, and potentially biologically relevant, subspecies of tumors [75,76]; and for class prediction - for example, predicting the behavior and susceptibility to therapy of a tumor by comparison to tumors with known outcomes [77,78]. However, the simpler pair-wise approach of L2L has the advantages of extending well across different platforms and not requiring access to raw data - only to lists of differentially expressed genes. It is well suited, therefore, to its task of finding common patterns between diverse gene expression studies, and enabling biological inferences to be drawn from the commonalities it finds.
VennMapper, created by Smid et al. , is one recent attempt in this direction [32]. It is a software tool that identifies overlaps in lists of differentially expressed genes (defined by an arbitrary fold-change cutoff) from user-supplied heterologous datasets, and calculates the statistical significance of the results using a z-value derived from a normal binomial distribution. The statistical approach is similar to that used by a variety of data mining tools that examine a list of genes for over-representation of GO categories, like GOMiner, EASE, Onto-Express and GO::TermFinder [2-5]. VennMapper and EASE, like the L2L Microarray Analysis Tool, are really general-purpose tools for comparing any given list of genes with any other list of genes. The authors of both tools suggest extending their use to comparing a user's data with 'previously published gene lists' [2], or 'comparing microarray data studying apoptosis, hypoxia, etc. with microarray data focusing on clinical backgrounds, like cancer, (viral) infections or neurological disease' [32]. L2L was conceived and developed independently of either of these tools, but fills the need that their authors, and others, have identified. Moreover, it does so in a way that is at once flexible, powerful, and extensible, yet simple enough to be accessible to every user of microarrays.
Acknowledgements
We are indebted to Roger Bumgarner of the University of Washington Center for Expression Arrays for generous support, suggestions and critiques throughout. We are also grateful for the support of Peter Rabinovich and the Nathan Shock Center of Excellence for the Basic Biology of Aging, at the University of Washington. This work was supported by the NIGMS Medical Scientist Training Program, a fellowship from the Cora May Poncin Foundation (J.C.N), and by NIH GM41624 (A.M.W.).
Figures and Tables
Figure 1 L2L and the L2L microarray database. (a) The centerpiece of L2L is the L2L Microarray Database, a collection of published microarray data in the form of lists of genes that are up- or downregulated in some condition. The L2L Microarray Analysis Tool (MAT) is a program that compares those lists with a user's microarray data, and reports statistically significant overlaps. The analysis tool includes a web browser interface, but the L2L application itself can be downloaded and run directly from the command line for batch or customized analyses. Three additional sets of lists, based on the three organizing principles of Gene Ontology, can also be used with the analysis tool. (b) The L2L Microarray Database contains over 350 lists compiled from over 100 selected microarray publications. A wide variety of topics are represented, from chromatin modifications and DNA damage to the immune response and adipocyte differentiation.
Figure 2 L2L uses a simple web-based interface, and generates easy-to-navigate, annotated HTML pages as output. (a) The L2L web interface. (b) The Results summary page displays each list from the database that significantly matched the data, along with links to list annotations and Listmatch pages. (c) An example Listmatch page, which displays all of the probes on a list that match the data, with a variety of annotations and links to Probematch pages. (d) Probematch pages show all of the lists on which a probe is found, with links back to their Listmatch pages. Arrows indicate sample navigation paths between the output pages.
Figure 3 The L2L application sequentially compares each list in the database with the input data, and records the overlap between the two lists of genes. (a) Each list in the database is a list of HUGO symbols. These are first translated to the corresponding microarray probes that represent those genes. Depending on the microarray, some genes on a list are represented by multiple probes and some by none at all. (b) The program finds the intersection between the translated list of probes from the database and the user's list of probes. The results are logged and written to a raw output file. The program then proceeds to the next list in the database. (c) Once all lists in the database have been compared with the user's data, the program creates a set of HTML pages to browse the output.
Figure 4 L2L analysis of gene expression changes in diabetic nephropathy (DN). (a) Three major conclusions of Baelde et al. [36] revisited. L2L finds support for cytoskeletal dysfunction, but no evidence of reduced nucleotide metabolism. Evidence for the central thesis, reduced tissue repair capacity, is mixed. L2L found reduced expression of IGF-binding proteins, suggesting a defect in response to these growth factors. However, L2L also found a correlation between genes repressed by the serum-response and genes downregulated in DN, as well as a correlation between genes upregulated in DN and genes induced by EGF and VEGF - despite reduced expression of VEGF itself in DN kidneys. (b) Three new biological themes in DN found by L2L. 1. Interferon, TNFα, and their associated apoptotic pathways are all downregulated in DN. 2. The hypoxia response is impaired in DN. 3. Pathways associated with adipogenesis and adipocyte function are downregulated in DN. Complete results, along with descriptions and annotations for all lists, can be found on the L2L website [6]. Red or green denote reduced or increased expression, respectively, in DN or in the condition represented by a list.
Figure 5 L2L analysis of gene expression changes in two studies of the ageing human brain. Lists of differentially expressed genes from Lu et al. (ageing_brain) [61] and Blalock et al. (alzheimers_disease and alzheimers_incipient) [62] were compared with all ageing-related lists in the L2L Microarray Database, including each other (all data are available on the L2L website [6]). Numbers represent binomial p values for significance of overlap. Green denotes overlap between lists of genes upregulated with ageing; red denotes overlap between lists of genes downregulated with ageing; black denotes overlap between lists of contrary directions; yellow denotes self-self comparisons.
Table 1 Sample data subjected to p value adjustment by Bonferroni correction or random-data simulation
Actual diabetic nephropathy (downregulated) data Random-data simulation (10,891 trials)
Name of L2L database list Total U95Av2 probes on list Expected matches Actual matches Binomial p value Hypergeometric p value Poisson p value Bonferroni-adjusted binomial p value Median binomial p value p value (list-specific) of actual binomial p value p value (all lists) of actual binomial p value FDR of actual binomial p value
ifn_beta_up 135 5.83 31 1.9E-14 n/a 4.3E-14 6.7E-12 0.53 <9.2E-05 <2.6E-07 <9.2E-06
ifn_alpha_up 111 4.79 28 2.7E-14 1.4E-14 6.1E-14 9.5E-12 0.52 <9.2E-05 <2.6E-07 <9.2E-06
ifn_all_up 73 3.15 19 2.0E-10 1.4E-10 1.9E-10 7.0E-08 0.61 <9.2E-05 <2.6E-07 <9.2E-06
nf90_up 59 2.55 17 3.1E-10 2.3E-10 2.9E-10 1.1E-07 0.73 <9.2E-05 <2.6E-07 <9.2E-06
ifnalpha_both_up 36 1.55 13 1.6E-09 1.3E-09 1.3E-09 5.9E-07 0.80 <9.2E-05 <2.6E-07 <9.2E-06
adip_diff_cluster2 43 1.86 12 1.8E-07 1.5E-07 9.0E-08 6.6E-05 0.56 <9.2E-05 <2.6E-07 <9.2E-06
emt_up 104 4.49 18 5.1E-07 3.4E-07 2.9E-07 1.8E-04 0.66 <9.2E-05 <2.6E-07 <9.2E-06
hpv31_dn 69 2.98 14 1.3E-06 9.6E-07 6.2E-07 4.6E-04 0.58 <9.2E-05 2.6E-07 9.2E-06
ifnalpha_either_up 83 3.58 15 2.5E-06 1.8E-06 1.2E-06 9.0E-04 0.70 <9.2E-05 1.3E-06 4.2E-05
adip_vs_fibro_up 55 2.38 12 3.2E-06 2.5E-06 1.4E-06 1.2E-03 0.69 <9.2E-05 1.5E-06 4.6E-05
hypoxia_normal_up 243 10.49 27 8.3E-06 n/a 5.5E-06 3.0E-03 0.60 9.2E-05 3.9E-06 9.8E-05
hypoxia_reg 60 2.59 12 8.5E-06 6.5E-06 3.5E-06 3.0E-03 0.74 <9.2E-05 4.4E-06 9.8E-05
tnfalpha_adip_up 17 0.73 6 5.3E-05 4.8E-05 1.2E-05 0.019 0.53 <9.2E-05 2.2E-05 3.4E-04
cmv_up 88 3.80 13 1.0E-04 7.4E-05 4.5E-05 0.037 0.53 <9.2E-05 3.8E-05 5.1E-04
vhl_normal_up 230 9.93 23 1.8E-04 n/a 1.1E-04 0.066 0.53 1.8E-04 6.9E-05 7.9E-04
dsrna_up 48 2.07 9 1.9E-04 1.5E-04 6.2E-05 0.068 0.62 2.8E-04 7.4E-05 8.2E-04
tnfalpha_tgz_adip_up 23 0.99 6 3.5E-04 3.0E-04 8.0E-05 0.12 0.64 5.5E-04 1.5E-04 1.4E-03
tgz_adip_up 26 1.12 6 7.1E-04 6.1E-04 1.7E-04 0.25 0.68 5.5E-04 2.9E-04 2.4E-03
hif1_targets 60 2.59 9 1.0E-03 7.9E-04 3.7E-04 0.37 0.74 6.4E-04 4.5E-04 3.6E-03
adip_vs_preadip_up 53 2.29 8 1.9E-03 1.4E-03 6.2E-04 0.67 0.67 1.7E-03 9.3E-04 6.2E-03
serum_fibroblast_core_dn 148 6.39 14 5.1E-03 n/a 2.5E-03 1 0.62 4.2E-03 2.4E-03 0.013
hypoxia_fibro_up 29 1.25 5 7.5E-03 6.1E-03 1.9E-03 1 0.72 7.3E-03 3.5E-03 0.018
n/a, calculation too complex to perform precisely.
Table 2 Sample data subjected to permutation analysis or comparison by gene symbol instead of probe ID
10% Data permutation (10,891 trials) Comparison by gene symbol
Name of L2L database list Binomial p value (actual) Median permutation binomial p value p value (list-specific) of actual binomial p value p value (all lists) of actual binomial p value FDR of actual binomial p value Total gene symbols on list Expected matches Actual matches Binomial p value
ifn_beta_up 1.9E-14 4.7E-12 0.10 3.2E-03 0.58 93 4.46 25 1.3E-12
ifn_alpha_up 2.7E-14 1.4E-12 0.11 3.5E-03 0.31 71 3.41 22 1.2E-12
ifn_all_up 2.0E-10 1.1E-08 0.20 0.011 0.80 48 2.30 14 3.5E-08
nf90_up 3.1E-10 2.7E-09 0.21 0.012 0.70 37 1.78 11 8.3E-07
ifnalpha_both_up 1.6E-09 2.0E-08 0.28 0.014 0.70 21 1.01 8 3.3E-06
adip_diff_cluster2 1.8E-07 1.5E-06 0.32 0.019 0.86 30 1.44 9 7.7E-06
emt_up 5.1E-07 2.4E-06 0.25 0.021 0.84 68 3.26 11 3.8E-04
hpv31_dn 1.3E-06 7.2E-06 0.30 0.022 0.79 49 2.35 9 4.8E-04
ifnalpha_either_up 2.5E-06 1.2E-05 0.30 0.025 0.83 50 2.40 9 5.6E-04
adip_vs_fibro_up 3.2E-06 2.0E-05 0.34 0.027 0.79 35 1.68 4 0.085
hypoxia_normal_up 8.3E-06 6.3E-05 0.24 0.030 0.77 168 8.06 23 6.5E-06
hypoxia_reg 8.5E-06 4.7E-05 0.35 0.031 0.74 39 1.87 9 7.7E-05
tnfalpha_adip_up 5.3E-05 5.3E-05 0.54 0.048 0.75 8 0.38 1 0.33
cmv_up 1.0E-04 4.1E-04 0.37 0.056 0.74 59 2.83 11 1.0E-04
vhl_normal_up 1.8E-04 4.6E-04 0.31 0.065 0.78 155 7.44 19 1.8E-04
dsrna_up 1.9E-04 9.7E-04 0.45 0.068 0.75 33 1.58 8 1.3E-04
tnfalpha_tgz_adip_up 3.5E-04 3.5E-04 0.54 0.079 0.78 11 0.53 1 0.42
tgz_adip_up 7.1E-04 7.1E-04 0.55 0.093 0.77 14 0.67 1 0.50
hif1_targets 1.0E-03 4.2E-03 0.46 0.10 0.80 34 1.63 6 5.2E-03
adip_vs_preadip_up 1.9E-03 7.6E-03 0.49 0.12 0.80 35 1.68 3 0.24
serum_fibroblast_core_dn 5.1E-03 0.012 0.42 0.16 0.87 114 5.47 9 0.098
hypoxia_fibro_up 7.5E-03 7.5E-03 0.62 0.17 0.86 22 1.06 3 0.086
==== Refs
Ashburner M Ball CA Blake JA Botstein D Butler H Cherry JM Davis AP Dolinski K Dwight SS Eppig JT Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet 2000 25 25 29 10802651 10.1038/75556
Hosack DA Dennis G JrSherman BT Lane HC Lempicki RA Identifying biological themes within lists of genes with EASE. Genome Biol 2003 4 R70 14519205 10.1186/gb-2003-4-10-r70
Zeeberg BR Feng W Wang G Wang MD Fojo AT Sunshine M Narasimhan S Kane DW Reinhold WC Lababidi S GoMiner: a resource for biological interpretation of genomic and proteomic data. Genome Biol 2003 4 R28 12702209 10.1186/gb-2003-4-4-r28
Khatri P Draghici S Ostermeier GC Krawetz SA Profiling gene expression using onto-express. Genomics 2002 79 266 270 11829497 10.1006/geno.2002.6698
Boyle EI Weng S Gollub J Jin H Botstein D Cherry JM Sherlock G GO::TermFinder - open source software for accessing Gene Ontology information and finding significantly enriched Gene Ontology terms associated with a list of genes. Bioinformatics 2004 20 3710 3715 15297299 10.1093/bioinformatics/bth123
L2L Microarray Analysis Tool
Microarray Gene Expression Data Society - MGED Society
Brazma A Hingamp P Quackenbush J Sherlock G Spellman P Stoeckert C Aach J Ansorge W Ball CA Causton HC Minimum information about a microarray experiment (MIAME)-toward standards for microarray data. Nat Genet 2001 29 365 371 11726920 10.1038/ng1201-365
Spellman PT Miller M Stewart J Troup C Sarkans U Chervitz S Bernhart D Sherlock G Ball C Lepage M Design and implementation of microarray gene expression markup language (MAGE-ML). Genome Biol 2002 3 RESEARCH0046 12225585 10.1186/gb-2002-3-9-research0046
Ikeo K Ishi-i J Tamura T Gojobori T Tateno Y CIBEX: center for information biology gene expression database. C R Biol 2003 326 1079 1082 14744116
Brazma A Parkinson H Sarkans U Shojatalab M Vilo J Abeygunawardena N Holloway E Kapushesky M Kemmeren P Lara GG ArrayExpress - a public repository for microarray gene expression data at the EBI. Nucleic Acids Res 2003 31 68 71 12519949 10.1093/nar/gkg091
Rhodes DR Yu J Shanker K Deshpande N Varambally R Ghosh D Barrette T Pandey A Chinnaiyan AM ONCOMINE: a cancer microarray database and integrated data-mining platform. Neoplasia 2004 6 1 6 15068665
Edgar R Domrachev M Lash AE Gene Expression Omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Res 2002 30 207 210 11752295 10.1093/nar/30.1.207
Irizarry RA Bolstad BM Collin F Cope LM Hobbs B Speed TP Summaries of Affymetrix GeneChip probe level data. Nucleic Acids Res 2003 31 e15 12582260 10.1093/nar/gng015
Wain HM Lush MJ Ducluzeau F Khodiyar VK Povey S Genew: the Human Gene Nomenclature Database, 2004 updates. Nucleic Acids Res 2004 32 Database issue D255 D257 14681406 10.1093/nar/gkh072
Tan PK Downey TJ Spitznagel EL JrXu P Fu D Dimitrov DS Lempicki RA Raaka BM Cam MC Evaluation of gene expression measurements from commercial microarray platforms. Nucleic Acids Res 2003 31 5676 5684 14500831 10.1093/nar/gkg763
The Cancer Genome Anatomy Project Batch Gene Finder
MatchMiner
NCBI HomoloGene
Xenarios I Salwinski L Duan XJ Higney P Kim SM Eisenberg D DIP, the Database of Interacting Proteins: a research tool for studying cellular networks of protein interactions. Nucleic Acids Res 2002 30 303 305 11752321 10.1093/nar/30.1.303
KEGG BRITE Database
Bader GD Betel D Hogue CW BIND: the Biomolecular Interaction Network Database. Nucleic Acids Res 2003 31 248 250 12519993 10.1093/nar/gkg056
The Cancer Genome Anatomy Project - Pathways
Dahlquist KD Salomonis N Vranizan K Lawlor SC Conklin BR GenMAPP, a new tool for viewing and analyzing microarray data on biological pathways. Nat Genet 2002 31 19 20 11984561 10.1038/ng0502-19
Segal E Friedman N Koller D Regev A A module map showing conditional activity of expression modules in cancer. Nat Genet 2004 36 1090 1098 15448693
Jenssen TK Laegreid A Komorowski J Hovig E A literature network of human genes for high-throughput analysis of gene expression. Nat Genet 2001 28 21 28 11326270 10.1038/88213
Glenisson P Coessens B Van Vooren S Mathys J Moreau Y De Moor B TXTGate: profiling gene groups with text-based information. Genome Biol 2004 5 R43 15186494 10.1186/gb-2004-5-6-r43
Ge H Liu Z Church GM Vidal M Correlation between transcriptome and interactome mapping data from Saccharomyces cerevisiae. Nat Genet 2001 29 482 486 11694880 10.1038/ng776
Jansen R Greenbaum D Gerstein M Relating whole-genome expression data with protein-protein interactions. Genome Res 2002 12 37 46 11779829 10.1101/gr.205602
Kemmeren P van Berkum NL Vilo J Bijma T Donders R Brazma A Holstege FC Protein interaction verification and functional annotation by integrated analysis of genome-scale data. Mol Cell 2002 9 1133 1143 12049748 10.1016/S1097-2765(02)00531-2
Lee HK Hsu AK Sajdak J Qin J Pavlidis P Coexpression analysis of human genes across many microarray data sets. Genome Res 2004 14 1085 1094 15173114 10.1101/gr.1910904
Smid M Dorssers LC Jenster G Venn Mapping: clustering of heterologous microarray data based on the number of co-occurring differentially expressed genes. Bioinformatics 2003 19 2065 2071 14594711 10.1093/bioinformatics/btg282
AmiGO
GeneCards
Entrez Gene
Baelde HJ Eikmans M Doran PP Lappin DW de Heer E Bruijn JA Gene expression profiling in glomeruli from human kidneys with diabetic nephropathy. Am J Kidney Dis 2004 43 636 650 15042541 10.1053/j.ajkd.2003.12.028
Doniger SW Salomonis N Dahlquist KD Vranizan K Lawlor SC Conklin BR MAPPFinder: using Gene Ontology and GenMAPP to create a global gene-expression profile from microarray data. Genome Biol 2003 4 R7 12540299 10.1186/gb-2003-4-1-r7
Kukreja A Cost G Marker J Zhang C Sun Z Lin-Su K Ten S Sanz M Exley M Wilson B Multiple immuno-regulatory defects in type-1 diabetes. J Clin Invest 2002 109 131 140 11781358 10.1172/JCI200213605
Chang FY Shaio MF Decreased cell-mediated immunity in patients with non-insulin-dependent diabetes mellitus. Diabetes Res Clin Pract 1995 28 137 146 7587921 10.1016/0168-8227(95)00168-8
Eibl N Spatz M Fischer GF Mayr WR Samstag A Wolf HM Schernthaner G Eibl MM Impaired primary immune response in type-1 diabetes: results from a controlled vaccination study. Clin Immunol 2002 103 249 259 12173299 10.1006/clim.2002.5220
Attallah AM Abdelghaffar H Fawzy A Alghraoui F Alijani MR Mahmoud LA Ghoneim MA Helfrich GB Cell-mediated immunity and biological response modifiers in insulin-dependent diabetes mellitus complicated by end-stage renal disease. Int Arch Allergy Appl Immunol 1987 83 278 283 3110074
Donath MY Storling J Maedler K Mandrup-Poulsen T Inflammatory mediators and islet beta-cell failure: a link between type 1 and type 2 diabetes. J Mol Med 2003 81 455 470 12879149 10.1007/s00109-003-0450-y
Rhodes CJ Type 2 diabetes-a matter of beta-cell life and death? Science 2005 307 380 384 15662003 10.1126/science.1104345
Bays H Mandarino L DeFronzo RA Role of the adipocyte, free fatty acids, and ectopic fat in pathogenesis of type 2 diabetes mellitus: peroxisomal proliferator-activated receptor agonists provide a rational therapeutic approach. J Clin Endocrinol Metab 2004 89 463 478 14764748 10.1210/jc.2003-030723
Lazar MA How obesity causes diabetes: not a tall tale. Science 2005 307 373 375 15662001 10.1126/science.1104342
Evans RM Barish GD Wang YX PPARs and the complex journey to obesity. Nat Med 2004 10 355 361 15057233 10.1038/nm1025
Mishra R Emancipator SN Miller C Kern T Simonson MS Adipose differentiation-related protein and regulators of lipid homeostasis identified by gene expression profiling in the murine db/db diabetic kidney. Am J Physiol Renal Physiol 2004 286 F913 F921 15075187 10.1152/ajprenal.00323.2003
Asano T Wakisaka M Yoshinari M Iino K Sonoki K Iwase M Fujishima M Peroxisome proliferator-activated receptor gamma1 (PPARgamma1) expresses in rat mesangial cells and PPARgamma agonists modulate its differentiation. Biochim Biophys Acta 2000 1497 148 154 10838168 10.1016/S0167-4889(00)00054-9
Guan Y Zhang Y Schneider A Davis L Breyer RM Breyer MD Peroxisome proliferator-activated receptor-gamma activity is associated with renal microvasculature. Am J Physiol Renal Physiol 2001 281 F1036 F1046 11704554
Isshiki K Haneda M Koya D Maeda S Sugimoto T Kikkawa R Thiazolidinedione compounds ameliorate glomerular dysfunction independent of their insulin-sensitizing action in diabetic rats. Diabetes 2000 49 1022 1032 10866056
Imano E Kanda T Nakatani Y Nishida T Arai K Motomura M Kajimoto Y Yamasaki Y Hori M Effect of troglitazone on microalbuminuria in patients with incipient diabetic nephropathy. Diabetes Care 1998 21 2135 2139 9839106
Bakris G Viberti G Weston WM Heise M Porter LE Freed MI Rosiglitazone reduces urinary albumin excretion in type II diabetes. J Hum Hypertens 2003 17 7 12 12571611 10.1038/sj.jhh.1001444
Issa JP Epigenetic variation and human disease. J Nutr 2002 132 8 Suppl 2388S 2392S 12163698
Imai S Kitano H Heterochromatin islands and their dynamic reorganization: a hypothesis for three distinctive features of cellular aging. Exp Gerontol 1998 33 555 570 9789733 10.1016/S0531-5565(98)00037-0
Richardson B Impact of aging on DNA methylation. Ageing Res Rev 2003 2 245 261 12726774 10.1016/S1568-1637(03)00010-2
Jameson CW Towards a unified and interdiciplinary model of ageing. Med Hypotheses 2004 63 83 86 15193354 10.1016/j.mehy.2004.01.021
Roy AK Oh T Rivera O Mubiru J Song CS Chatterjee B Impacts of transcriptional regulation on aging and senescence. Ageing Res Rev 2002 1 367 380 12067592 10.1016/S1568-1637(02)00006-5
Hasty P Campisi J Hoeijmakers J van Steeg H Vijg J Aging and genome maintenance: lessons from the mouse? Science 2003 299 1355 1359 12610296 10.1126/science.1079161
Vijg J Calder RB Transcripts of aging. Trends Genet 2004 20 221 224 15145572 10.1016/j.tig.2004.04.007
Kyng KJ May A Kolvraa S Bohr VA Gene expression profiling in Werner syndrome closely resembles that of normal aging. Proc Natl Acad Sci USA 2003 100 12259 12264 14527998 10.1073/pnas.2130723100
Lu T Pan Y Kao SY Li C Kohane I Chan J Yankner BA Gene regulation and DNA damage in the ageing human brain. Nature 2004 429 883 891 15190254 10.1038/nature02661
Blalock EM Geddes JW Chen KC Porter NM Markesbery WR Landfield PW Incipient Alzheimer's disease: microarray correlation analyses reveal major transcriptional and tumor suppressor responses. Proc Natl Acad Sci USA 2004 101 2173 2178 14769913 10.1073/pnas.0308512100
Lee CK Klopp RG Weindruch R Prolla TA Gene expression profile of aging and its retardation by caloric restriction. Science 1999 285 1390 1393 10464095 10.1126/science.285.5432.1390
Kayo T Allison DB Weindruch R Prolla TA Influences of aging and caloric restriction on the transcriptional profile of skeletal muscle from rhesus monkeys. Proc Natl Acad Sci USA 2001 98 5093 5098 11309484 10.1073/pnas.081061898
Jiang CH Tsien JZ Schultz PG Hu Y The effects of aging on gene expression in the hypothalamus and cortex of mice. Proc Natl Acad Sci USA 2001 98 1930 1934 11172053 10.1073/pnas.98.4.1930
Lee CK Weindruch R Prolla TA Gene-expression profile of the ageing brain in mice. Nat Genet 2000 25 294 297 10888876 10.1038/77046
Bandyopadhyay D Medrano EE The emerging role of epigenetics in cellular and organismal aging. Exp Gerontol 2003 38 1299 1307 14698809 10.1016/j.exger.2003.09.009
DCDFLIB
CPAN - Math-CDF
Ewens WJ Grant GR Statistical Methods in Bioinformatics: An Introduction 2005 2 New York: Springer Science+Business Media
Dudoit S Shaffer JP Boldrick JC Multiple hypothesis testing in microarray experiments. Stat Sci 2003 18 71 103 10.1214/ss/1056397487
Random.org
Benjamini Y Hochberg Y Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Statist Soc B 1995 57 289 300
Quackenbush J Computational analysis of microarray data. Nat Rev Genet 2001 2 418 427 11389458 10.1038/35076576
van't Veer LJ Dai H van de Vijver MJ He YD Hart AA Mao M Peterse HL van der Kooy K Marton MJ Witteveen AT Gene expression profiling predicts clinical outcome of breast cancer. Nature 2002 415 530 536 11823860 10.1038/415530a
Alizadeh AA Eisen MB Davis RE Ma C Lossos IS Rosenwald A Boldrick JC Sabet H Tran T Yu X Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature 2000 403 503 511 10676951 10.1038/35000501
Sorlie T Perou CM Tibshirani R Aas T Geisler S Johnsen H Hastie T Eisen MB van de Rijn M Jeffrey SS Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications. Proc Natl Acad Sci USA 2001 98 10869 10874 11553815 10.1073/pnas.191367098
Shipp MA Ross KN Tamayo P Weng AP Kutok JL Aguiar RC Gaasenbeek M Angelo M Reich M Pinkus GS Diffuse large B-cell lymphoma outcome prediction by gene-expression profiling and supervised machine learning. Nat Med 2002 8 68 74 11786909 10.1038/nm0102-68
|
16168088
|
PMC1242216
|
CC BY
|
2021-01-04 16:05:42
|
no
|
Genome Biol. 2005 Aug 31; 6(9):R81
|
utf-8
|
Genome Biol
| 2,005 |
10.1186/gb-2005-6-9-r81
|
oa_comm
|
==== Front
Ann Gen PsychiatryAnnals of General Psychiatry1744-859XBioMed Central London 1744-859X-4-161616805510.1186/1744-859X-4-16Case ReportPregnancy and delivery while receiving vagus nerve stimulation for the treatment of major depression: a case report Husain Mustafa M [email protected] Diane [email protected] Kenneth [email protected] University of Texas Southwestern Medical Center at Dallas, 5323 Harry Hines Blvd, Dallas, Texas 75390-8898, USA2005 16 9 2005 4 16 16 25 7 2005 16 9 2005 Copyright © 2005 Husain et al; licensee BioMed Central Ltd.2005Husain et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Depression during pregnancy can have significant health consequences for the mother and her infant. Antidepressant medications, which pass through the placenta, may increase the risk of low birth weight and preterm delivery. The use of selective serotonin reuptake inhibitors (SSRIs) during pregnancy may induce serotonergic symptoms in the infant after delivery. Antidepressant medications in breast milk may also be passed to an infant. Vagus nerve stimulation (VNS) therapy is an effective non-pharmacologic treatment for treatment-resistant depression (TRD), but little information exists regarding the use of VNS therapy during pregnancy.
Case presentation
The patient began receiving VNS therapy for TRD in March 1999. The therapy was effective, producing substantial reductions in depressive symptoms and improvement of function. In 2002, the patient reported that she was pregnant. She continued receiving VNS therapy throughout her pregnancy, labor, and delivery, which enabled the sustained remission of her depression. The pregnancy was uneventful; a healthy daughter was delivered at full term.
Conclusion
In this case, VNS therapy provided effective treatment for TRD during pregnancy and delivery. VNS was safe for the patient and her child.
==== Body
Background
A pregnant patient with major depression requires effective management of depressive symptoms for her own health and that of her child. Estimates of the prevalence of depression among pregnant women vary widely, ranging from 3.3% for major depression [1] to 20% for any type of depression [2]. Rates of depression may be as high as 51% in selected populations [3]. These rates compare with a 12-month worldwide prevalence of depression of 9.5% in women [4]. Among pregnant women with depression, many are untreated, sometimes discontinuing treatment for depression after becoming pregnant [1,5].
Depression during pregnancy can have many serious consequences. For the mother, depression is associated with an overall decline in general health, physical and social functioning, an increase in the experience of pain [3], and obstetric complications [6-8]. Depression in late pregnancy is associated with post-partum depression [2], while depression in early pregnancy increases the risk of preeclampsia, a major complication characterized by rapidly progressive hypertension with proteinuria, edema, or both [9]. For the infant, maternal depression during pregnancy was associated with admission to a neonatal intensive care unit [7] and with spontaneous preterm delivery in one study [10] but not in another [11].
Because of the importance of managing depression during pregnancy, numerous studies have examined the effects of antidepressant medications on fetal and infant development. Antidepressants and their metabolites pass through the placenta [12] and increase the risk of low birth weight [13,14] and preterm delivery [14,15]. Use of selective serotonin reuptake inhibitors (SSRIs) by mothers during pregnancy has been associated with substantially reduced levels of platelet serotonin in newborns [16], which may account for SSRI-induced serotonergic symptoms [17], serotonin withdrawal syndrome[18], tremulousness, reduced motor activity, heart rate variability [15], and blunted pain response [19,20]. Increased dosing of SSRIs may be required to maintain euthymia during later stages of pregnancy [21], which may exacerbate some effects. Antidepressants are transmitted to infants in breast milk, where they usually have no discernible clinical effect. However, in isolated reports, antidepressants in breast milk have been associated with reduced feeding, somnolence, reduced growth, and possible seizure [22]. Because both depression and its treatment with pharmacologic interventions may pose risks to the patient and her child, it is important to identify safe nonpharmacologic therapies for that may be used to treat major depressive episodes during pregnancy.
Vagus nerve stimulation (VNS) therapy has been evaluated for use in TRD [23-25]. A small pulse generator implanted subcutaneously in the left thoracic area delivers mild programmed pulses through an implanted lead to the left vagus nerve in the neck. Approved for the treatment of epilepsy since 1997, VNS therapy has been administered to more than 32,000 patients worldwide [26]. Several clinical studies have evaluated the use of adjunctive VNS therapy in chronic or recurrent TRD.
In a 3-month open-label pilot study of patients with chronic or recurrent TRD (bipolar or unipolar, defined by Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (DSM-IV) criteria [27], patients receiving adjunctive VNS therapy exhibited statistically significant improvements in average scores on the Hamilton 28-Item Rating Scale for Depression (HRSD28), Montgomery Asberg Depressive Rating Scale (MADRS), Global Assessment of Function (GAF), and Clinical Global Impression – Severity (CGI-S) scales [24]. After a year of follow-up, adjunctive VNS therapy was associated with sustained symptomatic benefit and sustained or enhanced functional status [25].
Because pregnancy was a contraindication for enrollment in the VNS studies of patients with TRD, there have been no studies of the use of VNS therapy among pregnant patients. A report of eight pregnancies in patients receiving VNS therapy for pharmacoresistent epilepsy has been reported [28] concluded that VNS therapy does not prevent conception and is not associated with any adverse effects on the pregnancy or the neonate.
Case presentation
The study from which this case report was derived was conducted in accordance with the ethical principles defined in the Declaration of Helsinki of the World Medical Association. The protocol was approved by the Institutional Review Boards (IRBs) of participating institutions, and each patient provided written informed consent.
The case study patient, a Caucasian woman aged 28 years with a DSM-IV diagnosis of unipolar depression, was enrolled in the acute and long-term phases of the pilot study of VNS therapy for TRD. At acute-phase study entry, she was noted to be obese and to have mild bronchoconstriction, as well as hypertension, sleep apnea, and arthritis in her knees, ankles, and feet. She reported that she had suffered from recurring depression for 10 years, confounded by obesity, despite pharmacologic treatment and psychotherapy. Her current depressive episode, which had begun 22 months before study enrollment, was found to be resistant to six different antidepressants (citalopram, sertraline, venlafaxine, paroxetine, bupropion, and clonazepam) and the atypical antipsychotic risperidone. In the year preceding enrollment in the study, she had been hospitalized twice for depression. The patient's baseline physical and clinical characteristics are summarized in Table 1.
Table 1 Baseline Physical and Clinical Characteristics of the Patient
Characteristic
Age 28 years
Height 168 cm
Weight 160 kg
Heart rate (BPM) 106
Blood pressure Systolic: 122
Diastolic: 88
Neurological examination Sad/depressed affect; other parameters within normal limits
Clinical assessment
HRSD28 49
MADRS 36
GAF 42
CGI-S 6
cm: centimeters; kg: kilograms; BPM: Beats per minute; HRSD28: Hamilton 28-Item Rating Scale for Depression; MADRS: Montgomery Asberg Depressive Rating Scale; GAF: Global Assessment of Function; CGI-S: Clinical Global Impression – Severity
The VNS therapy device and leads were surgically implanted on February 26, 1999. After recovery, the patient started receiving VNS therapy on March 17, 1999, with the initial stimulation parameters set as shown in Table 2. Substantial improvement was evident after 4 weeks, with depressive symptoms reduced and functioning improved as indicated by HRSD28, MADRS, CGI, and GAF scores (Figures 1 through 3). After 11 months, her HRSD28 score had decreased to 7 and her GAF score had reached 96. The patient's VNS output current was reduced to 0.25 milliamperes (mA) on August 31, 2001; other stimulation parameters were unchanged.
Figure 1 The patient experienced a substantial reduction in symptoms after receiving VNS therapy, as indicated by HRSD28 and MADRS scores. VNS therapy was initiated on March 17, 1999. The patient reported her pregnancy on May 30, 2002, and delivered a healthy child on January 24, 2003; remission of depression was sustained during the pregnancy.
Figure 2 The patient's score on the CGI-S scale also indicated a substantial improvement after the initiation of VNS therapy. Pregnancy from May 30, 2002 to January 24, 2003 did not significantly affect the patient's CGI scores.
Figure 3 Improvement in functioning is demonstrated by increases in GAF scores, beginning shortly after the start of VNS therapy and continuing through her pregnancy and delivery.
Table 2 Vagus Nerve Stimulation Therapy Parameter Values
Parameter Value
March 17, 1999 (beginning of stimulation) August 2001 During Pregnancy
Output current 0.50 mA 0.25 mA 0.25 mA
Signal frequency 20 Hz 20 Hz 20 Hz
Pulse width 500 μsec 250 μsec 250 μsec
Signal ON time 30 sec 30 sec 30 sec
Signal OFF time 5 min 5 min 5 min
mA: milliamperes; Hz: Hertz; μsec: microseconds; sec: seconds
With her depression in remission, the patient underwent gastric bypass surgery for obesity on December 22, 2000. The pulse generator was turned off on December 21, 2000 in preparation for her surgery; VNS therapy was resumed on March 27, 2001. Twoyears after the surgery, she had lost approximately 55 kg.
On May 30, 2002, the patient reported that she was pregnant with her first child. She was informed that, while information was limited on the effects of VNS therapy during pregnancy, no safety issues were known that would affect the pregnancy. The patient decided to continue receiving VNS therapy during the pregnancy; no changes were made in stimulation parameters. In addition, she continued to receive citalopram 80 mg per day and bupropion 400 mg per day, after dosage reductions were considered and rejected by her physicians. She remained in clinical remission of depression throughout her pregnancy (Figures 1 through 3). In compliance with the clinical study protocol, the pregnancy was reported as a serious adverse event that was not related to VNS therapy.
After an uneventful gestation period and normal spontaneous vaginal delivery with epidural anesthesia, the patient delivered a healthy daughter at full term on January 24, 2003. The infant weighed 3.1 kg and was approximately 49 cm long.
VNS therapy was administered at the patient's normal settings throughout labor and delivery. Contingency plans had been made to discontinue stimulation if the patient had required a Caesarian section procedure in which electrocautery might be used. (To avoid damage to the pulse generator and leads, the manufacturer recommends that electrosurgery electrodes be placed as far as possible from the implant, out of the direct path of current flow. Confirmation of correct programmed function of the device after electrosurgery is also recommended.) However, the patient did not require a Caesarian section, and programmed VNS therapy was continued during labor and delivery.
The patient reported an episode of postpartum depression lasting 11 days after delivery. She attributed the depressive episode to difficulties in breast-feeding, and the episode resolved without specific treatment. The child, now aged approximately two years, exhibits normal age-appropriate development.
Another serious adverse event, which was not considered to be associated with VNS therapy, occurred after implant: an episode of thrombophlebitis that resolved with medical therapy. Mild adverse events that the study investigator considered possibly or definitely related to the implant procedure or to VNS therapy were one episode each of moderate leg pain, discomfort in the lower incisors during stimulation, dizziness that resulted in a fall, and surgical wound opening. The patient also experienced periodic hoarseness, a common side effect associated with VNS therapy that is considered tolerable by most patients. No adverse events associated with the VNS therapy occurred during pregnancy, labor, or delivery.
Conclusion
Management of pregnancy in a woman with depression requires careful monitoring and treatment of depressive symptoms in addition to other aspects of the patient's condition. This patient, who had severe depression, experienced sustained remission of her TRD during pregnancy while receiving VNS therapy in combination with citalopram 80 mg per day and bupropion 400 mg per day. In this case, VNS therapy provided effective adjunctive treatment for the patient's depression during pregnancy and delivery; VNS was safe for the patient and her child.
Competing interests
Mustafa Husain declares that, in the last five years, he has received research grants from Cyberonics, Inc. and is on the Cyberonics Speakers' Bureau. Cyberonics, Inc. is funding the development and article processing fees associated with this manuscript. Dr. Husain further declares that he does not own any stock in Cyberonics, Inc.
Diane Stegman declares that, in the last five years, she has received fees from Cyberonics, Inc. for her work as clinical study coordinator. Ms. Stegman further declares that she does not own any stock in Cyberonics, Inc.
Kenneth Trevino declares that he has no competing interests.
Authors' contributions
MMH was the principal investigator of the VNS pilot study. DS was the study coordinator. KT assisted in data analysis and manuscript preparation.
Acknowledgements
The authors gratefully acknowledge the cooperation of the patient described in this case report, from whom written consent was obtained for the publication of this study.
This clinical study was supported in part by a grant from Cyberonics, Inc. Medical writing assistance was provided by Sue Hudson, whose services were funded by Cyberonics, Inc. Clinical monitors from Cyberonics, Inc., collected the data for this pilot study and encouraged the authors to submit this case report to help increase the understanding of VNS therapy and pregnancy.
==== Refs
Andersson L Sundstrom-Poromaa I Bixo M Wulff M Bondestam K aStrom M Point prevalence of psychiatric disorders during the second trimester of pregnancy: a population-based study Am J Obstet Gynecol 2003 189 148 154 12861154 10.1067/mob.2003.336
Josefsson A Berg G Nordin C Sydsjo G Prevalence of depressive symptoms in late pregnancy and postpartum Acta Obstet Gynecol Scand 2001 80 251 255 11207491 10.1034/j.1600-0412.2001.080003251.x
McKee MD Cunningham M Jankowski KR Zayas L Health-related functional status in pregnancy: relationship to depression and social support in a multi-ethnic population Obstet Gynecol 2001 97 988 993 11384708 10.1016/S0029-7844(01)01377-1
World Health Organization (WHO) Mental and neurological disorders [prevalence of depression]
Marcus SM Flynn HA Blow FC Barry KL Depressive symptoms among pregnant women screened in obstetrics settings J Womens Health (Larchmt) 2003 12 373 380 12804344
Preti A Cardascia L Zen T Pellizzari P Marchetti M Favaretto G Miotto P Obstetric complications in patients with depression--a population-based case-control study J Affect Disord 2000 61 101 106 11099747 10.1016/S0165-0327(99)00185-8
Chung TK Lau TK Yip AS Chiu HF Lee DT Antepartum depressive symptomatology is associated with adverse obstetric and neonatal outcomes Psychosom Med 2001 63 830 834 11573032
Jablensky AV Morgan V Zubrick SR Bower C Yellachich LA Pregnancy, delivery, and neonatal complications in a population cohort of women with schizophrenia and major affective disorders Am J Psychiatry 2005 162 79 91 15625205 10.1176/appi.ajp.162.1.79
Kurki T Hiilesmaa V Raitasalo R Mattila H Ylikorkala O Depression and anxiety in early pregnancy and risk for preeclampsia Obstet Gynecol 2000 95 487 490 10725477 10.1016/S0029-7844(99)00602-X
Orr ST James SA Blackmore Prince C Maternal prenatal depressive symptoms and spontaneous preterm births among African-American women in Baltimore, Maryland Am J Epidemiol 2002 156 797 802 12396996 10.1093/aje/kwf131
Andersson L Sundstrom-Poromaa I Wulff M Astrom M Bixo M Neonatal outcome following maternal antenatal depression and anxiety: a population-based study Am J Epidemiol 2004 159 872 881 15105180 10.1093/aje/kwh122
Hendrick V Stowe ZN Altshuler LL Hwang S Lee E Haynes D Placental passage of antidepressant medications Am J Psychiatry 2003 160 993 996 12727706 10.1176/appi.ajp.160.5.993
Chambers CD Johnson KA Dick LM Felix RJ Jones KL Birth outcomes in pregnant women taking fluoxetine N Engl J Med 1996 335 1010 1015 8793924 10.1056/NEJM199610033351402
Hendrick V Smith LM Suri R Hwang S Haynes D Altshuler L Birth outcomes after prenatal exposure to antidepressant medication Am J Obstet Gynecol 2003 188 812 815 12634662 10.1067/mob.2003.172
Zeskind PS Stephens LE Maternal selective serotonin reuptake inhibitor use during pregnancy and newborn neurobehavior Pediatrics 2004 113 368 375 14754951 10.1542/peds.113.2.368
Anderson GM Czarkowski K Ravski N Epperson CN Platelet serotonin in newborns and infants: ontogeny, heritability, and effect of in utero exposure to selective serotonin reuptake inhibitors Pediatr Res 2004 56 418 422 15240861
Laine K Heikkinen T Ekblad U Kero P Effects of exposure to selective serotonin reuptake inhibitors during pregnancy on serotonergic symptoms in newborns and cord blood monoamine and prolactin concentrations Arch Gen Psychiatry 2003 60 720 726 12860776 10.1001/archpsyc.60.7.720
Sanz EJ De-las-Cuevas C Kiuru A Bate A Edwards R Selective serotonin reuptake inhibitors in pregnant women and neonatal withdrawal syndrome: a database analysis Lancet 2005 365 482 487 15705457
Morag I Batash D Keidar R Bulkowstein M Heyman E Paroxetine use throughout pregnancy: does it pose any risk to the neonate? J Toxicol Clin Toxicol 2004 42 97 100 15083945 10.1081/CLT-120028753
Oberlander TF Grunau RE Fitzgerald C Papsdorf M Rurak D Riggs W Pain reactivity in 2-month-old infants after prenatal and postnatal serotonin reuptake inhibitor medication exposure Pediatrics 2005 115 411 425 15687451 10.1542/peds.2004-0420
Hostetter A Stowe ZN Strader JRJ McLaughlin E Llewellyn A Dose of selective serotonin uptake inhibitors across pregnancy: clinical implications Depress Anxiety 2000 11 51 57 10812529 10.1002/(SICI)1520-6394(2000)11:2<51::AID-DA1>3.0.CO;2-R
Gentile S The safety of newer antidepressants in pregnancy and breastfeeding Drug Saf 2005 28 137 152 15691224
Rush AJ George MS Sackeim HA Marangell LB Husain MM Giller C Nahas Z Haines S Simpson RKJ Goodman R Vagus nerve stimulation (VNS) for treatment-resistant depressions: a multicenter study Biol Psychiatry 2000 47 276 286 10686262 10.1016/S0006-3223(99)00304-2
Sackeim HA Rush AJ George MS Marangell LB Husain MM Nahas Z Johnson CR Seidman S Giller C Haines S Simpson RKJ Goodman RR Vagus nerve stimulation (VNS) for treatment-resistant depression: efficacy, side effects, and predictors of outcome Neuropsychopharmacology 2001 25 713 728 11682255 10.1016/S0893-133X(01)00271-8
Marangell LB Rush AJ George MS Sackeim HA Johnson CR Husain MM Nahas Z Lisanby SH Vagus nerve stimulation (VNS) for major depressive episodes: one year outcomes Biol Psychiatry 2002 51 280 287 11958778 10.1016/S0006-3223(01)01343-9
Cyberonics I Cyberonics Reports Financial Results for Q4 and FY 2005 and Postpones Investor Day Pending Final FDA TRD Decision
American Psychiatric Association Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (DSM-IV) 1994 Washington, DC , American Psychiatric Association
Ben-Menachem B Ristanovic R Murphy J Gestational outcomes in patients with epilepsy receiving vagus nerve stimulation [abstract] Epilepsia 1998 39 Suppl 6: Epilepsia 180
|
16168055
|
PMC1242217
|
CC BY
|
2021-01-04 16:39:07
|
no
|
Ann Gen Psychiatry. 2005 Sep 16; 4:16
|
utf-8
|
Ann Gen Psychiatry
| 2,005 |
10.1186/1744-859X-4-16
|
oa_comm
|
==== Front
Ann Clin Microbiol AntimicrobAnnals of Clinical Microbiology and Antimicrobials1476-0711BioMed Central London 1476-0711-4-141616805910.1186/1476-0711-4-14Case ReportNeurobrucellosis presenting as an intra-medullary spinal cord abscess Vajramani Girish V [email protected] Mahantesh B [email protected] Chidanand S [email protected] Department of Neurosurgery, Jawaharlal Nehru Medical College and KLES Hospital, Belgaum, Karnataka, India2 Department of Microbiology, Jawaharlal Nehru Medical College and KLES Hospital, Belgaum, Karnataka, India2005 16 9 2005 4 14 14 5 6 2005 16 9 2005 Copyright © 2005 Vajramani et al; licensee BioMed Central Ltd.2005Vajramani et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Of the diverse presentation of neurobrucellosis, intra-medullary spinal cord abscess is extremely rare. Only four other cases have been reported so far. We present a case of spinal cord intra-medullary abscess due to Brucella melitensis.
Case presentation
A forty-year-old female presented with progressive weakness of both lower limb with urinary incontinence of 6 months duration. She was febrile. Neurological examination revealed flaccid areflexic paraplegia with T10 below sensory impairment including perianal region. An intramedullary mass was diagnosed on Magnetic Resonance Image (MRI) scan extending from T12 to L2. At surgery, a large abscess was encountered at the conus medullaris, from which Brucella melitensis was grown on culture. She was started on streptomycin and doxycycline for 1 month, followed by rifampicin and doxycycline for 1 month. At 2-year follow-up, she had recovered only partially and continued to have impaired bladder function.
Conclusion
Neurobrucellosis, if not treated early, can result in severe neurological morbidity and sequale, which may be irreversible. Hence it is important to consider the possibility of neurobrucellosis in endemic region and treat aggressively.
NeurobrucellosisBrucella melitensisspinal cordintra-medullary abscessMRI
==== Body
Background
The presentation of neurobrucellosis, which encompass neuro-psychiatric disorders in brucellosis, is varied. It can present in acute form as meningoencephalitis or in chronic form, where there is involvement of peripheral nervous system or central nervous system. Chronic form includes epidural granuloma, demyelination of brain or spinal nerve roots, long tract degeneration etc. Meningitis has been the most frequent presentation, occurring in about 50% of the cases [1]. Extensive intra-medullary spinal cord abscess due to Brucella, is exceedingly rare. Only four other cases have been reported so far [2-5]. We present a case of intra-medullary spinal cord abscess in which Brucella melitensis was cultured from pus.
Case presentation
A forty-year-old housewife, from a very low socio-economic status group, presented with history of gradually progressive weakness of left lower limb of 6 months duration, rapidly progressive weakness of right lower limb of 8 days duration and urinary incontinence of 6 months duration. She had been living near a very unhygienic abattoir and admitted to drinking unpasteurised goat's milk. There was history of fever on and off with night sweats. There was no history trauma or past history of tuberculosis. On examination, she was moderately built and nourished. General physical examination was normal. She was febrile with a temperature of 99°F (37.2°C). Vital parameters were normal. Neurologically she was conscious, alert and orientated. Cranial nerve examination was normal. There was no papilledema and meningeal signs were absent. She had flaccid areflexic paraplegia with power 0/5 (MRC grade). She had impaired sensations in both lower limbs with a level at T10. Perianal sensations were impaired and she had poor anal tone. Routine haematological parameters revealed a total white blood cell (WBC) count of 13,980/cu mm with neutrophil predominance. Erythrocyte sedimentation rate ESR (Westergreen) was 50 mm in 1 hour. Standard agglutination test (tube) titer was 1:320 and 2-mercaptoethanol agglutination test titer was 1:80. Plain radiograph of lumbosacral spine was normal. MRI scan of the spine showed a lesion in the spinal cord extending from lower part of T12 to L2. It was hyper-intense on T1WI and iso-intense on T2WI There was cord edema extending cranially up to T10 (figure 1).
Figure 1 MRI scan showing the intramedullary lesion at conus medullaris. The lesion is hyperintense on T1WI and isointense on T2WI.
She underwent T11 to L3 laminectomy. The lower end of the cord and the conus medullaris were swollen and the cauda equina nerve roots were pushed to the right side. Myelotomy was done at the conus level. At a depth of about 0.5 cm, purulent fluid was encountered, which was sent immediately for microbiological analysis. Under operating microscope the abscess cavity was visualized through the limited myelotomy (fig 2). The abscess was completely evacuated, after which the cord and conus had become lax and pulsating well. Dura was closed completely.
Figure 2 Operative microphotograph showing the intra-medullary abscess.
Pus revealed gram-negative bacilli. It was inoculated aerobically [Brucella agar, chocolate and MacConkey media], and anaerobically [Kanamycin-vancomycin laked sheep blood agar (KVLB) and Bacteroides bile esculin agar (BBE)]. Brucella agar and CA were incubated in CO2 jar and after 2 days minute translucent colonies were seen. Gram stain from culture showed gram-negative bacilli. Oxidase, catalase, and urease test were positive. There was no H2S production and it was resistant to dye inhibition. The organism was confirmed as Brucella melitensis [6,7]. The organism isolated in blood culture taken preoperatively, also was identified as Brucella melitensis.
Postoperatively she had fever, headaches and vomiting lasting for about 1 week. It subsided once antibiotics were instituted. She was started on injection streptomycin 1 gm once a day for 1 month with oral doxycycline 100 mgm twice a day for 1 month. After one month she received oral rifampicin 450 mgm once a day with oral doxycycline 100 mgm twice a day for 1 month. Dexamethasone was given only perioperatively and was rapidly tapered and stopped in the post-operative period. Post operatively she gradually improved in neurological status. At 2-year follow up she had grade 3/5 power in both lower limbs and was mobilising on a wheel chair. The urinary symptoms did not resolve and she continues to be on Foley catheter. She refused a repeat MRI scan, as she could not afford it.
Discussion
Brucellosis is a common zoonosis in many parts of the world. Various types of central nervous system (CNS) involvement in brucellosis have been reported [1,8-11], the estimated incidence varying from 5–25% in different series, with an average of 3–5% [1,11,12]. The exact pathogenesis of CNS involvement is not clear. Various mechanisms have been postulated. It is known that Brucella organisms are capable of prolonged intracellular survival within phagocytes. Decreased host immunity may allow the organisms to proliferate [13]. The organism may act directly or indirectly through its endotoxins [1]. Immune mediated demyelination has been proposed to explain certain chronic forms of neurobrucellosis [11]. The cord or nerve root may secondarily be involved due to spondylitis, vasculitis and arachnoiditis [10].
Brucellosis is not uncommon in Belgaum district of Karnataka. Various forms of brucellosis including neurobrucellosis, have been reported from this region [8,9,12]. None of them had an intramedullary involvement of spinal cord. There are only four previous reports, in the world literature, of intramedullary involvement by Brucella (table 1). Systemic brucellosis was seen in all of them including the present case. Only in the present case was Brucella melitensis cultured from the intramedullary pus-in others either, it was not biopsied or there was no growth, the infection being suspected because of presence of systemic brucellosis. In the case reported in 1994 by Cokca et al, a 17-year old boy had an intramedullary dermoid that was infected with Brucella abortus type 3. At surgery, multiple cavitary abscesses containing hair was drained. He responded well to surgical drainage and medical treatment [2]. Bingol et al, in 1999, presented a 40-year old female who had a 10 × 30 mm intramedullary granuloma with surrounding oedema on MRI scan. This was presumed to be a Brucella granuloma as she was diagnosed and treated for systemic brucellosis about 4 months ago. She needed extended period of antibiotics with which there was complete resolution of the lesion as seen on the follow-up scan [3]. Novati et al, in 2002, described a 24-year old man who was diagnosed to have a focal abscess, 15 mm in diameter, within the dorsal tract of the spinal cord with perilesional oedema. Blood and bone marrow aspirate had grown Brucella melitensis and the patient was started on antibiotics for a period of 6 months, following which the abscess resolved [4]. Helvaci et al, in 2002, described a 15-year-old girl with systemic brucellosis, who had a well-circumscribed intramedullary mass at T11–T12. This was drained and biopsied as she did not respond to antibiotics alone. Histopathology showed non-caseating granuloma with chronic inflammation. Cultures of the purulent material were negative. She was treated with a total of 8 weeks of antibiotics with which she recovered considerably [5].
Table 1 Cases of neurobrucellosis with intramedullary spinal cord involvement.
Case Age/Sex Risk factor Lesion Pus culture Blood culture Serology Treatment Systemic brucellosis
Cokca et al, 1994 17-year-old boy Regular consumption of cows milk, living in rural area near breeding animals Intramedullary dermoid cyst (T11-L2) Brucella abortus biotype 3 Brucella abortus biotype 3 not mentioned Surgical+Medical present
Bingol et al, 1999. 40-year-old female Raising sheep and consuming raw milk Intramedullary granuloma at T5 not done No growth positive Medical present
Novati et al, 2002. 24-year-old male Consumption of fresh goats cheese Intramedullary abscess at T3 not done Brucella melitensis positive Medical present
Helvaci et al, 2002 15-year-old girl Consumption of cheese made from raw goats milk Intramedullary abscess at T11-T12 no growth no growth positive Surgical+Medical present
Present case, 2005 40-year-old female Living near abattoir, consumption of meat and goat's milk. Intramedullary abscess at Conus Brucella melitensis Brucella melitensis positive Surgical+Medical present
In the present case, there was direct involvement of spinal cord by the Brucella melitensis leading to a chronic progressive neurological impairment. Despite the raised pre-operative titres of antibody to brucella antigen and high prevalence of brucellosis in this region, the possibility of the cord lesion to be of Brucella origin was not considered pre-operatively, probably because of the rarity of brucella abscess in the spinal cord. The post-operative fever and headache could have been due to transient meningitis resulting from contamination of the CSF space during surgery. This, fortunately, responded well to the antibiotics.
There are no specific guidelines regarding the antibiotic regimens and duration of treatment for neurobrucellosis. The duration of treatment varies from 8 weeks to 2 year depending upon individual cases, surgical or medical line of treatment and response to the treatment. Drugs such as doxycycline, rifampicin and trimethoprim-sulfamethoxazole have been found effective due to their good CNS penetration and synergistic actions [10,14]. Tetracycline and streptomycin are good for systemic brucellosis, although their CNS penetration is poor. However, as most of these patients have systemic brucellosis as well, they should be covered with these antibiotics, especially in initial stages. In the present case, streptomycin and doxycycline was given for 1 month followed by rifampicin and doxycycline for 1 month. As the abscess was emptied of it contents completely under operating microscope, antibiotics were given for only 2 months in the post-operative period.
Conclusion
1. It is important to consider the possibility of intra-medullary abscess as a presentation of neurobrucellosis, especially in endemic region.
2. Prompt detection and neurosurgical drainage with antibiotics usually results in resolution of the infection.
3. The duration of antibiotics is variable and depends upon, a. the type of lesion, abscess or granuloma, b. whether surgically drained or not, and c. response to treatment. Steroids should be considered initially, especially if edema is demonstrated on the scan.
4. Increased awareness and early intervention could prevent the neurological disability and improve the functional outcome.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
GVV cared for the patient and did the operation. He did the literature search, and was involved in the inception of the paper. MBN was involved in microbiological analysis, literature search and drafting the manuscript. CSP was involved in microbial testing, critical review and supervising the paper.
Acknowledgements
We acknowledge Dr Anjali Vajramani for her help with manuscript preparation.
==== Refs
Al Deeb SM Yaqub BA Sharif HS Phadke JG Neurobrucellosis. Clinical characteristics, diagnosis, and outcome Neurology 1989 498 501 2927673
Cokca F Meco O Arasil E Unlu A An intramedullary dermoid cyst abscess due to Brucella abortus biotype 3 at T11-L2 spinal levels. Case report Infection 1994 5 359 360 7843817 10.1007/BF01715549
Bingol A Yucemen N Meco O Medically treated intraspinal "brucella" granuloma Surg Neurol 1999 52 570 576 10660022 10.1016/S0090-3019(99)00110-X
Novati R Vigano MG De Bona A Nocita B Finazzi R Lazzarin A Neurobrucellosis with spinal cord abscess of the dorsal tract: a case report Int J Infect Dis 2002 6 149 150 12146501 10.1016/S1201-9712(02)90079-2
Helvaci M KasIrga E Cetin N Yaprak I Intramedullary spinal cord abscess suspected of Brucella infectio Pediatr Int 2002 44 446 448 12139575 10.1046/j.1442-200X.2002.01569.x
Coerbel MJ MacMillan AP Collier L, Balows A, Sussman M Brucellosis Topley and Wilson's, Microbiology and Microbiological infections 1998 9 Arnold publications (Hooder headline group) 819 847
Farrel ID Collie JG, Marmion BP, Fraser AG, Simmons A Brucella Mackie and McCartney Practical Medical Microbiology 1996 14 Churchill Livingstone; Edinburgh 473, 478
Kuchabal DS Joglekar MD Nagalotimath SJ Brucellosis (A Cause of Peripheral Neuritis) Bull Ind Med Asso Bombay Branch 1974 3 11 463
Joglekar MD Nagalotimath SJ Neurobrucellosis J Asso Phys Ind 1976 24 183 186
Bashir R Zuheir Al-Kawi M Harder EJ Jinkins J Nervous system brucellosis: Diagnosis and treatment Neurology 1985 35 1576 81 3877254
Shakir RA Al-Din AS Araj GF Lulu AR Mousa AR Saadah MA Clinical categories of Neurobrucellosis. A report on 19 cases Brain 1987 110 213 223 3801851
Nagalothimath SJ Joglekar MD Prevalence of brucellosis in Belgaum Bull Ind Med Asso Bombay Branch 1974 3 457
Spink WW Most biological and clinical problems are related to intracellular parasitism in Brucellosis N Eng J Med 1952 247 603 610
Perez MAH Rodriguez BA Garcia AF Diez-Tejedor E Tella PB Treatment of nervous system brucellosis with rifampicin and doxycycline Neurology 1986 36 28 31
|
16168059
|
PMC1242218
|
CC BY
|
2021-01-04 16:38:20
|
no
|
Ann Clin Microbiol Antimicrob. 2005 Sep 16; 4:14
|
utf-8
|
Ann Clin Microbiol Antimicrob
| 2,005 |
10.1186/1476-0711-4-14
|
oa_comm
|
==== Front
BMC CancerBMC Cancer1471-2407BioMed Central London 1471-2407-5-1141615938410.1186/1471-2407-5-114Research ArticleAnthracyclines, proteasome activity and multi-drug-resistance Fekete Mirela R [email protected] William H [email protected] Frank [email protected] Department of Neurology, Bürgerhospital, Tunzhofer Str. 14-16, 70191 Stuttgart, Germany2 Department of Radiation Oncology, Roy E. Coats Labs., David Geffen School of Medicine at UCLA, 10833 Le Conte Avenue, Los Angeles, CA90095-1714, USA2005 13 9 2005 5 114 114 24 5 2005 13 9 2005 Copyright © 2005 Fekete et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
P-glycoprotein is responsible for the ATP-dependent export of certain structurally unrelated compounds including many chemotherapeutic drugs. Amplification of P-glycoprotein activity can result in multi-drug resistance and is a common cause of chemotherapy treatment failure. Therefore, there is an ongoing search for inhibitors of P-glycoprotein. Observations that cyclosporin A, and certain other substances, inhibit both the proteasome and P-glycoprotein led us to investigate whether anthracyclines, well known substrates of P-gp, also inhibit the function of the proteasome.
Methods
Proteasome function was measured in cell lysates from ECV304 cells incubated with different doses of verapamil, doxorubicin, daunorubicin, idarubicin, epirubicin, topotecan, mitomycin C, and gemcitabine using a fluorogenic peptide assay. Proteasome function in living cells was monitored using ECV304 cells stably transfected with the gene for an ubiquitin/green fluorescent protein fusion protein. The ability of the proteasome inhibitor MG-132 to affect P-glycoprotein function was monitored by fluorescence due to accumulation of daunorubicin in P-glycoprotein overexpressing KB 8-5 cells.
Results
Verapamil, daunorubicin, doxorubicin, idarubicin, and epirubicin inhibited 26S chymotrypsin-like function in ECV304 extracts in a dose-dependent fashion. With the exception of daunorubicin, 20S proteasome function was also suppressed. The proteasome inhibitor MG-132 caused a dose-dependent accumulation of daunorubicin in KB 8-5 cells that overexpress P-glycoprotein, suggesting that it blocked P-glycoprotein function.
Conclusion
Our data indicate that anthracyclines inhibit the 26S proteasome as well as P-glycoprotein. Use of inhibitors of either pathway in cancer therapy should take this into consideration and perhaps use it to advantage, for example during chemosensitization by proteasome inhibitors.
==== Body
Background
Multi-drug-resistance (MDR) is a common reason for chemotherapy treatment failure in breast cancer, leukemia, and non-Hodgkin lymphoma patients. MDR can often be attributed to over-expression of the mdr1 gene that codes for an ATP-dependent, transmembrane P-glycoprotein (P-gp) efflux pump pathway, which rapidly exports man structurally un-related drugs from the cell, including anthracyclines [1,2].
Numerous pre-clinical and clinical studies using P-gp modulating compounds like verapamil, cyclosporin A, reserpine, staurosporine, propafenone, phenoxazine, chloroquine, phenothiazine and their derivates have been undertaken to overcome MDR and several substances have been identified that are effective in vitro (reviewed in [3]). However, to revert MDR in vivo, most MDR-modulating drugs require serum concentrations that have unacceptable toxicity and therefore they are currently not used in standard chemotherapy regimens. The development of better, less toxic inhibitors might be aided by insights into the specificity of these inhibitors for other molecules and the spectrum of molecules bound by P-glycoprotein.
Two of the most commonly used MDR-modulating substances are verapamil and cyclosporin A (CsA), or their derivates. Interestingly, CsA has recently been identified as an inhibitor of the 26S proteasome [4]. The 26S proteasome is a highly conserved multicatalytic protease responsible for ATP- and ubiquitin-dependent degradation of all short-lived and 70–90% of all long lived proteins including cyclin A, B and E, p21 and p27, p53, cJun, cFos, and IκB. As such, the 26S proteasome controls cell cycle, signal transduction pathways, apoptosis and major functions of the immune system. Indeed some of the immunosuppressive properties of CsA, such as decreases in the expression of MHC-I molecules on the surface of target cells [5] and apoptotic death of lymphocytes through inhibition of the transcription factor NF-κB [6], may be due to its inhibitory effect on proteasome function. Vinblastine, a known P-gp substrate has also been shown to inhibit proteasome activity [7]. And, remarkably, the HIV protease inhibitor ritonavir was identified as an inhibitor of P-gp [8] and the proteasome [9]. Since CsA and ritonavir have been shown to inhibit both proteasome and P-gp activities, we questioned whether there was cross specificity between P-gp and proteasome activities. Cross specificity might explain effects of P-gp inhibitors on multiple cellular parameters that seem extrinsic to a pumping function of P-gp. Insights into substrate cross specificity of P-gp could offer a basis for the development of more selective P-gp inhibitors. They could also indicate reasons for the toxicity of these inhibitors, and why they affect cellular functions other than those related to P-gp.
Using an in vitro model, we show that anthracyclines and verapamil both inhibit proteasome function. Additionally, we demonstrate that the proteasome inhibitor MG-132 inhibits P-gp function, thereby increasing the uptake of doxorubicin in the cytoplasm and the nucleus.
Methods
Cell culture
KB 8.5 human epitheloid carcinoma cells that overexpress P-gp were a generous gift from Dr. Peter Hafkemeyer (University Clinic Freiburg, Germany). Every 21 days P-gp-positive KB 8.5 cells were selected by addition of colchicine (10 ng/ml, Sigma). 24 hours before drug treatment cells were plated into 6-well plates (Costar) at a density of 106 cells/well.
EVC 304 human bladder carcinoma cells and PC-3 prostate cancer cells were obtained from the German Microorganism and Tissue Culture Collection (German collection of microorganism and cell cultures, DSMZ, Braunschweig). Cells were grown in 75 cm2 flasks (Falcon) at 37°C in a humidified atmosphere at 5 % CO2 in DMEM medium (Sigma) supplemented with 10 % heat inactivated FCS (Sigma) and 1 % penicillin/streptomycin (Gibco BRL).
Drug treatment
Stock solutions of all cytotoxic drugs were obtained from the hospital pharmacy of the University Clinic Freiburg. MG-132 (Calbiochem) was dissolved at 10 mM in DMSO and stored as small aliquots (10–30 μl) at -20°C. In drug accumulation assays doxorubincin (10 μM), daunorubicin (2–16 μM) or MG-132 (0.5–50 μM, 0.5% DMSO) were added to cells at the indicated times. Control cells were subjected to DMSO treatment alone (0.5 %).
Proteasome function assays
20S and 26S proteasome function was measured as described previously (20). Briefly, cells were washed with PBS, then with buffer I (50 mM Tris, pH 7.4, 2 mM DTT, 5 mM MgCl2, 2 mM ATP), and pelleted by centrifugation. Glass beads and homogenization buffer (50 mM Tris, pH 7.4, 1 mM DTT, 5 mM MgCl2, 2 mM ATP, 250 mM sucrose) were added and vortexed for 1 minute. Beads and cell debris were removed by centrifugation at 1,000 × g for 5 minutes and 10,000 × g for 20 minutes. Protein concentration was determined by the BCA protocol (Pierce). One hundred μg protein of each sample was diluted with buffer I to a final volume of 1000 μl and the fluorogenic proteasome substrate SucLLVY-7-amido-4-methylcoumarin (chymotrypsin-like, Sigma) was added in a final concentration of 80 μM in 1% DMSO. To access 20S function, buffer I was replaced by an ATP-free buffer containing SDS (20 mM HEPES, pH 7.8; 0.5 mM EDTA, 0.03% SDS) [10]. Cleavage activity was monitored continuously by detection of free 7-amido-4-methylcoumarin using a fluorescence plate reader (Gemini, Molecular Devices) at 380/460 nm and 37°C. As controls for drug studies, 7-amido-4-methylcoumarin (AMC, 2 μM) was incubated with drugs in buffer I without cell extracts and measurements of proteasome function were corrected when necessary.
Drug accumulation assay
Total cellular daunorubicin content and accumulation of doxorubicin in the cytoplasm and nucleus were determined as described elsewhere [11] with some minor modifications. Growth medium on cells was replaced by PBS for 40 minutes at 37°C. This was replaced by fresh PBS containing daunorubicin or doxorubicin and MG-132 or anthracyclines alone. In some experiments, cells were washed with PBS after daunorubicin treatment and incubated in PBS containing MG-132 for an additional 40 minutes at 37°C. After drug treatment, the cells were washed twice with PBS, re-suspended in either 4 ml lysis buffer (0.3 M sucrose, 0.05 mM EGTA pH 8.0, 60 mM KCl, 15 mM NaCl, 15 mM HEPES pH 7.5, 150 μM spermine, 50 μM spermidine) containing 20 μl triton X-100 for nuclear isolation or 400 μl of 50% ethanol in 1 M HCl (v/v) for whole cell lysis. For the latter, cells were vortexed and diluted with water to a final volume of 1.4 ml. The cells in lysis buffer were mixed and left on ice for 15 minutes before centrifuging. The nuclei (pellet) were then vortexed with 400 μl HCl/isopropanol. Fluorescence derived from daunorubicin or doxorubicin was measured in quadruplicates of 200 μl using a fluorescence plate reader (Gemini, Molecular Devices) at 480/575 nm.
Transfection
ECV304 cells were plated at a density of 250.000 cells/well into six-well plates twelve hours before transfection. Cells were transfected with 5 μg of a plasmid (pEGFP-N1, Clontech) coding for an ubiquitin (Ub)-R-GFP fusion protein under control of a CMV promoter [12] (a kind gift from Dr. M. Masucci, Karolinska Institute, Sweden) using the Superfect transfection kit (Qiagen) and following the manufacturer's instructions. Transfected cells were maintained in DMEM (10 % FSC, 1 % penicillin/streptomycin) supplemented with 500 μg/ml G418 (Sigma) and clones were obtained. Expression of Ub-R-GFP was analyzed by flow cytometry (FL1-H, FACSCalibur, Becton Dickinson) using CellQuest Software before and after treatment with the proteasome inhibitor MG-132 (50 μM) for 10 hours at 37°C. Clone #10 (ECV304/10), which showed low background and high MG-132-induced expression of Ub-R-GFP, was used for inhibition experiments.
Statistics
Experimental data are presented as mean ± standard error of the mean from at least three independent experiments. A p-value <0.05 in a two-sided student's t-test was considered as 'statistically significant'.
Results
Verapamil is an inhibitor of 20S and 26S proteasome function
In order to test the hypothesis that the P-gp inhibitor verapamil inhibits proteasome function, proteasome extracts of ECV304 and PC-3 cells were incubated with different concentrations of the drug (0, 50, 100 and 200 μM) and immediately tested for their chymotrypsin-like activity against the fluorogenic substrate SucLLVY-7-amido-4-methylcoumarin. There was a dose-dependent inhibition of MG-132-sensitive 26S (Fig. 1A) and 20S (data not shown) proteasome function, consistent with a direct inhibitory effect of verapamil on the proteasome.
Anthracyclines inhibit 20S and 26S proteasome function in a dose-dependent manner
Since verapamil, vinblastine, and CsA have been found to inhibit 20S and 26S proteasome function and vinblastine and CsA serve as substrates of P-gp [3], we asked if anthracyclines in general have an inhibitory effect on this protease. When crude extracts of ECV304 cells were incubated with different doses (0 – 100 μM) of the anthracyclines doxorubicin, daunorubicin, idarubicin and epirubicin we observed dose-dependent inhibition of 26S proteasome function with IC50 values of 65.5 μM for doxorubicin, 13.7 μM for daunorubicin, 38.6 μM for idarubicin and 29.2 μM for epirubicin (Table 1). Topotecan, mitomycin C, and gemcitabine had no measurable effect on 26S proteasome function (data not shown). 20S proteasome function was inhibited by doxorubicin (IC50 5.8 μM), idarubicin (IC50 92 μM), epirubicin (IC50 12.5 μM) but not by daunorubicin (Table 2).
In order to demonstrate if this inhibition could be observed in living cells, we incubated ECV304/10 cells, stably transfected with an expression plasmid for an Ub-GFP fusion protein with doxorubicin (100 μM) for 12 hours. When analyzed by fluorescence microscopy, the cells showed perinuclear accumulation of doxorubicin while GFP accumulated throughout the cytoplasm, indicating inhibition of proteasome function (Fig. 2).
MG-132 treatment reverts multi-drug-resistance in P-gp expressing KB 8-5 cells
The human epitheloid carcinoma cell line KB 8-5 is a well-characterized tumor cell line that over-expresses mdr-1 with associated MDR. Preliminary experiments showed that treatment of KB 8.5 cells with the reversible proteasome inhibitor MG-132 (3.125 to 50 μM) induced apoptosis within 24 hours. This is in accord with numerous studies reporting induction of apoptosis in cancer cells by proteasome inhibitors [13], and indicated that MG-132 enters KB 8-5 cells and that they are not abnormally resistant to its effects based on enhanced P-gp function. After 45 minutes of incubation with MG-132 (50 μM), no morphological signs of toxicity were observed. KB 8-5 cells treated with different doses of MG-132 and daunorubicin (10 μM) for 45 minutes showed increased, dose-dependent accumulation of daunorubicin in the cytoplasm (e.g. a 4-fold increase at 50 μM MG-132, Fig. 3) indicating that MG-132 could block P-gp function. This was further supported by the observation that incubation of ECV304 cell with MG-132 (25 μM) caused an increased uptake of doxorubicin in the cytoplasm and in the nuclear fraction of the cells (Fig. 4).
Discussion
The observations that CsA [4] and vinblastine [7] have inhibitory effects on the cleavage activity of the 26S proteasome led us investigate the effects of anthracycline anticancer agents and verapamil on the activity of this protease. Verapamil caused a concentration-dependent inhibition of 20S and 26S function. Additionally, we found a concentration-dependent inhibition of 26S proteasome function for all four anthracyclines tested. Comparable results showing doxorubicin to be a non-competitive inhibitor of the proteasome have been reported previously [14]. With the exception of daunorubicin, anthracyclines also inhibited 20S chymotryptic function in a dose-dependent manner. It is known that doxorubicin is co-transported into the nucleus along with proteasomes [15,16] but our observation of a general direct inhibitory effect of anthracycline anticancer agents on the proteasome sheds a totally new light on the actions of these drugs.
The inhibitory effects of the reversible inhibitor of the proteasome, MG-132, on P-glycoprotein function, supports the view that P-glycoprotein and the proteasome can both be targeted by this new class of chemotherapeutic drugs. This was further supported by the observation that verapamil, another established inhibitor of P-gp, inhibited the chymotryptic 20S and 26S function of the proteasome. The fact that both P-glycoprotein and proteasome activities can both be regulated by pro-inflammatory cytokines and oxidative stress suggests [17-21] that studies on co-ordinate regulation of these activities might be illuminating.
These findings lead to interesting possibilities with respect to the possible use of proteasome inhibitors, which are just entering their first clinical trials [22,23], in combination therapy, as well as to the mechanism of action and toxicity of P-gp inhibitors. Using an in vitro system, we showed that the proteasome inhibitor MG-132 caused intracellular accumulation of anthracyclines, indicating inhibition of P-gp function. Proteasome inhibitors may interfere with drug-resistance at additional levels as P-gp and also topisomerase II are degraded in a proteasome-dependent manner and degradation is blocked by proteasome inhibitors [24,25]. However, given the long half-life of P-gp of 14–24 hours [26], the effects observed in our study after short-time incubation of the cells with MG-132 are probably not caused by an increased degradation of P-gp. The extent of the increase of anthracyclin accumulation in mdr1-overexpressing KB-8.5 treated with 25 μM concentrations of MG-132 cells in our study was comparable to the effect of verapamil at 50 μM [27]. Future studies have to clarify if similar effects can be obtained using clinically used proteasome inhibitors at concentrations typically reached in the serum of patients.
Tumor cells in general exhibit altered patterns of expression of proteasome subunits and their distribution between cytoplasm and nucleus often differs from normal cells [28-30]. This may explain why specific proteasome inhibitors like PS-341 are usually clinically well tolerated. Inhibition of proteasome function induces apoptosis of tumor cells [31-34] and sensitizes the surviving tumor cells to the actions of both chemotherapy [35] and radiation therapy [36,37]. Therefore, proteasome inhibitors might overcome P-gp-related MDR, with accompanying chemo- and radiosensitizing effects. Also, since tumor microvasculature expresses high levels of mdr-1 [38,39], the possibility exists that the neovasculature is a target for these drugs in vivo. On the other hand, direct inhibition of proteasome function might be an additional major mechanism of action for anthracyclines. Such inhibition could contribute to their ability to enhance the efficacy of other chemotherapeutic drugs, independent of their ability to reverse MDR.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MF carried out the molecular studies and helped to draft the manuscript. WMB participated in the design of the study and helped to draft the manuscript. FP designed the study and drafted the manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This investigation was supported in part by Grants of the German Research Foundation (DFG) Pa 723/1 Pa 723/3 (FP) and PHS Grant number CA-87887 awarded to WMc by the National Cancer Institute;
Figures and Tables
Figure 1 Verapamil is an inhibitor of 26S proteasome function. Incubation of crude extracts of ECV304 cells containing proteasomes with different doses of verapamil (50, 60, 80, 100, 200 μM) inhibited proteolysis of the chymotrypsin-like substrate SucLLVY-AMC in a dose-dependent manner, indicating inhibition of 26S proteasome function.
Figure 2 Anthracyclines are inhibitors of proteasome function. Incubation of ECV304 cells stably transfected with an Ub-GFP fusion protein with daunorubicin (100 μM, 16 h), caused accumulation of GFP throughout the cytoplasm (lower picture), indicating proteasome inhibition in living cells while untreated controls cells showed only little accumulation of GFP (A/B). Daunorubicin accumulated in the perinuclear region (C).
Figure 3 MG-132 treatment of KB 8-5 causes intracellular accumulation of anthracyclines. Incubation of KB 8-5 cells, which overexpress P-gp, with increasing doses of MG-132 (0, 6.25, 12.5, 25, 50 μM) caused a dose-dependent accumulation of daunorubicin, as measured by fluorescence, indicating inhibition of P-gp function by this proteasome inhibitor.
Figure 4 Accumulation of doxorubicin in the presence or absence of MG-132 (25 μM) in the cytoplasm and the nuclear fraction of ECV304 cells.
Table 1 Chymotryptic 26S Proteasome Activity in Lysates from ECV 304 Cells
μM Doxorubincin Daunorubicin Epirubicin Idarubicin
0 1 1 1 1
6.25 0.92 ± 0.03* 0.85 ± 0.05* 0.79 ± 0.1 n.s. 0.97 ± 0.03 n.s.
12.5 0.84 ± 0.05* 0.53 ± 0.11* 0.57 ± 0.31 n.s. 0.96 ± 0.02 n.s.
25 0.72 ± 0.04** 0.22 ± 0.09** 0.59 ± 0.14* 0.79 ± 0.12 n.s.
50 0.59 ± 0.1* 0.12 ± 0.06** 0.36 ± 0.2* 0.28 ± 0.09 **
100 0.36 ± 0.03*** 0.11 ± 0.06** 0.27 ± 0.17* 0.1 ± 0.06 **
n.s. not significant, *p < 0.05, **p < 0.01, ***p < 0.001 (two-sided student's t-test)
Table 2 Chymotryptic 20S Proteasome Activity in Lysates from ECV 304 Cells
μM Doxorubincin Daunorubicin Epirubicin Idarubicin
0 1 1 1 1
6.25 0.49 ± 0.1* 0.62 ± 0.19 n.s. 0.71 ± 0.21 n.s. 0.67 ± 0.07*
12.5 0.28 ± 0.09** 0.64 ± 0.07 n.s. 0.46 ± 0.2 n.s. 0.67 ± 0.08*
25 0.28 ± 0.13* 0.81 ± 0.05 n.s. 0.37 ± 0.08** 0.69 ± 0.05**
50 0.24 ± 0.06** 0.95 ± 0.04 n.s. 0.33 ± 0.07** 0.67 ± 0.05 **
100 0.25 ± 0.07** 1.05 ± 0.1 n.s. 0.35 ± 0.07** 0.49 ± 0.11*
n.s. not significant, *p < 0.05, **p < 0.01, ***p < 0.001 (two-sided student's t-test)
==== Refs
Gottesman MM Fojo T Bates SE Multidrug resistance in cancer: role of ATP-dependent transporters Nat Rev Cancer 2002 2 48 58 11902585 10.1038/nrc706
Lacayo NJ Duran GE Sikic BI Modulation of resistance to idarubicin by the cyclosporin PSC 833 (valspodar) in multidrug-resistant cells J Exp Ther Oncol 2003 3 127 135 14641819 10.1046/j.1359-4117.2003.01088.x
Ambudkar SV Dey S Hrycyna CA Ramachandra M Pastan I Gottesman MM Biochemical, cellular, and pharmacological aspects of the multidrug transporter Annu Rev Pharmacol Toxicol 1999 39 361 398 10331089 10.1146/annurev.pharmtox.39.1.361
Meyer S Kohler NG Joly A Cyclosporine A is an uncompetitive inhibitor of proteasome activity and prevents NF-kappa B activation FEBS Lett 1997 413 354 358 9280312 10.1016/S0014-5793(97)00930-7
Harding CV France J Song R Farah JM Chatterjee S Iqbal M Siman R Novel dipeptide aldehydes are proteasome inhibitors and block the MHC-I antigen-processing pathway J Immunol 1995 155 1767 1775 7636233
Marienfeld R Neumann M Chuvpilo S Escher C Kneitz B Avots A Schimpl A Serfling E Cyclosporin A interferes with the inducible degradation of NF-kappa B inhibitors, but not with the processing of p105/NF-kappa B1 in T cells Eur J Immunol 1997 27 1601 1609 9247567
Piccinini M Tazartes O Mezzatesta C Ricotti E Bedino S Grosso F Dianzani U Tovo PA Mostert M Musso A Rinaudo MT Proteasomes are a target of the anti-tumor drug vinblastine Biochem J 2001 356 835 841 11389692 10.1042/0264-6021:3560835
Drewe J Gutmann H Fricker G Torok M Beglinger C Huwyler J HIV protease inhibitor ritonavir: a more potent inhibitor of P- glycoprotein than the cyclosporine analog SDZ PSC 833 Biochem Pharmacol 1999 57 1147 1152 11230802 10.1016/S0006-2952(99)00026-X
Schmidtke G Holzhutter HG Bogyo M Kairies N Groll M de Giuli R Emch S Groettrup M How an inhibitor of the HIV-I protease modulates proteasome activity J Biol Chem 1999 274 35734 35740 10585454 10.1074/jbc.274.50.35734
Stein RL Melandri F Dick L Kinetic characterization of the chymotryptic activity of the 20S proteasome Biochemistry 1996 35 3899 3908 8672420 10.1021/bi952262x
Gieseler F Nussler V Brieden T Kunze J Valsamas S Intracellular pharmacokinetics of anthracyclines in human leukemia cells: correlation of DNA-binding with apoptotic cell death Int J Clin Pharmacol Ther 1998 36 25 28 9476145
Dantuma NP Lindsten K Glas R Jellne M Masucci MG Short-lived green fluorescent proteins for quantifying ubiquitin/proteasome-dependent proteolysis in living cells Nat Biotechnol 2000 18 538 543 10802622 10.1038/75406
Shinohara K Tomioka M Nakano H Tone S Ito H Kawashima S Apoptosis induction resulting from proteasome inhibition Biochem J 1996 317 385 388 8713062
Kiyomiya K Satoh J Horie H Kurebe M Nakagawa H Matsuo S Correlation between nuclear action of anthracycline anticancer agents and their binding affinity to the proteasome Int J Oncol 2002 21 1081 1085 12370758
Kiyomiya K Matsuo S Kurebe M Mechanism of specific nuclear transport of adriamycin: the mode of nuclear translocation of adriamycin-proteasome complex Cancer Res 2001 61 2467 2471 11289116
Kiyomiya K Matsuo S Kurebe M Proteasome is a carrier to translocate doxorubicin from cytoplasm into nucleus Life Sci 1998 62 1853 1860 9600327 10.1016/S0024-3205(98)00151-9
Sukhai M Piquette-Miller M Regulation of the multidrug resistance genes by stress signals J Pharm Pharm Sci 2000 3 268 280 10994039
Stohwasser R Kloetzel PM Cytokine induced changes in proteasome subunit composition are concentration dependent Biol Chem 1996 377 571 577 9067255
Yang Y Waters JB Fruh K Peterson PA Proteasomes are regulated by interferon gamma: implications for antigen processing Proc Natl Acad Sci U S A 1992 89 4928 4932 1594596
Pajonk F McBride WH Ionizing radiation affects 26s proteasome function and associated molecular responses, even at low doses Radiother Oncol 2001 59 203 212 11325450 10.1016/S0167-8140(01)00311-5
Grune T Oxidative stress, aging and the proteasomal system Biogerontology 2000 1 31 40 11707918 10.1023/A:1010037908060
Teicher BA Ara G Herbst R Palombella VJ Adams J The proteasome inhibitor PS-341 in cancer therapy Clin Cancer Res 1999 5 2638 2645 10499643
Adams J Palombella VJ Elliott PJ Proteasome inhibition: a new strategy in cancer treatment Invest New Drugs 2000 18 109 121 10857991 10.1023/A:1006321828515
Zhang Z Wu JY Hait WN Yang JM Regulation of the stability of P-glycoprotein by ubiquitination Mol Pharmacol 2004 66 395 403 15322230 10.1124/mol.104.001966
Ogiso Y Tomida A Lei S Omura S Tsuruo T Proteasome inhibition circumvents solid tumor resistance to topoisomerase II-directed drugs Cancer Res 2000 60 2429 2434 10811120
Yoshimura A Kuwazuru Y Sumizawa T Ikeda S Ichikawa M Usagawa T Akiyama S Biosynthesis, processing and half-life of P-glycoprotein in a human multidrug-resistant KB cell Biochim Biophys Acta 1989 992 307 314 2570611
Ghauharali RI Westerhoff HV Dekker H Lankelma J Saturable P-glycoprotein kinetics assayed by fluorescence studies of drug efflux from suspended human KB8-5 cells Biochim Biophys Acta 1996 1278 213 222 8593279
Amsterdam A Pitzer F Baumeister W Changes in intracellular localization of proteasomes in immortalized ovarian granulosa cells during mitosis associated with a role in cell cycle control Proc Natl Acad Sci U S A 1993 90 99 103 8380501
Kumatori A Tanaka K Inamura N Sone S Ogura T Matsumoto T Tachikawa T Shin S Ichihara A Abnormally high expression of proteasomes in human leukemic cells Proc Natl Acad Sci U S A 1990 87 7071 7075 2205851
Kanayama H Tanaka K Aki M Kagawa S Miyaji H Satoh M Okada F Sato S Shimbara N Ichihara A Changes in expressions of proteasome and ubiquitin genes in human renal cancer cells Cancer Res 1991 51 6677 6685 1660345
Qiu JH Asai A Chi S Saito N Hamada H Kirino T Proteasome inhibitors induce cytochrome c-caspase-3-like protease- mediated apoptosis in cultured cortical neurons J Neurosci 2000 20 259 265 10627603
Pasquini LA Besio Moreno M Adamo AM Pasquini JM Soto EF Lactacystin, a specific inhibitor of the proteasome, induces apoptosis and activates caspase-3 in cultured cerebellar granule cells J Neurosci Res 2000 59 601 611 10686588 10.1002/(SICI)1097-4547(20000301)59:5<601::AID-JNR3>3.0.CO;2-1
Zhang XM Lin H Chen C Chen BD Inhibition of ubiquitin-proteasome pathway activates a caspase-3-like protease and induces Bcl-2 cleavage in human M-07e leukaemic cells Biochem J 1999 340 127 133 10229667 10.1042/0264-6021:3400127
Delic J Masdehors P Omura S Cosset JM Dumont J Binet JL Magdelenat H The proteasome inhibitor lactacystin induces apoptosis and sensitizes chemo- and radioresistant human chronic lymphocytic leukaemia lymphocytes to TNF-alpha-initiated apoptosis [see comments] Br J Cancer 1998 77 1103 1107 9569046
Bold RJ Virudachalam S McConkey DJ Chemosensitization of pancreatic cancer by inhibition of the 26S proteasome J Surg Res 2001 100 11 17 11516199 10.1006/jsre.2001.6194
Pajonk F Pajonk K McBride W Apoptosis and radiosensitization of Hodgkin’s cells by proteasome inhibition Int J Radiat Oncol Biol Phys 2000 47 1025 1032 10863075 10.1016/S0360-3016(00)00516-2
Russo SM Tepper JE Baldwin ASJ Liu R Adams J Elliott P Cusack JCJ Enhancement of radiosensitivity by proteasome inhibition: implications for a role of NF-kappaB Int J Radiat Oncol Biol Phys 2001 50 183 193 11316563 10.1016/S0360-3016(01)01446-8
Iwahana M Utoguchi N Mayumi T Goryo M Okada K Drug resistance and P-glycoprotein expression in endothelial cells of newly formed capillaries induced by tumors Anticancer Res 1998 18 2977 2980 9713495
Toth K Vaughan MM Peress NS Slocum HK Rustum YM MDR1 P-glycoprotein is expressed by endothelial cells of newly formed capillaries in human gliomas but is not expressed in the neovasculature of other primary tumors Am J Pathol 1996 149 853 858 8780389
|
16159384
|
PMC1242219
|
CC BY
|
2021-01-04 16:03:03
|
no
|
BMC Cancer. 2005 Sep 13; 5:114
|
utf-8
|
BMC Cancer
| 2,005 |
10.1186/1471-2407-5-114
|
oa_comm
|
==== Front
BMC CancerBMC Cancer1471-2407BioMed Central London 1471-2407-5-1141615938410.1186/1471-2407-5-114Research ArticleAnthracyclines, proteasome activity and multi-drug-resistance Fekete Mirela R [email protected] William H [email protected] Frank [email protected] Department of Neurology, Bürgerhospital, Tunzhofer Str. 14-16, 70191 Stuttgart, Germany2 Department of Radiation Oncology, Roy E. Coats Labs., David Geffen School of Medicine at UCLA, 10833 Le Conte Avenue, Los Angeles, CA90095-1714, USA2005 13 9 2005 5 114 114 24 5 2005 13 9 2005 Copyright © 2005 Fekete et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
P-glycoprotein is responsible for the ATP-dependent export of certain structurally unrelated compounds including many chemotherapeutic drugs. Amplification of P-glycoprotein activity can result in multi-drug resistance and is a common cause of chemotherapy treatment failure. Therefore, there is an ongoing search for inhibitors of P-glycoprotein. Observations that cyclosporin A, and certain other substances, inhibit both the proteasome and P-glycoprotein led us to investigate whether anthracyclines, well known substrates of P-gp, also inhibit the function of the proteasome.
Methods
Proteasome function was measured in cell lysates from ECV304 cells incubated with different doses of verapamil, doxorubicin, daunorubicin, idarubicin, epirubicin, topotecan, mitomycin C, and gemcitabine using a fluorogenic peptide assay. Proteasome function in living cells was monitored using ECV304 cells stably transfected with the gene for an ubiquitin/green fluorescent protein fusion protein. The ability of the proteasome inhibitor MG-132 to affect P-glycoprotein function was monitored by fluorescence due to accumulation of daunorubicin in P-glycoprotein overexpressing KB 8-5 cells.
Results
Verapamil, daunorubicin, doxorubicin, idarubicin, and epirubicin inhibited 26S chymotrypsin-like function in ECV304 extracts in a dose-dependent fashion. With the exception of daunorubicin, 20S proteasome function was also suppressed. The proteasome inhibitor MG-132 caused a dose-dependent accumulation of daunorubicin in KB 8-5 cells that overexpress P-glycoprotein, suggesting that it blocked P-glycoprotein function.
Conclusion
Our data indicate that anthracyclines inhibit the 26S proteasome as well as P-glycoprotein. Use of inhibitors of either pathway in cancer therapy should take this into consideration and perhaps use it to advantage, for example during chemosensitization by proteasome inhibitors.
==== Body
Background
Multi-drug-resistance (MDR) is a common reason for chemotherapy treatment failure in breast cancer, leukemia, and non-Hodgkin lymphoma patients. MDR can often be attributed to over-expression of the mdr1 gene that codes for an ATP-dependent, transmembrane P-glycoprotein (P-gp) efflux pump pathway, which rapidly exports man structurally un-related drugs from the cell, including anthracyclines [1,2].
Numerous pre-clinical and clinical studies using P-gp modulating compounds like verapamil, cyclosporin A, reserpine, staurosporine, propafenone, phenoxazine, chloroquine, phenothiazine and their derivates have been undertaken to overcome MDR and several substances have been identified that are effective in vitro (reviewed in [3]). However, to revert MDR in vivo, most MDR-modulating drugs require serum concentrations that have unacceptable toxicity and therefore they are currently not used in standard chemotherapy regimens. The development of better, less toxic inhibitors might be aided by insights into the specificity of these inhibitors for other molecules and the spectrum of molecules bound by P-glycoprotein.
Two of the most commonly used MDR-modulating substances are verapamil and cyclosporin A (CsA), or their derivates. Interestingly, CsA has recently been identified as an inhibitor of the 26S proteasome [4]. The 26S proteasome is a highly conserved multicatalytic protease responsible for ATP- and ubiquitin-dependent degradation of all short-lived and 70–90% of all long lived proteins including cyclin A, B and E, p21 and p27, p53, cJun, cFos, and IκB. As such, the 26S proteasome controls cell cycle, signal transduction pathways, apoptosis and major functions of the immune system. Indeed some of the immunosuppressive properties of CsA, such as decreases in the expression of MHC-I molecules on the surface of target cells [5] and apoptotic death of lymphocytes through inhibition of the transcription factor NF-κB [6], may be due to its inhibitory effect on proteasome function. Vinblastine, a known P-gp substrate has also been shown to inhibit proteasome activity [7]. And, remarkably, the HIV protease inhibitor ritonavir was identified as an inhibitor of P-gp [8] and the proteasome [9]. Since CsA and ritonavir have been shown to inhibit both proteasome and P-gp activities, we questioned whether there was cross specificity between P-gp and proteasome activities. Cross specificity might explain effects of P-gp inhibitors on multiple cellular parameters that seem extrinsic to a pumping function of P-gp. Insights into substrate cross specificity of P-gp could offer a basis for the development of more selective P-gp inhibitors. They could also indicate reasons for the toxicity of these inhibitors, and why they affect cellular functions other than those related to P-gp.
Using an in vitro model, we show that anthracyclines and verapamil both inhibit proteasome function. Additionally, we demonstrate that the proteasome inhibitor MG-132 inhibits P-gp function, thereby increasing the uptake of doxorubicin in the cytoplasm and the nucleus.
Methods
Cell culture
KB 8.5 human epitheloid carcinoma cells that overexpress P-gp were a generous gift from Dr. Peter Hafkemeyer (University Clinic Freiburg, Germany). Every 21 days P-gp-positive KB 8.5 cells were selected by addition of colchicine (10 ng/ml, Sigma). 24 hours before drug treatment cells were plated into 6-well plates (Costar) at a density of 106 cells/well.
EVC 304 human bladder carcinoma cells and PC-3 prostate cancer cells were obtained from the German Microorganism and Tissue Culture Collection (German collection of microorganism and cell cultures, DSMZ, Braunschweig). Cells were grown in 75 cm2 flasks (Falcon) at 37°C in a humidified atmosphere at 5 % CO2 in DMEM medium (Sigma) supplemented with 10 % heat inactivated FCS (Sigma) and 1 % penicillin/streptomycin (Gibco BRL).
Drug treatment
Stock solutions of all cytotoxic drugs were obtained from the hospital pharmacy of the University Clinic Freiburg. MG-132 (Calbiochem) was dissolved at 10 mM in DMSO and stored as small aliquots (10–30 μl) at -20°C. In drug accumulation assays doxorubincin (10 μM), daunorubicin (2–16 μM) or MG-132 (0.5–50 μM, 0.5% DMSO) were added to cells at the indicated times. Control cells were subjected to DMSO treatment alone (0.5 %).
Proteasome function assays
20S and 26S proteasome function was measured as described previously (20). Briefly, cells were washed with PBS, then with buffer I (50 mM Tris, pH 7.4, 2 mM DTT, 5 mM MgCl2, 2 mM ATP), and pelleted by centrifugation. Glass beads and homogenization buffer (50 mM Tris, pH 7.4, 1 mM DTT, 5 mM MgCl2, 2 mM ATP, 250 mM sucrose) were added and vortexed for 1 minute. Beads and cell debris were removed by centrifugation at 1,000 × g for 5 minutes and 10,000 × g for 20 minutes. Protein concentration was determined by the BCA protocol (Pierce). One hundred μg protein of each sample was diluted with buffer I to a final volume of 1000 μl and the fluorogenic proteasome substrate SucLLVY-7-amido-4-methylcoumarin (chymotrypsin-like, Sigma) was added in a final concentration of 80 μM in 1% DMSO. To access 20S function, buffer I was replaced by an ATP-free buffer containing SDS (20 mM HEPES, pH 7.8; 0.5 mM EDTA, 0.03% SDS) [10]. Cleavage activity was monitored continuously by detection of free 7-amido-4-methylcoumarin using a fluorescence plate reader (Gemini, Molecular Devices) at 380/460 nm and 37°C. As controls for drug studies, 7-amido-4-methylcoumarin (AMC, 2 μM) was incubated with drugs in buffer I without cell extracts and measurements of proteasome function were corrected when necessary.
Drug accumulation assay
Total cellular daunorubicin content and accumulation of doxorubicin in the cytoplasm and nucleus were determined as described elsewhere [11] with some minor modifications. Growth medium on cells was replaced by PBS for 40 minutes at 37°C. This was replaced by fresh PBS containing daunorubicin or doxorubicin and MG-132 or anthracyclines alone. In some experiments, cells were washed with PBS after daunorubicin treatment and incubated in PBS containing MG-132 for an additional 40 minutes at 37°C. After drug treatment, the cells were washed twice with PBS, re-suspended in either 4 ml lysis buffer (0.3 M sucrose, 0.05 mM EGTA pH 8.0, 60 mM KCl, 15 mM NaCl, 15 mM HEPES pH 7.5, 150 μM spermine, 50 μM spermidine) containing 20 μl triton X-100 for nuclear isolation or 400 μl of 50% ethanol in 1 M HCl (v/v) for whole cell lysis. For the latter, cells were vortexed and diluted with water to a final volume of 1.4 ml. The cells in lysis buffer were mixed and left on ice for 15 minutes before centrifuging. The nuclei (pellet) were then vortexed with 400 μl HCl/isopropanol. Fluorescence derived from daunorubicin or doxorubicin was measured in quadruplicates of 200 μl using a fluorescence plate reader (Gemini, Molecular Devices) at 480/575 nm.
Transfection
ECV304 cells were plated at a density of 250.000 cells/well into six-well plates twelve hours before transfection. Cells were transfected with 5 μg of a plasmid (pEGFP-N1, Clontech) coding for an ubiquitin (Ub)-R-GFP fusion protein under control of a CMV promoter [12] (a kind gift from Dr. M. Masucci, Karolinska Institute, Sweden) using the Superfect transfection kit (Qiagen) and following the manufacturer's instructions. Transfected cells were maintained in DMEM (10 % FSC, 1 % penicillin/streptomycin) supplemented with 500 μg/ml G418 (Sigma) and clones were obtained. Expression of Ub-R-GFP was analyzed by flow cytometry (FL1-H, FACSCalibur, Becton Dickinson) using CellQuest Software before and after treatment with the proteasome inhibitor MG-132 (50 μM) for 10 hours at 37°C. Clone #10 (ECV304/10), which showed low background and high MG-132-induced expression of Ub-R-GFP, was used for inhibition experiments.
Statistics
Experimental data are presented as mean ± standard error of the mean from at least three independent experiments. A p-value <0.05 in a two-sided student's t-test was considered as 'statistically significant'.
Results
Verapamil is an inhibitor of 20S and 26S proteasome function
In order to test the hypothesis that the P-gp inhibitor verapamil inhibits proteasome function, proteasome extracts of ECV304 and PC-3 cells were incubated with different concentrations of the drug (0, 50, 100 and 200 μM) and immediately tested for their chymotrypsin-like activity against the fluorogenic substrate SucLLVY-7-amido-4-methylcoumarin. There was a dose-dependent inhibition of MG-132-sensitive 26S (Fig. 1A) and 20S (data not shown) proteasome function, consistent with a direct inhibitory effect of verapamil on the proteasome.
Anthracyclines inhibit 20S and 26S proteasome function in a dose-dependent manner
Since verapamil, vinblastine, and CsA have been found to inhibit 20S and 26S proteasome function and vinblastine and CsA serve as substrates of P-gp [3], we asked if anthracyclines in general have an inhibitory effect on this protease. When crude extracts of ECV304 cells were incubated with different doses (0 – 100 μM) of the anthracyclines doxorubicin, daunorubicin, idarubicin and epirubicin we observed dose-dependent inhibition of 26S proteasome function with IC50 values of 65.5 μM for doxorubicin, 13.7 μM for daunorubicin, 38.6 μM for idarubicin and 29.2 μM for epirubicin (Table 1). Topotecan, mitomycin C, and gemcitabine had no measurable effect on 26S proteasome function (data not shown). 20S proteasome function was inhibited by doxorubicin (IC50 5.8 μM), idarubicin (IC50 92 μM), epirubicin (IC50 12.5 μM) but not by daunorubicin (Table 2).
In order to demonstrate if this inhibition could be observed in living cells, we incubated ECV304/10 cells, stably transfected with an expression plasmid for an Ub-GFP fusion protein with doxorubicin (100 μM) for 12 hours. When analyzed by fluorescence microscopy, the cells showed perinuclear accumulation of doxorubicin while GFP accumulated throughout the cytoplasm, indicating inhibition of proteasome function (Fig. 2).
MG-132 treatment reverts multi-drug-resistance in P-gp expressing KB 8-5 cells
The human epitheloid carcinoma cell line KB 8-5 is a well-characterized tumor cell line that over-expresses mdr-1 with associated MDR. Preliminary experiments showed that treatment of KB 8.5 cells with the reversible proteasome inhibitor MG-132 (3.125 to 50 μM) induced apoptosis within 24 hours. This is in accord with numerous studies reporting induction of apoptosis in cancer cells by proteasome inhibitors [13], and indicated that MG-132 enters KB 8-5 cells and that they are not abnormally resistant to its effects based on enhanced P-gp function. After 45 minutes of incubation with MG-132 (50 μM), no morphological signs of toxicity were observed. KB 8-5 cells treated with different doses of MG-132 and daunorubicin (10 μM) for 45 minutes showed increased, dose-dependent accumulation of daunorubicin in the cytoplasm (e.g. a 4-fold increase at 50 μM MG-132, Fig. 3) indicating that MG-132 could block P-gp function. This was further supported by the observation that incubation of ECV304 cell with MG-132 (25 μM) caused an increased uptake of doxorubicin in the cytoplasm and in the nuclear fraction of the cells (Fig. 4).
Discussion
The observations that CsA [4] and vinblastine [7] have inhibitory effects on the cleavage activity of the 26S proteasome led us investigate the effects of anthracycline anticancer agents and verapamil on the activity of this protease. Verapamil caused a concentration-dependent inhibition of 20S and 26S function. Additionally, we found a concentration-dependent inhibition of 26S proteasome function for all four anthracyclines tested. Comparable results showing doxorubicin to be a non-competitive inhibitor of the proteasome have been reported previously [14]. With the exception of daunorubicin, anthracyclines also inhibited 20S chymotryptic function in a dose-dependent manner. It is known that doxorubicin is co-transported into the nucleus along with proteasomes [15,16] but our observation of a general direct inhibitory effect of anthracycline anticancer agents on the proteasome sheds a totally new light on the actions of these drugs.
The inhibitory effects of the reversible inhibitor of the proteasome, MG-132, on P-glycoprotein function, supports the view that P-glycoprotein and the proteasome can both be targeted by this new class of chemotherapeutic drugs. This was further supported by the observation that verapamil, another established inhibitor of P-gp, inhibited the chymotryptic 20S and 26S function of the proteasome. The fact that both P-glycoprotein and proteasome activities can both be regulated by pro-inflammatory cytokines and oxidative stress suggests [17-21] that studies on co-ordinate regulation of these activities might be illuminating.
These findings lead to interesting possibilities with respect to the possible use of proteasome inhibitors, which are just entering their first clinical trials [22,23], in combination therapy, as well as to the mechanism of action and toxicity of P-gp inhibitors. Using an in vitro system, we showed that the proteasome inhibitor MG-132 caused intracellular accumulation of anthracyclines, indicating inhibition of P-gp function. Proteasome inhibitors may interfere with drug-resistance at additional levels as P-gp and also topisomerase II are degraded in a proteasome-dependent manner and degradation is blocked by proteasome inhibitors [24,25]. However, given the long half-life of P-gp of 14–24 hours [26], the effects observed in our study after short-time incubation of the cells with MG-132 are probably not caused by an increased degradation of P-gp. The extent of the increase of anthracyclin accumulation in mdr1-overexpressing KB-8.5 treated with 25 μM concentrations of MG-132 cells in our study was comparable to the effect of verapamil at 50 μM [27]. Future studies have to clarify if similar effects can be obtained using clinically used proteasome inhibitors at concentrations typically reached in the serum of patients.
Tumor cells in general exhibit altered patterns of expression of proteasome subunits and their distribution between cytoplasm and nucleus often differs from normal cells [28-30]. This may explain why specific proteasome inhibitors like PS-341 are usually clinically well tolerated. Inhibition of proteasome function induces apoptosis of tumor cells [31-34] and sensitizes the surviving tumor cells to the actions of both chemotherapy [35] and radiation therapy [36,37]. Therefore, proteasome inhibitors might overcome P-gp-related MDR, with accompanying chemo- and radiosensitizing effects. Also, since tumor microvasculature expresses high levels of mdr-1 [38,39], the possibility exists that the neovasculature is a target for these drugs in vivo. On the other hand, direct inhibition of proteasome function might be an additional major mechanism of action for anthracyclines. Such inhibition could contribute to their ability to enhance the efficacy of other chemotherapeutic drugs, independent of their ability to reverse MDR.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MF carried out the molecular studies and helped to draft the manuscript. WMB participated in the design of the study and helped to draft the manuscript. FP designed the study and drafted the manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This investigation was supported in part by Grants of the German Research Foundation (DFG) Pa 723/1 Pa 723/3 (FP) and PHS Grant number CA-87887 awarded to WMc by the National Cancer Institute;
Figures and Tables
Figure 1 Verapamil is an inhibitor of 26S proteasome function. Incubation of crude extracts of ECV304 cells containing proteasomes with different doses of verapamil (50, 60, 80, 100, 200 μM) inhibited proteolysis of the chymotrypsin-like substrate SucLLVY-AMC in a dose-dependent manner, indicating inhibition of 26S proteasome function.
Figure 2 Anthracyclines are inhibitors of proteasome function. Incubation of ECV304 cells stably transfected with an Ub-GFP fusion protein with daunorubicin (100 μM, 16 h), caused accumulation of GFP throughout the cytoplasm (lower picture), indicating proteasome inhibition in living cells while untreated controls cells showed only little accumulation of GFP (A/B). Daunorubicin accumulated in the perinuclear region (C).
Figure 3 MG-132 treatment of KB 8-5 causes intracellular accumulation of anthracyclines. Incubation of KB 8-5 cells, which overexpress P-gp, with increasing doses of MG-132 (0, 6.25, 12.5, 25, 50 μM) caused a dose-dependent accumulation of daunorubicin, as measured by fluorescence, indicating inhibition of P-gp function by this proteasome inhibitor.
Figure 4 Accumulation of doxorubicin in the presence or absence of MG-132 (25 μM) in the cytoplasm and the nuclear fraction of ECV304 cells.
Table 1 Chymotryptic 26S Proteasome Activity in Lysates from ECV 304 Cells
μM Doxorubincin Daunorubicin Epirubicin Idarubicin
0 1 1 1 1
6.25 0.92 ± 0.03* 0.85 ± 0.05* 0.79 ± 0.1 n.s. 0.97 ± 0.03 n.s.
12.5 0.84 ± 0.05* 0.53 ± 0.11* 0.57 ± 0.31 n.s. 0.96 ± 0.02 n.s.
25 0.72 ± 0.04** 0.22 ± 0.09** 0.59 ± 0.14* 0.79 ± 0.12 n.s.
50 0.59 ± 0.1* 0.12 ± 0.06** 0.36 ± 0.2* 0.28 ± 0.09 **
100 0.36 ± 0.03*** 0.11 ± 0.06** 0.27 ± 0.17* 0.1 ± 0.06 **
n.s. not significant, *p < 0.05, **p < 0.01, ***p < 0.001 (two-sided student's t-test)
Table 2 Chymotryptic 20S Proteasome Activity in Lysates from ECV 304 Cells
μM Doxorubincin Daunorubicin Epirubicin Idarubicin
0 1 1 1 1
6.25 0.49 ± 0.1* 0.62 ± 0.19 n.s. 0.71 ± 0.21 n.s. 0.67 ± 0.07*
12.5 0.28 ± 0.09** 0.64 ± 0.07 n.s. 0.46 ± 0.2 n.s. 0.67 ± 0.08*
25 0.28 ± 0.13* 0.81 ± 0.05 n.s. 0.37 ± 0.08** 0.69 ± 0.05**
50 0.24 ± 0.06** 0.95 ± 0.04 n.s. 0.33 ± 0.07** 0.67 ± 0.05 **
100 0.25 ± 0.07** 1.05 ± 0.1 n.s. 0.35 ± 0.07** 0.49 ± 0.11*
n.s. not significant, *p < 0.05, **p < 0.01, ***p < 0.001 (two-sided student's t-test)
==== Refs
Gottesman MM Fojo T Bates SE Multidrug resistance in cancer: role of ATP-dependent transporters Nat Rev Cancer 2002 2 48 58 11902585 10.1038/nrc706
Lacayo NJ Duran GE Sikic BI Modulation of resistance to idarubicin by the cyclosporin PSC 833 (valspodar) in multidrug-resistant cells J Exp Ther Oncol 2003 3 127 135 14641819 10.1046/j.1359-4117.2003.01088.x
Ambudkar SV Dey S Hrycyna CA Ramachandra M Pastan I Gottesman MM Biochemical, cellular, and pharmacological aspects of the multidrug transporter Annu Rev Pharmacol Toxicol 1999 39 361 398 10331089 10.1146/annurev.pharmtox.39.1.361
Meyer S Kohler NG Joly A Cyclosporine A is an uncompetitive inhibitor of proteasome activity and prevents NF-kappa B activation FEBS Lett 1997 413 354 358 9280312 10.1016/S0014-5793(97)00930-7
Harding CV France J Song R Farah JM Chatterjee S Iqbal M Siman R Novel dipeptide aldehydes are proteasome inhibitors and block the MHC-I antigen-processing pathway J Immunol 1995 155 1767 1775 7636233
Marienfeld R Neumann M Chuvpilo S Escher C Kneitz B Avots A Schimpl A Serfling E Cyclosporin A interferes with the inducible degradation of NF-kappa B inhibitors, but not with the processing of p105/NF-kappa B1 in T cells Eur J Immunol 1997 27 1601 1609 9247567
Piccinini M Tazartes O Mezzatesta C Ricotti E Bedino S Grosso F Dianzani U Tovo PA Mostert M Musso A Rinaudo MT Proteasomes are a target of the anti-tumor drug vinblastine Biochem J 2001 356 835 841 11389692 10.1042/0264-6021:3560835
Drewe J Gutmann H Fricker G Torok M Beglinger C Huwyler J HIV protease inhibitor ritonavir: a more potent inhibitor of P- glycoprotein than the cyclosporine analog SDZ PSC 833 Biochem Pharmacol 1999 57 1147 1152 11230802 10.1016/S0006-2952(99)00026-X
Schmidtke G Holzhutter HG Bogyo M Kairies N Groll M de Giuli R Emch S Groettrup M How an inhibitor of the HIV-I protease modulates proteasome activity J Biol Chem 1999 274 35734 35740 10585454 10.1074/jbc.274.50.35734
Stein RL Melandri F Dick L Kinetic characterization of the chymotryptic activity of the 20S proteasome Biochemistry 1996 35 3899 3908 8672420 10.1021/bi952262x
Gieseler F Nussler V Brieden T Kunze J Valsamas S Intracellular pharmacokinetics of anthracyclines in human leukemia cells: correlation of DNA-binding with apoptotic cell death Int J Clin Pharmacol Ther 1998 36 25 28 9476145
Dantuma NP Lindsten K Glas R Jellne M Masucci MG Short-lived green fluorescent proteins for quantifying ubiquitin/proteasome-dependent proteolysis in living cells Nat Biotechnol 2000 18 538 543 10802622 10.1038/75406
Shinohara K Tomioka M Nakano H Tone S Ito H Kawashima S Apoptosis induction resulting from proteasome inhibition Biochem J 1996 317 385 388 8713062
Kiyomiya K Satoh J Horie H Kurebe M Nakagawa H Matsuo S Correlation between nuclear action of anthracycline anticancer agents and their binding affinity to the proteasome Int J Oncol 2002 21 1081 1085 12370758
Kiyomiya K Matsuo S Kurebe M Mechanism of specific nuclear transport of adriamycin: the mode of nuclear translocation of adriamycin-proteasome complex Cancer Res 2001 61 2467 2471 11289116
Kiyomiya K Matsuo S Kurebe M Proteasome is a carrier to translocate doxorubicin from cytoplasm into nucleus Life Sci 1998 62 1853 1860 9600327 10.1016/S0024-3205(98)00151-9
Sukhai M Piquette-Miller M Regulation of the multidrug resistance genes by stress signals J Pharm Pharm Sci 2000 3 268 280 10994039
Stohwasser R Kloetzel PM Cytokine induced changes in proteasome subunit composition are concentration dependent Biol Chem 1996 377 571 577 9067255
Yang Y Waters JB Fruh K Peterson PA Proteasomes are regulated by interferon gamma: implications for antigen processing Proc Natl Acad Sci U S A 1992 89 4928 4932 1594596
Pajonk F McBride WH Ionizing radiation affects 26s proteasome function and associated molecular responses, even at low doses Radiother Oncol 2001 59 203 212 11325450 10.1016/S0167-8140(01)00311-5
Grune T Oxidative stress, aging and the proteasomal system Biogerontology 2000 1 31 40 11707918 10.1023/A:1010037908060
Teicher BA Ara G Herbst R Palombella VJ Adams J The proteasome inhibitor PS-341 in cancer therapy Clin Cancer Res 1999 5 2638 2645 10499643
Adams J Palombella VJ Elliott PJ Proteasome inhibition: a new strategy in cancer treatment Invest New Drugs 2000 18 109 121 10857991 10.1023/A:1006321828515
Zhang Z Wu JY Hait WN Yang JM Regulation of the stability of P-glycoprotein by ubiquitination Mol Pharmacol 2004 66 395 403 15322230 10.1124/mol.104.001966
Ogiso Y Tomida A Lei S Omura S Tsuruo T Proteasome inhibition circumvents solid tumor resistance to topoisomerase II-directed drugs Cancer Res 2000 60 2429 2434 10811120
Yoshimura A Kuwazuru Y Sumizawa T Ikeda S Ichikawa M Usagawa T Akiyama S Biosynthesis, processing and half-life of P-glycoprotein in a human multidrug-resistant KB cell Biochim Biophys Acta 1989 992 307 314 2570611
Ghauharali RI Westerhoff HV Dekker H Lankelma J Saturable P-glycoprotein kinetics assayed by fluorescence studies of drug efflux from suspended human KB8-5 cells Biochim Biophys Acta 1996 1278 213 222 8593279
Amsterdam A Pitzer F Baumeister W Changes in intracellular localization of proteasomes in immortalized ovarian granulosa cells during mitosis associated with a role in cell cycle control Proc Natl Acad Sci U S A 1993 90 99 103 8380501
Kumatori A Tanaka K Inamura N Sone S Ogura T Matsumoto T Tachikawa T Shin S Ichihara A Abnormally high expression of proteasomes in human leukemic cells Proc Natl Acad Sci U S A 1990 87 7071 7075 2205851
Kanayama H Tanaka K Aki M Kagawa S Miyaji H Satoh M Okada F Sato S Shimbara N Ichihara A Changes in expressions of proteasome and ubiquitin genes in human renal cancer cells Cancer Res 1991 51 6677 6685 1660345
Qiu JH Asai A Chi S Saito N Hamada H Kirino T Proteasome inhibitors induce cytochrome c-caspase-3-like protease- mediated apoptosis in cultured cortical neurons J Neurosci 2000 20 259 265 10627603
Pasquini LA Besio Moreno M Adamo AM Pasquini JM Soto EF Lactacystin, a specific inhibitor of the proteasome, induces apoptosis and activates caspase-3 in cultured cerebellar granule cells J Neurosci Res 2000 59 601 611 10686588 10.1002/(SICI)1097-4547(20000301)59:5<601::AID-JNR3>3.0.CO;2-1
Zhang XM Lin H Chen C Chen BD Inhibition of ubiquitin-proteasome pathway activates a caspase-3-like protease and induces Bcl-2 cleavage in human M-07e leukaemic cells Biochem J 1999 340 127 133 10229667 10.1042/0264-6021:3400127
Delic J Masdehors P Omura S Cosset JM Dumont J Binet JL Magdelenat H The proteasome inhibitor lactacystin induces apoptosis and sensitizes chemo- and radioresistant human chronic lymphocytic leukaemia lymphocytes to TNF-alpha-initiated apoptosis [see comments] Br J Cancer 1998 77 1103 1107 9569046
Bold RJ Virudachalam S McConkey DJ Chemosensitization of pancreatic cancer by inhibition of the 26S proteasome J Surg Res 2001 100 11 17 11516199 10.1006/jsre.2001.6194
Pajonk F Pajonk K McBride W Apoptosis and radiosensitization of Hodgkin’s cells by proteasome inhibition Int J Radiat Oncol Biol Phys 2000 47 1025 1032 10863075 10.1016/S0360-3016(00)00516-2
Russo SM Tepper JE Baldwin ASJ Liu R Adams J Elliott P Cusack JCJ Enhancement of radiosensitivity by proteasome inhibition: implications for a role of NF-kappaB Int J Radiat Oncol Biol Phys 2001 50 183 193 11316563 10.1016/S0360-3016(01)01446-8
Iwahana M Utoguchi N Mayumi T Goryo M Okada K Drug resistance and P-glycoprotein expression in endothelial cells of newly formed capillaries induced by tumors Anticancer Res 1998 18 2977 2980 9713495
Toth K Vaughan MM Peress NS Slocum HK Rustum YM MDR1 P-glycoprotein is expressed by endothelial cells of newly formed capillaries in human gliomas but is not expressed in the neovasculature of other primary tumors Am J Pathol 1996 149 853 858 8780389
|
16171522
|
PMC1242220
|
CC BY
|
2021-01-04 16:03:04
|
no
|
BMC Cancer. 2005 Sep 19; 5:117
|
latin-1
|
BMC Cancer
| 2,005 |
10.1186/1471-2407-5-117
|
oa_comm
|
==== Front
BMC CancerBMC Cancer1471-2407BioMed Central London 1471-2407-5-1211617658210.1186/1471-2407-5-121Research ArticleDegranulating mast cells in fibrotic regions of human tumors and evidence that mast cell heparin interferes with the growth of tumor cells through a mechanism involving fibroblasts Samoszuk Michael [email protected] Emi [email protected] John K [email protected] Department of Pathology and Laboratory Medicine, University of California, Irvine, California USA2 Department of Obstetrics and Gynecology, Stanford University, Stanford, California USA2005 21 9 2005 5 121 121 20 6 2005 21 9 2005 Copyright © 2005 Samoszuk et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The purpose of this study was to test the hypothesis that mast cells that are present in fibrotic regions of cancer can suppress the growth of tumor cells through an indirect mechanism involving peri-tumoral fibroblasts.
Methods
We first immunostained a wide variety of human cancers for the presence of degranulated mast cells. In a subsequent series of controlled in vitro experiments, we then co-cultured UACC-812 human breast cancer cells with normal fibroblasts in the presence or absence of different combinations and doses of mast cell tryptase, mast cell heparin, a lysate of the human mast cell line HMC-1, and fibroblast growth factor-7 (FGF-7), a powerful, heparin-binding growth factor for breast epithelial cells.
Results
Degranulating mast cells were localized predominantly in the fibrous tissue of every case of breast cancer, head and neck cancer, lung cancer, ovarian cancer, non-Hodgkin's lymphoma, and Hodgkin's disease that we examined. Mast cell tryptase and HMC-1 lysate had no significant effect on the clonogenic growth of cancer cells co-cultured with fibroblasts. By contrast, mast cell heparin at multiple doses significantly reduced the size and number of colonies of tumor cells co-cultured with fibroblasts, especially in the presence of FGF-7. Neither heparin nor FGF-7, individually or in combination, produced any significant effect on the clonogenic growth of breast cancer cells cultured without fibroblasts.
Conclusion
Degranulating mast cells are restricted to peri-tumoral fibrous tissue, and mast cell heparin is a powerful inhibitor of clonogenic growth of tumor cells co-cultured with fibroblasts. These results may help to explain the well-known ability of heparin to inhibit the growth of primary and metastatic tumors.
==== Body
Background
In animal models of carcinogenesis, mast cells have been shown to play an important role in modulating angiogenesis, connective tissue remodeling, blood clotting, and the growth of tumors [1-3]. Similarly, a number of clinical investigators, including our laboratory, have documented the natural presence of mast cells in various human tumors, including breast cancer and melanoma [4-6]. In these studies, the mast cells were described as being predominantly located in the fibrous tissue stroma adjacent to the tumor cells rather than within the core of the tumor.
Collectively, these observations prompted us to speculate that the mast cells that are naturally present next to tumors might modulate the growth of the tumor cells through an indirect mechanism involving the fibroblasts that are adjacent to the tumor cells. It is now well-established, for example, that fibroblasts play a pivotal role in promoting the growth of many types of cancer cells [7-9] and that many mast cell granule components exert powerful effects on fibroblasts [2,10]. Therefore, in order to test this hypothesis, we first performed a comprehensive, immunohistochemical survey of a variety of human tumors to determine if mast cells are generally restricted to the fibrous tissue adjacent to tumor cells. To this end, tumor sections were immunostained for the presence of mast cell tryptase, a serine protease that is indicates the presence of intact and degranulated mast cells in tissues. This marker was selected instead of CD117 or other markers of intact mast cells because it can specifically identify intact mast cells as well as regions where mast cells have disintegrated and released the contents of their granules into the tissue. Because mast cell granules contain tryptase that is strongly bound to heparin [11,12], tryptase co-localizes with heparin in tissue sections and can therefore serve as an effective surrogate marker for the presence of heparin in tissues. This property is important because reliable immunohistochemical assays to detect the presence of mast cell heparin in tissue sections are generally not available.
In the second part of this study, we explored the effects of two major components of mast cell granules (tryptase and heparin) and FGF-7 on the clonogenic growth of a human breast cancer cell line co-cultured with normal human fibroblasts in a simple co-culture system that we have previously described in detail [13]. We specifically selected mast cell tryptase for study because it exerts profound effects on fibroblast function and proliferation [2,10]. Heparin, a highly sulfated proteoglycan produced exclusively by mast cells, was included in this study because our previous experiments with mice that were genetically or enzymatically depleted of mast cell heparin [5,14] have consistently demonstrated that tumors grow significantly faster in heparin-depleted mice than in control mice. FGF-7 was incorporated into our experiments because it is a powerful, heparin-binding growth factor for breast epithelial cells that is produced by fibroblasts [15-18]. Moreover, previous studies have clearly demonstrated that fibroblasts are the predominant source of FGF-7 in malignant breast tissues [18,19].
Here we demonstrate that degranulating mast cells are restricted almost exclusively to the peri-tumoral fibrous tissue in human cancers. Moreover, we show that heparin significantly interferes with the clonogenic growth of breast cancer cells co-cultured with fibroblasts in the presence or absence of exogenous FGF-7. These results lead us to conclude that heparin from mast cells can suppress the growth of tumor cells through an indirect mechanism involving the fibroblasts adjacent to the tumor.
Methods
Primary human tumor and normal tissue samples
After obtaining institutional review board approval for this project, we retrieved the most recent available paraffin blocks from the surgical pathology archives of UCI Medical Center (Orange, CA) and Long Beach Memorial Medical Center (Long Beach, CA) for the following types of tumors: breast cancer (n = 10); head and neck cancer (n = 10); lung cancer (n = 10), non-Hodgkin's lymphoma (n = 5); Hodgkin's lymphoma (n = 8); ovarian cancer (n = 10) and their normal tissue counterparts (n = 3 for each tumor type).
Immunostaining for mast cell tryptase
Representative sections of each tumor and normal tissue counterpart were immunostained for mast cell tryptase using a procedure that has been previously described in detail [5,6] and also histochemically stained with Giemsa stain to highlight intact mast cells (identifiable by their dark purple granules). In brief, de-paraffinized tissue sections of each tumor and normal tissue were first subjected to antigen retrieval using microwave treatment in citrate buffer. The tissue sections were then incubated with a murine monoclonal antibody directed against human mast cell tryptase (Dako-Cytomation, Carpinteria, California USA) at a concentration of 10 micrograms/mL for 2 hours. Bound primary antibody was then detected using the Dako LSAB+ peroxidase kit-universal (Dako-Cytomation) in accordance with the manufacturer's directions. The positive control for this antibody was a section of human breast cancer with extensive infiltration by mast cells noted on routine histologic examination [6]. Because virtually all normal tissues contain at least a few mast cells, the negative controls were serial sections of the same tissues incubated with isotype-matched, non-immune IgG from a mouse. The specificity of the immunostaining was confirmed on two slides by competitive inhibition with human mast cell tryptase at a concentration of 1 microgram/mL (Sigma-Aldrich, St. Louis, Missouri, USA).
An experienced pathologist then counted the numbers of mast cells in three representative, non-overlapping, 40 × microscopic fields of the peripheral and central regions (core) of each tumor section. For the normal tissue counterparts, the numbers of mast cells were counted in three, randomly selected and non-overlapping 40 × microscopic fields of each slide. The means and standard deviations of mast cell numbers in the two tumor regions and in the normal tissue counterparts were then calculated and compared statistically, using a two-tailed, unpaired t-test for significance.
Co-culture assay
The co-culture assay was performed as previously described [13] In brief, UACC-812 breast cancer cells (ATCC, Manassas, VA) were seeded at a density of 100 cells per well of a 96-well microtiter plate containing a confluent monolayer of allegedly normal human fibroblasts (CCD 1068 SK), also from ATCC. The allegedly normal human fibroblasts were derived from the skin of the breast from a woman who underwent a mastectomy for breast cancer. The cells were co-cultured in Eagles minimal essential medium (ATCC) containing 10% fetal calf serum (Sigma-Aldrich) and penicillin-streptomycin (ATCC) that was then supplemented on days 1, 4, 7, 10, and 13 after seeding with human mast cell tryptase derived from human lung (1, 5, or 10 μgrams/mL; Sigma-Aldrich, St. Louis, MO), or purified porcine intestinal derived heparin, high molecular weight type (0.1,1.0, 10.0 and 100 units/mL; Sigma-Aldrich); and/or human fibroblast growth factor-7 (100 ng/mL;Chemicon International, Temecula, CA). It should be noted that the lung-derived tryptase used in these experiments contained 0.05 mM heparin, but the final concentration of heparin in the tryptase solution used in our experiments was approximately 20-fold less than the lowest concentrations of porcine heparin used alone. On day 14, the cells were fixed in situ by incubating the wells with ice-cold methanol for 10 minutes. The cells were then stained in situ with 0.1% toluidine blue, and we used routine light microscopy to count the numbers of colonies of tumor cells containing 10 or more tumor cells in each well. Each assay was performed in triplicate, and the means and standard deviations for the numbers of colonies in each treatment and control group were then calculated and compared using a two-tailed unpaired t-test for significance.
The positive control for the assay consisted of identical numbers of tumor cells co-cultured on a monolayer of fibroblasts without any additional supplements to the growth medium. In order to determine if any effects that we observed were attributable to the fibroblasts, we also performed the clonogenic assay as described above, but in the absence of a monolayer of fibroblasts.
For comparative purposes, we also performed the co-culture assay in the presence of 50 microliters of lysate equivalent to 106 HMC-1 cells/mL. The HMC-1 cell line is a human mast cell line derived from a patient with mast cell leukemia [20] and was the generous gift of Dr. J.H. Butterfield from the Mayo Clinic (Rochester, Minnesota, USA). It expresses many of the markers and cytoplasmic components of normal mast cells [20] and was cultured under conditions that have been previously described [20]. The HMC-1 lysate was produced by a procedure described by Huttunen et al [21]. To confirm the presence of heparin in the HMC-1 cells, we stained cytopreparations of the cultured cells for 30 seconds with 0.1% toluidine blue, a metachromatic dye that binds strongly to heparin and produces an intense blue color in the granules of normal mast cells [11].
Measurement of tumor colony size
Digital images of 10–50 representative colonies of tumor cells in each well were acquired at 40 × magnification with a Nikon Eclipse E600 microscope equipped with a Spot II digital camera (Diagnostic Instruments, Sterling Heights, MI). Because some treatments resulted in far fewer colonies (see below) than the control group, the numbers of colonies that were analyzed among the different treatment groups varied from group to group. The digital images were then analyzed using a digital tool (Image Pro Plus version 4.5, Media Cybernetics, Inc., Silver Spring, MD) to outline the area of the tumor cell colony in each image. The area in each colony was electronically integrated, and we plotted the distribution of areas as a frequency histogram for each treatment and control group. The data were then analyzed statistically using the Mann-Whitney U-test for two samples and ranked observations, not paired.
Detection of FGF-7 in cultured human fibroblasts
A number of other investigators have previously documented with Western blotting and in situ hybridization that fibroblasts secrete FGF-7 and are the predominant source of FGF-7 in normal tissues and in breast cancers [18,19]. In order to confirm that the monolayer of fibroblasts that was used in our experiments was also producing FGF-7, we first immunostained the fibroblasts in situ with a monoclonal antibody directed against FGF-7 (Chemicon International), using a procedure that we have previously described in detail [13]. The negative controls consisted of an irrelevant monoclonal antibody or blocking buffer alone. As an additional confirmation for the production of FGF-7 by the fibroblasts, we analyzed the gene expression profile of the cultured fibroblasts, using a DNA microarray as previously described by us [13].
In order to provide additional confirmation of the presence of FGF-7 in the fibroblasts, we performed Western blotting of an immunoprecipitate of cultured fibroblasts. In brief, the 1068SK cells were lysed using RIPA Lysis Buffer (Upstate Biochemicals, Lake Placid, NY USA) supplemented with 1 mM PMSF. The cell lysates was then incubated with 0.3 ug/ml anti-FGF7 (Chemicon) for 1 hour then bound to Protein G beads (Sigma) for 1 hour. Beads were washed and spun down at 1,400 rpm for 1 minute in a microcentrifuge. Pelleted beads were boiled for 10 minutes at 95°C in sample buffer, centrifuged at top speed for 5 minutes and the resulting supernatant was collected for separation by SDS-PAGE on a 12% separating gel. Proteins were transferred to PVDF membrane (BioRad, Hercules, CA USA), incubated with rabbit anti-human KGF primary antibody (Chemicon) then with HRP conjugated goat anti-rabbit secondary antibody (BioRad), and visualized by Opti-4CN Detection Kit (BioRad). The negative control consisted of a lysate of a murine melanoma cell line, B16. The positive control consisted of human FGF-7 (Chemicon) as described above.
Results
Mast cells in fibrotic regions of tumors
A summary of the results of the immunostaining for mast cell tryptase is provided in Table 1. In all of the human tumors that we examined, mast cell tryptase and intact mast cells were located predominantly at the edges of the tumors and within the fibrous stroma next to tumors (Figure 1a–e). Notably, the mast cell infiltration was significantly greater in the tumors than in the normal tissue counterparts for almost all of the tumor types that we examined. Only the Hodgkin's lymphomas and non-Hodgkin's lymphomas consistently had substantial amounts of tryptase detectable within the core of the tumors, but even in these tumors, there were more mast cells at the edges of the tumor and in the fibrous stroma subdividing the tumor mass (Table 1 and Figure 1f). The Giemsa stain confirmed that intact mast cells were almost exclusively found embedded within the fibrous stroma within tumors (Figure 1d).
Reduction of clonogenic growth of breast cancer cells by heparin
The results of a representative clonogenic assay performed in triplicate with 1 unit/mL of heparin are presented in Figure 2. The experiment was repeated three times, with similar results. The consistent and notable finding was that this low dose of heparin significantly decreased the number of colonies of tumor cells in the co-culture assay with fibroblasts. The lowest concentration of heparin (0.1 units/mL) produced no detectable effect (average number of colonies = 45 ± 6, n = 3), while the higher doses of heparin (10 and 100 units/mL) produced effects very similar to 1 unit/mL (data not separately shown).
FGF-7 slightly increased the number of colonies, but this effect was neutralized by heparin. Tryptase at all of the concentrations that we tested produced no significant change in the numbers or sizes of tumor cell colonies in the co-culture assay (data not shown), despite the presence of small amounts of heparin in the tryptase solution. When the clonogenic assay was performed in the absence of fibroblasts, an average of only 23 ± 4 (n = 4) colonies were formed. Neither heparin nor FGF-7, individually (18 ± 5 colonies, n = 3) or in combination (20 ± 7 colonies, n = 3) significantly changed the clonogenic growth of breast cancer cells cultured without fibroblasts.
Importantly, the lysate of HMC-1 cells produced no significant effect on the clonogenic growth of the breast cancer cells in the co-culture assay (average number of colonies = 24 ± 4, n = 3). A toluidine blue stain of the cultured mast cell line revealed only weak staining of small numbers of intracytoplasmic granules (Figure 3) compared to the intense staining that is normally seen in mast cell granules [11,14]. This result suggests that the HMC-1 cell line grown in our laboratory produced only small amounts of heparin.
Reduction of size of tumor cell colonies by heparin
Visual microscopic examination of the tumor cell colonies in the clonogenic co-culture assay suggested that heparin produced a striking reduction in the size of the tumor cell colonies (Figure 4). This effect appeared to be particularly pronounced when heparin was added to co-cultures supplemented with FGF-7 that, by itself, markedly increased the size of the colonies (Figures 4c, d vs 4e). A statistical analysis of the distribution of colony sizes confirmed that heparin significantly reduced the size of the tumor cell colonies relative to the control or to FGF-7 treatment (Figure 5).
FGF-7 on membrane of cultured fibroblasts
Immunostaining studies confirmed that the fibroblasts expressed abundant FGF-7 on their cell membranes (Figure 4f). The negative control antibody and blocking buffer yielded no staining of the fibroblasts. The DNA microarray studies performed in triplicate indicated the presence of abundant mRNA coding for FGF-7 in the cultured fibroblasts (mean signal intensity 3940).
Western blotting (Figure 6) confirmed that our anti-FGF-7 antibody specifically recognized FGF-7 (lane 3) and not an immunoprecipitate of a negative control cell line (lane 2). The immunoprecipitate of the fibroblasts (lane 1) yielded a band at approximately 66 kD, similar to the results described by Palmieri et al [18]. The higher molecular weight of the FGF-7 in the immunoprecipitate of fibroblasts is probably attributable to the presence of tightly bound heparan sulfate from the cell membrane, as described by Palmieri et al [18].
Discussion
Our immunohistochemical studies have demonstrated that degranulating mast cells are located primarily in the peri-tumoral fibrous tissue in a wide variety of human cancers. Moreover, we have shown that heparin (a proteoglycan that is produced exclusively by mast cells) inhibits the clonogenic growth of human breast cancer cells co-cultured with normal fibroblasts but not tumor cells cultured alone. Significantly, the lysate of a human mast cell line, HMC-1, did not have any detectable effect on the clonogenic growth of the co-cultured tumor cells. We believe that this is probably due to the low levels of heparin in these cells, but we cannot exclude the possibility that some other component of mast cells may have neutralized the inhibitory effect of heparin. Similarly, various doses of tryptase had no effect on clonogenic growth. Taken together, these results provide strong evidence that mast cells can suppress the growth of tumor cells through an indirect mechanism that involves heparin and fibroblasts adjacent to the tumor cells.
This conclusion is consistent with our previous reports that depletion of endogenous heparin results in accelerated tumor growth in mice [5,14]. In specific, we previously showed that syngeneic tumor cells implanted into NDST-2 knockout mice grew faster than tumor cells implanted into wild-type mice that synthesized normal heparin [5,14]. NDST-2 knockout mice are unable to synthesize mast cell heparin and express abnormal mast cells with severely reduced amounts of histamine and mast cell proteases [11], probably because highly anionic heparin is required to stabilize the cationic compounds histamine, tryptase and chymase. Moreover, enzymatic depletion of mast cell heparin by injection of heparinase enzyme into tumor implants also accelerated tumor growth [14] and increased blood clotting within the tumors. Based on the results of the current study, we now propose that the accelerated tumor growth that we observed in heparin-depleted mice could be attributable, at least in part, to the absence of heparin-mediated inhibition of the growth-promoting interaction between fibroblasts and adjacent tumor cells.
Our findings are important because they may help to explain the well known ability of heparin and its derivatives to inhibit the growth of primary and metastatic tumors in various animal models and in humans with cancer [22-29]. Although the exact mechanism of this anti-tumor effect mediated by heparin remains uncertain, a number of possible explanations [30] have been proposed, including inhibition of thrombin and fibrin formation, immune system modulation, blockade of tumor cell adhesion to platelets, inhibition of angiogenesis, and functional inhibition of selectin-mediated cell-cell interactions leading to metastasis [31]. Heparins can also directly inhibit proliferation of various normal cell types, including smooth muscle cells and epithelial cells [reviewed in [32]]. Under some conditions, however, heparin actually stimulated the growth of epithelial cells [21]. On the other hand, only a few studies have directly evaluated the effects of heparin on the proliferation of cancer cells [also reviewed in [32]], and the results of these studies are generally inconclusive. Therefore, we propose that another possible mechanism for the anti-tumor effect of heparin observed in vivo may be the disruption of the critical, growth- promoting interactions that are known to occur between tumor cells and adjacent fibroblasts [7-9].
Although this study was not designed to explore the molecular mechanism of how heparin disrupts the interaction between fibroblasts and tumor cells, one possible explanation comes immediately to mind. Based on our findings with FGF-7, we speculate that heparin may disrupt the interaction of heparin-binding growth factors such as FGF-7 with heparan sulfate proteoglycans that are produced by fibroblasts and that act as essential co-factors for the growth factors. This explanation is consistent with our recent observation that optimal clonogenic growth of breast cancer cells in vitro requires direct physical contact between fibroblasts and breast cancer cells [13]. Moreover, previous reports from other laboratories have confirmed that human breast cancer cells are generally in close contact with fibroblasts that express abundant FGF-7 [18,19]. Finally, various heparan sulfate proteoglycans such as syndecan-1 and glypican are known to be expressed by stromal cells in cancer and to modulate the mitogenic effects of multiple heparin-binding growth factors [33,34]. In this regard, it is especially notable that heparin had no effect on FGF-7-mediated stimulation of mammary epithelial cells grown in a collagen gel matrix in the absence of fibroblasts [6]. Hence, it is reasonable to propose that heparin might interfere with the natural binding of heparin-binding growth factors to heparan sulfate proteoglycans produced by fibroblasts. Clearly, additional studies will be needed to test this hypothesis.
The possible biological and clinical significance of our in vitro experimental findings with regard to the naturally occurring mast cells that we observed around tumors remain speculative. Nevertheless, our observation that mast cells are abundantly present within the fibrous regions of tumors raises the intriguing possibility that a growth inhibitory mechanism similar to the one that we observed in our in vitro studies could also be naturally operative within tumors in vivo. The possible connection between our in vitro experiments and naturally occurring tumors is further strengthened by the reports from other laboratories that peri-tumoral fibroblasts express abundant FGF-7 [18,19] and that mast cell tryptase is intimately associated with the concurrent presence of heparin [11,12].
We acknowledge, however, that this possibility seems to be at odds with the accumulating evidence that mast cells promote rather than inhibit tumor growth [1,2]. In this regard, it should be emphasized that mast cell granules contain numerous biologically active compounds in addition to heparin, such as histamine, tryptase, and chymase. Some of these mast cell compounds and metabolites are likely to have significant effects on fibroblasts that remain to be defined. In addition, a number of other mediators from fibroblasts and mast cells could potentially influence tumor growth through a variety of mechanisms such as cyclooxygenase metabolites, heparanases, etc. The net effect on tumor growth, therefore, is likely to be the result of multiple complex interactions between the various components of mast cell granules and adjacent stromal cells such as vascular endothelium and fibroblasts. Indeed, our immunohistochemical studies also demonstrated close proximity of mast cells to peri-tumoral blood vessels as well as to fibroblasts. Consequently, it is entirely conceivable that the stimulatory effects of mast cells on angiogenesis or fibroblasts or other functions within tumors might outweigh the inhibitory effects mediated by mast cell heparin and fibroblasts.
Conclusion
Multiple independent lines of evidence strongly suggest that heparin from mast cells can suppress tumor cell growth through an indirect mechanism involving adjacent fibroblasts. The evidence includes the localization of mast cells to fibrous regions of tumors, the ability of heparin to inhibit tumor growth in vivo and in vitro in the presence of fibroblasts, and the accelerated growth of tumors in mice that were genetically or enzymatically depleted of heparin. Thus, we propose that further molecular analysis of the interaction between fibroblasts and heparin is warranted and may lead to improved insights into how heparin mediates its anti-tumor effect and, ultimately, to improved anti-tumor therapies.
Competing interests
MS is a paid consultant for Chemicon International (Temecula, CA) which supplied some of the reagents used in this study.
Authors' contributions
MS conceived of the study, designed the experiments, and drafted the manuscript. EK performed the experiments and helped to draft the paper. JKC provided the ovarian cancer samples, participated in the design of this study, and critically reviewed the manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This work was supported by a grant from the University of California Cancer Research Coordinating Committee (CRCC#35219 to MS).
Figures and Tables
Figure 1 Representative photomicrographs of mast cell infiltration in human head and neck cancer (1a), lung cancer (1b, c), ovarian cancer (1d, e), and Hodgkin's disease (1f). In Figures a-c, e and f, the sections were immunostained for mast cell tryptase and revealed extensive evidence of mast cell degranulation (brown stain) in the fibrous tissue adjacent to nests of tumor cells and blood vessels. A Giemsa stain (1d) confirmed the presence of intact mast cells with dense purple granules embedded within fibrous stroma between nests of tumor cells. Hematoxylin counterstain (all figures, except 1d).
Figure 2 Results of clonogenic assay of breast cancer cells co-cultured with fibroblasts in the presence or absence of heparin and/or FGF-7. After 14 days of co-culture, the numbers of colonies containing 10 or more tumor cells were counted and averaged. Heparin significantly reduced the numbers of tumor cell colonies in the presence or absence of exogenous FGF-7. When the tumor cells were cultured in the absence of fibroblasts, only an average of 24 colonies were formed, and the heparin had no significant effect on the numbers of colonies.
Figure 3 Toluidine blue stain of cytopreparation of HMC-1 cells. Relatively few blue granules were detected in the cultured human mast cell line, suggesting that the cultured mast cell line produced only small amounts of heparin.
Figure 4 Photomicrographs of tumor cell colonies (3a-e) and fibroblasts immunostained for FGF-7 (3f). In the absence of heparin, breast cancer cells co-cultured on a monolayer of fibroblasts generally produced irregularly shaped large clusters of tumor cells (3a). When heparin was added to the co-culture system, the tumor cells generally formed compact clusters (3b). FGF-7 added to the co-culture system resulted in large glandular-like structures composed of a peripheral rim of compact tumor cells and a central core of plate-like epithelial cells (3c, d). Heparin completely reversed the effects of FGF-7 and produced small compact clusters of darkly staining tumor cells (3e). Immunostaining for FGF-7 (3f) confirmed the abundant presence of FGF-7 (pink stain) on the membrane of the fibroblasts. Toluidine blue stain (3a-e).
Figure 5 Distribution of tumor cell colony sizes. The average sizes of the tumor cell colonies as measured by digital image analysis were 16,428 pixels for the control; 9917 pixels for heparin alone; 35,760 pixels for FGF-7 alone; and 5551 pixels for the cells treated with heparin and FGF-7. Heparin significantly reduced the distribution of colony sizes in the presence or absence or FGF-7 as determined by a Mann-Whitney U-test.
Figure 6 Western blot of FGF-7 in immunoprecipitate of fibroblasts. The polyclonal antibody recognized the positive control (approximately 22 kD; human FGF-7, lane 3) and did not bind to an immunoprecipitate of an irrelevant cell line (IP negative control, lane 2). The immunoprecipitate of the fibroblasts in lane 1 yielded a band at approximately 66 kD, which we attributed possibly to binding of FGF-7 to a heparan sulfate proteoglycan in the cell membrane.
Table 1 Comparison of mast cell numbers in tumors and corresponding normal tissues.
Mean Numbers of Mast Cells/40 × Field (± 1 standard deviation)
Peri-tumoral tissue Core of tumor Normal tissue
Breast (n = 10) 4.9 ± 1.1 0.3 ± 0.3* 1.2 ± 0.9**
Head & neck (n = 10) 6.7 ± 3.6 0.5 ± 0.3* 2.8 ± 2.1**
Ovarian (n = 10) 3.0 ± 0.8 0.7 ± 0.6* 0.9 ± 0.8**
Lung (n = 10) 4.4 ± 1.9 1.0 ± 0.3* 2.5 ± 0.6 **
Non-Hodgkins (n = 5) 2.1 ± 1.0 1.8 ± 0.5 NS 1.0 ± 0.5 NS
Hodgkin's (n = 8) 4.3 ± 1.3 1.9 ± 0.8* 1.0 ± 0.5**
*p value < 0.05, comparing mean numbers of mast cells in core of tumor versus per-tumoral regions.
** p value < 0.05, comparing mean numbers of mast cells in normal tissue counterparts (n = 3) to peri-tumoral tissue.
==== Refs
Wedemeyer J Galli SJ Decreased susceptibility of mast cell-deficient KitW/KitW-v mice to the development of 1, 2-dimethylhydrazine-induced intestinal tumors Lab Invest 2005 85 388 96 15696191 10.1038/labinvest.3700232
Coussens LM Raymond WW Bergers G Laig-Webster M Behrendtsen O Werb Z Caughey GH Hanahan D Inflammatory mast cells up-regulate angiogenesis during squamous epithelial carcinogenesis Genes Dev 1999 13 1382 97 10364156
Ribatti D Vacca A Nico B Crivellato E Roncali L Dammacco F The role of mast cells in tumor angiogenesis Br J Haematol 2001 115 514 21 11736931 10.1046/j.1365-2141.2001.03202.x
Kankkunen JP Harvima IT Naukkarinen A Quantitative analysis of tryptase and chymase containing mast cells in benign and malignant breast lesions Int J Cancer 1997 72 385 8 9247277 10.1002/(SICI)1097-0215(19970729)72:3<385::AID-IJC1>3.0.CO;2-L
Samoszuk M Corwin M Yu H Wang J Nalcioglu O Su MY Inhibition of thrombosis in melanoma allografts in mice by endogenous mast cell heparin Thromb Haemost 2003 90 351 60 12888884
Samoszuk M Corwin M Mast cell inhibitor cromolyn increases blood clotting and hypoxia in murine breast cancer Int J Cancer 2003 107 159 63 12925973 10.1002/ijc.11340
Bhowmick NA Neilson EG Moses HL Stromal fibroblasts in cancer initiation and progression Nature 2004 432 332 7 15549095 10.1038/nature03096
Grum-Schwensen B Klingelhofer J Berg CH El-Naaman C Grigorian M Lukanidin E Ambartsumian N Suppression of tumor development and metastasis formation in mice lacking the S100A4(mts1) gene Cancer Res 2005 65 3772 80 15867373 10.1158/0008-5472.CAN-04-4510
Orimo A Gupta PB Sgroi DC Arenzana-Seisdedos F Delaunay T Naeem R Carey VJ Richardson AL Weinberg RA. Stromal fibroblasts present in invasive human breast carcinomas promote tumor growth and angiogenesis through elevated SDF-1/CXCL12 secretion Cell 2005 121 335 348 15882617 10.1016/j.cell.2005.02.034
Frungieri MB Weidinger S Meineke V Köhn FM Mayerhofer A Proliferative action of mast-cell tryptase is mediated by PAR2, COX2, prostaglandins, and PPARγ: possible relevance to human fibrotic disorders PNAS 2002 99 15072 7 12397176 10.1073/pnas.232422999
Forsberg E Pejler G Ringvall M Lunderius C Tomasini-Johansson B Kusche-Gullberg M Eriksson I Ledin J Hellman L Kjellen L Abnormal mast cells in mice deficient in a heparin-synthesizing enzyme Nature 1999 400 773 6 10466727 10.1038/23488
Schwartz LB Bradford TR Regulation of tryptase from human lung mast cells by heparin J Biol Chem 1986 261 7372 9 3519608
Samoszuk M Tan J Chorn G Clonogenic growth of human breast cancer cells co-cultured in direct contact with serum-activated fibroblasts Breast Cancer Res 2005 7 R274 83 15987422 10.1186/bcr995
Samoszuk M Corwin MA Acceleration of tumor growth and peri-tumoral blood clotting by imatinib mesylate (Gleevec™) Int J Cancer 2003 106 647 52 12866022 10.1002/ijc.11282
Ulich TR Yi ES Cardiff R Yin S Bikhazi N Biltz R Morris CF Pierce GF Keratinocyte growth factor is a growth factor for mammary epithelium in vivo. The mammary epithelium of lactating rats is resistant to the proliferative action of keratinocyte growth factor Am J Pathol 1994 144 862 8 8178937
Imagawa W Cunha GR Young P Nandi S Keratinocyte growth factor and acidic fibroblast growth factor are mitogens for primary cultures of mammary epithelium Biochem Biophys Res Commun 1994 204 1165 9 7526860 10.1006/bbrc.1994.2585
Bansal GS Cox HC Marsh S Gomm JJ Yiangou C Luqmani Y Coombers RC Johnston CL Expression of keratinocyte growth factor and its receptor in human breast cancer Br J Cancer 1997 75 1567 74 9184170
Palmieri C Roberts-Clark D Assadi-Sabet A Coope RC O'Hare M Sunters A Hanby A Slade MJ Gomm JJ Lam EW Coombes RC Fibroblast growth factor 7, secreted by breast fibroblasts, is an interleukin-1β-induced paracrine growth factor for human breast cells J Endocrinol 2003 177 65 81 12697038 10.1677/joe.0.1770065
Bansal GS Cox HC Marsh S Gomm JJ Yiangou C Luqmani Y Coombes RC Johnston CL Expression of keratinocyte growth factor and its receptor in human breast cancer Br J Cancer 1997 75 1567 74 9184170
Nilsson G Blom T Kusche-Gullberg M Kjellen J Butterfield JH Sundstrom C Nilsson K Hellman L Phenotypic characterization of the human mast-cell line HMC-1 Scan J Immunol 1994 39 489 98
Huttunen M Hyttinen M Nilsson G Butterfield JH Horsmanheimo M Harvima IT Inhibition of keratinocyte growth in cell culture and whole skin culture by mast cell mediators Exp Dermatol 2001 10 184 92 11380614 10.1034/j.1600-0625.2001.010003184.x
Zacharski LR Ornstein DL Heparin and cancer Thromb Haemost 1998 80 10 23 9684778
Hejna M Raderer M Zielinski CC Inhibition of metastases by anticoagulants J Natl Cancer Inst 1999 91 22 36 9890167 10.1093/jnci/91.1.22
Zacharski LR Ornstein DL Mamourian AC Low-molecular-weight heparin and cancer Semin Thromb Haemost 2000 26 69 77 10.1055/s-2000-9499
Zacharski LR Anticoagulants in cancer treatment: malignancy as a solid phase coagulopathy Cancer Lett 2002 186 1 9 12183069 10.1016/S0304-3835(02)00258-6
Kakkar AK Antithrombotics and cancer: evidence for survival benefit Pathophysiol Haemost Thromb 2003 33 67 8 12955010 10.1159/000073299
Kakkar AK Levine MN Thrombosis and cancer: implications beyond Trousseau J Thromb Haemost 2004 2 1261 2 15304027 10.1111/j.1538-7836.2004.00867.x
Klerk CPW Smorenburg SM Otten HM Lensing AW Prins MH Piovella F Prandoni P Bos MM Richel DJ van Tienhoven G Buller HR The effect of low molecular weight heparin on survival in patients with advanced malignancy J Clin Oncol 2005 23 2130 2135 15699479 10.1200/JCO.2005.03.134
Lee AY Rickles FR Julian JA Gent M Baker RI Bowden C Kakkar AK Prins M Levine MN Randomized comparison of low molecular weight heparin and coumarin derivatives on the survival of patients with cancer and venous thromboembolism J Clin Oncol 2005 23 2123 2129 15699480 10.1200/JCO.2005.03.133
Falanga A The effect of anticoagulant drugs on cancer J Thromb Haemost 2004 2 1263 5 15304028 10.1111/j.1538-7836.2004.00868.x
Borsig L Non-anticoagulant effects of heparin in carcinoma metastasis and Trousseau's syndrome Pathophysiol Haemost Thromb 2003 33 64 66 12955009 10.1159/000073298
Smorenburg SM van Noorden CJF The complex effects of heparins on cancer progression and metastasis in experimental studies Pharmacol Reviews 2001 53 93 105
Maeda T Alexander CM Friedl A Induction of syndecan-1 expression in stromal fibroblasts promotes proliferation of human breast cancer cells Cancer Res 2004 64 612 21 14744776 10.1158/0008-5472.CAN-03-2439
Matsuda K Maruyama H Guo F Kleeff J Itakura J Matsumoto Y Lander AD Korc M Glypican-1 is overexpressed in human breast cancer and modulates the mitogenic effects of multiple heparin-binding growth factors in breast cancer cells Cancer Res 2001 61 5562 9 11454708
|
16176582
|
PMC1242221
|
CC BY
|
2021-01-04 16:03:05
|
no
|
BMC Cancer. 2005 Sep 21; 5:121
|
utf-8
|
BMC Cancer
| 2,005 |
10.1186/1471-2407-5-121
|
oa_comm
|
==== Front
BMC Ear Nose Throat DisordBMC Ear, Nose and Throat Disorders1472-6815BioMed Central London 1472-6815-5-71615939110.1186/1472-6815-5-7Research ArticleSelection of indicators for tonsillectomy in adults with recurrent tonsillitis Kasenõmm Priit [email protected] Andres [email protected] Mart [email protected] Mart [email protected] Marika [email protected] Department of Microbiology, Tartu University, Ravila St. 19, Tartu 50411, Estonia2 Department of Otorhinolaryngology, Tartu University Clinicum, Kuperjanovi St. 1, Tartu 51003, Estonia3 Department of General and Molecular Pathology, Medical Faculty, Tartu University, Ravila St. 19, Tartu 50411, Estonia2005 13 9 2005 5 7 7 9 4 2005 13 9 2005 Copyright © 2005 Kasenõmm et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
We aimed to find some new indicators for tonsillectomy (TE) in adults with recurrent tonsillitis (RT) by exploring whether the frequency of tonsillitis episodes and the length of morbidity period are associated with the macroscopic signs of sclerotic process in tonsils and microbiological data assessed by culture, molecular (PCR) and transmission electron microscopy (EM) methods.
Methods
The study involved 62 RT patients admitted for TE (age range 15–35, median 22 years) and 54 healthy volunteers (age range 18–24, median 20 years). The index of tonsillitis (IT) was calculated by multiplying the number of tonsillitis episodes per year by the morbidity period in years. On oropharyngeal examination the presence or absence of three sclerotic signs was evaluated: tonsillar sclerosis, obstruction of tonsillar crypts and scar tissue on the tonsils. The occurrence of Streptococcus pyogenes was assessed by culture and PCR methods in 24 tonsillar core specimens. The samples for EM investigation of crypt epithelium were taken from 10 removed tonsils.
Results
The IT values were in positive correlation with the number of sclerotic signs on oropharyngeal examination (r = 0.325, P = 0.010). Based on the IT values and the presence or absence of tonsillar sclerosis and obstruction of tonsillar crypts the receiver-operating curve (ROC) was constructed. It revealed that an IT score of 36 is an optimal cut-off value for prediction of sclerotic type tonsils. S. pyogenes was never found by culture, but its presence by PCR in nearly one third (29%) of diseased tonsillar tissue specimens was tightly associated with longer morbidity. EM revealed coccoid forms of intracellular bacteria in the crypt epithelium, which was accompanied with the damage of tight junctions between epithelial cells.
Conclusion
The index of tonsillitis ≥36, being a combination between the frequency of tonsillitis and the length of morbidity period, predicts the sclerotic process in recurrently inflamed tonsils. Therefore, the high IT values could serve as an indicator for TE in adults. The correlation between the longer morbidity period and the presence of S. pyogenes by PCR suggests that persistent infection may have a role in maintenance of recurrent inflammation in tonsils.
==== Body
Background
Recurrent tonsillitis (RT) is a chronic inflammatory process in palatine tonsils. A leading therapeutic approach for such a condition has been tonsillectomy (TE). Traditionally, recommendation for TE has depended primarily on the frequency of tonsillitis episodes. Patients with at least three episodes per year, despite adequate medical therapy, may be considered as candidates for TE, and surgical treatment is definitely recommended for patients with more than four or five episodes per year [1,2]. Adult patients often have fewer or less severe tonsillitis episodes, yet the dominance of other indices of chronic disease, such as poor general health, time loss from school or work, decreased life quality due to systemic effects or comorbid diseases, group A β-hemolytic streptococcus carriage state, and increased serum concentrations of antibodies, which have also been considered as appropriate indicators for TE [3-5]. Unfortunately, there is no consensus for these arbitrarily used criteria, hinting at a need for more precise indicators.
Palatine tonsils are a part of the mucosa-associated lymphatic tissue (MALT), a specialized compartment of the immune system that serves as a first line of defence against environmental harmful factors, including pathogenic microbes [6]. Paradoxically, palatine tonsils themselves are quite frequently affected by bacterial and viral infections causing local inflammation and systemic reactions. Recurrent or chronic inflammation in the tonsillar tissue results in obstruction of tonsillar crypts due to tissue fibrosis, accompanied by distension of the crypts' bottom and retention of its content [7,8]. In our previous study, the sclerotic and inflammatory type tonsils were discriminated, based on the presence or absence of tonsillar sclerosis, obstruction of tonsillar crypts and scar tissue on the tonsils. The sclerotic process in tonsils was further evidenced by increased collagen content. It revealed that sclerotic type tonsils had remarkably lower count of neutrophils in its tissue, which increased the risk for bacteraemia during tonsillectomy [9]. Thus, extensive tissue fibrosis seemed to be a critical point in the RT pathogenesis where the defensive function of tonsils becomes impaired. Unfortunately, there are no studies showing whether the frequency of tonsillitis episodes and the length of morbidity period have an association with sclerotic process in recurrently inflamed tonsils. We suggest that such an approach might be helpful in finding some new indicators for TE.
Despite the high frequency in population, the etiology of RT has remained unclear. The surface and deep bacterial flora of recurrently inflamed tonsils consist of an abundance of potentially pathogenic aerobic and anaerobic bacteria [10-14]. Surprisingly, the isolation rate of Streptococcus pyogenes from adults with RT, the most important pathogen in acute tonsillar infection, is lower by conventional culture methods [15-17]. As the pathogenesis of various infectious diseases has been attributed to intracellularly residing bacteria, applying some modern methods could provide advantages to determine the occurrence of S. pyogenes and its role in the pathogenesis of RT.
The aim of this study was to assess whether the frequency of tonsillitis episodes and the length of morbidity period are associated with the macroscopic signs of sclerotic process in tonsils and microbiological data, assessed by culture, molecular (PCR) and transmission electron microscopy (EM) methods in RT patients admitted for TE (RT-TE).
Methods
Clinical cohort and follow-up
Patients
The study involved 62 RT-TE patients (age range 15–35, median 22 years; 41 females and 21 males) selected among 486 adults referred for TE due to recurrent attacks of tonsillitis episodes between October to December 2000, March to June and September to December 2001 at the Department of Otorhinolaryngology, Tartu University Clinicum. Every third patient (≥15 years of age) was selected from the operation list on two particular days of the week. Each patient had a history of recurrent tonsillitis episodes for at least one year, characterized by sore throat or swollen painful tonsils with fever or symptoms of systemic illness during exacerbations, but the absence of symptoms of possible viral upper respiratory tract infection, such as running nose and cough. As routinely throat cultures were not taken from adults with RT during each exacerbation, the episodes were considered of unknown cause. All patients had been referred for TE by an ENT surgeon from the Department of Otorhinolaryngology. The particular number of tonsillitis episodes per year was not set as an inclusion criterion in the present study. The exclusion criteria were the acute tonsillitis exacerbation, acute respiratory infection, and antibiotic therapy within the two previous months.
Control subjects
The control group consisted of 54 volunteer students (age range 18–24, median 20 years; 36 female and 18 male) who were not suffering from recurrent tonsillitis episodes. The study had approval from the Tartu University Research Ethics Committee, and in each case written informed consent was obtained from each participant.
Collection of history data and oropharyngeal examination
In RT-TE patients, the disease history data such as the number of tonsillitis episodes per year, the length of morbidity period in years, presence of documented comorbid diseases, usage of antibiotics and changes in life quality due to tonsillitis episodes were collected by one examiner (MK Jr), and the oropharyngeal examinations were performed by another (PK), who was blinded to the type of patients seen. In healthy controls, the same examiner conducted the oropharyngeal examinations separately. On oropharyngeal examination the presence or absence of the three characteristic signs of sclerotic process was evaluated: tonsillar sclerosis, obstruction of tonsillar crypts, and scar tissue on the tonsils. Tonsillar sclerosis was defined as increased tightness of tonsillar and peritonsillar tissue together with the fixation of palatine tonsil in the tonsillar fossa. The obstruction of the tonsillar crypts was documented when narrowing of the crypts' mouth was observed resulting in loss of clear cryptic pattern of the tonsillar surface. The scar tissue on the tonsils was defined as white tissue-spots or streaks on the tonsillar surface.
Analysis of S. pyogenes occurrence in core tonsils
Bacteriological analysis
The occurrence of S. pyogenes in tonsillar core tissue was assessed in the first 24 tonsils removed from RT-TE patients. The bacteriological culture was performed according to the previously described method [18]. Briefly, after excision, one of the tonsils was placed in a sterile Petri dish and taken immediately to the microbiology laboratory. For a tonsillar core culture, approximately 0.2 g of tissue was aseptically excised and homogenized in a sterile mortar with a known amount of pre-reduced phosphate-buffered saline (PBS; pH 7.2) in the anaerobic glove box (Sheldon Manufacturing Inc., USA, with a gas mixture: 5% CO2, 5% H2, 90% N2) and was further serially diluted (10-2 – 10-7). The dilutions were seeded on Columbia horse blood agar plates enriched with streptococcus selective supplement (Oxoid Ldt., UK). All plates were incubated for 48 hours at 36°C in an atmosphere enriched with 10% CO2 in Jouan IG150 incubator (Jouan, France). The culture plates were examined for the growth of β-hemolytic streptococci and selected colonies were Gram stained and subjected to microscopy. The β-hemolytic streptococci were distinguished from α-hemolytic streptococci by the type of hemolysis and were grouped using streptococcus latex agglutination test (Oxoid Ltd., UK).
PCR amplification
For molecular detection of S. pyogenes, the total genomic DNA was extracted from the tonsillar tissue samples of 24 selected RT-TE patients according to the previously described method [19]. For the amplification of specific S. pyogenes mitogenic factor (mf) gene [20], the following primers were used: forward, 5'-CTA CTT GGA TCA AGA CGG-3'; and reverse, 5'-TTA GGG TTT CCA GTC CAT CC-3'. The PCR was performed in a 25-ml volume with ~10 ng DNA sample, in an automated thermal cycler (Biometra, Eppendorf) by using a Ready-To-Go PCR Bead (Amersham Pharmacia Biotech Inc., USA). Extracted DNA of S. pyogenes ATCC 19615 served as a positive control.
Transmission electron microscopy
For detection of the putative intracellular location of bacteria, the transmission electron microscopy (EM) of the crypt epithelium of 10 randomly selected tonsils from RT-TE patients was performed. The interactions between epithelial cells, infiltrating nonepithelial cells and bacteria were studied. Approximately 1-mm3 samples from PTs were fixed in 2.5% glutaraldehyde (0.1 M cacodylate puffer, pH 7.4) at 4°C for 2.5 h and postfixed in 1% osmium tetraoxide. After dehydration through an ethanol series and acetone, samples were embedded in epoxy resin. Sections were cut with ultratome MT-LX (RMC, USA). Semithin sections (1 μm) were stained with methylene blue, azure II eosin and basic fuchsin for light microscopy. Ultrathin sections were stained with uranyl acetate and lead citrate and were examined by TEM using Tecnai 10 electron microscope (FEI, Netherlands).
Statistical analyses
Statistical analyses were performed using 'Excel' (Microsoft Corp.) and 'R' (The R Development Core Team) software, employing Chi-square, Mann-Whitney rank sum and Pearson's rank correlation tests. Comparing the presence of sclerotic signs in RT-TE patients and in healthy controls the sensitivity, specificity, positive (PPV) and negative predictive value (NPV) of the signs were calculated. Based on the disease history data and the presence of sclerotic signs, the receiver-operating curve (ROC) and the area under the curve (AUC) were constructed for prediction of sclerotic type tonsils [21]. All differences were considered statistically significant for P-values less than 0.05.
Results
Disease history in RT-TE patients
Out of 62 RT-TE patients, 26 (42%) patients had six or more, 10 (16%) had four to five and 26 (42%) patients had three or less tonsillitis episodes per year. The median number of tonsillitis episodes in the whole group of RT-TE patients was 4.5 per year. The duration of morbidity ranged from 1 to 23 years; the median being 6 years. There was no difference in the length of morbidity between patients with four or more and patients with three or less tonsillitis episodes per year; the median being 7 and 5 years respectively. No correlation was found between the frequency of tonsillitis episodes and the duration of morbidity.
The index of tonsillitis (IT) was calculated by multiplying the number of tonsillitis episodes per year by the morbidity period in years [22]. The median IT in the whole group of RT-TE patients was 30 (range 6–138). The comorbid disease was documented in 14 (22%) RT patients: rheumatic heart disease in 7, unspecified polyarthritis in 5, rheumatoid arthritis and glomerulonephritis both in one patient.
Associations between residing bacteria and epithelial damage
Occurrence of S. pyogenes in tonsillar core tissue
S. pyogenes was not cultivated from any of the core tonsils, but it was found in 29% (7 out of 24) of the diseased tonsillar tissue specimens by PCR.
EM
EM revealed various morphotypes of bacteria on the surfaces of epithelial cells, and many of them were in intimate contact with the cell membrane. Many coccoid forms of bacteria were either penetrating into the cells or were locating completely intracellularly (Figure 1). The bacteria within cells were usually intact and surrounded by cytoplasmatic tonofibrils (Figure 2).
The nonepithelial cells in the intact crypt epithelium, including neutrophilic granulocytes, were tightly packed between epithelial cells (Figure 3). However, in case of damage of tight junctions between epithelial cells, with remaining desmosomes only on the projections, free spaces between adjacent epithelial cells appeared (Figure 4). These gaps were frequently occupied by degenerating granulocytes with intact granules and bacteria (Figure 5).
Optimal cut-off score of IT
As expected, the presence of sclerotic signs on oropharyngeal examination was more common in RT-TE patients than in the healthy controls. The most common sign in RT-TE patients was the scar tissue on tonsils, but it was also frequently found in healthy controls. The tonsillar sclerosis and obstruction of crypts were less frequently found in healthy controls, but were observed in nearly half of RT-TE patients. Accordingly, tonsillar sclerosis had the highest specificity and PPV, while scars on tonsils showed the highest sensitivity and NPV (Table 1). We found that the higher frequency of tonsillitis episodes was in strong correlation with the occurrence of obstructed tonsillar crypts and the longer morbidity period with tonsillar sclerosis and with the presence of S. pyogenes in tonsillar tissue by PCR (Table 2). Further, the higher IT values were expectedly in good correlation with the number of sclerotic signs on oropharyngeal examination (r = 0.325, P = 0.010). The frequency of tonsillitis episodes per year and the length of morbidity period showed no correlation with scars on the tonsillar surface. The presence of comorbid diseases had no association with the sclerotic signs and PCR data.
Based on the IT values and the presence or absence of tonsillar sclerosis and obstruction of tonsillar crypts the ROC curve with AUC was constructed to ascertain the cut-off score of IT. It revealed that an IT score of 36 is an optimal cut-off value for prediction of sclerotic type tonsils (AUC = 0.716). It had a sensitivity of 52.5%, specificity of 86.1%, positive predictive value of 87.5% and negative predictive value of 50.0% (Figure 6).
Finally, out of 26 (42%) RT-TE patients with only three or less tonsillitis episodes per year, 13 had tonsillar sclerosis and obstruction of tonsillar crypts on oropharyngeal examination; one patient had a documented rheumatic fever and another one positive PCR for S. pyogenes. The remaining 11 patients with lower rate of recurrences had only the scars on their tonsils or had no signs of sclerotic process. None of them had documented comorbid diseases or evidences of S. pyogenes persistence in their tonsils.
Discussion
The present study revealed that the higher frequency of tonsillitis episodes per year has a strong correlation with the presence of obstructed tonsillar crypts while the longer disease history correlates well with the presence of tonsillar sclerosis on oropharyngeal examination. Generally, these findings are in accordance with current knowledge of RT pathogenesis. The continuous exacerbations of chronic inflammation in tonsillar tissue come in long term down to parenchymal fibrosis, followed by stenosis of branched, blind-ended and narrow tonsillar crypts [7,8]. The subsequent retention of crypts' contents sets up an ideal culture medium for microorganisms, resulting in the formation of small abscesses, sacks full of different microorganisms. The obstruction of tonsillar crypts and their chronic suppuration potentially promote more easily the exacerbations of chronic inflammation than widely opened and freely drained crypts. However, the present study demonstrated that sclerotic type tonsils can be expected not only in patients with high number of tonsillitis episodes per year, but also in patients with lower number of episodes if combined with long morbidity period. The signs of sclerotic process in tonsils were found in a half of RT-TE patients with only three or less tonsillitis episodes per year. It indicates that a gradual accumulation of exacerbations after long years of suffering is also a factor for the development of sclerotic type tonsils.
As the parenchymal fibrosis leads to lowered count of neutrophils in tonsillar tissue, increasing the risk for spread of bacteria into the bloodstream and infection generalization [9], removal of such functionally compromised tonsils could be justified. However, considering the sclerotic signs as the only indicator for TE, particularly in adults with lower rate of tonsillitis episodes, may lead to an overestimation of the need for surgery. Although the sclerotic signs were very frequently found in RT-TE patients, they were also encountered in a significant proportion of healthy persons. For instance, as the scars were commonly found on tonsils in both groups, it had low specificity and PPV for RT diagnosis. Therefore, the recommendations for TE should be based on detailed disease history, taking both the frequency of tonsillitis episodes per year and the length of morbidity period into account, and the presence of sclerotic signs could only strengthen the decision.
In order to add up different disease history data, the frequency of tonsillitis episodes per year was multiplied by the morbidity period in years. Basically, it represents a total number of tonsillitis episodes the patient has ever had and was called the index of tonsillitis in an earlier study [22]. Although it seemed to be successful to characterize RT severity in adults, a specific cut-off score of IT as an indicator for tonsillectomy was not provided. In the present study, the IT values were compared with the presence or absence of most characteristic sclerotic signs, the tonsillar sclerosis and obstruction of tonsillar crypts, in order to construct the ROC curve for prediction of sclerotic type tonsils. An optimal cut-off score of IT was found to be 36, which had balanced sensitivity, specificity and predictive values. This cut-off score indicates that a minimum of 36 tonsillitis episodes could be enough for the development of sclerotic type tonsils. We suggest that specificity of 86.1% and PPV of 87.5% of this score are high enough to use it for differentiating patients with advanced tonsillitis from less severe cases.
The interesting finding in the present study was that nearly one-third of culture negative tonsillar core specimens were positive for S. pyogenes by PCR. The failure of conventional culture to reveal the presence of S. pyogenes has been attributed to its ability for intracellular penetration [24-26]. EM revealed several coccoid forms of intracellular bacteria in the crypt epithelium of diseased tonsils, which was frequently accompanied with the damage of connections between epithelial cells, called tight junctions. Although the type of intracellular bacteria is unknown, a correlation between the presence of S. pyogenes by PCR and the longer morbidity period of RT suggests that hidden persistence of S. pyogenes in tonsils may in some cases be responsible for continuous inflammation in its tissue and formation of sclerosis.
In the pathogenesis of concomitant inflammatory diseases of other tissues and organs, such as glomerulonephritis and IgA nephropathy, reactive and rheumatoid arthritis, chronic inflammatory and autoimmune neurological disorders, the key role has been attributed to S. pyogenes [27-31]. The high rate of comorbid diseases in our RT-TE patients suggest that TE is often undertaken to eliminate the reservoir of putative pathogen, e.g. S. pyogenes, despite a negative throat culture. Seemingly, one of the solutions to prevent repeated attacks of tonsillar infections in the early stages of RT is to apply treatment plans with antibiotics effective against intracellular bacteria, particularly among patients with a high-risk for comorbidity. Among our patients, the candidates for such a conservative therapy could have been 18% of patients who had three or less tonsillitis episodes, but had neither the signs of sclerotic process on oropharyngeal examination nor supporting comorbidity yet evidence of S. pyogenes in tonsillar tissue.
We conclude an IT score ≥36, being a combination between the frequency of tonsillitis and the length of morbidity period, predicts the sclerotic process in recurrently inflamed tonsils. Therefore, the high IT values could serve as an indicator for TE in adults. The correlation between the longer morbidity period and the presence of S. pyogenes by PCR suggests that persistent infection may have a role in maintenance of recurrent inflammation in tonsils.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
PK was principal investigator, carried out oropharyngeal examinations, collected samples for microbiological and molecular studies, performed PCR reactions, participated in electronmicroscopical studies, and drafted the manuscript. AP carried out the electronmicroscopical studies and interpreted the data. MK participated in the design of the study and revised critically the manuscript. MK Jr. carried out collection of disease history data from patients and participated in microbiological and molecular studies. MM coordinated the study and helped to draft the manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors wish to thank the members of the Department of Otorhinolaryngology, Tartu University Clinicum for their contribution and providing study patients, and Krista Fischer from the Department of Public Health, Tartu University for excellent help performing the statistical analyses. This work was supported by Grant No. 4898 from the Estonian Science Foundation, Estonian Target funding No. 0418 from the Estonian Ministry of Education and the Centre of Molecular and Clinical Medicine, Faculty of Medicine, University of Tartu.
Figures and Tables
Figure 1 Transmission electron microscopy of the crypt epithelium of palatine tonsils showed coccoid forms of bacteria within epithelial cells (original ×10000).
Figure 2 High power magnification showed that intracellular bacteria were surrounded by cytoplasmatic tonofibrils (original ×73000).
Figure 3 The intact crypt epithelium contained neutrophilic granulocytes and other nonepithelial cells, which were tightly packed between epithelial cells (original ×2100).
Figure 4 Damage of tight junctions between epithelial cells, with remaining desmosomes only on the projections, led to formation of free spaces between adjacent epithelial cells (original ×7000).
Figure 5 High power magnification showed that the gaps between epithelial cells were occupied by degenerating granulocytes and their intact granules (original ×27000).
Figure 6 The ROC curve of IT scores for prediction of sclerotic type tonsils. An optimal cut-off score of IT was 36 (AUC = 0.716), with sensitivity of 52.5%, specificity of 86.1%, positive predictive value of 87.5% and negative predictive value of 50.0%.
Table 1 Prevalence of sclerotic signs in RT-TE patients and healthy controls with their sensitivity, specificity and predictive values.
Prevalence, n (%)
Signs of sclerotic process Patients
(n = 62) Healthy controls
(n = 54) Sensitivity Specificity PPVa NPVb
Tonsillar sclerosis 29 (47) 2 (4) 0.47 0.96 0.94 0.61
Crypts' obstruction 34 (55) 8 (15) 0.55 0.85 0.81 0.62
Scar tissue on tonsils 49 (79) 11 (20) 0.79 0.80 0.82 0.77
RT-TE patient – patients who were referred for tonsillectomy due to recurrent tonsillitis episodes, PPV – positive predictive value, NPV – negative predictive value
Table 2 Correlation between the patients' disease history data, the presence of sclerotic signs in tonsils and PCR data on Streptococcus pyogenes.
Signs of sclerotic process
History data Tonsillar sclerosis Obstruction of crypts All three signs PCR for S. pyogenes
Frequency of tonsillitis episodes NS Rp = 0.354
P = 0.005 Rp = 0.299
P = 0.018 NS
Morbidity period Rp = 0.437
P = 0.001 NS Rp = 0.318
P = 0.011 Rp = 0.503
P = 0.012
Index of tonsillitis Rp = 0.384
P = 0.002 NS Rp = 0.325
P = 0.01 NS
Rp – Pearson correlation coefficient. NS – statistically nonsignificant correlation
==== Refs
American Academy of Otolaryngology-Head and Neck Surgery
Scottish Intercollegiate Guidelines Network Management of sore throat and indications for tonsillectomy A national clinical guideline Edinburgh 1999
Mui S Rasgon BM Hilsinger RL Efficiacy of tonsillectomy for recurrent throat infection in adults Laryngoscope 1998 108 1325 1328 9738750 10.1097/00005537-199809000-00012
Bhattacharyya N Kepnes LJ Shapiro J Efficacy and quality-of-life impact of adult tonsillectomy Arch Otolaryngol Head Neck Surg 2001 127 1347 1350 11701072
Bhattacharyya N Kepnes LJ Economic benefit of tonsillectomy in adults with chronic tonsillitis Ann Otol Rhinol Laryngol 2002 111 983 988 12450171
Perry ME Whyte A Immunology of the tonsils Immunol Today 1998 19 414 421 9745205 10.1016/S0167-5699(98)01307-3
Michaels L Ear, nose and throat histopathology 2001 New York: Springer-Verlag
Altemani A Endo LH Chone C Idagawa E Histopathological concept of chronic tonsillitis in children Acta Otolaryngol (Stockh) 1996 14 16
Kasenõmm P Mesila I Piirsoo A Kull M Mikelsaar M Mikelsaar R-H Macroscopic oropharyngeal signs indicating impaired defensive function of palatine tonsils in adults suffering from recurrent tonsillitis APMIS 2004 112 248 256 15233639 10.1111/j.1600-0463.2004.apm11204-0504.x
Surow JB Handler SD Telian SA Fleisher GR Baranak CC Bacteriology of tonsil surface and core in children Laryngoscope 1989 99 261 266 2645491
Gaffney RJ Freeman DJ Walsh MA Cafferkey MT Differences in tonsillar core bacteriology in adults and children: a postoperative study of 262 patients Respir Med 1991 85 383 388 1759002
Mitchelmore IJ Reilly PG Hay AJ Tabaqchali S Tonsil surface and core cultures in recurrent tonsillitis: prevalence of anaerobes and beta-lactamase producing organisms Eur J Clin Microbiol Infect Dis 1994 13 542 548 7805681 10.1007/BF01971304
Brook I Yokum P Foote PA Changes in the core tonsillar bacteriology of recurrent tonsillitis: 1977–93 Clin Infect Dis 1995 21 171 176 7578726
Stjernquist-Desatnik A Holst E Tonsillar microbial flora: comparison of recurrent tonsillitis and normal tonsils Acta Otolaryngol (Stockh) 1999 119 102 106
Brook I Yocum P Bacteriology of chronic tonsillitis in young adults Arch Otolaryngol 1984 110 803 805 6334513
Lildholdt T Doessing H Lyster M Outzen KE The natural history of recurrent acute tonsillitis and a clinical trial of azitromycin for antibiotic prophylaxis Clin Otolaryngol 2003 28 371 373 12871256 10.1046/j.1365-2273.2003.00728.x
Podbielski A Beckert S Schattke R Leithauser F Lestin F Gossler B Kreikemeyer B Epidemiology and virulence gene expression of intracellular group A streptococci in tonsils of recurrently infected adults Int J Med Microbiol 2003 293 179 190 12868654 10.1078/1438-4221-00253
Kasenõmm P Kull M Mikelsaar M Association between tonsillar core microflora and post-tonsillectomy bacteremia Microb Ecol Health Dis 2002 14 122 127 10.1080/08910600260081784
Louie L Simor AE Louie M McGeer A Low DE Diagnosis of group A streptococcal necrotizing fasciitis by using PCR to amplify the streptococcal pyrogenic exotoxin B gene J Clin Microbiol 1998 36 1769 1771 9620418
Iwasaki M Igarashi H Hinuma Y Yutsudo T Cloning, characterization and overexpression of a Streptococcus pyogenes gene encoding a new type of mitogenic factor FEBS Lett 1993 331 187 192 8405402 10.1016/0014-5793(93)80323-M
Van der Schouw YT Verbeek AL Ruijs JH ROC curves for the initial assessment of new diagnostic tests Fam Pract 1992 9 506 511 1490547
Fujihara K Goto H Hotomi M Kobayashi M Hayashi M Tamura S Kuki K Yamanaka N Immunological derangement in tonsils with recurrent infections. A study of co-stimulatory factors on tonsillar B lymphocytes International Concress Series 2003 1257 49 53 10.1016/S0531-5131(03)01620-0
Becker W Naumann HH Pfaltz CR Ear, nose, and throat diseases 1994 New York: Thieme Medical Publishers Inc
La Penta D Rubens G Chi E Cleary PP Group A streptococci efficiently invade human respiratory epithelial cells Proc Natl Acad Sci USA 1994 91 12115 12119 7991594
Osterlund A Popa R Nikkila T Scheynius A Engstrand L Intracellular reservoir of Streptococcus pyogenes in vivo: a possible explanation for recurrent pharyngotonsillitis Laryngoscope 1997 107 640 647 9149167 10.1097/00005537-199705000-00016
Norrby-Teglund A Kotb M Host-microbe interactions in the pathogenesis of invasive group A streptococcal infections J Med Microbiol 2000 49 849 852 11023181
Bisno AL Group A streptococcal infections: the changing scene Curr Opin Infect Dis 1995 8 117 122
Kobayashi S Tamura N Akimoto T Ichikawa G Xi G Takasaki Y Hashimoto H Reactive arthritis induced by tonsillitis Acta Otolaryngol Suppl 1996 523 206 211 9082784
Swedo SE Leonard HL Garvey M Mittleman B Allen AJ Perlmutter S Pediatric autoimmune neuropsychiatric disorders associated with streptococcal infections: clinical description of the first 50 cases Am J Psychiatry 1998 155 264 271 9464208
Harsha WJ Goco PE Crawford JV Remission of chronic inflammatory demyelinating polyneuropathy following tonsillectomy Ear Nose Throat J 2003 82 520 521 12955838
Kawano M Okada K Muramoto H Morishita H Omura T Inoue R Katajima S Katano K Koni I Mabuchi H Yachie A Simultaneous, clonally identical T cell expansion in tonsil and synovium in a patient with rheumatoid arthritis and chronic tonsillitis Arthritis Rheum 2003 48 2483 2488 13130467 10.1002/art.11212
|
16159391
|
PMC1242222
|
CC BY
|
2021-01-04 16:03:26
|
no
|
BMC Ear Nose Throat Disord. 2005 Sep 13; 5:7
|
utf-8
|
BMC Ear Nose Throat Disord
| 2,005 |
10.1186/1472-6815-5-7
|
oa_comm
|
==== Front
BMC Ear Nose Throat DisordBMC Ear, Nose and Throat Disorders1472-6815BioMed Central London 1472-6815-5-91616806510.1186/1472-6815-5-9Case ReportAcute unilateral hearing loss as an unusual presentation of cholesteatoma Thio Daniel [email protected] Shahzada K [email protected] Richard C [email protected] Department of Otorhinolaryngology, South Warwickshire General Hospitals NHS Trust Warwick CV34 5BW UK2 Department of Otorhinolaryngology, South Warwickshire General Hospitals NHS Trust Warwick CV34 5BW UK3 Department of Otorhinolaryngology, South Warwickshire General Hospitals NHS Trust Warwick CV34 5BW UK2005 18 9 2005 5 9 9 10 7 2005 18 9 2005 Copyright © 2005 Thio et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Cholesteatomas are epithelial cysts that contain desquamated keratin. Patients commonly present with progressive hearing loss and a chronically discharging ear. We report an unusual presentation of the disease with an acute hearing loss suffered immediately after prolonged use of a pneumatic drill.
Case presentation
A 41 year old man with no previous history of ear problems presented with a sudden loss of hearing in his right ear immediately following the prolonged use of a pneumatic drill on concrete.
The cause was found to be a fractured long process of incus which had been eroded by the presence of an attic cholesteatoma.
A tympanomastoidectomy and ossiculoplasty was performed with good result.
Conclusion
Cholesteatomas may be asymptomatic and insidious in their onset. This case illustrates the point that an indolent disease such as this may present in unusual ways and the clinician must always have a high index of suspicion combined with thorough assessment and examination of every patient.
==== Body
Background
The definition of cholesteatoma is the occurrence of keratinizing, stratified, squamous epithelium within the middle-ear cavity where otherwise only modified respiratory epithelium ought to be present [1]. Patients normally present with a chronically discharging ear and may complain of hearing loss. We present the case of a 41 year old man whose primary presentation of cholesteatoma was a sudden unilateral hearing loss following extended use of a pneumatic drill – a feature not associated with chronic middle ear disease. This mode of presentation has never previously been reported in the literature.
Case presentation
A 41-year-old gentleman was referred to the otolaryngology outpatient clinic with a one-month history of acute right-sided hearing loss. At the time of onset, he noticed a sudden loss of hearing in his right ear immediately following the prolonged use of a pneumatic drill on concrete without the benefit of ear protection. There was no previous history of noise exposure or any ear problems.
Assessment of the right tympanic membrane revealed a retraction pocket in the attic containing dry cholesteatoma with the suspicion of the cholesteatoma passing postero-inferiorly. His left tympanic membrane was normal at otoscopy. Weber testing lateralised to the right side and Rinne's test was negative on the right using a 512Hz tuning fork. The rest of the examination was unremarkable. Pure tone audiometry confirmed a 45-decibel mean conductive hearing loss on the right (Figure 1) with normal hearing levels in the left ear.
He subsequently underwent a right tympanomastoidectomy and ossiculoplasty. Operative findings confirmed extensive erosion and fracture of the long process of incus (LPI) by a moderate size cholesteatoma. The stapes footplate was mobile. Following careful dissection and removal of the entire cholesteatoma together with the head of the malleus, ossicular continuity was restored with a Goldenberg partial ossicular replacement prosthesis connecting the stapes suprastructure to the handle of malleus. A repeat audiogram 7 weeks post-surgery showed a 15-decibel improvement in his right ear (Figure 2). The patient went on to make a good recovery with no recurrence of cholesteatoma at 12 months with good hearing thresholds
Discussion
The history of acute noise exposure without ear protection may normally be expected to result in a bilateral sensorineural hearing loss. Furthermore, our patient had no previous history of otological problems, which is unusual in somebody with such extensive cholesteatoma. His presenting complaint of sudden hearing loss was a result of vibration from the pneumatic drill, fracturing the already eroded long process of incus (LPI), thus resulting in immediate ossicular discontinuity and the subsequent sudden hearing loss. This was confirmed at operation.
Cholesteatomas are keratin-containing epidermoid cysts that classically arise from the pars flaccida or postero-superior segment of the tympanic membrane. They have the propensity to expand into the middle ear cleft (MEC) and beyond, leading to both intracranial and extracranial complications. This is commonly compounded by the presence of infection.
Clinical features vary and arise as a result of the disease itself or its complications. The most common presenting features are an offensive smelling discharge and hearing loss [2]. Hearing loss may be a feature due to ossicular chain disruption or may result from accumulation of toxic inflammatory mediators which pass through the round window into the cochlea. Dizziness and vertigo may be due to the presence of a labyrinthine fistula [3].
Signs include crusting in the attic region, granulomatous polyps and marginal granulomas, the presence of keratin debris in a retraction pocket, and marginal and attic perforations [3,4].
Management of cholesteatomas is primarily surgical. The aim of surgery is to convert unsafe disease to safe disease, with restoration of hearing a secondary priority. Surgery is directed toward the eradication of entrapped keratinising epithelium and debris from the middle ear and mastoid air spaces and tympanomastoidectomy is the operation of choice [5]. The decision as to whether to perform an ossiculoplasty at the initial operation depends on the findings at surgery and the surgeon's preference.
Conclusion
This case emphasises the point that cholesteatomas can be largely asymptomatic. It was fortuitous that vibrations from a pneumatic drill caused the already eroded LPI to fracture, facilitating the diagnosis and treatment of this indolent and potentially dangerous condition. A presentation of such unlikely coincidences has not been reported in the literature. The clinician must always have a high index of suspicion combined with thorough assessment and examination of every patient to ensure that this indolent disease is not missed.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
DT drafted the manuscript and prepared the figures
SKA revised the manuscript
RCB performed the ossiculoplasty and proof-read the manuscript
All authors have read and approved of the manuscript
Pre-publication history
The pre-publication history for this paper can be accessed here:
Figures and Tables
Figure 1 Pre-operative pure tone audiogram.
Figure 2 Post-operative pure tone audiogram.
==== Refs
Stenfors LE Does occurrence of keratinizing stratified squamous epithelium in the middle-ear cavity always indicate a cholesteatoma? J Laryngol Otol 2004 118 757 763 15550180 10.1258/0022215042450805
Sheahan P Donnelly M Kane R Clinical features of newly presenting cases of chronic otitis media J Laryngol Otol 2001 115 962 966 11779324 10.1258/0022215011909774
Lesser THJ Roland NJ, McRae RDR, McCombe AW Cholesteatoma Key topics in Otolaryngology 2001 2 Aberystwyth: BIOS Scientific Publishers 35 37
Kerr AG Booth JB editors Scott-Brown's Otolaryngology Otology 1997 3 6 Bath: Butterworth-Heinemann Publishers
Schroeder A Darrow DH Management of the draining ear in children Pediatric annals 2004 33 843 853 15615311
|
16168065
|
PMC1242223
|
CC BY
|
2021-01-04 16:03:26
|
no
|
BMC Ear Nose Throat Disord. 2005 Sep 18; 5:9
|
utf-8
|
BMC Ear Nose Throat Disord
| 2,005 |
10.1186/1472-6815-5-9
|
oa_comm
|
==== Front
BMC GastroenterolBMC Gastroenterology1471-230XBioMed Central London 1471-230X-5-291616474610.1186/1471-230X-5-29Research ArticleHigh rates of early HBeAg seroconversion and relapse in Indian patients of chronic hepatitis B treated with Lamivudine: results of an open labeled trial Alexander George [email protected] Chalamalasetty S [email protected] Kamal [email protected] TS [email protected] Gourdas [email protected] Department of Gastroenterology, Sanjay Gandhi Postgraduate Institute of Medical Sciences, Lucknow, India2005 15 9 2005 5 29 29 31 12 2004 15 9 2005 Copyright © 2005 Alexander et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The use of Lamivudine in chronic hepatitis B (CHB) is well known, however the reported rate of HBeAg sero-conversion and its durability post-treatment have varied considerably. We undertook the present study to study the effect of Lamivudine on HBeAg loss and seroconversion rates in Indian patients of CHB in relation to frequency, predictors and durability.
Methods
We treated 60 patients of e antigen positive CHB (with active viral replication and ongoing necro-inflammatory activity) with Lamivudine. They were followed up by monthly aminotransferases, and 3 monthly HBeAg and anti-HBe. Those who attained HBeAg sero-conversion were advised to discontinue Lamivudine after 6 months and followed up every 3 months thereafter, to see for relapse. Treatment was given for maximum of 3 years if not sero-converted.
Results
The annual incremental loss of HBeAg in patients receiving Lamivudine was 25 (41.6%) at end of 1st year, 33 (55%) at 2nd year and 35 (58.3%) at 3rd year. The corresponding rates for full sero-conversion were 17/60 (28.6%), 22/60 (36.6%) and 24/60 (40%) in the 3 years. HBeAg loss correlated with increased pre-therapy ALT levels (p = 0.002) and decreased pretreatment HBV-DNA levels (p = 0.004). The presence of cirrhosis had no influence on the rate of HBeAg loss. Relapse occurred in 35% (7/20) post-treatment at median time of 6 months.
Conclusion
Indian patients showed a higher rate of HBeAg sero-conversion in the first year of Lamivudine treatment. This correlated with baseline ALT and inversely with HBV-DNA levels. Relapse rate after treatment was high and occurred soon after stopping treatment.
==== Body
Background
Hepatitis B virus infects more than 300 million people worldwide, contributing to debilitating illness and death, and more than 75% of those affected are of Asian origin [1]. Chronically infected persons with viral replication are at highest risk for progressive liver disease and it is estimated that up to 25% of persons with chronic hepatitis B virus infection will die prematurely of cirrhosis or hepatocellular carcinoma. Cirrhosis and hepatocellular carcinoma account for more than 50% of deaths in Asian men with chronic Hepatitis B infection [2].
The risk of chronicity after acute HBV infection is low in immuno-competent adults and is reported to be less than 5%[3]. However, the natural history of HBV infection differs between Asian and Western patients. Asians usually become infected perinatally, rarely have an acute hepatitis-like clinical syndrome but almost invariably remain chronically infected and run substantial risk for developing cirrhosis and hepatocellular carcinoma. Western patients who are usually infected in adult life by percutaneous or sexual exposure, typically have an acute hepatitis-like clinical illness, clear the virus and rarely become chronically infected [4,5]. Chronically infected persons with evidence of active viral replication are at highest risk for progressive liver disease. Cirrhosis develops in 15 to 20 % of them within 5 years, even if histologic liver damage is initially mild [6]. HBeAg seroconversion, which can occur spontaneously or post-treatment, is associated with a substantial reduction in the risk of liver failure [7]. Spontaneous sero-conversion occurs only in 7 to 20 % per year [8].
Interferon -alfa was the first drug specifically licensed to treat chronic Hepatitis B. The efficacy of Interferon-alfa was found to be variable, but a meta-analysis showed that 33% receiving it lost HBeAg compared with 12% of untreated controls [9]. Interferon-alfa was least effective in Asian patients [10]. It has to be given by injection and has potential dose-limiting side effects. Patients with liver cirrhosis often deteriorated on initiating treatment [7]. Enthusiasm with Interferon for treatment for chronic hepatitis B therefore waned with the wide availability and use of Lamivudine.
Lamivudine, an oral nucleoside analogue, inhibits viral DNA replication. In doses of 100 mg/day in adults (1.5 mg/Kg/day in children), the median suppression of serum HBV-DNA is greater than 98% in most patients during treatment [11]. It produces rapid decrease in serum HBV- DNA and aminotransferase levels, improves liver histology and enhances the rate of HBeAg loss compared with placebo treatment [12,13]. The virologic end points of treatment have been the sustained disappearance of serum HBV DNA by a conventional hybridization assay and either the disappearance of hepatitis B e antigen from serum [HBeAg loss] or the loss of HBeAg accompanied by the detection of anti-HBe [HBeAg sero-conversion] [11,12]. The proportion of patients achieving HBeAg sero-conversion after 12 months of Lamivudine treatment [100 mg/day] has ranged between 16% and 18% in Western and Asian studies [11-13]. Several workers have shown pre-treatment high serum ALT and low serum HBV-DNA levels to be independently associated with increased rate of HBeAg loss and sero-conversion when treated with Lamivudine [13,14].
The safety of Lamivudine has led to the suggestion that continuous therapy may be beneficial, particularly in patients who do not lose HBeAg. Extended treatment with Lamivudine beyond 1 year has shown good results in various studies with results showing an incremental response in HBeAg sero-conversion rate [15,16]. The study by Leung NW et al showed sero-conversion rates of 22% after 1 year, increasing to 29% after 2 years and 40% after 3 years [16]. A major limitation of chronic therapy however is the development of viral resistance, marked virologically by rise in HBV DNA levels despite continuation of therapy and clinically by increases in serum transaminases [11,12,15,17]. Lamivudine induced HBeAg seroconversion was reported to be durable by several Western studies, 80 to 90%[12] and 73%[18]. However, studies in Asian countries show that relapse rates are much higher post treatment, 37.5% at 1 year [19] and 45.8%[20].
It has been observed that Hepatitis B virus may behave differently in different geographic regions [19,21]. This may be due to some host factors or to the viral genotypic differences [21]. Therefore it is important for each country to determine to determine its own rate and pattern of seroconversion following treatment of chronic hepatitis B. The aims of the present study was to study the effect of Lamivudine on chronic hepatitis B in Indian patients with regard to 1) rate of seroconversion and HBeAg loss 2) predictors of seroconversion 3) the durability of seroconversion post treatment.
Methods
Patients
Sixty patients [50 men and 10 women, median age 40 years, range 4–80 years, cirrhosis 23] with HBeAg positive chronic HBV infection who were started on Lamivudine 100 mg per day (1.5 mg/Kg/day in children) during the time period from August 1998 to June 2001 were followed up. All patients were positive for HBsAg for more than six months, had active replicative status [HBeAg positive, AntiHBe negative and HBV-DNA positive by PCR technique] with alanine aminotransferase levels that were less than 10 times the upper limit of normal for at least the previous 3 months. All had evidence of ongoing necro-inflammatory activity either on liver biopsy (knodell ishak score>4) or suggested by raised alanine aminotransferase levels. Cirrhosis was diagnosed on basis of clinical [evidence of portal hypertension, liver decompensation], biochemical, endoscopic [varices] and imaging [ultrasound] evidence. Patients were excluded if they had any of the following: previous antiviral treatment for hepatitis B; immunomodulatory drugs or corticosteroids within 6 months before Lamivudine treatment; co infection with hepatitis C virus or the human immunodeficiency virus; or the presence of other types of liver disease.
The study was approved by the local ethics committee and all patients provided informed consent before treatment.
Methods
Lamivudine was given at a dose of 100 mg/day during the study period. Serum ALT was checked monthly and serum HBeAg, anti-HBe [measured by the commercially available ELISA kit; Organon Teknika] 3 monthly till seroconversion. HBV DNA testing by the In-house PCR technique was done to confirm viral suppression, initially at 3 months and then annually. It was also done at the time of seroconversion or biochemical breakthrough. HBV-DNA levels were done by Quantiplex branched DNA assay at baseline and at 3 months. HBeAg seroconversion was defined as disappearance of HBeAg and appearance of anti-HBe antibody, while HBeAg loss was defined as disappearance of HBeAg only. Viral breakthrough was identified on the basis of ALT rise (greater than 2 × upper limit of normal) with re-emergence of HBV-DNA positivity (by PCR) and detectable levels of HBV DNA. HBeAg seroconversion was confirmed by repeat testing after 3 months. Lamivudine was continued for 6 months after seroconversion was achieved or for maximum of 3 years. Thereafter post treatment monitoring in those who had seroconverted continued by monthly ALT and 3 monthly HBeAg, anti- HBe. This was done to look for relapse, which was defined as re-emergence of HBV DNA positivity by PCR technique and/or HBeAg positivity after Lamivudine was stopped post seroconversion.
Statistics
The baseline factors evaluated were age, sex, BMI, weight, ALT, HBV DNA levels and presence of cirrhosis. Data were expressed as median (range) or mean+/- SD. For statistical significance, nominal variables were analyzed by Chi square test with Yates correction. For numerical variables Wilcoxon Rank Sum Test was used, as the data was not expected to have a Gaussian distribution. A p value of < 0.05 was taken as significant.
Results
Patient population
Sixty patients who received Lamivudine were followed up. All of them completed at least 1 year of treatment. In the second year 4 patients who had not seroconverted, did not come for follow up and similarly in the 3 rd year 6 patients dropped out. [See table 1 for baseline characteristics] The majority, 50 [83.3%], were men, and median age was 40 years [range 4–80 years]. The mode of transmission of HBV was unknown in most [61.7%] with blood transfusion history being present in 11.7% and history of hemodialysis in 13.3%. Elevated ALT was present in 50 [83.3%] of the 60 patients. The median ALT level was 72 U/L [range 27 to 394] and median HBV DNA level 920 [range: 0.8–4500] mEq/ml. The median × Upper limit of normal of ALT was 1.8. Cirrhosis was present in 23 [38.3%] of the patients.
HBeAg response
Sero-conversion of HBeAg to anti HBe occurred in 17 of 60 patients (28.33%) at the end of first year and incrementally rose to 36.6%(22/60) by second year and to 40%(24/60) by third year as shown in Figure 1. Of all sero-conversions maximum occurred in the first year (17/24) 71%, 21% in the 2nd year and only 8% in the 3rd year. The rate of HBeAg loss was 41.66%(25/60) in first year and incrementally rose to 58.3% (35/60) by third year of treatment. Similar to that seen for seroconversion, maximum loss of HBeAg occurred in the 1st year (71.42%) as compared to 2nd and 3rd years of treatment (22.8% and 5.71 % respectively). Onset of seroconversion occurred at a mean of 10.17 months after starting treatment.
Pretreatment factors influencing HBeAg loss
The association of age, sex, weight, BMI, baseline ALT, baseline HBV-DNA level and presence of cirrhosis with HBeAg loss was analyzed. [See table 2] Only baseline ALT and HBV DNA level were associated significantly with HBeAg clearance, with median ALT among those who lost their HBeAg being 94 U/L compared with median ALT in those who did not, 45 U/L (p = 0.002). Likewise the median HBV DNA level was 111.3 mEq/ml among those who lost their HBeAg as compared to that in those who did not, 958 mEq/ml (p = 0.004). Table 3 depicts the frequency of HBeAg loss and seroconversion according to baseline ALT level. HBeAg response rates increased with increase in level of pretreatment ALT. Among patients with pretreatment ALT levels greater than 1 to 2 times the ULN, HBeAg loss occurred in 50%, which increased to 70% among those with ALT levels greater than 2 to 5 times the ULN. The rate of HBeAg loss was highest among those with ALT levels greater than 5 times ULN, occurring in 80 %. Similar trends were observed with HBeAg seroconversion, though seroconversion occurred less frequently than HBeAg loss. Figure 2 depicts the HBeAg seroconversion rates year wise according to baseline ALT levels. After 1 year, 15% (3 of 20) of patients with baseline serum ALT >1–2 × ULN, 40% (8 of 20) with ALT>2 – 5 × ULN and 60% (6 of 10) had achieved seroconversion, increasing to 30% (6 of 20), 55% (11 of 20) and 70% (7 of 10) respectively after 3 years of treatment.
ALT normalization and HBV-DNA levels
Alanine aminotransferase normalization occurred in 56% (28/50) of patients with elevated baseline ALT levels. This occurred more significantly in those who cleared their HBeAg (71.8%) than those who did not (27.7%), p = 0.02 [see table 2] Among the 24 who seroconverted, in 22 (91%), ALT normalization occurred in the first year. HBV DNA levels became undetectable within 3 months after initiating Lamivudine treatment in 19 out of 21 patients (90.4%) in whom it was done. The 2 patients in whom it was still detectable, one of whom had chronic renal failure and was on maintenance peritoneal dialysis, did not attain seroconversion on long-term treatment.
Follow up and relapse
Median follow-up of 8 months [Range 3–18 months] was done in 20 patients who had seroconverted and stopped Lamivudine. Two patients did not come for follow up after serconversion and in the other two, minimum follow up of 3 months was not available at time of analysis. Relapse was seen in 7 patients (35%), which occurred at a median of 6 months after stopping treatment [Range 3–8 months].
Breakthrough
Breakthrough infection [i.e. re- emergence of DNA positivity, increase in viral load and increase in ALT] during treatment occurred in 6 patients out of the 25 (24%) who did not achieve HBeAg loss. These 6 had undetectable DNA levels initially at 3 months after starting Lamivudine. The onset of viral breakthrough was at 10 months in 2, 15 months in 1, 18 months in 1, and at 27 months in 2. None of these patients achieved seroconversion on long-term Lamivudine and were persistently HBeAg positive.
Discussion
The present study in Indian patients show high seroconversion rates in the first year (28.6%), reaching 40% at end of 3 years. The first year rate is higher than previous Western and Asian studies but by the 3rd year the cumulative rate becomes similar to that mentioned by Leung NW et al [16]. The rate of HBeAg loss, cumulatively rising from 36.6% in the first year to 58.3% by third year is also more than that showed by Perrillo RP et al, 25% in the first year [13] and Dienstag JL et al, 32%[12]. These results clearly indicate that Indian patients have higher HBeAg loss and seroconversion rates in the first year of treatment but the cumulative response seen on extended treatment is not much and by 3 years, the seroconversion rate reaches only 40%, which is mentioned in the previous studies. Seroconversion rates were further enhanced in patients with elevated pretreatment ALT, reaching 60% by 3 years in those with baseline ALT>2 × ULN. The 1st year seroconversion rate in these patients is higher than that seen by Leung NW et al (27%) but again; by the 2nd and 3rd years the rates become similar [16]. It is also seen that the maximum percentage of seroconversions occurred in the first year (70.8% of all seroconversions), which is more than that seen by Leung et al, 56.5% [16]. It is not clear why the seroconversion rate in Indian patients, as seen in the present study is different from the previous published Western and Asian rates. The baseline median ALT concentration in the present study (1.8 × ULN) is comparable to that in previous studies, [14,16] and less than reported by Perrillo et al (2.2 × ULN)[13]. Also taking only those with elevated ALT, the 1st year seroconversion rate in the in the present study is still higher than that mentioned in previous studies [12,16]. The mean HBV DNA level (920 mEq/ml) is actually more than that seen by Lau et al (587 mEq/ml)[15]. Therefore differences in these baseline characteristics that influence seroconversion are not the reason for the higher seroconversion rate seen. Differences in the HBV genotype can account for changing patterns of HBV response to treatment, in different geographic locations and is mentioned later.
The effect that progressively higher levels of pretreatment ALT had on HBeAg loss was striking. Previous studies with Lamivudine have shown that there is a significant correlation between pretherapy ALT levels and HBeAg seroconversion as well as HBeAg loss. Among patients with pretreatment ALT 2–5 × ULN, 70% achieved HBeAg loss and at the highest level (> 5 × ULN), 80% experienced HBeAg loss. This is much more than seen by Chien RN et al -64%[14] and Perrillo RP et al; 56%[13], but similar to that seen by Liaw YF et al; 80%[23]. As the number of patients with ALT levels greater than 5 times the ULN was relatively small, a larger sample size would be required for a more accurate estimate. As ALT elevations in patients with chronic hepatitis B are the results of T-cell -mediated hepatocytolysis, [24] the level of ALT elevation reflects the level of T-cell immune response of the patients to HBV. As also shown by the present analysis, antiviral agents like Lamivudine are more effective, in terms of HBeAg seroconversion, in patients who have mounted an ongoing immune response to HBV. The high HBeAg seroconversion in patients with high baseline ALT levels seems to the result of a concerted effort of (1) the immune -mediated killing of the hepatocytes harboring cccDNA by the antiviral defenses of the host, and (2) the potent direct antiviral effect of Lamivudine and the enhanced CD-4 responses resulting from Lamivudine therapy. This suggests that alternative treatment strategies need to be defined for patients in the immune tolerance phase or with a low anti-HBV immune response. The present study also showed significant correlation between low baseline HBV-DNA levels and HBeAg loss. This has been shown in previous studies [11-15]. That the other baseline factors like age, sex, BMI and weight had no effect on seroconversion has also been seen in previous studies [13,14].
Some studies have shown that the presence of cirrhosis could be a predictor of HBeAg sero-conversion [13,14]. As the number of patients with cirrhosis was high (38 %) in the present study as compared to previous studies, we had a doubt whether it was contributing to the high sero-conversion rates seen. But on analysis, in the present study, presence of cirrhosis had no statistical significant correlation with HBeAg loss.
Normalization of ALT was seen to occur more significantly in patients who had HBeAg loss than in those without, as seen in previous studies [15,16,23]. A disturbing finding in the present study was the high relapse rate of 35% post treatment. This is in variance with Western studies showing durability of HBeAg seroconversion post treatment [12,18]. However studies from South East Asia have reported similar high relapse rates [19,20]. The cause of the high relapse rate is not clear. It may be caused by immune tolerance, which is caused by a long-standing viral infection [25,26]. It has been suggested that Lamivudine treatment can restore immune response to HBV with reduction in viral load [27]. However, long-standing infections in vertically transmitted patients may make this immune response incomplete [25,26]. Even in patients with spontaneous HBeAg seroconversion, frequent relapses were observed in patients with long-standing HBV infection [28]. The duration of additional Lamivudine treatment after HBeAg seroconversion and pretreatment HBV-DNA levels are 2 independent predictive factors for relapse [19]. Although Lamivudine can inhibit viral replication, it cannot eliminate covalently closed circular DNA (cccDNA) in hepatocytes [29]. Studies using in vitro and in vivo model systems have shown that chronic infection is maintained by the cccDNA in hepatocytes [29]. The minimum half-life of the infected cells was estimated to exceed 10 to 100 days. Therefore it was suggested that prolonged treatment over 12 months might be needed till viral clearance, otherwise the chances for relapse [30].
Although the HBeAg seroconversion rate was high in our study, it was not durable. This observation quite contrasts with results in Western countries, in which the therapeutically induced seroconversion is usually maintained [12,18]. Similar high relapse rates were seen in the study by Song et al in Korean patients [19]. The cause of the high relapse rate after HBeAg seroconversion is not clear. It may be caused by immune tolerance, which is caused by a long-standing viral infection [25,26]. It has been suggested that lamivudine therapy can restore immune response to HBV with reduction of viral load. However, long standing infections in vertically transmitted patients may make this immune response incomplete [25,26]. Even in patients with spontaneous HBeAg seroconversion, frequent relapses were observed in patients with long-standing HBV infection [28]. Therefore, it can be suggested that HBeAg seroconversion does not necessarily guarantee prolonged suppression of HBV infection in those endemic areas in which perinatal transmissions are common.
Viral breakthrough was seen in 6 patients. In all of these six, HBV DNA had become undetectable initially at 3 months. All six did not achieve seroconversion with Lamivudine treatment. Sequencing of HBV genome for YMDD mutations was not performed to confirm viral resistance emerging. Resistance to Lamivudine typically develops after 6 months of treatment and is associated with mutations in the highly conserved catalytic region of the HBV polymerase gene [15]. Previous studies show the development of resistance in a high proportion of patients [31,32]. In studies from Asia, resistance was reported to occur in 17% of patients after 1 year [11], 26% after 2 years [33] and 49% after 3 years of treatment [34]. In the Indian study by Wakil et al, frequency of emergence of YM5521/VDD mutations was 29% and presence of normal ALT and low levels of HBV-DNA did not exclude the existence of resistant mutants [35]. Therefore only looking at biochemical and virologic breakthrough, as in the present study will miss out on identifying the emergence of viral resistance in most. The majority of patients in these studies, who developed resistance, still had biochemical and virologic evidence of improvement in the liver disease [23,33,34].
The genetic heterogeneity of the HBV genome has been established and seven genotypes (A to G) can be classified, based on comparison of complete HBV genomes [36]. The geographic distribution of these genotypes is heterogeneous with genotypes A and D being more common in India [39]. Studies are now coming out showing differences in the natural history and response to treatment among the various genotypes. Studies have shown that genotype C is associated with more severe liver injury as compared with genotype B [37]. In a study by Yuen MF et al it was seen that genotype C was associated with lower rate of HBeAg seroconversion whereas genotype B had earlier onset of seroconversion [21]. HBV genotype has also been related to interferon treatment. In a study on German subjects, interferon induced HBeAg seroconversion was higher among patients with genotype A than those with genotype D [40]. Another report from Taiwan found that the rate of HBeAg loss was significantly higher in patients with genotype B compared with C [38]. A third study in HBeAg negative patients found that patients with genotype A responded better than genotype D (70% vs. 40%) [41]. With regards to Lamivudine, reports are less and only one study mentions better response of genotype B as compared to genotype C [42]. Therefore geographic differences in the natural history and response to treatment of chronic hepatitis B could be explained by the genotypic variations of HBV between different geographic regions. This may explain the higher early sero-conversion rates seen in this present study as compared to other Western and Asian studies. Also whether any host genetic factor could be influencing the rate of seroconversion has to be ascertained. Future work has to address these issues.
Conclusion
The results show that the HBeAg seroconversion rate in the first year in Indian patients is higher than that published from previous Western and Asian studies, but by three years the seroconversion rates become similar. Maximum seroconversion occurs in the first year with not much additional benefit on continuing treatment into the 2nd and 3rd years. Pretreatment ALT and HBV DNA levels were significant predictors of HBeAg loss. Relapse rates after treatment were high and were comparable to previous Asian studies. Future work has to done to elucidate the cause for the geographic variations in the rates of HBeAg seroconversion, especially with regard to influence of genotypes and any host genetic factors identified.
Abbreviations
CHB: chronic hepatitis B
HBeAg: hepatitis B e antigen
ALT: alanine amino-transferase
HBV-DNA: hepatitis B virus DNA
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
GA: Participated in the study design, collection of data and drafting the manuscript
CSB, KS, TSN: Participated in the study design and collection of data
GC: Conceived the study, participated in its design, coordination, drafting the manuscript
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
Uday C Ghoshal for statistical analysis help
Figures and Tables
Figure 1 Cumulative HBeAg seroconversion and loss in 60 patients.
Figure 2 HBeAg Seroconversion Year wise According to Baseline ALT.
Table 1 Base-line Characteristics of the Patients
Age (yr)
Median 40
Range 4–80
Male sex (%)
Weight (kg) 83.3
Median 58.5
Range 14–101
Route of HBV acquisition (%)
Blood Transfusion 11.7
Renal Dialysis 13.3
Sexual 5
Perinatal 3.3
Others 5
Unknown 61.7
BMI (kg/m2)
Median 22.8
Range 16.4–29.5
ALT (U/L)
Abnormal (%) 83.3
Median 72
Range 27–394
Median × ULN 1.8
HBV-DNA (mEq/ml)
Mean 920
Range 0.8–4500
Cirrhosis (%) 38.33
Table 2 Pre-treatment Variables Influencing HBeAg Loss
Variable HBeAg Loss (n = 35) No HBeAg Loss (n = 25) p value
Age (years) 37.5 44 ns
Sex
Male 29 21 ns
Female 6 4
Weight (Kg) 60 55 ns
BMI 23.1 22.7 ns
Median ALT (U/L) 94 45 0.002
Range 32–394 27–254
Median HBV-DNA (mEq/ml) 111.3 958 0.004
Range 0.8–4500 4.2–3500
Cirrhosis
Present 13 10 ns
Absent 22 15
ALT Normalization
Yes 23 5 0.02(df = 1)
No 9 13
Table 3 HBeAg Seroconversion and Loss by Pretreatment ALT Level
HbeAg Seroconversion HBeAg Loss
ALT Level Number % Number %
<= 1 × ULN 0/10 - 3/10 30
>1-<=2 × ULN 6/20 30 10/20 50
>2-<=5 × ULN 11/20 55 14/20 70
> 5 × ULN 7/10 70 8/10 80
Total 24/60 40 35/60 58.3
==== Refs
Maynard JE Hepatitis B: global importance and need for control Vaccine 1990 8 S18 S20 2139281 10.1016/0264-410X(90)90209-5
Beasley RP Hwang LY Lin CC Chien CS Hepatocellular Carcinoma and Hepatitis B: a prospective study of 22,707 men in Taiwan Lancet 1981 2 1129 33 6118576 10.1016/S0140-6736(81)90585-7
Seef B Beebe GW Hoofnagle JH A serological follow up of the 1942 epidemic of post-vaccination hepatitis in the United States Army N Engl J Med 1987 316 965 2436048
Beasley RP Hwang GY Postnatal infectivity of hepatitis B surface antigen carrier mothers J Infect Dis 1983 147 185 90 6827135
Margoli HS Alter MJ Hadler SC Hepatitis B: evolving epidemiology and implications for control Semin Liver Dis 1991 11 84 92 1832236
Maddrey WC Hepatitis B: an important public health issue J Med Virol 2000 61 362 366 10861647 10.1002/1096-9071(200007)61:3<362::AID-JMV14>3.0.CO;2-I
Niederau C Heintges T Lange S Goldman G Mohr L Haussinger D Long-term follow up of HBeAg positive patients treated with IFN-alpha for Chronic Hepatitis B N Eng J Med 1996 334 1422 1427 10.1056/NEJM199605303342202
Lok A Lai CL Wu P Leung EKY Lam TS Spontaneous hepatitis B e antigen to antibody seroconversion and reversion in Chinese patients with chronic hepatitis B virus infection Gastroenterology 1987 92 1839 1843 3569757
Wong DKH Cheung AM O'Rourke Naylor CD Detsky AS Heathcote J Effect of alpha-interferon treatment in patients with hepatitis B e antigen-positive chronic hepatitis B: a meta-analysis Ann Intern Med 1993 119 312 23 8328741
Lok ASF Wu PC Lai CL Lau JY Leung EK Wong LS Ma OC Lauder IJ Ng CP Chung HT A controlled trial of interferon with or without prednisone priming for chronic hepatitis B Gastroenterology 1992 102 2091 2097 1587429
Lai C-L Chien R-N Nancy WY Chang TT Guan R Tai DI Ng KY Wu PC Dent JC Barber J Stephenson SL Gray F A one year trial of Lamivudine for chronic hepatitis B N Engl J Med 1998 339 61 68 9654535 10.1056/NEJM199807093390201
Dienstag JL Schiff ER Wright TL Perillo RP Hann HW Goodman Z Crowther L Condreay LD Woessner M Rubin M Brown NA Lamivudine as initial treatment for chronic hepatitis B in the United States N Engl J Med 1999 341 1256 1263 10528035 10.1056/NEJM199910213411702
Perillo RP Lai CL Liaw YF Dienstag JL Schiff ER Schalm SW Heathcote EJ Brown NA Atkins M Woessner M Gardner SD Predictors of HBeAg loss after Lamivudine treatment for Chronic Hepatitis B Hepatology 2002 36 186 194 12085364 10.1053/jhep.2002.34294
Chien RN Liaw Y-F Atkins M Pretreatment ALT level as a determinant for Hepatitis B eAg seroconversion during Lamivudine treatment in patients with Hepatitis B Hepatology 1999 30 770 774 10462384 10.1002/hep.510300313
Lau DT-Y Khokhar F Doo E Ghany MG Herion D Park Y Kleiner D Schmid P Condreay L Gauthier J Kuhns M Liang TJ Hoofnagle J Long term treatment of Chronic Hepatitis B with Lamivudine Hepatology 2000 32 828 834 11003630 10.1053/jhep.2000.17912
Leung NW Lai C-L Chang T-T Guan R Lee CM Ng KY Lim SG Wu PC Dent JC Edmondson S Condreay L Chien RN Extended Lamivudine treatment in patients with Chronic Hepatitis B enhances HBeAg seroconversion rates: Results after 3 years of treatment Hepatology 2001 33 1527 1532 11391543 10.1053/jhep.2001.25084
Tassopoullous NC Volpes R Pastore G Heathcote J Buti M Goldin RD Hawley S Barber J Condreay L Gray DF Efficacy of lamivudine with patients with HBeAg -ve hepatitis B virus DNA positive [precore mutant] Chronic Hepatitis B Hepatology 1999 29 889 896 10051494 10.1002/hep.510290321
Schiff E Chancier J Rowley K Norkrans G Perillo R Tong M Crowther L Durability of HBeAg seroconversion after Lamivudine monotherapy in phase II and III trial Hepatology 1998 28 163A (abstract 1)
Song BC Suh DJ Lee HC Chung YH Lee YS HBeAg seroconversion after Lamivudine treatment is not durable in patients with chronic hepatitis B in Korea Hepatology 2000 32 803 6 11003626 10.1053/jhep.2000.16665
Lee CM Ong GY Lu SN Durability of Lamivudine induced HBeAg seroconversion for chronic hepatitis B patients with acute exacerbation J Hepatol 2002 32 669 74 10.1016/S0168-8278(02)00267-2
Man FY Erwin S He-Jun Y Decraemer H Lai CL Significance of hepatitis B genotype in acute exacerbation, HBeAg seroconversion, and cirrhosis related complications and HCC Hepatology 2003 37 562 567 12601354 10.1053/jhep.2003.50098
Kao JH Chen PJ Lai MY Chen DS Hepatitis B genotypes correlate with clinical outcomes in patients with chronic hepatitis B Gastroenterology 2000 118 554 559 10702206
Liaw YF Leung NW Chang T-T Guan R Tai DI Ng KY Chien RN Dent J Roman L Edmondson S Condreay L Chien RN Effects of extended Lam treatment in Asian patients with Chronic Hepatitis B Gastroenterology 2000 119 172 180 10889166 10.1053/gast.2000.8559
Liaw YF Tsai SL Pathogenesis and significance of acute exacerbation and remission in patients with Chronic Hepatitis B virus infection Viral Hep Rev 1997 3 143 154
Chisari FV Ferrari Hepatitis B virus immunopathogenesis Annu Rev Immunol 1995 13 29 60 7612225 10.1146/annurev.iy.13.040195.000333
Davis GL Hoofnagle JH Reactivation of chronic hepatitis B infection Gastroenterology 1987 92 2028 2031 3569777
Boni C Bertoletti A Penn A Cavalli A Pilli M Urbani S Scognamiglio P Boehme R Panebianco R Fiaccadori F Ferrari C Lamivudine treatment can restore T cell responsiveness in chronic hepatitis B J Clin Invest 1998 102 968 975 9727065
Davis GL Hoofnagle JH Waggoner JG Spontaneous reactivation of chronic hepatitis B virus infection Gastroenterology 1984 86 230 235 6690350
Doong SL Tsai CH Schinazi RF Liotta DC Cheng YC Inhibition of the replication of hepatitis B virus in vitro by 2' 3'-dideoxy-3'-thiacytidine and related analogues Proc Natl Acad Sci USA 1991 88 8495 8499 1656445
Zeuzem S deMan RA Honkoop P Roth W Schalm S Schmidt J Dynamics of hepatitis B virus infection in vivo J Hepatol 1997 27 431 436 9314118 10.1016/S0168-8278(97)80345-5
Liaw YF Chien R-N Yeh C-T Tsai S-L Acute exacerbation and hepatitis B virus clearance after emergence of YMDD motif mutation during Lamivudine treatment Hepatology 1999 30 567 572 10421670 10.1002/hep.510300221
Ling R Mutimer D Ahmed M Boxall EH Elias E Dushieko GM Harrison TJ Selection of mutations in the hepatitis B virus polymerase during treatment of transplant recepients with Lamivudine Hepatology 1996 24 711 713 8781347
Liaw Y-F Lai CL Leung NW Chang TT Guan R Tau DI Ng KY Chien RN Dent J Roman L Edmondson S Condreay L Chien RN Two year Lamivudine treatment in Chronic Hepatitis B infection- results of a placebo controlled multicentre study in Asia Gastroenterology 1998 114 A1289
Chang T-T Lai CL Liaw Y-F Leung NWY Guan R Lim SG Lee CM Enhanced HBeAg seroconversion rates in Chinese patients on Lamivudine Hepatology 1999 30 420A
Wakil JM Kazim SN Khan CA Raisuddin S Parvez M Guptan R Thakur V Hasnain S Sarin SK Prevalence and treatment profile of Mutations associated with Lamivudine treatment in Indian patients with Chronic Hepatitis B in the surface and Polymerase genes of HBV J Med Virol 2002 68 311 318 12226816 10.1002/jmv.10205
Stuyver L De Gendt S Van Geyt C Zoulim F Fried M Schinazi RF Rossau R A new genotype of hepatitis B virus: complete genomes and phylogenetic relatedness J Gen Virol 2000 81 67 74 10640543
Lindh M Hannoun C Dhillon AP Norkrans G Horal P Core Promoter mutations and genotypes in relation to viral replication and liver damage in East Asian hepatitis B virus carriers J Infect Dis 1999 179 775 82 10068571 10.1086/314688
Kao JH Wu NH Chen PJ Lai MY Chen DS Hepatitis B genotypes and the response to interferon therapy J Hepatol 2000 33 998 1002 11131465 10.1016/S0168-8278(00)80135-X
Thakur V Guptan RC Kazim SN Malhotra V Sarin SK Profile, spectrum and significance of HBV genotypes in chronic liver disease patients in the Indian subcontinent J Gastroenterol Hepatol 2002 17 165 170 11966946 10.1046/j.1440-1746.2002.02605.x
Erhardt A Reineke U Blondin D Mutations of core promoter and response to interferon treatment in chronic replicative hepatitis B Hepatology 2000 31 716 725 10706563 10.1002/hep.510310323
Zhang X Zoulim F Habersetzer F Analysis of hepatitis B virus genotypes and pre-core region variability during IFN treatment of HBe antigen negative CHB J Med Virol 1996 48 8 16 8825704 10.1002/(SICI)1096-9071(199601)48:1<8::AID-JMV2>3.0.CO;2-E
Kao JH Liu CJ Chen DS Hepatitis B viral genotypes and lamivudine resistance J Hepatol 2002 36 303 4 11830346 10.1016/S0168-8278(01)00246-X
|
16164746
|
PMC1242224
|
CC BY
|
2021-01-04 16:03:27
|
no
|
BMC Gastroenterol. 2005 Sep 15; 5:29
|
utf-8
|
BMC Gastroenterol
| 2,005 |
10.1186/1471-230X-5-29
|
oa_comm
|
==== Front
BMC GastroenterolBMC Gastroenterology1471-230XBioMed Central London 1471-230X-5-301617152510.1186/1471-230X-5-30Research ArticleBone mineral density and cytokine levels during interferon therapy in children with chronic hepatitis B: does interferon therapy prevent from osteoporosis? Gur Ali [email protected] Bünyamin [email protected] Kemal [email protected] Mehmet [email protected] Kenan [email protected] Aysegul Jale [email protected] Department of Physical Medicine and Rehabilitation, Medical Faculty, Dicle University, Diyarbakir – Turkey2 Department of Pediatrics, Medical Faculty, Dicle University, Diyarbakir – Turkey2005 19 9 2005 5 30 30 4 1 2005 19 9 2005 Copyright © 2005 Gur et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Our aim was to determinate bone mineral density (BMD), levels of biochemical markers and cytokines in children with chronic hepatitis B treated with interferon (IFN)-alpha and to investigate effect of IFN-alpha therapy on these variables. To the best of our knowledge, this is first study carried out about BMD and cytokine levels in pediatric patients with chronic hepatitis B treated with IFN-alpha.
Methods
BMD, levels of parathyroid hormone (PTH), osteocalcin, C-terminal cross-linking telopeptide of type I collagen (CTX), calcium, alkaline phosphates (ALP), cytokines as TNF-alpha, interleukin (IL)-1β, IL-2r, IL-6, and IL-8 were studied in 54 children with chronic hepatitis B (4–15 years old) treated with interferon alone (n = 19) or in combination with lamivudine (n = 35) for six months and as controls in 50 age-matched healthy children.
Results
There was no significant difference in respect to serum IL-1β, TNF-α and osteocalcin levels while serum IL-2r (p = 0.002), IL-6 (p = 0.001), IL-8 (p = 0.013), PTH (p = 0.029), and CTX (p = 0.021) levels were higher in children with chronic hepatitis B than in healthy controls. BMD of femur neck (p = 0.012) and trochanter (p = 0.046) in patients were higher than in healthy controls. There was a statistically significant correlation between serum IL-1β and osteocalcin (r = -0.355, p < 0.01); between serum IL-8 and CTX levels (r = 0.372, p = 0.01), and ALP (r = 0.361, p = 0.01); between serum ALP and femur neck BMD (r = 0.303, p = 0.05), and trochanter BMD (r = 0.365, p = 0.01); between spine BMD and IL-2R (r = -0.330, p < 0.05).
Conclusion
In conclusion, our study suggest that BMD of femur, serum IL-2r, IL-6, IL-8, PTH, and CTX levels were higher in children with chronic hepatitis B treated with IFN-alpha alone or combination with lamivudine than in healthy children. High femur BMD measurements found in patients may suggest that IFN-alpha therapy in children with chronic hepatitis B could contribute indirectly to prevent from hip osteoporosis. Additionally, further investigations on effects of IFN-alpha for bone structure in children should be performed in the future.
==== Body
Background
Bone development during childhood and adolescence is a key determinant of adult skeleton health. A reduced bone mass is associated with increased fracture risk in adults as well as in children. Peak bone mass, which is reached by early adulthood, serves as a bone reserve for the remainder of life, therefore childhood and adolescence are crucial periods for bone development. Strategies implemented for optimization of bone acquisition, as well as factors adversely affecting bone growth during these susceptible periods can have potentially long-standing consequences [1].
Recent studies have reported that osteodystrophy occurs not only in patients with alcoholic cirrhosis, but also in those with cirrhosis induced by hepatitis B and C viruses [2,3]. Because of improved treatment, patients with cirrhosis are living longer, an increasing proportion of such patients are found to have bone disease [4].
It is postulated that chronic liver disease and its complications might be responsible for activating some mediators [5,6]. It is further postulated that these mediators, such as some cytokines, might be the final common pathway leading to bone loss in parenchimal liver disorders [7].
A variety of compounds, including hormones and nutrients, are known to modulate bone remodelling. In addition, to these well-characterized substances, the immune system plays a role in this process through the involvement of pro-inflammatory cytokines [8]. Much interest has been focused on the role of the immune system in bone remodeling, and in particular, on the potential influence of cytokines upon the autocrine and paracrine regulation of bone cell activity [9-12]. Cytokines possess an important role in the regulation of bone resorption and formation during pathologic bone remodeling, and they also play a role during normal bone remodeling [13]. Significantly, IL-6 is a potent activator of ostoclasts and bone resorption. Similarly, other cytokines, such as IL-1, IL-11 and TNF influence ostoclast function and the age associated dysregulation of these cytokines may also contribute to the development of osteoporotic bone disease [8]. IL-8 is a chemokine of importance in inflammatory processes, and causes an increase in the levels of parathyroid hormone (PTH) mRNA. This suggest that IL-8 and inflammatory events may play a role in bone homeostasis by acting upon the parathyroid gland [14-16].
Interferon (IFN) has been shown to be effective in inducing inhibition of viral replication, normalization of liver tests and even improvement of liver histology in HBV-related liver diseases and it is known that IFN-alpha may affect bone turnover. There is limited information about the long-term effect of IFN-alpha therapy on bone metabolism.
A large number of studies on hepatic osteodystrophy in adult have reported recent advances in research on bone metabolism. However, bone metabolism in children has been regarded as differential diagnosis of bone resorption, pathological mechanisms and effects of IFN-alpha has not been elucidated.
Our aim was to determinate bone mineral density (BMD), levels of biochemical markers and cytokine in children with chronic hepatitis B treated with IFN-alpha and to investigate effect of IFN-alpha therapy on these variables. To the best of our knowledge, this is first study carried out about BMD and cytokine levels in pediatric patients with chronic hepatitis B treated with IFN-alpha. In view of the cost and widespread universal use of this drug in all age groups, especially with the epidemic of hepatitis B and C, we feel that such a detailed study is important.
Methods
BMD, levels of PTH, osteocalcin, CTX, calcium, cytokines as TNF-alpha, interleukin (IL)-1β, IL-2r, IL-6, and IL-8 were studied in 54 children with chronic hepatitis B (4–15 years old) treated with interferon alone (n = 19) or in combination with lamivudine (n = 35) for six months and as controls in 50 sex and age-matched healthy children.
This study was performed in Dicle University, Diyarbakir, Turkey. Informed consent was taken from the parents of patients and sufficient information was given to them about the disease course and the treatment procedure at the beginning of the study. The study was approved by the local ethics committee.
BMD of the spine and hip (neck, trochanter) were measured by dual-energy x-ray absorptiometry (DEXA) (NORLAND, 6938CE, New York, USA). The variation coefficient for consecutive determinations on spine and femur images in our laboratory was 1.9% at the lumbar spine and 1.6 % at the femur region. All spinal scans were reviewed for evidence of vertebrae with collapse or focal sclerosis by an experienced radiologist.
The diagnostic criteria for chronic HBV infection were seropositivity for hepatitis B surface antigen (HBsAg), lack of anti-hepatitis B surface antibodies (anti-HBs), and presence of anticore IgG antibodies (anti-HBc). All patients had been infected with HBV for more than 2 yr. The mean time from the presumed onset of HBV infection, defined as at least from the first documented elevation of serum liver enzyme levels, to the study was 3.9 ± 3.2 yr. Knodell's histological activity index was used to evaluate necro-inflammation and fibrosis in biopsy samples from all patients. Mean inflammatory score was 5.1 ± 2.4, and the mean fibrosis score, 1.2 ± 1.1. None of all patients had cirrhosis.
The diagnosis of all patients was confirmed after a thorough laboratory investigation for their symptoms of icterus, abdominal pain, fatigue and loss of appetite. Moreover, family members of patients were tested for serologic parameters of HBV in order to determine possible vertical or horizontal transmission.
Patients were excluded from the study, after the screening, if they were more than 16 years old; if they had having positive test results for antibody to hepatitis D virus, hepatitis C virus, or human immunodeficiency virus; having decompensate liver disease (defined by a serum bilirubin level more than 2.5 times the upper limit of normal, a prothrombin time prolonged by more than 3 s and a serum albumin level lower than 3 g/dl or a history of ascites, variceal hemorrhage, or hepatic encephalopathy); if they have evidence of autoimmune hepatitis (defined as an anti-nuclear antibody titer higher than 1/160) or metabolic liver disease (Wilson's disease, hemochromatosis, deficit of α-1 antitrypsin); if they had received investigational drug within 30 days before enrollment. Patients were also excluded if they had a total white-blood-cell count less than 2500/m3, a neutrophil granulocyte count less than 1000/mm3 and a value of haemoglobin less than 10 g/dl; if they were in poor clinical condition and/or had serious medical diseases (e.g. malnutrition, cardiomyopathies, diabetes, hypertension, neurologic, metabolic, autoimmune and neoplastic diseases).
19 patients with hepatitis B who entered the study received ten million units/ body surface area (max 10 million units) three times per week of recombinant interferon alpha 2b alone and 35 patients received interferon alpha 2b in same dosage in combination with lamivudine 4 mg/kg (max 100 mg) for six months. Recombinant interferon alpha-2b was administered subcutaneously by qualified medical staff or by the parents of patients after adequate training.
Blood samples were obtained after an over-night fast; precautions were taken to avoid contamination. Freshly drawn blood (15 ml) samples were obtained and immediately centrifuged at 200 × g (20 min at 24°C). For these tests HBV antigens and antibodies were assessed by qualitative micro-particle enzyme immunoassay (Organon Teknika BV, Boxtel, The Netherlands) HBV-DNA by Digene Hybride Capture Systems (Beltsville, MD 20705, USA). Serum levels of cytokines were determined using IMMULITE diagnostic kits (DPC-Diagnostic Products Corporation, USA). This diagnostic kit is an in vitro enzyme-linked immunosorbent assay for the quantitative measurement of human cytokines in serum. The serum Osteocalcin level was measured with a commercially available N-MID Osteocalcin Electrochemiluminescence Immunoassay kit (Roche Diagnostics GmbH, Mannheim, Germany). The serum levels of CTX were determined by Elecsys β-Crosslaps commercially available immunoassay kit (Roche Diagnostics GmbH, Mannheim, Germany). Serum PTH was measured by a two-site immunoradiometric assay using a commercially available kit (Nichols Institute). Serum and urinary chemical estimations were performed using Beckman-Synchron CX-5 technology.
Statistical analysis
The data obtained were analyzed using the Statistical Package for the Social Sciences (SPSS 10.0). Results in patients with chronic hepatitis B and controls were compared using Student's unpaired t test. Results in patients treated with IFN-alpha alone and combination therapy were compared using Mann -Whitney -U test. Pearson's correlation test was used for correlation analysis. All statistical tests were 2-sided; p < 0.05 was considered to be statistically significant. Values are expressed as the mean ± standard deviation.
Results
The mean age of patients and healthy control groups were 10.23 ± 3.12 and 10.02 ± 2.84 years, respectively. Patients group consisted of 43 females and 11 males and there were 39 females and 11 males in controls group. In both groups, mean body mass index and other demographic characteristics were similar. There was no statistically significant difference in any demographic characteristics between the groups (p > 0.05).
There was no statistically significant difference between patients received recombinant interferon alpha-2b alone and combination with lamivudine (Table 1).
Laboratory data of both groups are shown in Table 2. While serum IL-2r (p = 0.002), IL-6 (p = 0.001), IL-8 (p = 0.013), PTH (p = 0.029), and CTX (p = 0.021) levels were higher in children with chronic hepatitis B than in healthy controls, there was no significant different in respect to serum IL-1β, TNF-α and osteocalcin levels. BMD of femur neck (p = 0.012) and trochanter (p = 0.046) in patients were higher than in healthy controls.
Correlation between laboratory data and BMD measurements of patients group are shown in Table 3. There was a statistically significant correlation between serum IL-1β and osteocalcin (r = -0.355, p < 0.01), ALT (r = 0.494, p = 0.01), and AST (r = 0.528, p = 0.01); between serum IL-8 and CTX levels (r = 0.372, p = 0.01), and ALP (r = 0.361, p = 0.01); between serum ALP and femur neck BMD (r = 0.303, p = 0.05), and trochanter BMD (r = 0.365, p = 0.01); between spine BMD and IL-2R (r = -0.330, p < 0.05).
Discussion
Osteoporosis is an otherwise healthy child or adolescent is rare, although cases of idiopathic osteoporosis have been described. Rather, pediatric osteoporosis is increasingly recognized in the setting of chronic illness related to the disease itself or its treatment. A large number of studies on primary osteoporosis have reported recent advances in research on bone metabolism. However, secondary osteoporosis has been regarded as a differential diagnosis of primary osteoporosis, and its pathological mechanisms have not been elucidated compared with those of primary osteoporosis [17].
Bone manifestations are well-known extrahepatic complications of chronic liver diseases [18,19]. In these patients, several factors contribute to the development of bone disease. In particular, malnutrition, immobilization, and hormonal changes are causes for deteriorating bone metabolism in patients with chronic liver diseases [19]. The mechanism leading to osteoporosis is still unclear. The equilibrium between bone formation and bone resorption is disturbed [20], and, apart from the decreased activity of osteoblasts [21], there are also studies indicating an increase in osteoclast activity [22]. In contrast to primary biliary cirrhosis and primary sclerosing cholangitis, no disease-specific association between chronic hepatitis B, C, and D virus infection and osteoporosis is documented. Only few studies on bone metabolism have been performed in patients suffering from chronic viral hepatitis, especially before and after liver transplantation [23-25].
Bone disease in patients with chronic active hepatitis is usually asymptomatic and is characterized by decreases in BMD. Histomorphometric analysis of bone biopsies from the iliac crest of patients with chronic active hepatitis shows osteoporosis with decreased trabecular bone volume and no osteomalacia. Bone remodeling is regulated by a number of growth factors, cytokines, systemic peptides and steroid hormones. Proinflammatory cytokines appear to have a role in the development of chronic liver disease. IL-1b and TNF-alpha are involved in liver fibrogenesis [26]. The activation of the cytokine cascade, induces fibroblast proliferation and parenchymal inflammatory response producing liver damage [26,27].
The prevalence of osteoporosis among patients with chronic liver diseases ranges from 10% to 60% [28-31], the highest being observed in cholestatic liver disease and alcoholic liver disease. A recent study revealed that the prevalence of osteoporosis in patients with cirrhosis secondary to hepatitis B or C was nearly 50% [32]. Most studies of bone disease were performed in patients with cirrhosis. Nevertheless, little is known about the occurrence of bone disease in non-cirrhotic patients with chronic hepatitis B or C [33].
Interleukins and lymphokines may play a role in the bone remodeling process [1]. The calcitonin-like effect of IFN-γ is difficult to interpret if one assumes that the immune interferon actually brings about the fusion of monocytes into osteoclasts. If this actually occurred in bone this would force the conclusion that these cells can not be activated in the presence of IFN-γ. This could also explain the reduced effectivity of PTH in the presence of the immune interferon. Although IL-1β and TNF-α may be involved in the bone remodeling process, we did not find any significant difference when we compared their serum levels in children with chronic hepatitis B and healthy controls. It should be borne in mind that several disputes exist among researches concerning cytokines and pathologic bone remodeling, especially concerning the secretion of cytokines into the peripheral blood. This may be related to effect of IFN-alpha therapy.
Since biochemical markers of bone turnover are important in the assessment of osteoblastic and osteoclastic functions, we measured the serum osteocalcin and C-terminal cross-linking telopeptide of type I collagen levels. Osteocalcin is a noncollagenous protein secreted by osteoblasts and is widely accepted as a marker for osteoblastic activity [34] and bone formation [35], whereas serum CTX, as a collagen-degradation product is a marker of bone resorption [36].
In some studies, it has been reported that the serum osteocalcin levels are higher in cirrhosis patients, which means that cirrhotic patients have high turnover osteoporosis [37]. However, some authors reported that the serum osteocalcin levels were lower in cirrhotic patients and the osteopenia in these patients was not due to a decrease in bone formation [38,39].
In hepatocellular dysfunction, some authors reported that the serum parathyroid hormone levels were higher [40], and others reported them as unchanged [29]. Another report showed that the increase in bone resorption might be the result of decreased PTH degradation [41]. In our study, while serum PTH and CTX levels were higher in children with chronic hepatitis B than in healthy controls, there was no significant different in respect to serum osteocalcin and ALP levels.
IFN-alpha has numerous clinical applications but is used most extensively in the treatment of chronic hepatitis B and chronic hepatitis C. Research into the effects of IFN-alpha on bone mineral metabolism has been very sparse, and the majority of studies reflect in vitro models. The exact mechanism of positive effect on bone mineral metabolism by IFN-alpha is not completely understood although a number have been postulated. Both in vivo and in vitro studies demonstrate that IFN-alpha decreases bone resorption, whereas osteoblast may or may not be affected in vivo [42]. An in vitro study on the effects of IFN-alpha on human bone marrow stromal cells showed that IFN-alpha decreased the production of IL-1b [43], which has been shown to stimulate osteoclastic bone resorption [44].
Takayanagi et al. [45] reported that there is cross-talk between the tumour necrosis factor and IFN families of cytokines, through which IFN-gamma provides a negative link between T-cell activation and bone resorption. Authors stated that the findings of their study may offer a therapeutic approach to treat the inflammation-induced tissue breakdown. IFN-alpha clearly decreases bone resorption, but in vitro data suggest that there is decreased formation with increased differentiation of osteoblasts, whereas the in vivo work suggests that osteoblasts are not suppressed by IFN-alpha [42]. Thus, IFN-alpha could be increased BMD in children with chronic hepatitis by one or a combination of these mechanisms. In our study, because there is no statistically significant difference between patients received IFN-alpha alone and combination therapy we think that changes in BMD biochemical markers and cytokines are related to IFN-alpha treatment.
Solis-Herruzo et al. [46] reported that adult male patients receiving ribavirin and IFN-alpha had a lower bone mass than those receiving IFN-alpha only; this suggests that ribavirin was responsible for the decrease in bone mineral density. This was, however, a cross-sectional study and did not evaluate patients before treatment, possibly leading to inconsistent conclusions. Trombetti et al. [47], on the other hand, did not find any effect of ribavirin in bone metabolism. The impact, therefore, of IFN-alpha and ribavirin in bone remains unclear [48]. In our study, there was no statistically significant difference between patients with hepatitis B received IFN-alpha alone and combination with lamivudine.
In our study, BMD values of femur, but was not spine, in patients were higher than in healthy controls. High BMD values of femur postulated that IFN-alpha therapy might be responsible for inhibiting some mediators. It is further postulated that these mediators, such as some cytokines, might be the final common pathway leading to bone loss in parenchimal liver disorders. Because interferon inhibits the formation of osteoclast-like cells [49], interferon treatment may increase BMD. However, interferon is expensive and is thus inappropriate for the treatment of bone lesions. But, it is postulated that in patients treated with IFN-alpha may not need additional therapy for the treatment of bone resorption.
Conclusion
In conclusion, our study suggest that BMD of femur, serum IL-2r, IL-6, IL-8, PTH, and CTX levels were higher in children with chronic hepatitis B treated with IFN-alpha alone or combination with lamivudine than in healthy children. High femur BMD measurements found in patients may suggest that IFN-alpha therapy in children with chronic hepatitis B could contribute indirectly to prevent from hip osteoporosis. Additionally, further investigations on effects of IFN-alpha for bone structure in children should be performed in the future.
Abbreviations
BMD, bone mineral density; IFN, interferon; PTH, parathyroid hormone; ALP, alkaline phosphates; CTX, C-terminal cross-linking telopeptide of type I collagen; IL, interleukin; DEXA, dual-energy x-ray absorptiometry; HBsAg, hepatitis B surface antigen; Anti-HBc, anticore IgG antibodies; HBV, Hepatitis B virus
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
AG participated in the design of the study and performed the statistical analyses.
BD participated in the design of the study and screened of subjects.
KH and AJS conceived of the study, and participated in its design and coordination.
KN and MB participated in the sequence alignment.
All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Figures and Tables
Table 1 Comparisons of clinical features in patients with chronic hepatitis B treated with IFN alone or IFN plus lamivudine
Variables IFN-alpha (n = 19) IFN plus Lamivudine (n = 35) P
TNF-α 24.48 ± 8.46 26.60 ± 7.43 NS
IL-1β 7.56 ± 10.86 7.08 ± 12.47 NS
IL-2r 1472.42480.47 1333.72 ± 472.51 NS
IL-6 24.85 ± 31.46 25.57 ± 29.43 NS
IL-8 18.19 ± 13.88 18.73 ± 14.27 NS
PTH 42.03 ± 29.41 40.99 ± 26.92 NS
s-CTX 1.88 ± 0.68 1.96 ± 0.75 NS
Osteocalcin 65.49 ± 34.45 63.85 ± 32.67 NS
Serum calcium 9.28 ± 1.76 9.86 ± 1.69 NS
ALP 249.45 ± 71.34 193.09 ± 72.88 NS
ALT 62.83 ± 76.39 61.45 ± 81.54 NS
AST 59.16 ± 61.48 57.66 ± 59.82 NS
L2–4 BMD 0.63 ± 0.09 0.61 ± 0.14 NS
Femur neck BMD 0.78 ± 0.12 0.82 ± 0.17 NS
Trochanter BMD 0.67 ± 0.16 0.69 ± 0.12 NS
Table 2 Comparisons of laboratory data and BMD measurements of children with chronic hepatitis B treated with IFN-alpha alone or combination with lamivudine and healthy controls.
Variables Patients (n = 54) Control (n = 50) P
TNF-α 25.54 ± 8.30 15.47 ± 6.77 NS
IL-1β 7.32 ± 12.13 5.48 ± 6.14 NS
IL-2r 1403.07 ± 455.67 1120.48 ± 492.37 0.002
IL-6 25.21 ± 39.27 5.98 ± 6.49 0.001
IL-8 18.46 ± 14.77 12.55 ± 7.36 0.013
PTH 41.51 ± 28.80 33.08 ± 18.22 0.029
s-CTX 1.92 ± 0.76 1.61 ± 0.52 0.021
Osteocalcin 64.67 ± 30.63 75.81 ± 31.10 NS
Serum calcium 9.57 ± 1.83 9.49 ± 1.52 NS
ALP 221.27 ± 67.27 219.66 ± 78.74 NS
ALT 62.14 ± 88.31 27.01 ± 42.41 0.009
AST 58.41 ± 65.57 31.72 ± 29.69 0.007
L2–4 BMD 0.62 ± 0.13 0.59 ± 0.12 NS
Femur neck BMD 0.80 ± 0.16 0.72 ± 0.15 0.012
Trochanter BMD 0.68 ± 0.14 0.62 ± 0.14 0.046
Table 3 Correlation between BMD and serum cytokines, and biochemical markers in children with chronic hepatitis C treated with IFN-alpha.
IL-1β .501**
IL-2r .075 .096
IL-6 .167 -.101 -.163
IL-8 .137 .251 .248 -.126
PTH .005 -.012 .009 .121 .088
s-CTX -.087 -.253 .140 -.060 .372** -.155
ALP -.013 -.049 -.055 -.092 .361** -.102 .179
Osteocalcin .105 -.355** .016 .135 -.205 -.101 .256 -.006
ALT .189 .494** .087 -.161 .243 .122 -.257 .251 -.206
AST .214 .528** .125 -.192 .253 .089 -.222 .252 -.155 .960**
L2–4 BMD -.159 .019 -.330* .036 .138 .226 .122 .072 -.118 .122 .084
Femur Neck BMD -.112 -.043 -.246 .025 .105 .179 .067 .303* -.034 .163 .097 .704**
Trochanter BMD -.111 -.039 -.237 -.006 .123 .090 .017 .365** -.125 .198 .125 .697** .943**
TNF-α IL-1β IL-2r IL-6 IL-8 PTH s-CTX ALP Osteocalcin ALT AST L2–4 BMD Femur Neck BMD
** Correlation is sigificant at the 0.01 level (2-tailed).
* Correlation is significant at the 0.05 level (2-tailed).
==== Refs
Hartman C Hochberg Z Shamir R Osteoporosis in pediatrics IMAJ 2003 5 509 515 12901249
Shiomi S Kuroki T Masaki K Takeda T Nishiguchi S Nakajima S Seki S Kobayashi K Okamura T Ochi H Osteopenia in primary biliary cirrhosis and cirrhosis of the liver in women, evaluated by dual-energy X-ray absorptiometry J Gastroenterol 1994 29 605 609 8000509
Tsuneoka K Tameda Y Takase K Nakano T Osteodystrophy in patients with chronic hepatitis and liver cirrhosis J Gastroenterol 1996 31 669 678 8887033
Shiomi S Nishiguchi S Kurooka H Tamori A Habu D Takeda T Ochi H Cyclical etidronate for treatment of osteopenia in patients with cirrhosis of the liver Hepatol Res 2002 22 102 106 11818249 10.1016/S1386-6346(01)00123-1
Manolagas SCV Jilka RL Bone marrow, cytokines and bone remodelling. Emerging insights into the pathophysiology of osteoporosis N Engl J Med 1995 332 305 311 7816067 10.1056/NEJM199502023320506
Salmayenli N Genç S Karan MA Med Sci Res 1998 26 207 208
Karan MA Erten N Tascioglu C Karan A Sindel D Dilsen G Osteodystrophy in posthepatic cirrhosis. in posthepatic cirrhosis YMJ 2001 42 547 552
Ershler WB Harman SM Keller ET Immunologic aspects of osteoporosis Dev Comp Immunol 1997 21 487 499 9463782 10.1016/S0145-305X(97)00029-3
Mac Donald BR Gowen M Cytokins and bone Br J Rheumatol 1992 31 149 155 1540781
Wallach S Avioli LV Feinblatt JD Cytokines and bone metabolism Calcif Tissue Int 1993 53 293 296 8287313 10.1007/BF01351830
Gruber HE Bone and immune system Proc Soc Exp Biol Med 1991 197 219 225 2068116
Roodman GD Role of cytokines in the regulation of bone resorption Calcif Tissue Int 1993 53 94 98 10.1007/BF01673412
Natale VM Filho WJ Duarte AJS Does the secretion of cytokines in the periphery reflect their role in bone metabolic diseases? Mech Ageing Dev 1997 94 14 23 10.1016/S0047-6374(96)01844-1
Angeletti RH D'Amico T Ashok S Russell J The chemokine interleukin-8 regulates parathyroid secretion J Bone Miner Res 1998 13 1232 1237 9718190
Gur A Denli A Nas K Cevik R Karakoc M Sarac Aj ERdogan F Possible pathogenetic role of new cytokines in postmenopausal osteoporosis and changes during calcitonin plus calcium therapy Rheumatol Int 2002 22 194 198 12215865 10.1007/s00296-002-0223-x
Gur A Denli A Cevik R Nas K Karakoc M Sarac AJ The effects of alendronate and calcitonin on cytokines in postmenopausal osteoporosis: a randomized and controlled study YMJ 2003 44 99 109
Hirano Y Kishimoto H Hagino H Teshima R The change of bone mineral density in secondary osteoporosis and vertebral fracture incidence JBMM 1999 17 119 124
McGaughan GW Feller RB Osteoporosis in chronic liver disease: Pathogenesis, risk factors, and management Dig Dis 1994 12 223 231 7851001
Hay JE Bone disease in liver transplant recipients Gastroenterol Clin North Am 1993 22 237 249
Rosen H Primary biliary cirrhosis and bone disease Hepatology 1996 21 253 255 10.1016/0270-9139(95)90435-2
Stellon AJ Webb A Compston J Williams R Low bone turnover state in primary biliary cirrhosis Hepatology 1987 7 137 142 3804193
Eastell R Dickson ER Hodgson SF Wiesner RH Porayko MK Wahner HW Cedel SL Riggs BL Krom RA Rates of vertebral bone loss before and after liver transplantation in women with primary biliary cirrhosis Hepatology 1991 14 296 300 1860685 10.1016/0270-9139(91)91418-Z
Meys E Fontanges E Fourcade N Thomasson A Pouyet M Delmas PD Bone loss after orthotopic liver transplantation Am J Med 1994 97 445 450 7977433 10.1016/0002-9343(94)90324-7
Monegal A Navasa M Guanabens N Peris P Pons F Martinez de Osanba MJ Rimola A Rodes J Munos-Gomes J Osteoporosis and bone mineral metabolism in cirrhotic patients referred for orthotopic liver transplantation Calcif Tissue Int 1997 60 148 154 9056162 10.1007/s002239900205
Trautwein C Possienke M Schlitt H Boker KH Horn R Raab R Manns MP Brabant G Bone density and metabolism in patients with viral hepatitis and cholestatic liver diseases before and after liver transplantation Am J Gastroenterol 2000 95 2343 2351 11007240 10.1111/j.1572-0241.2000.02269.x
Chojkier M Brenner DA Therapeutic strategies for hepatic fibrosis Hepatology 1988 8 176 182 3276587
Matsuoka M Pham NT Tsukamoto H Differential effects of interleukin-1 alpha, tumor necrosis factor alpha and transforming growth factor beta 1 on cell proliferation and collagen formation by cultured fat storing cells Liver 1989 9 71 78 2785237
Stellon AJ Webb A Compston JE Bone histomorphometry and structure in corticosteroid treated chronic active hepatitis Gut 1988 29 378 384 3356370
Hodgson SF Dickson ER Wahner HW Johnson KA Mann KG Riggs BL Bone loss and reduced osteoblast function in primary biliary cirrhosis Ann Intern Med 1985 103 855 860 4062087
Bonkovsky HL Hawkins M Steinberg K Hersh T Galambos JT Henderson JM Millikan WJ Gallowey JR Prevalence and prediction of osteopenia in chronic liver disease Hepatology 1990 12 273 280 2391068
Gonzalez-Calvin JL Garcia-Sanchez A Bellot V Munoz-Torres M Raya-Alvarez E Salvatierra-Rios D Mineral metabolism, osteoblastic function and bone mass in chronic alcoholism Alcohol Alcohol 1993 28 571 579 8274181
Gallego-Rojo FJ Gonzalez-Calvin JL Munoz-Torres M Mundi JL Fernandez-Perez R Rodrigo-Moreno D Bone mineral density, serum insulin-like growth factor I, and bone turnover markers in viral cirrhosis Hepatology 1998 28 695 699 9731561 10.1002/hep.510280315
Schiefke I Fach A Wiedmann M Aretin AV Schenker E Borte G Mossner J Caca K Reduced bone mineral density and altered bone turnover markers in patients with non-cirrhotic chronic hepatitis B or C infection World J Gastroenterol 2005 11 1843 1847 15793878
Garnero P Grimaux M Demiaux B Preaudat C Seguin P Delmas PD Measurement of serum osteocalcin with a human specific two-site-radioimmunassay J Bone Miner Res 1992 7 1389 1397 1481725
Kohsla S Kleerekoper M Favus MJ Biochemical markers of bone turnover Primer on the metabolic bone diseases and disorders of mineral metabolism 1999 4 Philadelphia: Lippincott Wlliams and Wilkins 128 134
Garnero P Gineyts E Riou JP Delmas PD Assessment of bone resorption with a new marker of collagen degradation in patients with metabolic bone disease J Clin Endocrinol Metab 1994 79 780 785 8077361 10.1210/jc.79.3.780
Resch H Pietschmann P Krexner E Woloszczuk W Willvonseder R Peripheral bone mineral content in patients with fatty liver and hepatic cirrhosis Scand J Gastroenterol 1990 25 412 416 2336550
Suzuki K Arakawa Y Chino S Yagi K Hepatic osteodystrophy Nippon Rhinsho 1998 56 1604 1608
Capra F Casaril M Gabrielli GB Stanzial A Ferrari S Gandini G Falezza G Corrocher R Plasma Osteocalcin levels in liver cirrhosis Ital J Gastroenterol 1991 23 124 127 1742505
Fonseca V Epstein O Gill DS Menon RK Thomas M Mclntyre N Dandona P Hyperparathyroidism and low serum Osteocalcin despite vitamin D replacement in primary biliary cirrhosis J Clin Endocrinol Metab 1987 64 873 877 3494037
Atkinson MJ Vido I Keck E Hesch RD Hepatic osteodystrophy in primary biliary cirrhosis, a possible defect in Kuppfer cell cleavage of parathyroid hormone Clin Endocrinol (Oxf) 1983 18 21 28
Goodman GR Dissanayake IR Gorodetsky E Zhou H Ma YF Jee WS Epstein S Interferon-alpha, unlike interferon-gamma, does not cause bone loss in the rat Bone 1999 25 459 463 10511113 10.1016/S8756-3282(99)00182-9
Aman MJ Keller U Derigs G Mohamadzadeh M Huber C Peschel C Regulation of cytokine expression by interferon-alpha in human bone marrow stromal cells: Inhibition of hematopoietic growth factors and induction of interleukin-1 receptor antagonist Blood 1994 84 4142 4150 7994029
Pfeilschifter J Chenu C Bird A Mundy GR Roodman GD Interleukin-1 and tumor necrosis factor stimulate the formation of human osteoclast-like cells in vitro J Bone Miner Res 1989 4 113 118 2785743
Takayanagi H Ogasawara K Hida S Chiba T Murata S Sato K Takaoka A T-cell-mediated regulation of osteoclastogenesis by signalling cross-talk between RANKL and IFN-g Nature 2000 408 600 605 11117749 10.1038/35046102
Solis-Herruzo JA Castellano G Fernandez I Munoz R Hawkins F Decreased bone mineral density after therapy with alpha interferon in combination with ribavirin for chronic hepatitis C J Hepatol 2000 33 812 817 11097491 10.1016/S0168-8278(00)80314-1
Trombetti A Giostra E Mentha G Negro F Rizzoli R Lack of evidence for ribavirin-induced bone loss Hepatology 2002 36 255 257 12085375 10.1053/jhep.2002.33711
Moreira RO Balduino A Martins HS Reis JS Duarte ME Farias ML Borojevic R Ribavirin, but not Interferon a-2b, is associated with impaired osteoblast proliferation and differentiation in vitro Calcif Tissue Int 2004 75 160 168 15148560
Kurihara N Roodman GD Intereferons-alpha and -gamma inhibit interleukin-1b-stimulated osteoclast-like cell formation in long-term human marrow cultures J Interferon Res 1990 10 541 547 2125633
|
16171525
|
PMC1242225
|
CC BY
|
2021-01-04 16:03:27
|
no
|
BMC Gastroenterol. 2005 Sep 19; 5:30
|
utf-8
|
BMC Gastroenterol
| 2,005 |
10.1186/1471-230X-5-30
|
oa_comm
|
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-1141615014510.1186/1471-2164-6-114Research ArticleIdentification and characterization of the fibrinogen-like domain of fibrinogen-related proteins in the mosquito, Anopheles gambiae, and the fruitfly, Drosophila melanogaster, genomes Wang Xinguo [email protected] Qin [email protected] Bruce M [email protected] Department of Animal Health and Biomedical Sciences, University of Wisconsin-Madison, 1656 Linden Dr., Madison, WI 53706, USA2 Department of Biochemistry, University of Wisconsin-Madison, 433 Babcock Drive Madison, WI 53706, USA3 Promega Corp., 2800 Woods Hollow Road, Madison, WI 53711, USA2005 8 9 2005 6 114 114 16 5 2005 8 9 2005 Copyright © 2005 Wang et al; licensee BioMed Central Ltd.2005Wang et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The fibrinogen-like (FBG) domain, which consists of approximately 200 amino acid residues, has high sequence similarity to the C-terminal halves of fibrinogen β and γ chains. Fibrinogen-related proteins (FREPs), which contain FBG domains in their C-terminal region, are found universally in vertebrates and invertebrates. In invertebrates, FREPs are involved in immune responses and other aspects of physiology. To understand the complexity of this family in insects, we analyzed FREPs in the mosquito genome and made comparisons to FREPs in the fruitfly genome.
Results
By using the genome data of the mosquito, Anopheles gambiae, 53 FREPs were identified, whereas only 20 members were found in the Drosophila melanogaster genome. Using sequence profile analysis, we found that FBG domains have high sequence similarity and are highly conserved throughout the FBG domain region. By secondary structure analysis and comparison, the FBG domains of FREPs are predicted to function in recognition of carbohydrates and their derivatives on the surface of microorganisms in innate immunity.
Conclusion
Detailed sequence and structural analysis discloses that the FREP family contains FBG domains that have high sequence similarity in the A. gambiae genome. Expansion of the FREP family in mosquitoes during evolutionary history is mainly accounted for by a major expansion of the FBG domain architecture. The characterization of the FBG domains in the FREP family is likely to aid in the experimental analysis of the ability of mosquitoes to recognize parasites in innate immunity and physiologies associated with blood feeding.
==== Body
Background
In mammals, fibrinogen, a soluble plasma protein, contains six polypeptide chains, two each of the Aα, Bβ and γ chains, linked by 29 disulfide bonds. Fibrinogen participates in both the cellular phase and the fluid phase of coagulation [1]. The fibrinogen-like (FBG) domain, which consists of approximately 200aa residues and has high similarity to the C-terminal halves of fibrinogen β and γ chains, has been found in a growing number of proteins [2]. Three distinct fibrinogen-related proteins (FREPs) have been identified in human: ficolin, tenascins, and microfibril-associated protein (MAP) [3-5]. These FREPs all contain a common C-terminal FBG domain with high sequence identity to the C-terminal regions of fibrinogen β and γ chains, but differ in their N-terminal regions. The FBG domain in ficolin can be brought together as clusters of three by collagen O-like triple helices, and is responsible for N-acetylglucosamine (GlcNAc) binding activity [6]. Recent studies have shown that human serum ficolins act as phagocytic receptors on circulating monocytes for microorganism recognition [7]. Tenascins are a family of multifunctional extracellular matrix (ECM) glycoproteins subject to complex spatial and temporal patterns of expression in the course of various organogenetic processes. These proteins mediate cell adhesion and show tissue-specific and cell growth-associated expression [4]. Microfibril-associated protein, another extracellular matrix protein, is a component of connective tissue microfibrils and a candidate for involvement in the etiology of inherited connective tissue diseases, which are associated with the Smith-magenis syndrome, a multiple congenital anomaly/mental retardation syndrome [8].
In invertebrates, several FREPs have been reported in various species, such as tachylectins from the horseshoe crab, Tachypleus tridentatus [9], fibrinogen-related proteins (FREP) from the snail, Biomphalaria glabrata [10], ficolins from the solitary ascidian, Halocynthia roretzi [11], tachylectin-related protein in the sponge, Suerites domuncula [12] and aslectin (AL-1) from the mosquito, Armigeres subalbatus [13]. All of these FREPs contain a common C-terminal FBG domain with high sequence identity to that of fibrinogen β and γ chains, but which differs in their N-terminal regions. These FREPs likely play an important role in the innate immune response against parasites [9,12,13]. The FBG domain of tachylectin is able to bind GlcNAc [9]. Aslectin, which also binds GlcNAc, is able to bind bacteria, and is likely involved in the antibacterial immune response in mosquitoes [13].
The rapid progress in the development of whole genome and expressed sequence tag (EST) databases provides an abundance of sequence data that greatly facilitates gene function studies. Using bioinformatics, one can mine the information from these databases to acquire an overview of each gene family and assess evolutionary relationships among its members [14]. Although the FREP family in the genomes of Anopheles gambiae and Drosophila melanogaster was briefly compared earlier [15], the FBG domains in this gene family have not been comparatively characterized. In this study, data derived from the genome and EST databases of the mosquito, A. gambiae, and the fruitfly, D. melanogaster, are presented here as an initial, yet exhaustive search for FREPs in both species. Provided is an overview of this protein family, including sequence alignments, patterns of conservation, and phylogenetic relationships. A further comparison between the annotated gene products from the genome sequences and the actual transcripts from the EST database also is made. In summary, these studies provide the first encompassing description of the FREP gene family in insects and establish a foundation for future studies that aim to define the role of these genes.
Results and discussion
Identification of FREP genes and characterization of the FBG domain in the A. gambiae genome
To identify FREP proteins encoded in the A. gambiae genome, a PSI-BLAST search was performed using AL-1 as a query sequence to screen the A. gambiae genome database at NCBI. Sixty amino acids were used as the minimum length of homology, and protein sequences having 35% or greater amino acid identity with AL-1 were added to the gene family list. To find FREPs that may have been overlooked due to low sequence identity to AL-1, we selected each sequence from the search results as a new seed to search the A. gambiae genome database again. Additional sequences were identified as homologs of the queries and added to the original list. This gene family list was manually examined to eliminate redundant sequences generated by repeated searching. This search revealed the presence of 53 genes encoding hypothetical FREP proteins in the A. gambiae genome (Table 1).
Table 1 Fibrinogen-related proteins in A. gambiae and D. melanogaster
Gene ID Length (aa)1 FBG domain2 Chromosomal location Transcription3
P M EST cDNA library
A. gambiae
EAA10385 201 full 2L 20D -
EAA10406 217 full 2L 20D -
EAA04425 186 full 2L 26A + Hemocyte
EAA10466 865 848 3' truncated 2L 21A + Development
EAA14231 226 full 3R 35B + NAP1
EAA44096 190 full 2L 23B + NAP1,NAH, Blood1,NAFB
EAA05203 296 273 full 3L 42B -
EAA05102 363 341 full 3L 42A + 4A3A,NAP1,NAH, Blood1,NAFB
EAA05205 308 full 3L 42A -
EAA05224 310 full 3L 42A + 4A3B, NAH, Blood1
EAA43404 314 292 full 3R 33C -
EAA01903 236 full Unknown + NAP1
EAL39348 202 full 3L 40A -
EAA10360 688 660 full 2L 21A -
EAA00222 173 full Unknown -
EAA13725 182 full 3L 40A -
EAA05204 543 3' truncated 3L 42A -
EAA13743 187 full 3L 40A -
EAA01418 362 337 3' truncated 2R 10A -
EAA05160 216 3' truncated 3L 42B + NAH, IRB, Blood1
EAA04072 280 258 full 2L 26B + NAH, blood1
EAL39349 262 3' truncated 3L 40A -
EAA05042 777 756 full 3L 42A + Blood1, cDNA1
EAA03931 178 full 2L 26D + Blood1, cDNA1, NAH
EAA02818 144 3' truncated Unknown + NAP1
EAA09906 171 5' truncated 3L 39A + NAH, NAFB, Blood1
EAL39350 330 308 full 3L 40A -
EAL39343 284 3' truncated 3L 40A -
EAA13689 178 3' truncated 3L 40A -
EAA04169 234 3' truncated 2L 26A -
EAL41889 339 full 2L 26D -
EAA05087 211 3' truncated 3L 42A -
EAA06922 323 267 3' truncated X 5A + NAH, NAFB, Blood1
EAA01294 185 full 2R 8C -
EAA15009 183 5' truncated 3R 33B + NAP1
EAL39347 242 3' truncated 3L 40A -
EAA13749 180 3' truncated 3L 40A -
EAA05439 266 3' truncated 3L 40B -
EAA05095 259 230 3' truncated 3L 42A -
AAR01125 268 3' truncated Unknown -
EAA13688 1020 3' truncated 3L 40A + cDNA1
EAA05097 166 3' truncated 3L 42A -
EAL39030 81 3' truncated 3R 33B + NAP1
EAA05065 116 3' truncated 3L 42A -
EAL40630 94 3' truncated Unknown -
EAA13692 441 Full 3L + NAP1
EAA02970 321 300 Full Unknown -
EAA13755 596 Full 3L -
EAA13691 231 Full 3L 40A -
EAA13726 212 Full 3L 40A + NAFB
EAA13760 271 Full 3L + cDNA1
EAA10480 284 265 Full 2L -
EAA05069 227 204 3' truncated 3L + NAP1
D. melanogaster
AAM68209 291 271 Full 2R 58B9 + GH
AAF57948 246 225 Full 2R 53D1 + RE
AAF44911 187 167 3' truncated 2L 34C4 -
AAF59068 347 Full 2R 44D4 -
AAF52372 176 5' truncated 2L 26C3 -
AAF48780 358 335 Full X 16F1 + LP
AAM52597 195 Full X 9A3 + RE, GH
AAF46536 332 310 Full X 9A3 + RH, GH, EK
AAN09619 241 Full X 9A3 + RH, GH, EK
AAL48972 198 177 3' truncated 2R 53D1 + RE
AAF47782 459 436 Full 3L 63E5 + RE, GM, EK,LP,CA
AAF58455 799 758 Full 2R 49D3 + RE, SD,RE,EK,LP
AAF55227 363 Full 3R 89A5 +
AAF49079 422 Full 3L 76E1 + RE, GM, EK, EC
AAN11645 406 Full 3L 76E1 + EK, GM
AAM11109 154 5' truncated 3L 76E1 + EK, GM
AAF46535 334 315 Full X 9A3 + RE, GH
AAN09447 251 Full X 16F1 + LP
AAF46801 157 5' truncated 2R 58B8 -
AAA28880 774 752 Full 2R 49D3 + RE, SD,RE,EK,LP
1. P represents precursor form of the predicted protein. M represents mature form of the predicted protein. Blank in column M indicates that signal is unpredictable.
2. FBG domain classified three categories. Full is the protein containing entire FBG domain; 5' truncated is the protein containing part of FBG domain which is truncated at the 5 primer region; 3' truncated is the protein containing part of FBG domain which is truncated at the 3 primer region.
3. In the transcription, + indicates matched transcript in EST database, – indicates no matched transcript in EST database. Tissue distribution was represented with the short-written of the EST library that was described in table 2.
To define the FBG domain in the FREP family, all 53 FREP and the human fibrinogen chain γ were aligned with the T_Coffee program. The results showed that most of the FREP genes have a C-terminal region composed of approximately 200aa with high sequence similarity with the C-terminus of human fibrinogen chain γ. Based on the alignment, the highly conserved region of 200aa residues in FREP was defined as the FBG domain in this study. A selected number of the FBG domains of the FREP were aligned and the highly conserved regions are illustrated in Fig. 1. This definition also is supported by the FBG domains in human and mouse ficolins [2]. In the FREP gene family, 28 of the 53 FREP genes were found in complete open reading frames and with a full FBG domain, and the remaining 25 FREP genes have truncated FBG domains, either in the 5'-region or the 3'-region (Table 1). Using a signal peptide prediction program, 14 of the 53 FREPs were predicted to contain secretion signal peptides (Table 1), suggesting that FREPs can be extracellular or intracellular.
Figure 1 Multiple sequence alignment of a representative set of the FBG domains of the FREP family in A. gambiae. Multiple sequence alignment was constructed using T-Coffee program. The 100% consensus sequence was boxed with black in the alignment. The PHD secondary structure is shown above the alignment with H representing an α-helix and E representing a β-strand. The sequences are denoted by their gene names in GenBank.
Conserved structure of the FBG domain in the FREP family and variation in some members
To construct an optimal multiple alignment of the FBG domain, we first aligned selected sequences with the T_Coffee program; this was followed by refinements on the basis of the PSI-BLAST search results. The selected multiple sequence alignment is shown in Fig. 1. The multiple alignment of the FBG domain sequences shows that FBG domains are highly similar throughout. Strikingly, 53% (28/53) of the FREPs contained a full FBG domain in their C-terminus (Table 1). Interestingly, some of the FREPs contain more than one FBG domain, although most of them are all not full FBG domains (Fig. 1). The distribution of the multiple FBG domains in these proteins shows certain patterns. Some of them contain two FBG domains that are connected by a 150aa hinge, e.g., EAA10360 and EAA05204. However, the two FBG domains in EAA10466 are located in the center of the protein, and are hinged together by approximately 20aa residues. There are also some members that contain 3 FBG domains. In EAA05042, three equivalent length regions of the FBG domain were repeated in the sequence (Fig. 2). Some of the FREPs also are composed of other domains in addition to the FBG domain, such as Lipase in EAA10466 (Fig. 2). In invertebrates, several FREP proteins have been reported to play an important role in innate immunity and in particular in the recognition of parasites (TL5A, AL-1). AL-1 can be upregulated by bacterial challenge and is able to bind GlcNAc and bacteria [13]. The FBG domain of TL5A can form a ligand-binding pocket specifically recognizing the acetyl-group in eliciting an immune response [16]. These data suggest that the FBG domains of FREPs probably function in recognizing carbohydrate moieties as part of the role they play in the mosquito immune response.
Figure 2 Distribution of multiple FBG domains in the members of FREP family in A. gambiae. The protein is represented by a line with the number above corresponding to amino acids which start from the N-terminus of each protein. The identified domains are shown under the line. FReD represents FBG domain. ZnMc represents Zinc-dependent metalloprotease domain. The sequences are denoted by their gene name in GenBank.
Using the multiple alignment of the FBG domains as queries, the secondary structure was predicted with the PHD program. The results show that the FBG domains have a highly conserved structure profile throughout the FBG domain (Fig. 1). By comparison of the predicted secondary structure with multiple alignment, most of these secondary structures fall in the conserved region, suggesting that FBG domains have similar domain architectures in the FREP gene family. To further compare the predicted secondary structure of the FBG domains with known structures, we found that the FBG domain is structurally related to the human fibrinogen γ fragment and the FBG domain of TL5A in the protein data bank (PDB) (Fig. 3A and 3B) [16,17]. The FBG domains of human fibrinogen γ fragment and TL5A compose the central and larger domain B and a relatively smaller domain P (Fig. 4). The domain B is predominantly built up by a twisted seven-stranded antiparallel β-sheet (strands β3-β7, β9 and β12) and helices α4 and α5 (Fig. 3A), and their tertiary structure is very similar (Fig. 3B). The domain P possesses only a few short elements of secondary structure, and comprises the major functional site forming a binding pocket [16]. The predicted secondary structures of the FBG domains in the FREP gene family approximately correspond to the domain architectures of FBG domains in human fibrinogen γ chain and TL5A. The β-sheets and α-helices in the predicted structure of the FBG domain are highly conserved with the corresponding structures in TL5A, especially in the domain B (Fig. 1 and Fig. 3B). For example, the central strand β12, which extends the C terminus of domain P back to domain B and brings both polypeptide termini in close proximity, was also seen in the FBG domains (Fig. 1 and Fig. 4). This suggests that the FBG domain architecture is conserved between houseshoe crab and mosquitoes. The projection of some of the highly conserved domains that form the ligand-binding pocket suggests that the core structure of the ligand-binding pocket is also likely to be conserved across these FBG domains (Fig. 1, Fig. 3B and Fig. 4). These observations imply that the FBG domains are most likely to function as receptors for carbohydrates or their derivatives. Beyond the common core, FBG domains also show great diversity in terms of the insertions and deletions among the conserved domains. Some FBG domains lose a conserved domain due to deletion, such as EAL39350. Other members have a short insertion located in the loop region, such as EAA10406 and EAA15009 (Fig. 1). By comparison of amino acids in the FBG domains of FREP corresponding to the P domain binding site in TL5A, we found that the domain architectures of these FBG domains have considerable diversity that is incorporated into a shared basic architectural blueprint (Fig. 1).
Figure 3 Ribbon representation of the core structure of the FBG domain of tachylectin 5A (PDB: 1JC9) and recombinant human γ-fibrinogen carboxyl terminal fragment (PDB: 2FIB). A. Ribbon plot of the FBG domain of TL5A. The domain shown here is a cartoon representation from the crystal structure. Main α-helices and β-sheets were shown in the figure. The residues forming the ligand-binding packet are depicted in the stick format and labeled in red. B. Superposition of the crystal structure of the FBG domain of TL5A (grey) and human γ-fibrinogen carboxyl terminal fragment (golden). By aligning TL5A and the γ chain fragment, the region composed of 178aa residues at the C-terminal regions of both proteins was used to generate superposition ribbon plot. Loop P-1 and P-3 in fibrinogen γ chain fragment are represented in green.
Figure 4 Topology diagram showing the arrangement of secondary-structure elements in the FBG domains of TL 5A. Domains named in analogy to human fibrinogen γ chain fragment. α-helix is represented in green and β-sheet is represented in brown. Domain B and domain P are separated by a red line. Starting position of amino acid in each secondary structure is shown in the figure with single letter. The disulfide bridge (Cys-206-Cys-219) in the domain P is represented by a dot line.
In domain P of TL5A, a disulfide bridge Cys-206-Cys-219 is an important structure to connect the metal-binding site to the acetyl group recognition site. These two conserved cysteines were seen in the FBG domains of the FREP family (Fig. 1). Furthermore, four aromatic side chains (Tyr-210, Tyr-236, Tyr-248, and His-220), which can form a funnel to the acetyl-group in TL5A, were also seen in most of the FBG domains of FREPs (Fig. 1). In some of the FBG domains, the amino acids corresponding to the binding sites have mutated. This great diversity probably provides the variability necessary for these FBG domains to form slightly different binding sites that could recognize different carbohydrates. This provides a diverse and potential flexible arsenal for the host to recognize a variety of correspondingly diverse carbohydrates on the surface of pathogens. Alternatively, it is likely that some of the FBG domains have other unknown functions besides recognition. Beyond the conservation of the full FBG domains in the FREP gene family, FBG domains show great variety in terms of their lengths. Multiple sequence alignment shows that 24 of the 53 FREPs consist of truncated FBG domains (data not shown). Multiple sequence alignment shows that 24 of the 53 FREPs consist of truncated FBG domains (Table 1). The lengths vary from 30 to 160 amino acids. Many of them are truncated in the C-terminus. By scanning the corresponding genome sequences using Artemis, we found some of the truncated parts of the FBG domain exist in the genome in close relation to the annotated fragment, suggesting that the truncation probably was a missannotation of the genome. By comparison of sequence similarity and structural profile, the recognition sites in the FBG domains of FREP and TL5A correspond structurally to the polymerization pocket in the fibrinogen γ fragment (Fig. 3A and 3B). Five of the seven amino acids that form the polymerization pocket are structurally equivalent to amino acids in the sugar-binding site of TL5A. The long loops P-1 and P-3 in the fibrinogen γ fragment are shortened by 14 and 7 amino acids respectively in the FBG domains and TL5A, and represent the major structural differences found in the functionally important domain P. The domain P also has very different surface charge in the two structures. On 1FIB, it forms a highly negative charged patch (Fig. 5A), while it is mainly hydrophobic on 1JC9 (Fig. 5B), which probably contributed to their target specificities. Variability in this domain points to a potential evolutionary transition from a carbohydrate to a protein-binding module [16,17].
Figure 5 Recombinant human γ-fibrinogen carboxyl terminal fragment (A) and surface of electrostatic potential of tachylectin 5A (B). A. Negative charged patch was outlined in circle. B. Hydrophobic groove was outlined in circle. The orientation is the same in both A and B. Red is for negative charge, blue is for positive charge and grey is non-polar areas.
Phylogenetic relationships of the FBG domains in A. gambiae
To understand the evolutionary history of this gene family, an attempt was made to identify correlations between chromosomal locations of FREP and FBG domain sequence similarities among the family members. The genes for the FREP family in A. gambiae have been mapped to specific A. gambiae chromosomal locations by retrieving Locuslink from Ensembl. Of the 53 FREP genes, chromosomal locations could be determined for (Fig. 6). The majority of FREP genes are found in clusters on chromosomes 2L and 3L, and some of these genes are arrayed in tandem. Twenty three genes located on chromosome 3L form 2 large clusters and 10 genes located on the chromosome 2L form two small clusters. This suggests that the FREP gene family evolved by expansion. FBG domains tandemly linked present a target for mispairing and unequal crossover, which could have resulted in duplication and divergence of the genes over time. These tandemly duplicated FBG domains could then become physically separated through chromosomal rearrangements and translocation. This suggests a dynamic history for the FBG domains that is likely to have involved gene expansion, with the FREP gene family evolving through vast expansion of the FBG domain.
Figure 6 Genomic distribution of FREP family members in A. gambiae. Chromosomes are represented with a line and chromosomal numbers are shown on the top of each chromosome. Chromosomal loci of the FREP genes are shown with their name. The proteins are denoted by their gene name in GenBank.
To analyze the evolutionary history of FBG domains in the FREP family, a phylogenetic tree was constructed with the alignments of the conserved FBG domains using maximum-likelihood methods (Fig. 7). This tree showed that the FBG domains were grouped into several branches. However, a major branch was observed in the evolutionary tree of the FBG domains. This branch is comprised largely of FBG domains of the FREP family from the A. gambiae genome. If the number of FBG domains increased mainly by tandem duplication, we would expect the domains which are physically clustered in the genome to form a monophyletic group. However, by examinating the relationships between phyletic pattern and chromosomal location of the FBG domains, it is found that some FBG domains grouped together in the phylogenetic tree are located on different chromosomes, such as EAA09906 and EAA04072, EAA43404 and EAA13725 (Fig. 6 and Fig. 7). This suggests that a dynamic history for the FBG domains likely involved shuffling among chromosomes. The predicted role, for at least a subset of these FBG domains, is in carbohydrate sensing. This expansion in the A. gambiae genome may have been a response to the diversity of carbohydrates encountered, resulting in the utilization of numerous FBG domain variations in order to recognize a broad range of different carbohydrates.
Figure 7 Phylogenetic tree of the FBG domains of the FREP family in A. gambiae. Phylogenetic relationships of the FBG domains are shown. The seed alignment used for constructing the tree was the multiple alignment sequences shown in Fig. 1. Maximum-likelihood approach was used to construct the tree with the proml program of the PHYLIP package, which uses the Jones-Taylor-Thornton model of change between amino acids and a Hidden Markov Model (HMM) method of inferring different rates of evolution at different amino acid positions. The FBG domains of each FREP are denoted by their gene name in GenBank.
ESTs for FREPs in mosquitoes
To confirm that the conceptual FREP proteins predicted from the genome are actually transcribed in mosquitoes, we searched the A. anopheles EST database. Twenty one of the 53 predicted genes were identified to have transcripts (Table 1). Examination of the transcript resources reveals that these genes are likely to be expressed in different tissues in mosquitoes, such as fat body, midgut and head (Table 1 and 2). Some of these genes also are expressed following immune challenges and a blood meal. These results suggest that FREP genes probably play a role in immune responses or any of the diverse array of physiologies associated with blood feeding. However, more than 50% of the predicted FREP genes have not been identified transcriptionally in the EST database. It is possible that the EST database does not cover the entire transcriptome and greater coverage is needed. To compare the actual transcripts of the FREP genes in different mosquito species, the FREP transcripts in Ar. subalbatus and Aedes aegypti were searched in the immune challenged hemocyte EST databases at ASAP [18]. Five and 12 different genes were transcribed respectively in the bacteria-challenged hemocytes. This suggests that some of the FREP genes are hemocyte-associated and possibly involved in innate immune responses post bacteria inoculation [19].
Table 2 Description of EST libraries from A. gambiae and D. melanogaster
Name Description Supplier
A. gambiae
NAP1 mix developmental stages European Molecular Biology
NAFB Normalized Fat Body Library University of Notre Dame
cDNA1 Adult cDNA1 Celera Genomics
4A3B cDNA libraries derived from immune-responsive hemocyte-like cell lines
blood1 Adult with blood-fed cDNA Celera Genomics
NAH Normalized Anopheles Head University of Notre Dame
IRB Infected Rat Blood-fed 30 hr Abdomen, Female adult 5–7 days post eclosion University of Notre Dame
D. melanogaster
GH Adult male and female head
RE normalized Embryo from male and female, 0–24 hours mixed stage embryonic Lawrence Berkeley National lab
LP Whole body Larval-early pupal from male and female
RH Adult male and female normalized Head pFlc-1 Lawrence Berkeley National lab
EK Mixed stage embryos, imaginal disks and adult head Lawrence Berkeley National lab
GM Ovary, newly eclosed females, germarium-stage 6, female.
SD Schneider L2 cell culture pOT2, cell line British Columbia Cancer A
CA Male and female salivary gland, 16, 18, 20, 22, and 24 hrs after puparium formation
EC Fat body-3rd instar larva Lawrence Berkeley National lab
Fibrinogen-related proteins in D. melanogaster
D. melanogaster is an important experimental insect and is used as a standard research model in the biomedical sciences. D. melanogaster is closely related to mosquitoes, with both insects belonging to the order Diptera. To compare the evolutionary development of the FBG domains between mosquito and fruitfly, detailed analyses of conserved segments were conducted. By searching the NCBI database, 20 FREP conceptual proteins were predicted in the D. melanogaster genome (Table 1). The multiple alignment of the FBG domain sequences showed that conservation exists throughout the FBG domain region (Fig. 8). Truncated FBG domains also exist in FREPs in D. melanogaster (Table 1). For example, two members of the FREP gene family have 3'-truncated FBG domains (AAF44911 and AAL48972). To further understand the relationships of the FBG domains between A. gambiae and D. melanogaster, a phylogenetic tree was constructed by using the conserved FBG domains from both species. The most striking pattern observed in the evolutionary tree was the presence of multiple branches comprised largely of proteins from a single organism (Fig. 10). These lineage specific expansions accounted for most of the FBG domains in A. gambiae and D. melanogaster. Furthermore, a branch comprised of the FBG domains from both A. gambiae and D. melanogaster was also noted (EAA01294, EEA15009, AAF55227, AAA28880) (Fig. 10).
Figure 8 Multiple sequence alignment of a representative set of the FBG domains of FREP in D. melanogaster. Multiple sequence alignment was constructed using T-Coffee program. The 100% consensus sequence was boxed with black in the alignment. The PHD secondary structure is shown above the alignment with H representing an α-helix and E representing a β-strand. The sequences are denoted by their gene name in GenBank.
Figure 9 Genomic distribution of FREP family members in D. melanogaster. Alternative spliced transcripts from the same gene are represented with [. The others are as detailed in Figure 6.
To determine genomic distribution of FREP members, the chromosomal location for every sequence was found by using the Locuslink program. Position information showed that some genes have more than one transcript (Fig. 9). To get the detailed information about these genes, a comparison of mRNA and genome sequences was performed by using the Spidey program at NCBI. The results showed that the predicted proteins from the same genes are generated by alternative splicing among exons and introns post transcription. Some of the FBG domains come from the same transcription region, such as AAF46535 and AAM52597. This would generate the same FBG domains. However, some of the FBG domains are generated from different regions. For example, transcription of the FBG domain in AAN09447 is located in a big intron between the first two exons in AAF48780, resulting in different FBG domains. To determine the actual fully processed transcripts of these genes, a search of the EST database was conducted. Thirteen of the 20 FREP proteins were identified in the D. melanogaster transcript database. By examining the genomic location, we found that AAF49079, AAN11645 and AAM11109 are transcribed from the same gene. The actual transcripts of these 3 gene products are also represented in the EST database. This further illustrates the complexity of gene regulation post transcription, which could provide multiple protein products from a single gene, thereby, further increasing variation in the FREP family.
Figure 10 Phylogenitic tree of the FBG domains from A. gambiae and D. melanogaster. The seed alignment used for constructing the tree was the multiple alignment sequences of representative set of the FBG domains of FREP families in A. gambiae and D. melanogaster. The phylogenetic tree was constructed as described in methods and detailed in Fig. 5. The FBG domains of each FREP are denoted by their gene name in GenBank. The name of the FREP from A. gambiae start with E, and the name of the FREP from D. melanogaster start with A.
Compared with the D. melanogaster FREP gene family, the massive expansion of the FREP gene family in mosquitoes probably is associated with particular aspects of the mosquito's biology, possibly hematophagy and exposure to parasites [15]. The blood meal imposes challenges associated with proliferation of the microbial flora in the gut and coagulation of ingested blood and penetration of the midgut by blood-born pathogens. A FREP protein (e.g AL-1) in the mosquito Ar. subalbatus has bacteria binding properties, and it has been suggested that FREP may be important in controlling bacteria infections in mosquitoes [13]. However, mosquitoes may use a number of FREP proteins as anticoagulants, for instance, as competitive inhibitors preventing polymerization of blood [15]. Some mosquito FREP genes are up-regulated by invading malaria parasites [20,21], suggesting a possible role in an antimalarial defense system.
Conclusion
The detailed sequence and structural analyses disclose that the FREP family contains highly similar FBG domains in the A. gambiae genome. FBG domains are predicted to recognize carbohydrates and their derivatives. The sequence divergence seen in the binding domains of FBG domains makes it possible to recognize a wide range of carbohydrate derivatives. This suggests that the FREP family may play an important role in innate immunity. Expansion of the family during evolutionary history is mainly accounted for by a major expansion of the FBG domain architectures. Further analysis of the chromosomal locations and phyletic patterns of the FBG domains suggest that they have been acquired by tandem duplication and shuffling. Compared with D. melanogaster, the massive expansion of the FREP family in A. gambiae probably is associated with particular aspects of the mosquito's biology, such as exposure to parasites and hematophagy. Experimental investigations of these proteins are likely to be of interest in understanding insect innate immunity and physiology.
Methods
Database searching and sequence retrieving for fibrinogen-related protein
A PSI-BLAST search [22] of the A. gambiae and D. melanogaster genome database at the National Center for Biotechnology Information (NCBI) [23] was performed using AL-1 as a query. To obtain the recent progress of FREP in A. gambiae genome, the A. gambiae database at Ensembl [24] was also searched. Following accumulation of the complete list of accession numbers, the corresponding protein sequence was retrieved from GenBank at NCBI and Ensembl.
Signal peptide prediction
Signal peptides were predicted using the SignalPv3.0 [25,26].
Searching for ESTs database
To determine the actual transcripts for individual FREP genes, BLAST search of an EST database at Berkeley Drosophila Genome Project and TIGR A. gambiae Gene Index (AgGI) was performed [27,28]. The annotated cDNA sequences encoding FREPs identified in the PSI-BLAST search were used as queries for individual BLAST search in these EST database. The availability of EST was determined based on sequence similarity with the query: a 97% or greater identity was considered to be an EST corresponding to a specific gene. To get information about FREP transcripts in the mosquito, Ar. Subalbatus and Ae. aegypti, hemocyte EST databases at ASAP in both species were searched using AL-1 as a seed [18,29].
Multiple sequence alignment and phylogenetic analysis
Multiple sequence alignment was performed using the T-Coffee program [30,31]. Phylogenetic analysis was carried out with the maximum-likelihood algorithm [32]. The package used for phylogenetic analysis was proml program from PHYLIP [33], and the unrooted tree was draw using drawtree program in this package.
View of DNA sequence annotation
To verify the annotation of truncated genes, the corresponding genomic sequences was scanned by Artemis [34].
Secondary structure prediction
Secondary structure prediction was produced with the PHD program [35], with multiple alignment of individual FBG domains of FREP family. The structure data of TL5A and recombinant human γ-fibrinogen carboxyl terminal fragment were obtained from protein data bank (PBD) [36] and the ribbon diagrams were constructed with Molmol program [37].
Chromosomal location and alternative splice transcripts
The chromosomal location of the FREP genes in A. gambiae genome was retrieved at Ensembl [24]. The chromosomal location of the FREP genes in D. melanogaster was retrieved at NCBI [23]. To identify alternative spliced transcripts for each gene, spidey, a cDNA-to-genomic alignment program, was used to align spliced sequences to genomic sequences, using local alignment algorithms and heuristics to put together a global spliced alignment [38].
Abbreviations
FBG domain, fibrinogen-like domain; FREP, fibrinogen-related protein; AL-1, aslectin, TL5A, tachylectin 5A; GlcNAc, N-acetylglucosamine; MAP, microfibril-associated protein; aa, amino acid; BLAST, basic local alignment search tool; PSI-BLAST, position specific iterative BLAST; EST, expressed sequence tag; PDB, protein data bank; Molmol, molecule analysis and molecule display.
Authors' contributions
XW carried out the database survey. He identified and analyzed the FBG domains, and prepared the manuscript. QZ generated ribbon diagram and did structure analyses. BMC conceived the study and contributed to the preparation of the manuscript. All authors read and approved the final manuscript.
Acknowledgements
We thank Thomas A. Rocheleau and George Mayhew for critically reading the manuscript and useful discussion. We are grateful to Anthony Nappy for assistance with graphics. This study was supported by NIH grant AI 19769.
==== Refs
Gorkun OV Veklich YI Weisel JW Lord ST The conversion of fibrinogen to fibrin: recombinant fibrinogen typifies plasma fibrinogen Blood 1997 89 4407 14 9192765
Lu J Le Y Ficolins and the fibrinogen-like domain Immunobiology 1998 199 190 199 9777405
Matsushita M Fujita T The role of ficolins in innate immunity Immunobiology 2002 205 490 497 12396010 10.1078/0171-2985-00149
Erickson HP Tenascin-C, tenascin-R and tenascin-X: a family of talented proteins in search of functions Curr Opin Cell Biol 1993 5 869 76 7694605 10.1016/0955-0674(93)90037-Q
Kobayashi R Mizutani A Hidaka H Isolation and characterization of a 36-kDa microfibril-associated glycoprotein by the newly synthesized isoquinolinesulfonamide affinity chromatography Biochem Biophys Res Commun 1994 198 1262 6 8117283 10.1006/bbrc.1994.1178
Lu J Teh C Kishore U Reid KB Collectins and ficolins: sugar pattern recognition molecules of the mammalian innate immune system Biochim Biophys Acta 2002 1572 387 400 12223281
Teh C Le Y Lee SH Lu J M-ficolin is expressed on monocytes and is a lectin binding to N-acetyl-D-glucosamine and mediates monocyte adhesion and phagocytosis of Escherichia coli Immunology 2000 101 225 32 11012776 10.1046/j.1365-2567.2000.00099.x
Zhao Z Lee CC Jiralerspong S Juyal RC Lu F Baldini A Greenberg F Caskey CT Patel PI The gene for a human microfibril-associated glycoprotein is commonly deleted in Smith-Magenis syndrome patients Hum Mol Genet 1995 4 589 97 7633408
Gokudan S Muta T Tsuda R Koori K Kawahara T Seki N Mizunoe Y Wai SN Iwanaga S Kawabata S Horseshoe crab acetyl group-recognizing lectins involved in innate immunity are structurally related to fibrinogen Proc Natl Acad Sci USA 1999 96 10086 10091 10468566 10.1073/pnas.96.18.10086
Adema CM Hertel LA Miller RD Loker ES A family of fibrinogen-related proteins that precipitates parasite-derived molecules is produced by an invertebrate after infection Proc Natl Acad Sci USA 1997 94 8691 8696 9238039 10.1073/pnas.94.16.8691
Kenjo A Takahashi M Matsushita M Endo Y Nakata M Mizuochi T Fujita T Cloning and characterization of novel ficolins from the solitary ascidian, Halocynthia roretzi J Biol Chem 2001 276 19959 19965 11259430 10.1074/jbc.M011723200
Schroder HC Ushijima H Krasko A Gamulin V Thakur NL Diehl-Seifert B Muller IM Muller WE Emergence and disappearance of an immune molecule, an antimicrobial lectin, in basal metazoa. A tachylectin-related protein in the sponge Suberites domuncula J Biol Chem 2003 278 32810 7 12805362 10.1074/jbc.M304116200
Wang X Rocheleau TA Fuchs JF Hillyer JF Chen CC Christensen BM A novel lectin with a fibrinogen-like domain and its potential involvement in the innate immune response of Armigeres subalbatus against bacteria Insect Mol Biol 2004 13 273 82 15157228 10.1111/j.0962-1075.2004.00484.x
Redfern O Grant A Maibaum M Orengo C Survey of current protein family databases and their application in comparative, structural and functional genomics J Chromatogr B Analyt Technol Biomed Life Sci 2005 815 97 107 15652801
Zdobnov EM von Mering C Letunic I Torrents D Suyama M Copley RR Christophides GK Thomasova D Holt RA Subramanian GM Mueller HM Dimopoulos G Law JH Wells MA Birney E Charlab R Halpern AL Kokoza E Kraft CL Lai Z Lewis S Louis C Barillas-Mury C Nusskern D Rubin GM Salzberg SL Sutton GG Topalis P Wides R Wincker P Yandell M Collins FH Ribeiro J Gelbart WM Kafatos FC Bork P Comparative genome and proteome analysis of Anopheles gambiae and Drosophila melanogaster Science 2002 298 149 59 12364792 10.1126/science.1077061
Kairies N Beisel HG Fuentes-Prior P Tsuda R Muta T Iwanaga S Bode W Huber R Kawabata S The 2.0-A crystal structure of tachylectin 5A provides evidence for the common origin of the innate immunity and the blood coagulation systems Proc Natl Acad Sci USA 2001 98 13519 24 11707569 10.1073/pnas.201523798
Yee VC Pratt KP Cote HC Trong IL Chung DW Davie EW Stenkamp RE Teller DC Crystal structure of a 30 kDa C-terminal fragment from the gamma chain of human fibrinogen Structure 1997 5 125 38 9016719 10.1016/S0969-2126(97)00171-8
Glasner JD Liss P Plunkett G 3rdDarling A Prasad T Rusch M Byrnes A Gilson M Biehl B Blattner FR Perna NT ASAP, a systematic annotation package for community analysis of genomes Nucleic Acids Res 2003 31 147 51 12519969 10.1093/nar/gkg125
Bartholomay LC Cho WL Rocheleau TA Boyle JP Beck ET Fuchs JF Liss P Rusch M Butler KM Wu RC Lin SP Kuo HY Tsao IY Huang CY Liu TT Hsiao KJ Tsai SF Yang UC Nappi AJ Perna NT Chen CC Christensen BM Description of the transcriptomes of immune response-activated hemocytes from the mosquito vectors Aedes aegypti and Armigeres subalbatus Infect Immun 2004 72 4114 26 15213157 10.1128/IAI.72.7.4114-4126.2004
Dimopoulos G Christophides GK Meister S Schultz J White KP Barillas-Mury C Kafatos FC Genome expression analysis of Anopheles gambiae : responses to injury, bacterial challenge, and malaria infection Proc Natl Acad Sci USA 2002 99 8814 9 12077297 10.1073/pnas.092274999
Srinivasan P Abraham EG Ghosh AK Valenzuela J Ribeiro JM Dimopoulos G Kafatos FC Adams JH Fujioka H Jacobs-Lorena M Analysis of the Plasmodium and Anopheles transcriptomes during oocyst differentiation J Biol Chem 2004 279 5581 7 14627711 10.1074/jbc.M307587200
Altschul SF Madden TL Schaffer AA Zhang J Zhang Z Miller W Lipman DJ Gapped BLAST and PSI-BLAST: a new generation of protein database search programs Nucleic Acids Res 1997 25 3389 402 9254694 10.1093/nar/25.17.3389
The National Center for Biotechnology Information (NCBI)
The A. gambiae database at Ensembl
Nielsen H Engelbrecht J Brunak S von Heijne G Identification of prokaryotic and eukaryotic signal peptides and prediction of their cleavage sites Protein Engineering 1997 10 1 6 9051728 10.1093/protein/10.1.1
The SignalPv3.0
Berkeley Drosophila Genome Project
TIGR A. gambiae Gene Index (AgGI)
The mosquito, Ar. Subalbatus and Ae. aegypti, hemocyte EST databases at ASAP
Notredame C Higgins DG Heringa J T-Coffee: A novel method for fast and accurate multiple sequence alignment J Mol Biol 2000 302 205 17 10964570 10.1006/jmbi.2000.4042
The T-Coffee program
Felsenstein J Inferring phylogenies from protein sequences by parsimony, distance, and likelihood methods Methods Enzymol 1996 266 418 27 8743697
PHYLIP
Artemis
Rost B Sander C Prediction of protein secondary structure at better than 70% accuracy J Mol Biol 1993 232 584 99 8345525 10.1006/jmbi.1993.1413
Guex N Peitsch MC SWISS-MODEL and the Swiss-PdbViewer: an environment for comparative protein modeling Electrophoresis 1997 18 2714 23 9504803 10.1002/elps.1150181505
Koradi R Billeter M Wuthrich K MOLMOL: a program for display and analysis of macromolecular structures J Mol Graph 1996 14 51 5 8744573 10.1016/0263-7855(96)00009-4
Spidey
|
16150145
|
PMC1242226
|
CC BY
|
2021-01-04 16:32:49
|
no
|
BMC Genomics. 2005 Sep 8; 6:114
|
utf-8
|
BMC Genomics
| 2,005 |
10.1186/1471-2164-6-114
|
oa_comm
|
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-1181615690110.1186/1471-2164-6-118Methodology ArticleRapid single nucleotide polymorphism mapping in C. elegans Davis M Wayne [email protected] Marc [email protected] Tracey [email protected] Patrick [email protected] Shawn [email protected] Erik M [email protected] Department of Biology, University of Utah, Salt Lake City, Utah 84112-0840, USA2005 12 9 2005 6 118 118 3 8 2005 12 9 2005 Copyright © 2005 Davis et al; licensee BioMed Central Ltd.2005Davis et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
In C. elegans, single nucleotide polymorphisms (SNPs) can function as silent genetic markers, with applications ranging from classical two- and three-factor mapping to measuring recombination across whole chromosomes.
Results
Here, we describe a set of 48 primer pairs that flank SNPs evenly spaced across the C. elegans genome and that work under identical PCR conditions. Each SNP in this set alters a DraI site, enabling rapid and parallel scoring. We describe a procedure using these reagents to quickly and reliably map mutations. We show that these techniques correctly map a known gene, dpy-5. We then use these techniques to map mutations in an uncharacterized strain, and show that its behavioral phenotype can be simultaneously mapped to three loci.
Conclusion
Together, the reagents and methods described represent a significant advance in the accurate, rapid and inexpensive mapping of genes in C. elegans.
==== Body
Background
Single Nucleotide Polymorphism (SNP) mapping has transformed studies of genetic linkage in C. elegans since its introduction in 2001[1]. In SNP mapping, DNA sequence polymorphisms between the wild-type C. elegans strain (N2 Bristol) and a closely related strain (CB4856 Hawaiian) are used as genetic markers. Compared to other markers that have been used for genetic mapping, SNPs have two distinct advantages. First, unlike conventional marker mutations that cause visible phenotypes, SNPs in general have no associated phenotype. Thus, mutant phenotypes that are masked by conventional marker mutations, such as those with subtle behavioral defects, can be mapped using SNPs. Second, SNPs are far denser than other markers, including both visible markers and DNA polymorphisms such as Tc1 insertions. Because SNPs are approximately as dense as genes, SNP mapping can in theory provide single-gene resolution [2]. Together, these two advantages have made SNP mapping the technique of choice for many C. elegans researchers.
SNP mapping is usually done in two phases. The first phase, chromosome mapping, is similar to traditional two-factor mapping and seeks to identify the relevant chromosome and rough position of the gene of interest. The second phase, interval mapping, seeks to place the gene of interest in an interval between two SNPs, and can be used iteratively to fine map the gene. SNP detection in both phases is typically performed by using only SNPs that alter a restriction site, which are also known as snip-SNPs [1]. Although other SNP detection methods have been used for mapping in C. elegans, such as fluorescence polarimetry and indel detection [3,4], snip-SNPs are attractive because they require low initial investment and do not require specialized equipment. However, snip-SNP detection requires PCR amplification of the SNP region, digestion with the appropriate restriction enzyme, and gel electrophoresis. These multiple steps can be daunting, particularly during chromosome mapping when many SNPs need to be assayed simultaneously.
The method described here streamlines the procedure for detecting snip-SNPs, making faster, more efficient mapping possible. First, we identified 48 SNPs that met our criteria (8 per chromosome). This means that every part of the genome is linked to multiple SNPs, so that adjacent SNPs serve as internal controls, and also that sub-chromosome position can be determined. Second, we simplified the PCR step by identifying primer sets that all work at identical amplification conditions, and that are tolerant of being added to reactions by pin-replication. This enabled us to quickly amplify across 48 snip-SNPs in a single 96-well PCR plate. Third, we streamlined restriction digestion by using only snip-SNPs that can be distinguished by a single restriction enzyme, DraI. This enzyme is relatively inexpensive and tolerant of PCR buffer salts. This allowed us to perform digestion in the original PCR plate, by adding to each well an identical digestion cocktail. Finally, primer locations were designed so that the informative digestion products can be resolved on an agarose gel. The accessibility of this mapping procedure, together with its speed, low cost, robustness, and accuracy, should make it a preferred option for most C. elegans labs.
Results and discussion
Chromosome mapping
To facilitate mapping a mutation onto a chromosome, we designed a set of PCR reagents based on modifications of the principle of bulk segregant analysis described by Wicks et al. [1]. Our primary goal was to simplify the procedure by performing all steps in a 96-well format PCR plate, and designing each SNP reaction to be performed under identical conditions. To design a set of primers that can be used for SNP mapping in a 96-well format, two conditions must be met: the primers must all use the same conditions for polymorphism detection, and the primers must all use the same conditions for amplification. First, we simplified SNP detection by using only SNPs that could be detected using a single restriction enzyme. Since SNPs are concentrated in non-coding A/T-rich regions of the genome, we reasoned that good coverage would be obtained from the enzyme DraI, which recognizes the sequence TTT^AAA. We identified all DraI SNPs in a custom database (available as supplementary material) that incorporated all SNPs identified by the Genome Sequencing Center, Washington U, St. Louis, MO (6,333 total SNPs, 248 DraI SNPs in our database) [5] and by Exelixis, South San Francisco, CA (9,295 total SNPs, 257 DraI SNPs in our database) [3]. From among these we selected eight candidate DraI SNPs on each chromosome that were far enough from nearby DraI sites (typically >200 bp on one side, >50 bp on the other) to enable detection of cleavage at the SNP DraI site. The genetic positions of the resulting 48 DraI SNPs are shown in Figure 1.
Figure 1 Genetic position of SNPs used for mapping. The genetic position of each SNP was obtained from Wormbase Release 143.
Next, to enable simultaneous amplification of all selected SNPs in a 96-well format, we chose primer pairs with similar annealing temperatures and product length. The program Primer3 [6] was used to design primers to amplify short sequences (typically 300–500 bp) containing each selected DraI SNP. Optimum Tm was set to 60°C. Primer pairs were tested, and unsatisfactory pairs were redesigned until all 48 primer pairs amplified robustly in simultaneous PCR reactions in a single plate. The resulting primer sequences are shown in Table 1.
Table 1 DraI SNP primers, locations and band sizes. In each pair of primers, the left primer is listed first; all primer sequences are given 5' to 3'. Interpolated genetic positions are from [9] release WS143.
Genetic Location Physical Location Clone N2 digest CB4856 digest Primers Wormbase Identifier
I, -19 169, 017 F56C11 354, 146 500 ATGCCAGTGATAAGGAACGG snp_F56C11 [4]
TCACATCCCTTGTCGATGAA
I, -12 1,905,969 Y71G12A 503, 72 377, 126, 72 TCGAAATCAGGGAAAAATTGA snp_Y71G12 [3]
ACGATTTTCGGGGAGTTTTT
I, -6 2,818,973 W03D8 395, 144 538 GTTTTCACTTTTGCCGGTGT pkP1052
TGAAGGCGCATATACAGCAG
I, -1 4,594,014 D1007 325, 134, 41 459, 41 AAAATATCAGGAAAGAGTTTCGG snp_D1007 [7]
TTTAAAGATTAAGGGTGGAGCG
I, 5 10,722,146 B0205 494 365, 129 ATCTGGCACCAAATATGAGTCG CE1-247
AATCTCGATTTTCAAGGAGTGG (rs3139013)**
I, 13 12,047,594 F58D5 445 295, 151 TCCTGGATAATCCCCAAAAA snp_F58D5 [4]
CCCTGCCATTGATCTTGTTT
I, 14 12,729,812 T06G6 236, 99, 78 335, 78 TTGAAATCCCCTTTAAAATCCC uCE1-1361
ACACTGGGTACCTGACTCATGC
I, 26 14,682,016 Y105E8B 360, 114, 27 474, 27 ATTATTAACGGCCACGGTGA snp_Y105E8B [3]
CCCACACACTCTCACCTTCA
II, -18 176,720 T01D1 263, 112 375 CCGAATTTTCAAATGGATGC pkP2101
CCATTGGAATTGCACACAAA
II, -14 2,121,018 R52 345 236, 109 CTGTGCTGTTGACGATATTGG snp_R52 [5]
ATGTCTCATTGCAAAATTCGG
II, -6 3,828,599 F54D10 516 387, 129 TTGTGAGCTTATATCTCAGTTGTCG pkP2103
AGATTTGGTTAGAAATATCACCGC
II, 1 9,052,466 T24B8 373, 121 494 TCAAAAACTTACAATCAATCGTCG snp_T24B8 [1]
CCAGAAAATCTGCACAGAAGG
II, 4 11,827,835 Y6D1A 224, 117, 124, 44 340, 124, 44 TTCTTCAAAAAGTCTAGGTTCAGCA snp_Y6D1 [1]
GGGGACGAAAACGGAGTTTG
II, 11 12,605,350 Y38E10A 483 352, 132 ACCGTTTAATAGGATTATTTGGG uCE2-2131
AAGTCTGCGGAATAATTGATGG
II, 16 13,235,564 F15D4 500 368, 132 TTCCAGGTAATACACATACAACTCC pkP2116
AAAAACACAAAGTTCAAAAACCC
II, 22 14,132,466 K09E4 365, 119 484 CCACTGGCTATAAGCTTTTCTAGG CE2-215
TAAGGATTTCAGGCTTTTAGGC (rs3139227)**
III, -25 939,698 T12B5 206, 189 395 TATCATCGAAATCCCGGAAA uCE3-637
TTCGGACGGGAGTAGAATTG
III, -19 1,827,732 Y39A3CL 342, 78, 76 272, 78, 76, 70 TCCCAATTTCCCTCTAAAAACC uCE3-735
TTGAATTTGGACCATTTTGAGG
III, -12 2,599,699 Y71H2B 368, 105 473 GAGGAACCAAATCTGGCGTA snp_Y71H2B [2]
TGAAAACTTGGAAAATCGGTG
III, -7 3,359,033 F45H7 239, 85, 27 196, 85, 43, 27 AATTTGAATCAGTGACTTTTGGC CE3-127
TTTCTGCAAACATTTTTCTTCG (rs3139272)**
III, -1 7,320,107 F56C9 486 354, 132 AAAAATACATGTCTACACAACCCG snp_F56C9 [1]
TTTCTTATCACTGTGCAGTCTTACC
III, 4 10,652,476 Y39A1A 355, 142, 30 497, 30 AGCGTTAAAGTATCGGTTATTTCG snp_Y39A1 [9]
TAAATTCATTTCAAACAATCGAGC
III, 12 11,656,188 Y41C4A 339, 156 495 ATCAAGTTTCTGATTGCTCTTTCC snp_Y41C4 [2]
AAAAACGTGATTTTTCAATTTTGC
III, +21 13,715,622 W06F12 273, 137, 78 200, 137, 78, 73 AGCAGGCTCACCATCATCATCA uCE3-1426
GACATTACGGTAGAGGAGATGGA
IV, -24 795,461 F56B3 301, 128, 71 429, 71 TGATGGTGTGTCTGCGTACC uCE4-515
AGAGCTGGAGAGCACGGATA
IV, -16 1,799,032 Y38C1BA 187, 304 491 CGCATAAATCCAACGTTCTCTG snp_Y38C1B [2]
AATCCATAAGTTTCGTGTTGGG
IV, -7 2,761,525 Y54G2A 498 250, 248 ACTCGGCATCCTCACGC snp_Y54G2 [5]
GTTGAAAATTTTTTCATAGCTATCATC
IV, -5 3,347,952 F42A6 295, 124 419 TGCTGAAATATTGGAAAATTGAGG pkP4055
TTATATCGTCGAGGAGGTTAGAGG
IV, 1 4,991,851 E03H12 376 300, 76 AAAATGGGAAGCGTACCAAA pkP4071
TGCTTGTAGCGTTTCCAAGA
IV, 8 13,049,020 F49E11 313, 77 390 GACACGACTTTAGAAACAACAGC snp_F49E11 [1]
TGGTATGGAGTCCCTATTTTGG
IV, 12 14,566,396 Y57G11B 284, 162, 52* 327, 119, 52* TGTAAATACCCCACATTTCAAGC snp_Y57G11B [2]
AAATTTCCAATTGTTCAAAGCC pkP4095*
IV, 14 16,085,085 Y105C5B 241, 108, 78, 48 319, 108, 48 TCGAATTGTTGTGTTTCTTTTGA pkP4099
TTCCAATTTTCTCGGTTTGG
V, -17 1,773,464 F36H9 307, 87, 79 386, 87 TTTCGGAAAATTGCGACTGT pkP5076
CGCGTTTTGGAGAATTGTTT
V, -13 2,726,662 C24B9 288, 167 455 TCATCTGTTATTTCGTCTCTTGC uCE5-828
CGGTAATAATATGCTTTGTGGG
V, -5 4,550,757 Y61A9LA 454 307, 147 GAGATTCTAGAGAAATGGACACCC snp_Y61A9L [1]
AAAAATCGACTACACCACTTTTAGC
V, 1 7,089,411 VC5 435, 70 300, 135, 70 AGAAATGATCCGATGAAAAAGC pkP5097
CCGATAGTGTTCATAGCATCCC
V, 6 13,951,850 R10D12 500 348, 152 CAAATTAAATATTTCTCAAAGTTTCGG ***
ACATAAGCGCCATAACAAGTCG
V, 10 16,321,481 F57G8 475 288, 187 TAAAGCCGCTACGGAAATACTC pkP5129
ATTTTCTCCCTAATTCCAGGTG
V, 13 17,610,508 Y6G8 282, 205 487 CATTCATTTCACCTGTTGGTTG uCE5-2609
TCGGGAAGATAATCAAAATTCG
V, 18 18,782,547 Y17D7B 324, 164 488 GAAATTCAAATTTTTGAGAAACCC snp_Y17D7B [3]
TTCAGACCATTTTTAGAATATTCAGG
X, -17 2,065,464 F49H12 540 321, 219 ATATGTGAGTTTACCATCACTGGG pkP6143
ACGTTTTGAAAAATTTGGTTGC
X, -8 4,161,493 ZK470 422, 72, 40 326, 96, 72, 40 CCAAAACGGCCAAGTATCAG pkP6105
TTGCACTCTTTCTCCTTCCG
X, -4 5,934,688 C46F4 169, 54, 51, 35, 22 223, 51, 35, 22 AAGTGTTCAATGATTTTGTCTAATTG uCE6-981
TGACAGGAGAATACTTTTGAAGG
X, 2 10,637,922 F11A1 409, 133 542 AGCAACAAACAATGCAACTATGG snp_F11A1 [2]
TAAACAAGAGGGTACAAGGTATCG
X, 8 12,750,713 F22E10 341, 126 467 TTAAAACCATACAATTCTTCTCAGC snp_F22E10 [1]
GAATTCCCAATCAACAGAGAGC
X, 11 13,339,566 F46G10 318, 191, 37 509, 37 ACTGTTTACCGCGTCTTCTGC pkP6132
CCGTGTATATAAGAAAATGTGTTCG
X, 17 14,547,382 T24C2 409, 34 302, 107, 34 GCTGGGATTTTGAAGAGTTGTT uCE6-1459
CAGTGAATCATCCGTTGAATTT
X, 23 15,500,013 H13N06 358, 134 492 CAAATACCAAAGTTGATCGTGG uCE6-1554
TTGTTGCAATTAAATCAAACGG
*Y57G11B has two DraI SNPs within a single PCR product.
**dbSNP IDs are given where the SNP has been submitted to the NCBI dbSNP database [10].
***The SNP on R10D12 has not been added to wormbase. It can be found at [11]
Finally, we devised a set of procedures that maximize speed and minimize the potential for error during reaction set up and gel loading (Figure 2). Our chromosome mapping procedure begins with the same genetic manipulations as other SNP mapping protocols. Hawaiian males are crossed into the mutant strain to produce heterozygous F1 animals. Homozygous F2 animals from the heterozygous F1 animals are identified based on their mutant phenotype. At the same time, animals with a non-mutant phenotype, which are enriched for Hawaiian sequences at the locus of interest, are also isolated (Figure 2). Thirty to fifty animals of each class are combined into two tubes and lysed using detergent and proteinase K (Figure 2). A PCR master mix, not including PCR primers, is then assembled and added to each lysate. These PCR mixes are then dispensed into alternating rows of a 96-well PCR plate using a multi-channel pipettor (Figure 2). These three simple steps generate a set of 96 PCR master mixes ready for the final addition of primers.
Figure 2 Procedure for chromosome mapping. (A), Method. Typically 30 mutant animals (homozygous Bristol DNA surrounding the mutation) and 30 wild-type animals (heterozygous Bristol/Hawaiian or homozygous Hawaiian DNA) are lysed in 20 μL lysis buffer. The lysate is then added to a PCR mix lacking primers, and the mix is aliquoted into every other row of a 96 well plate. Primers are added by pin replication from a master plate. Because the 8-channel pipette loads every other lane of the gel, each mutant reaction is placed next to its control. DNA ladder is typically placed in lanes 17 and 34. (B), Results from homozygous N2 Bristol and CB4856 Hawaiian genotypes. 50 Bristol adults and 50 Hawaiian adults were lysed in 20 μL lysis buffer, and used for the DNA template for the 48 PCR reactions covering all six chromosomes. Note that pure Bristol and Hawaiian DNA was used for each PCR reaction in the gel shown. When mapping a recessive mutant in the Bristol background against the Hawaiian strain, unlinked SNPs will display a 50-50 mix of Bristol bands and Hawaiian bands in both mutant and non-mutant lanes. Linked SNPs will display an enrichment of Bristol bands in the mutant lane, approaching 100% Bristol for tight linkage. The non-mutant lane will display a 2/3 to 1/3 enrichment of Hawaiian compared to Bristol DNA.
Because there are 96 separate reactions, each requiring addition of a specific primer pair, we generated a pre-arrayed set of primers, which are then added to the PCR master mixes by pin replication. Primer pairs described above are arrayed in pairs into a microtiter plate at 10 μM each primer ('primers' Figure 2A), with each row containing the primer pairs for the eight SNPs along a single chromosome. Adjacent rows contain a duplicate set of primers for a particular chromosome, and the plate of primers is pin-replicated into the master PCR mix. After amplification, PCR products are digested in the plate with DraI in a final volume of 15 μL and loaded onto a 2.5% agarose gel using an 8-channel pipette. Because we use a gel comb with wells spaced half the distance between pipette tips of the multi-channel pipette, we can automatically load the mutant samples from the upper row and wild-type samples from the lower row for each SNP pair in adjacent wells. The resulting gel displays all 48 SNP markers, from left to right and from chromosome I to X (Figure 2B). Each mutant SNP is next to its non-mutant control, so that the whole genome can be quickly scanned for linkage.
To validate the final set of primers for chromosome mapping experiments, we used them to map dpy-5, a mutation with a well known genetic position. We crossed CB4856 Hawaiian males to a triply-marked mapping strain, EG1000 dpy-5(e61) I; rol-6(e187) II; lon-1(e1820) III. We allowed the heterozygous F1s to self, and from the F2 generation we picked 50 Dpy and 50 non-Dpy animals into separate lysis reactions. We performed chromosome mapping PCR on these lysates using primer sets from LGI and LGII (Figure 3A). As expected, we found linkage to the center of LGI and no linkage to LGII.
Figure 3 SNP mapping dpy-5(e61). A, chromosome mapping. Each pair of lanes shows results from the SNP at the indicated genetic map position, using either the Dumpy (D) or the wild-type (+) template. Linkage is visible as an increase in the proportion of Bristol N2 DNA in Dumpy lanes compared to the wild-type lanes, and is visible on LGI from -12.2 to 15.5. B and C, interval mapping. Each column in B is an individual Dpy recombinant, assayed for the three SNPs W03D8 (top row), D1007 (middle row), and B0205 (bottom row). Most (31/47) recombinants show Bristol DNA at all three SNPs. This indicates that these recombinants were homozygous Bristol at these loci, as expected for tightly linked markers. Sixteen animals show half Bristol and half Hawaiian DNA at one or more loci, indicating that they have one chromosome that is recombinant in this interval. Columns marked "a" are recombinant in the W03D8-D1007 interval, those marked "b" in the D1007-dpy-5 interval, and those marked "c" in the dpy-5-B0205 interval. These data are summarized in C, which depicts the three recombinant genotypes using blue for Bristol DNA and red for Hawaiian DNA. One recombinant, marked with an asterisk, is homozygous Hawaiian at two SNP loci and heterozygous at the third, and is thus very unlikely to be homozygous Bristol at the dpy-5 gene (see Results for explanation).
Interval mapping
After determining the rough position of a mutation on a chromosome using chromosome mapping, mutations can be quickly mapped to a genetic interval using the same efficiencies of the 96-well format employed in chromosome mapping. Interval mapping differs from chromosome mapping in that the genotype of individual mutant animals, rather than the genotype of pooled animals, must be determined. Also, it is necessary to assay these mutant DNAs for many SNPs within the interval for which linkage has been established. Therefore, it is most convenient to pin replicate the DNA templates, rather than the primers (Figure 4).
Figure 4 Interval mapping. Individual recombinants are singled from heterozygotes and the animal, or a representative sample of their progeny, are placed in wells of a 96-well PCR plate and lysed. The plate may also contain three control wells, with Bristol, Hawaiian, and a 50-50 mix of animals. DNAs from the lysed animals are pin-replicated into a PCR master mix containing primers for the desired SNP. The plates are processed for PCR amplification, digested with DraI and samples run on an agarose gel.
Briefly, we crossed Hawaiian males into our mutant strain (isolated from Bristol N2) to generate a heterozygous strain. From the progeny of these heterozygotes, we singled 96 mutant animals onto worm growth plates and allowed them to lay self-fertilized embryos. After the F3 progeny had grown to the adult stage, we washed about a quarter of the F3 progeny from the plate into individual wells of the 96-well plate (see Methods). Deriving templates from the self progeny of a homozygous mutant has the advantage of allowing each mutant recombinant to be scored as a population, rather than a single animal. Also, having additional template is convenient for additional rounds of PCR if higher resolution is desired. We found that 96-well plates containing the lysed worms can be frozen at -80°C and reused successfully after many rounds of thawing and re-freezing. If more rapid mapping was required, we found it possible to remove the single F2 animal (after it had laid sufficient embryos to ensure propagation of the strain) and to lyse it for analysis of its SNPs. Specifically, a drawn out and sealed Pasteur pipette was used as a pick to place each F2 adult individually into 5 μL of lysis buffer in a 96-well plate. The embryos on the worm plate could later be used as a source of DNA for additional mapping experiments.
Lysed DNA from each well was pin replicated into 96-well PCR plates containing complete PCR cocktail minus template. Each plate included the primer set for a single SNP. Typically we use plates representing four adjacent SNPs from the section of a particular chromosome that had shown linkage in the chromosome mapping experiments. For each plate, PCR, digestion and gel electrophoresis were performed as for chromosome mapping.
To validate the technique for mapping a mutation to an interval, we once again used the previously mapped gene dpy-5. We crossed Hawaiian males into the strain EG1000 as described above, and allowed the heterozygous progeny to self-fertilize. From the F2 progeny, we singled 48 Dpy animals onto individual plates. When these plates had starved, we washed each population of progeny into a well containing lysis buffer in a 96-well plate. The lysed DNA was assayed using three SNP primer pairs from the center of LGI (Figure 3B). Since each well contained progeny of a single F2 animal, it was possible to determine whether that animal was homozygous Bristol, homozygous Hawaiian, or heterozygous Bristol/Hawaiian at each SNP. From these data we could identify dpy-5-containing Bristol chromosomes that have recombined with Hawaiian DNA to the left of dpy-5 (recombinant types 'a' and 'b', Figure 3B and 3C), and to the right of dpy-5 (recombinant type 'c', Figure 3B and 3C). Keep in mind that each worm contains two dpy-5 containing chromosomes, but at regions near the mutation, usually only one is recombinant as illustrated in Figure 3C. We found, as expected, that our map data placed dpy-5 between -1 and 5 on LGI at approximately 0.3, very close to the known map position of 0.0 for dpy-5 (Figure 3C). Interestingly, one of the 48 Dpy F2 animals showed no linkage to LGI (see * in Figure 3B). The plate of worms that had been used to generate that lysate was chunked onto a new plate to verify the Dpy phenotype. Surprisingly, they were Dpy, but less so than dpy-5 homozygotes. We have observed that this Dpy phenotype segregates at a low frequency from several unrelated crosses between Hawaiian CB4856 and Bristol N2. Indeed, the plate segregated Rol-6 animals that were also Dpy. This confirms that the phenotype is not due to a dpy-5 mutation, since dpy-5 is epistatic to rol-6, but apparently the synthetic Dpy phenotype is not.
Mapping unknown mutations
To illustrate the utility of these methods we mapped a suppressor mutation of a behavioral phenotype. The map data demonstrate that these methods were able to map the original uncoordinated mutation and two other loci that synthetically suppress it in a single experiment. This strain, KY5029, was isolated in a screen for suppressors of unc-31(e928) (gift of Liakot Khan and Kouichi Iwasaki). unc-31 encodes the C. elegans homolog of CAPS, a protein required for dense-core vesicle release [7]. unc-31(e928) mutants are very inactive – almost paralyzed – on food. The suppressor strain KY5029 moves well, in fact it is slightly hyperactive. The strain KY5029 was crossed to Hawaiian males. Heterozygous F1 progeny were singled to plates. 85 Unc-31 animals were singled from among the F2 progeny. Most of these plates of Unc-31 animals segregated active suppressed animals, demonstrating that the suppressors are recessive. Two relatively active animals from each of the 85 plates were singled. From among the 170 plates, 30 plates were found to have the hyperactive phenotype of KY5029. Animals from these 30 plates were combined and used for chromosome mapping. From the chromosome mapping experiment we found that the suppressed unc-31 phenotype of KY5029 animals was linked to three genetic regions, on chromosomes I, II, and IV (data not shown). The region on chromosome IV contains unc-31, and linkage to IV is expected since Unc-31 animals were selected from among the F2 generation. Thus, the regions on chromosomes I and II must contain mutations suppressing the Unc-31 phenotype.
We found that the individual suppressor loci are weak unc-31 suppressors on their own. From the unc-31(e928) plates we singled animals that were less strongly suppressed; specifically, they were slightly sluggish rather than hyperactive. Linkage to the Bristol N2 genotype in these animals was observed on chromosomes I and IV, while linkage to the Hawaiian genotype was observed on chromosome II (data not shown). Thus, chromosome I contains a novel suppressor of unc-31 that partially suppresses the unc-31(e928) phenotype. Chromosome II contains a mutation that, in combination with the mutation on chromosome I, suppresses the unc-31(e928) phenotype to produce hyperactive worms. Further analysis determined that the suppressor on II could also partially suppress unc-31(e928) on its own. To simplify future mapping experiments we mapped the independent suppressing activities of the mutations on chromosome I and II in the unc-31(e928) background. Both suppressors were then fine mapped using primer sets on I and II independently to the genetic intervals depicted in Figure 5A and 5B. Together, these data suggest that KY5029 contains two mutant loci in addition to unc-31(e928). These complex interactions were deciphered with a minimum of time, effort, and confusion.
Figure 5 SNP mapping two synthetic unc-31 suppressor mutations. A, EG5296 ox300; unc-31(e928) interval mapping data. B, EG5297 dpy-5(e61); ox305; unc-31(e928) interval mapping data. The two suppressor mutations were separated from KY5029 and the suppressing activities were separately mapped to chromosome I and chromosome II, respectively. Each row illustrates the results from a single recombinant animal, and each colored box represents the genotype of a recombinant at the indicated SNP. Bristol is represented by blue, heterozygotes by purple, Hawaiian by red, and PCR failures by gray. F12B6 is wormbase allele snp_F12B6[1], F28H1 is wormbase allele snp_F28H1[1], T10D4 is a polymorphism identified by the St. Louis SNP consortium[5] (see Methods).
Conclusion
In summary, these methods comprise an accurate and fast technique for mapping that has advantages over both traditional mapping experiments and over other SNP mapping approaches that have been previously described. Compared to traditional mapping, this technique offers standardized, efficient mapping to small intervals for large numbers of mutations, such as might result from a genetic screen. In addition, it allows the mapping of subtle phenotypes (such as behavior) and complex genotypes (such as suppressor or synthetic mutations).
Compared to other SNP mapping methods, the technique described here occupies a comfortable midpoint between the simple but less efficient method described by Wicks et al. [1], and the high-throughput but complex techniques published by Swan et al. [3] and Zipperlen et al. [4]. Although the SNP detection technique described by Wicks is simple and inexpensive, it requires setting up a number of individual PCR reactions and matching the correct PCR product with the correct restriction enzyme. In the methods described here, once the primers (in chromosome mapping) or templates (in interval mapping) are arrayed into a plate, reaction components are accurately dispensed automatically and repeatably. Errors in matching restriction enzymes and buffers to primer sets have been eliminated by using DraI for all reactions. Further, we have found that assaying 8 SNPs on each chromosome means that every mutation is linked to multiple SNPs, giving a high level of redundancy. In fact, because of its cost and accuracy advantages, our technique has been successfully applied in an undergraduate teaching lab setting (M. Peters, personal communication).
The fluorescent polarimetry technique described by Swan et al., and the indel detection technique described by Zipperlen et al., enable high throughput SNP detection in C. elegans. However, these techniques require specialized equipment (a fluorescence polarimeter or capillary sequencer) that are not accessible to every laboratory and that require significant operator knowledge. Compared to those SNP mapping approaches, the technique described here is cheaper and more accessible, since it relies on methods that most labs already use.
In this paper, we present a happy medium between previous approaches to SNP mapping in worms. We build upon the simple, inexpensive, accessible and robust restriction digestion SNP detection technique of Wicks. However our primer sets, equipment and techniques substantially reduce user effort relative to the Wicks method, and so provide the efficiency and low error rate of the Swan or Zipperlen technologies.
Methods
Chromosome mapping
Hawaiian CB4856 males were crossed into EG1000 dpy-5(e61) I; rol-6(e187) II; lon-1(e1820) III. Fifty Dpy animals and fifty non-Dpy animals from among the self-progeny of EG1000/CB4856 heterozygote hermaphrodites were picked into separate tubes, each containing 20 μL single-worm lysis buffer (50 mM KCl, 10 mM Tris pH 8.3, 2.5 mM MgCl2, 0.45% IGEPAL CA-630, 0.45% Tween 20, 0.01% (w/v) gelatin, 60 ug/ml proteinase K). A further 96 Dpy animals were picked to individual plates for use in interval mapping (see below). They were lysed by freezing at -80°C followed by incubation at 65°C 1 hour and proteinase was inactivated by incubation at 95°C 15 minutes. The Dpy lysate DNA templates were then added to a PCR master mix containing 424 μL water, 52 μL 10X PCR buffer (10X: 22.5 mM MgCl2, 500 mM Tris-HCl, 140 mM (NH4)2SO4, pH 9.2 at 25°C), 10.4 μL 10 mM dNTPs, and 3.12 μL Taq (5 units/μl). A similar mix was made with the 50 non-Mutant animals. 9.8 μL of the mutant mix or the non-mutant mix was aliquoted into alternate rows of a 96-well PCR plate (Figure 1A). Primer pairs were arrayed into a microtiter plate at 10 μM each primer, so that neighboring rows contain duplicate pairs, and pin-replicated into the master mix. PCR reactions were done using the cycling conditions: 2' 94°C, 35 cycles of (15" 94°C, 45" 60°C, 1' 72°C), 5' 72°C. After amplification, PCR products were digested in the plate with the restriction enzyme DraI in a final volume of 16 μL (10 μL PCR product, 4.15 μL H2O 1.6 μL 10X DraI buffer (New England Biolabs), 0.25 μL DraI (10 units/μL, New England Biolabs)). This was accomplished by adding 6 μL of the enzyme plus enzyme buffer mix to each well using a multi-channel pipette followed by brief centrifugation in a Sorval RT6000D centrifuge with an H1000B rotor. Digestion reactions were incubated at 37°C at least 4 hours. Samples were then loaded onto a 2.5% agarose gel using an 8-channel pipette. The resulting gel displays all 48 SNP markers, from left to right and from chromosome I to X. Each Mutant SNP is next to its non-Mutant control, so that the whole genome can be quickly scanned for linkage.
Interval mapping
PCR templates were generated by cloning mutant animals from among the self-progeny of EG1000/CB4856 F1 hermaphrodites (described above) onto individual seeded plates. After 5 days, self progeny were washed from each plate using water (>100 worms / plate) and placed in a single well of a 96-well plate. Worms were allowed to settle to the bottom of the wells for 15' at 4°C then excess water was pipetted off to leave 45 μl in each well. The plates were frozen and stored at -80°C. The plates were thawed and 15 μl of 4X lysis buffer (200 mM KCl, 40 mM Tris pH 8.3, 10 mM MgCl2, 1.8% IGEPAL CA-630, 1.8% Tween 20, 0.04% (w/v) gelatin, 240 ug/ml proteinase K) was then added to each well to give 1X lysis buffer. The plates were covered with sealing tape and briefly vortexed to break up the worm pellet. The worms were lysed by incubation at 65°C 1 hour and 95°C 15 minutes. These PCR templates were stored frozen at -80°C and thawed prior to each use. For each PCR, each well of the 96-well plate received 9.8 μL of a PCR mix containing 8.5 μL water, 1 μL 10X buffer, 0.2 μL 10 mM dNTP, 0.02 μL each primer (100 μM), and 0.06 μL Taq (5 units/μl). Templates were then pin-replicated from the lysis plate. PCR conditions and DraI digests were the same as in chromosome mapping.
unc-31(e928) suppression mapping
We cloned 85 animals with an unc-31(e928)-like phenotype from the self progeny of KY5029/CB4586 hermaphrodites. unc-31(e928) animals are lethargic and uncoordinated; however, most animals exhibit periodic moments of coordinated movement making it difficult to distinguish between plates with no suppressed progeny and plates with weakly suppressed progeny. Therefore, two animals exhibiting coordinated movement were cloned from each plate. We scored the progeny of these 170 animals and divided them into five classes: uncoordinated, sluggish yet coordinated, coordinated, hyperactive, and mixed. For the sluggish and hyperactive phenotypes, we collected animals for chromosome mapping by combining two animals from each plate. Chromosome mapping was performed as described above, suggesting the presence of two suppressors located on chromosomes I and II (data not shown). To confirm the mapping results, individual recombinants were assayed for SNPs on chromosomes I, II, and IV (data not shown). To simplify further interval mapping experiments these suppressors were crossed away form each other to generate two partially suppressed strains, EG5296 ox300; unc-31(e928) and EG5297 dpy-5(e61); ox305; unc-31(e928). For EG5297, we verified the loss of the ox300 chromosome by homozygosing a dpy-5(e61) marker. Interval mapping of ox300 was carried out by cloning 11 unc-31(e928)-like self progeny from EG5296/CB4586 hermaphrodites. From these self-progeny 35 sluggish yet coordinated animals were cloned and assayed at SNPs flanking the suppressor (Figure 5A). ox300 is located on Chromosome I between W03D8 and B0205. ox305 was mapped by cloning 8 unc-31(e928)-like self progeny from EG5297/CB4586 hermaphrodites. From these self-progeny 21 partially suppressed animals were cloned and assayed at SNPs flanking the suppressor (Figure 5B). The suppressor in ox305 is located on Chromosome II between R52 and T10D4. Four new SNPs that are not part of the chromosome mapping set were used. Wormbase allele snp_F12B6[1] was amplified with primers 5'-caggttggtttttggcaagt-3' 5'-tgattgaacatatccggcaa -3' and detected with MfeI. Wormbase allele snp_F28H1[1] was amplified with primers 5'-gcagtaggcaagagtcaggc-3' and 5'-tattgcacttggctcacagc -3' and was detected with HpyCH4V. pkP2135 was amplified with primers 5'-tttgcagatttccgatactgtg -3' and 5'-ttttgtcgtaagacctttggtg -3' and was detected with DraI. A polymorphism on T10D4, referenced at the web page [8] was amplified with primers 5'-gtacgcctcaaaaagtggag -3' and 5'-accacccaacacaatctctg -3' and was detected with MseI.
Abbreviations used
SNP: Single nucleotide polymorphism.
Authors' contributions
MWD MH conceived and designed the SNP methods, carried out and supervised the experiments and drafted the manuscript. MWD wrote the scripts that extracted and formatted the SNP data. TH and SO tested SNP primers PH carried out the unc-31 mapping experiments. EMJ helped coordinate the experiments and draft the manuscript. All authors read and approved the final manuscript.
Acknowledgements
We would like to thank the Caenorhabditis Genetics Center (CGC) for providing strains, S. Wicks for early advocation of SNP mapping in C. elegans, the St. Louis and Exelixis SNP discovery projects for shotgun sequencing of CB4856, L. Khan and K. Iwasaki for KY5029, other laboratories that provided feedback on early versions of our primer sets, and T. Harris for providing initial analysis of SNP sequence data.
==== Refs
Wicks SR Yeh RT Gish WR Waterston RH Plasterk RH Rapid gene mapping in Caenorhabditis elegans using a high density polymorphism map Nat Genet 2001 28 160 164 11381264 10.1038/88878
Jakubowski J Kornfeld K A local, high-density, single-nucleotide polymorphism map used to clone Caenorhabditis elegans cdf-1 Genetics 1999 153 743 752 10511554
Swan KA Curtis DE McKusick KB Voinov AV Mapa FA Cancilla MR High-throughput gene mapping in Caenorhabditis elegans Genome Res 2002 12 1100 1105 12097347
Zipperlen P Nairz K Rimann I Basler K Hafen E Hengartner M Hajnal A A universal method for automated gene mapping Genome Biol 2005 6 R19 15693948 10.1186/gb-2005-6-2-r19
C. elegans Single Nucleotide Polymorphism Data
Rozen S Skaletsky H Primer3 on the WWW for general users and for biologist programmers Methods Mol Biol 2000 132 365 386 10547847
Livingstone D Studies on the UNC-31 Gene of Caenorhabditis elegans PhD thesis 1991 University of Cambridge, Darwin
SNP Info
Wormbase
dbSNP Home
SNP Info
|
16156901
|
PMC1242227
|
CC BY
|
2021-01-04 16:32:46
|
no
|
BMC Genomics. 2005 Sep 12; 6:118
|
utf-8
|
BMC Genomics
| 2,005 |
10.1186/1471-2164-6-118
|
oa_comm
|
==== Front
BMC GeriatrBMC Geriatrics1471-2318BioMed Central London 1471-2318-5-111615014710.1186/1471-2318-5-11Study ProtocolDesign and pilot results of a single blind randomized controlled trial of systematic demand-led home visits by nurses to frail elderly persons in primary care [ISRCTN05358495] van Hout Hein PJ [email protected] Giel [email protected] Marwijk Harm WJ [email protected] Aaltje PD [email protected]'t Veer Petronella J [email protected] Willemijn [email protected] Wim AB [email protected] VU University medical center Amsterdam, Institute for Research in Extramural Medicine, Department of General Practice, Van der Boechorststraat 7, 1081 BT Amsterdam, The Netherlands2005 8 9 2005 5 11 11 31 5 2005 8 9 2005 Copyright © 2005 van Hout et al; licensee BioMed Central Ltd.2005van Hout et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The objective of this article is to describe the design of an evaluation of the cost-effectiveness of systematic home visits by nurses to frail elderly primary care patients. Pilot objectives were: 1. To determine the feasibility of postal multidimensional frailty screening instruments; 2. to identify the need for home visits to elderly.
Methods
Main study: The main study concerns a randomized controlled in primary care practices (PCP) with 18 months follow-up and blinded PCPs. Frail persons aged 75 years or older and living at home but neither terminally ill nor demented from 33 PCPs were eligible. Trained community nurses (1) visit patients at home and assess the care needs with the Resident Assessment Instrument-Home Care, a multidimensional computerized geriatric assessment instrument, enabling direct identification of problem areas; (2) determine the care priorities together with the patient; (3) design and execute interventions according to protocols; (4) and visit patients at least five times during a year in order to execute and monitor the care-plan. Controls receive usual care. Outcome measures are Quality of life, and Quality Adjusted Life Years; time to nursing home admission; mortality; hospital admissions; health care utilization.
Pilot 1: Three brief postal multidimensional screening measures to identify frail health among elderly persons were tested on percentage complete item response (selected after a literature search): 1) Vulnerable Elders Screen, 2) Strawbridge's frailty screen, and 3) COOP-WONCA charts.
Pilot 2: Three nurses visited elderly frail patients as identified by PCPs in a health center of 5400 patients and used an assessment protocol to identify psychosocial and medical problems. The needs and experiences of all participants were gathered by semi-structured interviews.
Discussion
The design holds several unique elements such as early identification of frail persons combined with case-management by nurses.
From two pilots we learned that of three potential postal frailty measures, the COOP-WONCA charts were completed best by elderly and that preventive home visits by nurses were positively evaluated to have potential for quality of care improvement.
==== Body
Background
Publishing the design of a study
Publishing the design and protocol of a study before results are available is important for several reasons. A published protocol allows easier comparison between what was originally intended and hypothesized and what was actually done, and it gives readers greater insight into the methodological quality of a study [1]. Furthermore, it has often been recognized that negative or adverse outcomes are less likely to be published [2]. Publishing the design of a study before its start announces the study will be undertaken, which encourages publication of the results and in any case informs researchers where they can find the data for inclusion in systematic reviews [1,1,2]. Thus, publishing a design article can prevent publication bias. In addition, publishing pilot results provides a better insight in the choices for particular instruments and interventions.
Primary care and elderly
In the Netherlands all people are registered in a primary care practice and Primary Care Physicians (PCPs) act as gatekeepers to specialist care, whereas for example in the USA most persons are not registered in a primary care practice [3]. There are few barriers to primary care facilities in the Netherlands and in the Netherlands about 86% of older people contact their PCP yearly [4]. Older persons in primary care are, therefore, a good representation of the total older population at risk. Primary care is confronted with increasing numbers of frail elderly because of the aging of the population, and their wish to live independently for as long as possible. Frailty poses a complex problem for primary care. Up to about 20% of the elderly, defined here as those 75 years of age and over are vulnerable for further deterioration of functional abilities and quality of life accompanied by a substantial increased risk of institutionalization [5]. This implies an exploding need for care.
Primary care is insufficiently equipped for a potential explosion of care needs. GPs are often unaware of the health status and functional limitations of their elderly patients [6,7]. Several studies reported a considerable amount of undetected morbidity both among consulting and non-consulting patients [8,9]. Moreover, PCPs, as the medically responsible person, do not regard themselves suited for systematic management and long-term monitoring for chronic diseases and disabilities associated with frail health [10].
Proactive detection of care needs in elderly but still competent persons who do not explicitly seek help is at odds with the prevailing reactive paradigm in primary care. However, as perhaps many frail elderly are unaware of the types of help available, there is a need for care experiments with transmural collaboration among health professionals, which might increase the quality of (primary) care for frail persons at home.
Earlier interventions
Systematic home visits to frail elderly by nurses can reduce mortality and nursing home admissions provided that a substantial number of home visits are paid and care plans are based on multidimensional assessments [11,12]. In addition, accumulating evidence shows that preventive home visits are mostly accompanied by a reduction of health care costs [13]. Both from patient (health gains) and societal (cost savings) perspective this is a desirable situation.
Frailty and preventive mechanisms
Frailty is the result of reduced ability to maintain a physiological and psychosocial equilibrium, thereby increasing the risk of functional disability, temporary or permanent loss of the ability to cope, morbidity, and mortality [14-16]. Frailty is strongly associated with aging [17,18]. The potential preventive mechanisms of home visits comprise early detection of worsening health conditions and modifiable risk factors, enabling concerted actions with responsible health professionals to optimize treatment, improve life style and increase support for family caregivers to persevere informal care.
Costs
Aging is costly. About one third of the health care expenditures in industrialized countries relate to persons 70 years or older [19]. Nursing homes, homes for the elderly and hospital beds are occupied mainly by elderly. Elderly are massive consumers of medication. Elderly consume most home care. When the number and portion of (frail) elderly increases the health care costs will explode. Among community dwelling elderly with usual care the median annual nursing home and hospital admission rate is 2.4% (range 0–40%) respectively 26% (range 5–56%). Among frail elderly median annual admission rates are 15% and 45% respectively [11,12].
Objectives
To describe the design of an evaluation of the cost-effectiveness of systematic home visits by nurses to frail elderly primary care patients. Pilot objectives were to determine the feasibility of postal multidimensional frailty screening instruments and to identify the need for home visits to frail elderly. This article describes the background, design and pilot results.
Methods
Pilot 1: Selecting frail patients by postal questionnaires (Table 2)
Table 2 First Pilot: Selecting frail patients
Background: Measuring frailty is subject to debate and various operational definitions were proposed [15]. For our purpose we sought a valid easy administrable self-report instrument.
Objective: To determine the feasibility of multidimensional frailty screening instruments that could be sent by mail.
Methods: After a literature search three multidimensional screening instruments were selected and tested in one general practice among all 75+ patients: 1) VES-13, 2) Strawbridge's frailty screen, and 3) COOP-WONCA charts. Feasibility was expressed in percentage complete item response [20–22]. Our goal was to identify the worst quarter. This point of departure was based on studies by Fried and Rockwood who reported between 20–30% of 75+ people to be frail according to their measures [14,17].
Results: Of 116 patients 85 (81%) agreed to participate and 69 actually returned the questionnaire. The complete item response on the COOP-WONCA, Strawbridge screen, and VES-13 were 87%, 60% and 56% respectively. In order to identify a quarter of persons with the worst health on the COOP-WONCA, all persons were selected who scored in the worst quartile of at least two of the six charts (overall health ≥4; physical fitness ≥5; changes in health ≥4; daily activities ≥4; Feelings ≥3; social activities ≥3). This resulted in 23 persons who were further assessed at home by the RAI-HC. 90% had at least one chronic disease, two thirds had at least one ADL limitation, 60% had depressive symptoms (CESD>16) and 30% had cognitive impairment (MMSE<24) [37].
Conclusion: The COOP-WONCA was the most feasible screener. Our selection rule identified a frail group. The geriatric assessment identified new potentially treatable problems.
COOP-WONCA = COOP functional health assessment charts – World Organization of Family Doctors
VES-13 = Vulnerable Elders Survey-13
RAI-HC = Resident Assessment Instrument – Home Care version
MMSE = Mini Mental Screen Examination
After a literature search three multidimensional screening instruments were selected and tested in one primary care practice among all 118 75+ patients: 1) VES-13, 2) Strawbridge's frailty screen, and 3) COOP-WONCA charts [20-22]. Feasibility was expressed in percentage complete item response. Our goal was to identify the worst quarter.
Pilot 2: Exploring the potential for quality of care improvement of preventive home visits among elderly persons (Table 3)
Table 3 Second Pilot : Exploring the potential for quality of care improvement of preventive home visits among elderly persons.
Objective: To identify the need and possible benefit of home visits for frail patients, PCPs and nurses.
Method: The setting was a health center of 5400 patients with 3 PCPs and a practice nurse. Possible frailty was determined by the PCPs among their 75+ patients in the following cases: beginning dementia, active carcinoma, two or more medications for organ indication, treatments by two or more medical specialists, being 85+ and not contacted the PCP over the last three years, uncertainty regarding the ability to manage oneself, and all other persons the PCP felt it necessary to pay attention to. The nurses visited the patients and used an elaborate geriatric assessment protocol to identify psychosocial and medical problems. The nurses and the PCPs designed a care plan. The experiences of all participants were gathered by semi-structured interviews.
Results: The participants (PCPs, nurses, patients) evaluated this approach positively. The PCPs gained better insight in medical and care situation of their elderly patients and experienced less work pressure. The nurses experienced better quality of care. The patients felt safer and more independent. The PCP also selected a number of healthy persons.
Conclusion: Home visits by nurses were regarded by all to have potential for quality of care improvement. Point of concern was the inadequate selection of frail patients by the PCPs. Also, the assessment protocol used by the nurses provided no triggers on when actions should follow.
PCP = Primary Care Physician
The setting was a health center of 5400 patients with 3 PCPs and a practice nurse. Possible frailty was determined by the PCPs among their 75+ patients in the following cases: beginning dementia, active carcinoma, two or more medications for organ indication, treatments by two or more medical specialists, being 85+ and not contacted the PCP over the last three years, uncertainty regarding the ability to manage oneself, and all other persons the PCP felt it necessary to pay attention to. The nurses visited the patients and used an elaborate geriatric assessment protocol to identify psychosocial and medical problems. The nurses and the PCPs designed a care plan. The experiences of all participants were gathered by semi-structured interviews.
Main study
Design
A randomized controlled trial in 33 primary care practices (55 primary care physicians) among frail 75+ patients at home who responded to a Health Screener, with 18 months follow-up. Frail persons living at the same address were randomized as one unit. PCPs are held blind for the group assignment. Block-randomization ensured equal numbers of intervention and usual care patients per practice. Random number tables were used by and independent person for randomization. The ethical committee of the VU medical center approved the study.
Study population main study
The PCPs provided the names and addresses of all their listed patients of 75 years or older and living at home. All persons received a health survey including the COOP-WONCA charts in order to identify the 20–25% elderly with the frailest functional health. The cut-offs per chart were based on a combination of reference data and our pilot data [22]. Inclusion and exclusion criteria are summarized in Table 1.
Table 1 Inclusion and exclusion criteria
Inclusion: • Age 75 years and over and listed as general practice patient
• Living at home
• Frail: Self reported Health score in the worst quartile of at least two of six COOP-WONCA charts (scoring range: 1, excellent, to 5 very bad): Overall health ≥4; Physical fitness ≥5; Changes in health ≥4; Daily activities ≥4; Mental health ≥3; Social activities ≥3
Exclusion: • Terminally ill as determined by PCPs
• Persons with dementia symptoms according to MMSE or 7-minute screen
• Living in residential homes.
• Participating in other research projects
Intervention(s)
The scores of all persons who filled out the health survey and positively responded to the care offer were analyzed. The intervention consisted of 7 elements; (1) All frail persons and randomized to the intervention were contacted by one of eight trained nurses. In the first visit the nurses assess the health status and care needs by the Resident Assessment Inventory Home Care version (RAI-HC), a structured and computerized multidimensional geriatric instrument that enables direct and validated identification of problem areas [23,24]. The RAI-HC holds about 120 items and 30 domains of health and service needs (Table 5). It takes between 45 to 60 minutes to complete; (2) In our intervention the list of problems is discussed with the patient to determine whether additional care is needed.
Table 5 Case example of assessment by a nurse with the RAI-HC: triggered health risks
Client Assessed Problem Observed Action undertaken earlier? Relevant action now? Immediate action? Action later?
1. ADL / Revalidation potential X X
2. IADL / more formal care
3. Health promotion
4. Risk intramural admission
5. Communication impairment
6. Visual impairment
7. Alcohol abuse
8. Cognition
9. Behavior
10. Depression and Anxiety
11. Abuse
12. Social functioning
13. Heart and lungs X X
14. Dehydration X X
15. Falls
16. Nutrition
17. Dental health
18. Pain
19. Bedsores X X
20. Skin and food problems X X
21. Compliance
22. Vulnerable support system X
23. Medication management
24. Palliative care
25. Preventive health X X
26. Psychofarmaca use X X
27. Reduced service package
28. Environment
29. Feces incontinence
30. Urinal incontinence catheter X
Therefore the nurse and the patient make a hierarchy of the problems; (3) The nurses design and execute individual suited care-plans that comply with patient priorities; (4) The nurses are case-managers and offer to visit the patients at least 5 times in a year in order to execute and monitor the care-plan, to evaluate whether the care-needs have changed and adapt the care/plan when needed; (5) The nurses also meet the PCPs on a regular basis to discus the care plans and to assure that medical actions are carried out by the PCPs; (6) To assure the quality of care, the nurses receive regularly educational updates and organize monthly meetings to discuss problematic cases. Two staff members supervise them. A national Dutch guideline on home care nursing of frail elderly patients was available [25]. This guideline was used to protocolize nurse interventions whenever possible; (7) The care plan is left at the patients' house to enable other visiting health professionals to take notice of and report on the care plan.
Outcomes and measurements
Table 4 provides an overview of all outcomes and measurements in the study.
Table 4 Measurement scheme
Health screener Instrument T-1 T0 T1 T2
6 months 18 months
Functional Health status COOP-WONCA X X X
ADL & IADL GARS X X X
Cognitive decline IQCODE self report X
Depressive symptoms CES-D X X X
Chronic diseases Chronic diseases list X X
Mobility and Falls Questionnaire X X X
Body Mass Index Questionnaire X
Weight change Questionnaire X
Demographics Questionnaire X
Behavioral problems Questionnaire X
Incontinence Questionnaire X X X
Main Outcomes
a. Health related Quality of life SF36 + EQ5D X X X
b. Hospital admissions Patient + hospital database X X X
c. (Days until) Institutionalization PCP + nursing homes X
d. (Days until) Mortality Relatives + PCP X
e. Health resource utilization Self report + PCP + hospital + pharmacy databases X X X
PCP = Primary Care Physician
GARS = Groningen Activity Restriction Scale
CES-D = Center for Epidemiological Studies – Depression Scale
IQCODE = Informant Questionnaire on Cognitive Decline in the Elderly
SF36 = Short Form 36 item version
EQ5D = EuroQuality of life
X = measurement
T-1 = pre randomization health screening
T0 = Measurement immediately after randomization
T1, T2 = Follow-up measurements
Outcomes are:
1. Health related quality of life as measured with the Short Form 36 (SF-36), and Quality Adjusted Life Years by health utilities based on Euroqol (EQ-5D) [26,27];
3. (Days until) institutionalization: Hospital stay, placement in nursing home or home for the elderly are surveyed and crosschecked at institutes;
4. (Days until) mortality as checked with the PCPs;
5. Direct costs as measured by patient questionnaires with three-monthly recall periods. These self-report data are supplemented by data from the centralized regional pharmacy database (medication use), regional hospital check, and nursing home checks. In case patients are not able to fill out the forms themselves a close relative will be approached (Table 4).
Sample size calculation
For an anticipated Health related Quality of Life benefit on at least two SF-36 domains with minimal relevant effect size Cohen's D = 0.5, 64 persons per group are required with a two sided alpha of 0.05 and 80% probability. Anticipating on an annual attrition of 20% (mortality, inability to respond, unwilling) 75 persons per group will be needed.
In a trial of 650 persons a reduction of 10% in hospital admission, institutionalization and mortality can be detected with a two-sided alpha of 0.05 and 80% probability (320 persons needed per group). Effect estimates are based on previous meta-analyses [11,12].
Data-analysis
According to the 'intention-to-treat' principle differences between intervention and usual care patients on mortality, hospitalization and nursing home placement (dichotomous outcomes) are tested by both chi-square tests and logistic regression analysis. Differences in time until these events will be analyzed with Cox-proportional hazard modeling. For quality of life (continuous outcome: SF-36, EQ-5D) General Linear Models (GLM), a technique for repeated measures is used to analyze group differences. Possible baseline differences in the outcome measures will be accounted for in GLM. Additional subgroup-analyses will be performed on types of recommendations in the care-plans.
Potential confounding and effect-modification is checked for sociodemographic characteristics, number and type of chronic disease, (I)ADL functioning (GARS), cognitive decline (IQCODE), mood (CES-D), behavioral problems (incontinence, sleep, agitation en aggression), medication use (centralized pharmacy data base) (Table 4) [28-30].
Quality of the data was assured by independent double checks of all forms. Also our institute employs a quality assurance policy. In this respect guidelines on all aspects of research were issued and all projects are subject to audits.
Economic evaluation
The economic evaluation will be performed alongside the randomized trial from a societal perspective. Data on resource use are collected in several ways: self report questionnaires, hospital and nursing home registration, and community pharmacy records. Only direct healthcare costs will be considered such as costs of consultations of the general practitioner, nursing home physician, medical specialist, hospitalizations, and medical department of the nursing home, and use of medication and medical aids. Medication data are retrieved from the centralized pharmacy files in the research region. If available, Dutch guideline prices are used to value resource use [31,32]. Otherwise, tariffs are used. Medication costs are valued using prices of the Royal Dutch Society for Pharmacy [33]. Contacts with GPs and referrals will be checked as well in GPs' patient information files.
Cost analysis
To compare costs between the two groups, confidence intervals for the differences in mean costs are calculated using bias-corrected and accelerated bootstrapping with 5000 replications. [34] For the cost-effectiveness analysis the difference in total costs between the intervention and usual care group are compared with the difference over 18 months in improvement of quality of life, reduced institutionalization, hospitalization and mortality. In addition, a cost-utility analysis will be done to assess the incremental costs per Quality Adjusted Life Years (QALY). Uncertainty around the cost-effectiveness and cost-utility ratios is calculated using the bias-corrected percentile method (5000 replications) and presented in a cost-effectiveness plane [35].
Patient outcome analysis
QALYs are calculated by multiplying the utility based on EQ-5D scores with the amount of time a patient spent in this particular health state [36]. Transitions between health states are linearly interpolated.
Recruitment
The recruitment phase yielded a total of 33 PCP practices that were willing to participate. Inclusion started in spring 2003 and lasted until summer 2004. Figure 1 provides an overview of the recruitment and randomization. The health questionnaire was mailed to 4823 patients. Of the 2949 (61%) responders, 658 frail patients were detected and randomized.
Figure 1 Flow chart PIKO.
Non-response
Females more often non-responded (41.3% versus 35.8%, X2 = 13.4 p < 0.001), as did slightly older persons (83.5 versus 81.7 t = 13.8 p < 0.001).
Intervention
Eight nurses were trained in the use of the RAI-HC on laptops. The nurses were relatively unskilled in computer use so specific training was provided. Table 5 shows an example of a patient's problems list. Based on the detected problems the nurses design care plans.
Discussion
In this paper we describe the design of a randomized cost-effectiveness trial of effect of preventive home visits by nurses to frail elderly primary care patients as well as the results of two pilot studies to determine the need and the feasibility instruments to select the frailest portion among elderly. The main study main study holds unique elements. The intervention concerns pro-active care that should guarantee timely detection of patients with frailer health, followed by structured nurse-led care focusing on patients. It focuses on the client as well as on the system around the client. The nurses use the Resident Assessment Instrument; a comprehensive geriatric assessment protocol that, by computerization, triggers health risks and, hereby, guides care planning.
The main study achieved a substantial response on the postal COOP-WONCA screening. In our pilot study, home visits by nurses were regarded to have potential for quality of care improvement. Point of concern was the inadequate selection of frail patients by the PCPs. Also, the assessment protocol used by the nurses provided no triggers on when actions should follow. In a second pilot study, the COOP-WONCA emerged as the most feasible postal screener compared to vulnerable elders survey (VES-13) and the Frailty screening list of Strawbridge. Our selection rule identified a frail group.
Limitations to the generalizability of our future findings are firstly the non-response to the mailed health survey. A number of persons (n = 100) responded but did not want to participate because of their good health. We remain uncertain about other non-responders being frailer and perhaps in greater need of nurse support. In future projects alternative means for health screening may be tested in older PCP patients such as contacting consulting patients. Secondly, our RAI assessment demanded a structured approach of computer skilled persons. The computer skills of the nurses were rather limited which led to extensive additional training. Some nurses had difficulties with the computerized assessments and disliked the structured format in which the assessment took place. Thirdly, we remain uncertain about the best frailty measure to identify persons that can benefit most from preventive actions. We selected the COOP-WONCA because of its broad health definition and very good feasibility. Last, whether PCPs remain blind for group assignment during he follow up remains to be seen. It is possible that the regular contacts with the nurses will reveal some of the assignments.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
HPJvH, GN and HWJvM designed the study. HPJvH drafted the article and all authors contributed to the final concept.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
Funding was received from the Foundation to improve Primary care medicine of the VU University medical center and the Netherlands Organization for Health Research and Development (ZONmw). We like to thank all participating PCPs, nurses, patients and especially Home Care Organization 'De Omring' for their cooperation to evaluate this project.
==== Refs
Godlee F Publishing study protocols: making them visible will improve registration, reporting and recruitment BMC News and Views 2001 2 4
Thornton A Lee P Publication bias in meta-analysis: its causes and consequences J Clin Epidemiol 2000 53 207 216 10729693 10.1016/S0895-4356(99)00161-4
van Londen J Dutch health care – a study on purple Lancet 1996 347 1239
Cardol M van Dijk L De Jong JD De Bakker DH Westert GP 2nd Dutch National Study on morbidity and performance in primary care: what does the family phyhsician do?)[Tweede Nationale Studie naar Ziekten en Verrichtingen in de huisartspraktijk: Huisartsenzorg, wat doet de huisarts] Utrecht, The Netherlands, NIVEL 2004 2nd Dutch National Study on morbidity and performance in primary care.
Rockwood K Stadnyk K MacKnight C McDowell I Hebert R Hogan DB A brief clinical instrument to classify frailty in elderly people Lancet 1999 353 205 206 9923878 10.1016/S0140-6736(98)04402-X
Stalenhoef PA Kuijnen LGJ Schonck RSM KJ van de Horst FG Crebolder HFJM Homevisits by community nurse can contribute to the family physician's knowledge of elderly patients [Huisbezoek door de wijkverpleegkundige kan bijdragen aan de kennis van de huisarts over oudere patiënten] Ned Tijdschr Geneesk 1996 140 2464 2467
Tremellen J Assessment of patients aged over 75 years in general practice BMJ 1993 305 621 624 1393077
Eekhof J De Bock G Schaapveld K Springer M Effects of screening for disorders among the elderly: an intervention study in general practice Fam Pract 2000 17 329 333 10934182 10.1093/fampra/17.4.329
Lagaay AM van der Meij JC Hijmans W Validation of medical history taking as part of a population based survey in subjects aged 85 and over BMJ 1992 304 1091 1092 1586823
Iliffe S Manthorpe J Eden A Sooner or later? Issues in the early diagnosis of dementia in general practice: a qualitative study J Gerontol A Biol Sci Med Sci 2003 20 376 381
Elkan R Kendrick D Dewey M Hewitt M Robinson J BM Effectiveness of home based support for older people: systematic review and meta-analysis BMJ 2001 323 719 725 11576978 10.1136/bmj.323.7315.719
Stuck AE Egger M Hammer A Minder CE Beck JC Home visits to prevent nursing home admission and functional decline in elderly people JAMA 2002 287 1022 1028 11866651 10.1001/jama.287.8.1022
Johri M Beland F Bergman H International experiments in integrated care for the elderly: a synthesis of the evidence Int J Geriatr Psychiatry 2003 18 222 235 12642892 10.1002/gps.819
Fried LP Tangen CM Walston J Newman AB Hirsch C Gottdiener J Frailty in older adults: evidence for a phenotype J Gerontol A Biol Sci Med Sci 2001 56 M146 M156 11253156
Hogan DB MacKnight C Bergman H Steering Committee CIoFaA Models, definitions, and criteria of frailty Aging Clin Exp Res 2003 15 1 29 14580013
Ligthart G Broese van Groenou MI, Deeg DJG, Knipscheer CPM, Ligthart GJ Frailty a clinically feasible concept? [Kwetsbaarheid een klinische bruikbaar begrip?] VU-visions on aging [VU-visies op veroudering] 1999 Amsterdam: VU University 25 28
Rockwood K Howlett SE MacKnight C Beattie BL Bergman H Hebert R Prevalence, attributes, and outcomes of fitness and frailty in community-dwelling older adults: report from the Canadian study of health and aging J Gerontol A Biol Sci Med Sci 2004 59 1310 1317 15699531
Mitnitski AB Graham JE Mogilner AJ Rockwood K Frailty, fitness and late-life mortality in relation to chronological and biological age BMC Geriatr 2002 2 1 11897015 10.1186/1471-2318-2-1
Polder JJ Achterberg PW Cost of illness in the Netherlands 2004 Bilthoven, The Netherlands, RIVM, National Institute for Health and Environment.
Saliba D Elliott M Rubinstein LZ Solomon DH The vulnerable elders survey: a tool for identifying vulnerable older people in the community JAGS 2001 49 1691 1699 10.1046/j.1532-5415.2001.49281.x
Strawbridge WJ Sheme SJ Balfour JL Higby HR Kaplan GA Antecedents of frailty over three decades in an older cohort Sci Soc Sci 1998 53 S9 16
Van Weel C Konig-Zahn C Touw-Otten F van Duijn N Meyboom-de Jong B Measuring functional health with the COOP/WONCA charts, a manual 1995 Groningen, Northern Centre for Health Care Research, University Groningen
InterRai Overview Committee Minimal Data Set, Resident Assessment Interview, Home Care version 2.0 2005
Landi F Tua E Onder G Carrara B Sgadari A Rinaldi C Minimum data set for home care: a valid instrument to assess frail older people living in the community Med Care 2000 38 1184 1190 11186297 10.1097/00005650-200012000-00005
KITZZ Community Nurse Guideline for dementia [Wijkverpleegkundige standaard Dementie] 2005 Groningen, KITTZ, KwaliteitsInstituut voor Toegepaste ThuisZorgvernieuwing
Kind P The EuroQol instrument: an index of health related quality of life Quality of life and pharmacoeconomics in clinical trials 1996 Philadelphia: Lippincott-Raven Publishers 191 201
McHorney CA Ware JE JrRaczek AE The MOS 36-Item Short-Form Health Survey (SF-36): II. Psychometric and clinical tests of validity in measuring physical and mental health constructs Med Care 1993 31 247 263 8450681
Beekman AT Deeg DJ Van Limbeek J Braam AW De Vries MZ van Tilburg W Criterion validity of the Center for Epidemiologic Studies Depression scale (CES-D): results from a community-based sample of older subjects in The Netherlands Psychol Med 1997 27 231 235 9122304 10.1017/S0033291796003510
de Jonghe JF Schmand B Ooms ME Ribbe MW Abbreviated form of the Informant Questionnaire on cognitive decline in the elderly Tijdschr Gerontol Geriatr 1997 28 224 229 9526793
Kempen GI Miedema I Ormel J Molenaar W The assessment of disability with the Groningen Activity Restriction Scale. Conceptual framework and psychometric properties Soc Sci Med 1996 43 1601 1610 8961404 10.1016/S0277-9536(96)00057-3
Oostenbrink JB Koopmanschap MA Rutten FF Standardisation of costs: the Dutch Manual for Costing in economic evaluations Pharmacoeconomics 2002 20 454
Oostenbrink JB Bouwmans CAM Koopmanschap MA Rutten FFH Manual for costing research 2004 Amstelveen, The Netherlands: Health Care Insurance Board
Z-index: G-Standaard for medication Den Haag, The Netherlands 2002
Efron B Tibshirani RJ An introduction to the bootstrap 1993 New York/London: Chapman & Hall
Chaudhary MA Stearns SC Estimating confidence intervals for cost-effectiveness ratios: an example from a randomized trial Statistics in Medicine 1996 15 1447 1458 8841654 10.1002/(SICI)1097-0258(19960715)15:13<1447::AID-SIM267>3.3.CO;2-M
Dolan P Modeling valuations for EuroQol health states Med Care 1997 35 1095 1108 9366889 10.1097/00005650-199711000-00002
Folstein MF Folstein SE McHugh PR A practical method for grading the cognitive state of patients for the clinician J Psychiatr Res 1975 12 189 198 1202204 10.1016/0022-3956(75)90026-6
|
16150147
|
PMC1242228
|
CC BY
|
2021-01-04 16:30:32
|
no
|
BMC Geriatr. 2005 Sep 8; 5:11
|
utf-8
|
BMC Geriatr
| 2,005 |
10.1186/1471-2318-5-11
|
oa_comm
|
==== Front
BMC Int Health Hum RightsBMC International Health and Human Rights1472-698XBioMed Central London 1472-698X-5-61612020810.1186/1472-698X-5-6Research ArticleMedia reporting of tenofovir trials in Cambodia and Cameroon Mills Edward [email protected] Beth [email protected] Ping [email protected] Elaine [email protected] Kumanan [email protected] Sonal [email protected] Dept. of Clinical Epidemiology & Biostatistics, McMaster University, Hamilton, Canada2 Centre for International Human Rights Law, University of Oxford, Oxford, UK3 Department of Epidemiology, London School of Hygiene and Tropical Medicine, London, UK4 Development Studies Institute, London School of Economics, London, UK5 Departments of Medicine, Health Policy, Management and Evaluation, University of Toronto, Toronto, Canada6 Bloomberg School of Public Health, Johns Hopkins University, Baltimore, USA2005 24 8 2005 5 6 6 19 5 2005 24 8 2005 Copyright © 2005 Mills et al; licensee BioMed Central Ltd.2005Mills et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Two planned trials of pre-exposure prophylaxis tenofovir in Cambodia and Cameroon to prevent HIV infection in high-risk populations were closed due to activist pressure on host country governments. The international news media contributed substantially as the primary source of knowledge transfer regarding the trials. We aimed to characterize the nature of reporting, specifically focusing on the issues identified by media reports regarding each trial.
Methods
With the aid of an information specialist, we searched 3 electronic media databases, 5 electronic medical databases and extensively searched the Internet. In addition we contacted stakeholder groups. We included media reports addressing the trial closures, the reasons for the trial closures, and who was interviewed. We extracted data using content analysis independently, in duplicate.
Results
We included 24 reports on the Cambodian trial closure and 13 reports on the Cameroon trial closure. One academic news account incorrectly reported that it was an HIV vaccine trial that closed early. The primary reasons cited for the Cambodian trial closure were: a lack of medical insurance for trial related injuries (71%); human rights considerations (71%); study protocol concerns (46%); general suspicions regarding trial location (37%) and inadequate prevention counseling (29%). The primary reasons cited for the Cameroon trial closure were: inadequate access to care for seroconverters (69%); participants not sufficiently informed of risks (69%); inadequate number of staff (46%); participants being exploited (46%) and an unethical study design (38%). Only 3/23 (13%) reports acknowledged interviewing research personnel regarding the Cambodian trial, while 4/13 (30.8%) reports interviewed researchers involved in the Cameroon trial.
Conclusion
Our review indicates that the issues addressed and validity of the media reports of these trials is highly variable. Given the potential impact of the media in formulation of health policy related to HIV, efforts are needed to effectively engage the media during periods of controversy in the HIV/AIDS epidemic.
==== Body
Background
With almost 5 million new HIV infections and 3 million AIDS deaths occurring every year worldwide – almost 600 every hour -, the development of safe, effective, and accessible prevention methods has become one of the most urgent global public health needs[1]. One innovative method, aimed at reducing HIV infection through chemoprophylaxis, is the use of the antiretroviral drug tenofovir (Viread), a nucleotide reverse transcriptase inhibitor produced by Gilead Sciences Inc. Tenofovir has been chosen as a promising agent for Pre-exposure Prophylaxis (PREP) therapy for several reasons: it is taken once daily, can be taken without food, and also has a strong safety record, limited side effects, and a favorable resistance profile in HIV patients.
Research in animals indicates that tenofovir used as a PREP may be effective in reducing the risk of HIV infection[2,3], although recent data have questioned the longevity of tenofovir's protective effect if there is one[4]. In order to determine if tenofovir may prevent HIV infection in humans, the National Institutes of Health, the Bill and Melinda Gates Foundation and the Centres for Disease Control funded a series of randomized trials (See Table 1). The conduct of these trials have, however, received widespread criticism by activist groups citing ethical concerns and a lack of community involvement in the planning of the trials[5,6]. This emerging opposition has halted the progress of 2 PREP trials, in Cambodia and Cameroon, and threatened the stability of planned PREP trials, and related recruiting effort, in other developing nations[7].
Table 1 Current tenofovir PREP Studies
Study Location Population Group Sponsor Study Goal Expected Initial Results
Cambodia Commercial sex workers 960 volunteers NIH Safety & efficacy 2007
Cameroon, Ghana High-risk women 800 volunteers FHI Safety & efficacy 2007
Malawi High-risk men 500 volunteers FHI Safety & efficacy 2007
Botswana Young adults 1,200 volunteers CDC Safety &efficacy 2007
Thailand Injection drug users 1,600 volunteers CDC Safety & efficacy 2007
United States Men who have sex with men CDC Safety 2007
Peru Men who have sex with men 2,100 volunteers NIH Safety & efficacy 2008
Trial closures
The first trial to close early was a randomized trial to assess the safety and efficacy of tenofovir as a PREP agent in Commercial Sex Workers (CSWs) in Phnom Penh, Cambodia. The trial planned to recruit 960 CSWs and was led by investigators from the United States (US) and Australia. In July 2004, however, a protest staged at the XV International AIDS Society conference in Bangkok, Thailand, brought worldwide media attention to the trials[8,9]. International activist and local representatives of participant groups led the protest. This protest, as well as subsequent demonstrations directed at the Cambodian Ministry of Health, resulted in the early closure of the trial by the Cambodian Prime Minister, prior to recruitment. The Cambodian Ministry of Health has provided no official reasons for their decision to cancel the trial.
In February, 2005, an extension site of the PREP trial in Cameroon was halted by the Cameroon national Ministry of Public Health[6]. The trial was being conducted by Family Health International (FHI) and had begun recruiting participants. At the time of closure, the trial had enrolled 400 high-risk sexual behaviour participants. In this case, media attention again acted as a catalyst to raising concerns about the quality of treatment provided to participants and the quality of care that might be provided afterwards. Protests, led by ACT-UP Paris, an international AIDS activist group based in several countries, and collaborating with Réseau Éthique Droit et Santé (REDS), a Cameroon based AIDS activist group highlighted the concerns with the conduct of the trial and delivery of care[10]. A documentary examining the activists' allegations aired on French TV-2 and made the trial international news. In response to allegations of trial misconduct and ethical violations, the Cameroon government established an independent inquiry into the trial conduct. The independent inquiry reported on February 23, 2005, that the trial cannot proceed without regular reporting and a formal study site accreditation for the satellite trial clinic[11], issues that that had not been addressed by the activist groups or media. The committee did however, identify that many of the allegations made by the media about a lack of safety were false. The inquiry has since recommended that the trials resume after the trial administrators dealt with the reporting issues and attain site accreditation. ACT-UP Paris reports that they will continue to protest the PREP trials taking place in other countries[12].
The media's role as a disseminator of scientific research is particularly important in areas of scientific conduct[13,14]. Because no peer-reviewed publication had been published at the time of conducting this study and the media was a strong catalyst in bringing attention to these trials, we aimed to characterize the nature of their reporting. We determined the sources of information that the media utilized in reporting on the tenofovir trial and the extent to which they consulted activists, researchers and participants involved in the trial.
Methods
In order to identify all relevant media articles, we searched 3 media databases (POPLINE, PROQUEST and LexisNexis), 5 electronic medical databases (AMED, CINAHL, E-Psyche, EMBASE and MEDLINE) as well as extensively searched the Internet using Google from inception to March 10, 2005. Our specific search terms included, but not limited to: "tenofovir", "viread", "Gilead", "Cambodia" and "Cameroon." We limited articles to those published in the English language. We contacted advocacy agencies and FHI to determine recent interviews they may have provided. We additionally reviewed CDC factsheets.
To be included in our review, articles had to report on the trial controversy and closures in Cambodia or Cameroon. We included articles from any media source, but excluded non-media articles posted on Non-Government Organization (NGO) and activist websites as these contain mostly blogs or position papers and are not intended to be objective. We additionally excluded articles addressing the trial stopped in Nigeria, since this trial was stopped by FHI for logistical reasons, as well as the planned trials of PREP in other developing countries, because they have not been halted at the time we conducted our search [7].
Three authors independently reviewed articles and reviewed them for relevance (PW, BR, EM). Using content analysis, we developed a coding template and extracted the following information from each article: source, date, location of article, author of article, individuals cited, organizations cited, events reported, and source of evidence. We extracted data on the initial coding template through a first reading of the articles to identify major themes. We agreed on the theme categories through consensus. We then reread the articles and coded them appropriately independently, in duplicate. We focused on reporting the claims and counter-claims reported concerning the trials. We aimed to determine the extent to which media reports sought input from stakeholders. As much misleading information was available on the internet through web-blogs, we believed that contact with stakeholders would provide greater inferences into the actual reasons for contentiousness or conduct of the trials. We specifically determined how many articles reported speaking with the following individuals involved in the trial closures: activist groups, researcher groups involved in the trial, local ministry officials and participants. All disagreements were resolved by consensus.
Results
Our systematic searches yielded 52 relevant articles addressing the trial in Cambodia. We excluded 22, leaving 30 reports, of which 6 are duplicates. In total, we included 24 distinct, original reports addressing the Cambodian trial [15-34]. We included 13 reports that specifically addressed the trials in Cameroon [33,35-45].
Allegations
Cambodia (See Additional file 1)
Thirteen of the Cambodian reports alluded directly to the Prime minister (PM), Hun Sen [5,8,19,21-23,27,29,30,46-48]. The PM was quoted numerous times as being concerned with the study's effects on the human rights of the Cambodian people. With pressure from the Cambodian Sex Workers (CSW) to suspend the study, the PM finally conceded and the trial was halted in August 2004. The representatives from the CSW NGO, Women's Network for Unity, were against using CSWs from poor countries like Cambodia for experimental drug testing. Numerous allegations were made supporting the early termination of the trial and are outlined in Additional file 1. Most were made by CSWs and members of various activist groups opposed to the study. Seventeen reports discussed the demand for 30–40 years of medical insurance to counter any adverse events [15-17,22,25-31,33-35,49,50]. Seventeen papers referred to the claim that some aspects of the study violated basic human rights [5,8,16,19,22-24,27-30,47,48,51]. Twelve reports, in particular, discussed the allegations regarding the participation and potential 'exploitation' of an already vulnerable population [5,16,17,19,23,26,27,29,30,48]. Eleven reports discussed claims made against the study's protocol. For potential participants and many activists, the use of a placebo or a 'dummy pill' predominantly became an ethical concern [5,8,21,24,27,28,30,33,51-53]. Nine reports discussed suspicions regarding the need to run the trial in Cambodia when there were high-risk populations residing in the US and Europe [5,16,19,22,26,30,47,50,52]. Finally, seven reports discussed the allegations that researchers were purposefully providing insufficient counseling and educational resources, ensuring that some women would, in fact become HIV positive during the trial [5,16,23,27,28,33,52]. Of note, one news report published in an academic journal, inaccurately reported that tenofovir was a vaccine[31].
Counter claims
The demand for long term medical insurance was counterclaimed by an investigator affiliated with the trial in 1 report[50]. She mentioned that this appeal was impossible to meet due to the enormous costs this would entail. She further mentions that there is no place in the world, where participants in clinical trials receive coverage for life. A representative from FHI directly counters the allegations made concerning the violation of human rights[47]. The FHI representative countered that the sex workers involved in the trial are offered an enhanced standard of care well beyond what is typically offered in Cambodia and other HIV prevention trials. No investigators or officials were reported responding to the allegations regarding the use of a placebo nor the claim that counseling is purposely limited to ensure some seroconversions.
Cameroon (see Additional file 2)
There were 13 reports addressing the tenofovir trial in Cameroon [12,33,35-37,39-45,54]. Nine reported allegations that there was no care for those who seroconvert during the trial [12,33,35-37,39-41,54]. The charge that participants are not being sufficiently informed of the risks involved was reported in 9 articles [12,33,35-37,40,41,43,44]. Other major allegations included an inadequate number of support staff (6/13) [12,36,39-41,54], participants being exploited and treated like 'guinea pigs'(6/13) [12,36,39-41,54], an unethical study design (5/13) [12,36,40-42,44] and finally, the belief that trial is only being conducted in Cameroon to promote Gilead's commercial prospect (5/13) [12,40-42,44].
Counter claims
Two reports interviewed investigators who directly responded to the claim that no help is provided for those who seroconvert during the trial [36,43]. The study co-ordinator in Douala was cited, saying all volunteers who have become infected are referred to approved medical centers for treatment[36]. A representatives from FHI in another report [43], responded citing that study-support services are offered and that women can decide if they want to continue on this path. In addition, it was reported that women have access to medical care and treatment including referral to services where they can receive care for HIV. The representative emphasized that women who have volunteered to participate will have "life-long access to HIV care and treatment". To counter the claim that participants are not fully nor sufficiently informed of the risks involved, an FHI representative in one report directly responds to this allegation. She states that all potential participants "were counseled before the trial started to make sure they understood the potential risks and benefits of study participation" [43].
Sources of information (see Table 2)
Table 2 Sources of information cited in media reports
Study Total Investigator Potential participant Representatives of participant groups Activist group Others
Cambodia 24 3 (12.5%) 1 (4.1%) 1 (4.1%) 3 (12.5%) 3 (12.5%) NIAID representative;
Cambodian AIDS authority;
Bio-consultant
Cameroon 13 4 (30.8%) 0 1 (7.7%) 6 (46.2%) 1 (7.7%) Local physician
Cambodia
In order to evaluate the soundness of the reports, we identified the source of the information to determine whom, if anyone had been interviewed. Potential subjects for the interviews included trial investigators, potential participants (ie. CSWs), representatives or activist group members. Of 24 reports, 14 did not identify any primary source of information [8,16,21-23,27-30,33,50,52,55,56]. In 3 reports trial investigators or officials were interviewed [47,48,50] All investigators emphasized that the concurrent trial care provided was beyond the standard of care in Cambodia and in-addition to what is offered in other HIV prevention trials.
One report interviewed potential participants [24]. In three reports that cited activists, their comments came from web blogs [5,26,53]. Representatives of NGOs were interviewed in one report [48].
Cameroon
Of the 13 reports, 4 did not report a primary source of information[33,35,44,45]. Two reports interviewed both researchers and opponents[36,37]. Four reports concerning the Cameroon trials surveyed investigators[36,37,41,43]. There were no interviews with potential participants in any reports. An interview with a participant representative was recorded in one report [36] and HIV activists in 6 reports, either in person or through Internet forums [12,36,37,40,42,54].
Discussion
The findings of this review should be of interest to clinical trialists, advocates and policy makers. The media are important communicators of information and can raise awareness of emerging issues concerning health and risk[57] We observed that the media brought forth a variety of concerns related to the conduct of the trials in both Cambodia and Cameroon. In several instances these concerns were not supported by evidence and were possibly inaccurate. We also observed that the media involved stakeholders to a comparatively small degree, suggesting that the viewpoints they reported may not accurately represent the views of all those concerned about the conduct of the trial. Given the potential impact of the media reporting on the conduct of future tenofovir trials, the nature of their reporting deserves closer scrutiny.
There are several limitations to consider in interpreting our review. Although we systematically searched many databases and extensively searched the Internet, it is possible that we were unable to identify some reports. Our searches were limited to the English and French languages, and as both Cambodia and Cameroon have media in other languages, we may have missed articles published in their respective languages. We conducted a content analysis to identify issues that we deemed trial related. It is possible that other issues exist, and that other individuals were interviewed, but were not reported in the articles. There are also several strengths to consider in interpreting our review. This is, we believe, the first systematic review to assess media reporting related to the closed trials. We conducted extensive searches and extracted data independently, in duplicate, to remove investigator driven biases. We spoke with representatives from each stakeholder group to determine if they were aware of additional reports, as well as searched the CDC reports.
The media reports consistently reported concerns with ethical issues in the conduct of the trials. Both the Cambodian reports and the Cameroon reports cite access to appropriate standards of care for those who become infected during the trial and appropriate standard of prophylaxis. Indeed these same issues are being echoed in the impending trial of tenofovir in drug users in Thailand[9]. Further to this, most media reports regarding both trials reported that participants were not fully aware of the risks involved with participation. Clinical trialist's planning prevention trials, whether of chemoprophylaxis or microbicides should be aware that many populations are not in situations in which to make informed decisions about clinical trial methodologies and the risks related to trial participation. Enrolling these communities in developing nations requires the establishment of community advisory boards and gender advisory boards, to ensure that the information is being provided, and interpreted, in an accurate manner[58,59].
The tenofovir trial media attention does however, have implications for prevention trials, such as HIV vaccines. Prevention methods are urgently needed to stem the onslaught of new HIV infections. Novel interventions, such as vaccines, microbicides, or chemoprophylaxis, will all require clinical trial validation. The media reporting of the tenofovir trials may threaten trial recruitment and potentially stigmatize trial participation.
There are important qualitative differences between the media reports of the trial in Cambodia compared to the trial in Cameroon. It is important to note that the trial in Cambodia was closed before recruiting any participants, whereas the trial in Cameroon had met its recruitment goals (n = 400). Most criticisms regarding the trial in Cambodia related to a lack of access to adequate care during the trials and post-trial, for those who may become infected. These concerns for potential infections appeared to be the overriding concerns as activists cited that these trials would be conducted to a different level of protection and insurance if they were conducted in a developed country. The articles addressing the trial closure in Cameroon were somewhat different and were largely based on reports of trial misconduct, in the context of inadequate informed consent, inadequate access to counseling and inadequate access to male and female condoms. The trials did however; have overlapping concerns about the ethical standards for trials in developing nations compared to Western nations and the protection of the participants human rights.
The tenofovir trials raise important questions about the relationship between researchers, participants and activists[60]. With worldwide media attention, these trial closures demonstrate the ability of activists to engage the media and bring about important consequences for the conduct of trials. Activists are experienced at bringing about change and have strong lobbying potential that can impact researchers and the HIV community. The activist communities have substantial experience using the media to address topics of importance in HIV and are well-educated regarding ethical standards and trial designs. Indeed, it is important to note that several of the ethical issues raised in the media reports: standard of care, access to proven prophylaxis and access to treatment for seroconverters; have not yet been resolved within the academic community [61,62].
The Media, accuracy of communication and influence on policy
The media plays an important role at the interface of science and policy. Indeed, the media has been utilized to change behaviour in HIV public health campaigns[60]. Most scientists and non-scientists receive information from media sources [63]. One policy making model specifically addresses this issue and describes how information is generated by researchers and then disseminated by advocacy networks and the media [64]. The information provided by the media is eventually interpreted by policy makers and government officials, and those informally involved, such as patient groups and stakeholder groups. Although media information provides weak levels of evidence, multiple sources of media information may strengthen personal inferences. The interpretation of this information is influenced by the value systems of the individuals receiving the information.
This review displays the important role that the media play in knowledge dissemination, and the responsibilities that they have in accurate reporting. In our review, 20 reports did not acknowledge interviewing any stakeholder and in no case did any report interview those supporting the trials and those against the trials. It is impossible to determine the extent to which the allegations made regarding the trials were wholly inaccurate. Our review contributes to a large body of evidence indicating that the media at times misrepresents scientific information and presents the science in a varied manner [65], as in the case of the trial in Cameroon where the reports suggest inadequate counseling and informed consent sheets that were not available in the local languages, issues that were not upheld by the ministry of health. It should be noted however, that not only standard news sources, but also academic news sources were susceptible to inaccuracies in their reporting. In one case, a top academic journal reported tenofovir to be a vaccine trial[31].
Indeed the media plays an important role in highlighting HIV issues so that it remains of topic of high profile and can be used as a strategy to convey AIDS awareness and prevention[60]. The media has a responsibility to report in a factual and objective manner [66]. Media agencies should be engaged to educate them on trial protocols and ethics so that reporting can be discriminating. The International AIDS Vaccine Initiative (IAVI) has now initiated training with journalists to ensure knowledgeable reporting related to vaccines [67]. This seems like a logical step towards ensuring adequate reporting of prevention trials.
We recommend further analysis of this emerging issue of media reporting of contentious clinical trials. The impact of such reporting, as we have described, clearly demonstrates that it deserves closer attention. Next steps would engaging journalists involved in the reporting to determine what strategies clinical trialists, participants and stakeholders could utilize to improve reporting. An analysis of the strategies used by stakeholders to provide information to the media and the relative successes of these strategies would also be worth studying.
Conclusion
In conclusion, our review identifies the heterogeneous reporting of issues related to these trial closures and the relatively poor involvement of stakeholders for the purpose of interviews. In several instances these concerns were not supported by evidence and were possibly inaccurate. However, the overall issues reported in the media reports is a sense of distrust and concern for the well-being of participants. Given the role of the media in reporting on the conduct of future prevention trials, serious initiatives are warranted to engage the media and preempt difficulties in the transference of information.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
EM, SS, EW conceptualized the study. EM, EW, SS, PW and BR carried out the searches and data abstraction. EM, EW, BR, PW, SS, and KW wrote the drafts of the manuscript. EM, EW, BR, PW, SS, and KW dealt with critical revisions and final submission of the manuscript. All authors approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Supplementary Material
Additional File 1
Concerns cited in Cambodian reports.
Click here for file
Additional File 2
Concerns cited in Cameroon reports.
Click here for file
==== Refs
UNAIDS: AIDS epidemic update Geneva: UNAIDS 2004
Van Rompay KK McChesney MB Aguirre NL Schmidt KA Bischofberger N Marthas ML Two low doses of tenofovir protect newborn macaques against oral simian immunodeficiency virus infection J Infect Dis 2001 184 429 438 11471100 10.1086/322781
Van Rompay KK Schmidt KA Lawson JR Singh R Bischofberger N Marthas ML Topical administration of low-dose tenofovir disoproxil fumarate to protect infant macaques against multiple oral exposures of low doses of simian immunodeficiency virus J Infect Dis 2002 186 1508 1513 12404171 10.1086/344360
Subbarao S Otten R Ramos A Jackson E Adams D Kim C Bashirian S Johnson J Monsour M Janssen R Paxton L Greenberg A Folks T Chemoprophylaxis with oral tenofovir disoproxil fumarate (TDF) delays but does not prevent infection in Rhesus Macaques given repeated rectal challenges of SIV 12th Conference on Retroviruses and Opportunistic Infections 2005 Boston 136LB
Cohen J AIDS research. Cambodian leader throws novel prevention trial into limbo Science 2004 305 1092 15326323 10.1126/science.305.5687.1092a
Cohen J Cameroon Suspends AIDS Study Science, Science Now 2005 Feb. 4, 2005
Cohen J AIDS clinical trials. More woes for novel HIV prevention approach Science 2005 307 1708 15774734 10.1126/science.307.5716.1708a
Chase M Key AIDS study in Cambodia now in jeopardy The Wall Street Journal 2004 New York
Singh JA Mills EJ Abandoned trials of Pre-Exposure Prophylaxis for HIV: What went wrong? PLOS Medicine 2005 2 e234 16008507 10.1371/journal.pmed.0020234
IRIN-News CAMEROON: Clinical trial of anti-HIV drug on sex workers in question 2005
Atatah C Douala AIDS Drug Controversy: Medical Council Says Trials Violated Ethical Norms The Post News Line 2005
ACT-UP-Paris tenofovir: inethical trial ACT-UP Paris 2004
Wilson K Code C Dornan C Ahmad N Hebert P Graham I The reporting of theoretical health risks by the media: Canadian newspaper reporting of potential blood transmission of Creutzfeldt-Jakob disease BMC Public Health 2004 4 1 14706119 10.1186/1471-2458-4-1
Nerlich B Clarke DD Anatomy of a media event: how arguments clashed in the 2001 human cloning debate New Genet Soc 2003 22 43 59 15282898 10.1080/1463677032000069709
James JS Cambodia stops important tenofovir prevention trial AIDS Treatment News 2004
Munthit K Cambodian PM orders anti-AIDS drug test halted; cites human values 2004 Phnom Penh: Canadian Press
Naik MCaG Key AIDS Study in Cambodia Now in Jeopardy Wall Street Journal 2004 B1 B1
Ahmad K Trial of antiretroviral for HIV prevention on hold The Lancet Infectious Diseases 2004 4 597 15470760 10.1016/S1473-3099(04)01159-4
Stephan H Growing Pains in AIDS drug Development: HIV drug trial in Cambodia halted Nature Biotechnology 2004
McKew M Drug Trial to start in Cambodia The 7:30 report 2004
Russell S AIDS Chicago Antiviral Drug Used to Treat AIDS to be tested as Vaccine San Francisco Chronicle 2004
JMO Pro-abortion Group Thwarted in attempt to Test Dubious HIV drugs on Cambodian Prostitutes Phnom Penh: LifeSiteNews 2004
Alcorn K Cambodian study of tenofovir for HIV prevention halted, PM cites 'human rights' AIDSMAP 2004
Timberg C Dose of Prevention where HIV thrives: Nigeria Brothel is test site for new pill 2004 Washington Post. Washington A15 A15
Cohen J Cambodian Leader Throws Novel Prevention Trial into Limbo Science 2004 305 1092 1092 15326323 10.1126/science.305.5687.1092a
Jeffreys E Cambodian Sex Workers win against Bill Gates Scarlet Alliance 2004
Comments IN Cambodia's premier halts planned trials of AIDS drug Life Issues Net 2004
Cairns G Sex Workers force end to drug trial Positive Nation 2004
Canada.com Cambodia's prime minister halts AIDS drug trial Science and Development Network 2004
Thomas J Cambodian HIV Clinical Trial impasse: UNAIDS mediation is essential AIDS Asia 2004
Ojcius D HIV vaccine trial halted Nat Rev Microbiol 2004 2 776 15468461 10.1038/nrmicro1010
Munthit K PM wants anti-AIDS drug trials stopped The Gazette 2004 Montreal A14 A14
Reenie S The Gilead Saga American Journal of Bioethics Editors Blog February 18, 2005
Cannistra SA The ethics of early stopping rules: who is protecting whom? J Clin Oncol 2004 22 1542 1545 15117977 10.1200/JCO.2004.02.150
Bernard EJ Cameroon study of tenofovir for HIV prevention suspended, Health Minister claims 'dysfunctions' in study protocol Douala: Aidsmap 2005
AEGiS-IRIN CAMEROON: Clinical trial of anti-HIV drug on sex workers in question Yaounde: Integrated Regional Information Network 2005
Colombant N Cameroon Activists Want Protection of Drug Test Victims Abidjan: VOA 2005
Coalition GCfMatAVA Statement on the PREP trials in Cambodia and Cameroon AF-AIDS 2005
publique J-LR-Mds Reply to posting by Act-UP Paris 2004
Samba-Kounzi R [email protected] 2004
Shetty P Cameroon suspends trial of AIDS drug after protests Science and Development Network 2005
Tall DCT "Gilead Scandal in Cameroun" AfriCASO condemns Actions Treatments 2005
AF-AIDS Tenofovir trials: Interview with Beth Robinson from FHI HealthGap February 5 2005
Chase M AIDS-Preventative Test Halted by Cameroon Amid Protests The Wall Street Journal 2005 16134271
Simeon V SIDA: Le Scandale est dans l'eprouvette La nouvelle Expression 2005 Douala
AIDS drug trial stopped in Cambodia Washington Times 2004
Major HIV drug trial to be halted BBC News 2004
J C The Asian Equation: One size doesn't fit all International Antiviral Therapy Evaluation Center 2004
News B Major HIV drug trial to be halted BBC News 2004
Cambodian sex workers reject HIV drug trials Asian Labor News 2004
Ahmad K Trial of antiretroviral for HIV prevention on hold Lancet Infect Dis 2004 4 597 15470760 10.1016/S1473-3099(04)01159-4
Ethics, drug trials and HIV: the case of tenofovir
James JS Cambodia stops important tenofovir prevention trial AIDS Treat News 2004 4 5
Group NHVaMA Phase 2 trial of oral tenofovir use on a chemoprophylaxis for HIV infection in Nigeria: The outcome of community involvement with the scientific research process 2004
Sex workers trump Gates' humanitarian bid Business Report 2004 March 30
Munthit K PM wants anti-AIDS drug trials stopped The Gazette
McCombs M Shaw D The Agenda-Setting Function of the Mass Media Public Opinion Quarterly 1972 36 176 187 10.1086/267990
Mills E Cooper C Guyatt G Gilchrist A Rachlis B Sulway C Wilson K Barriers to participating in an HIV vaccine trial: a systematic review Aids 2004 18 2235 2242 15577535 10.1097/00002030-200411190-00003
Mills EJ Singh S Dolma S Nixon S Kapoor S Enrolling women into HIV prevention trials: A logistical challenge but ethical imperative Presented at the 2005 Canadian HIV AIDS Research Conference 2005 Vancouver
Duan N Listening to consumers and HIV vaccine preparedness Lancet 2005 365 1119 1121 15794955 10.1016/S0140-6736(05)71857-2
Schuklenk U The standard of care debate: against the myth of an "international consensus opinion" J Med Ethics 2004 30 194 197 15082817 10.1136/jme.2003.006981
Lie RK Emanuel E Grady C Wendler D The standard of care debate: the Declaration of Helsinki versus the international consensus opinion J Med Ethics 2004 30 190 193 15082816 10.1136/jme.2003.006031
Steinbrook R Medical journals and medical reporting N Engl J Med 2000 342 1668 1671 10833217 10.1056/NEJM200006013422212
Lomas J Connecting Research and Policy Canadian Journal of Policy Research 2000 1 140 144
Wilkie T Sources in science: who can we trust? Lancet 1996 347 1308 1311 8622510 10.1016/S0140-6736(96)90947-2
Social Issues Research Centre, Royal Society, Royal Institution of Great Britain. Guidelines on science and health communication. Oxford: Social Issues Research Centre 2001 Accessed April 28, 2005
IAVI International AIDS Vaccine Initiative 2005 Accessed April 28, 2005
|
16120208
|
PMC1242229
|
CC BY
|
2021-01-04 16:29:56
|
no
|
BMC Int Health Hum Rights. 2005 Aug 24; 5:6
|
utf-8
|
BMC Int Health Hum Rights
| 2,005 |
10.1186/1472-698X-5-6
|
oa_comm
|
==== Front
BMC ImmunolBMC Immunology1471-2172BioMed Central London 1471-2172-6-221617658110.1186/1471-2172-6-22Research ArticleReciprocal role of cyclins and cyclin kinase inhibitor p21WAF1/CIP1 on lymphocyte proliferation, allo-immune activation and inflammation Khanna Ashwani K [email protected] Department of Medicine (Nephrology), Medical College of Wisconsin, Milwaukee WI-53226 USA2005 21 9 2005 6 22 22 9 7 2005 21 9 2005 Copyright © 2005 Khanna; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Immune activation that results due to the aberrant proliferation of lymphocytes leads to inflammation and graft rejection in organ transplant recipients. We hypothesize that the cell cycle control and inflammation are parallel events, inhibition of cellular proliferation by cyclin kinase inhibitor specifically p21 will limit inflammation and prevent allograft rejection.
Methods
We performed in vitro and in vivo studies using lymphocytes, and rat heart transplant model to understand the role of cyclins and p21 on mitogen and allo-induced lymphocyte activation and inflammation. Lymphocyte proliferation was studied by 3H-thymidine uptake assay and mRNA expression was studied RT-PCR.
Results
Activation of allo- and mitogen stimulated lymphocytes resulted in increased expression of cyclins, IL-2 and pro-inflammatory cytokines, which was inhibited by cyclosporine. The over-expression of p21 prolonged graft survival in a completely mismatched rat heart transplant model resulted by inhibiting circulating and intra-graft expression of proinflammatory cytokines.
Conclusion
Cyclins play a significant role in transplant-induced immune activation and p21 over-expression has potential to inhibit T cell activation and inflammation. The results from this study will permit the design of alternate strategies by controlling cell cycle progression to achieve immunosuppression in transplantation.
==== Body
Background
Alloimmune activation, caused by aberrant T lymphocyte proliferation is one of the key post transplant events in organ transplant recipients. Current immunosuppressive drugs are therefore designed to inhibit T lymphocyte proliferation. Our previous studies have demonstrated that immunosuppressive drugs, cyclosporine (CsA) tacrolimus (TAC), and sirolimus (SRL) besides inhibiting lymphocyte proliferation and IL-2 also induce the expression of TGF-β and other fibrogenic molecules [1-3] leading to nephrotoxicity and chronic rejection. Therefore, there is a need to develop alternate strategies to achieve immunosuppression for increased graft survival with least nephrotoxicity. The most effective immunosuppression can be achieved by the direct inhibition of T lymphocyte proliferation. Since the expression of cyclins and cyclin-dependent kinases and pro-inflammatory cytokines is increased during T lymphocyte proliferation (4), control of T cell proliferation by regulating the expression of cyclins would potentially inhibit allo-immune activation and inflammation. p21WAF1/CIP1 is one of the most potent cyclin kinase inhibitor and therefore has potential to control the expression of cyclins and T cell activation.
We have demonstrated that CsA, TAC and SRL [4-6] induces the expression of cyclin kinase inhibitor p21WAF1/CIP1 and also in vitro and in vivo over-expression of p21WAF1/CIP1 in lymphocytes results in decreased response to mitogenic stimuli and greater sensitive to the inhibitory effects of cyclosporine [7]. The present study was designed to study the expression of cyclins during T cell activation, allograft rejection, and the effect of p21WAF1/CIP1 on mitogenic and allogeneic stimulation, pro-inflammatory cytokines and graft survival in a rat heart transplant model.
Methods
Preparation of lymphocytes and experimental protocol
Lymphocytes from the normal individuals (n = 4, obtained from Blood Center of Greater Milwaukee, Milwaukee) were separated as described [1]. For mRNA studies, lymphocytes (1 × 106/ml) were cultured with PHA (2 μg/ml), with and without CsA (100 mg/ml) for 4 h, cells were washed twice with PBS and the cells were stored in Trizol at -80 C for RNA preparation.
Rat cardiac transplantation
Hetrotopic heart transplants were performed as described by us [8]. We used Lewis (LEW, RT11) and Wistar-Furth (WF RT1u) rats, which represent complete genetic disparity at both major and minor histocompatibility loci. Isografts were performed in LEW-LEW while allogeneic transplantations were performed in WF-LEW transplant combinations. Immunosuppression was accomplished using CsA at a dosage of 2.5 mg/kg for the whole duration of the experiment described. Rats were monitored daily for evidence of allograft slowing and failure and at the time of rejection, animal s were sacrificed and spleens were used to prepare lymphocytes and organs were snap frozen in liquid nitrogen and stored at -80 C till isolation of RNA.
Detection of mRNA by reverse transcription and polymerase chain reaction (PCR) in allografts and lymphocytes
Total RNA was isolated from lymphocytes with Trizol (Invitrogen, Carlsbad, CA) and tissues with SV RNA isolation kit (Promega, Madison, WI). Purity of RNA was confirmed by a ratio of 260/280 nm. 1 μg of RNA was reverse transcribed into cDNA using a superscript reverse transcription kit from Invitrogen (Carlsbad CA). The amplification of specific mRNA expression was achieved by polymerase chain reaction (PCR) using specific primer sequences for p21WAF1/CIP1, β-actin, IL-2, TNF-α; Cyclin G, Cyclin E, Cyclin D3, IL-6, and IL-10 are described by us [4-6]. The PCR products were resolved in 1% agarose gel electrophoresis, ethidium bromide stained specific bands were visualized under UV light and photographed. The densitometric analysis of specific bands was made using Alpha-Imager (Alpha Innotech Corp, San Leandro, CA) and data are represented as the ratio of the specific gene to β-actin. We performed cycle analysis for each primer pair to select a cycle number for amplification for each gene studied.
Jurkat T cells proliferation assay
Cell proliferation was determined using 3H-thymidine incorporation as previously described (1–2). All assays were performed in triplicate. A total of 3 individual experiments investigating the proliferation of unaltered and p21WAF1/CIP1 Jurkat cells [described in ref. 7] were performed in unstimulated and activated with PHA. Briefly, 200,000 cells were added to each well of a round bottom 96-well plate. PHA (2 μg/ml) was added to the wells, controls were without PHA. The cells were cultured for 64 h at 37°C in 95% air and 5% CO2 enriched environment. The cultures were pulsed with 3H Thymidine (1 μCi/well) for the last 16 h of incubation, cells were harvested and radioactivity counted using a scintillation counter. 3H -Thymidine uptake was expressed as the mean counts per minute of triplicate samples. The magnitude of Jurkat T cells proliferation from unaltered and p21WAF1/CIP1-augmented cells were investigated at rest following mitogen stimulation.
In vivo transfection of p21WAF1/CIP1
p21 sense plasmid DNA
Plasmid DNA was isolated from competent E-coli cells transformed with either the empty pcDNA3.1/Zeo vector (Invitrogen, Carlsbad CA) or the vector containing the full-length p21WAF1/CIP1 gene in the sense direction described by us [7].
Mixed Lymphocyte Reaction (MLR)
MLR was performed with splenocytes from isografts, untreated and CsA treated transplant recipients (LEW) as responders and irradiated splenocytes from donor strain (WF) as stimulators. The stimulator or responder cells were cultured alone as negative controls. 3H Thymidine uptake was expressed as the median counts per minute of triplicate samples. The extent of proliferation determined the allo-reactivity of among these groups of mice and rats.
Data analysis
Differences between groups were determined using two-tailed unpaired T test with significance considered present at a p value of less than 0.05. Statistical analysis was performed using a software program from GraphPad Software, Inc., San Diego, CA 92121 USA. The results are expressed as M ± SEM.
Results
Effect of inhibition of lymphocyte proliferation on IL-2, cyclins, TNF-α, IL-6 and p21WAF1/CIP1 mRNA
To understand the relationship between cyclins, pro-inflammatory cytokines and p21WAF1/CIP1 during lymphocyte proliferation and inhibition, we studied the mRNA expression of cyclin G, cyclin D3, IL-2, IL-6, TNF-α, and p21WAF1/CIP1 in lymphocytes activated in the presence or absence of CsA. The expression of IL-2 mRNA was used as a control for activation and inhibition of lymphocytes. The results from a representative of three consecutive experiments are shown in Figure 1A. Mitogen activation of lymphocytes resulted in an increase in IL-2, cyclins G and D3, TNF-α, IL-6 and p21WAF1/CIP1 mRNA (lanes 2), when compared to untreated lymphocytes (lane 1). As expected, CsA inhibited the expression of IL-2, and TNF-α mRNA however IL-6 mRNA was not completely inhibited. More interestingly, whereas the inhibition of lymphocytes by CsA also resulted in a significantly decreased expression of cyclins G, and D3 mRNA, and an increased expression of cyclin kinase inhibitor p21WAF1/CIP1 mRNA (lane 3) was observed.
The results obtained from three consecutive experiments as the ratio of each gene with β-actin (Mean ± SEM) are presented in Figure 1B. A statistically significant decrease in pro-inflammatory cytokines IL-2 (p < 0.016), IL-6 (p < 0.02), TNF-α (p < 0.03) and cyclins; Cyclin D3 (p < 0.01) and Cyclin G (p < 0.008) was observed in sharp contrast to a significant increase in p21 (p < 0.03) in lymphocytes activated in the presence of CsA. CsA treatment also resulted in increased expression of p21 protein in activated lymphocytes (Figure 1C). This increase was about 2 fold.
Expression of cyclins in lymphocytes from rejecting and non-rejecting rats
The rationale for these experiments is our hypothesis that in untreated and rejecting recipients of cardiac transplants, alloimmune activation will result in an increased expression of mRNA for cyclins in lymphocytes and the treatment with CsA will inhibit alloimmune activation and hence mRNA expression of cyclins leading to the prolongation of graft survival. To test this, we studied the mRNA expression of cyclins G, D3 and E in lymphocytes from rejecting (untreated) and non-rejecting recipients (CsA treated) in a heterotopic rat cardiac allograft model using a fully MHC mismatched [WF (RTlu) into LEW (RTll)] strain combination. The results (Figure 2A) demonstrate that the expression of cyclins G, D3 and E was significantly higher in lymphocytes obtained from untreated recipients of cardiac allografts (lanes 4, 5) compared to those treated with CsA (2.5 mg/kg) for 30 days (lanes 1–3). The untreated rats rejected graft between 8–10 days after transplant whereas CsA treated rats did not reject and were sacrificed on day 30th. Almost identical expression of the housekeeping gene β-actin in the lymphocytes of these rats is also shown (Figure 2A). The results from the untreated and CsA treated recipients are also presented as the ratio of each cyclin to β-actin (Mean ± SEM); cyclin D3 (0.43 ± 0.05 vs 0.16 ± 0.04, p < 0.03); cyclin G (0.45 ± 0.02 vs 0.1 ± 0.03, p < 0.003) and cyclin E (0.46 ± 0.03 vs 0.1 ± 0.04, p < 0.007). These results (Figure 2B) obtained from a semi-quantitative PCR analysis demonstrate that the aberrant alloimmune activation responsible for graft rejection is associated with the increased expression of cyclins. More significantly, CsA treatment significantly decreased mRNA expression of cyclins, alloimmune activation and prolongation of graft survival.
Correlation of the expression of cyclins and pro-inflammatory cytokines in allo-immune
To confirm that the increased expression of cyclins in lymphocytes from rejecting rats was due to alloimmune activation, we performed mixed lymphocyte reaction (MLR) using spleen cells from donor animals (WF) as stimulators and splenocytes from recipients (LEW) as responders. Three groups were studied; isografts (control), untreated allografts (A), and CsA-treated allografts (B). After five-day MLR lymphocytes were harvested and washed. RNA was prepared; reverse transcribed to cDNA, and was amplified by RT-PCR for cyclin D3, IFN-γ, TNF-α, IL-6 and IL-10 mRNA. As shown in the Figure 2C the mRNA expression of cyclin D3 was higher in lymphocytes from untreated allografts group A as compared to lymphocytes from isografts group B. This increased expression of cyclin D3 correlated with the pro-inflammatory cytokines IFN-γ and TNF-α mRNA expression. The expression of IL-6 and IL-10 mRNA was not statistically significant between groups A and B. CsA treatment decreased cyclin D3 expression by 50% and inhibited statistically significant (p < 0.03) IFN-γ mRNA expression. Lymphocyte activation in MLR assay was quantified by 3H-thymidine uptake assay. Proliferation of lymphocytes from untreated allografts was significantly higher (two tailed p value = 0.02) compared to CsA- treated allografts (Mean ± SEM of counts per minute, n = 3, 11796 ± 728 vs 7575 ± 360). These results support our conclusions from rat transplant studies that the alloimmune results in increased expression of cyclins mRNA that correlates with production of pro-inflammatory cytokines. Allo-immune activation is demonstrated by increased lymphocyte proliferation from MLR assay, which decreased in CsA treated animals.
p21WAF1/CIP1 over-expression, lymphocyte proliferation and IL-2 expression
To confirm our previous in vitro and in vivo studies that p21WAF1/CIP1 over-expression will inhibit lymphocyte proliferation and IL-2 expression, we conducted studies using Jurkat T cells. Jurkat T cells were transfected with empty vector plasmid DNA (control DNA) and p21WAF1/CIP1 sense plasmid DNA. We used Jurkat T cells for these experiments because of the ease with which these cells can be transfected as compared to primary T cells; these p21WAF1/CIP1 over-expressing Jurkat T cells are described previously [7]. No differences were observed in the proliferation of normal Jurkat cells and Jurkat cells transfected with empty vector DNA. Control and p21WAF1/CIP1 over-expressing Jurkat T cells were activated with PHA (2 μg/ml) for 4 h for IL-2 mRNA expression studies and 24 h for proliferation studies using 3[H]-thymidine uptake assay. The results demonstrate that the control Jurkat cells, but not p21WAF1/CIP1 over-expressing Jurkat T cells, responded to mitogenic stimulation by PHA (Figure 3A). PHA stimulation resulted in an increased expression of IL-2 mRNA in Jurkat cells, not in p21WAF1/CIP1 over-expressing Jurkat cells (Figure 3B). These results indicated that the p21WAF1/CIP1 over-expression rendered Jurkat cells unresponsive to mitogenic stimuli, possibly p21WAF1/CIP1 over-expression did not allow increased cyclin expression, thereby preventing lymphocyte activation by PHA. An increased expression of p21 protein in four different sets of p21-overexpressing Jurkat T cells (lanes 2–4) compared to Jurkat T cells transfected with empty vector plasmid DNA (lane 1) is also shown (Figure 3C) suggesting the increased p21 protein expression in these p21 overexpressing cells.
Effect of p21WAF1/CIP1 over-expression on graft survival in a rat heart transplant model
p21WAF1/CIP1 over-expression in Rats
Encouraged by our studies with mice, which demonstrated that transfection with p21WAF1/CIP1 sense plasmid DNA resulted in decreased lymphocyte proliferation, we performed pilot experiments to determine if in vivo over expression of p21 will also result in improved graft survival in a rat cardiac transplant model. We injected (intramuscularly) either p21WAF1/CIP1 sense plasmid DNA or empty vector plasmid DNA (1 mg) to 4 rats in each group. Since in our experiments with mice we used 100 μg of DNA, based on difference in the average weights of a rat and mouse, we used 10 times more DNA in rats. Seven days after the injection, animals were sacrificed, RNA was prepared from heart (h), liver (l), kidney (k) and spleen (s), reverse transcribed to cDNA and amplified for p21WAF1/CIP1 mRNA. Results shown in Figure 4A demonstrate that injection with p21WAF1/CIP1 sense plasmid DNA but not with empty vector plasmid DNA resulted in an over-expression of p21WAF1/CIP1 mRNA. These results also demonstrate that p21WAF1/CIP1 transgenesis using intramuscular injection of plasmid DNA can be achieved in rats. Since during isolation of RNA the contaminating DNA is treated with DNAse, the amplification of injected p21 sense plamsid DNA can be ruled out. The expression of p21WAF1/CIP1 protein using western blot was also detected in spleens, which was the only tissue analyzed (results not shown).
We then studied the effect of modulation of p21WAF1/CIP1 on alloimmunity in a rat heart transplant recipients. We have extensive experience using the completely MHC mismatched WF (RTlu) into LEW (RTll) strain combination and have well defined thresholds of cyclosporine-based immunosuppression. In this model, animals reject within 7 to 10 days in the absence of immunosuppression and as late as 180 days with immunosuppression (CsA 2.5 mg/kg). A total of 12 rat transplants divided into four groups (A-D) were performed. Recipients in Group A were given one intramuscular injection of empty vector plasmid DNA (1 mg); Group B rats were given a daily dose of CsA (2.5 mg/kg). Rats in Group C rats received three weekly injections of p21WAF1/CIP1 sense plasmid DNA (0.5 mg); and Group D rats were given one intra-muscular injection of p21WAF1/CIP1 sense plasmid DNA and a daily injection of CsA (2.5 mg/Kg). The allografts were followed by palpitation, and an arbitrary scale of 1–4 was used to rate the heartbeat to determine the time of graft rejection. The rats were sacrificed when a heartbeat of 1–2 was recorded, which was considered as a cutoff for rejection. Though the number of transplants is low, yet as shown in Figure 4B, p21WAF1/CIP1 alone (*p < 0.04) or in combination with CsA (**p < 0.005) significantly prolonged the graft survival.
To confirm that this effect was due to the inhibition of alloimmune activation, we studied the expression mRNA of IL-2 in lymphocytes isolated from spleens and heart allografts. We also examined the expression of IL-10 mRNA in lymphocytes and allografts. The results are shown in the Figure 4C. The expression of IL-2 mRNA both in lymphocytes and allografts was higher in animals injected with empty vector plasmid DNA demonstrating increased allo-immune activation. IL-2 mRNA expression decreased significantly in recipients treated with p21WAF1/CIP1 sense plasmid DNA alone or together with CsA. The expression of IL-2, correlated with rejection, which indicated an increased immune activity due to allo-immune response resulting in the rejection as compared to p21WAF1/CIP1 or p21WAF1/CIP1 /CsA treated recipients. We did not observe any significant changes in the expression of IL-10 mRNA in allografts, which decreased in animals treated with p21WAF1/CIP1 sense plasmid DNA alone or with CsA, however it did not reach a level of significance (Figure 4C).
Discussion
The experiments performed in this study were designed to understand the role of cyclins on mitogen and allo-stimulation of immune cells and also, if the inhibition of cyclins will correlate with pro-inflammatory cytokines. We also studied if p21WAF1/CIP1 modulation in recipients of cardiac transplantation modulates allo- and mitogenic stimuli and allograft survival. The results demonstrate that during lymphocyte activation, mRNA expression of cyclins and pro-inflammatory cytokines is significantly increased and CsA inhibited lymphocyte activation, mRNA expression of cyclins, pro inflammatory cytokines but induced p21 mRNA and protein expression.
Studies [9-12] have demonstrated that the expression of cyclin D3, cdk6, and cyclin E is activated in IL-2-stimulated T lymphocytes. However, the novel finding of this present study is that mRNA expression of cyclins in activated lymphocytes correlates with that of pro-inflammatory cytokines, and the expression of both the cyclins and pro-inflammatory cytokines is inhibited by immunosuppressive agent CsA. These are novel findings not demonstrated previously. Our results emphasize that the cell cycle progression and inflammation are concerted events thus regulation of cell cycle control could result in decreased inflammation.
Our in vitro findings on the increased expression of cyclins mRNA in activated lymphocytes were reproduced in our in vivo studies. The mRNA expressions of cyclin D3, G and E in lymphocytes (possible predominantly T cells, CsA inhibits proliferation of T lymphocytes) isolated from spleens from untreated recipients of rat cardiac transplant were significantly higher than those treated with cyclosporine. The increased expression of cyclins may represent an uncontrolled allo-immune activation in these rats. Since CsA treatment resulted in the inhibition of allo-immune activation and increased graft survival accompanied by a significant inhibition of mRNA expression of cyclins in CsA treated. This possibly was due to the CsA mediated inhibition of alloimmune activation. These results indicate the presence of an active cell cycle progression during allo-immune activation. Therefore, the control of cell cycle progression should prevent inflammation leading to an improved graft survival. These results are supported by our studies with MLR cultures using lymphocytes from rat heart transplant recipients. An increased proliferation of lymphocytes accompanied increased expression of cyclins and pro-inflammatory cytokine mRNA when responders lymphocytes were used from untreated rats as compared to those from isografts or CsA treated rat heart transplant recipients. Again, these activated lymphocytes were possible predominantly T cells, T lymphocyte proliferation is a key component of allo-immune activation. Therefore, these results lend credence to our thinking that the inhibition of allo-immune activation accompanies decreased expression of cyclins and pro-inflammatory cytokines.
These results confirm that control of cell cycle progression plays a significant role in T cell proliferation/activation. Role of p21 in other aspects of lymphocyte proliferation has been studied. Studies of Balomenos et.al, [13] Santiago-Raber et al [14] and Brian et al [15] demonstrated that T lymphocytes from p21WAF1/CIP1-/- mice proliferated significantly more than from wild type mice upon stimulation. These results support our studies that p21WAF1/CIP1 modulation alters cell cycle progression and the immune system. Jackson et al [16] showed that increased levels of p21WAF1/CIP1 at the end of G (1) could prevent cdk-mediated entry into S phase, leading to proliferative unresponsiveness also found in our experiments with p21WAF1/CIP1 over-expressing Jurkat T cells.
The results from this study are of significance because p21 is one of the most potent regulators of the cell cycle and is known to inhibit cell proliferation in two different ways. p21 binds to Cdk2 and inhibits PCNA (proliferating cell nuclear antigen), which is an auxiliary protein in DNA polymerase needed for DNA synthesis and nucleotide excise-n- repair [17]. PCNA has 6 binding sites for p21 [18]. Studies also [19] demonstrated that the PCNA binding and inhibitory activities reside in the C-terminal domain of p21, compared to the location of the CDK inhibitory activity in the conserved N-terminal domain. The authors also concluded that the CDK and PCNA inhibitory domains prevented DNA replication suggested a dual function of p21 as a cell-cycle inhibitor in vivo. We conducted these studies exclusively with cyclin kinase inhibitor p21WAF1/CIP1, though p53 and cyclin kinase inhibitors (p27, p16) have been shown to inhibit cell cycle yet p16 and p21WAF1/CIP1 inhibit cell cycle progression through distinct mechanisms [20]. The specific target for p16 is the Cdk/4cyclin D complex and in a tumor model, p21WAF1/CIP1 and p16 did not show additive or synergistic effects [21]. Furthermore in contrast to p21WAF1/CIP1, the expression of p27 is not under transcriptional control and its mRNA expression remains unchanged during cell cycle [22]. Also, high levels of p27 but not p21WAF1/CIP1 are observed in most quiescent cells and the inhibition of p27 levels precedes the progression of cell cycle [23]. Though both p21WAF1/CIP1 and p27 are critical in the response of cells to mitogens, p21WAF1/CIP1 provides a better balance between cyclins and cyclin kinase inhibitors [24] stressing its significance in inhibition of proliferation/immunosuppression. It is therefore possible that p21WAF1/CIP1 over-expression could interrupt the cell cycle progress and also prevent inflammation. It is well known that during T cell activation, expression of pro-inflammatory cytokines IFN-γ, TNF-α and IL-6 is significantly increased. Since T cells are the key mediators of allo-immune activation, this increased expression of cytokines in organ transplant recipient results in graft rejection [25-27]. Our results demonstrate a parallel increase in the expression of cyclins and pro-inflammatory cytokines. Therefore an inhibition/regulation of cell cycle progression of immune cells by over-expression of cyclin kinase inhibitor p21WAF1/CIP1 would decrease both allo-immune activation and inflammation in transplant recipients.
We also demonstrate that rats transfected with p21WAF1/CIP1 plasmid DNA over expressed p21WAF1/CIP1 mRNA in different tissues. The recipients of cardiac allograft animals who received intra-muscular injection of p21WAF1/CIP1 sense plasmid DNA had significantly increased graft survival compared to the recipients transfected with empty plasmid DNA. These very preliminary studies suggest that p21 overexpression can prolong graft survival to a degree comparable to prolongation by CsA. Further studies will surely be required to confirm and quantify the effect of p21 on graft survival. We also present unique results that the expression of IL-2 mRNA was significantly decreased in both lymphocytes and allografts isolated from p21WAF1/CIP1 over-expressing recipients of rat heart transplants.
Our method of using plasmid DNA to obtain in vitro and in vivo transfection of p21WAF1/CIP1 is based on the data supporting the efficacy of intra-muscular injection of plasmid DNA for a number of genes [28]. A number of studies [29-32] have demonstrated that non-viral plasmid DNA provides a simple, safe, and viable alternative for gene therapy involving muscle tissue resulting in high level of expression. More significantly plasmids do not induce neutralizing immunity, which permits repeated administration. Rauh et al [33] tested the hypothesis that intramuscular injection of naked DNA could result in the distribution remote from the site of needle placement, facilitating intramuscular gene transfer. Using transcutaneous ultrasound imaging the authors demonstrated that a solution of plasmid DNA administered by direct intramuscular injection into the skeletal muscles of the limb is distributed well beyond the site of needle entry and persisted for 8–10 weeks. Therefore, based on these and our own studies [7], we believe that the p21WAF1/CIP1 over-expression can be obtained through intramuscular injections of plasmid DNA, which could result in the decreased responsiveness of T lymphocytes to allo-and mitogenic stimuli.
In summary, the results from this study uniquely demonstrate that during lymphocyte activation, expression of cyclins is increased and the inhibition of lymphocyte activation by cyclosporine inhibits the expression of cyclins and increases the expression of cyclin kinase inhibitor p21WAF1/CIP1. The expression of cyclins correlates with that of pro-inflammatory cytokines like TNF-α and IFN-γ in vitro in activated lymphocytes and in vivo in lymphocytes from animals with rejecting rat heart transplants. These studies demonstrate that cyclins and pro-inflammatory cytokines are key mediators of allo-immune activation and the alteration of p21WAF1/CIP1 expression can modulate lymphocyte proliferation and allo-immune activation. These studies uniquely provide evidence on the role of cell cycle control molecules on allo-immune activation and will allow the development of alternate strategies to obtain improved graft survival in organ transplantation. Moreover based on our previously published studies, the results presented in this study, and our recently published study [34] we believe that p21WAF1/CIP1 might provide better immunosuppression with least side effects observed with the currently clinically used immunosuppressive drugs for the organ transplant recipients.
Acknowledgements
Author is thankful to Mr. Matthew Plummer for skillful technical assistance. This work was supported in part by research grant from the National Institutes of Health RO1AI41703.
Figures and Tables
Figure 1 CsA inhibits cyclins, IL-6 and TNF-α and induces p21WAF1/CIP1 mRNA expression in activated lymphocytes. A). A picture representative of three different experiments demonstrating cyclins (D3 and G), p21WAF1/CIP1, IL-6, TNF-α and induction of mRNA expression in activated lymphocytes (lane 2) with CsA (lanes 3). The lower bands in IL-2 mRNA area are primer dimers. B) Mean ± SEM (n = 3) of the ratio of cyclins, IL-6 and TNF-α and induces p21WAF1/CIP1 with β-actin is shown for lymphocytes either unctivated or activated with and without CsA. The p values signify the statistical significance for mRNA expression in lymphocytes activated with and with out CsA for each gene. C) Western blot analysis for p21 protein in lymphocytes unctivated (lane 1), activated (lane 2) with CsA (lane 3).
Figure 2 In vitro and in vivo mRNA expression of Cyclins relates to alloimmune activation: A: Cyclins; D3, E and G and β-actin mRNA expression in lymphocytes isolated from spleens of rat heart transplant recipients. Untreated (lanes 4,5), CsA treated (lanes 1,2,3). B: Mean ± SEM of the ratio cyclins with β-actin, p values are calculated between densitometric numbers of untreated and CsA treated transplant recipient. C: Cyclin D3 and pro-inflammatory cytokines mRNA expression in lymphocyte from MLR assay using stimulants from donor strain with responders from donor strain (Control) untreated rejecting transplant recipients (A) and CsA treated non rejecting transplant recipients (B). The p values represent the statistical significance between ratio of densitometric numbers for each gene with β-actin from untreated (Group A) vs CsA (Group B) treated transplant recipients.
Figure 3 Over expression of p21WAF1/CIP1 inhibits proliferation and IL-2 mRNA expression in Jurkat T cells: A: A comparison of the proliferation of Jurkat T cells with and without p21 overexpression. B. Il-2 mRNA expression in PHA activated normal and p21 over-expressing Jurkat T cells. An identical expression of house keeping gene β-actin is also shown. C: p21 protein expression in four different clones of p21 overexpressing Jurkat T cells, Untreated (lane 1), p21 overexpressing (lane 2–5) and transfected with empty vector DNA (lane 6).
Figure 4 p21WAF1/CIP1 over-expression prolongs allograft survival: A: p21 Injection of p21 sense plasmid DNA injected mice (set 2not empty vector plasmid DNA induces p21 mRNA expression in heart (h), liver (l), kidney (k) and spleen (s). B: Kaplan-Meyer survival graph for rat cardiac transplant recipients. Significant difference in the survival of allografts in p21WAF1/CIP1 transfected recipients compared to controls (* = p < 0.04) and p21WAF1/CIP1 together with CsA (* * = p < 0.005) can be seen. C: Effect of p21WAF1/CIP1 over-expression and CsA treatment on mRNA expression of IL-2 in lymphocytes and allografts. A significant decreased expression of IL-2 mRNA expression in lymphocytes and allografts compared to controls is shown (* = p < 0.01) and (* * = p < 0.001).
==== Refs
Khanna A Cairns V Hosenpud JD Tacrolimus induces increased expression of transforming growth factor-beta in lymphoid and non-lymphoid cells Transplantation 1999 67 614 619 10071036 10.1097/00007890-199902270-00021
Khanna A Cairns V Becker CG Hosenpud JD Transforming Growth Factor-beta (TGF-β) mimics and anti-TGF-β antibody abrogates the in-vivo effects of Cyclosporine: demonstration of a direct role of TGF-β in immunosuppression and nephrotoxicity of CsA Transplantation 1999 67 882 882 10199738 10.1097/00007890-199903270-00016
Khanna AK Mechanism of the combination immunosuppressive effects of Rapamycin with either cyclosporine or tacrolimus Transplantation 2000 70 690 694 10972232 10.1097/00007890-200008270-00027
Khanna A Hosenpud JD Cyclosporine induces the expression of the cyclin inhibitor p21WAF1/CIP1 Transplantation 1999 67 1262 1268 10342319 10.1097/00007890-199905150-00011
Khanna AK The immunosuppressive agent tacrolimus induces p21WAF/CIP1 via TGF-beta secretion Biochem Biophys Res Commun 2003 303 266 272 12646197 10.1016/S0006-291X(03)00340-1
Khanna A Plummer M Bromberek K Woodliff J Hariharan S Immunomodulation in stable renal transplant recipients with concomitant tacrolimus and sirolimus therapy BMC Med Immunol 2002 1 3 9 10.1186/1476-9433-1-3
Khanna AK Hosenpud JD In vitro and in vivo transfection of p21WAF1/CIP1 gene enhances Cyclosporine mediated inhibition of lymphocyte proliferation J Immunology 2001 165 1882 1889
Hosenpud JD Boyle TM Hensler H Sanford G Khanna AK The relationship between acute rejection and chronic rejection is highly dependent on specific MHC matching: A multi-strain rat heterotopic heart transplant study Transplantation 2000 69 2173 2178 10852619 10.1097/00007890-200005270-00037
Shipman PM Sabath DE Fischer AH Comber PG Sullivan KE Tan M Prystowsky MB Cyclin mRNA and protein expression in recombinant interleukin 2-stimulated cloned murine T lymphocytes J Cell Biochem 1988 38 189 198 2906640 10.1002/jcb.240380306
Nourse J Firpo E Flanagan WM Coats S Poyak K Lee MH Massague J Crabtree GR Roberts JM Interleukin-2-mediated elimination of the p27Kip1 cyclin-dependent kinase inhibitor prevented by rapamycin Nature 1994 372 570 573 7990932 10.1038/372570a0
Fan W Richter G Cereseto A Beadling C Smith KA Cytokine response gene 6 induces p21WAF1/CIP1 and regulates both cell growth and arrest Oncogene 1999 18 6573 6582 10597261 10.1038/sj.onc.1203054
Mohapatra S Agrawal D Pledger WJ p27Kip1 Regulates T cell proliferation J Biol Chem 2001 276 21976 21973 11297537 10.1074/jbc.M009788200
Balomenos D Martin-Caballero J Garcia MI Prieto I Flores JM Serrano M Martinez AC The cell cycle inhibitor p21WAF1/CIP1 controls T-cell proliferation and sex-linked lupus development Nat Med 2000 6 171 176 10655105 10.1038/72272
Santiago-Raber ML Lawson BR Dummer W Barnhouse M Koundouris S Wilson CB Kono DH Theofilopoulos AN Role of Cyclin Kinase Inhibitor p21WAF1/CIP1 in Systemic Autoimmunity J Immunol 2001 167 4067 4074 11564828
Brian R Lawson H Kono D Theofilopoulos AN Deletion of p21WAF1/CIP1 (WAF-1/Cip1) Does Not Induce Systemic Autoimmunity in Female BXSB Mice Journal of Immunology 2002 168 5928 5932
Jackson SK DeLoose A Gilbert KM Induction of anergy in Th1 cells associated with increased levels of cyclin-dependent kinase inhibitors p21WAF1/CIP1 Cip1 and p27Kip1 J Immunol 2001 166 952 958 11145672
Flores-Rozas H Kelman Z Dean FB Pan Z Harper JW Elledge SJ O'Donnell M Hurwitz J Cdk-interacting protein 1 directly binds with proliferating cell nuclear antigen and inhibits DNA replication catalyzed by the DNA polymerase delta holoenzyme Proc Natl Acad Sci USA 1994 91 8655 8659 7915843
Waga S Hannon GJ Beach D Stillman E The p21 inhibitor of cyclin-dependent kinases controls DNA replication by interaction with PCNA Nature 1994 369 574 578 7911228 10.1038/369574a0
Luo Y Hurwitz J Massague J Cell-cycle inhibition by independent CDK and PCNA binding domains in p21Cip1 Nature 1995 375 159 161 7753174 10.1038/375159a0
Sherr CJ Roberts JM Inhibition of mammalian G1 cyclin dependent kinases Gene Dev 1995 9 1149 1165 7758941
Mobley SR Liu TJ Hudson JM Clayman GL In vitro growth suppression by adenoviral transduction of p21WAF1/CIP1 and p16 in squamous cell carcinoma of the head and neck: a research model for combination gene therapy Arch Otolaryngol Head Neck Surg 1998 124 88 92 9440786
Hall M Bates S Peters G Evidence for different modes of action of cyclin-dependent kinase inhibitor p15 and p16 bind to kinases, p21WAF1/CIP1 and p27 bind to cyclins Oncogene 1995 11 1581 1588 7478582
Hengst I Reed SI Translational control of p27 accumulation during the cell cycle Science 1996 271 1861 1864 8596954
Toycshima VL Hunter T p27, a novel inhibitor of G1 cyclin-dependent kinase activity, is related to p21WAF1/CIP1 Cell 1994 78 67 74 8033213 10.1016/0092-8674(94)90573-8
Grenz A Schenk M Zipfel A Viebahn R TNF-alpha and its receptors mediate graft rejection and loss after liver transplantation Clin Chem Lab Med 2000 38 1183 1185 11156356 10.1515/CCLM.2000.184
Kaden J Priesterjahn R Increasing urinary IL-6 levels announce kidney graft rejection Transpl Int 2000 13 S34 41 11111958 10.1007/s001470050271
Hamano K Ito H Shirasawa B Gohra H Katoh T Fujimura Y Esato K Correlations among expression of intercellular adhesion molecule 1, cellular infiltration, and coronary arteriosclerosis during chronic rejection using the rat heart transplantation model Eur Surg Res 1999 30 235 242 10.1159/000008582
Herweijer H Wolff JA Progress and prospects: naked DNA gene transfer and therapy Gene Ther 2003 10 453 458 12621449 10.1038/sj.gt.3301983
Chun S Mdaheshia M Lee S Rouse BT Immune modulation by IL-10 gene transfer via viral vector and plasmid DNA: implication for gene therapy Cell Immunol 1999 194 194 204 10383822 10.1006/cimm.1999.1500
Budker V Zhang Danko I Williams P Wolff J The efficient expression of intravascularly delivered DNA in rat muscle Gene Ther 1998 5 272 276 9578848 10.1038/sj.gt.3300572
Davis HL Demeneix BA Quantin B Coulombe J Whalen RG Plasmid DNA is superior to viral vectors for direct gene transfer into adult mouse skeletal muscle Hum Gene Ther 1993 4 733 740 8186288
Prud'homme GJ Gene therapy of autoimmune diseases with vectors encoding regulatory cytokines or inflammatory cytokine inhibitors J Gene Med 2000 2 222 232 10953913 10.1002/1521-2254(200007/08)2:4<222::AID-JGM117>3.0.CO;2-P
Rauh G Pieczek A Irwin W Schainfeld R Isner JM In vivo analysis of intramuscular gene transfer in human subjects studied by on-line ultrasound imaging Hum Gene Ther 2001 12 1543 1519 11506696 10.1089/10430340152480267
Khanna AK Plummer M Nilakantan V Pieper G Recombinant p21 protein inhibits lymphocyte activation and transcription factors J Immunol 2005 174 7610 7617 15944261
|
16176581
|
PMC1242230
|
CC BY
|
2021-01-04 16:03:31
|
no
|
BMC Immunol. 2005 Sep 21; 6:22
|
utf-8
|
BMC Immunol
| 2,005 |
10.1186/1471-2172-6-22
|
oa_comm
|
==== Front
BMC Mol BiolBMC Molecular Biology1471-2199BioMed Central London 1471-2199-6-191615690210.1186/1471-2199-6-19Research ArticleThe radioresistance kinase TLK1B protects the cells by promoting repair of double strand breaks Sunavala-Dossabhoy Gulshan [email protected] Sri Kripa [email protected] Siddhartha [email protected] Sam [email protected] Benedetti Arrigo [email protected] Department of Biochemistry and Molecular Biology and the Feist-Weiller Cancer Center, Louisiana State University Health Sciences Center. 1501 Kings Highway, Shreveport, LA 71130-3932, USA2005 12 9 2005 6 19 19 15 3 2005 12 9 2005 Copyright © 2005 Sunavala-Dossabhoy et al; licensee BioMed Central Ltd.2005Sunavala-Dossabhoy et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The mammalian protein kinase TLK1 is a homologue of Tousled, a gene involved in flower development in Arabidopsis thaliana. The function of TLK1 is not well known, although knockout of the gene in Drosophila or expression of a dominant negative mutant in mouse cells causes loss of nuclear divisions and missegregation of chromosomes probably, due to alterations in chromatin remodeling capacity. Overexpression of TLK1B, a spliced variant of the TLK1 mRNA, in a model mouse cell line increases it's resistance to ionizing radiation (IR) or the radiomimetic drug doxorubicin, also likely due to changes in chromatin remodeling. TLK1B is translationally regulated by the availability of the translation factor eIF4E, and its synthesis is activated by IR. The reason for this mechanism of regulation is likely to provide a rapid means of promoting repair of DSBs. TLK1B specifically phosphorylates histone H3 and Asf1, likely resulting in changes in chromatin structure, particularly at double strand breaks (DSB) sites.
Results
In this work, we provide several lines of evidence that TLK1B protects the cells from IR by facilitating the repair of DSBs. First, the pattern of phosphorylation and dephosphorylation of H2AX and H3 indicated that cells overexpressing TLK1B return to pre-IR steady state much more rapidly than controls. Second, the repair of episomes damaged with DSBs was much more rapid in cells overexpressing TLK1B. This was also true for repair of genomic damage. Lastly, we demonstrate with an in vitro repair system that the addition of recombinant TLK1B promotes repair of a linearized plasmid incubated with nuclear extract. In addition, TLK1B in this in vitro system promotes the assembly of chromatin as shown by the formation of more highly supercoiled topomers of the plasmid.
Conclusion
In this work, we provide evidence that TLK1B promotes the repair of DSBs, likely as a consequence of a change in chromatin remodeling capacity that must precede the assembly of repair complexes at the sites of damage.
==== Body
Background
The Tousled gene of Arabidopsis thaliana encodes a protein kinase which, when mutated, results in abnormal flower development characterized by a stochastic loss of floral meristem and organs [1]. Two mammalian Tousled-like kinases (TLK1 and TLK2) were cloned by Sillje et al., 1999 [2] during a PCR-based search for human kinases, who also reported that the activity of these kinases is maximal in S phase, and more recently, these kinases were reported to be targets of checkpoint kinases, ATM and Chk1 [3]. Specifically, it was reported that TLK1 is inhibited by Chk1 by direct phosphorylation at S695. These findings identify a functional cooperation between ATM and Chk1 in propagation of a checkpoint response to DNA damage and suggest that through transient inhibition of TLK1 the ATM-CHK1-TLK pathway may regulate processes involved in chromatin assembly [4]. Indeed, in AT cells (cells deficient in ATM protein) TLK1 was not inhibited after genotoxic stress [4]. Since ATM and Chk1 are involved in the DNA damage checkpoint upon radiation, this suggests that TLKs may be involved in some aspect of genome surveillance, particularly chromatin remodeling concurrent with DNA repair (see below). The function of TLK1 is not well known, although knockout of the gene in Drosophila or expression of a dominant negative mutant in mouse cells causes loss of nuclear divisions and missegregation of chromosomes [5,6] likely through changes in chromatin remodeling capacity. The importance of TLK1 in chromosome segregation was recently confirmed in C. elegans embryos [7].
We recently cloned a cDNA encoding a mammalian Tousled-like kinase (TLK1B) through a very different scheme than the one used by Sillje et al. [2], based upon polysomal redistribution of weakly translated transcripts that become preferentially recruited upon overexpression of eIF4E [8]. Indeed, the human TLK1B mRNA (a splice variant of the TLK1 mRNA cloned by Sillje et al.) contains a 5'UTR 1088-nt-long with two upstream AUG codons, which was found to be very inhibitory for translation [8,9]. The inhibition of translation could be relieved by either overexpressing eIF4E, or by deleting a large section of the 5'UTR [8]. We subsequently discovered that TLK1B overexpression protects the cells from the genotoxic effects of ionizing radiation (IR) or the radiomimetic drug, doxorubicin. TLK1B probably exerts these effects by phosphorylating histone H3 [8,10] and the histone H3 chaperone Asf1 [10,11], and thereby promoting chromatin remodeling concurrent with repair of DNA damage. Interestingly, synthesis of TLK1B is induced at the translation level by the presence of double strand breaks [DSBs; [12]].
The discovery that TLK1B is a kinase that phosphorylates histone H3 came first in a series of experiments aimed at identifying the potential substrates of TLK1B. We carried out kinase assays with recombinant GST-TLK1B and various typical substrates, only a few of which were phosphorylated efficiently. However, we found that TLK1B phosphorylated very well histone H3 at S10 but not the other core histones or H1 [8]. We subsequently confirmed that the MM3MG cells overexpressing TLK1B had a higher constitutive level of H3 phosphorylation in asynchronous cells [8]. We could also show genetically that TLK1B is a histone H3 kinase. We showed that inducible expression of TLK1B in a yeast strain carrying a temperature-sensitive allele of the major H3 kinase [Ipl-1; [13]] could rescue growth at the non-permissive temperature [8]. In addition, it restored normal levels of histone H3 phosphorylation [8]. Significantly, expression of TLK1B in yeast increased radioresistance, indicating a conservation of function and substrates. Interestingly, we also found that IR results in a loss of H3 phosphorylation, but the significance of this result was unclear. The dephosphorylation of H3 after IR was reported also by another group [14]. Recently, a possible explanation for this effect was described [3,4]. These authors showed that TLK1 (the larger isoform) is inhibited by γ-radiation. The inhibition is presumably mediated by ATM and Chk1 by direct phosphorylation at S695 [4]. It seemed possible that physiologically the increased TLK1B synthesis following IR can help offset the loss of TLK1 activity resulting from IR and restore appropriate levels of histone H3 phosphorylation.
In this paper we present evidence that overexpression of TLK1B protects the cells from IR by facilitating the repair of DSBs, and that the steady state phosphorylation of H3 and H2AX is restored to pre-IR much more rapidly in TLK1B overexpressing cells.
Results
TLK1B protects cells from IR
We previously published initial evidence that TLK1B protects a normal mammary cell line (MM3MG) from IR [8]. This result is now reproduced in Fig. 1 in a more complete assay with serially diluted cells plated for clonogenic assays. We show that 90% of untransfected MM3MG were killed at 2 Gray (Gy), in contrast to only 55% of those transfected with TLK1B (P = 0.001 by one-tailed t-test). At 4 Gy, very few MM3MG cells survived, in contrast to 4% of cells expressing TLK1B (P = 0.0001). This confirmed that overexpression of TLK1B significantly increased their radioresistance.
Inverse phosphorylation of H2AX and H3 after IR
We previously reported that TLK1B phosphorylates very well histone H3 at S10 but not the other core histones or H1 [8]. Interestingly, we also found that IR results in a loss of H3 phosphorylation, but the significance of this result was unclear. If TLK1 is the major kinase involved in a "chromosomal response" to DNA damage, then inhibition of TLK1 or TLK1B activity by IR through ATM is expected to result in a loss of phosphorylation of histone H3. To probe for the significance of the H3 dephosphorylation, we carried out a time course of recovery from IR and monitored the phosphorylation of H2AX and H3 in the MM3MG cells that overexpress TLK1B. Radiation-induced damage has been shown to result in rapid phosphorylation of histone H2AX at DSB sites [15,16], while BRCA2, a protein that localizes to DSBs, was found to accumulate to sites of condensed chromatin colocalized with phosphorylated H3 [17]. As seen in Fig. 2, phosphorylation of H2AX is, as expected, very rapid after IR in both control and TLK1B cells. In the control, phosphorylation of H2AX remains elevated for at least 8 hr and slightly reduced after 16 hr of recovery. Further, phosphorylation of H3 drops drastically after IR and only recovers after 16 hr in the control. In TLK1B cells, basal phosphorylation of H3 is elevated at t = 0, as previously reported [8] and then decreases after IR (note that the Chk1 phosphorylation site is conserved in TLK1B). However, between 4 hr and 8 hr, the phosphorylation of H3 recovers completely. At the same time, phosphorylation of H2AX drops precipitously. This suggests that repair of DNA is much more rapid in TLK1B cells, if the phosphorylation state of histones H3 and H2AX and presumed remodeling of chromatin can be taken as an indicator of DSB repair.
Blocking ATM activity with wortmannin prevents the loss of H3 phosphorylation after IR
If TLK1 and TLK1B are kinases that are largely responsible for the phosphorylation of H3 in unstressed conditions (excluding mitosis), and if activation of ATM following IR is responsible for inhibiting TLK1/B [as previously published; [3,4]], then inhibition of ATM with wortmannin should prevent in large part the dephosphorylation of H3 following IR. Fig. 3 shows that this is indeed the case, since cells pretreated with wortmannin and subjected to IR showed only a very modest decrease in H3 phosphorylation throughout the time course of IR and recovery. Contrary to the expectation, phosphorylation of H2AX was not inhibited by wortmannin in either control or cells overexpressing TLK1B. This was somewhat surprising since ATM is believed to directly phosphorylate this histone after IR [18]. However, very recently, it was demonstrated that ATR, which requires much higher concentrations of wortmannin for inhibition (0.1 mM), also phosphorylates H2AX after radiation, although with somewhat delayed kinetics [16].
Episomal vectors as reporters for DSBs
To overexpress TLK1B, we used an episomal vector called BK-Shuttle [19]. An important advantage is that these plasmids can be easily rescued from mammalian cells by the Hirt's supernatant protocol [19]. Most stable cell lines carrying these vectors typically have hundreds of copies of the episomes, which have the typical structure of mini-chromosomes and thus behave like genomic DNA [20,21]. We previously showed that the extraction of episomes is a very simple and efficient protocol that yields very consistent amounts of plasmids that are linear with respect to cell numbers [19]
In Fig. 4, we show an example of how these episomes can be used to study protection from IR-mediated DSBs. We have found that IR causes significant damage to the episomes, as shown by a large proportion of linearized molecules due to DSBs. Plasmids isolated from untreated cells migrate as a predominant supercoiled fast-migrating form and a relaxed circular form. Plasmids isolated from irradiated cells show a dose-dependent loss of the supercoiled form due to DSBs, which convert the plasmid to a linearized form (and some nicked circular). The conversion of closed (circular and supercoiled forms) to linearized plasmids can be exploited to monitor the activity of TLK1B. Since the episomes are assembled in typical chromatin, the function of TLK1B in repair (e.g., chromatin remodeling to produce ends suitable for ligation) would mirror its function on genomic DNA. Note that the intensities of the bands were quantified with an imaging program (ImageQuant), and that the sum of the bands (supercoiled, linear, or circular) was a constant, indicating very consistent yields of episomes.
Episomal vectors as reporters for DSB repair
The episomal vectors can be used as reporters for repair of DSBs and provide a useful system to test if this is the likely mechanism of radioresistance. Cells were irradiated and allowed to recover during an 8 hr time course (repair time), and plasmids were isolated for analysis by gel electrophoresis. In this experiment, however, the episomes were extracted from cells by alkaline lysis (essentially the same protocol used to extract plasmids from bacteria). Plasmids that are linearized with a DSB or plasmids with nicks are not recovered from alkaline lysis because of strand separation. In contrast, supercoiled/covalently closed plasmids are recovered from alkali treatment. Fig. 5 shows that IR causes an immediate and significant loss of supercoiled plasmids in control MM3MG cells containing the empty vector. Religation of the linear plasmids and formation of the supercoiled molecules does not occur until 4 to 8 hr of recovery from IR, and then, the recovery is only partial. In contrast, IR causes a similar loss of supercoiled molecules at t = 0 in MM3MG cells expressing TLK1B, but the recovery is very fast. At 2 hr (R2) the recovery of supercoiled molecules is almost complete and it is fully restored by 8 hr (R8). The recovery of the supercoiled forms is almost certainly due to the repair of pre-existing, damaged plasmids and not due to de novo synthesis. First, the repair time was too short to account for a large fraction of newly synthesized plasmids, and second, IR results in a dramatic arrest in DNA replication in normal cells [22], making it highly unlikely that any newly replicated plasmid was achieved. This experiment was repeated with identical results, indicating that the loss of supercoiled species after IR can be easily detected by alkaline lysis with great reproducibility.
Enhanced repair of genomic damage in TLK1B cells
To assess genomic repair, a modified terminal deoxytransferase (TdT) fill-in reaction was adopted. Briefly, the cells were irradiated (or not) to create genomic breaks, which were then processed with TdT and biotinylated dNTPs according to the manufacturer's protocol (see Methods). Macroscopic deposit of diaminobenzidine (DAB) at breakage sites was used to determine the extent of genomic damage and repair during a time course of recovery from IR (Fig. 6 and Table 1). No DAB deposits were visible in non-irradiated cells. In irradiated cells, 9–13 spots per cell were visible immediately after IR in both control and cells expressing TLK1B. Already at two hr of recovery from IR, in TLK1B cells the number of spots decreased by more than 50%, in contrast to control cells in which there was no appreciable recovery.
Rapid repair and chromatin assembly in vitro
The direct way to probe the function of TLK1B in repair of a DSB is in vitro. Repair assays were carried out as described in Methods, based on a system that demonstrated a synergism between CAF-1 and Asf1 in a repair-coupled nucleosome assembly [23]. If TLK1B increases the activity of Asf1 in vitro, then reactions supplemented with TLK1B should show a greater proportion of supercoiled topomers in addition to more ligated (closed) plasmid. Briefly, repair/nucleosome assembly was carried out on Bluescript plasmid that is linearized with EcoRI. We wished to monitor simultaneously: 1) Processing/ligation of the ends; and 2) superhelicity of the plasmid by the formation of nucleosomes on the template. Reactions contained nuclear extract, an energy mix, and additional purified TLK1B. The plasmid was then re-extracted and analyzed by electrophoresis in agarose gels. In Fig. 7A, linearized plasmids were incubated with cell extract +/- TLK1B, and repair was monitored during a time course. Note that the addition of TLK1B greatly stimulated the speed of the reaction in the formation of dimer molecules and the ligated/supercoiled forms of the plasmids, which already appear at 20 min instead of 40–60 min for extract without TLK1B. For shorter time points of incubation we used a more sensitive and quantitative method to monitor repair of the cut plasmids, i.e., a bacterial transformation assay. Linear plasmids do not transform bacteria, whereas repaired, covalently closed plasmids do. After 10 min of incubation with cell extract, in a typical experiment, we obtained 31 colonies vs. 260 for extract supplemented with exogenous TLK1B. Therefore, it appears that TLK1B improves the accessibility of the repair and ligation machinery to the free ends of the DNA or that it stimulates the deposition of histones [by Asf1; [11]] to assemble chromatin. This lends strong support to our hypothesis that the principal mechanism of increased radioresistance is through more efficient repair of DSBs likely linked to chromatin remodeling.
In Fig. 7B, we show an assay of supercoiling activity, which is a measure of chromatin assembly. In this assay, the plasmid Bluescript (nearly 100% supercoiled, input) is used as a template for the deposition of core histones in the presence of nuclear extract and an energy mix. In the absence of histones, the extract causes the bacterially supercoiled form to convert to mostly relaxed forms due to endogenous topoisomerases (data not shown). However, after incubation in the presence of histones, the plasmid migrates as a series of discrete supercoiled forms due to the formation of nucleosomes, which decrease the linking number by one integer per nucleosome. The addition of recombinant TLK1B stimulated the formation of the more highly supercoiled forms, particularly the form that runs like bacterially supercoiled plasmid.
Discussion
In this work, we have provided four lines of evidence that the Tousled kinase, TLK1B, protects the cells from IR by facilitating the repair of DSBs. First, the pattern of phosphorylation/dephosphorylation of H2AX and H3 indicated that cells overexpressing TLK1B return to pre-IR phosphorylation state much more rapidly than controls. Second, the repair of episomes damaged with DSBs was much more rapid and complete by 8 hr of recovery in cells overexpressing TLK1B. Third, we have found that the repair of genomic breaks occurs more rapidly in cells overexpressing TLK1B, and with kinetics that are similar to those of repair of episomes. Lastly, we demonstrated with an in vitro repair system that the addition of recombinant TLK1B promotes repair of a linearized plasmid incubated with nuclear extract. Consistent with the results published by Groth [4] and Kodym [24] we found that TLK1 activity is inhibited by IR, as shown by loss of phosphorylation of histone H3, which is one of the best substrates of TLK1. Nonetheless, when TLK1B is overexpressed (about 6-fold in our stably transfected MM3MG cells), the recovery of H3 phosphorylation was quite rapid (about 4 hr) probably because of mass action due to higher levels of the kinase.
The role of TLK1B in radioresistance is particularly intriguing based on the recent findings that Asf1 is also a specific TLK1 substrate [11] known to participate with Rad53 in chromatin remodeling at sites of DSBs [25]. Furthermore, the importance of histone H3 kinases cannot be overstated. Phosphorylation of H3 at S10 is becoming one of the most intensely studied aspects of chromatin remodeling, both during segregation of chromosomes at mitosis, and in aspects of transcription [13,26-28]. Therefore, studies of TLK1B and the family of Tousled kinases are bound to become the center of much attention. Elevated phosphorylation of H3 has also been reported in several lines of oncogenically transformed fibroblasts [29], although the underlying mechanism is unknown. We have found that elevated expression of TLK1B did not oncogenically transform MM3MG cells, but the cells became highly resistant to IR or doxorubicin [8]. We have preliminary results that demonstrate that TLK1B is elevated in some breast carcinomas and it is possible that this may result in a disease refractory to treatment [30]. We currently favor a mechanism by which TLK1B protects the cells from DSBs, by promoting repair-coupled chromatin remodeling which depends on the phosphorylation of histone H3 and Asf1.
Radiation-induced damage has been shown to result in rapid phosphorylation of histone H2AX. Phosphorylated foci of H2AX co-localize with DNA repair and signaling proteins, and H2AX has been demonstrated to be involved in their recruitment to sites of DSBs [31]. Knockout of H2AX in mice resulted in defects in concentration of repair proteins (53BP1, BRCA1, NBS1) at the sites of DNA damage, and H2AX-deficient cells are IR sensitive demonstrating the requirement of H2AX in DNA repair [15]. The importance of radiation-induced phosphorylation of H2AX in repair of DSBs has been shown by the fact that H2AX deficiency results in genomic instability [32]. In yeast, H2AX is not involved in activation of the S-phase checkpoint by DSBs, but rather in the efficiency of DNA repair [33]. The importance of histone H3 phosphorylation in DNA damage has not been investigated as well as that of H2AX, but the overall effect of these modifications is likely to be chromatin remodeling and recruitment of repair proteins. Recently, H2AX was found to recruit the chromatin remodeling protein INO80 [34].
Our current model is that in normal cells TLK1 (the constitutively expressed larger isoform) performs the normal functions of this kinase, which may have a role in chromatin remodeling and genome surveillance in unstressed conditions. Following DNA damage by IR or doxorubicin, synthesis of TLK1B is induced through a translational control mechanism [12]. TLK1B can then facilitate repair of DNA damage. This would greatly accelerate the response of those cells to DNA damage and the efficiency with which repair is implemented, significantly increasing their resistance to IR.
An alternative that we considered for a role of TLK1B in radioprotection is that TLK1B functions in a signaling pathway that protects cells from undergoing apoptosis. There are two compelling reasons why this is not likely (at least not directly). First, overexpression of TLK1B did not change the transcriptome in MM3MG cells (microarray analysis, data not shown), indicating that its protective effect is post-transcriptional. There were no changes in the expression of pro- and anti-apoptotic genes, or cell cycle regulators. Second, expression of TLK1B conferred protection against IR even in yeast. Whereas proteins involved in sensing and repairing DNA damage are conserved between mammals and yeast, prototypical proteins involved in the apoptotic and antiapoptotic pathways are not found in yeast.
Conclusion
Studies of the Tousled kinases are only now beginning to shed light upon their function, despite the early discovery of a role in flower and leaf morphology inferred by mutations in plants. In this work, we have provided four lines of evidence that the Tousled kinase, TLK1B, protects the cells from IR by facilitating the repair of DSBs. First, the pattern of phosphorylation/dephosphorylation of H2AX and H3 indicated that cells overexpressing TLK1B return to pre-IR phosphorylation state much more rapidly than controls. Second, the repair of episomes damaged with DSBs was much more rapid and complete by 8 hr of recovery in cells overexpressing TLK1B. Third, we have found that the repair of genomic breaks occurs more rapidly in cells overexpressing TLK1B, and with kinetics that are similar to those of repair of episomes. Lastly, we demonstrated with an in vitro repair system that the addition of recombinant TLK1B promotes repair of a linearized plasmid incubated with nuclear extract. Therefore, it appears that TLK1 and TLK1B have a role in genome surveillance, particularly upon genotoxic stress, which induces the expression of TLK1B.
Methods
Cell lines and tissue culture
Normal breast epithelial cells, MM3MG, transfected or not with TLK1B were cultured as described in Li et al. [8].
Radiation experiments
Control MM3MG cells and the cells overexpressing TLK1B [MM3MG-TLK1B; [8]] were harvested with PBS/EDTA and adjusted to 10,000 cells/tube in DMEM/10% FCS. Cells were irradiated in the Radiation department at LSUHSC with Elekta Precise linear accelerator at 6 MV. For each radiation dose levels (0 to 8 Gy), aliquots of serially diluted cells (100–5000) were plated on 6-well plates in triplicate. After a period of 10 days of incubation, the wells were rinsed with PBS and stained with crystal violet, and the colonies counted. The experiment was repeated thrice, and the results were expressed as the fraction of surviving cells compared to the number of colonies formed in the non-irradiated samples (plating efficiency).
Western blots
The anti-histone H3 phosphorylated at Ser-10 and anti-histone H2AX phosphorylated at Ser-139 were from Upstate Cell Signaling (Lake Placid, NY). For Western blots, 30 μg of protein of each sample was separated on a 15% SDS/PAGE gel. The proteins were transferred to Immobilon-P membranes (Millipore, Bedford, MA) and incubated overnight with primary antiserum and for 1 hr with secondary antisera (1:1000 dil.). Finally, the membranes were washed three times and developed with Opti-4CN reagent (Bio-Rad, Hercules, CA).
Extraction of episomes
Episomes were isolated from 2 × 107 cells (90% confluent flasks) stably transfected with empty vector (BK-Shuttle) or vector carrying TLK1B. Two methods were used to extract the episomes: either the standard Hirt's supernatant protocol [35], or alkaline lysis. Briefly, the cells were resuspended in 0.1 ml TE, lysed at room temperature with solution 1 (0.2 ml of 0.2 M NaOH, 1% SDS), which was then neutralized with solution 2 (0.15 ml of 3 M K-Acetate/glacial acetic acid). After a brief centrifugation at 10,000 × g to remove insoluble material (including genomic DNA), remaining nucleic acids were extracted with Phenol/Chloroform (1:1), and precipitated with cold EtOH. The episomes were analyzed by gel electrophoresis on 1% agarose/TAE, and stained with EtBr.
Assay of genomic repair
To assess the repair of DNA damage in vivo, the modified TUNEL assay (terminal deoxynucleotidyltransferase-mediated dUTP nick end labeling) was applied. MM3MG or TLK1B cells were grown to 50% confluence on tissue culture slides prior to exposing them to ionizing irradiation (20 Gy). After radiation, the cells were allowed to recover for varying times (0, 2, 8 hr). Subsequently, cells were fixed in 4% formalin/PBS and permeabilized in 0.2% Triton-X100/PBS. For labeling DNA breaks in situ, the DeadEnd Colorimetric TUNEL System (Promega, Madison, WI) was used according to the manufacturer's protocol. The biotinylated nucleotides incorporated at 3'OH ends were reacted with horseradish peroxidase labeled-streptavidin, which was then detected by diaminobenzidine (DAB). The nuclear DNA-labeled sites (brown spots) within each cell were counted under a light microscope (40 × magnification), and the average (± SD) number of DNA breaks per cell was calculated. At least 10 cells per dose were counted.
Assay of DSB repair in vitro
Nuclear extract from MM3MG cells was prepared as described in [36]. Repair assays were carried out as described by Mello [23]. Briefly, repair/nucloesome assembly was carried out on 0.1 μg of Bluescript plasmid (per reaction) that was linearized with EcoRI. We monitored simultaneously ligation of the ends and superhelicity of the plasmid by the formation of nucleosomes on the template. Reactions contain 20 μg of nuclear extract, 5 mM MgCl2, 40 mM Hepes, pH7.8, 0.5 mM DTT, 4 mM ATP, 20 μM dNTPs, 4 mM phosphocreatine, 2 u of creatine phosphokinase, and additional recombinant TLK1B. After incubation at 37°C for the indicated amount of time, the reactions were deproteinized with phenol and the plasmid was re-precipitated with cold EtOH.
Assay of chromatin assembly
Nucloesomes assembly was carried out on 2 μg of Bluescript plasmid. Reactions contained 15 μg of MM3MG cell extract (which already contains sufficient amounts of topoisomerases), 5 mM MgCl2, 40 mM Hepes, pH 7.8, 0.5 mM DTT, 4 mM ATP, 20 μM dNTPs, 4 mM phosphocreatine, 2 u of creatine phosphokinase, and additional purified proteins (200 ng TLK1B and 2 μg supplemental HeLa histones). The reactions were incubated at 37°C for 0.5 hr. The plasmid was re- extracted with GeneClean III kit (Bio 101, Vista, CA), separated on an agarose gel and subsequently stained with EtBr.
Authors' contributions
GSD prepared figures 1, 2, 3, 4. SKB prepared figure 7A; SS and SN prepared figure 5; SS prepared figure 6 and figure 7B. ADB wrote the paper.
Acknowledgements
This work was supported by an LSUHC-S Office of Research Institutional award. GSD is supported by funds from the Louisiana Gene Therapy Consortium, and SKB and SS are supported by the Feist-Weiller Cancer Center.
Figures and Tables
Figure 1 Pattern of γ-radiation sensitivity by clonogenic assays. 104 untransfected MM3MG and cells overexpressing TLK1B were irradiated with the indicated doses and plated in triplicates of 100, 500, 1000, or 5000 cells in multiwell plates. The colonies were counted 10 days later. The average from 2 independent experiments is shown. S.F. = surviving fraction.
Figure 2 Phosphorylation of H3 and H2AX after irradiation. MM3MG and MM3MG over-expressing TLK1B cells were grown to 80% confluence prior to gamma-radiation (10 Gy). Cells were harvested at different times (0 h, 30 min, 1 h, 2 h, 4 h, 8 h and 16 h) after irradiation and lysed in RIPA buffer. Equal amount of protein of each sample was loaded on a 15% SDS-PAGE gel and electrophoresed. Blots were probed with phospho-Histone H3 (Ser-10) antibody, or phospho-Histone H2AX (Ser-139) antibody (Upstate Cell Signaling). Equal loading of proteins was confirmed by staining the blots with Ponceau S prior to processing. These blots are representative of two separate experiments.
Figure 3 Phosphorylation of H3 and H2AX after irradiation in the presence of wortmannin. The experiment was carried out as detailed in the legend to Fig. 2, but the cells were pre-treated with 30 μM wortmannin to inhibit ATM.
Figure 4 Analysis of episomes. 2 × 107 cells transformed with BK-Shuttle or TLK1B were irradiated with the indicated dose of γ-radiation. The cells were returned to the incubator and the plasmids were isolated 1 hr later by the Hirt's protocol and separated on a 1% agarose/TAE gel. The mobility of the forms (circular, linear, and supercoiled) is indicated. The structure of the BK-Shuttle episomal vector is shown on the right. The bands were quantified with ImageQuant vs. 5 (Molecular Dynamics).
Figure 5 Analysis of episomes during a time course of recovery from IR. 2 × 107 cells transformed with BK-Shuttle or TLK1B were irradiated (or not, C) with 20 Gy of γ-radiation. The cells were returned to the incubator and the plasmids were isolated immediately after radiation (IR), or after the indicated times of recovery, 2 to 8 hr (R2-R8). The episomes were recovered by alkaline lysis, which removes plasmids with DSBs by strand separation followed by the rapid renaturation. This is because it is not always easy to separate on agarose gels the linearized from the supercoiled form of the plasmid (these plasmids are 11–14 kb in size, without and with insert). The bands were quantified with ImageQuant. The intensities of the bands relative to control are indicated underneath each lane. Size markers (Novagen 1 kb ladder) are shown in the first lane.
Figure 6 Assay of repair of genomic damage. Examples of non-irradiated cells (A, D) or cells irradiated and allowed to recover for 0 hr (B, E) or 2 hr (C, F) are shown. MM3MG cells are shown in A-C; TLK1B cells are shown in D-F. An arrow points to one of the DAB spots.
Figure 7 A. Ligation reactions and supercoiling. Reactions were prepared as described Experimental Procedures. The position of linearized plasmid (lane 1, input), bacterial supercoiled (lane 2) and dimers is indicated. The various topomeric forms of supercoiled plasmid (the result of deposition of nucleosomes on the template) are not resolved well in this gel without chloroquine, and appear as a slight smear (lanes 6, 9, 10). B. Assay of chromatin assembly. Nucloesomes assembly was carried out on 2 μg of Bluescript plasmid as described in Methods. Each band corresponds to the addition of one nucleosome, which decreases the linking number. These gels are representative of two different experiments.
Table 1 RECOVERY TIME (hr) MM3MG (number of breaks) TLK1B (number of breaks)
0 13 +/- 0.35 11 +/- 2.83
2 11 +/- 0.64 5 +/- 1.06
8 7 +/- 1.91 4 +/- 1.27
* Non irradiated cells had no convincing labeled spots.
==== Refs
Roe JL Nemhauser JL Zambryski PC TOUSLED participates in apical tissue formation during gynoecium development in Arabidopsis Plant Cell 1997 9 335 353 9090879 10.1105/tpc.9.3.335
Sillje HHW Takahashi K Tanaka K Van Houwe G Nigg EA Mammalian homologues of the plant Tousled gene code for cell-cycle-regulated kinases with maximal activities linked to ongoing DNA replication EMBO J 1999 18 5691 5702 10523312 10.1093/emboj/18.20.5691
Krause DR Jonnalagadda JC Gatei MH Sillje HH Zhou BB Nigg EA Khanna K Supppression of Tousled-like kinase activity after DNA damage or replication block requires ATM, NBS1 and Chk1 Oncogene 2003 22 5927 5937 12955071 10.1038/sj.onc.1206691
Groth A Lukas J Nigg EA Sillje HHW Wernstedt C Bartek J Hansen K Human Tousled like kinases are targeted by an ATM- and Chk1- dependent DNA damage checkpoint EMBO J 2003 22 1676 1687 12660173 10.1093/emboj/cdg151
Carrera P Moshkin YM Gronke S Sillje HH Nigg EA Jackle H Karch F Tousled-like kinase functions with the chromatin assembly pathway regulating nuclear divisions Genes Dev 2003 17 2578 2590 14561777 10.1101/gad.276703
Sunavala-Dossabhoy G Li Y Williams B De Benedetti A A dominant negative mutant of TLK1 causes chromosome missegregation and aneuploidy in normal breast epithelial cells BMC Cell Biol 2003 4 16 14583098 10.1186/1471-2121-4-16
Han Z Riefler GM Saam JR Mango SE Schumacher JM The C. elegans Tousled-like kinase contributes to chromosome segregation as a substrate and regulator of the Aurora B kinase Curr Biol 2005 15 894 904 15916946 10.1016/j.cub.2005.04.019
Li Y DeFatta RJ Sunavala G Anthony C De Benedetti A A translationally regulated Tousled kinase phosphorylates histone H3 and confers radioresistance when overexpressed Oncogene 2001 20 726 738 11314006 10.1038/sj.onc.1204147
De Benedetti A Graff JR eIF-4E expression and its role in malignancies and metastases Oncogene 2004 23 3189 3199 15094768 10.1038/sj.onc.1207545
Ehsan H Reichheld JP Durfee T Roe JL TOUSLED kinase activity oscillates during the cell cycle and interacts with chromatin regulators Plant Physiol 2004 134 1488 1499 15047893 10.1104/pp.103.038117
Sillje HH Nigg EA Identification of human Asf1 chromatin assembly factors as substrates of Tousled-like kinases Curr Biol 2001 11 1068 1073 11470414 10.1016/S0960-9822(01)00298-6
Sunavala-Dossabhoy G Fowler M De Benedetti A Translation of the radioresistance kinase TLK1B is induced by gamma-irradiation through the activation of mTOR and phosphorylation of 4E-BP1 BMC Mol Biol 2004 5 1 15070431 10.1186/1471-2199-5-1
Hsu JY Sun ZW Li X Reuben M Tatchell K Douglas K Bishop DK Grushcow JM Brame CJ Caldwell JA Hunt DF Lin R Smith MM Allis CD Mitotic Phosphorylation of Histone H3 Is Governed by Ipl1/aurora Kinase and Glc7/PP1 Phosphatase in Budding Yeast and Nematodes Cell 2000 102 279 291 10975519 10.1016/S0092-8674(00)00034-9
Guo CY Mizzen C Wang Y Larner JM Histone H1 and H3 dephosphorylation are differentially regulated by radiation-induced signal transduction pathways Cancer Res 2000 60 5667 5672 11059758
Fernandez-Capetillo O Chen HT Celeste A Ward I Romanienko PJ Morales JC Naka K Xia Z Camerini-Otero RD Motoyama N Carpenter PB Bonner WM Chen J Nussenweig A DNA damage-induced G2-M checkpoint activation by histone H2AX and 53BP1 Nat Cell Biol 2002 4 993 997 12447390 10.1038/ncb884
Wang H Wang M Wang H Bocker W Iliakis G Complex H2AX phosphorylation patterns by multiple kinases including ATM and DNA-PK in human cells exposed to ionizing radiation and treated with kinase inhibitors J Cell Physiol 2005 202 492 502 15389585 10.1002/jcp.20141
Marmorstein LY Kinev AV Chan GK Bochar DA Beniya H Epstein JA Yen TJ Shiekhattar R A human BRCA2 complex containing a structural DNA binding component influences cell cycle progression Cell 2001 104 247 257 11207365 10.1016/S0092-8674(01)00209-4
Jeggo PA Lobrich M Artemis Links ATM to Double Strand Break Rejoining Cell Cycle 2005 4 359 362 15684609
De Benedetti A Rhoads RE A novel BK virus-based episomal vector for the expression of foreign genes in mammalian cells Nucleic Acids Res 1991 19 1925 1931 1709496
Keller W Determination of the number of superhelical turns in simian virus 40 DNA by gel electrophoresis Proc Natl Acad Sci 1975 72 4876 4880 174079
Milavetz B Hyperacetylation and differential deacetylation of histones H4 and H3 define two distinct classes of acetylated SV40 chromosomes early in infection Virology 2004 319 324 336 14980492 10.1016/j.virol.2003.11.010
Sorensen CS Syljuasen RG Falck J Schroeder T Ronnstrand L Khanna KK Zhou BB Bartek J Lukas J Chk1 regulates the S phase checkpoint by coupling the physiological turnover and ionizing radiation-induced accelerated proteolysis of Cdc25A Cancer Cell 2003 3 247 258 12676583 10.1016/S1535-6108(03)00048-5
Mello JA Sillje HH Roche DM Kirschner DB Nigg EA Almouzni G Human Asf1 and CAF-1 interact and synergize in a repair-coupled nucleosome assembly pathway EMBO Rep 2002 3 329 334 11897662 10.1093/embo-reports/kvf068
Kodym R Mayerhofer T Ortmann E Purification and identification of a protein kinase activity modulated by ionizing radiation Biochem Biophys Res Commun 2004 313 97 103 14672703 10.1016/j.bbrc.2003.11.090
Hu F Alcasabas AA Elledge SJ Asf1 links Rad53 to control of chromatin assembly Genes Dev 2001 15 1061 1066 11331602 10.1101/gad.873201
Wei Y Yu L Bowen J Gorovsky MA Allis CD Phosphorylation of histone H3 is required for proper chromosome condensation and segregation Cell 1999 97 99 109 10199406 10.1016/S0092-8674(00)80718-7
De Souza CP Osmani AH Wu LP Spotts JL Osmani SA Mitotic histone H3 phosphorylation by the NIMA kinase in Aspergillus nidulans Cell 2000 102 293 302 10975520 10.1016/S0092-8674(00)00035-0
De La Barre AE Gerson V Gout S Creaven M Allis CD Dimitrov S Core histone N-termini play an essential role in mitotic chromosome condensation EMBO J 2000 19 379 391 10654937 10.1093/emboj/19.3.379
Chadee DN Hendzel MJ Tylipski CP Allis CD Bazett-Jones DP Wright JA Davie JR Increased Ser-10 phosphorylation of histone H3 in mitogen-stimulated and oncogene-transformed mouse fibroblasts J Biol Chem 1999 274 24914 24920 10455166 10.1074/jbc.274.35.24914
Norton KS McClusky D Sen S Yu H Meschonat C De Benedetti A Li BDL TLK1B is Elevated With eIF4E Overexpression in Breast Cancer J Surg Res 2004 116 98 103 14732354 10.1016/j.jss.2003.08.001
Ward IM Minn K Jorda KG Chen J Accumulation of checkpoint protein 53BP1 at DNA breaks involves its binding to phosphorylated histone H2AX J Biol Chem 2003 278 19579 19582 12697768 10.1074/jbc.C300117200
Celeste A Fernandez-Capetillo O Kruhlak MJ Pilch DR Staudt DW Lee A Bonner RF Bonner WM Nussenzweig A H2AX phosphorylation is dispensable for the initial recognition of DNA breaks Nat Cell Biol 2003 5 675 679 12792649 10.1038/ncb1004
Redon C Pilch DR Rogakou EP Orr AH Lowndes NF Bonner WM Yeast histone 2A serine 129 is essential for the efficient repair of checkpoint-blind DNA damage EMBO Rep 2003 4 678 84 12792653 10.1038/sj.embor.embor871
Morrison AJ Highland J Krogan NJ Arbel-Eden A Greenblatt JF Haber JE Shen X INO80 and gamma-H2AX interaction links ATP-dependent chromatin remodeling to DNA damage repair Cell 2004 119 767 775 15607974 10.1016/j.cell.2004.11.037
Hirt B Selective extraction of polyoma DNA from infected mouse cell cultures J Mol Biol 1967 26 365 369 4291934 10.1016/0022-2836(67)90307-5
Gaymes TJ Mufti GJ Rassool FV Myeloid leukemias have increased activity of the nonhomologous end-joining pathway and concomitant DNA misrepair that is dependent on the Ku70/86 heterodimer Cancer Res 2002 62 2791 2797 12019155
|
16156902
|
PMC1242231
|
CC BY
|
2021-01-04 16:22:25
|
no
|
BMC Mol Biol. 2005 Sep 12; 6:19
|
utf-8
|
BMC Mol Biol
| 2,005 |
10.1186/1471-2199-6-19
|
oa_comm
|
==== Front
BMC Med EducBMC Medical Education1472-6920BioMed Central London 1472-6920-5-341616230010.1186/1472-6920-5-34Research ArticleDoctor-patient interaction in Finnish primary health care as perceived by first year medical students Miettola Juhani [email protected]äntyselkä Pekka [email protected] Tuula [email protected] Department of Public Health and General Practice, University of Kuopio, P.O.Box 1627, 70211 Kuopio, Finland2 Unit of General Practice, Kuopio University Hospital, P.O.Box 1777, 70211 Kuopio, Finland2005 15 9 2005 5 34 34 24 2 2005 15 9 2005 Copyright © 2005 Miettola et al; licensee BioMed Central Ltd.2005Miettola et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
In Finland, public health care is the responsibility of primary health care centres, which render a wide range of community level preventive, curative and rehabilitative medical care. Since 1990's, medical studies have involved early familiarization of medical students with general practice from the beginning of the studies, as this pre-clinical familiarisation helps medical students understand patients as human beings, recognise the importance of the doctor-patient relationship and identify practicing general practitioners (GPs) as role models for their professional development. Focused on doctor-patient relationship, we analysed the reports of 2002 first year medical students in the University of Kuopio. The students observed GPs' work during their 2-day visit to primary health care centres.
Methods
We analysed systematically the texts of 127 written reports of 2002, which represents 95.5% of the 133 first year pre-clinical medical students reports. The reports of 2003 (N = 118) and 2004 (N = 130) were used as reference material.
Results
Majority of the students reported GPs as positive role models. Some students reported GPs' poor attitudes, which they, however, regarded as a learning opportunity. Students generally observed a great variety of responsibilities in general practice, and expressed admiration for the skills and abilities required. They appreciated the GPs' interest in patients concerns. GPs' communication styles were found to vary considerably. Students reported some factors disturbing the consultation session, such as the GP staring at the computer screen and other team members entering the room. Working with marginalized groups, the chronically and terminally ill, and dying patients was seen as an area for development in the busy Finnish primary health care centres.
Conclusion
During the analysis, we discovered that medical students' perceptions in this study are in line with the previous findings about the importance of role model (good or bad) in making good doctors. Therefore, medical students' pre-clinical primary health care centre visits may influence their attitudes towards primary health care work and the doctor-patient relationship. We welcome more European studies on the role of early pre-clinical general practice exposure on medical students' primary care specialty choice.
==== Body
Background
In Finland, public health care is the responsibility of primary health care centres which render a wide range of first level preventive, curative and rehabilitative medical care. In order to improve continuity of care, most Finnish municipalities have switched from the traditional primary health-care system to a family doctor system [1,2]. In Finland, medical studies have traditionally involved an initial two-year preclinical period of mainly theoretical courses. However, students now have contact with patients from the beginning of their studies [3].
The first international reports on the subject of doctor-patient relationship training during pre-clinical medical education originate from the 1970s [4]. Presently, medical schools in several countries integrate training in communication skills into the medical curriculum, and provide medical students with the opportunity to meet patients in a real primary health care setting during the early stages of their studies [5,6]. Pre-clinical familiarisation of medical students with general practice considerably improves students' communication skills and understanding of patients' perceptions, helps students understand patients as human beings, and helps them recognise the importance of the doctor-patient relationship [7-9]. There is also much international documentation concerning the importance of role models in making good doctors [10]. However, some international reports claim that early exposure of medical students to family practice faculty or to family practitioners in their own clinics does not influence primary care specialty choice [11].
In Finnish society, there are quite alarming signs of a fall in public appreciation of public primary health care [12]. Authorities have devoted more attention to efforts to restore the traditional high esteem in which the demanding work of general practitioners and their teams in primary health care centres has been held. Medical schools have also acknowledged the problem.
As a continuation of the community oriented medical training program since early 70's at the University of Kuopio, a two-day primary health care centre visit was introduced in 1996 to the Introduction to Medicine Course, which is held in the first year of the medical studies. The purpose of the field visit is to familiarise students with the Finnish health care system and its operation, the everyday work of GPs, and in particular what they do and how they do it in their interaction with patients. Moreover, the purpose is to underline the significance of the interaction between doctor and patient. The orientation programme involves theoretical and practical teaching of communication skills prior to the primary health care centre visit.
In the primary health care centres, students observe the operation of the institution and the work of GPs alone and as a team member, interview one primary care patient and write a report about their findings based on the accompanying checklist (Table 1). In the reports, students also make their comments on the doctor-patient interaction. They observe encounter situations and report their perceptions based on the checklist.
Table 1 Check-list for the health care centre visit (the focus areas of this study with bold font).
Location of the health care centre
Catchment area and population
Number of general and other medical practitioners
Contents of the working day of a general practitioner
working in different sectors (accident and emergency department, outpatient department, maternal, child, school, occupational health care, family planning, home care, wards)
participation in training, meetings and workshops
Attitude of the practitioners towards their work
Attitude of the practitioners towards their patients
Doctor-patient communication
Practitioner as a team member
teamwork
attitude towards colleagues, nurses, physiotherapists and other key persons
Students' personal experiences of the visit
In our study, we focused on 1st year medical students' perceptions of the attitude of the general practitioners towards their patients and doctor-patient communication. We aimed at reaching students' own experiences, perceptions and meanings as original as possible. Therefore, we gave them only a tentative list of the main themes (categories) to be observed and emphasised their freedom to express themselves openly. In our study, there was no theoretical guiding principle, but we utilised Garfinkel's (1967) ethnometodological ideas as the source on gradually development of theoretical thinking [13].
Methods
In 2002, 2003 and 2004 like in the previous years since 1996, students observed the work of GPs in eastern and central Finland. As a part of their assignment, they wrote a short report of their experiences.
In 2002, altogether 64 primary health care centres were involved in the programme, of which 35 are official university partner institutions. Of the 133 first year students, 127 students (95.5 % of all in the class) submitted their reports, of which 88 were female students (69 %) and 39 male students (31 %). In the reference years 2003 and 2004, we received reports from 118 students (92.2% of 128 students) and 130 students (98.5% of 132 students) respectively.
Each author (JM, PM, TV) worked out one third of the 2002 reports. The texts were analysed independently and systematically by each author finding out the main themes/categories. All authors found out the same main categories, namely doctor-patient relationship, doctor as an ideal role model, perception on their own medical capabilities, observations and on organisational setting of the primary health care centre. Finally, the contents were coded by agreed subcategories (Table 2).
Table 2 The categories of doctor-patient interaction.
A. Personal qualities of practitioners
general interest, empathy, adaptability, approach
personality factors
B. Communication, interaction, dialogue
communication styles
patient information, authority
influence of continuity of care in doctor-patient relationship
disturbances in doctor-patient communication
C. Context-related topics
specific patient groups
chronic and terminal illness, dying patients
physical and psychological environment
haste
teamwork/working alone
comprehensiveness of the work
organisational issues
list system versus conventional system
continuity of care
Our purpose was to reflect the first impression of the students on the primary health care culture. In addition, the authors analysed the doctor-patient interaction against the contextual framework (Figure 1). During the process, we quantified and analysed the presence of concepts within texts putting more emphasis on qualitative description of the contents. A native English speaker with proficiency in Finnish checked the translated citations.
Figure 1 The four elements of the doctor-patient-relationship.
The reports of 2003 and 2004 were used as reference material to check whether the findings are in line with those of 2003 reports.
Results
The primary health care centre visit was reported as very useful or useful by 90% of all reporting students. No student reported the visit as useless or waste of time.
Attitudes of the practitioners towards their patients
The GPs were described as: good listener, actively present, interested, funny, cheerful, respectful, genial, warm, cosy, caring, friendly, empathic, neutral, objective, correct, effective, competent, and diplomatic. Some students reported that the consultation was pleasant, even though negative expressions were used for the practitioner such as reserved or distant. Only a few students observed attitude problems, such as a dismissive attitude, unwillingness, sneering at the patient or anxiety.
Two thirds (70%) of the students reported positive or very positive attitude of GPs towards the concerns of their patients. Students reported that the GPs treated patients as human beings, not only as patients or clients. Moreover, students observed that GPs allowed patients to express themselves freely: In my opinion doctor Y's attitude towards the patients was objective but human. She did not remain distant. When I observed her work, I felt that she really cares, and concentrates thoroughly on each and every case. She had a good contact with her patients, and the patients described their complaints to her openly. Doctor Y spared enough time for explaining facts to patients, and the patients actively inquired for more information, if anything was left unclear (43/F).
Students perceived that patients could feel safe and comfortable with GPs with differing attitudes: The younger doctor was more spontaneous and quicker in a way. He normally made jokes and chatted, and at times he could act very efficiently and solve a problem quickly. Perhaps he could see who needed support, and who only needed treatment. The older doctor was very correct; his social distance to patients seemed greater, but most probably this did not adversely affect the encounter. Although he was warm to the patient, he made sure not to step on the patient's toes. This appeared as slow movements and considerateness. On the other hand this made me think the doctor was listening and allowing enough time for the patient (60/M).
Doctor-patient interaction
GPs adjusted their communication at the patient's level. Medical students observed different ways of communication depending on the patient: The doctor knows how to handle patients and is able to adapt her communication. She normally listens to a talkative patient and at times makes focusing questions. She encourages a shy patient to talk about his/her symptoms and asks more questions (55/F).
In some cases, GPs seemed to clarify their messages with drawings: Doctor S gives clear and comprehensible instructions which she also reinforces with various drawings if necessary. At the end of the session she also checks with some questions that the patient has understood the message (114/F). But in some cases they did not give adequate guidance in health matters: It was strange how passively patients made use of immunization services, even though they are free of charge for certain high-risk groups, and how little people know about vaccinations and other procedures. A young man belonging to the high-risk group because of his asthma came for a specific reason. He asked if he should get an influenza vaccination, and the doctor recommended it. The man suspected that such a vaccine could weaken his own resistance to diseases. Is there a principle in Finland that only the doctor knows what to do with the patient? As far as I am concerned, I want to rectify this problem. In my opinion the patient should not even need to ask what happens, information should be given automatically, but of course hectic work makes interaction difficult (122/M).
Several students reported GPs as having strong opinions about unhealthy life-styles, but at the same time behaving diplomatically: The doctor was very empathic, although strict with some patients when necessary. A couple of times I noticed that the doctor looked amused when the patient asked a medically funny question. This episode did not have any importance anyway, and at the end of the day the doctor explained the medical conception patiently, and at times looked for additional information for the patient. The doctor tried to explain things clearly, and patients had the opportunity to inquire about their concerns. The doctor as a family practitioner knows the patients and thus is able to adjust his behaviour and communication style according to the patient (66/F).
Some students reported that the GP whose communication with ordinary patients was emphatic turned to more distant with children, mentally retarded, elderly, and socially marginalised patients. Similar aloofness was observed with accident and emergency cases.
Students reported that the working pace of GPs in accident and emergency departments is hectic: Both doctors behaved differently in the A&E department than in the non-acute clinical sessions. In my opinion, this was merely due to the high number of patients in the A&E department, where the hectic working pace simply could not be handled. The fairly cold attitude put me off in the beginning, but afterwards I realized that it is the only way to cope with the chaotic situation (94/M). In addition, students perceived and reported GPs working style as mechanical and distant: Also on the ward the same attitude of help and support was present, but one could not help getting the impression that chronic patients are like things, plants or animals, features which the doctor investigates with the nurse and introduces to a guest (89/M).
The computer screen and the concentration of GPs on other matters (incoming telephone calls, and other staff members entering the consultation room) were experienced as disturbing factors in doctor-patient communication: I did not feel comfortable with the habit of all the doctors of staring at the computer screen while listening to the patient. Many times the patients, while they were still speaking, started to look at me as if they were explaining their concerns to me (63/F).
Students reported continuity of care as an important element of the doctor-patient-relationship: ... Patients seem to be very open with regard to their illnesses. Doctors behave like patients' family members. Based on our discussions with patients, they feel comfortable in seeking medical help from their own family doctors and are bold enough to ask them for help, unlike in the case of an unknown doctor. Also doctors gave positive feedback about the family doctor system; it often helps when one knows the concerns of the patient in advance. So, no time is wasted on getting to know each other (54/F).
The reports suggest that the primary health care centre visit may have a positive effect on the early professional growth of medical students and seems to turn negative attitudes towards primary health care into more positive attitudes. Several students observed identity forming rituals; they appreciated that they were introduced to the clients as GPs' "colleagues". Similarly, wearing white uniform, stethoscopes, getting small gifts from pharmaceutical representatives etc. and recognition by other staff members were reported to help them adapt to the primary health care centre work as team members:It was nice to dress in the white uniform for the first time and feel what it is like to be a practising doctor (114). The best thing was that I had a chance to examine sinuses or auscultate lungs and heart on my own (57).
Discussion
The reports reflect perceptions of first year medical students during the 2-day primary health care centre visit. They entered the institutions with their eyes open, like newcomers in a foreign culture. Thus their observations represent a mixture of lay and professional conceptions, although the students had already gone through the introductory theoretical studies and practical exercises in communication skills prior to the primary health care centre visit.
In our study, in spite of the fact that each researcher analysed one-third of the material independently, the researchers' final classification of the material was similar, and the conclusions were parallel. No conflicting facts were found with the reference data of 2003 and 2004.
There are two main reliability problems in our study. Firstly, the students may have acquired a slightly more positive impression of the doctor-patient interaction than is actually the case, because the presence of a student in the consultation room may have made the GPs more polite and careful. Secondly, due to the timing of the health care centre visit in the first weeks of medical studies, the general practice culture may have been so fascinating and exciting to the students that they may have picked more positive than negative experiences ("honeymoon effect ").
The students' reports are well in line with previous findings that Finnish primary health care centre doctors are highly committed to their work and the concerns of their patients, but are also under heavy psychological stress caused by the ever-increasing demands made on them by the public health care system [12,14]. Medical students observed with admiration the diversity of GP work, and seemed to realise the great variety of skills and capabilities that are required in general practice.
The students regarded the two-day primary health care centre visit as useful and motivating. They reported the interaction between the GP and the patient as keys to a successful clinical encounter. They reported primary health care physicians as adaptable and skilful medical experts, but also as good supporters, even friends in some cases. The approach of the individual GPs differed, but very few were reported to have big difficulties in inter-personal relationships.
The students perceived some weak areas in the doctor-patient interaction, such as the behaviour of GPs in a busy working environment, with socially marginalized groups, and with seriously ill and dying patients.
The students seemed to have – a priori – an impression what a good GP is like, and how patients should be treated. Evidently, the early pre-clinical contact of medical students with the real general practice helps students adopt a holistic approach in their future clinical work. As a result, they treat patients like human beings or real friends, and not like numbers, machines or strangers [15].
Conclusion
Our findings are in line with the previously documented importance of pre-clinical primary care orientation programme. We discovered that the positive role model given by senior general practitioners obviously strengthens the confidence of first year medical students in GP-work; with the result that working as a GP can be one of their realistic career options. We welcome more European research on the influence of the early pre-clinical family practice exposure on medical students' primary care specialty choice.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
The reports of 2002 were divided randomly into three parts of equal number. Each author (JM, PM, TV) read through his/her share. Consequently, series of working sessions took place, and consensus of the findings was achieved. JM read through the reports of 2003 and 2004 for reference, and drafted the manuscript. PM and TV contributed with their additional comments. Finally, all authors read and approved the manuscript. The reviewers' reports were discussed intensively. After consulting some additional literature, revisions were made as a joint venture.
Pre-publication history
The pre-publication history for this paper can be accessed here:
==== Refs
Health Care in Finland Brochures of the Ministry of Social Affairs and Health 2004 11
Finnish Medical Association Health Care and Medical Education in Finland
Finnish Medical Association Medical Education in Finland
Armstrong D Hicks BH Higgins PM Weinman JA Teaching communication skills to pre-clinical medical students: a general practice based approach Med Educ 1979 13 82 85 431420
Cade J An evaluation of early patient contact for medical students Med Educ 1993 27 205 210 8336569
Wilkinson TJ Gower S Sainsbury R The earlier, the better: the effect of early community contact on attitudes of medical students to older people Med Educ 2002 36 540 542 12047668 10.1046/j.1365-2923.2002.01226.x
Lassen LC Larsen JH Almind G Backer P Medical students experience early patient contact in general practice. A description and evaluation of a new course in the medical curriculum Scand J Prim Health Care 1989 7 53 55 2727462
Valkova L First early patient contact for medical students in Prague Fam Pract 1997 14 394 396 9472374 10.1093/fampra/14.5.394
Hampshire AJ Providing early clinical experience in primary care Med Educ 1998 32 495 501 10211291 10.1046/j.1365-2923.1998.00243.x
Paice E Heard S Moss F How important are role models in making good doctors? Br Med J 2002 325 707 710 12351368
Bland CJ Meurer LN Maldonado G Bland CJ Meurer LN Maldonado G Determinants of primary care specialty choice: a non-statistical meta-analysis of the literature Acad Med 1995 70 620 641 7612128
Kumpusalo E Haggren O Vehviläinen A Liukko M Kujala S Takala J Miten terveyskeskuslääkärit viihtyvät työssään? Terveyskeskustyön nykytila ja tulevaisuus 2002 – tutkimus I. [How satisfied are primary health care centre doctors with their work? The current state and future prospects of general practice 2002 – study I] Suom Lääkäril 2002 57 4235 4239
Garfinkel H Studies in ethnomethodology Englewood Cliffs, Orientice Hall 1967
Virjo I Mattila K Kumpusalo E Neittaanmäki L Hyppölä H Kujala S Luhtala R Halila H ja Isokoski M Lääkäri 98 – tutkimustietoa lääkärikunnasta. [Physician 98 – research data on the medical profession] Suom Lääkäril 1999 54 3997 4002
Sered S Tabory E "You are a number, not a human being". Israeli breast cancer patients' experiences with the medical establishment Med Anthropol Q 1999 13 223 252 10440031 10.1525/maq.1999.13.2.223
|
16162300
|
PMC1242232
|
CC BY
|
2021-01-04 16:30:56
|
no
|
BMC Med Educ. 2005 Sep 15; 5:34
|
utf-8
|
BMC Med Educ
| 2,005 |
10.1186/1472-6920-5-34
|
oa_comm
|
==== Front
BMC Med Res MethodolBMC Medical Research Methodology1471-2288BioMed Central London 1471-2288-5-291616227910.1186/1471-2288-5-29Research ArticleModeling repeated ordinal responses using a family of power transformations: application to neonatal hypothermia data Zayeri Farid [email protected] Anoshirvan [email protected] Navid [email protected] Fatemeh [email protected] Department of Biostatistics, School of Medical Sciences, Tarbiat Modarres University, Tehran, Iran2 Department of Biostatistics, School of Medical Sciences, Tarbiat Modarres University, Tehran, Iran3 Department of Obstetrics and Gynecology, Tehran University of Medical Sciences, Tehran, Iran4 Department of Neonatology, Tehran University of Medical Sciences, Tehran, Iran2005 14 9 2005 5 29 29 25 1 2005 14 9 2005 Copyright © 2005 Zayeri et al; licensee BioMed Central Ltd.2005Zayeri et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
For analyzing a repeated ordinal response, it is common to use a multivariate cumulative logit model. This model may fit poorly, especially when a nonsymmetric response is available. In these cases, alternative strategies should be utilized.
Methods
In this paper, we present a family of power transformations for the cumulative probabilities to model asymmetric departures from the random-intercept cumulative logit model. To illustrate this method, we analyze the data from an epidemiologic study to identify risk factors of hypothermia among newly born infants in some referral university hospitals in Tehran, Iran.
Results
For hypothermia data, using this family of transformations and comparing the goodness-of-fit statistics showed that a model with the cumulative complementary log-log link gives us a better fit compared to a model with the cumulative logit link.
Conclusion
In some areas, using the ordinary cumulative logit link function does not lead to the best fit. So, other link functions should be evaluated to discover the best transformation for the cumulative probabilities.
==== Body
Background
Hypothermia is an important cause of morbidity, and occasionally mortality, in the newborn [1]. In 1958, Silverman et al. [2] and in 1964, Buetow and Klein [3] reported the adverse effects of hypothermia on viability and hope for life in premature and low birth weight neonates. Low body temperature in newborns can lead to an increased rate of basal metabolism, peripheral vasoconstriction, decreased peripheral perfusion, tissue ischemia and finally metabolic acidosis [4]. Vascular changes in the lungs may result in decreased ventilation, increased demand for oxygen and worsening of respiratory distress [5]. Meanwhile, acidosis and hypoxia can predispose to pulmonary hemorrhage and disseminated intravascular coagulation (DIC) [4]. Hepatocyte ischemia affects liver functions and may cause indirect hyperbilirubinemia. In addition, the high metabolic rate leads to higher glucose consumption and hypoglycemia [5]. In many parts of the world, health personnel are not aware of the importance of keeping babies warm by simple methods such as drying and wrapping immediately after birth, avoiding harmful practices, encouraging early breast feeding and keeping newborns in close contact with their mothers [6].
Considering the prevalence of hypothermia experienced by Iranian neonates, and regarding that there is not adequate information about this health problem in our country, we decided to design an epidemiologic study to estimate the incidence rate and identify some of the most important risk factors of neonatal hypothermia in different referral hospitals of Tehran, Iran. In this longitudinal study, the body temperature of the newborns was measured repeatedly at several occasions. At each time of measurement, the ordinal outcome was defined as the severity of hypothermia for each newborn.
To analyze the dependence of a categorical response data on explanatory variables it is common to fit transforms of the probabilities by linear functions of parameters. In this context, the logistic transform is probably the one most commonly used. However, as for all models, it is tentative and therefore some consideration of adequacy is needed. If some non-logistic model gives a better or simpler fit, it is important to discover that. In this paper, we introduce a family of random-effects models to describe the relationship between repeated ordinal response data and a host of covariates. Using this family of statistical models, we are able to model asymmetric departures from the cumulative logit model. This approach is a simple and straightforward extension of the family of power transformations introduced by Aranda-Ordaz [7] to model asymmetric departures from the ordinary logistic regression model. We also develop the necessary computer program to obtain the maximum likelihood estimates of the model parameters.
Methods
The study of hypothermia in the newborns
The study of hypothermia was an epidemiologic research in some referral university teaching hospitals of Tehran, Iran. In this study, the researchers aimed to estimate the incidence rate of hypothermia and identify some of the most important risk factors of this health problem. To do this, a random sample of 900 newborns was selected in these hospitals from August 2003 to May 2004. After obtaining consent from the neonates' parents, the rectal temperature of the newborns was measured using a low-reading rectal thermometer at the following occasions;
i) Immediately after birth in the operating room
ii) After admission to the neonatal unit (levels I, II, III of nursery care)
iii) One hour after admission to the neonatal unit
iv) Two hours after admission to the neonatal unit
If a newborn was hypothermic, she/he was re-warmed according to WHO recommendations [6]. The ordinal response variable was defined as the severity of hypothermia at each occasion; 1 = moderate to severe hypothermia (temperature less than or equal to 35°C), 2 = mild hypothermia (temperature between 35°C and 36.5°C), 3 = normal body temperature (temperature between 36.5°C and 38°C). In addition, the following covariates were considered as the potential risk factors or risk indicators for neonatal hypothermia; sex (0 = male, 1 = female), weight (0 = more than or equal to 2500 gr, 1 = less than 2500 gr), gestational age (0 = more than or equal to 37 weeks, 1 = less than 37 weeks), environmental temperature at each time of measurement (0 = more than or equal to 27°C, 1 = less than 27°C, where 27°C was the mean temperature of the operating room and neonatal unit during the study), apgar score (a quick method of assessing the state of newborn infant. This score comprises five components: heart rate, respiratory effort, muscle tone, reflex irritability, and color, each of which is given score of 0, 1, or 2 [8]) immediately and five minutes after birth (0 = more than or equal to 8, 1 = less than 8) and cardiopulmonary resuscitation (CPR) (0 = not received, 1 = received).
Note that, the recorded outcomes for each newborn (severity of hypothermia at different occasions) are positively correlated ordinal observations, so convenient statistical approaches should be utilized to model the relationship between this response data and the described factors.
Family of transformations for repeated ordinal response data
Suppose, in a longitudinal study, there are T occasions (times) of measurement, and the ordinal response at each time has j = 1,2,...,J levels. This ordinal response for ith individual (i = 1,2,...,N) at tth time of measurement (t = 1,2,...,T) can be denoted by Yit. Each individual has T covariate vectors xit, each of dimension P × 1; the vector xit contains all the relevant covariates at time t, including time-dependent and time-stationary covariates. We also let Xi = (xi1,...,xiT)' represent the T × P matrix of covariates for subject i.
In 1981, Aranda-Ordaz [7] introduced a family of asymmetric transformations for binary response data in the form of
w(π) = {(1-π)-λ -1}/λ (1)
where 0 <π < 1 denotes the probability of success and λ is the transformation parameter. This seems to be a useful transformation when it is desirable to treat successes and failures asymmetrically. Using equation (1), one can denote the GLM form of this family as
log w(π) = η (2)
where η = β'X is the linear systematic part of the model, and β is a vector of unknown regression parameters. Using equations (1) and (2), the family of asymmetric transformations for univariate binary response data can be written as
log{[(1-π)-λ -1]/λ} = X'β (3)
Using simple calculations, one can show that for λ = 1 equation (3) reduces to the ordinary logistic model, while for λ→0 the complementary log-log model is obtained.
Now, we generalize the described family of power transformations to repeated ordinal response data. Suppose πitj = pr(Yit = j | ui) denotes the probability of ordinal response j for ith individual at time t. Now, using this definition and considering the equations (2) and (3), a family of random-intercept models for repeated ordinal response data can be defined as
log w(γitj | ui) = ui + αj + β'xit (4)
where γitj = pr(Yit ≤ j | ui) is the cumulative probability of response category j for individual i at time t. Here, ui denotes the random term for cluster i, αj's are known as model cut-off points and β is the common P × 1 vector of fixed-effect regression parameters.
Using equation (4) for hypothermia data, the family of random-effects models can be written as
log{[(1-γitj)-λ -1]/λ | ui} = ui + αj + β1Sexi + β2Weighti + β3Gestational_agei + β4Environmental_Tempit + β5Apgar1i + β6Apgar5i + β7Multiple_pregi + β8CPRi (5)
for i = 1,2,...,900, t = 1,2,3,4 and j = 1,2. Here, γit1 = πit1 is the probability of being moderate to severe hypothermic for newborn i at time t (probability of being in the first category of ordinal response) and γit2 = πit1 + πit2 is the probability of being hypothermic (including mild, moderate or severe) for ith neonate at tth time of measurement (the probability of being in the first or second category of ordinal response). We also suppose that ui ~ N(0,σ2), where σ is an unknown scale parameter which should be estimated in the model fitting process.
Maximum likelihood estimators and computer programs
The model in equation (4) has P unknown fixed-effect regression parameters (β), J-1 unknown cut-off points (αj), and a parameter pertaining to the distribution of ui (σ, the standard deviation of the random-effect terms).
Consider the random-intercept model in equation (4). Assuming η = ui + αj + β'xit, it is easy to show that
and, therefore, the marginal probabilities can be computed by the following equation
πitj = γitj - γit(j-1) (7)
To write the required likelihood function, one can form J indicator random variables yitj, where yitj = 1 if Yit= j, and yitj = 0 if otherwise. The marginal distribution of Yit is assumed to be multinomial (with sample size yit+=1), that is
Now, the necessary log-likelihood function for estimating the model parameters can be written as below
where ψ is a vector including all the unknown model parameters.
Solving the above score function in a random-effects model generally is not trouble-free, especially with an unknown transformation parameter. In this context, the procedure NLMIXED in statistical software SAS usually works well. By a user-friend programming, the likelihood function in equation (8) can be defined and then available estimating methods in this procedure help us to find the parameter estimates and related standard errors. For hypothermia data set, we used a Dual Quasi-Newton as the optimization technique and an Adaptive Gaussian Quadrature as the integration method. In addition, some useful goodness of fit statistics such as Akaike's Information Criterion (AIC) and Schwartz's Bayesian Information Criterion (BIC) are available in this procedure. A model with the smallest value of AIC or BIC shows a better fit compared to other random-effects models. To find more detailed descriptions about fitting the random-effects models, the interested reader can refer to Agresti [9]. In addition, the SAS code for fitting the random-intercept model in equation (5) is available in Appendix. The NLMIXED procedure estimates all the unknown parameters in the log-likelihood function. We denoted the unknown fixed-effect regression parameters by b1, b2, ..., b8 and the unknown transformation parameter (λ) by b9.
Results
Description of the data
The study sample consisted of 900 neonates (452 male and 448 female newborns). Of these, 298 newborns (33.1 percent) had low or very low birth weight (weight less than 2500 gr), and 323 newborns (35.9 percent) were preterm (gestational age less than 37 weeks). The mean temperature of the operating rooms and neonatal units was about 27°C (SD = 2.1). In addition, 726 neonates (80.7 percent) had apgar score more than or equal to 8 at the first minute after birth and 844 newborns (93.8 percent) had apgar score 8 or higher five minutes after birth. In this sample, the rate of multiple pregnancies was about 3 percent. Additionally, 63 newborns (7 percent) received CPR during the study. It should be noted that 42 hypothermic newborns (9 percent) died in a short period after birth, while this rate was about 2.7 percent (11 newborns) for the non-hypothermic neonates.
As mentioned before, for each newborn the ordinal response variable was the severity of hypothermia at each time of measurement. Table 1 shows the incidence rate and severity of hypothermia among these newborns, separately in four consecutive measurements. Summing over mild and moderate-severe rows in this table shows that 53.4, 13.7, 2.9 and 0.7 percent of these newborns were hypothermic, respectively at these four consecutive measurements.
Table 1 Severity of hypothermia among the sample neonates
Occasion
Time 1 Time 2 Time 3 Time 4 Total
Severity of Hypothermia Normal 419 777 874 894 2964
Mild 329 105 20 4 458
Moderate-Severe 152 18 6 2 178
Total 900 900 900 900 3600
Analysis of risk factors
To evaluate the fit of the illustrated model and to identify the significant risk factors of neonatal hypothermia, we first fit the described model in equation (5) with unknown transformation parameter in order to estimate this parameter and provide preliminary information about the appropriate link function for this data set. Table 2 shows the estimates, standard errors, p-values, and goodness-of-fit statistics. The estimate of transformation parameter, that is , shows serious departure from the cumulative logit model. In other words, since the estimate of transformation parameter is very close to zero, we can conclude that a model with the cumulative complementary log-log link (λ→0) seems to be more convenient for this data compared to a model with the cumulative logit link (λ = 1).
Table 2 Estimates from the longitudinal hypothermia data
Unknown λ λ = 1 λ→0
Parameter Est† SE‡ P§ Est SE P Est SE P
α1/Cutoff 1 2.932 0.256 _ 4.212 0.308 _ 3.611 0.235 _
α2/Cutoff 2 5.839 0.261 _ 14.144 0.306 _ 6.733 0.269 _
β1/Sex 0.118 0.126 0.346 0.585 0.562 0.298 0.230 0.206 0.264
β2/Weight 1.178 0.153 <0.001 3.005 0.773 <0.001 1.242 0.157 <0.001
β3/Gest Age# 0.351 0.174 0.044 1.224 0.654 0.061 0.249 0.112 0.025
β4/Env Temp* 2.976 0.501 <0.001 4.368 0.648 <0.001 2.247 0.511 <0.001
β5/Apgar1** 0.370 0.153 0.016 1.096 0.533 0.040 0.400 0.185 0.031
β6/Apgar5*** 1.037 0.366 0.005 2.629 1.198 0.028 1.009 0.436 0.021
β7/Multi Preg$ 0.802 0.352 0.023 1.659 0.805 0.039 0.776 0.336 0.021
β8/CPR 0.302 0.104 0.004 0.272 0.137 0.047 0.277 0.114 0.015
σ 3.584 0.395 <0.001 5.987 0.648 <0.001 3.977 0.410 <0.001
λ 0.031 0.013 0.020 _ _ _ _ _ _
-2log.likelihood 3882.6 3906.8 3889.5
AIC 3904.6 3928.8 3911.5
BIC 3972.7 3996.9 3979.6
§ p for two-sided p-value
† Est for estimate of the model parameter
‡ Se for standard error of the estimate
# Gest Age for gestational age of the neonate
* Env Temp for environmental temperature
** Apgar1 for apgar score at the first minute after neonate's birth
*** Apgar5 for apgar score five minutes after neonate's birth
$ Multi Preg for multiple pregnancy
In the next step, we used the cumulative complementary log-log and logit link functions to model hypothermia data (Table 2). The obtained results tell us that there are substantial differences between the parameter estimates in these two models. Furthermore, comparing the goodness-of-fit statistics reveals that the model with the cumulative complementary log-log link gives a better fit compared to the model with the cumulative logit link function. This is not in contrast with the obtained results from the model with unknown transformation parameter.
Here, it should be noted that the fitted models did not lead to the same significant risk factors for hypothermia. Regarding to the column of p-values in Table 2 for the model with unknown λ and the model with the cumulative complementary log-log link, we can conclude that all the described factors, except sex, are significantly associated with hypothermia. But in the cumulative logit model (the model with λ = 1), gestational age of the neonates did not show significant effect on hypothermia.
Discussion
Longitudinal studies are now widespread in many areas of medical research. The statistical analysis of these studies is usually difficult, especially when a repeated ordinal response is available [10,11]. In this context, generalized estimation equations methodology, introduced by Liang and Zeger [12], is a helpful strategy for analyzing repeated binary response data. In 1994, Lipsittz et al. extended this methodology to repeated categorical responses [13]. The choice of appropriate method for analyzing repeated categorical responses depends heavily on the aims of the study. Marginal and random-effects models are probably the most common approaches for the analysis of correlated categorical response data. Carrière and Bouyer presented helpful strategies for choosing marginal and random-effects models in longitudinal binary responses. They demonstrated that if the main goal of the study is to predict a mean prevalence of a specific disease over time by sex, age group or other characteristics, the marginal models are suitable. In contrast, if the goal is to study the individual risk factors for etiological considerations, the random-effects models are more appropriate because they allow adjustment on non-observed individual characteristics and a better understanding of the underlying mechanism [14].
In the present article, we introduced a family of power transformations to model the repeated ordinal response data. Depending on the main aim of the study, when a random-effects model is chosen for analyzing a repeated ordinal response, the first option is probably a model with the cumulative logit link function. Our main goal in the present study was to obtain more efficient estimates compared to those obtained using the cumulative logit link function. Nowadays, by using powerful statistical softwares such as SAS and S-PLUS, the fitting process is not too difficult. The required time for running process, even in large sample size data sets, is not more time-consuming compared to the ordinary random-effects models.
Supposing J = 2, the model in equation (4) reduces to a model for the analysis of repeated binary response data. In this situation, Yit is a scalar. Omitting the random terms ui, with T = 1 and J = 2, this approach reduces to the model presented by Aranda-Ordaz.
In our proposed model, assuming an unknown transformation parameter, λ, and estimating this parameter in the model fitting process is an appropriate strategy for checking asymmetric departures from the logistic model. If the estimate of the transformation parameter showed a serious departure from 1, then it can be concluded that the logistic transformation is not an appropriate option. For a given data, if the estimate of the transformation parameter is significant but not close to 0 or 1, then it can be concluded that neither the logit nor the complementary log-log link is appropriate. In this situation, one can fit the model in equation (4) with the estimated transformation parameter. Otherwise, if the estimate of this parameter is not significant or the standard error of the estimate is too large, the traditional methods of choosing the link function (for example, fitting the ordinary cumulative models with the common link functions such as logit, probit, complementary or negative log-log and then choosing the model with the best fit) may be preferable. To decide about the proper choice of transformation parameter for a given data set, an alternative strategy may be fitting this model with a sequence of values of λ and comparing the obtained goodness-of-fit statistics to determine a model with the best fit. In this context, drawing a graph of lambda versus common goodness-of-statistics (such as deviance) is a convenience approach for choosing a model with the proper transformation parameter.
As we mentioned before, the results of the present study showed that low birth weight and premature newborns, neonates with low apgar scores and those who received cardiopulmonary resuscitation had higher risk for being hypothermic. The same findings have been already reported in other research [15-18]. Moreover, the results of regression analysis revealed that the environmental temperature is significantly associated with neonatal hypothermia. It appears that newborns have higher risk for hypothermia when the operating room or neonatal unit temperature is not warm enough (in our study at least 27°C). This finding shows the importance of keeping the operating room and neonatal unit warm enough to reduce the risk of hypothermia among newly born infants. In general, theses results help us to train the medical care personnel for a better management of high risk newborn babies.
Appendix
SAS code for fitting the proposed random-effects model
data set1;
infile 'a:\hypothermia.dat';
y1 = 0; y2 = 0; y3 = 0;
if response = 1 then y1 = 1;
if response = 2 then y2 = 1;
if response = 3 then y3 = 1;
proc nlmixed qpoints = 100;
bounds i2>0;
bounds b9>0; *** b9 is the unknown transformation parameter,λ***
eta1 = i1+ sx*b1+ wt*b2+ ga*b3+ et*b4+ ap1*b5+ ap5*b6+ mp*b7+ cpr*b8+ u;
eta2 = i1+i2+ sx*b1+ wt*b2+ ga*b3+ et*b4+ ap1*b5+ ap5*b6+ mp*b7+ cpr*b8+ u;
p1 = 1-((1+ b9*exp(eta1))**(-1/b9));
p2 = (1+ b9*exp(eta1))**(-1/b9)) - ((1+ b9*exp(eta2))**(-1/b9));
p3 = (1+ b9*exp(eta2))**(-1/b9));
LL = y1*log(p1)+ y2*log(p2)+ y3*log(p3);
*** LL is the log-likelihood function ***
model response~general(LL);
estimate 'intercept2' i1+i2;
random u~normal(0, sigma*sigma) subject = case;
run;
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
FZ performed the statistical analysis, interpreted the results and drafted the main body of the manuscript. AK was the chief coordinator of the study and participated in the drafting of the manuscript. NK participated in the data collection, the drafting of the manuscript and coordinated the communications. FN coordinated the medical part of the research, designed the questionnaire and cooperated in the drafting of the manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors wish to acknowledge Dr. Masoud Salehi for his kind help in data analysis and drafting the manuscript.
==== Refs
Hackman PS Recognizing and understanding the cold-stressed term infants Neonatal Netw 2001 20 35 41 12144102
Silverman WA Fertig JW Berger AP The influence of the thermal environment upon the survival of newly born premature infants Pediatrics 1958 22 876 886 13600915
Buetow KC Klein SW Effect of maintenance of "normal" skin temperature on survival of infants of low birth weight Pediatrics 1964 34 163 170 14211076
LeBlanc MH The physical environment Neonatal-Perinatal medicine 2002 1 7 St Louis, Missouri: Mosby 512 524
Loughead MK Loughead JL Reinhart MJ Incidence and physiologic characteristics of hypothermia in very low birth weight infants Pediatric Nursing 1997 23 11 15 9137016
WHO/FHF/SM Thermal control of the newborn: a practical guide Geneva 1993
Aranda-Ordaz FJ On two families of transformations to additivity for binary response data Biometrika 1981 68 357 363
Cunningham FG Gant NF Leveno KJ Gilstrap LC Hauth JC Wenstrom KD Williams Obstetrics 2001 I 21 New York: McGraw-Hill
Agresti A Categorical data analysis 2002 2 New York: John Wiley & Sons
Diggle PJ Liang KY Zeger SL Analysis of longitudinal data 1994 New York: Oxford
Dwyer JH Fienleib M Lippert P Hoffmeister H Statistical models for longitudinal studies of health 1992 New York: Oxford
Liang KY Zeger SL Longitudinal data analysis using generalized linear models Biometrika 1986 73 13 22
Lipsitz SR Kim K Zhao L Analysis of repeated categorical data using generalized estimating equations Stat Med 1994 13 1149 1163 8091041
Carrière I Bouyer J Choosing marginal or random-effects models for longitudinal binary responses: application to self-reported disability among older persons BMC Med Res Methodol 2002 2 15 12466027 10.1186/1471-2288-2-15
Kambarami R Chidede O Neonatal hypothermia levels and risk factors for mortality in a tropical country Cent Afr J Med 2003 49 103 106 15298464
Manji KP Kisenge R Neonatal hypothermia on admission to a special care unit in Dar-es-Salam, Tanzania: a cause for concern Cent Afr J Med 2003 49 23 27 14562586
Ondoa-Onama C Tumwine JK Immediate outcome of babies with low apgar score in Mulago Hospital, Uganda East Afr Med J 2003 80 22 29 12755238
Doctor BA O'Riordan MA Kirchner HL Shah D Hack M Perinatal correlates and neonatal outcomes of small for gestational age infants born at term gestation Am J Obstet Gynecol 2001 185 652 659 11568794 10.1067/mob.2001.116749
|
16162279
|
PMC1242233
|
CC BY
|
2021-01-04 16:32:52
|
no
|
BMC Med Res Methodol. 2005 Sep 14; 5:29
|
utf-8
|
BMC Med Res Methodol
| 2,005 |
10.1186/1471-2288-5-29
|
oa_comm
|
==== Front
BMC Med Res MethodolBMC Medical Research Methodology1471-2288BioMed Central London 1471-2288-5-301617151810.1186/1471-2288-5-30Research ArticleAdaptive designs based on the truncated product method Neuhäuser Markus [email protected] Frank [email protected] Institute for Medical Informatics, Biometry and Epidemiology, University of Duisburg-Essen, Hufelandstr. 55, D-45122 Essen, Germany2 Novartis Pharma AG, WSJ-27.1.005, 4002 Basel, Switzerland2005 19 9 2005 5 30 30 12 12 2004 19 9 2005 Copyright © 2005 Neuhäuser and Bretz; licensee BioMed Central Ltd.2005Neuhäuser and Bretz; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Adaptive designs are becoming increasingly important in clinical research. One approach subdivides the study into several (two or more) stages and combines the p-values of the different stages using Fisher's combination test.
Methods
Alternatively to Fisher's test, the recently proposed truncated product method (TPM) can be applied to combine the p-values. The TPM uses the product of only those p-values that do not exceed some fixed cut-off value. Here, these two competing analyses are compared.
Results
When an early termination due to insufficient effects is not appropriate, such as in dose-response analyses, the probability to stop the trial early with the rejection of the null hypothesis is increased when the TPM is applied. Therefore, the expected total sample size is decreased. This decrease in the sample size is not connected with a loss in power. The TPM turns out to be less advantageous, when an early termination of the study due to insufficient effects is possible. This is due to a decrease of the probability to stop the trial early.
Conclusion
It is recommended to apply the TPM rather than Fisher's combination test whenever an early termination due to insufficient effects is not suitable within the adaptive design.
==== Body
Background
Randomized controlled experiments were introduced by Sir Ronald A. Fisher in the 1920s for agricultural studies and not in order to compare the effects of different treatments in humans. However, according to Palmer [1] the way clinical trials are conducted today is essentially unchanged from Fisher's day. In contrast to agricultural studies most clinical trials require periodic monitoring of the accumulating data, e.g. to minimize the number of experimental patients who will continue with an inferior treatment [[2], p. 360].
Adaptive designs with at least one interim analysis can potentially be used for periodic monitoring. All information from the first stage(s) can be used to plan the following stage(s). A number of adaptive designs have been proposed recently, for an overview see Bauer et al. [3]. Here, we consider the adaptive procedure according to Bauer and Köhne [4] that uses Fisher's product test.
Let k be the number of stages (i.e., there are k - 1 interim analyses), and let pi be the one-sided p-value observed with the i-th stage's data, i = 1, ..., k. According to Fisher's product criterion [[5], pp. 37–39] the null hypothesis H0 can be rejected at the end of the trial if
,
where is the (1 - α)-quantile of the central χ2-distribution with 2k degrees of freedom.
In clinical trials boundaries for early stopping after an interim analysis may be incorporated. Obviously, in the case of p1 ≤ cα early stopping with the rejection of H0 is possible after stage one. In general, H0 can be rejected after the j-th stage if . In addition, one may terminate the trial due to insufficient effects. A lower limit α0 can be included so that the trial is terminated without rejecting H0 if p1 ≥ α0. According to Bauer and Köhne [[4], p. 1031] a value of 0.5 may be a suitable choice for α0. Bauer and Röhmel [[6], p. 1596] recommended α0 = 1 for establishing a dose-response relationship, that is, no early stopping without rejecting H0 at all. In this context, an early stopping due to insufficient effects is not feasible since doses in a plateau region could have been used. In that case, different doses may be used in the following stage.
Note that, in case of α0 < 1, larger boundaries for apply for early stopping with the rejection of H0. For a two-stage design, one can reject H0 after stage one if p1 ≤ α1 for a value of α1 that lies between cα and α [4,6]. This value can be calculated iteratively using the formula [[4], p. 1032]
As an alternative to Fisher's product test, Zaykin et al. [7] recently introduced a truncated product method for combining p-values. To be precise, instead of calculating the product of all p-values, they suggested the use of the product of only those p-values that do not exceed some fixed cut-off value τ, 0 < τ ≤ 1. The truncated product Wτ is defined as
where I(.) is the indicator function. Since the p-values of the different stages are independent,
[[7], p. 173] holds for w < 1 under the overall null hypothesis (i.e., under the assumption that each stage's null hypothesis is true). Figure 1 displays the rejection region for k = 2 and τ = 0.5.
Figure 1 The rejection region of the truncated product method for k = 2 and τ = 0.5.
When using the truncated product method, the (1-α)-quantile of the distribution of Wτ, , is the critical value for the combination test. Analogous to Fisher's combination test an can be calculated for given α0 such that the overall type I error rate is α.
Zaykin et al. [7] and Neuhäuser [8] investigated the truncated product method for combining a large number of p-values and demonstrated by simulation that it can provide high power. In this paper we investigate whether the truncated product method is also useful for the adaptive design described above. In contrast to previous applications [7,8] we consider classical experimental questions involving only few p-values. Very recently, a rank truncated product was proposed as a further alternative [9]. That method uses the product of the K most significant p-values where K can be chosen. Since we consider the combination of 2 to 4 p-values only, the rank truncated product does not seem to be appropriate for our aim.
We first present the comparison of the combinations with and without truncation for designs with two stages. Afterwards, designs with more than two stages are investigated. We then illustrate the method using two examples, and conclusions are given in final section.
Methods
In order to compare the adaptive procedures with and without truncation we consider the situation of two parallel groups with means μ1 and μ2. There are 100 observations per stage. These observations are subdivided into two groups and are assumed to be normally distributed with a common, but unknown variance σ2. Student's t test is performed in each of the two stages with a one-sided significance level of α = 5%.
The overall p-value, i.e. the p-value of the combination test, is defined as follows [10]: In case the study stops after stage 1, the overall p-value equals p1. Otherwise, the overall p-value is for Fisher's combination test and for the truncated product test.
The case α0 = 0.5
First, we consider a study that is terminated early due to insufficient effects if p1 ≥ α0 = 0.5. Without any truncation (i.e., τ = 1) we have cα = 0.0087 and α1 = 0.0233 in this case [4]. However, when we set τ = α0 = 0.5, a smaller value for α1 but a larger boundary for is obtained. To be precise, the trial can be terminated early with the rejection of H0 if p1 ≤ = 0.0190, and there is a significance at the end of the trial if Wτ = 0.5 ≤ = 0.0095.
Although α1 is decreased the overall power can increase in case of truncation as the boundary for Wτ = 0.5 is larger than that for Wτ = 1. Table I displays the overall power, that is, the power to reject H0 after any stage, for different alternatives (see the appendix for details about the calculation of the power). The power is slightly higher in case of truncation. The difference is very small when the ratio (sample size in stage one)/(sample size in stage two) is large. The reason is that the probability to stop already after the first stage depends on the sample size in stage one.
Table 1 Power to reject H0 in a two-stage design with α0 = 0.5 (combination of t tests, one-sided, α = 0.05)
δ = 0.1 0.2 0.3 0.4 0.5
25 observations per group in stage one, 75 observations per group in stage two
τ = 1 0.149 0.343 0.595 0.808 0.929
τ = 0.5 0.153 0.352 0.605 0.815 0.931
50 observations per group and stage
τ = 1 0.162 0.377 0.644 0.854 0.959
τ = 0.5 0.165 0.384 0.652 0.860 0.961
75 observations per group in stage one, 25 observations per group in stage two
τ = 1 0.166 0.386 0.654 0.860 0.962
τ = 0.5 0.167 0.389 0.657 0.863 0.963
The area of the rejection region of Fisher's test that can be relocated in case of α0 < 1 [[4], p.1032] has, under H0, the probability Pr(p1 ≥ α0 and p1p2 ≤ cα) = cα(-lnα0). In case of truncation with τ = α0, an area with probability Pr(p1 ≥ α0 and p2 ≤ ) = (1 - τ) can be relocated. Since cα(-lnα0) > (1 - τ) for practically relevant situations (see e.g. Table II), we have < α1. Hence, the probability to terminate the trial after stage one is lower in case of truncation with τ = α0.
Table 2 Boundaries cα and for two to four stages
Number of stages (k) cα for τ = 0.5
α = 0.025
2 0.00380 0.00408
3 0.00072 0.00085
4 0.00015 0.00020
α = 0.05
2 0.00870 0.00948
3 0.00184 0.00222
4 0.00042 0.00057
For instance, in the case of 50 observations per group and stage and δ = 0.4 (α = 0.05) the probabilities to reject H0 after the first stage are Pr(p1 ≤ α1) = 0.496 and Pr(p1 ≤ ) = 0.461, respectively. The probability to stop without rejecting H0 is Pr(p1 ≥ α0) = 0.023 irrespective of truncation. With the fixed sample size of 100 per stage the expected total sample size is 200 - 100·Pr(stop after first stage). This expected total sample size is 148 for τ = 1, but 152 in case of truncation. Hence, the slight increase in power is connected with a larger expected total sample size.
An a priori fixed sample size for stage two is uncommon within an adaptive design. Instead, a sample size reassessment can be carried out during the interim analysis [11]. Using p1 and the difference and variability observed in stage one, we simulated the sample size for stage two needed for an overall power of 80%. The results (not shown) indicate that, in this case, the application of the truncated product method can lead to a smaller expected total sample size.
Nevertheless, there is still a smaller probability to stop the trial after the first stage when the truncation is applied. That is a clear disadvantage in clinical development where early decisions are desirable. Therefore, despite the (small) improvement in terms of power, a truncation does not seem to be preferable within a two-stage adaptive design when α0 < 1.
The case α0 = 1
As mentioned in the introduction, α0 = 1 can be a suitable choice, for example when establishing a dose-response relationship. The choice α0 = 1 leads to the same rejection boundary cα for the interim and the final analysis, respectively. Hence, there is α1 = cα and . Since cα <, the expected total sample size is decreased due to truncation even in case of a fixed sample size for stage two. For instance, in the case of 50 observations per group and stage and δ = 0.4 (α = 0.05) the probability to reject H0 after the first stage is Pr(p1 ≤ α1) = 0.342 for τ = 1, but Pr(p1 ≤ ) = 0.354 for τ = 0.5. The resultant expected total sample sizes are 166 and 165, respectively. Therefore, a gain in power would be of more importance in case α0 = 1.
However, as demonstrated in Figure 2 there is hardly any difference in power between the choices τ = 0.5 and τ = 1. Nevertheless, the application of the truncated product method is preferable in the case α0 = 1 because there is a lower expected total sample size and a higher probability to reject H0 already after the first stage.
Figure 2 Power to reject H0 in a two-stage design with α0 = 1. (50 observations per group and stage, combination of t tests, one-sided, α = 0.05)
The value increases with a decreasing truncation point τ. Hence, in order to increase the probability to reject H0 after stage one, one may argue that a smaller value of τ is preferable. However, this is not the case because the overall power depends on the choice of τ, too. For example, consider 50 observations per group and stage and δ = 0.4 (α = 0.05) again. In this case, the overall power is 0.861 for τ = 1, 0.864 for τ = 0.5, but only 0.830 for τ = 0.2.
We now present results for adaptive designs with three and four stages, respectively, and α0 = 1. Again, the behaviour of the strategies is investigated for fixed sample sizes in the separate study stages without including the option for sample size reassessment. The trial can be terminated with the rejection of H0 after the j-th stage if in case of τ = 1 or if in case of truncation. For up to four stages, Table II displays the boundaries cα and for τ = 0.5.
The choices τ = 1 and τ = 0.5 were compared in a Monte Carlo simulation study performed using SAS version 8.2. For each configuration, 10,000 simulation runs were created. Table III shows the overall power and the expected total sample sizes. Always, the truncation is more powerful than the choice τ = 1, however, the difference in power is small. Furthermore, as in the case of k = 2, the expected total sample size is smaller when the truncated product method is applied (α0 = 1). The decrease of the expected total sample size is more pronounced for larger values of k. Therefore, the truncation can be recommended again. It reduces the expected total sample size without a loss in power.
Table 3 Simulated power to reject H0 and expected total sample sizes in three- and four-stage designs with α0 = 1 (50 observations per group and stage, combination of t tests, one-sided, α = 0.05)
δ = 0.1 0.2 0.3 0.4 0.5
3 stages
Overall power τ = 1 0.198 0.498 0.789 0.950 0.993
τ = 0.5 0.198 0.502 0.799 0.953 0.993
Expected total sample τ = 1 293.3 278.7 250.0 213.7 179.6
size τ = 0.5 292.7 276.9 247.1 209.9 176.1
4 stages
Overall power τ = 1 0.230 0.590 0.883 0.984 0.999
τ = 0.5 0.233 0.596 0.888 0.985 0.999
Expected total sample τ = 1 389.0 360.0 308.5 254.1 207.6
size τ = 0.5 387.6 356.2 302.5 246.8 202.3
Discussion
In this section we only consider the case α = 0.025 and α0 = 1. The first example discussed in this section was presented by Bauer and Röhmel [6]. In a two-stage dose-response study the effect of a new drug on blood pressure was investigated. Assume that the trial would have started with two medium doses. The p-value for the one-sided t test between these two doses in the interim analysis was p1 = 0.206. Thus, the study continued with the comparison placebo vs. a higher dose, and the second stage led to p2 = 0.0178. The product in the final analysis was p1p2 = 0.00367, the corresponding overall p-value of the non-truncated product test is 0.024. Hence, the combination test is significant even at the 0.025 level.
Figure 3 shows the overall p-value of the combination test in case of truncation. Note that TPM p-values may be calculated using a C++ code offered by Zaykin et al. [7] which is available at , in addition, the method is implemented in the SAS procedure psmooth. There is no large influence of τ as long as this truncation point is larger than max(p1,p2). When τ is slightly smaller than max(p1,p2), i.e. for τ → max(p1,p2) with τ < max(p1,p2), the p-value reaches a local maximum of 0.061. For τ < min(p1,p2) the p-value equals 1. Hence, a too small choice of τ is risky. Thus, the analysis of this example may be a further indication that the choice τ = 0.5 is reasonable. In fact, in this example any τ > 0.206 would have been a powerful alternative to Fisher's criterion.
Figure 3 The overall p-value of the final analysis based on the combination of p1 = 0.206 and p2 = 0.0178 (first example) in dependence of the truncation point τ. The horizontal reference line corresponds to Fisher's product criterion.
The second example is a hypothetical clinical study with two stages. We consider a scenario as Bauer and Köhne [[4], p. 1038] in their example. A clinical trial investigates a new therapy for an indication in which no efficient standard therapy is available. For the first stage five individual endpoints have been selected. The first stage's sample size is 30 each in the therapy and the control group. The changes to the baseline measurements of the five endpoints were combined into a single generalized least squares (GLS) criterion according to O'Brien [12], and the first stage's p-value was p1 = 0.1758. Hence, the study continued.
For the second stage the set of five endpoints may be reduced for different reasons such as observed effects and variability, burden to the patients, and costs. The test statistic for the second stage was again the corresponding GLS criterion. In this example, this led to a p-value of a similar magnitude as in the first stage: p2 = 0.1517. Therefore, in the final analysis we have p1p2 = 0.0267, and the corresponding overall p-value is 0.1233 when Fisher's original product test is applied. The overall p-values for different choices of τ are displayed in Table IV. Here, the TPM gives a smaller overall p-value than Fisher's method for all considered values of the truncation point with the exception of τ = 0.1. However, that value is smaller than min(p1, p2). In this example α0 = 1 may be appropriate because no efficient standard therapy is available, the sample size of stage 1 is relatively small, and there might be only one endpoint showing a difference between the therapy and the control group.
Table 4 The overall p-value of the final analysis based on the combination of p1 = 0.1758 and p2 = 0.1517 (second example) in dependence of the truncation point τ, τ = 1 corresponds to Fisher's product criterion.
τ p-value for TPM
0.1 1
0.2 0.0801
0.3 0.0964
0.4 0.1064
0.5 0.1130
0.6 0.1174
0.7 0.1203
0.8 0.1221
0.9 0.1230
1.0 0.1233
Conclusion
The application of the truncated product method instead of Fisher's combination test within an adaptive design hardly changes the overall power. Therefore, to decide whether or not a truncation is useful one should focus on the probability to stop early and on the expected total sample size. According to these criteria, a truncation seems to be preferable in case of α0 = 1, but not for α0 < 1.
A variety of other combination functions exists [13], for example, the inverse normal method was proposed for adaptive designs [14]. According to Rice [15] Fisher's test is "inappropriate when asking whether a set of tests, on balance, supports or refutes a common null hypothesis ... because ... Fisher's statistic is more sensitive to smaller, as compared to larger, P-values" [[15], p. 303–305]. In contrast, the inverse normal method is not differentially sensitive to data that support or refute a common null hypothesis. Thus, one may argue that the inverse normal method is more appropriate for an adaptive design if each stage tests the same null hypothesis. However, in the context of a dose-response study, discussed here as a motivation for α0 = 1, different doses may be tested in different stages, that is, the hypotheses tested change. The resultant question is whether at least one stage is significant, and a high sensitivity to small p-values is desirable. Consequently, Fisher's test or TPM are appropriate. An additional advantage of these two combination methods is that an early termination with rejection of the null hypothesis is possible with α0 = 1 and a full level α combination test at the end.
There is also some literature related to the efficiency of adaptive designs, and to the choice of combination functions. Wassmer [16], for example, compared Fisher's product criterion with an alternative adaptive design proposed by Proschan and Hunsberger [17] based on a conditional power function. Wassmer [16] concluded that "no substantial differences between the procedures were found in terms of rejection regions, power, and expected sample sizes". One of the first to investigate optimal adaptive designs for the control of conditional power were Brannath and Bauer [18]. They constructed two-stage designs with overall and conditional power, which minimize the expected sample size for different specifications of the alternative. It transpires that there is a variety of different options to combine P-values and there is no consensus on the best method to use. In this paper we improve under special conditions Fisher's combination test using the truncated product method.
It is worthwhile to note that the truncation point τ must be specified a priori in the study protocol. Unless determined a priori, the truncated product method can be misused to alleviate an observed significance. A post-hoc choice based on the observed maximum of the individual p-values is therefore not permitted. As discussed above, τ = 0.5 may be a suitable choice. A further argument for this choice is that those p-values are excluded from the product that indicate a difference in the unanticipated direction. Note that the truncated product does not follow a χ2-distribution. Thus, a penalty results for the exclusion of large p-values. Nevertheless, this exclusion can be advantageous as demonstrated by Zaykin et al. [7] and above for the case of adaptive designs.
For the presentation of the power a one-sided significance level of α = 5% was chosen in this paper. However, completely analogous results can be found in case of α = 2.5%. Regarding the choice of α for one-sided tests it is referred to Neuhäuser [19].
Appendix
The power of a two-stage test according to Bauer and Köhne [4], that is, a combination with τ = 1, is given e.g. by Wassmer [[20], p. 833].
In case of truncation with τ = α0 > α the power is
where fδ denotes the respective density under the alternative δ [20]. In case of truncation with τ > α, but α0 = 1, the power is
Wassmer [20] presented a SAS/IML program to calculate the power for the two-stage test without truncation. Modifications of this program were used to calculate the different powers given above.
Abbreviations
TPM – truncated product method
Authors' contributions
MN performed most of the statistical analyses and drafted the manuscript. FB participated in the statistical analyses and helped to draft the manuscript. Both authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors would like to thank Roswitha Senske for technical support and a reviewer for helpful comments and suggestions.
==== Refs
Palmer CR Ethics, data-dependent designs, and the strategy of clinical trials: time to start learning-as-we-go? Statistical Methods in Medical Research 2002 11 381 402 12357585 10.1191/0962280202sm298ra
Gauch HG Scientific method in practice 2002 Cambridge University Press: Cambridge
Bauer P Brannath W Posch M Flexible two-stage designs: an overview Methods of Information in Medicine 2001 40 117 121 11424295
Bauer P Köhne K Evaluation of experiments with adaptive interim analyses Biometrics 1994 50 1029 1041 Correction in Biometrics 52:380 7786985
Hedges LV Olkin I Statistical methods for meta-analysis 1985 Academic Press: Orlando
Bauer P Röhmel J An adaptive method for establishing a dose-response relationship Statistics in Medicine 1995 14 1595 1607 7481196
Zaykin DV Zhivotovsky LA Westfall PH Weir BS Truncated product method for combining P-values Genetic Epidemiology 2002 22 170 185 11788962 10.1002/gepi.0042
Neuhäuser M Tests for genetic differentiation Biometrical Journal 2003 45 974 984 10.1002/bimj.200390064
Dudbridge F Koeleman BPC Rank truncated product of P-values, with application to genomewide association scans Genetic Epidemiology 2003 25 360 366 14639705 10.1002/gepi.10264
Brannath W Posch M Bauer P Recursive combination tests Journal of the American Statistical Association 2002 97 236 244 10.1198/016214502753479374
Friede T Kieser M A comparison of methods for adaptive sample size adjustment Statistics in Medicine 2001 20 3861 3873 11782039 10.1002/sim.972
O'Brien PC Procedures for comparing samples with multiple endpoints Biometrics 1984 40 1079 1087 6534410
Loughin TM A systematic comparison of methods for combining p-values from independent test Computational Statistics and Data Analysis 2004 47 467 485 10.1016/j.csda.2003.11.020
Lehmacher W Wassmer G Adaptive sample size calculations in group sequential trials Biometrics 1999 55 1286 1290 11315085 10.1111/j.0006-341X.1999.01286.x
Rice WR A consensus combined P-value test and the family-wide significance of component tests Biometrics 1990 46 303 308
Wassmer G A comparison of two methods for adaptive interim analyses in clinical trials Biometrics 1998 54 696 705 9629649
Proschan MA Hunsberger SA Designed extension of studies based on conditional power Biometrics 1995 51 1315 1324 8589224
Brannath W Bauer P Optimal conditional error functions for the control of conditional power Biometrics 2004 60 715 723 15339294 10.1111/j.0006-341X.2004.00221.x
Neuhäuser M The choice of α for one-sided tests Drug Information Journal 2004 38 57 60
Wassmer G A technical note on the power determination for Fisher's combination test Biometrical Journal 1997 39 831 838
|
16171518
|
PMC1242234
|
CC BY
|
2021-01-04 16:32:51
|
no
|
BMC Med Res Methodol. 2005 Sep 19; 5:30
|
utf-8
|
BMC Med Res Methodol
| 2,005 |
10.1186/1471-2288-5-30
|
oa_comm
|
==== Front
BMC Musculoskelet DisordBMC Musculoskeletal Disorders1471-2474BioMed Central London 1471-2474-6-481614455010.1186/1471-2474-6-48Research ArticleClassifying health-related quality of life outcomes of total hip arthroplasty Xu Min [email protected] Donald S [email protected] Lisa [email protected] Boris [email protected] Arthritis Research Centre of Canada, Vancouver, BC, Canada2 Department of Orthopaedics, University of British Columbia, Vancouver, BC, Canada3 Centre for Clinical Epidemiology & Evaluation, Vancouver, BC, Canada2005 6 9 2005 6 48 48 15 2 2005 6 9 2005 Copyright © 2005 Xu et al; licensee BioMed Central Ltd.2005Xu et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Primary total hip arthroplasty (THA) is an effective treatment for hip osteoarthritis, assessed by whatever distribution-based measures of responsiveness. Yet, the group level evaluation has provided very little evidence contributes to our understanding of the large variation of treatment outcome. The objective is to develop criteria that classify individual treatment health related quality of life (HRQOL) outcome after primary THA, adjusted by preoperative scores.
Methods
We prospectively measured 147 patients' disease specific HRQOL on the date of consultation and 12 months post operation by Western Ontario McMaster Universities Osteoarthritis Index (WOMAC). Regression models were used to determine the "expected" outcome for a certain individual baseline score. The ceiling effect of WOMAC measurement is addressed by implementing a left-censoring method.
Results
The classification criteria are chosen to be the lower boundary of the 95% confidence interval (CI) of the estimated median from the regression. The robustness of the classification criteria was demonstrated using the Monte-Carlo simulation.
Conclusion
The classification criteria are robust and can be applied in general orthopaedic research when the sample size is reasonable large (over 500).
==== Body
Background
Statistical tests are frequently called upon to assess treatments whose effect size is small or whose reduction of risk is modest, as is often the case with emerging treatments. But what of the evaluation of mature treatments whose effect is known to be substantial?
Primary total hip arthroplasty (THA) is an effective treatment for patients with severe hip osteoarthritis (OA). The improvement is large by any of the measures of responsiveness commonly used in orthopaedic research [1-6]. Yet, there are large variations reported in treatment outcome and very little good-quality evidence contributes to our understanding of this variation [7,8]. The evidence is limited mostly to patient- and implant-related factors [7]. The role of variations in service delivery practices and other factors remains unclear. On the whole, patients' outcome is good. Nevertheless, the development of classification criteria to differentiate between overall good results is necessary to achieve a better understanding of the variance and ultimately to reduce the number of relatively poor performers.
Because THA aims to improve physical function and relieve pain, and because it is broadly successful in its goal, health-related quality of life (HRQOL) is generally acknowledged to be the primary outcome of interest [9,10]. Assessment of HRQOL is typically made at the group level, that is, by measures such as the t-test, effect size (ES), and standard response mean (SRM), that are characteristic of a group. However, using such methods may not provide the best evidence to explain the association between postoperative outcomes and risk factors [8]. In contrast, the measurement of individual changes is an increasingly attractive method of quantifying HRQOL outcomes because it has the potential to document objectively the patient-perceived impact of treatment. Expectation and satisfaction are highly individualized; they contribute significantly to self-assessed quality of life. But these individualizing influences are lost in statistics such as pre- and postoperative mean scores that only express a group [11]. Raising the mean outcome is a worthwhile objective, especially when the mean badly needs improvement. When, as with effective treatments, the mean is not an overriding concern, it is appropriate to turn our attention to individuals within the mean [12]. Even groups whose mean change due to treatment is equivalent are likely to contain individuals who did substantially better and worse than others [13]. Developing statistical methods to assess these differences rather than the means themselves is a natural accompaniment to the refinement of treatments such as hip arthroplasty.
Two recent studies have found an association between preoperative health status and postoperative outcomes [14-16]. Fortin et al. examined the relationship between preoperative functional status and postoperative outcomes in a prospective cohort study using the Western Ontario McMaster Universities Osteoarthritis Index (WOMAC) and Short Form 36 (SF-36). They found that poorer preoperative function was the strongest predictor of pain and functional outcomes at 6 and 24 months after THA [15,16]. The authors concluded that surgery performed later in the natural history of functional decline results in worse postoperative functional status. They also noted that function and pain in patients with lower preoperative function did not improve after the operation to the level achieved by those with higher preoperative scores [15,16]. Thus, measures of postoperative HRQOL outcome need to be adjusted by preoperative functional status.
Methods
The objective of this study is to develop a tool for classifying the HRQOL outcome of THA based on the individual's preoperative HRQOL score. The development of the tool and the results of a simulation study are presented in the methods section. We describe the design of a case study evaluating the postoperative outcome for THA. In the development of the instrument section, a left-censored linear regression model is employed as a means of understanding and communicating the relationship between baseline and expected outcome. An expected postoperative HRQOL score for each individual preoperative score is estimated using this left-censored linear regression model. By using the expected HRQOL outcome, we identify patients whose benefit from THA is "better than expected." The performance of these classification criteria is evaluated in difference sample sizes by simulation. In the development of these classification criteria we adjust the postoperative outcome by its preoperative score. The result of this simulation study shows that these classification criteria are robust.
Study population
Data from a prospective cohort study were used for a case study. This study included 201 patients registered on the wait list for THA between March, 2001 and May, 2003 with 147 patients completed follow-up ending in March, 2004. This study was conducted at the Vancouver Hospital & Health Sciences Centre. Ethical approval was issued by the University of British Columbia Clinical Ethics Review Board. Patients presenting during this period at the Division of Reconstructive Orthopedics at Vancouver Hospital (VH) with a diagnosis of osteoarthritis (OA) and requiring primary THA are included in the study. OA is defined by the American College of Rheumatology's (ACR) clinical classification criteria for OA of the hip [17]. Patients were excluded for the following reasons: previous THA to the index joint; inflammatory arthritis; bilateral THA performed simultaneously; inability to respond to a questionnaire in English; and urgent surgery performed within 28 days after the decision for THA.
Every patient requiring hip arthroplasty was requested to complete the WOMAC questionnaire on the date of consultation. The questionnaire is self-administered. Medical office assistants handed each patient a WOMAC questionnaire once the decision was reached to enter the wait list. To assess postoperative outcomes, WOMAC questionnaires were mailed at 12 months following surgery. WOMAC is recommended for OA-specific outcomes [18,19]. It contains dimensions for pain (5 items), stiffness (2 items), and function (17 items). Dimensions are equally weighted and reported as sums, where the higher number indicates a greater burden of OA. At present it is the most frequently used measure of pain and self-reported disability among arthroplasty patients [10]. The WOMAC questionnaire has 24 questions, each question is given a Likert scale response from 0 (best health state) to 4 (worst health state). The WOMAC score for each subscale is calculated as the sum of the scores of each question included in the subscale. The range of each subscale is as follows: function: 0–68; pain: 0–20; stiffness: 0–8.
Patients' names and provincial health numbers were used to obtain age and gender through the medical office administrative database. Co-morbidity information was obtained through medical chart review using the Charnley classification, which stratifies patients by the presence of OA in one or both hips, or a co-morbid condition that impairs walking. This scale allows a meaningful comparison between groups [20]. The Charnley classes we used are:
A: Single hip with osteoarthritis
B1: Bilateral hips with arthritis
B2: Previous THA on the contra-lateral hip
C: Multiple joints affected with arthritis or a chronic disease that affects HRQOL (specifically walking)
Statistical analysis
Log-linear regression model
In the following, we aim at building a linear model to explore the relationship between follow-up score and baseline score. Since the distribution is skewed (Fig. 1 & Fig. 2), one cannot use the follow-up score in a linear regression analysis as an outcome variable. We found that the logarithms of the follow-up WOMAC functional scores follow a symmetrical distribution (Fig. 3). Therefore we build a log-linear regression as following:
Figure 1 The distribution of baseline WOMAC functional scores.
Figure 2 The distribution of follow-up WOMAC function scores.
Figure 3 The distribution of the logarithms of follow-up WOMAC functional scores.
log(Follow-up) = α+β*Baseline+σ*ε,
where Follow-up is the follow-up WOMAC score and Baseline is the baseline WOMAC score. The error term ε follows a normal distribution with a mean of 0 and a standard deviation of 1, and σ is a fixed constant that changes the variability of the expected value.
However the observed follow-up data has some WOMAC function score equal to zero and the logarithm of 0 is infinite. So we can not censor the postoperative score at 0. Moreover, the WOMAC function 0 is corresponding to complete freedom from joint symptoms. It is unlikely that patients before and after THA would have no detectable impairment in their hip. Therefore, as a measurement tool, the WOMAC questionnaire is limited in providing HRQOL information at the extreme low end of the scale (score of 0). Since a true score is unknown when the score is between 0 and 1, we regard measurements below 1 as left-censored observations. In our model, we chose 0.9 to be the censoring point so that 1 was preserved in the model and 0 was censored. We transformed the observed follow-up score as follows:
Follow-up = 0.9, if Follow-up <= 0.9;
Follow-up = Follow-up, if Follow-up > 0.9,
The Tobit regression model is a well known instrument for measuring left-censored variables in economic research [21]. In order to incorporate the left-censored observations in the regression analysis, we built a Tobit model to incorporate the left-censored observations in the regression analysis. The maximum likelihood method was used to estimate the probabilities of log(Follow-up) given the baseline WOMAC score. The regression analysis was conducted using the SAS 8.1 PROC LIFEREG procedure.
Instrument for classifying function outcomes
Through this regression analysis, an expected postoperative outcome for each baseline WOMAC functional score was obtained. Due to the skewed distribution of follow-up scores, we used the median of the follow-up score instead of the mean as the classification criteria. The mean of the predicted logarithm of follow-up scores was estimated through the model, and the median of estimated follow-up scores is exp (mean of log(Follow-up)) according to the mathematical transformation.
Since the model is derived from a rather small size sample, the variation of the estimated median of the follow-up scores should be taken into consideration. Using the lower 95% confidence interval of the median as a cutoff point associated with the baseline score, the study patients were divided into two groups. Group I: Patients below the line were considered to have achieved a "better than expected" outcome. Group II: Patients above the line were considered to have achieved a "not better than expected" outcome.
Assessment of the classification instrument
We implemented the Monte-Carlo simulation method to investigate the robustness of our classification criteria. Our intention was to assess the robustness across baseline scores and for different sample sizes. We generated random postoperative WOMAC functional scores assuming a systematic relationship between the baseline scores and postoperative WOMAC functional scores and adding a random component. The systematic relationship and the parameters for the random component were specified from the Tobit model estimates in our case study. In each data set, postoperative scores were generated for baseline WOMAC functional scores fixed at 10, 17, 34, 51, and 68. We chose 10 since it is the lowest baseline score that is eligible for surgery and available in the case study. The functional subscale contains 17 questions; each question has a response on Likert scale from 0 to 4. Therefore, we chose the folds of 17 as the baseline levels for simulation. The postoperative score was left-censored at 0.9. We looked at sample sizes increasing from 100 to 500 in increments of 100. For each sample size, we generated 1000 data sets. Then, for each data set, the regression model was fit and the median postoperative score and the cutoff points (ie. the lower bound of the 95% CI for the median score) were estimated at each baseline score.
We also tested the model using same method with different censoring points (0.9, 2, 3, 4, and 5) while the sample size was fixed at 500. For each censoring point, we generated 1000 datasets and the cutoff points were estimated for each data set.
Results
Study population
This study included 201 patients, among which there are 147 patients completed follow-up ending in March, 2004. The average age is 64.8 years and there are 83 females (56%) and 66 males (44%) in the study. Seventy-two patients (50%) have only one joint involved with OA; 34 patients have bilateral disease. Of these 34 patients, there are 18 with contra-lateral hip replacement prior to the index surgery and 16 patients with moderate to severe OA in the contra-lateral hip. Thirty-nine patients (27%) have multiple joints involved with OA or have a chronic systematic disease. When compare the component of age, gender and disease statues, there are no statistical differences of between the 147 patients and 54 patients who did not complete follow-up. In the following analysis, all the results are based on the 147 patients who completed follow-up. We found that the distribution of baseline WOMAC functional scores (scale 0–68) follows a symmetrical distribution and its' mean and standard deviation (SD) are 39 and 13 respectively (Fig. 1). Its minimum is 10 points and median is 41 points. While at the end of the follow-up, the distribution of WOMAC functional scores (scale 0–68) shows a truncated distribution because the follow-up outcome is nearly as good as a full recovery or normal function; that is, the follow-up outcomes have a limit as a score of 0 (best function). The mean follow-up WOMAC functional score is 14 (SD = 14). Its minimum is 0 points and median is 8.5 points. Since the distribution is skewed, one cannot use the follow-up score in a linear regression analysis as an outcome variable.
Log-linear regression model
Table 1 shows the parameter estimates obtained through this regression analysis. For the sample population, the estimate of the expected value of the lognormal distribution is given by:
Table 1 Parameter estimation for the log-linear regression
Parameters Estimates 95% CI
Intercept 0.98 0.29–1.67
Coefficient 0.03 0.01–0.05
Scale 1.30 1.14–1.48
Log(Follow-up) = 0.98+0.03*Baseline.
Based on the model, the baseline WOMAC functional score is a significant predictor of the follow-up WOMAC functional score (p = 0.0005). Increasing the baseline score by 10 points raises the estimated postoperative score by approximately 35%. The estimated median score line and its 95% confidence interval are shown in Fig. 4.
Figure 4 95% confidence interval for the median of expected function outcomes.
Age, gender, co-morbidity, and waiting time were also tested as covariates in the log-linear regression model. None of these variables were significant predictors of the follow-up WOMAC functional score and there was no difference in the regression coefficient for baseline WOMAC functional scores with or without these covariates. A goodness of fit test showed that the Tobit model is well fitted. In the model, the outliers are detected by the studentized residual; those observation having an absolute studentized residual over 3.5 were removed.
Instrument for classifying function outcomes
The simulation results are summarized in Fig. 5, 6, 7, 8 and Table 2. Fig. 5 summarizes that the median estimation is very consistent despite the increase in sample size. Table 2 represents the same information as Fig. 5, but provides the actual values for classification criteria that can be used as a reference table for future researchers.
Figure 5 The average of the median estimates from simulation.
Figure 6 Average of the estimated median and classification criteria.
Figure 7 The coefficient of variation of the classification criteria from simulation (Different sample size).
Figure 8 The coefficient of variation of the classification criteria from simulation (Different censor point).
Table 2 Average of the classification criteria
Average of the classification criteria
Baseline N = 100 N = 200 N = 300 N = 400 N = 500
10 3.5 3.5 3.4 3.4 3.5
17 4.2 4.2 4.2 4.2 4.2
34 6.8 6.8 6.8 6.8 6.8
51 11.1 11.1 11 11 11
68 18.2 18.1 17.9 17.8 17.8
Fig. 6 summarizes the results of classification criteria in simulated data sets with different sample size. While the sample size increases, the mean of the cutoff points approaches the mean of the median estimation. That is, when the sample size is reasonable large (n = 500), the cut off points are almost equivalent to the estimated median.
Fig. 7 summarizes the coefficient of variation (CV) for the distribution of classification criteria, in simulated data sets with different sample size. The lower the CV is, the higher the precision of the estimation is. This plot shows two trends. First, the CV is lowest at the median baseline level and increases toward the extreme values in both directions, as expected. For example, in a sample of 100 patients, the CV of the estimated cutoff point is 12.5% at baseline 34, 20.2% at baseline 10 and 22.8% at baseline 68. That is, the precision of this estimation is the highest when the baseline is around 34 and reduced toward both extremes. Second, we also found that the CV decreases with the sample size. For example, at baseline 34, the CV is 12.5% for 100 scores and 5.3% for 500 scores. This indicates that the precision of the estimation increases with a larger sample size.
Fig. 8 summarizes the CV for the distribution of classification criteria, in simulated data sets with different censoring points. While sample sizes being fixed at 500, we found that the CV increases with higher censoring level. This indicates that the precision of the estimation increases with a lower censoring point. Therefore censoring postoperative scores at 0.9 is preferred over censoring at a higher level.
Discussion
In the past decade the orthopaedic community has shifted toward the inclusion of patient-based measures of outcome assessments [22]. It was typical of earlier orthopedic practice that the patient's perspective received less attention than did clinician's measures of disease and impairment [23,24]. Clinicians used complication rates, mortality, most frequently revision rates and clinical judgment to assess the degree of improvement [25]. Since THA, in most cases, aims explicitly to improve HRQOL, using HRQOL measures as endpoint in orthopaedic research on evaluation of treatment outcome is now seen as a necessity to fully understand the effects of this intervention [26].
THA is an effective treatment by any of the distribution-based measures of responsiveness [6]. Yet, there are large variations reported in treatment outcome. Why some patients do better than others post-operation? Group level perspective on evaluation of treatment outcome has provided very little evidence contributes to our understanding of this question.
Most reports of the HRQOL outcome of THA use distribution-based approaches that test the significance of the change due to treatment [22]. However, distribution-based approaches are based on the statistical characteristics of the sample. For example, paired t-statistics are frequently used to estimate the statistical significance of the change [27]. The problem with using the t-test as a measure of change is that it focuses exclusively on the significance which will inevitably increase with sample size [28]. A different problem exists in determining the minimal clinically significant difference for THA. In clinical drug trials, a 9.3 point change in WOMAC functional score was accepted as a minimally significant improvement in arthritis symptoms [29]. But the 9.3-point change is too small to be applied to the outcomes of THA which typically show a 60–100% improvement over baseline [15,16]. The expected change in WOMAC functional scores after THA is four times larger than the minimal clinically important difference derived from drug trials in OA. Effect size (ES) and standard response mean (SRM) are also common measures for responsiveness at the group level. Cohen's criteria can be used to classify responsiveness as mild, moderate, and large [30]. But these statistics may be influenced by the heterogeneity of the sample. Moreover, Cohen's magnitude of effect does not suit the nature of orthopaedic surgery. An effect size larger than 0.8 is considered a "large effect". However, by that criterion, the majority of patients in our case study would be considered to have experienced a "large effect" both by ES and SRM statistics. Such criteria are inadequate for documenting the positive impacts of treatment. Characteristics of the baseline distribution will strongly influence the effect size, while variability of the change in the sample may influence the standard response mean.
We developed a method to classify the HRQOL outcome on an individual level. Group distributions can have a negligible mean difference with large variance. Therefore, the large differences that are important to individuals are not measured by group level, whereas the individual level takes them into account. This makes the individual perspective important for clinical treatment decisions [13,28]. We have shown that improvement after hip arthroplasty is not as big when the patients have a better preoperative score; therefore, postoperative outcomes are not evaluated at the group level but rather at each individual baseline level, so that for each individual patient an expected outcome can be generated.
We addressed the ceiling effect of the WOMAC instrument in the measurement of postoperative outcome of THA, as 10% of patients in our case study recorded a postoperative WOMAC score of 0. A ceiling effect occurs when a patient can improve only minimally or not at all. In the presence of a ceiling effect, the paper by Austin et al. suggests that the coefficient estimates from the left-censored regression model are better than the estimates from a least square regression [31]. We address the ceiling effect by implementing a linear regression of log-transformed WOMAC function score while treating postoperative scores as left-censored at 0.9. The regression model represents the relationship between baseline and postoperative outcome.
The estimated median of postoperative scores was chosen to distinguish between those who are able to benefit fully from treatment and those who are not. Due to the small sample size, the classification criteria in this case study is the lower boundary of the 95% confidence interval (CI) of the estimated median. Our study results agrees with the previous literature in that postoperative HRQOL scores were found to be strongly associated with their baseline values. We evaluated the changes in the WOMAC dimensions of pain, stiffness, and function from pre- to post operation. The effects of age, gender, and co-morbidity on follow-up WOMAC scores were not statistically significant, so these are excluded from the regression model.
The performance of the classification criteria was demonstrated using the Monte-Carlo simulation. The variation of the classification criteria will decrease with increasing sample size; likewise, the classification criteria become closer to the estimated median with increasing sample size. Thus, with a small sample set, researchers could use the lower boundary of 95% CI of the estimated median as the classification criteria. When there is a reasonable larger sample (bigger than 500), one could use the estimated median itself as the classification criteria.
The limitation of this research is that the estimated classification criteria were not validated in a different clinical setting. Instead, they were evaluated through simulation. Therefore, we are recommending that clinicians use only the methods rather than the actual values of the classification criteria until further research is done in this area.
Conclusion
The contribution of this paper is two-fold. First, the criteria for classify individual treatment outcome adjusted by baseline score was proposed. The development of these classification criteria also addresses the ceiling effect of the HRQOL measurement. Second, the performance of the classification criteria was found to be more precise with a reasonable larger sample size (n > 500). Vancouver Hospital (VH) is a tertiary referral centre and teaching hospital for the University of British Columbia (UBC). The demographics of arthroplasty patients, however, are not different from elsewhere. The study result is expected to be generalizable to a similar clinical setting.
This paper provides intuitive criteria for classifying HRQOL outcomes based on individual scores before surgery. The result of this method is an individual outcome which can serve as a standard advice for patient counseling based on HRQOL status at consultation. It gives orthopaedic researchers a means of defining "success" of effective surgery. In the future, we will evaluate this method in different populations and with other HRQOL instruments such as Oxford Hip Score and the Short Form 12 questionnaire.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
Analysis of data, interpretation and the original draft were completed by Min Xu. Donald Garbuz conceived the study, participated in the design and contributed to clinical conception and interpretation. Lisa Kuramoto performed the Monte-Carlo simulation in the statistical analysis. Boris Sobolev also participated in the design, provided critical evaluation of methodological content and revision of the manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We thank Mr. James Latteier and Mr. Francisco Luna for data collection. Findings previously presented at annual meeting of American College of Rheumatology, San Antonio, Texas, 2004. Abstract Title: Effect of Delays on Individual Quality of Life Outcome after Primary Total Hip Arthroplasty.
==== Refs
Cabanela ME Campbell DC 2ndHenderson ED Total joint arthroplasty. The hip Mayo Clin Proc 1979 54 559 63 470452
NIH consensus conference Total hip replacement. NIH Consensus Development Panel on Total Hip Replacement JAMA 1995 273 1950 6 7783307 10.1001/jama.273.24.1950
Harris WH Sledge CB Total hip and total knee replacement (1) N Engl J Med 1990 323 725 31 2201916
Harris WH Sledge CB Total hip and total knee replacement (2) N Engl J Med 1990 323 801 7 2136367
Jones CA Voaklander DC Johnston DW Health related quality of life outcomes after total hip and knee arthroplasties in a community based population J Rheumatol 2000 27 1745 52 10914862
Laupacis A Bourne R Rorabeck C The effect of elective total hip replacement on health-related quality of life J Bone Joint Surg Am 1993 75 1619 26 8245054
Liang MH Cullen KE Poss R Primary total hip or knee replacement: evaluation of patients Ann Intern Med 1982 97 735 6753682
Fitzpatrick R Norquist JM Dawson J Jenkinson C Rasch scoring of outcomes of total hip replacement J Clin Epidemiol 2003 56 68 74 12589872 10.1016/S0895-4356(02)00532-2
Gartland JJ Orthopaedic clinical research. Deficiencies in experimental design and determinations of outcome J Bone Joint Surg Am 1988 70 1357 64 3182887
Ethgen O Bruyere O Richy F Dardennes C Reginster JY Health-related quality of life in total hip and total knee arthroplasty. A qualitative and systematic review of the literature J Bone Joint Surg Am 2004 86-A 963 74 15118039
Bowling A Measuring health. A review of quality of life measurement scales 1992 Buckingham, England: Open University Press
Hageman WJ Arrindell WA Establishing clinically significant change: increment of precision and the distinction between individual and group level of analysis Behav Res Ther 1999 37 1169 93 10596464 10.1016/S0005-7967(99)00032-7
Samsa G Edelman D Rothman ML Williams GR Lipscomb J Matchar D Determining clinically important differences in health status measures: a general approach with illustration to the Health Utilities Index Mark II Pharmacoeconomics 1999 15 141 55 10351188
MacWilliam CH Yood UM Verner JJ McCarthy BD Ward RE Patient-related risk factors that predict poor outcome after total hip replacement Health Serv Res 1996 31 623 638 8943994
Fortin PR Clarke AE Joseph L Liang MH Tanzer M Ferland D Outcomes of total hip and knee replacement: Preoperative functional status predicts outcomes at six months after surgery Arthritis Rheum 1999 42 1722 1728 10446873 10.1002/1529-0131(199908)42:8<1722::AID-ANR22>3.0.CO;2-R
Fortin PR Penrod JR Clarke AE St-Pierre Y Joseph L Belisle P Timing of total joint replacement affects clinical outcomes among patients with osteoarthritis of the hip or knee Arthritis Rheum 2002 46 3327 30 12483739 10.1002/art.10631
Altman R Alarcon G Appelrouth D Bloch D Borenstein D Brandt K The American College of Rheumatology criteria for the classification of reporting of osteoarthritis of the hip Arthritis Rheum 1991 34 505 514 2025304
Bellamy N Kirwan J Boers M Brooks P Strand V Tugwell P Recommendations for a core set of outcome measures for future phase III clinical trials in knee, hip, and hand osteoarthritis. Consensus development at OMERACT III J Rheumatol 1997 24 799 802 9101522
Bellamy N Buchanan W Goldsmith CH Campbell J Stitt LW Validation study of WOMAC: A health status instrument for measuring clinically important patient relevant outcomes to antirheumatic drug therapy in patients with osteoarthritis of the hip or the knee J Rheumatol 1988 15 1833 40 3068365
Charnley J The long-term results of low-friction arthroplasty of the hip performed as a primary intervention J Bone Joint Surg Br 1972 54 61 76 5011747
Tobin J Estimation of Relationships for Limited Dependent Variables Econometrica 1958 26 24 36
Swiontkowski MF Outcomes measurement in orthopaedic trauma surgery Injury 2003 26 653 657 8745800 10.1016/0020-1383(95)00127-1
Cole B Finch E Gowland C Physical Rehabilitation Outcome Measures 1994 Toronto, Ontario, Canadian Physiotherapy Association
Ponzer S Nasell H Bergman B Functional outcome and quality of life in patients with Type B ankle fractures: A two-year follow-up study J Orthop Trauma 1999 13 363 368 10406704 10.1097/00005131-199906000-00007
Beaton DE Bombardier C Katz JN Wright JG A taxonomy for responsiveness Journal of Clinical Epidemiology 2001 54 1204 1217 11750189 10.1016/S0895-4356(01)00407-3
Rejeski WJ Shumaker S Knee osteoarthritis and health-related quality of life Med Sci Sports Exerc 1994 26 1441 5 7869876
Liang M Larson MG Cullen KE Comparative measurement efficiency and sensitivity of five health status instruments for arthritis research Arthritis Rheum 1985 28 542 547 4004963
Crosby RossD Kolotkin RonetteL Williams G Rhys Defining clinically meaningful change in health-related quality of life Journal of Clinical Epidemiology 2003 56 395 407 12812812 10.1016/S0895-4356(03)00044-1
Ehrich EW Davies GM Watson DJ Minimal perceptible clinical improvement with the Western Ontario and McMaster Universities Osteoarthritis Index questionnaire and global assessments in patients with osteoarthritis J Rheumatol 2000 27 2635 41 11093446
Cohen J Statistical power analysis for the behavioral sciences 1988 2 New York: Academic Press
Austin PC Escobar M Kopec JA The use of the Tobit model for analyzing measures of health status Qual Life Res 2000 9 901 10 11284209 10.1023/A:1008938326604
|
16144550
|
PMC1242235
|
CC BY
|
2021-01-04 16:32:04
|
no
|
BMC Musculoskelet Disord. 2005 Sep 6; 6:48
|
utf-8
|
BMC Musculoskelet Disord
| 2,005 |
10.1186/1471-2474-6-48
|
oa_comm
|
==== Front
BMC Musculoskelet DisordBMC Musculoskeletal Disorders1471-2474BioMed Central London 1471-2474-6-491617429710.1186/1471-2474-6-49Research ArticleComparison of plasma endothelin levels between osteoporotic, osteopenic and normal subjects Muratli Hasan Hilmi [email protected]Çelebi Levent [email protected] Onur [email protected]çimoğlu Ali [email protected] 3rd Orthopaedics and Traumatology Clinic, Ankara Numune Education and Research Hospital, Talatpaşa Bulvarı, Sıhhiye, Ankara, Turkey2005 20 9 2005 6 49 49 6 1 2005 20 9 2005 Copyright © 2005 Muratli et al; licensee BioMed Central Ltd.2005Muratli et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
It has been demonstrated that endothelins (ET) have significant roles in bone remodeling, metabolism and physiopathology of several bone diseases. We aimed to investigate if there was any difference between the plasma ET levels of osteoporotic patients and normals.
Methods
86 patients (70 women and 16 men) with a mean age of 62.6 (ranges: 51–90) years were included in this study. Patients were divided into groups of osteoporosis, osteopenia and normal regarding reported T scores of DEXA evaluation according to the suggestions of World Health Organization. According to these criteria 19, 43 and 24 were normal, osteopenic and osteoporotic respectively. Then total plasma level of ET was measured in all patients with monoclonal antibody based sandwich immunoassay (EIA) method. One-way analysis of variance test was used to compare endothelin values between normals, osteopenics and osteoporotics.
Results
Endothelin total plasma level in patients was a mean of 98.36 ± 63.96, 100.92 ± 47.2 and 99.56 ± 56.6 pg/ml in osteoporotic, osteopenic and normal groups respectively. The difference between groups was not significant (p > 0.05).
Conclusion
No significant differences in plasma ET levels among three groups of study participants could be detected in this study.
==== Body
Background
The endothelins (ET) are a family of 21-aminoacid peptides consisting of endothelin-1 (ET-1), the related peptides ET-2 and ET-3 [1]. In addition to being among the most potent vasoconstrictor agents known, ET have been found to possess a wide range of pharmacological activities on different tissues [1-3]. The close proximity of cells on the bone surface to vascular endothelial cells exposes bone cells to endothelial cell products such as the polypeptide ET. It is well recognized that ET play an important role in bone metabolism [4-8].
The receptors for ET on the osteoclasts, osteoblasts and their intracellular signal systems were predominantly found out by detailed in vitro and in vivo studies [3,8,9]. ET-A and ET-B receptor subtypes are expressed in bone cells. Stimulation of phospholipids turnover and activation of tyrosine kinases are used by ET as a major way while transducting the intracellular signals [3,9]. It was also demonstrated that osteoblasts, osteoclasts and osteocyts contain measurable amount of ET [10,11].
It was shown that ET regulates bone blood flow in the intact vascularized bone preparations [12]. However ETs' effects are not limited with their vasoactivity in the bone tissue. ET stimulates proliferation of the capillary endothelial cells [2,13-15], osteoblasts and osteoprogenitor cells [3,16]. They also stimulate differentiation of osteoprogenitor cells to osteoblasts [3,4,6].
Osteoblastic activity is increased by ET as this effect is demonstrated by stimulation of synthesis of collagen and non-collagen proteins [6] as well as osteocalcin and osteopontin messages in bone tissue [17].
It was demonstrated that ET has also certain interactions with 1,25 dihydroxyvitamin D3. Upon in vitro observations it has been addressed that ET together with vascular endothelial growth factor (VEGF) and 1,25 dihydroxyvitamin D3 may have an in vivo activity on bone formation and remodeling process [7].
There are some controversial reports regarding the effects of ET on bone resorption. They clearly inhibit motile process of the osteoclasts [5,8]. Osteoclastic bone resorption is inhibited by ET with similar doses that produce vasoconstriction [5]. Furthermore ET inhibits parathyroid hormone secretion in the parathyroid adenoma cells [18]. However along with these antiresorptive effects it was shown that they lead to prostaglandin (PG) [19,20] and interleukin-6 (IL-6) [21] related stimulation of resorption.
There are also controversial reports about ET effects on the mineralization process of bone. It is thought that mineralization is inhibited by ET through the stimulation of ET-A receptor [22,23]. On the contrary blockage of ET-A receptors are reported to cause osteopenia in experimental studies [24].
In the basis of above mentioned effects of ET in the bone, we thought that ET may have an important role in the physiopathology of osteoporosis and according to our investigation there is no epidemiologic study available in the literature, in particular to investigate the associations between ET and bone mineral density. We aimed to find out if there was any difference of the ET plasma levels between osteoporotic and normal people.
Methods
Groups of patients
242 patients who were over 50 years of age were referred by us to examine in our hospital's radiology department just for screening by dual energy X-ray absorbsiometry (DEXA) during from March to June 2004 were invited to the present study as soon as the results of DEXA were obtained. Patients with systemic diseases (diabetes, hypertension, renal disease, or clinical manifestation of atherosclerosis or known another diseases) and patients with abnormal laboratory results (regarding routine hemogram parameters and routine biochemical test) were planned to excluded. Other exclusion criteria were receival of any medication for osteoporosis previously or another drug in the last 3 months before the study and smoking or drinking alcoholic beverages for at least 48 hours before the blood sample receival. Presence of any anamnestic or clinical signs of osteoporosis (pain and previous fractures) and presence of any differences from reference values of the biochemical markers of bone remodeling in the patients with normal bone density according to our accepted criteria described below was accepted as another exclusion criteria.
The study was performed cross-sectionally. After first interview 12 of 242 invited patients refused to participate in the study. 44 patients who had known systemic diseases (diabetes, hypertension, renal disease, or clinical manifestation of atherosclerosis) and 22 patients who had received drugs in the last 3 months before the study were excluded at the beginning. 24 patients who received any treatment for osteoporosis before the study were also excluded. 16 participants who were evaluated as normal regarding bone mineral densitometry evaluation but who had any anamnestic or clinical signs of osteoporosis (pain and previous fractures) were also excluded from the study.
Remaining 124 patients who accepted to join our study were informed of the nature of the study. Consent was obtained from each participant. Then all patients were analyzed on clinical and biochemical basis. Then systemic blood pressures were measured in all other participants and 16 patients were excluded because of high blood measurement. Then Complete blood count and biochemical profiles including routine biochemical tests and biochemical markers of bone remodeling were assessed and 20 patients were excluded because of the pathological findings in this analysis. Although they were instructed not to use, 2 patients who smoked or drank alcoholic beverages in the period of 48 hours before the blood sample collection were excluded from the study.
At the end of these initial evaluation procedures remaining 86 patients were included in this study according to our accepted criteria. There were 16 males and 70 females. Mean age was 62.6 (ranges: 51–90) years. Patients were divided into 3 groups regarding reported T scores in DEXA evaluation. Consistent data base regarding young normal value and the population standard deviation were used in order to calculate T score. T-scores less than -2.5 on either total lumbar spine or total hip were accepted as osteoporosis, while scores between -1 and -2.5 were accepted as osteopenia and scores above -1 were accepted as normal according to the suggestions of World Health Organization (WHO) [25]. According to suggested criteria of WHO [25] 19 of 86 were normal, 43 were osteopenic and 24 were osteoporotic. All patients' demographic and anthropometrical characteristics were noted.
Methods
After the insertion of a teflon cannula into the antecubital vein all subjects remained recumbent for at least 30 minutes. Then 5 ml of venous blood was withdrawn in vacutainer K2-EDTA plasma tubes from all patients 8 hours after overnight fasting. Blood samples were centrifuged immediately for 15 minute at 2000 × g. Then plasma samples were stored frozen at -80°C until EIA.
We had quantified the total amount of human ET fasting plasma level by using commercially available Endothelin-1 EIA (Endothelin EIA Kit, Catalog No:583151, Cayman Chemical Company, Michigan, USA) following instructions of the manufacturer. This immunometric assay is based on a double-antibody 'sandwich' technique and permits endothelin measurements within the range of 0–250 pg/ml, typically with a limit of detection of 1,5 pg/ml. Monoclonal anti ET-1 antibody in this kit had a cross reactivity of 100% with ET-2 and 100% with ET-3. Samples were assessed with no prior purification.
The intra- and inter-assay coefficients of variation of the method were 5 and 6%, respectively.
In this method monoclonal antibody specific to endothelin and acetylcholinesterase: Fab' Conjugate (AChE:Fab') bind to different epitopes on the Endothelin-1 molecule and forming sandwich. This sandwich is immobilized on the plate so the excess reagents are washed away. The concentration of analyte is detected by measuring the enzymatic activity of the AChE by adding Ellman's Reagent which contains the substrate for AChE. Addition of Ellman's Reagent produces a yellow-colored product which can be measured spectrophotometrically. The intensity of the color is directly proportional to the amount of bound conjugate which in turn is proportional to the concentration of the Endothelin.
Enzyme immunometric analysises were run twice from the same sample (in Düzen Laboratories Chain, Ankara, Turkey).
Data analysis
Data analyses were done by SPSS for Windows version 11.5. Prior to the analysis, all the data were examined for accuracy of data entry and fit between their distributions and the assumptions of univariate analysis. To improve pairwaise linearity and to reduce the extreme skewness and kurtosis, the z score for all variables was computed. It was found that all dependent variables are normally distributed.
Data was analyzed using two-way analysis of variance (two-way ANOVA) to assess statistical significance. Independent variables were group of subject (osteoporotics, osteopenics and normals), and gender (male and female); dependent variable was endothelin value.
This analysis was also used to compare weight, height and body mass index parameters between the groups of subjects.
By using the formula y = y'-b.(x-x') (y: new endothelin, y': old endothelin, b: regression coefficient x: weight or height or body mass index or age, x': mean weight or height or body mass index or age) we had made endothelin independent of weight, height and body mass index because these variables were treated as covariate variables. Osteoporotics, osteopenics and normals were compared for statistical significance in terms of endothelin levels using one-way ANOVA test after controlling for covariates. Two-way ANOVA test was used to confirm the difference of endothelin level between the males and females for each group separately after controlling for covariates. A value of p < 0.05 was considered as significant.
Pearson Product Moment Correlation coefficient analysis was used to detect relation between age, body mass index, height, weight, T scores and endothelin levels. All values were given as mean ± S.D.
Results
Age, gender and anthropometrical parameters of the groups were summarized in Table 1.
Table 1 Demographical and anthropometrical characteristics of the groups.
Number of cases Gender Age (years) Mean ± S.D. Height (cm) mean ± S.D. Weight (kg) mean ± S.D. BMI (kg/m2) mean ± S.D.
M F
Osteoporotics 24 4 20 65.95 ± 9.4a 153.29 ± 7.69b 65.79 ± 12.4c 28.02 ± 5.14a
Osteopenics 43 7 36 62.60 ± 7.9 156.09 ± 6.48 75.90 ± 11.72 31.19 ± 4.68
Normal 19 5 14 58.20 ± 7.6 160.0 ± 13.27 83.15 ± 11.5 33.01 ± 6.61
a: There is a significant difference between normals, osteopenics and osteoporotics (p < 0.01)
b: There is a significant difference between normals, osteopenics and osteoporotics (p < 0.05)
c: There is a significant difference between normals, osteopenics and osteoporotics (p < 0.001)
Average systemic blood pressure of study participants was 126 ± 8 (Mean ± S.D.) mmHg and 75 ± 9 mmHg for systolic and diastolic levels.
Endothelin fasting plasma levels
Endothelin fasting plasma levels were found comparable in both runs of analysises. Unadjusted ET total plasma levels were a mean of 98.36 ± 63.96 pg/ml in osteoporotic group, a mean of 100.92 ± 47.2 pg/ml in osteopenic group and a mean of 99.56 ± 56.6 pg/ml in normal group. The difference between groups was not significant. (p > 0.05) (Figure 1)
Figure 1 Plasma ET levels (mean ± S.D.) before the adjustments for age, weight and height of each group are presented diagrammatically.
Levels according to gender
In men with osteoporosis mean unadjusted ET level was 185.7 ± 17.2 pg/ml and this was significantly higher than in osteopenic men (124.8 ± 59.6 pg/ml) and in normal men (93.0 ± 50.1 pg/ml) (p < 0.05). In women there was not any significant difference between groups (normal:102.0 ± 60.7 pg/ml, osteopenics: 94.7 ± 42.7 pg/ml, osteoporotics: 79.9 ± 53.8 pg/ml, p > 0.05). (Figure 1)
Independent of osteoporotic status mean ET level was significantly higher in men (130.1 ± 58.7 pg/ml) than women (91.5 ± 50.2 pg/ml). (p < 0.05)
In osteoporotics mean ET level was significantly higher in men than in women. (p < 0.001). In osteopenic men mean ET level was higher than women but it was not significant (p > 0.05). In normal group mean ET level was lower in men than in women but it was not significant too (p > 0.05).
Levels according to age
When adjusted to age, ET levels did not differ significantly between groups. (p > 0.05)
Levels according to anthropometrical parameters
Osteoporotics had significantly lower values of weight than normals and osteopenics. (p < 0.001) Osteoporotics were significantly shorter than normals and osteopenics. (p < 0.05) As a result of these osteoporotics had significantly lower body mass index than normals and osteopenics. (p < 0.01)
After separately adjustments to body weight, length and body mass index, ET levels did not differ significantly between groups. (p > 0.05)
Regardless of groups there was no significant correlation between neither weight nor height and ET values. (p > 0.05)
In normal groups no correlation was found between weight, length, body mass index and ET levels. (p > 0.05) In osteopenic patients we have found negative correlation of weight (r:-0.36, p < 0.05), body mass index (r:-0.45, p < 0.01) and ET values. In osteoporotics we have found positive correlation of height and ET levels. (r:0.41, p < 0.05)
Other correlation studies
Regardless of groups and in osteopenics and osteoporotics there was no correlation between vertebral or hip T scores and ET values (p > 0.05) but a correlation between hip T scores and ET values of normal group (r:-0.5, p < 0.05).
Discussion
After demonstration of ETs' important effects on bone tissue it was begun to be considered that they may have certain roles in the physiopathology of some clinical entities [26,27]. As a matter of fact Tarquini et al. measured the circulating ET-1 levels of patients with Paget Disease in which the bone turn over is extremely increased and they found out that ET-1 level was significantly increased in patients with Paget Disease when compared to controls [27]. So they believed that ET-1 may have a role in the physiopathology of this disease and it could be used as a marker.
We investigated if ET had a role in the physiopathology of osteoporosis. We found out that comparison regardless of gender among osteoporotics, osteopenics and normals and comparison of female osteoporotics, osteopenics and normals yielded no significant differences regarding plasma ET levels. In addition regardless of groups (according to suggested criteria of WHO [25]) no correlation was found between vertebral or hip T scores and ET values. Although plasma ET levels of osteoporotic men were found significantly higher than normal men we believed that it could be speculative to make conclusion with these findings because there were only 4 osteoporotic men in the series.
Studies about the men osteoporosis demonstrated us that although total estradiol levels do not change substantially over life in men, bioavailable estradiol levels decrease to 50% of the levels in young men in the older ages. It is thought that this decline in bioavailable estradiol levels may be the major cause of bone loss in elderly osteoporotic men [28,29]. In previous laboratory studies it was shown that estrogens down regulated ET-1 both through the secretion from the vascular endothelial cells and m-RNA expression levels [30,31]. Upon these observations we believed that possible reason for higher plasma ET levels of osteoporotic men then the normal men in our study may be because of the lower bioavaliable estrogen concentration of the males in these ages and as a result possible decrease of estrogens effect in down regulation in ET and consequently increase of ET amount and effects. In fact bone loss is more accelerated in women after menopause as a result of a decline in circulating estrogens levels then the men in the same ages period [32]. However considering presence of no difference in the plasma ET levels between the osteoporotics, osteopenics and normals in the women population of our study participants it is not possible to say same mechanism is true for women regarding the ET and estrogens interaction.
Regarding osteoporosis physiopathology there are many effects of ET which can cause bone resorption and inhibition on the mineralization process as follows. ET-1's effect through the PG system [19,20] and IL-6 expression [21] are in favor of stimulating bone resorption. Recently it was described that expression of mRNA of PG endoperoxide G/H synthase is stimulated by ET-1 and this stimulation is a way of leading increase of PGE-2 production [20]. Tatrai and Stern [19] showed that ET-1 modulates the intracellular calcium signalization of PGE-1 and if the cells confronts with ET-1 then PGE-1 comes out. As a result ET-1 causes bone resorption depended with PGs.
Hierl et al. [21] showed that ET-1 has a dose dependent stimulatory effect on IL-6 expression in human osteblastic cell (HOC) cultures. IL-6 leads to bone resorption potently and it's this action was described with detailed in vivo and in vitro studies. And ET-1s' bone resorptive effect in the cell culture environment was mainly attributed to its effect on the IL-6 expression stimulation.
There are also reports about inhibitory activity of ET on the mineralization process of bone. Hiruma et al. [22] demonstrated that calcium deposition into the bone cells is decreased by ETs in rat calvarial osteoblast like cell cultures. They concluded that mineralization process in the osteoblasts may be inhibited by ET-1 through the ET-A receptor. Inoue et al. [23] also reported similar findings.
Considering findings of reports cited above which are in favor of ET's stimulatory effect on bone resorption and inhibitory effects on the mineralization process on bone tissue we thought ET can be the important peptide in the osteoporosis physiopathology and we thought we can find differences in the plasma ET levels between osteoporotics and normal subjects. But our findings do not support this idea.
According to manufacturer's instructions normal levels of ET-1 in human plasma are below the detection limit of the kit which we used; therefore purification and concentration of the sample is necessary for accurate measurement of ET-1 levels. Considering 100% cross reactions of this kit for all ET subtypes including ET-1, ET-2 and ET-3 and performing our analysis without prior purification process it should be addressed that our measurements reflects total endothelin measurements, not ET-1 alone. We did not perform purification because manufacturer states that samples can be assayed with no prior purification in general and they suggest performing purification process only for samples containing low concentration of endothelin (0–50 pg/ml). They also state that samples must be >50 pg/ml in order to be assayed accurately with this kit and all average ET levels of our groups was already in this range. In addition all samples obtained from both normals and pathologics regarding bone densitometric evaluation were evaluated with the same method, without prior purification, and all obtained measurements were within the detection range of this kit (0–250 pg/ml). So we believe that in the evaluation of our results and comparison of these findings with other studies these points should be taken into consideration.
Our study has certain limitations. Firstly a number of study participants were relatively small because osteoporosis is a complex disease, known to be affected by many factors such as age, gender, adiposity, menopause etc. We believe that subsequent studies should be performed with larger number of participants especially for men. In addition although we only included patients with no known disease and with normal laboratory findings in the routine evaluation tests and normal blood pressure, it was not possible to know with this limited evaluations if these patients had any disease which are not possible to diagnose with our screening tests for this study and which may also cause changes of plasma ET level as mentioned and referenced below.
Although it is known by many studies that [27,33,34] increased level of ET concentrations can be detected in the plasma as a result of overproduction of ET released from pathologic tissues and/or due to hypervascularization associated with the lesion in osseous or non-osseous pathologies, it can be thought that systemic circulation may thus not entirely reflect local changes in the bones. So in the substantial studies ET concentrations at the site of osteoporotic bone tissue should also be evaluated by biopsies.
Conclusion
According to our findings there is no significant difference between the ostoeporotics, osteopenics and normals regarding plasma ET levels. We think that particularly estrogens, prostaglandins and interleukin levels should also be measured in similar study designs in order to discuss the role of ET in osteoporosis and the difference of ET levels between males and females.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
HHM conceived of the study, and participated in its design and coordination and helped to draft and write the manuscript.
LÇ carried out the bone mineral density and immunoassay measurement organization.
OH carried out the immunoassay study organization and statistical analysis.
AB participated in the design of the study and helped to draft the manuscript.
All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
==== Refs
Yanagisawa M Kurihara H Kimura S Tomobe Y Kobayashi M Mitsui Y Yazaki Y Goto K Masaki T A novel potent vasoconstrictor peptide produced by vascular endothelial cells Nature 1988 332 411 415 2451132 10.1038/332411a0
Briggs PJ Moran CG Wood MB Actions of endothelin-1, 2, and 3 in the microvasculature of bone J Orthop Res 1998 16 340 347 9671929 10.1002/jor.1100160310
Stern PH Tatrai A Semler DE Lee SK Lakatos P Strieleman PJ Tarjan G Sanders JL Endotelin receptors, second messengers and actions in bone J Nutr 1995 2028S 2032S 7602388
Kasperk CH Borcsok I Schairer HU Schneider U Nawroth PP Niethard FU Ziegler R Endothelin-1 is a potent regulator of human bone cell metabolism in vitro Calcif Tissue Int 1997 60 368 374 9075635 10.1007/s002239900245
Alam AS Gallagher A Shankar V Ghatei MA Datta HK Huang CL Moonga BS Chambers TJ Bloom SR Zaidi M Endothelin inhibits osteoclastic bone resorption by a direct effect on cell motility: implications for the vascular control of bone resorption Endocrinology 1992 130 3617 3624 1597159 10.1210/en.130.6.3617
Tatrai A Foster S Lakatos P Shankar G Stern PH Endotelin-1 actions on resorption, collagen and noncollagen protein synthesis, and phosphatidylinositol turnover in bone organ cultures Endocrinology 1992 131 603 607 1639010 10.1210/en.131.2.603
Wang DS Miura M Demura H Sato K Anabolic effects of 1,25-dihydroxyvitamin D3 on osteoblasts are enhanced by vascular endotelial growth factor produced by osteoblasts and by growth factors produced by endotelial cells Endocrinology 1997 138 2953 2962 9202240 10.1210/en.138.7.2953
Zaidi M Alam AS Bax BE Shankar VS Bax CM Gill JS Pazianas M Huang CL Sahinoglu T Moonga BS Role of the endothelial cell in osteoclast control: new perspectives Bone 1993 14 97 102 8392855 10.1016/8756-3282(93)90234-2
Suzuki A Oiso Y Kozawa O Effects of Endotelin 1 on phospholipase D activity in osteoblast-like cells Mol Cell Endocrinol 1994 105 193 196 7859925 10.1016/0303-7207(94)90169-4
Sasaki T Hong MH Endothelin-1 localization in bone cells and vascular endothelial cells in rat bone marrow Anat Rec 1993 237 332 337 8291686 10.1002/ar.1092370306
Sasaki T Hong MH Localization of endothelin-1 in the osteoclast J Electron Microsc 1993 42 193 196
Coessens BC Miller VM Wood MB Endothelin induces vasoconstriction in the bone vasculature in vitro: an effect mediated by a single receptor population J Orthop Res 1996 14 611 617 8764871 10.1002/jor.1100140416
Tsui JC Baker DM Biecker E Shaw S Dashwood MR Potential role of endothelin 1 in ischaemia-induced angiogenesis in critical leg ischaemia Br J Surg 2002 89 741 747 12027984 10.1046/j.1365-2168.2002.02100.x
Medinger M Adler CP Schmidt-Gersbach C Soltau J Droll A Unger C Drevs J Angiogenesis and the ET-1/ETA receptor system: immunohistochemical expression analysis in bone metastases from patients with different primary tumors Angiogenesis 2003 6 225 231 15041798 10.1023/B:AGEN.0000021395.43438.44
Wang X Tokuda H Hirade K Kozawa O Stress-activated protein kinase/c-Jun N-terminal kinase (JNK) plays a part in endothelin-1-induced vascular endothelial growth factor synthesis in osteoblasts J Cell Biochem 2002 87 417 423 12397601 10.1002/jcb.10323
Takuwa Y Masaki T Yamashita K The effects of the endothelin family peptides on cultured osteoblastic cells from rat calvariae Biochem Biophys Res Commun 1990 170 998 1005 2202304 10.1016/0006-291X(90)90491-5
Shioide M Noda M Endothelin modulates osteopontin and osteocalcin messenger ribonucleic acid expression in rat osteoblastic osteosarcoma cells J Cell Biochem 1993 53 176 180 8227190 10.1002/jcb.240530211
Eguchi S Hirata Y Imai T Kanno K Akiba T Sakamoto A Yanagisawa M Masaki T Marumo F Endothelin receptors in human parathyroid gland Biochem Biophys Res Commun 1992 184 1448 1455 1375462 10.1016/S0006-291X(05)80045-2
Tatrai A Stern PH Endothelin-1 modulates calcium signaling by epidermal growth factor, alpha-thrombin, and prostaglandin E1 in UMR-106 osteoblastic cells J Bone Miner Res 1993 8 943 952 8213257
Windischhofer W Zach D Fauler G Raspotnig G Kofeler H Leis HJ Involvement of Rho and p38 MAPK in Endotelin-1-Induced Expression of PGHS-2 m RNA in osteoblast-like cells J Bone Miner Res 2002 17 1774 1784 12369781
Hierl T Borcsok I Sommer U Ziegler R Kasperk C Regulation of Interleukin-6 expression in human osteoblastic cells in vitro Exp Clin Endocrinol Diabetes 1998 106 324 333 9792466
Hiruma Y Inoue A Shiohama A Otsuka E Hirose S Yamaguchi A Hagiwara H Endothelins inhibit the mineralization of osteoblastic MC3T3-E1 cells through the A-type endothelin receptor Am J Physiol 1998 275 R1099 1105 9756539
Inoue A Kamiya A Ishiji A Hiruma Y Hirose S Hagiwara H Vasoactive peptide-regulated gene expression during osteoblastic differentiation J Cardiovasc Pharmacol 2000 36 S286 289 11078400
Tsukahara H Hori C Hiraoka M Yamamoto K Ishii Y Mayumi M Endothelin subtype A receptor antagonist induces osteopenia in growing rats Metabolism 1998 47 1403 1407 9826221 10.1016/S0026-0495(98)90313-4
The WHO Study Group Assessment of fracture risk and its application to screening for postmenopausal osteoporosis Geneva, WHO Technical report series 1994 843
Mohammad KS Guise TA Mechanisms of osteoblastic metastases: role of Endothelin-1 Clin Orthop 2003 415 S67 74 14600594
Tarquini R Perfetto F Tarquini B Endotelin-1 and Paget's bone disease: is there a link? Calcif Tissue Int 1998 63 118 120 9685515 10.1007/s002239900500
Khosla S Melton LJ 3rdAtkinson EJ O'Fallon WM Klee GG Riggs BL Relationship of serum sex steroid levels and bone turnover markers with bone mineral density in men and women: a key role for bioavailable estrogen J Clin Endocrinol Metab 1998 83 2266 2274 9661593 10.1210/jc.83.7.2266
Khosla S Melton LJ 3rdAtkinson EJ O'Fallon WM Relationship of serum sex steroid levels to longitudinal changes in bone density in young versus elderly men J Clin Endocrinol Metab 2001 86 3555 3561 11502778 10.1210/jc.86.8.3555
Grande M Carlstrom K Stege R Pousette A Faxen M Estrogens affect Endothelin-1 m RNA Expression in LNCaP human prostate carcinoma cells Eur Urol 2002 41 568 574 12074801 10.1016/S0302-2838(02)00030-1
Wingrove CS Stevenson JC 17 beta-Oestradiol inhibits stimulated endothelin release in human vascular endothelial cells Eur J Endocrinol 1997 137 205 208 9272111 10.1530/eje.0.1370205
Bjornerem A Straume B Midtby M Fonnebo V Sundsfjord J Svartberg J Acharya G Oian P Berntsen GK Endogenous sex hormones in relation to age, sex, lifestyle factors, and chronic diseases in a general population: the Tromso Study J Clin Endocrinol Metab 2004 89 6039 6047 15579756 10.1210/jc.2004-0735
Asham E Shankar A Loizidou M Fredericks S Miller K Boulos PB Burnstock G Taylor I Increased endothelin-1 in colorectal cancer and reduction of tumour growth by ET(A) receptor antagonism Br J Cancer 2001 85 1759 1763 11742499 10.1054/bjoc.2001.2193
Inoue K Hirota M Kimura Y Kuwata K Ohmuraya M Ogawa M Further evidence for endothelin as an important mediator of pancreatic and intestinal ischemia in severe acute pancreatitis Pancreas 2003 26 218 223 12657945 10.1097/00006676-200304000-00002
|
16174297
|
PMC1242236
|
CC BY
|
2021-01-04 16:32:04
|
no
|
BMC Musculoskelet Disord. 2005 Sep 20; 6:49
|
utf-8
|
BMC Musculoskelet Disord
| 2,005 |
10.1186/1471-2474-6-49
|
oa_comm
|
==== Front
BMC PediatrBMC Pediatrics1471-2431BioMed Central London 1471-2431-5-341613733410.1186/1471-2431-5-34Research ArticleTourette syndrome and learning disabilities Burd Larry [email protected] Roger D [email protected] Marilyn G [email protected] Jacob [email protected] Department of Pediatrics, University of North Dakota School of Medicine and Health Sciences, Grand Forks, North Dakota, USA2 Department of Pediatrics, University of British Columbia, Vancouver, British Columbia, Canada3 Department of Neuroscience, University of North Dakota School of Medicine and Health Sciences, Grand Forks, North Dakota, USA2005 1 9 2005 5 34 34 13 12 2004 1 9 2005 Copyright © 2005 Burd et al; licensee BioMed Central Ltd.2005Burd et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Tourette Syndrome (TS) is a neurodevelopmental disorder of childhood. Learning disabilities are frequently comorbid with TS. Using the largest sample of TS patients ever reported, we sought to identify differences between subjects with TS only and subjects with TS and a comorbid learning disability.
Methods
We used the Tourette Syndrome International Consortium database (TIC) to compare subjects with comorbid Tourette Syndrome and learning disabilities (TS + LD) to subjects who did not have a comorbid learning disability (TS - LD). The TIC database contained 5,500 subjects. We had usable data on 5,450 subjects.
Results
We found 1,235 subjects with TS + LD. Significant differences between the TS + LD group and the TS - LD group were found for gender (.001), age onset (.030), age first seen (.001), age at diagnosis (.001), prenatal problems (.001), sibling or other family member with tics (.024), two or more affected family members (.009), and severe tics (.046). We used logistic modeling to identify the optimal prediction model of group membership. This resulted in a five variable model with the epidemiologic performance characteristics of accuracy 65.2% (model correctly classified 4,406 of 5,450 subjects), sensitivity 66.1%, and specificity 62.2%.
Conclusion
Subjects with TS have high prevalence rates of comorbid learning disabilities. We identified phenotype differences between the TS - LD group compared to TS + LD group. In the evaluation of subjects with TS, the presence of a learning disability should always be a consideration. ADHD may be an important comorbid condition in the diagnosis of LD or may also be a potential confounder. Further research on etiology, course and response to intervention for subjects with TS only and TS with learning disabilities is needed.
==== Body
Background
Tourette Syndrome (TS) is a complex developmental disorder defined by the childhood onset of motor and vocal tics with a longitudinal outcome of gradual improvement in most subjects [1-4]. The disorder is associated with increased prevalence rates of comorbid disorders, the most common of which is attention-deficit hyperactivity disorder (ADHD) [5,6]. Learning disabilities (LD) and obsessive-compulsive disorder (OCD) or obsessive and compulsive behaviors (OCB) are also common [7-10].
In previous work we have demonstrated that over time the presence of comorbidity is an important factor in syndromal severity and in the level of impairment from the disorder [11]. While several disorders have been demonstrated to occur as a manifestation of the broad TS phenotype (OCD, ADHD) the role of several other conditions is currently a contentious issue in the definition of that broad phenotype [8-10,12]. Previous research has demonstrated increased prevalence of LD in subjects with TS but the role of LD as a manifestation of the broad TS phenotype is not yet settled [8,10,13-25].
Previous research on TS and comorbid LD has relied on relatively small samples usually selected from 1 or 2 clinic sites [13-20,22-26]. The limitations of small sample size and selected catchments for these studies have led to concerns about the generalizability of the results of these studies. In order to minimize these limitations, we have elected to utilize a large international population of cases of TS and TS with comorbid LD to examine differences in subjects with TS without a comorbid learning disability and subjects with TS and a comorbid learning disability. We utilized data from the Tourette Syndrome International Consortium (TIC) to examine differences between subjects with Tourette Syndrome and learning disabilities (TS + LD) and subjects with Tourette Syndrome who did not have a comorbid learning disability (TS - LD). We have utilized data from this consortium for multiple other studies of TS including comorbid TS and pervasive developmental disorders, prediction of tic severity, and hereditary factors in tic severity [27].
Methods
The study population was comprised of consecutive subjects entered into the database since its inception. The 5,500 subjects in this study include the 3,500 subjects previously reported in the paper by Freeman and colleagues [28].
Registry reporting sites
Thirty-six sites have over 50 subjects and seventeen sites have over 100 subjects. Twenty-four sites have less than 50 cases and 19 are currently inactive. The geographic distribution of the consortium cases was: Canada 40.6%, United States 22.6%, Europe 25.1%, Middle East 3.6%, South America 1.8%, Asia 3.0%, Australia 3.0%, and Africa 0.3%. The clinicians who submit cases to the registry are either physicians (nearly all) or psychologists.
Subject selection
All subjects entered in the registry met the criteria for TS from the Tourette Syndrome Classification Study Group [29]. Each subject was reported utilizing a structured reporting format [see Additional file 1] to assure comparability of the data. A learning disabilities diagnosis entered into the registry was inclusive of specific learning disorders as defined in the DSM-IV, through the less precise and less verifiable category of learning disorders NOS [30]. The diagnosis of LD would only rarely include individuals with mental retardation (MR). In this paper we excluded the few subjects with both LD and MR. The TIC database does not have data on the proportion of subjects diagnosed with LD after psychometric testing or the proportion where LD was a clinical diagnosis or both.
Subject data were then forwarded to the consortium where each case was reviewed for inconsistencies prior to entry into the database. If errors were identified or suspected, the case file was returned to the clinical site for review. This data is not verified beyond the identification of errors in either data entry from the submitted form (data entry control procedures are utilized to minimize these errors) or unless an error is detectable by the field entry restriction values for each variable.
We utilized the method of Spady et al. for management of summary data [31]. As in most reported cohorts we were not able to detect diagnostic error. To minimize the potential impact of errors we do not report values from individual clinical sites. Thus, the results represent pooled data from multiple sites to reduce any potential impact from systematic or inadvertent error from any one site. This data pooling increases accuracy but does so by obscuring between site differences and as a result decreases precision.
Statistical analysis
For this study we had usable data on 5,450 subjects. Continuity corrected Chi-Square was used to test the association between gender, age of onset, age first seen, age diagnosed, clinician type, perinatal problems, heredity of TS, severity of TS, and fourteen comorbidities by group (TS + LD) and (TS - LD). Since LD was used to define one of the groups in this study LD was not counted as a comorbid disorder in the study. Thus, the variable comorbidity is comprised of all other comorbid disorders available from the dataset. Observations with missing values were deleted for each univariate analysis. After completion of the univariate analysis, we used logistic regression modeling to identify the optimal set of prediction variables to predict group membership TS + LD. We used the epidemiologic performance characteristics of accuracy, sensitivity and specificity to select a final logistic model.
Results
Of the 5,450 subjects with TS, the TS + LD group was comprised of 1,235 subjects (22.7%) while the TS - LD group had 4,215 (76.3%) subjects. In the TS - LD group, 3,774 subjects (69.2%) had other comorbid conditions. Four hundred and forty-one patients of the TS-LD group (8.1%) had no comorbid disorders or conditions. In the TS + LD group the average number of comorbidities other than LD was 3.04 (s.d. 2.07). The analysis includes all the variables included in the database. The average age of onset of TS was 6.37 years (s.d. = 2.82) and was determined by parental report of tic onset. The average age of diagnosis was 13.43 years (s.d. = 10.0). The TIC Registry population was 81.4% male, while 19.3% had perinatal problems and 53.9% had at least one family member with a history of tics or TS.
Table 1 shows variables with significant associations for TS + LD. Subjects with TS + LD had an increased proportion of males (p < .001), and have an age of onset of TS before eight years of age (p = .030). They also were first seen before 18 years of age (p < .001) and were diagnosed before they were thirteen years of age (p < .001). The average age of onset for those with TS + LD was 6.14 (s.d. = 2.56) and was comparable to 6.44 years (s.d. = 2.88) for those with TS - LD. The average age first seen for those with TS + LD was 12.5 (s.d. = 7.5), while it was over three years later (mean = 15.7, s.d. = 11.6) for those with TS - LD. The average age for diagnosis in TS + LD was also three years earlier (mean = 11.4, s.d. = 7.1) compared to those with TS - LD (mean = 14.0, s.d. = 10.6). Seventy-four percent of the cases were diagnosed by a psychiatrist and 19% were diagnosed by a neurologist (p < .001).
Table 1 Between group comparisons in 5,450 subjects with Tourette Syndrome and Learning Disabilities (TS + LD) and Tourette Syndrome without learning disabilities(TS-LD) by gender, age, perinatal problems, and family history of tics.
TS + LD TS - LD
n (%) n (%) p
Gender
Female 159 (12.9) 857 (20.3) <.001
Male 1,076 (87.1) 3,355 (79.7)
Age of Onset of TS
<= 4 261 (25.1) 922 (24.8) .030
5 to 7 517 (49.7) 1,709 (45.9)
8 or Older 263 (25.3) 1,089 (29.3)
Age First Seen
<= 17 1,071 (87.4) 3,118 (74.6) <.001
>17 154 (12.6) 1,063 (25.4)
Age Diagnosed
<= 8 416 (35.8) 1,275 (32.5) <.001
9 to 12 488 (42.0) 1,366 (34.8)
13 or Older 257 (22.1) 1,284 (32.7)
Perinatal Problems
Yes 280 (26.9) 607 (17.1) <.001
No 761 (73.1) 2,945 (82.9)
Has a Child with Tics
Yes 12 (1.0) 104 (2.6) .002
No 1,189 (99.0) 3,913 (97.4)
At Least One Family Member with Tics
Yes 613 (51.0) 2,201 (54.8) .024
No 588 (49.0) 1,816 (45.2)
Two or More Family Members with Tics
Yes 145 (12.1) 609 (15.2) .009
No 1,056 (87.9) 3,408 (84.8)
Has Severe Tics
Yes 226 (18.3) 668 (15.9) .046
No 1,009 (81.7) 3,544 (84.1)
Missing data alters the row and column totals for some variables.
Perinatal problems were prevalent in 27 percent of those with TS + LD, and only 17 percent for those with TS - LD (p < .001). The proportion of subjects with TS + LD were somewhat less likely to have a child with tics (p = .02) or have at least one family member with tics or TS (12 percent) when compared to the proportion of subjects with TS - LD (15 percent, p = .009). The proportion of subjects with severe tics in the TS + LD group was only slightly higher (18%) compared to those with TS - LD (16%), p = .046.
Subjects with TS + LD were more likely to have one or more of the fourteen comorbid disorders and conditions in the dataset when compared to those subjects with TS - LD (Table 2). Increases in comorbid conditions ranged from 0.7 percent for (psychotic disorder, p = .037) to 28.9 percent for (ADHD, p < .001). The mean number of comorbidities for subjects with TS + LD was 3.9 (s.d. = 2.2) and for subjects with TS - LD was 2.8 (s.d. = 2.0).
Table 2 Between group comparisons in 5,450 people with Tourette Syndrome and comorbid learning disabilities (TS + LD) and Tourette Syndrome without learning disabilities (TS - LD).
TS + LD TS - LD
n (%) n % p
ADHD 990 (80.2) 2,161 (51.3) <.001
Anger 570 (46.2) 1,424 (33.8) <.001
Sleep 372 (30.2) 998 (23.7) <.001
Mood 266 (21.5) 767 (18.2) <.001
Social Skills 409 (33.1) 620 (14.7) <.001
Anxiety 249 (20.2) 686 (16.3) .002
Sexual Behavior 85 (8.1) 137 (4.0) <.001
CD 250 (20.2) 504 (12.0) <.001
Coprolalia 199 (16.1) 531 (12.6) .002
Stutter 137 (11.1) 271 (6.4) <.001
Neurologic 104 (8.4) 218 (5.2) <.001
DevD 137 (11.1) 180 (4.3) <.001
PDD 101 (8.18) 167 (4.0) <.001
Psy 20 (1.6) 37 (0.9) .037
Attention deficit-hyperactivity disorder (ADHD), conduct disorder (CD), obsessive compulsive disorder (OCD), obsessive compulsive behavior (OCB), developmental disorder (DevD), learning disability (LD), mental retardation (MR), pervasive developmental disorder (PDD), psychosis (Psy) and neurological abnormality (Neurologic).
ADHD was the most prevalent comorbid disorder for subjects with TS + LD. In this population, 58% (3151) of the TS children had ADHD and 31% (990) of these had LD. The potential impact of ADHD on LD either as a causal factor or as a confounder for the diagnosis of LD is demonstrated by the finding that only 11 % (245) of the 2299 TS children without ADHD had LD.
The variables from Table 1 and the total number of comorbidities were entered into a logistic regression model. The optimal prediction model for the TS + LD group was comprised of five variables (being seen for evaluation before 18, being male, having fewer family members with tics or TS, having perinatal problems, and having more comorbidities). The logistic model performance characteristics were accuracy 65.2% (model correctly classified 4,406 of 5,450 subjects), sensitivity 66.1% and specificity of 62.2%.
Discussion
In a population of 5,450 subjects with TS, we found 1,235 subjects with comorbid LD (TS + LD). Using logistic regression, we produced a five variable model that accurately predicted group membership for 65.2% of the 5,450 subjects in this study. The model parameters were male gender, fewer affected family members, increased rates of pregnancy, labor and delivery complications, increased prevalence of comorbidities and younger age at diagnosis. The absolute differences in rates of comorbidities between the groups for individual variables were often small and as a result the differences may be of limited clinical relevance. However, the five variable model may well have relevance for risk assessment for clinicians caring for subjects with TS and possibly for healthcare policy makers as well. Confirmation of our estimates of the performance characteristics of this model as a screening tool would require further study in a clinical setting. However, the development of a screening tool would be beneficial since delayed identification of learning disabilities results in delayed initiation of intervention services and likely increases the educational difficulty experienced by a person with an unidentified learning disability [22, 32].
The etiology of learning disabilities and the appropriate conceptual view of these diverse disorders as comorbid disorders or as variably prevalent components of the broader TS phenotype has yet to be resolved [3, 9, 10, 14, 15, 33]. In this study, ADHD was the most prevalent comorbid disorder with TS occurring in 57.8% of subjects (n = 3151). In subjects with TS + LD, 80.2% also had a diagnosis of ADHD and in the TS - LD group, 51.3% had a diagnosis of ADHD. We found that 31% of subjects with ADHD also had a diagnosis of LD compared to only 11% in subjects with TS who did not have ADHD. Thus, the comorbidity rates in this study may not differ from those reported for ADHD and reading disorders alone [34, 35]. where the prevalence of comorbidity between reading disorders and ADHD is 25 to 40%. The increased rates of ADHD in the TS + LD group may have multiple explanations including the possibility that ADHD is a confounder and that most cases of LD in subjects with TS represent the additional impairments in learning from the ADHD. In which case LD is misdiagnosed or that ADHD is an important component in the causal chain for LD and that LD is very often under diagnosed in subjects with ADHD. Additional research is required to determine which, if either, of these possibilities is correct. Other data sets will likely be required to examine the role of ADHD on LD in subjects with TS and other combinations of LD, ADHD and TS.
Limitations
We have defined the two groups used in this study by the presence or absence of a clinically defined learning disability. We are unable to determine the accuracy of the diagnosis for subjects in this study. In this study we did not have the data to restrict LD cases to a single set of criteria. For example, we did not count only cases meeting the discrepancy criteria, which is a widely used strategy for the diagnosis of LD in the United States. We are not aware of a single diagnostic schema with wider acceptance around the world than the DSM criteria. As a result the prevalence estimates in this study may be biased. This might alter the accuracy of prevalence estimates or the significance testing for some variables. However, given the effect sizes found here, the bias would have to be consistent and quite large to alter the primary results.
Conclusion
We found the prevalence of LD to be increased in subjects with TS. Additional studies are required to improve our understanding of the etiologic factors resulting in the expression of the patterns of individual syndromal variability noted in this and in other studies. It would be of interest to examine hypotheses to determine if subjects with TS have different types of LD or have specific patterns of comorbidity with LD. However, developing test batteries for subjects from over 20 different languages, cultures and differing academic systems seems a formidable task.
Improved understanding of the factors associated with a later diagnosis of an LD may have important implications for prevention of secondary disabilities especially those which result from symptom expression prior to a diagnosis of either TS or of TS with comorbidity [22, 36]. This recognition may be years or in some cases decades delayed. Ongoing research is needed to identify appropriate medical, psychosocial or educational management strategies for the two broad groups discussed here in this paper.
Competing interests
The author(s) declare that they have no competing interest.
Authors' contributions
RF, LB, and JK designed the study, and RF developed and maintained the TIC Registry. MK completed the data analysis. RF, LB, MK and JK wrote the manuscript and contributed important intellectual content. All authors have read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Supplementary Material
Additional File 1
TIC Data Entry Form. The TIC Consortium Data Entry Form is a standardized form used by each center to extract data elements from the record for submission to the consortium database.
Click here for file
==== Refs
Bruun RD Budman CL The course and prognosis of Tourette syndrome Neurol Clin 1997 15 291 298 9115462 10.1016/S0733-8619(05)70313-3
Burd L Kerbeshian J Barth A Klug MG Avery PK Benz B Long-term follow-up of an epidemiologically defined cohort of patients with Tourette syndrome J Child Neurol 2001 16 431 437 11417610
Leckman JF Zhang H Vitale A Lahnin F Lynch K Bondi C Kim YS Peterson BS Course of tic severity in Tourette syndrome: the first two decades Pediatrics 1998 102 14 19 9651407 10.1542/peds.102.1.14
American Psychiatric Association Diagnostic and statistical manual of mental disorders DSM-IV-TR text revision 2000 4 Washington, DC: American Psychiatric Association
Kerbeshian J Burd L Epidemiology and comorbidity. The North Dakota prevalence studies of Tourette syndrome and other developmental disorders Adv Neurol 1992 58 67 74 1384283
Freeman RD Fast DK Burd L Kerbeshian J Robertson MM Sandor P An international perspective on Tourette syndrome: selected findings from 3,500 individuals in 22 countries Dev Med Child Neurol 2000 42 436 447 10972415 10.1017/S0012162200000839
Comings DE Comings BG TS, learning, and speech problems J Am Acad Child Adolesc Psychiatry 1994 33 429 430 8169192
Pauls DL Leckman JF Cohen DJ Familial relationship between Gilles de la Tourette's syndrome, attention deficit disorder, learning disabilities, speech disorders, and stuttering J Am Acad Child Adolesc Psychiatry 1993 32 1044 1050 8407750
Comings DE Comings BG A controlled study of Tourette syndrome. I. Attention-deficit disorder, learning disorders, and school problems Am J Hum Genet 1987 41 701 741 2890294
Como PG Neuropsychological function in Tourette syndrome Adv Neurol 2001 85 103 111 11530420
Casey MB Cohen M Schuerholz LJ Singer HS Denckla MB Language-based cognitive functioning in parents of offspring with ADHD comorbid for Tourette syndrome or learning disabilities Dev Neuropsychol 2000 17 85 110 10916577 10.1207/S15326942DN1701_06
Comings DE Gade-Andavolu R Gonzalez N Blake H Wu S MacMurray JP Additive effect of three noradrenergic genes (ADRA2a, ADRA2C, DBH) on attention-deficit hyperactivity disorder and learning disabilities in Tourette syndrome subjects Clin Genet 1999 55 160 172 10334470 10.1034/j.1399-0004.1999.550304.x
Yeates KO Bornstein RA Neuropsychological correlates of learning disability subtypes in children with Tourette's syndrome J Int Neuropsychol Soc 1996 2 375 382 9375162
Eapen V Robertson MM Zeitlin H Kurlan R Gilles de la Tourette's syndrome in special education schools: a United Kingdom study J Neurol 1997 244 378 382 9249623 10.1007/s004150050105
Walkup JT Scahill LD Riddle MA Disruptive behavior, hyperactivity, and learning disabilities in children with Tourette's syndrome Adv Neurol 1995 65 259 272 7872144
Singer HS Schuerholz LJ Denckla MB Learning difficulties in children with Tourette syndrome J Child Neurol 1995 10 S58 S61 7751556
Brookshire BL Butler IJ Ewing-Cobbs L Fletcher JM Neuropsychological characteristics of children with Tourette syndrome: evidence for a nonverbal learning disability? J Clin Exp Neuropsychol 1994 16 289 302 8021315
Burd L Kauffman DW Kerbeshian J Tourette syndrome and learning disabilities J Learn Disabil 1992 25 598 604 1431545
Burd L Kerbeshian J Educational management of children with Tourette syndrome Adv Neurol 1992 58 311 317 1414637
Comings DE Comings BG A controlled family history study of Tourette's syndrome, I: Attention-deficit hyperactivity disorder and learning disorders J Clin Psychiatry 1990 51 275 280 2365665
Matthews WS Attention deficits and learning disabilities in children with Tourette's syndrome Pediatr Ann 1988 17 410 1 414, 416 3164467
Lerer RJ Motor tics, Tourette syndrome, and learning disabilities J Learn Disabil 1987 20 266 267 3474329
Definitions and classification of tic disorders. The Tourette Syndrome Classification Study Group Arch Neurol 1993 50 1013 1016 8215958
American Psychiatric Association Diagnostic and statistical manual of mental disorders DSM-IV 1994 4 Washington, DC: American Psychiatric Association
Spady DW Saunders DL Schopflocher DP Svenson LW Patterns of injury in children: a population-based approach Pediatrics 2004 113 522 529 14993544 10.1542/peds.113.3.522
Dykman RA Ackerman PT Attention deficit disorder and specific reading disability: separate but often overlapping disorders J Learn Disabil 1991 24 96 103 2010680
Semrud-Clikeman M Biederman J Sprich-Buckminster S Lehman BK Faraone SV Norman D Comorbidity between ADDH and learning disability: a review and report in a clinically referred sample J Am Acad Child Adolesc Psychiatry 1992 31 439 448 1592775
|
16137334
|
PMC1242237
|
CC BY
|
2021-01-04 16:31:06
|
no
|
BMC Pediatr. 2005 Sep 1; 5:34
|
utf-8
|
BMC Pediatr
| 2,005 |
10.1186/1471-2431-5-34
|
oa_comm
|
==== Front
BMC Public HealthBMC Public Health1471-2458BioMed Central London 1471-2458-5-951615940010.1186/1471-2458-5-95Research ArticleAIDS knowledge and attitudes in a Turkish population: an epidemiological study Ayranci Unal [email protected] Medico-Social Center, Osmangazi University, 26480 Meselik-Eskisehir, Turkey2005 13 9 2005 5 95 95 14 5 2005 13 9 2005 Copyright © 2005 Ayranci; licensee BioMed Central Ltd.2005Ayranci; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The aim of this study was to investigate and present some pertinent comments concerning Acquired Immunodeficiency Syndrome (AIDS) knowledge, attitudes and misconceptions among the general population in a city of west Turkey. This study was deemed important and relevant due to the increasing importance of AIDS in Turkey and the other countries.
Methods
Using a multistage area sampling method, a random sample of individuals aged 11–83 years, living in 65 different quarters in the city of Eskisehir, Turkey during September, October and November 2004 were interviewed.
Results
In all, 1048 respondents completed the survey. In most items, respondents displayed a fairly good to excellent degree of knowledge about HIV/AIDS. Individuals with higher degrees of education indicated more correct responses in all items relating to knowledge of HIV/AIDS.
In general, the respondents' attitudes towards AIDS and people with AIDS were found to be tolerant and positive, with one answer choice showing that the majority of the respondents agreed with the statement that those with HIV/AIDS must be supported, treated and helped (90.7%). Moreover, the proportions of the respondents' misconceptions were found to be significantly low for all the items. However, nearly one fourth of the respondents agreed with the misconceptions 'AIDS is a punishment by God' and 'One is not infected with HIV/AIDS if engaged in sport and well nourished'.
Conclusion
In general HIV/AIDS related knowledge was high and people showed positive attitudes. However, people continue to hold misconceptions about AIDS and these need to be addressed by health education programs targeting those at higher risk.
==== Body
Background
The Acquired Immunodeficiency Syndrome (AIDS), one of the most complex health problems of the 21 st century, is in its third decade and has become a pandemic disease that threatens the world population. Moreover, with no treatment or cure in sight, the disease continues to spread at an alarming rate [1-3]. Recent epidemiological data indicates that an estimated 34–46 million individuals are living with Human Immunodeficiency Virus (HIV)/AIDS [2-4]. Over 30 million people have already died from AIDS, with the year 2003 alone seeing 3 million [4]. Four million children have been infected since the virus first appeared. Over 90% of these individuals are concentrated in the developing countries, mostly in countries least able to afford to care for infected people. Over 50% of the newly infected adults are in the age bracket 15 to 24 years old and more than 40% are women [2,4-6].
In Turkey, the importance of AIDS started with the diagnosis of two patients in the year 1985, and still continues to be on the agenda up to the present day. The number of AIDS cases increases every year: 34 new cases in 1990, 91 new cases in 1995, and 119 new cases in 1999. According to the June 2004 statistics of the Ministry of Health (MoH) the cumulative number of HIV positive patients is 1802, 76% of which are sexually active and social individuals aged between 15–49 years. Approximately 800 of these were AIDS patients [7]. These numbers, however, are only the official numbers of the MoH. As has been previously established, the number of patients with infectious diseases increases geometrically. Thus, based on the original two cases of 1985, the actual number of the patients in 2004 has been calculated at about 10,000. According to the information supplied by the MoH, AIDS in this country is considered to be in its beginning phase, compared with the calculation above. However, this figure does not reflect the real numbers afflicted due to the inadequacy of the registration system and the fact that people with sexually transmitted diseases do not generally attend health centers [8].
This article presents data from a study of the general population of Eskisehir, a city in western Turkey. Several studies, conducted among selected target populations, have evaluated HIV/AIDS knowledge and the attitudes of certain groups such as university students [9], health non-commissioned officer candidates [10], soldiers [11], street children/youths [12], and nurses [13] within the Turkish society. In Turkey, however, we are unaware of any earlier studies of HIV/AIDS-related knowledge, attitudes, and misconceptions or practices conducted among the general population of Turkey. This was the particular aim of this study. Therefore, the present study sought to address Turkish society's risk behavior, knowledge, attitudes and misconceptions about HIV/AIDS, to assess needs for sex education and to discover sources of information about AIDS for the people in a city of west Turkey.
Methods
Setting
Eskisehir is a semi rural province situated in the western part of Turkey, with a population of about 460,000. The socio-economical level of the city is average compared to other cities of the country. There are significant disparities in the socio-economic characteristics between the quarters of the city. It includes two universities, and also has a cosmopolitan structure.
The questionnaire
The questionnaire used in this survey, based on the WHO AIDS programme knowledge, attitudes, beliefs and practices (KABP) survey in 1988 [14] as well as literature [3,15-20], was modified to suit the Turkish culture and norms, in a way that covered all the profession groups in the general public living in the city, such as students, tradesmen, workers, housewives, lorry drivers etc. The questionnaire, consisting of 56 questions in Turkish, was divided into three broad sections: sociodemographic characteristics, knowledge concerning AIDS, and sources of information about AIDS. The sources of information included modes of transmission, attitudes towards AIDS and people with AIDS, and misconceptions or beliefs about AIDS and AIDS patients. Some questions on the questionnaire, such as those concerning the source of information of HIV/AIDS, were open-ended and covered sociodemographic characteristics, knowledge of the disease including the mode of transmission and populations at high risk, attitudes towards HIV-positive patients, and the source of knowledge and beliefs towards HIV/AIDS. The questionnaire was then pre-tested on a sample of 62 participants from different subpopulations of the city. Alpha coefficients for reliability and internal consistency of the questions were found to be 0.891, 0.734, and 0.603 for knowledge, attitudes, beliefs or misconceptions about HIV/AIDS, respectively. The completed questionnaires were checked for consistency and completeness. Questions were answered using the options "Agree/True", "Disagree/False", and "I don't know/I have no idea". Responses to all the items were converted to a percentage indicating the proportion of correct responses.
Sampling
1048 people were interviewed face-to-face between September and November 2004 for a study on the KABP relating to HIV/AIDS of the population in a city of western Turkey.
Due to the questionnaire being rather long, the survey being conducted on a general population and there being a possibility that some people in the city are illiterate, trained interviewers (4 females and one male) helped to explain any questions that the respondents found incomprehensible. 274 houses and 108 workplaces situated in the 65 quarters of the city, each having approximately equal populations, were determined using a stratified random sample method. During the study period, a total of 1,621 people were working or living in these places, with 479 people living in houses, and 1,142 people working in workplaces. The study was conducted at the participants' work place or home. Our objective was to contact the whole population of subjects in the aforementioned places. Criteria for inclusion in the study was having the ability to complete the questionnaire and completion of education to at least primary school level, working on the presumption that this would ensure that all participants had a basic knowledge level of sexuality, a basic level of maturity with regard to answering sexually related questions, or the ability to communicate with one another. Those who came to visit the city from other cities and those with hearing impairments were excluded from the study. In addition, children and those not willing to participate were also excluded. The sample was representative. It did not differ from the general population in terms of age or sex.
Procedures
All subjects (1048/1621, 64.7%) were told that participation in the investigation was strictly voluntary and were told that the data collected would not be used for anything except the research aim. Those who agreed to participate were given the questionnaire to complete. The duration for completing the questionnaire was between 20–25 minutes per subject. The principal investigator met weekly with the data collectors to ensure the quality of data collected.
Legal ethical consent
Ethical permission for the study was obtained prior to collect data, by contacting and receiving approval from the appropriate management authority, the health directorship of the city involved. Participants were assured of the confidentiality of their responses and provided informed verbal consent.
Statistical analyses
The statistical package for social sciences (SPSS) version 10.0 (Chicago, IL, USA) was used to enter and analyze the data on a personal computer. Obtained data were evaluated by frequency and percentages ratios, Chi-square (x2) and t tests. The measure for statistical significance was established a priori as P < 0.05.
Results
Sample characteristics
The mean (± SD) age of the respondents (n = 1048) was 29.9 ± 10.6, 95% CI (Confidence Interval) of 29.3–30.6, ranging from 11 to 83 years. It was significantly lower in women than in men (28.6 ± 10.9, 95% CI of 27.6–29.6 and 31.1 ± 10.3, 95% CI of 30.2–31.9, respectively), (t = 3.7, d.f. = 1046, p = 0.000, 95% CI of 1.13–3.71). Age ranged from 11 to 83 in men and from 12 to 76 in women. More respondents (56.7%) were male and single (48.1%) or married (46.9%), had attained the level of secondary education or above (86.2%). Most participants (64.8%) were working in a job and believed in God (87.9%). Household income levels were average or higher (71.6%). The characteristics of the participants are presented in Table 1.
Table 1 The respondents' characteristics (n = 1048)
Number Percentage
Gender
Male 594 56.7
Female 454 43.3
Age
≤24 396 37.8
25–34 367 35.0
35–44 184 17.6
≥45 101 9.6
Employment status
Employed 679 64.8
Housewife 155 14.8
Student 83 7.9
Unemployment 59 5.6
Retired 72 6.9
Number of those living at home
Between 1–3 345 32.9
Between 4–5 552 52.7
≥6 151 14.4
Marital status
Single 504 48.1
Married 492 46.9
Widowed/divorced/seperated 52 5.0
Presence of religious belief
Yes 921 87.9
No 98 9.4
Unsure 29 2.8
Educational levels
Illiterate 20 1.9
Primary 125 11.9
Secondary 119 11.4
High school 280 26.7
Higher education 504 48.1
Family's total income level
Low 298 28.4
Average 521 49.7
High 229 21.9
Sixty two percent of the sample believed they were not at any risk of contracting HIV (Unshown data).
Respondents' knowledge levels
The analysis of data indicated that in most items respondents had a fairly good to excellent knowledge about HIV/AIDS. The percentages of 'true' responses for all the knowledge items were higher than 'false' and 'don't know' responses, with the exception of the response for item 7. The vast majority had correct knowledge about items 2 (92.0%), 25 (94.3%), 28 (93.6%), 29 (91.5%), and 30 (90.6%). However, over 30% of the respondents thought that AIDS is a hereditary disease (37.6%); that AIDS is not generally seen in developing or underdeveloped countries (38.5%); that a person infected with HIV usually shows some symptom of the disease (39.2%); and that urine, X-ray, total blood count and biochemistry analyses are the tests used to check for the HIV virus in the blood (30.0%). Further misconceptions held were that HIV/AIDS can be contracted through sharing public toilets and swimming pools with an infected person (33.4%); using an infected person's belongings such as clothes, comb, underwear and towels (33.7%); sharing the food utensils of an infected person (37.7%); and exposure to an infected person who coughs or spits (34.9%), or the urine of an infected person (31.2%).
There was an important evidence of sex differences in responses regarding knowledge about HIV/AIDS for 13 items out of the 34. In statements relating to knowledge on items 1., 2., 3., 4., 6., 9., 11., 21., 22., 23., 29., 32., and 33, males gave significantly more correct responses than did female respondents.
Age groups were found to be significantly associated with AIDS related knowledge for 16 of the items. Younger respondents (24 years) responded better to items 6., 16., 17., and 31., whereas older respondents (45 years) responded better to items 5., 7., 8., 26., and 35. On the other hand, those in age group 25–34 years old had more correct responses for items 4., 9., 10., 12., and 20., and those in age group 35–44 years old had significantly more correct responses for items 19., and 27. The detailed data are presented in Table 2.
Table 2 The respondents' knowledge on HIV/AIDS
Knowledge items (Alpha 0.891 for the below 34 items) Yes n(%) No n(%) Don't know n(%)
General knowledge:
1 A virus causes AIDS 794(75.8)√ 87(8.3) 167(15.9)
2 AIDS is a contagious disease 964(92.0)√ 46(4.4) 38(3.6)
3 AIDS is a hereditary disease 394(37.6) 496(47.3)√ 158(15.1)
4 There is an active treatment for AIDS 305(29.1) 545(52.0)√ 198(18.9)
5 AIDS is mostly seen in the developing or underdeveloped countries, mostly in countries least able to afford to care for infected people 511(48.8)√ 403(38.5) 134(12.8)
6 AIDS is not a serious disease. It is a simple disease like the common cold 85(8.1) 904(86.3)√ 59(5.6)
7 A person infected with HIV does not usually show any symptoms of the disease 340(32.4)√ 411(39.2) 297(28.3)
8 Resistance to other diseases in an individual with AIDS is rather low 727(69.4)√ 138(13.2) 183(17.5)
9 There is a vaccine for AIDS 152(14.5) 737(70.3)√ 159(15.2)
10 We can distinguish AIDS patients from others by their appearance 242(23.1) 561(53.5)√ 245(23.4)
11 The ELISA test is used to check for the HIV virus in the blood 700(66.8)√ 92(8.8) 256(24.4)
12 Urine, X-ray, total blood count and biochemistry analyses are used to check for the HIV virus in the blood 314(30.0) 433(41.3)√ 301(28.7)
HIV/AIDS can be contacted through:
13 Sharing public toilets and swimming pools with an infected person 350(33.4) 560(53.4)√ 138(13.2)
14 Using an infected person's belongings such as clothes, comb, underwear and towel 353(33.7) 568(54.2)√ 127(12.1)
15 Sharing a razor blade with an infected person 780(74.4) √ 164(15.6) 104(9.9)
16 Touching an infected person, such as hugging, holding and shaking hands 243(23.2) 718(68.5)√ 87(8.3)
17 Sharing the food utensils of an infected person 395(37.7) 510(48.7)√ 143(13.6)
18 Exposure to an infected person who coughs or spits 366(34.9) 531(50.7)√ 151(14.4)
19 Having a tattoo done with the same devices after an infected person 798(76.1)√ 122(11.6) 128(12.2)
20 The bite of a mosquito 312(29.8) 468(44.7)√ 268(25.6)
21 Sharing injection needles or the surgical operation devices of an infected person 933(89.0)√ 50(4.8) 65(6.2)
22 Having a tooth extracted with the same devices after an infected person 896(85.5)√ 53(5.1) 99(9.4)
23 An infected pregnant woman's infecting her unborn baby 816(77.9)√ 82(7.8) 150(14.3)
24 Donating to another person the organs and tissue of an infected person 819(78.1)√ 84(8.0) 145(13.8)
25 Having vaginal sex with an infected person 988(94.3)√ 26(2.5) 34(3.2)
26 Having oral sex with an infected person 741(70.7)√ 148(14.1) 159(15.2)
27 Having anal sex with an infected person 821(78.3)√ 76(7.3) 151(14.4)
28 Receiving blood from an infected person 981(93.6)√ 28(2.7) 39(3.7)
29 The vaginal liquid of an infected person 959(91.5)√ 26(2.5) 63(6.0)
30 The sperm of an infected person 950(90.6)√ 45(4.3) 53(5.1)
31 The urine of an infected person 327(31.2) 505(48.2)√ 216(20.6)
32 The tears of an infected person 171(16.3) 659(62.9)√ 218(20.8)
33 The mucus or nasal fluid of an infected person 167(15.9) 652(62.2)√ 229(21.9)
34 The breast milk of an infected person 657(62.7)√ 191(18.2) 200(19.1)
√ True responses
Table 3 shows the respondents' AIDS knowledge levels according to their educational status. It revealed that the respondents' AIDS knowledge levels showed statistical significances according to their educational status for all but 8 knowledge items (5., 7., 20., 21., 26., 27., 29. and 30.). The knowledge levels of those who had attained a higher education level, such as university, were higher than those having lower educational levels, save for 6 items out of 34 (7., 26., 27., 29., 30., and 34.).
Table 3 AIDS knowledge levels by educational status of the respondents
General knowledge items on HIV/AIDS and the proportions of those who answered correctly Illiterate n(%) 20(1.9) Primary n(%) 125(11.9) Secondary n(%) 119(11.4) High school n(%) 280(26.7) Higher school n(%) 504(48.1) Those answering correctly n(%) 1048(100.0)
1 A virus causes AIDS‡ 15(75.0) 72(57.6) 65(54.6) 203(72.5) 439(87.1) 794(75.8)
2 AIDS is a contagious disease‡ 18(90.0) 108(86.4) 106(89.1) 249(88.9) 483(95.8) 964(92.0)
3 AIDS is a hereditary disease‡ 6(30.0) 34(27.2) 37(31.1) 125(44.6) 294(58.3) 496(47.3)
4 There is an active treatment for AIDS‡ 8(40.0) 50(40.0) 44(37.0) 153(54.6) 290(57.5) 545(52.0)
5 AIDS is mostly seen in developing or underdeveloped countries, mostly in countries least able to afford to care for infected peoplef 9(45.0) 62(49.6) 50(42.0) 138(49.3) 252(50.0) 511(48.8)
6 AIDS is not a serious disease. It is a simple disease like the common cold‡ 13(65.0) 93(74.4) 83(69.7) 237(84.6) 478(94.8) 904(86.3)
7 A person infected with HIV does not usually show any symptoms of the diseasef 10(50.0) 40(32.0) 29(24.4) 86(30.7) 175(34.7) 340(32.4)
8 Resistance to the other diseases in an individual with AIDS is rather low† 12(60.0) 87(69.6) 68(57.1) 198(70.7) 362(71.8) 727(69.4)
9 There is a vaccine for AIDS‡ 8(40.0) 56(44.8) 62(52.1) 195(69.6) 416(82.5) 737(70.3)
10 We can distinguish AIDS patients from others by their appearance‡ 7(35.0) 39(31.2) 51(42.9) 144(51.4) 320(63.5) 561(53.5)
11 The ELISA test is used to check for the HIV virus in the blood‡ 10(50.0) 58(46.4) 59(49.6) 165(58.9) 408(81.0) 700(66.8)
12 Urine, X-ray, total blood count and biochemistry analyses are used to check for the HIV virus in the blood‡ 3(15.0) 33(26.4) 38(31.9) 84(30.0) 275(54.6) 433(41.3)
HIV/AIDS can be contacted through:
13 Sharing public toilets and swimming pools with an infected person‡ 8(40.0) 43(34.4) 51(42.9) 145(51.8) 313(62.1) 560(53.4)
14 Using an infected person's belongings such as clothes, comb, underwear and towel‡ 5(25.0) 45(36.0) 51(42.9) 139(49.6) 328(65.1) 568(54.2)
15 Sharing a razor blade with an infected person‡ 13(65.0) 85(68.0) 72(60.5) 203(72.5) 407(80.8) 780(74.4)
16 Touching an infected person, such as hugging, holding and shaking hands‡ 9(45.0) 54(43.2) 60(50.4) 182(65.0) 413(81.9) 718(68.5)
17 Sharing the food utensils of an infected person‡ 6(30.0) 33(26.4) 39(32.8) 123(43.9) 309(61.3) 510(48.7)
18 Exposure to an infected person who coughs or spits‡ 7(35.0) 39(31.2) 45(37.8) 121(43.2) 319(63.3) 531(50.7)
19 Having a tattoo done with the same devices after an infected person‡ 13(65.0) 91(72.8) 76(63.9) 206(73.6) 412(81.7) 798(76.1)
20 The bite of a mosquitof 7(35.0) 49(39.2) 52(43.7) 126(45.0) 234(46.4) 468(44.7)
21 Sharing the injection needles or surgical operation devices of an infected personf 15(75.0) 110(88.0) 95(79.8) 243(86.8) 470(93.3) 933(89.0)
22 Having a tooth extracted with the same devices after an infected person† 15(75.0) 107(85.6) 95(79.8) 228(81.4) 451(89.5) 896(85.5)
23 An infected pregnant woman's infecting her unborn baby† 15(75.0) 91(72.8) 84(70.6) 207(73.9) 419(83.1) 816(77.9)
24 Donating the organs and tissue of an infected person to another person† 15(75.0) 91(72.8) 83(69.7) 211(75.4) 419(83.1) 819(78.1)
25 Having vaginal sex with an infected person† 18(90.0) 117(93.6) 105(88.2) 261(93.2) 487(96.6) 988(94.3)
26 Having oral sex with an infected person† 13(65.0) 100(80.0) 85(71.4) 211(75.4) 332(65.9) 741(70.7)
27 Having anal sex with an infected personf 13(65.0) 102(81.6) 91(76.5) 220(78.6) 395(78.4) 821(78.3)
28 Receiving blood from an infected person‡ 18(90.0) 114(91.2) 102(85.7) 260(92.9) 487(96.6) 981(93.6)
29 The vaginal liquid of an infected personf 16(80.0) 116(92.8) 104(87.4) 259(92.5) 464(92.1) 959(91.5)
30 The sperm of an infected personf 18(90.0) 111(88.8) 104(87.4) 263(93.9) 454(90.1) 950(90.6)
31 The urine of an infected person‡ 9(45.0) 42(33.6) 46(38.7) 131(46.8) 277(55.0) 505(48.2)
32 The tears of an infected person‡ 10(50.0) 62(49.6) 62(52.1) 167(59.6) 358(71.0) 659(62.9)
33 The mucus or nasal fluid of an infected person‡ 10(50.0) 56(44.8) 61(51.3) 163(58.2) 362(71.8) 652(62.2)
34 The breast milk of an infected person† 8(40.0) 56(44.8) 46(38.7) 112(40.0) 150(29.8) 657(62.7)
p < 0.001‡, p > 0.05f, p < 0.05†
Respondents' attitudes
In general, the respondents' attitudes towards AIDS and people with AIDS were found to be significantly tolerant and positive, with the exception of items 7 and 8. The majority of the respondents positively agreed with statement 5 (70.5%) and statement 9 (90.7%). On the other hand, a large number of the respondents agreed with the stigmas 1., 2., 3., 4., 6., 7., 8 and 10 with proportions of between 30% and 50%. These results are shown in Table 4.
Table 4 The respondents' attitudes towards HIV/AIDS
Attitudes to persons with HIV/AIDS (Alpha 0.734 for the below 10 items) Agree n(%) Disagree n(%) Neither agree nor disagree n(%)
1 Students with AIDS should go to special schools for those with AIDS 389(37.1) 504(48.1)√ 155(14.8)
2 If there is a student with AIDS in a school, I would delete the record of my child from that school 346(33.0) 554(52.9)√ 148(14.1)
3 I would not sit in the same armchair or desk with a person with AIDS 345(32.9) 588(56.1)√ 115(11.0)
4 I would not kiss someone with AIDS 400(38.2) 553(52.8)√ 95(9.1)
5 They should be locked up or isolated in a special center 187(17.8) 739(70.5)√ 122(11.6)
6 I would have personal contact with someone with AIDS as an ordinary person 601(57.3)√ 333(31.8) 114(10.9)
7 I would share public toilets and swimming pools with someone with AIDS 412(39.3)√ 506(48.3) 130(12.4)
8 I would wash my clothes with those of an individual with AIDS 437(41.7)√ 489(46.7) 122(11.6)
9 They must be supported, treated and helped 951(90.7)√ 50(4.8) 47(4.5)
10 Everybody must know about those with AIDS by means of national media 332(31.7) 579(55.2)√ 137(13.1)
√ Positive attitudes
Women were more positive in their attitudes towards HIV/AIDS or AIDS victims when answering all the items compared to men, and those attitudes were statistically significant for items 1., 3., 5., 8., and 10.
There were significant differences for items 2., 3., 4., and 7 between different age groups with regard to their attitudes towards AIDS and AIDS victims. Those in the age group 45 years old and above had significantly more positive attitudes to items 2., and 7., whereas those aged 24 years old and under had significantly more positive attitudes to items 3., and 4.
Upon comparison of people with different educational levels, it was found that there were significant differences between individuals with different levels of education for 8 items out of 10. Those with higher education were significantly more positive in their attitudes to items 1., 2., 3., 9., and 10., compared to less educated respondents, whereas illiterate individuals had significantly more positive attitudes to items 4., 5., and 7 when compared to higher educated respondents.
Respondents' misconceptions
In general, the proportions of the respondents' misconceptions were found to be significantly low for all the items. The majority of the respondents disagreed with all the statements. However, 23.2% and 24.2% agreed with misconceptions 2., and 6., respectively. The results are shown in Table 5.
Table 5 The respondents' misconceptions towards HIV/AIDS
Misconceptions to persons with HIV/AIDS (Alpha 0.603 for the below 6 items) Agree n(%) Disagree n(%) Neither agree nor disagree n(%)
1 If you are passionately in love with someone, you become immune to AIDS 73(7.3) 930(88.7) 45(4.3)
2 AIDS is a punishment from God 243(23.2) 707(67.5) 98(9.4)
3 AIDS does not influence the Turkish 46(4.4) 982(93.7) 20(1.9)
4 I will not be infected with AIDS come what may 100(9.5) 884(84.4) 64(6.1)
5 Married couples do not contract AIDS even if they have sex with others 43(4.1) 957(91.3) 48(4.6)
6 You cannot be infected with HIV/AIDS if you are engaged in sport and are well nourished 254(24.2) 620(59.2) 174(16.6)
Women more disagreed with the vast majority of the misconceptions, with the exception of item 4., where men more disagreed. There were significant differences between men and women to the misconception items 4., and 5..
Significant differences between different age groups were seen for only 2 of the items (2., and 6.) with regard to their misconceptions towards AIDS and AIDS victims. Those in the age group 24 years old and under disagreed more significantly with misconception item 2., whereas those in the age group 35–44 disagreed more significantly towards misconception item 6. However, no difference was observed between the other items (1., 3., 4., and 5.) and different age groups.
The proportions of those who neither agreed nor disagreed were higher for those agreeing with all items; however, there were no significant differences between religious belief and misconceptions, except for item 6.
Although the proportions of those who agreed with all the items were higher in men than in women, there were significant differences between religious beliefs and sex differences, with the exception of items 4., and 5.
There were no significant differences between different age groups and those who agreed with misconceptions towards AIDS and AIDS victims, barring item 2 where the proportion of those agreeing with item 2 was higher in those aged 45 years and over when compared to the other age groups.
While comparing people with different educational levels, it was found that there were significant differences between individuals with different levels of education in all of the misconception items.
Those with higher education disagreed significantly more with all the misconceptions, compared to less educated respondents.
Sources of HIV/AIDS information
The majority of the respondents indicated that their level of information about HIV/AIDS was average (49.3%). Most respondents reported that mass media (television, 68.9%; newspapers, 47.6%; magazines, 23.6%) was the major sources of their information about HIV/AIDS, followed by school and friends (18.8% and 18.7%, respectively). However, most indicated that they desired to learn more (86.6%). These results are shown in Table 6
Table 6 The respondents' source of information and their informational needs (n = 1048)
Number* Percentage
Level of information about HIV/AIDS
Average 517 49.3
Bad 446 42.6
Good 85 8.1
Source of information
Television 722 68.9
Newspaper 499 47.6
Magazine 247 23.6
School 197 18.8
Friend 196 18.7
Book 154 14.7
Radio 141 13.4
Doctor 119 11.3
Teacher 82 7.8
No 70 6.7
Family 49 4.7
Nurse 47 4.5
Internet 19 1.8
Poster 16 1.5
Seminar 14 1.3
AIDS Association 8 0.7
Workplace 3 0.2
Desire to learn more
Yes 908 86.6
No 140 13.4
*The total exceeds the sample size since each respondent could choose several response categories
Discussion
This paper reports data from a population-based study on AIDS knowledge, attitudes and misconceptions among the general population in Eskisehir, Turkey. The findings indicated that people in Turkey reported good knowledge about AIDS. However, nearly 40% and 30% of the respondents believed that AIDS is a hereditary disease and that there is an active treatment for AIDS, respectively. They also indicated that they believed that AIDS is not generally seen in the developing or underdeveloped countries, countries that would be least able to afford to care for infected people (38.5%), that people infected with HIV usually show some symptom of the disease (39.2%), and that 23.1% of them believed that AIDS patients differ from the normal population in their appearance (23.1%). Furthermore, respondents demonstrated a limited knowledge of how HIV/AIDS cannot be transmitted whereas a rather high rate said that AIDS could be contracted through items 13 (33.4%), 14 (33.7%), 17 (37.7%), 18 (34.9%), 20 (29.8%), and 31 (31.2%). It appears that a number of respondents in the city do not know about the risk of transmission of AIDS or HIV infection from different sources. These findings are consistent with other results in both our country and other countries [9,15,17,21].
The findings suggested that gender was not associated with the correct answering of questions on AIDS-related knowledge, excluding those of 13 items (1., 2., 3., 4., 6., 9., 11., 21., 22., 23., 29., 32., and 33.) where males preformed better. In contrast, as expected, those who had higher levels of education and younger respondents (those aged 34 and under) generally had more correct answers on questions relating to knowledge about AIDS. A study from the US among the general population also showed significant differences in AIDS-related knowledge with less-educated and older respondents being less likely to respond correctly to general AIDS knowledge questions [22]. Another reason for the higher proportion of true answers in men than women may be that men feel freer than women to talk about matters relating to sex and HIV/AIDS [23]. Furthermore, there were significant differences between those with different levels of education and knowledge about AIDS. In all items individuals with higher education levels, especially those with a university or college education, had more correct responses for 28 of the items, barring only 6 items (7., 26., 27., 29., 30., and 34.).
The most interesting finding from this survey was the fact that people in the city showed a more positive attitude towards AIDS and those with AIDS than expected, with the exception of a few items (7., and 8.). For example only a small number of the respondents disagreed with the attitude that people with AIDS must be supported, treated and helped (4.8%) or agreed with the attitude that people with AIDS should be locked up or isolated in a special center (17.8%). Furthermore, when taking into consideration that the majority of the respondents' educational levels were high school and above (74.8%) and that their levels of information about HIV/AIDS were average or above (57.4%) our study, in line with the study of Maswanya et al (2000) [24], may conclude that people with good knowledge or good education about AIDS do become more tolerant of people with AIDS.
On the other hand, there was a substantial negative attitude towards HIV and AIDS positive patients. A proportion of about 30% and 50% of the respondents, excluding item 5 where the proportion of negative attitude was only 17.8%, expressed negative attitudes. These findings are consistent with the findings of some studies conducted in Iran and Indian [3,23]. This can be explained by the similar sociocultural design of Turkish, Iranian and Indian attitudes towards HIV/AIDS, especially in the light of religious factors, and may also be explained by the respondents having confused opinions towards HIV virus and people with AIDS.
In our study, 17.8% of the respondents agreed with the statement suggesting that people with AIDS should be locked up or isolated in a special center. This result was in line with the study by Moatti et al (1998) [25], where the proportion of this attitude was 21.9%. One explanation for this attitude may be that some people in the community do not know much about AIDS or people with AIDS.
In this study, women were more positive in their attitudes towards HIV/AIDS or AIDS victims in all the items when compared to men. This finding is compatible with the study by Lester (1989) [26]. This may be explained by women being more sensitive due to naturally possessing the mothering instinct.
This survey demonstrated that in general only a small number of the respondents harbored misconceptions. For instance, only 23.2% of people in a Muslim country such as Turkey agreed with the statement that AIDS is a punishment from God. In parallel, this proportion was 14.2% in Montazeri (2004)'s study [20] conducted in Iran and 30.2% in the study conducted out by Tebourski and Alaya (2004) [15] in Tunisia. This means that these people believe that even religious factors could not prevent a person from HIV infection or simply that they do not believe that holding religious beliefs can be preventive. However, Nwokoji and Ajuwon (2004) [17] indicated that such misconceptions might encourage some individuals to take risks by creating the false impression that they will be cured if they become infected with AIDS [17]. It seems there is a need for further investigation into the role of religion in AIDS prevention, particularly in countries such as Turkey where religion plays an important role in people's everyday life.
In the current study, those having attained a higher education level disagreed significantly more with all the misconceptions, compared to less educated respondents. This is compatible with the study of Eshetu et al (2004) [27]. This indicates that raising educational levels is a key tool in fighting the epidemic.
In this study, the most important AIDS related knowledge source for the majority of the respondents was the mass media such as television (68.9%) and newspapers (47.6%), followed by magazines (23.6%). It appears that the mass media, especially television, has an important role in raising AIDS awareness within the Turkish community. In contrast, institutions such as school (18.8%) and the AIDS Association (0.7%) where better information about AIDS can be found, or from specialist persons such as doctors (8.1%), teachers (7.8%) and nurses (4.5%) had less importance. These should be more involved in AIDS education. In addition, only 4.5% said that their family informed them about the disease. The sources of knowledge of HIV/AIDS for the general population reported here are similar to those previously reported [9,11,16,28]. To explain these findings, we may say that the importance of the mass media is obvious as a means of maintaining information about AIDS-related problems, and also that very little communication regarding HIV/AIDS occurred between themselves and specialist institutions and persons or their family. Taking everything into account, the media should implement new methods for AIDS education in order to improve public knowledge of HIV/AIDS. Such findings show that media prevention campaigns should be encouraged and these have the potential role to limit the emergence of Turkey's HIV/AIDS epidemic. However, studies in Southeast Asia have shown that much of the media have done little to change existing cultural values and prejudice about the sexuality and situation of people who are living with HIV or AIDS [29]. In this study, overall, there were a lot of incorrect information, negative attitudes and misconceptions about HIV/AIDS. This is consistent with the findings of Agrawal et al (1999) [23], and can be attributed to the many false claims published in the media and other modes of advertisement.
In the present study, the respondents desired to learn more about HIV/AIDS (86.6%). This indicates that reports dealing with the rapid spread of AIDS in various populations, as well as in the Turkish population, have increased the level of anxiety over contagion among the respondents.
Sixty two percent of the sample believed they were at no risk of contracting HIV. This figure is comparable to the 2003 National HIV/AIDS and Reproductive Health Survey that reported 72% of the civilian population [30] and also Nwokoji and Ajuwon (2004)'s study [17]. This low perception of risk is probably influenced by the widespread denial of the existence of HIV in the country by governments or health authorities.
Conclusion
The level of HIV/AIDS related knowledge was relatively high throughout our study, with most people showing positive attitudes. Misconceptions about AIDS still exist and need to be addressed by health education programs targeting those at higher risk. It has previously been shown that health education and sustained prevention efforts, such as the introduction of HIV/AIDS education into schools, mass-media campaigns promoting the use of condoms, and the wearing down of individuals' ability to deny risk towards patients with HIV or AIDS positive can be effective in changing people negative attitudes or misconceptions towards AIDS [31-33].
Being in total agreement with the opinion reported by Tebourski and Ben Alaya (2004) in their study [15], we would also like to state that success in increasing the publics' positive attitude to people with HIV/AIDS is vital to obtaining our goal of having a more effective control in the spreading of the disease.
Competing interests
The author(s) declare that he has no competing interests
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The author wishes to thank the study participants for their valuable efforts and time, and also Dr. Hasan Colak and Kerin Turan for assistance with the language of this manuscript.
==== Refs
Tumer A Unal S Unal S Epidemiology of HIV infection in the World and in Turkey 2000 AIDS: modern medical seminars. Ankara: Gunes Publishers 1 10
UNAIDS/WHO Report on Global HIV/AIDS Epidemic 2003
Tavoosi A Zaferani A Enzevaei A Tajik P Ahmadinezhad Z Knowledge and attitude towards HIV/AIDS among Iranian students BMC Public Health 2004 4 17 15157281 10.1186/1471-2458-4-17
UNAIDS-WHO Improving estimates. Report on the global HIV/AIDS epidemic 2004
Fowler MG Melnick SL Mathieson BJ Women and HIV. Epidemiology and global overview Obstet Gynecol Clin North Am 1997 24 705 729 9430163 10.1016/S0889-8545(05)70340-5
UNAIDS HIV and AIDS-related stigmatization, discrimination and denial; forms, contexts and determinants 2000
T.C. Ministry of Health. Basic Health Services General Directorship. Contagious Illnesses Department HIV/AIDS Surveillance data 2003
Duyan V Agalar F Sayek I Surgeons' attitudes toward HIV/AIDS in Turkey AIDS care 2001 13 243 250 11304430 10.1080/09540120020027413
Ungan M Yaman H AIDS knowledge and educational needs of technical university students in Turkey Patient Educ Counsel 2003 51 163 167 10.1016/S0738-3991(02)00190-8
Kilic S Acikel CH Kir T Ogur R Ucar M Knowledge level and attitude of health non-commissioned officer students about HIV/AIDS TSK Koruyucu Hekimlik Bulteni 2004 3 111 118
Bakir B Tumerdem N Ozer M Tufekci F Ucar M Hasde M Approaches of a group of soldiers in Istanbul and Ankara about AIDS Gulhane Medical Journal 2003 45 19 24
Baybuga MS Celik SS The level of knowledge and views of the street children/youth about AIDS in Turkey Int J Nurs Study 2004 41 591 597 10.1016/j.ijnurstu.2003.12.010
Unsal A Metintas S Sariboyaci MA Inan OC Level of knowledge of nurses about AIDS in several health services in Eskisehir General Medical Journal 1999 9 53 58
WHO/GPA/SBR Interview schedule on knowledge, attitude, beliefs and practices on AIDS/KABP survey 1988
Tebourski F Ben Alaya D Knowledge and attitudes of high school students regarding HIV/AIDS in Tunisia: Does more knowledge lead to more positive attitudes? J Adolescent Health 2004 34 161 164 10.1016/j.jadohealth.2004.01.001
Kitaura H Adachi N Kobayashi K Yamada T Knowledge and attitudes of Japanese dental health care workers towards HIV-related disease J Dentistry 1997 25 279 283 10.1016/S0300-5712(96)00030-9
Nwokoji UA Ajuwon AJ Knowledge of AIDS and HIV risk-related sexual behavior among Nigerian naval personnel BMC Public Health 2004 4 24 15210055 10.1186/1471-2458-4-24
Nwokocha ARC Nwakoby BAN Knowledge, attitude, and behavior os secondary (high) school students concerning HIV/AIDS in Enugu, Nigeria, in the Year 2000 J Pediatr Adolesc Gynecol 2002 15 93 96 12057531 10.1016/S1083-3188(01)00150-4
Uwalaka E Matsuo H Impact of knowledge, attitude, and beliefs about AIDS on sexual behavioral change among college students in Nigeria: The case of the university of Nigeria Nsukka West Africa Review 2002 3 2
Montazeri A AIDS knowledge and attitudes in Iran: results from a population-based survey in Tehran Patient Educ Counsel 2005 57 199 203 10.1016/j.pec.2004.05.014
Bhattacharya G Cleland C Holland S Knowledge about HIV/AIDS, the perceived risks of infection and sources of information of Asian-Indian adolescents born in the USA AIDS Care 2000 12 203 209 10827861 10.1080/09540120050001887
Schoenborn CA Marsh SL Hardy AM AIDS knowledge and attitudes for 1992. Data from the National Health Interview Survey. Data from the National Health Interview Survey Adv Data 1994 23 1 16
Agrawal HK Rao RS Chandrashekar S Coulter JB Knowledge of and attitude to HIV/AIDS of senior secondary school pupils and trainee teachers in Udupi District, Karnataka, India Ann Trop Pediatr 1999 19 143 149 10.1080/02724939992464
Maswanya E Moji K Aoyagi K Yahata K Izumi T Takemoto T Knowledge and attitudes toward AIDS among female college students in Nagasaki J Health Educ Res 2000 15 5 11 10.1093/her/15.1.5
Moatti JP Manesse L Le Gales C Pages JP Fagnani F Social perception of AIDS in the general public: a French study Health Policy 1998 9 1 8 10286663 10.1016/0168-8510(88)90113-3
Lester D Attitudes toward AIDS. Personality and individual differences AIDS Care 1989 10 693 694
Eshetu M Kebede D Ismail S Sanders E Wolday D Meselse T Tegbaru B Worku A Behavioral survey for HIV/AIDS infection in Asosa, among the general population and commercial sex workers Ethiop J Health Dev 2004 18 75 81
Morishita T Mizutani S Tomita Y Murashima S AIDS-related knowledge among nurses in Mie Prefecture J Public Health 1993 40 323 329
Wolffers I Culture, media, and HIV/AIDS in Asia Lancet 1997 349 52 54 8988132 10.1016/S0140-6736(96)07032-8
Federal Ministry of Health National HIV/AIDS and Reproductive Health Survey 2003 Federal Ministry of Health, Abuja, Nigeria
Scheutz F Dental care of HIV-infected patients: attitudes and behavior among Danish dentists Community Dent Oral Epidemiol 1989 17 117 119 2525453
Stratigos JD Tzala E Global epidemiology of HIV infection and AIDS Clinics in Dermatology 2000 18 381 387 11024305 10.1016/S0738-081X(99)00133-9
St Lawrence J Marx B Scott C Uwakwe C Roberts A Rosenthal D Crosscultural comparision of US and Nigerian adolescents' HIV-related knowledge, attitudes, and risk behavior: Implications for risk reduction interventions AIDS Care 1995 7 449 461 8547360 10.1080/09540129550126407
|
16159400
|
PMC1242238
|
CC BY
|
2021-01-04 16:28:57
|
no
|
BMC Public Health. 2005 Sep 13; 5:95
|
utf-8
|
BMC Public Health
| 2,005 |
10.1186/1471-2458-5-95
|
oa_comm
|
==== Front
BMC SurgBMC Surgery1471-2482BioMed Central London 1471-2482-5-191616806310.1186/1471-2482-5-19Research ArticleBurn wounds infected with Pseudomonas aeruginosa triggers weight loss in rats Steinstraesser Lars [email protected] Olaf [email protected] Ming H [email protected] Frank [email protected] Marcus [email protected] Grace [email protected] Adrien [email protected] Hans U [email protected] Daniel [email protected] Stewart C [email protected] Dept. of Surgery, University of Michigan, 1150 W. Medical Center Drive, Ann Arbor, MI 48109-0666, USA2 Medicine, University of Michigan, 1150 W. Medical Center Drive, Ann Arbor, MI 48109-0666, USA3 Pathology, University of Michigan, 1150 W. Medical Center Drive, Ann Arbor, MI 48109-0666, USA4 Dept. Plastic Surgery Burn Center, Ruhr-University Bochum, Buerkle-de la-Camp Platz 1, 44789 Bochum, Germany2005 17 9 2005 5 19 19 16 4 2005 17 9 2005 Copyright © 2005 Steinstraesser et al; licensee BioMed Central Ltd.2005Steinstraesser et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Despite dramatic improvements in the management of burns, infection still remains a serious risk for the burn patient. The aim of this study was to shed light on the impact of acute burn injury with or without infection on cytokine profiles.
Methods
Sprague-Dawley rats (n = 21) were randomized into three groups: 1) burn only 2) burn and infection or 3) sham burn. Weight was monitored and blood was collected for cytokine ELISA, LPS quantification, and peripheral blood analysis. Animals were sacrificed either after 6 or 12 days.
Results
Infected animals showed substantial weight loss until day 6 post-burn as compared to burn alone. Endotoxin and TNF-α levels were elevated early in the infected burn group within 48 hours post-burn. In contrast, significant up-regulation of the anti-inflammatory cytokine IL-10 occurred later in the clinical course and was associated with the recovery from weight loss.
Conclusion
Our results suggest that in the presence of infection, you get a SIRS response possibly due to transient endotoxemia that is only seen in the infection group. In contrast, both burn and infection get a late IL-10 (CARS) response, which is then associated with a return to normal weight in the infection group.
==== Body
Background
While current procedures for burn injury management have improved patient prognosis, increased morbidity and mortality remains a major challenge for the clinician. Thus, identification of the mechanisms responsible for post-burn immune dysfunction and increased susceptibility to wound infection, subsequent sepsis and multiple organ failure under such conditions, is crucial for the development of improved treatment modalities. Thermal injury induces an immuncompromised state that predisposes patients to sepsis and multiple organ failure [1-6]. Skin, as the first line of defense against invading microbes, is equipped with an array of immune mediators capable of recruiting inflammatory cells to enable neutralization and clearance of bacteria and fungi [7,8]. Hence, immune failure in a burn patient who has lost the skin barrier is vulnerable to infection. These are major complications associated with burn trauma and recent evidence suggests that activation of a pro-inflammatory cascade plays an important role in their development [9].
For a long time the integument had been considered as an organ of passive protection but over the past few years several discoveries have shown that the skin is not only the target of diseases connected with immunological mechanisms but moreover, an important immunocompetent organ in itself [10]. The difficulty depends on the causal connectivity of findings since a clear distinction has not always been made between the early shock phase and the later continuance of organ failure. Various physiologic and immunologic alterations are observed in acute inflammations associated with infection, trauma and thermal injury. Recently, increasing attention has been directed to the role of cytokines in the mechanisms of such alterations. Tumor necrosis factor-α (TNF-α), interleukin-1β (IL-1β), interleukin-6 (IL-6), which are pro-inflammatory cytokines, and interleukin-10 (IL-10) are considered potential mediators of inflammation produced by immunoregulatory cells as well as by a variety of other cell types [11]. Cytokines play a major role in the regulation of immune response, hematopoesis and inflammation [12].
The aim of this paper is to shed light on the role of wound infection with Pseudomonas aeruginosa on the systemic inflammatory response and weight loss in burn injury.
Methods
Bacteria
A multi drug resistant strain of Pseudomonas aeruginosa isolated from human burn wound was used for this study. Over night culture was diluted into fresh TSB and incubated for 2.5 h at 37°C. The subculture was centrifuged (10 min, 4°C, 880 g). The bacterial pellet was washed once and resuspended in a cold sodium phosphate buffer (J.T.Baker, Deventer, Holland, pH 7.4). CFU were calculated by OD600 nm measurement (UV-VIS-Spektrometer, Perkin Elmer 555).
Animals
Adult male Sprague-Dawley rats (250–300 g) were obtained from the Unit for Laboratory Animal Medicine (University of Michigan, Ann Arbor, MI) and maintained under standard laboratory conditions. In addition to a resting period at the institutional vivarium, animals were acclimated to the laboratory environment for at least 48 hours before treatment. After treatment, the rats were placed in individual cages in a temperature controlled room with food and water provided ad libitum and a 12 hour light and a 12 hour dark diurnal cycle. All experiments were performed in accordance with National Institute of Health guidelines and approval was obtained from the University of Michigan Animal Care and Use Committee.
Experimental design
Sprague-Dawley rats (n = 21) were randomized into three groups: 1) burn only; 2) burn and infection (108 CFU of multi-drug resistant Pseudomonas aeruginosa) or 3) sham burn. Animals in the first two groups received a 30% (TBSA) partial thickness burn. The animals were anesthetized prior to each intervention with intraperitoneal Ketamine hydrochloride (100 mg/kg; Fort Dodge Laboratories, Fort Dodge, Iowa) and Xylazine (13 mg/kg; Bayer Corporation, Shawnee Mission, Kansas) injection. The skin of the whole torso was clipped and treated with depilatory cream (Sally Hansen®Div. Del Laboratories, Inc., Farmingdale. NY). Twenty-four hours later, rats were placed in a mold, which exposed an area of 30% total body surface area (TBSA). Rats, except the sham burn group 3, sustained a superficial partial scald injury at the defined area on both flanks and back at 60°C for twenty seconds. This extend of injury in this animal model is time dependent and has been previously described [13]. Additionally, the burn wound of the infected group was covered with a 2 cm × 2 cm gauze containing 108 CFU (colony forming units) of log-phase Pseudomonas aeruginosa. Injury was covered with Tegaderm™ HP (3 M Health Care, St. Paul, MN) and Flex-Wrap™ Self-Adherent Wrap (The Kendall Company, Mansfield, MA). After treatment, rats were resuscitated subcutaneously with 5 ml saline. For the remainder of the study, all animals received buprenorphine (0.3 mg/kg) intra peritoneal twice daily for pain control post-burn. In previous experiments we have not observed any abdominal injury with this scald burn model which could affect food intake or absorption [13-17]. Weight changes were monitored daily. Blood was collected after 12, 24, 48 and 72 h, then after 6 and 12 days. This was used for cytokine ELISA, LPS quantification, and peripheral blood analysis. After either 6 or 12 days the animals were sacrificed and treated areas of the infected and non infected wound tissues were harvested aseptically, weighed, homogenized, serially diluted and plated in triplicate on trypticase soy agar with 5% sheep blood and Pseudomonas isolation agar (both from Becton Dickinson). Bacterial plates were then incubated for 18 hours and the number of colony forming units were counted in blinded fashion. Results are expressed as CFU per gram infected skin tissue.
Peripheral blood analysis
At appropriate time points, 20 μl of EDTA anti-coagulated blood was collected from the rat tail vein. A Hemavet Mascot Multispecies Hematology System Counter 1500 R (CDC Technologies Inc., Oxford, CT) was used to measure the blood differential, WBC, RBC, hemoglobin and hematocrit. Furthermore, 200 μl of plasma was collected and stored at -20°C for cytokine and endotoxin screening.
Endotoxin levels
For the purpose of determining the amount of endotoxin present in the plasma, a QCL-1000 Limulus Amebocyte Lysate Assay (Bio Whittaker Inc., Walkersville, MD) was performed. This assay uses a modified Limulus Amebocyte Lysate and a coloring substrate to detect gram-negative bacterial endotoxin chromogenically. Following the manufacturer's instruction, 50 μl of plasma was diluted 1:10 in order to quantify the endotoxin levels. Concentrations were photospectrometrically analyzed at 410 nm and calculated corresponding to the standard curve.
Cytokines
With the purpose of quantifying changes in cytokine levels, ELISA's were performed for IL-1β, IL-6, Il-10 and TNF-α using Cytoscreen™ Immunoassay Kits (BioSource International, Inc., Camarillo, CA). Following the manufacturer's instruction, 50 μl of plasma was used in a 1:1 dilution. ELISA plates were read on a plate reader (Bio-Tek Instruments, Windoski, VT) at dual wavelengths of 465 nm and 590 nm. Sample concentrations were determined by comparison to a standard curve.
Statistical analysis
Data with normal distribution was analyzed using ANOVA and Student's t-test (StatView®, Abacus Concepts/SAS® Institute, Cary, N.C.). Results were considered to be significant at p < 0.05 and expressed as the mean ± SEM.
Results
Wound infection
Macroscopic evaluation of burned skin immediately after the scald injury revealed notable paleness. A few minutes later a reddish pink coloration appeared, giving way to progressive pallor over the next 30–60 minutes. Edema reached a peak after approximately 2 hours and then slowly regressed, disappearing completely during the ensuing 24 hours. At the day of sacrifice, the infected wounds showed clear signs of macroscopic wound infection with pus and redness in the inoculated and adjacent tissue. Tissue counts in the infected group showed 106-107 CFU of P. aeruginosa per gram of tissue compared to 101-102 CFU in the non infected groups. Throughout the study period no mortalities in any of the groups was observed.
Weight changes
Weight loss is a prominent metabolic response to thermal injury and infection. As shown in figure 1, weight loss was significant higher for 6 days postoperatively in the infected burn group compared to the other treatment groups. In the infected group, weight decreases constantly until day 6 (mean difference day 1–6: 38.820 g). In this period the weight loss shows significance (p < 0.05) compared to the control group at every time point. Although the infected burn group recovers from the first weight loss and regained weight after 7 days postoperatively, weight remains significantly lower than in the control group. No significant difference in weight loss was observed comparing the sham burn groups with the burn only group. Differences in food intake or direct metabolic effects comparing the burn and the infected burn group have not been observed.
Figure 1 Animal body weight. Body weight of each animal was measured every day during the whole experiment.
Peripheral blood analysis
Lymphocytes were significantly lower after 12 hours post treatment for the burn (5000/ml ± 900; p = 0.04) and the infected group (3000/ml ± 500; p = 0.02) compared to no treatment control (9000/ml ± 800). No significant differences were observed for neutrophils, monocytes, red blood cell count, platelets, hemoglobin and hematocrit (data not shown).
Endotoxin levels
In the infected animals, endotoxin levels were significantly higher 12 hours post treatment for the infected burn group (625 pg/ml ± 200) compared to the burn (172 pg ± 30; p = 0.02) and control (216 pg/ml ± 48; p = 0.02) group with no difference comparing burn only with the control group. 24 hours post treatment no difference between the groups was noticeable (Figure 2).
Figure 2 Endotoxin level. Blood serum endotoxin levels were measured 12, 24, 48, 72 h, 6 and 12 days post burn injury in infected and control group (each n = 6). Endotoxin amount was calculated in ng/ml serum. The values at 12 h, 6 and 12 days were further displayed as bar graph to demonstrate significant (p < 0,05) differences.
Cytokines
IL-1
Until 48 hours after thermal injury, IL-1 levels were detected in comparable concentrations (Figure 3). After 72 hours, IL-1 concentrations in the infected group (1.125 ng/ml ± 0.19) and in the burn group (0.997 ng/ml ± 0.07) peaked and remained significantly (p = 0.03) higher compared to the no treatment control group until they returned to baseline after 12 days.
Figure 3 Cytokine expression. Blood was obtained from each rat in the separated treatment groups after 24, 48 and 72 h, 6 and 12 days post injury. Serum was separated and analyzed for IL-1β, IL-6, IL-10 and TNF-α using ELISA detection system. A time course is shown including the whole follow up of the study. Amount cytokine expression is displayed in pg/ml serum.
IL-6
In the burn group and infected burn group, IL-6 concentrations showed a peak 48 h postburn followed by a descend (p < 0.02) reaching baseline levels on day 6 compared to no treatment control (Figure 3). No significant difference was observed comparing the burn vs. infected burn group.
IL-10
Between 24 and 72 hours postburn IL-10 levels increased significantly in the burn and infected burn group with a steady state after the 6th day postburn (Figure 3). No significant difference was seen comparing the burn vs. infected burn group.
TNF-α
Levels for TNF-α were significantly higher 48 h postburn in the infected burn group compared to the burn and no treatment control group. The initial endotoxemia is followed by 24 and 48 hr TNF rise which is then followed by gradual weight loss in the infection vs. burn alone group. After 48 hours TNF-α concentrations of the infected and burn groups were comparable and significantly higher compared to the control group until day 12 (p < 0,01).
Discussion
Extensive burn trauma induces an acute inflammatory response that is associated with a variety of systemic alterations including increased vascular permeability, myocardial dysfunction, hypermetabolism and altered hepatic synthetic activity [18-21].
This study demonstrates significantly higher TNF-α and endotoxin levels associated with substantial higher weight loss in infected burn wounds compared to non-infected burn wounds. In the utilized animal model, weight loss was observed for 6 days post burn in the infected group, whereas no differences in weight loss were observed in the non-infected burn group. However, no pair feeding was carried out within this study in order to explain differences in weight gain or loss. In a clinical setting it is impossible to investigate the reason for hypermetabolism in burn patients [22].
Previous in vivo experiments have demonstrated higher endotoxin levels after bacterial challenge [16,17,23,24]. Furthermore, it has been shown that cachectin and TNF is associated with weight loss [25], which can be reversed by blocking TNF and IL-1 receptors [26]. TNF-α is a multifunctional cytokine that is secreted in response to injury, inflammation or infection. In thermal injury, TNF-α is considered a potent mediator of other cytokines such as IL-6 [25]. Clinical and experimental studies have shown a significant elevation in IL-6 production after burn injury and sepsis, which correlates with suppressed cell-mediated immunity and increased mortality [27-29]. In addition, blockade of IL-6 biological activity following burn injury and/or sepsis has been shown to improve outcome [30]. In our study we did not see any significant difference in IL-6 production. This is due, in part, to the limited burn area (30% BSA) without differences in mortality and development of burn-associated immunosuppression. A number of experimental studies have shown implication of Th-2 cytokines, like IL-10, in immunosuppression after thermal injury as well as in sepsis [31-33]. Recent findings have demonstrated that the macrophages are resistant to the suppressive effects of IL-10 post burn [34,35]. It has been verified that block of endogenous IL-10 enhances IL-6 and TNF-α release in response to LPS. Addition of exogenous IL-10 to the macrophages cultures suppresses inflammatory mediator release [36]. In this study we were able to show that IL-10 expression levels were associated with weight gain in animals with infected burn wounds.
Conclusion
We hypothesize that infection control in burned patients is the most important factor for preventing burn-associated weight loss. Our results suggest that in the presence of infection, you get a SIRS response possibly due to transient endotoxemia that is only seen in the infection group. In contrast, both burn and infection get a late IL-10 (CARS) response, which is then associated with a return to normal weight in the infection group. More studies are clearly needed to fully understand the pathophysiology in infected burn wounds.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
LS, OB and MHF performed most of the experiments. LS, OB, MHF, FJ, ML, GS, AD, HUS, DR, SCW participated in the experimental design, data interpretation and writing of the manuscript
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We wish to express our gratitude to Dr. Nancy Gong and Mita Ghosh for their assistance with experiments. This work was supported in part by the National Institutes of Health Grants GM54911 (SCW), HL03803-01, DK02210 (GLS), DK53296 (GLS).
==== Refs
Baue AE Durham R Faist E Systemic inflammatory response syndrome (SIRS), multiple organ dysfunction syndrome (MODS), multiple organ failure (MOF): are we winning the battle? Shock 1998 10 79 89 9721973
Harris BH Gelfand JA The immune response to trauma Semin Pediatr Surg 1995 4 77 82 7633853
Nguyen TT Gilpin DA Meyer NA Herndon DN Current treatment of severely burned patients Ann Surg 1996 223 14 25 8554414 10.1097/00000658-199601000-00004
Saffle JR Sullivan JJ Tuohig GM Larson CM Multiple organ failure in patients with thermal injury Crit Care Med 1993 21 1673 1683 8222683
Still JMJ Belcher K Law EJ Experience with polymicrobial sepsis in a regional burn unit Burns 1993 19 434 436 8216775 10.1016/0305-4179(93)90069-K
Teodorczyk-Injeyan JA Sparkes BG Mills GB Peters WJ Falk RE Impairment of T cell activation in burn patients: a possible mechanism of thermal injury-induced immunosuppression Clin Exp Immunol 1986 65 570 581 2946498
Steinstraesser L Oezdogan Y Wang SC Steinau HU Host defense peptides in burns Burns 2004 30 619 627 15475133 10.1016/j.burns.2004.05.013
Nizet V Ohtake T Lauth X Trowbridge J Rudisill J Dorschner RA Pestonjamasp V Piraino J Huttner K Gallo RL Innate antimicrobial peptide protects the skin from invasive bacterial infection Nature 2001 414 454 457 11719807 10.1038/35106587
Meakins JL Etiology of multiple organ failure J Trauma 1990 30 S165 8 2254977
Hengge UR Benninghoff B Ruzicka T Goos M Topical immunomodulators--progress towards treating inflammation, infection, and cancer Lancet Infect Dis 2001 1 189 198 11871495 10.1016/S1473-3099(01)00095-0
Kawakami M Terai C Okada Y Changes of the interleukin-6 levels in skin at different sites after thermal injury J Trauma 1998 44 1056 1063 9637162
Reddy RC Chen GH Tekchandani PK Standiford TJ Sepsis-induced immunosuppression: from bad to worse Immunol Res 2001 24 273 287 11817325 10.1385/IR:24:3:273
Steinstraesser L Fohn M Klein RD Aminlari A Remick DG Su GL Wang SC Feasibility of biolistic gene therapy in burns Shock 2001 15 272 277 11303725
Jacobsen F Baraniskin A Mertens J Mittler D Mohammadi-Tabrisi A Schubert S Soltau M Lehnhardt M Behnke B Gatermann S Steinau HU Steinstraesser L Activity of histone H1.2 in infected burn wounds J Antimicrob Chemother 2005 55 735 741 15772144 10.1093/jac/dki067
Jacobsen F Mittler D Hirsch T Gerhards A Lehnhardt M Voss B Steinau HU Steinstraesser L Transient cutaneous adenoviral gene therapy with human host defense peptide hCAP-18/LL-37 is effective for the treatment of burn wound infections Gene Ther 2005 15973442
Steinstraesser L Klein RD Aminlari A Fan MH Khilanani V Remick DG Su GL Wang SC Protegrin-1 enhances bacterial killing in thermally injured skin Crit Care Med 2001 29 1431 1437 11445704 10.1097/00003246-200107000-00022
Steinstraesser L Tack BF Waring AJ Hong T Boo LM Fan MH Remick DI Su GL Lehrer RI Wang SC Activity of novispirin G10 against Pseudomonas aeruginosa in vitro and in infected burns Antimicrob Agents Chemother 2002 46 1837 1844 12019098 10.1128/AAC.46.6.1837-1844.2002
Dehne MG Sablotzki A Hoffmann A Muhling J Dietrich FE Hempelmann G Alterations of acute phase reaction and cytokine production in patients following severe burn injury Burns 2002 28 535 542 12220910 10.1016/S0305-4179(02)00050-5
Cohen J The immunopathogenesis of sepsis Nature 2002 420 885 891 12490963 10.1038/nature01326
Brun-Buisson C The epidemiology of the systemic inflammatory response Intensive Care Med 2000 26 Suppl 1 S64 74 10786961 10.1007/s001340051121
Angus DC Wax RS Epidemiology of sepsis: an update Crit Care Med 2001 29 S109 16 11445744 10.1097/00003246-200107001-00035
Long CL Energy balance and carbohydrate metabolism in infection and sepsis Am J Clin Nutr 1977 30 1301 1310 888781
Klein RD Su GL Aminlari A Zhang H Steinstraesser L Alarcon WH Wang SC Skin lipopolysaccharide-binding protein and IL-1beta production after thermal injury J Burn Care Rehabil 2000 21 345 352 10935817 10.1067/mbc.2000.107542
Steinstraesser L Burghard O Nemzek J Fan MH Merry A Remick DI Su GL Steinau HU Wang SC Protegrin-1 increases bacterial clearance in sepsis but decreases survival Crit Care Med 2003 31 221 226 12545019 10.1097/00003246-200301000-00034
Marano MA Moldawer LL Fong Y Wei H Minei J Yurt R Cerami A Lowry SF Cachectin/TNF production in experimental burns and Pseudomonas infection Arch Surg 1988 123 1383 1388 3178487
Fathalla B Hamada K Vannier E Smith D Edwards C Roubenoff R Effects of aging and cytokine blockade on inflammatory cachexia Clin Exp Rheumatol 2004 22 85 90 15005009
Drost AC Burleson DG Cioffi WGJ Mason ADJ Pruitt BAJ Plasma cytokines after thermal injury and their relationship to infection Ann Surg 1993 218 74 78 8328832
Biffl WL Moore EE Moore FA Peterson VM Interleukin-6 in the injured patient. Marker of injury or mediator of inflammation? Ann Surg 1996 224 647 664 8916880 10.1097/00000658-199611000-00009
Kowal-Vern A Walenga JM Hoppensteadt D Sharp-Pucci M Gamelli RL Interleukin-2 and interleukin-6 in relation to burn wound size in the acute phase of thermal injury J Am Coll Surg 1994 178 357 362 8149035
Dinarello CA Cannon JG Wolff SM Bernheim HA Beutler B Cerami A Figari IS Palladino MAJ O'Connor JV Tumor necrosis factor (cachectin) is an endogenous pyrogen and induces production of interleukin 1 J Exp Med 1986 163 1433 1450 3486936 10.1084/jem.163.6.1433
Friedman G Jankowski S Marchant A Goldman M Kahn RJ Vincent JL Blood interleukin 10 levels parallel the severity of septic shock J Crit Care 1997 12 183 187 9459114 10.1016/S0883-9441(97)90030-7
van der Poll T Speelman P van Deventer SJ [Immunology in clinical practice. VIII. Role of cytokines in the pathogenesis of bacterial infections] Ned Tijdschr Geneeskd 1998 142 14 17 9556983
Van der Poll T Lowry SF Epinephrine inhibits endotoxin-induced IL-1 beta production: roles of tumor necrosis factor-alpha and IL-10 Am J Physiol 1997 273 R1885 90 9435641
Schwacha MG Ayala A Chaudry IH Insights into the role of gammadelta T lymphocytes in the immunopathogenic response to thermal injury J Leukoc Biol 2000 67 644 650 10811004
Schwacha MG Somers SD Thermal injury induces macrophage hyperactivity through pertussis toxin-sensitive and -insensitive pathways Shock 1998 9 249 255 9565252
Seitz M Loetscher P Dewald B Towbin H Gallati H Baggiolini M Interleukin-10 differentially regulates cytokine inhibitor and chemokine release from blood mononuclear cells and fibroblasts Eur J Immunol 1995 25 1129 1132 7737285
|
16168063
|
PMC1242239
|
CC BY
|
2021-01-04 16:28:04
|
no
|
BMC Surg. 2005 Sep 17; 5:19
|
utf-8
|
BMC Surg
| 2,005 |
10.1186/1471-2482-5-19
|
oa_comm
|
==== Front
Cardiovasc UltrasoundCardiovascular Ultrasound1476-7120BioMed Central London 1476-7120-3-271615015010.1186/1476-7120-3-27ReviewMyocardial contractility in the echo lab: molecular, cellular and pathophysiological basis Bombardini Tonino [email protected] Department of Echocardiography, Institute of Clinical Physiology, National Council of Research, Pisa, Italy2005 8 9 2005 3 27 27 27 7 2005 8 9 2005 Copyright © 2005 Bombardini; licensee BioMed Central Ltd.2005Bombardini; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
In the standard accepted concept, contractility is the intrinsic ability of heart muscle to generate force and to shorten, independently of changes in the preload or afterload with fixed heart rates. At molecular level the crux of the contractile process lies in the changing concentrations of Ca2+ ions in the myocardial cytosol. Ca2+ ions enter through the calcium channel that opens in response to the wave of depolarization that travels along the sarcolemma. These Ca2+ ions "trigger" the release of more calcium from the sarcoplasmic reticulum (SR) and thereby initiate a contraction-relaxation cycle.
In the past, several attempts were made to transfer the pure physiological concept of contractility, expressed in the isolated myocardial fiber by the maximal velocity of contraction of unloaded muscle fiber (Vmax), to the in vivo beating heart. Suga and Sagawa achieved this aim by measuring pressure/volume loops in the intact heart: during a positive inotropic intervention, the pressure volume loop reflects a smaller end-systolic volume and a higher end-systolic pressure, so that the slope of the pressure volume relationship moves upward and to the left. The pressure volume relationship is the most reliable index for assessing myocardial contractility in the intact circulation and is almost insensitive to changes in preload and after load. This is widely used in animal studies and occasionally clinically. The limit of the pressure volume relationship is that it fails to take into account the frequency-dependent regulation of contractility: the frequency-dependent control of transmembrane Ca2+ entry via voltage-gated Ca2+ channels provides cardiac cells with a highly sophisticated short-term system for the regulation of intracellular Ca2+ homeostasis. An increased stimulation rate increases the force of contraction: the explanation is repetitive Ca2+ entry with each depolarization and, hence, an accumulation of cytosolic calcium. As the heart fails, there is a change in the gene expression from the normal adult pattern to that of fetal life with an inversion of the normal positive slope of the force-frequency relation: systolic calcium release and diastolic calcium reuptake process is lowered at the basal state and, instead of accelerating for increasing heart rates, slows down. Since the force-frequency relation uncovers initial alteration of contractility, as an intermediate step between normal and abnormal contractility at rest, a practical index to measure it is mandatory.
Measuring end-systolic elastance for increasing heart rates is impractical: increasing heart rates with atrial pacing has to be adjunct to the left ventricular conductance catheter, to the left ventricular pressure catheter, to the vena cava balloon, and to afterload changes. Furthermore, a noninvasive index is needed. Noninvasive measurement of the pressure/volume ratio for increasing heart rates during stress in the echo lab could be the practical answer to this new clinical demand in the current years of a dramatic increase in the number of heart failure patients.
==== Body
Calcium channels: evolutional aspects
When ~3,500,000,000 years ago prokaryotes appeared, the selection of an intracellular messenger preceded the appearance of ionic channels of enveloping lipid membranes.
Calcium had conformational and stechiometric advantages to be chosen as an intracellular messenger (=good messenger and modulator of intracellular processes), due to its high coordination number and irregular coordination geometry. If calcium was the first intracellular messenger, ionic channels for calcium had to appear for the cells to maintain constant intracellular calcium concentrations [1]. Ionic channels, other than function of intracellular ionic concentration surveyors, became to have the first function of reactive capability to outer stimulus, changing abruptly their functions. The primitive Ca2+ channels were activated by mechanical stimuli, present but slow and low efficient in reactions. But faster reactions are essential for survival.
When ~1,500,000,000 years ago the hydrosphere became aerobic and the primitive unicellular organisms developed mechanisms of energy production, eukaryotes developed active transport and voltage-gated channels as a result of selection for faster signaling [1,2].
When in this new evolutional cell a mechanical stimulus involving a portion of the membrane hits the cell, a depolarization produced by the mechanic-sensitive Ca2+ channels subsequently extends the effect to the entire cell by voltage gated Ca2+ channels.
Toward the Na channel: high-speed signaling required by multicellularity
With the appearance of multicellularity, even faster signaling had to appear for life competition. As a possible solution to this new requirement, increase in density of Ca2+ channels would increase the speed of the depolarizing wave, but would have compromised the role of Ca2+ as modulator of intracellular function.
The appearance of Na+ channels, capable of carrying greater ionic fluxes without interfering with intracellular processes, would be evolutionarily favorable. And, in fact, this model was the one chosen by evolution in nerve cells: earlier selection of Na+ channels to sustain potential changes [1]. In fact, also for a maximum concentration on the cell membrane of channels, Ca2+ channels mediated maximum velocity is equal to 0.10 m/sec. For Na+ channels mediated max. velocity is 3 m/sec.
Muscle fibers: a particular evolutionary aspect
In primitive muscle fibers all the activating Ca2+ ions for contraction came from outside the cell.
High density of Ca2+ channels would increase the speed of the depolarizing wave and at the same time would speed up the activation of the contractile machinery.
But with the evolutionary appearance of intracellular calcium stores (sarcoplasmic reticulum) in supplying Ca2+ ions for contraction, muscle fibers acquired the capability of stronger contraction with less energy consumption. As for force of contraction, also signaling speed was a problem in muscle cells: but in these cells less pronounced advantages are provided by an exclusively Na+-channels-based membrane potential changes. The simultaneous presence of fast Na+ channels for conduction function and slow Ca2+ channels for beginning the contraction mechanism through inward calcium flow was maintained [3].
At the present evolutionary state, once the impulse has formed in the sinus node, it spreads very rapidly throughout the atrium to reach the atrioventricular (AV) node and ventricles. In atrial tissue, the pattern of the action potential is dominated by a fast sodium channel. The action potential duration of atrial tissue is short when compared to that of ventricles, and the inward flow of calcium ions is less, with a lower force of contraction developed in the atria and lower activity of L-type calcium channels. Conduction of the wave of depolarization is rapid through conduction tissues where the action potential goes through fast sodium channel activity, whereas conduction is lower through the ventricular myocardium, where there is chiefly calcium channel activity with a slower rate of depolarization.
High frequency-induced upregulation of human cardiac calcium currents: the final evolution
Ultimately developed, the frequency-dependent control of transmembrane Ca2+ entry via voltage-gated Ca2+ channels provides mammalian cardiac cells with a highly sophisticated short-term system for regulation of intracellular Ca2+ homeostasis.
Up-regulation of Ca2+ entry through Ca2+ channels by high rates of beating (HFIUR of ICa) is involved in the frequency-dependent regulation of contractility: this process is crucial in adaptation to exercise and stress [4,5]. This regulation is rapid, (the steady state is reached rapidly within few seconds for each heart rate level), intrinsic to the myocardium cell, with no necessity to be driven from neuronal or hormonal controls (Fig. 1).
Figure 1 High frequency-induced upregulation of human cardiac calcium currents in isolated cardiomyocytes. Up regulation of Ca2+ entry through Ca2+ channels by high rates of beating is involved in the frequency-dependent regulation of contractility: for each increasing heart rate the steady state is reached rapidly (within few seconds, on the left : FFR). Beta-adrenergic receptor stimulation produces an important enhancement of the force-frequency relation on myocardial contractility: β-adrenergic stimulation, by means of cyclic adenosine monophosphate, promotes phosphorylation and the opening probability of the Ca2+ channel. The effect of increasing contractility by increasing heart rate ("pure" Bowditch treppe) is intrinsic to myocardium and takes few seconds to occur, while the β-adrenergic amplification of the force-frequency relation takes longer, i.e. 30–40 seconds, the time it takes for β-receptor activation and cAMP synthesis (on the right: FFR + ISO). (Modified from: Piot C, Lemaire S, Albat B, et al. High frequency-induced upregulation of human cardiac calcium currents. Circ 1996; 93:120–8)
Contractility
Molecular aspects: calcium ion fluxes in cardiac contraction-relaxation cycle
The crux of the contractile process lies in the changing concentrations of Ca2+ ions in the myocardial cytosol.
Crucial features are [6] entry of Ca2+ ions through the voltage-sensitive L-type Ca2+ channels, acting as a trigger for the release of Ca2+ ions from the sarcoplasmic reticulum (SR).
Relatively small amounts of calcium ions actually enter and leave the cell during each cardiac cycle, whereas much larger amounts move in and out of the sarcoplasmic reticulum. Calcium-induced calcium release explains most of the current available data. This process elevates by about tenfold the concentration of calcium ions in the cytosol. The result is the increasing interaction of calcium ions with troponin C to trigger the contractile proteins [6] (Fig. 2).
Figure 2 Molecular basis of contractility in normal heart. Crucial features are entry of Ca2+ ions through the voltage-sensitive L-type Ca2+ channels in response to the wave of depolarization, acting as a trigger for the release of Ca2+ ions from the sarcoplasmic reticulum (SR). The crux of the contractile process lies in the changing concentrations of Ca2+ ions in the myocardial cytosol. The varying actin-myosin overlap is shown for systole, when calcium ions arrive, and diastole, when calcium ions leave. At the end of systole, calcium stops interaction with troponin C and calcium ions are taken up into the SR by the activity of the pump called SERCA. Calcium taken up into the SR by the calcium uptake pump is stored within the SR before further release. The small amount of calcium that has entered the cell leaves predominantly through a Na+/Ca2+ exchanger. (Modified from Opie LH. Normal and abnormal cardiac function. Chapter 14, page 443. In Braunwald Zipes Libby Heart disease, 6th edition, W. B Saunders Company, 2001)
Left ventricular contraction
Left ventricular pressure starts to builds up when the arrival of calcium ions at the contractile proteins starts to trigger actin-myosine interaction. The thin actin filament interacts with the myosin head when Ca2+ ions arrive at troponin C (TnC). As more and more myofibers enter the contracted state, pressure development in the left ventricle proceeds. The interaction of actin and myosine increases, and cross-bridge cycling augments. As long as enough calcium ions are bound to troponin C, many repetitive cycles of this nature occur. The enhanced force development in response to a greater calcium ion concentration is due to recruitment of additional cross bridges.
When calcium ions depart from their binding sites on troponin C, cross-bridge cycling cannot occur and the diastolic phase of the cardiac cycle sets in.
Left ventricular relaxation
At the end of systole, calcium stops interacting with troponin C and calcium ions are taken up into the SR by the activity of the SERCA (sarcoplasmatic reticulum Ca2+ ATPase) pump that constitutes nearly 90% of the protein component of the SR. Calcium taken up into the SR by the calcium uptake pump is stored within the SR before further release. To balance the small quantity of calcium ions entering the heart cell with each depolarization, a similar quantity must leave the cell. First, calcium can be exchanged for sodium ions entering by the Na+/Ca2+ exchange and, second, an ATP-consuming sarcolemmal calcium pump can transfer calcium into this extra cellular space against a concentration gradient.
As the cytosolic calcium ion concentration starts to decline because of the uptake of calcium into the SR under the influence of activated phospholamban, more and more myofibers enter the state of relaxation (Fig. 2).
Preload and after load
The preload is the load present before contraction has started, at the end of diastole. When the preload increases, the left ventricle distends during diastole, and the stroke volume rises according to Starling's law [7]. The proposed explanation for the Starling effect, whereby a greater end-diastolic fiber length develops a greater force, is explained by an interaction between sarcomere length and calcium ions (length sensitization of the sarcomere): 1) increase in end-diastolic fiber length at any given free Ca2+ concentration would increase force by a small amount on the basis of the change in filament overlap; 2) when the fiber is stretched and the sarcomere length increases, for any given number of Ca2+ ions binding to TnC, there is greater force development. Length sensitization of the sarcomere explains how the sarcomere can "upgrade itself" to a higher force-length curve [6].
The afterload is the systolic load on the left ventricle after it has started to contract.
Increased afterload means that an increased intraventricular pressure has to be generated first to open the aortic valve and then during the ejection phase [8].
In the nonfailing heart, the left ventricle can overcome any physiological acute increase in load [6].
Contractility: how can it be defined?
"Contractility is the inherent capacity of the myocardium to contract independently of changes in the preload or afterload. Whatever the problems of measuring it, contractility remains an essential corner concept to separate the effects of a primary change in loading conditions from an intrinsic change in the force of contraction [6]". It is a basic property of cardiac muscle and is strictly linked to the activation quantity of actin myosin transverse bridges in the myocardial fibers, and to the velocity of cross-bridge activation at the systole onset [6,9]. Cytosolic calcium level is the determinant of:
- The myocardial fiber number involved in the contraction process.
- The maximal velocity of myocardial fibers shortening.
Increased contractility, is reflected in higher myocardial fiber shortening velocity, with a more highly developed tension peak and a steeper pressure rise, when preload, afterload, and heart rate are constant: in the cytosol calcium release is more and faster from SR with a higher cytosol calcium concentration in systole: more troponin is activated from higher levels of calcium with more acitn-myosin cross-bridges in the time unit, and ultimately myocardial fiber contraction is more and faster.
Decreased contractility is reflected in lower myocardial fiber shortening velocity, with a lower tension peak and a blunted pressure rise, when preload, after load, and heart rate are constant: in the cytosol calcium release is less and slower from SR with a lower cytosol calcium concentration in systole: less troponin is activated from lower levels of calcium with less actin-myosin cross-bridges in time unit, and ultimately myocardial fiber contraction is less and slower.
The isolated myocardial fiber: idealized contractility in the physio lab
Contractility expressed in the isolated myocardial fiber is the maximal velocity of contraction of unloaded muscle fiber (Vmax). This value is defined as the maximal velocity of contraction, when there is no load on the isolated muscle. This strictly preload and after load independent index, fulfills the theoretical requirements for contractility quantification and greatly contributes to this research field [9,10]. Nevertheless, this model is not usable in in-vivo conditions.
The in vivo, beating heart: how to measure contractility
In the past attempts were made to transfer the purely physiological concept of contractility expressed in the isolated myocardial fiber by the maximal velocity of contraction of unloaded muscle fiber (Vmax), to the in vivo beating heart. Suga and Sagawa achieved this aim by measuring pressure/volume loops in the intact heart: during a positive inotropic intervention, the pressure volume loop reflects a smaller end systolic volume and a higher end-systolic pressure, so that the slope of the pressure volume relationship (Ees) moves upward and to the left [11,12]. Ees is the most reliable index for assessing (standard) myocardial contractility at rest in the intact circulation and is almost insensitive to changes in preload, and after load.
This is widely used in animal studies and occasionally clinically [6].
A now-time conductance catheter is used for human studies [13]. The method is highly correct, but invasive, complex, and technically demanding. (Fig. 3)
Figure 3 Pressure-volume loops in the cath lab. A conductance catheter is used to measure pressure-volume loops in humans. The time landmarks during the cardiac cycle include the following: B, aortic valve opening and the beginning of ejection; C, aortic valve closure; D, mitral valve opening; and A, end-diastole. During diastole (D-A tract) LV filling occurs, with a low end-diastolic LV pressure increase in the normal heart. During isovolumic contraction, or pre-ejection systole, (A-B tract) LV volume is unchanged but LV pressure rises to point B when it equals aortic pressure, and the aortic valve opens: isotonic systole, or systolic ejection phase (B-C tract), starts. When LV systolic emptying ends (C point), the aortic valve closes, and isovolumic diastolic relaxation starts. (C-D tract). Smaller end-systolic volume and higher end-systolic pressure are typical markers of higher contractility. Counter-directional changes identify compromised contractility. Focusing on end-systolic volume and on end-systolic pressure it immediately appears that the upper left corner of the pressure volume loop (C point) quantifies both measures.
Focusing on cytosol calcium concentrations along the pressure-volume loop, (Fig 3) in diastole (D-A tract) cytosolic calcium is reuptake from cytoplasm and stored in the SR [6]. At the A end-diastolic point, the end-diastolic volume (or maximal myocardial fiber length) predicts contractile-proteins calcium-sensitivity of the upcoming systole according to the Starling's law [7]. The velocity of the pressure development in the isovolumic systole (A-B tract), and the ejection force in the isotonic systole (B-C tract) are both strictly linked to the contractile state. When LV systolic emptying ends (C point), the aortic valve closes, and isovolumic diastolic relaxation starts. (C-D tract).
More highly developed systemic pressure simultaneously with lower end-systolic volume is typical of higher contractility. Counter-directional changes identify compromised contractility.
If end-systolic volume is measured for different end-systolic pressure values, sequential end systolic pressure/volume values can be recorded (C, C1, C2, C... points). The upper left corners (C, C1, C2, and C... points) of the loops define the LV end-systolic pressure-volume relation (ESPVR). (Fig. 4) The ESPVR predicts in a heart with constant contractility the end-systolic volume when end-systolic pressure changes, and ultimately predicts the left ventricle ability to empty for different afterload values [14].
Figure 4 The end-systolic pressure-volume relationship (ESPVR). Suga and Sagawa were the first to use simultaneous LV pressure-volume measurements. These Authors, searching for a preload and afterload independent contractility index, measured pressure-volume loops during sudden preload and afterload changes. The upper left corners of the loops (C, C1, C2, C... points) define the LV end-systolic pressure-volume relation (ESPVR). ESPVR predicts the end-systolic volume in a heart with constant contractility when end-systolic pressure changes, and ultimately predicts the left ventricle ability to empty for different afterload values. The slope of the ESPVR line is the end-systolic elastance (Ees). In the clinical setting it is difficult to generate the end-systolic pressure-volume relationship (ESPVR) free of changes in reflex-mediated variations in contractility. It also requires a means to measure pressure and volume accurately and simultaneously.
Contractility is quantified by the angular coefficient (or slope) of the ESPVR relation: the Ees (end systolic elastance) (Fig. 5).
Figure 5 Load changes at constant contractility (left) and contractility changes at constant load (right). Left panel. The graph shows how two additional pressure-volume loops appear with an acute increase in afterload or preload. Contractility is quantified by the ESPVR slope: the Ees (end systolic elastance). Right panel. Increased contractility, is reflected in higher myocardial fiber shortening velocity, with a more highly developed tension peak and a steeper pressure rise, when preload, after load, and heart rate are constant: Ees moves upward and to the left. The left ventricular emptying fraction or ejection fraction (LVEF) is reflected in the ability of the left ventricle to empty. Because myocardial contractility is an important determinant of LVEF, LVEF and contractility are frequently considered to be interchangeable. But they are not the same: thus it is possible to have low LVEF despite normal contractility when LV afterload is excessive. Alternatively, LVEF may be nearly normal despite decreased myocardial contractility if LV afterload is low. (Modified from Little WC. Assessment of normal and abnormal cardiac function. Chapter 15, page 480. In Braunwald Zipes Libby Heart disease, 6th edition, W. B Saunders Company, 2001)
Contractility and heart rate
The heart contractility dependence on increasing heart rates has been established in most mammalians.
The inherent ability of ventricular myocardium to increase its strength of contraction independently of neurohormonal control, in response to an increase in contraction frequency is known as frequency treppe [4] (Fig. 6). In humans this myocardial property causes the contractile force to rise, as contraction frequency is increased from 60 to about 180 bpm and to then decline with further increase in frequency (the force-frequency relation "FFR") [6].
Figure 6 Force-frequency relation or Bowditch treppe. Developed force of contraction in the isolated papillary muscle at increasing stimulation rates. The stimulus rate is shown as the action potential duration on an analog analyzer. The tension developed by papillary muscle contraction is shown as developed force. An increased stimulation rate increases the force of contraction. On cessation of rapid stimulation, the contraction force gradually declines. Heart rate is a leading determinant of cytosol calcium concentration, and strictly linked to contractility. In the healthy heart, a frequency increase up to 180 beats per minute provides for faster systolic calcium SR release (increased contractility or developed force) and for faster diastolic SR calcium reuptake (positive lusitropic effect). (Modified from Opie LH. Normal and abnormal cardiac function. Chapter 14, page 443. In Braunwald Zipes Libby, Heart disease, 6th edition, W. B Saunders Company, 2001).
Molecular basis
Heart rate is a leading determinant of cytosol calcium concentration, and strictly linked to the contractility levels. In the healthy heart, a frequency increase up to 180 beats per minute provides systolic faster calcium SR release (increased contractility or developed force) and diastolic faster SR calcium reuptake (positive lusitropic effect).
Up-regulation of Ca2+ entry through Ca2+ channels by high rates of beating (HFIUR of ICa) is involved in the frequency-dependent regulation of contractility: this process is crucial in adaptation to exercise and stress [5]. This regulation is rapid, (the steady state is reached rapidly within few seconds for each heart rate level), intrinsic to the myocardium cell, with no need to be driven from neuronal or hormonal controls (Fig. 1, Fig. 6).
Cellular and myocardial fiber level
This property has been definitively established in the human heart in experimental settings using cardiomyopathic myocardial strips.
Measurements of twitch tension in isolated left-ventricular strips from explanted cardiomyopathic hearts compared with non-failing hearts show reduction in peak rates of generation and relaxation of twitch tension and a decrease in slope of tension rate vs. contraction frequency [15,16] (Fig. 7).
Figure 7 Plots of average steady-state isometric twitch tension versus stimulation frequency in non-failing and failing myocardium. Measurements of twitch tension in isolated left-ventricular strips from explanted cardiomyopathic hearts compared with non-failing hearts show reduction in peak rates of generation and relaxation of twitch tension and a decrease in slope of tension rate vs. contraction frequency The FFR of these failing groups both exhibit a negative treppe at contraction frequencies above about 100 bpm. The contraction frequency at which the FFR begins its descending limb ("optimum stimulation frequency") declines progressively in the order: ASD (atrial septal defect), CAD (coronary artery disease), IDDM (diabetic myopathy), MR (mitral regurgitation), DCM (dilated cardiomyopathy). (Modified from: Mulieri AL. In "Heart Metabolism in Failure" R.A. Howarth Ed. 1997. The role of myocardial force-frequency relation in left ventricular function and progression of human heart failure)
The FFR of these failing groups both exhibit a negative treppe at contraction frequencies above about 100 bpm.
Presence of a negative treppe in the working range of heart rates may constitute an additional liability beyond mere depression of the wall tension since this may contribute to an accelerated progression of heart failure. In patients in end stage failure the peak of the FFR occurs at such a low frequency that there is a negative treppe over the entire in vivo range of heart rates. The contraction frequency at which the FFR begins its descending limb ("optimum stimulation frequency") declines progressively in the order: atrial septal defect, coronary artery disease, diabetic myopathy, mitral regurgitation, dilated cardiomyopathy. This suggests that a correlation between severity of myocardial disease and optimum contraction frequency may exist [15,16].
In more severe heart failure the peak of the FFR is shifted sufficiently to lower frequencies so that it has a negative slope over the entire range of in vivo heart rates (i.e., 80–150 bpm). The weakening of contractile strength as heart rate rises suggest the possibility that in vivo, a sudden increase in heart rate could predispose the ventricle to being stretched by venous return.
While the FFR is well known in the physiological lab [5,17,18], with extensive studies in isolated strips of failing myocardium [15,16], in animal models of heart failure [19,20], till now its knowledge and use in the clinical setting is extremely limited [21-26].
Fetal gene program: back from the future
"As the ventricle fails, there is a change in the ventricular gene expression pattern from the normal adult pattern to that normally observed only during fetal life. There is a down regulation of the calcium uptake pump (SERCA2) and of the fast-contracting myosin heavy chain. The fetal program may be activated from cytosolic calcium overload, by adding phosphate groups to enzymes that normally inhibit the fetal program [6]" (Fig. 8).
Figure 8 Molecular basis of contractility in failing heart. There is increasing evidence that disturbances in calcium handling play a central role in the disturbed contractile function in myocardial failure. The sarcoplasmic reticulum calcium ATPase (SERCA) is depressed both in function, as well as in expression. At the same time the sarcolemmal sodium-calcium (Na+/Ca2+) exchanger is increased both in function and in expression. The result is a characteristic change in calcium homeostasis with decreased diastolic uptake of calcium into the sarcoplasmic reticulum with subsequently reduced calcium release during the next systole, resulting in reduced contractile performance. At the same time increased capacity of the sodium-calcium exchanger extrudes intracellular calcium ions to the extra-cellular space, thereby rendering these ions unavailable for the contractile cycle. Intracellular Ca2+ handling is abnormal in heart failure and cause systolic and diastolic dysfunction. The mRNA and protein levels of the Na+/Ca2+ exchanger are increased in myocites from heart failure patients and correlates inversely with the SERCA mRNA levels. The augmentation in Na+/Ca2+ exchange activity is a compensatory response to the reduction in Ca2+ reuptake caused by a decrease in SERCA2. But enhanced Na+/Ca2+ exchange instead of SRCa2+ reuptake is an energy-wasting process: ATP consumption to extrude cytosolic Ca2+ from the myocyte is almost doubled with respect to the normal SRCa2+ reuptake.
Changes in the calcium cycle are fundamental to the impaired contractile performance of the failing heart. The SR calcium stores are severely depleted because of the combined effects of depressed calcium uptake into the SR resulting from decreased SERCA activity, both down-regulated and inhibited. Thus, the calcium ions entering with depolarization are unable to trigger the release of enough calcium to generate a normal calcium transient (Fig. 9). There is a close relationship between the depression of SERCA in human heart failure and the depressed force-frequency relationship. Paradoxically, the diastolic calcium level is higher than normal. Starting from this higher level, as the heart rate increases, the calcium ions enter more rapidly through the calcium channels than can be extruded through the Na+/Ca2+ exchange, so that the diastolic levels rise, as does the diastolic tension.
Figure 9 Molecular pathopysiology, action potentials and calcium transients in isolated myocytes of normal (A) vs. failing (B) hearts. Upper panels. Left: normal myocyte. Right: failing heart myocytes show depressed SERCA both in function, and in expression; the sarcolemmal sodium-calcium (Na+/Ca2+) exchanger is increased both in function and in expression, and correlates inversely with the SERCA levels. Lower panels: action potential and intracellular calcium transient. The action potentials recorded in myocytes isolated from the failing hearts (right) are markedly prolonged compared with that in a myocyte from a normal heart (control, left). The intracellular calcium transients measured with the fluorescent calcium indicator fura-2 are also markedly abnormal in myocytes isolated from the failing heart (right). Compared with a normal myocyte (control, left), the failing myocyte shows (plot s) an attenuated cytosolic Ca2+ rise with depolarization and a markedly delayed return to baseline. The intracellular calcium transient (plot d) from a myocyte with isolated diastolic dysfunction (normal cytolsolic Ca2+ systolic release, delayed cytosolic Ca2+ diastolic removal) shows a normal rise with depolarization and a markedly delayed return to baseline. These abnormalities reflect the altered expression or function of key calcium-handling proteins and contribute to the abnormal action potential in the top illustration. (Modified from: O'Rourke B, Kass DA, Tomaselli GF, et al. Mechanisms of altered excitation-contraction coupling in canine tachycardia-induced heart failure I. Circ Res 1999; 84: 562–70.)
Muscle strips prepared from patients with severe heart failure behave very differently from normal muscle, in that there is hardly any response to an increased stimulation frequency. Whereas in strips from normal hearts, optimal force development is reached at rates of about 150 to 180 beats/min, in patients with cardiomyopathy an increased heart rate produces a decreased twitch tension (Fig. 7). In addition, the diastolic tension rises markedly with the stimulation frequency, compatible with a rate-induced cytosolic calcium overload causing diastolic dysfunction [6].
Present limits of end systolic elastance (ESPVR slope) for contractility measurement
This standard, historically accepted, rest-assessed contractility, is limited because it is an invasive index, but especially because it fails to take into account frequency-dependent regulation of contractility: ultimately developed in the evolutional scale, as a typical feature of more advanced mammalian species, absent in fetal life and in adults with heart failure-induced regression of the contractile mechanism, the frequency-dependent control of transmembrane Ca2+ entry via voltage-gated Ca2+ channels provides mammalian cardiac cells with a highly sophisticated short-term system for regulation of intracellular Ca2+ homeostasis.
The impossibility of separating the cellular mechanism of contractility changes from those of load or heart rate is now clear. "Thus, there is a clear overlap between contractility, which should be independent of load or heart rate, and the effects of load and heart rate on the cellular mechanism. Hence, the traditional separation of inotropic state from load or heart rate effects as two independent regulators of cardiac muscle performance is no longer simple now that the underlying cellular mechanisms have been uncovered [6]." This topic is not important only as a speculative concept, but especially clinically: in fact as the heart fails, there is a change in the ventricular gene expression pattern from the normal adult pattern to that normally observed only during fetal life, as a memory of primordial contraction patterns, with an inversion of the normal positive slope of the relation: the systolic calcium release and diastolic calcium reuptake process is lowered at the basal state and, instead of accelerating for increasing heart rates, it slows down. Since the assessment of FFR shows initial alteration of contractility, as an intermediate step between normal and abnormal contractility at rest, a practical index to measure it is mandatory.
Since end-systolic elastance (Ees), expressing the slope of the in-vivo, end-systolic ventricular pressure vs chamber volume relation, is the most "foolproof' window into in vivo myocardial contractility, Ees should be measured at each heart rate step increase, as made by Liu and coworkers [21] (Fig. 10).
Figure 10 Force-frequency relationship in the cath lab. During a pressure-volume loop study the contractility was quantified at baseline and during heart rate increase (atrial pacing). At each incremental heart rate the upper left corners of the loops define the LV end-systolic pressure-volume relation (ESPVR). The slope of the ESPVR is the end-systolic elastance (Ees). Upper left panel. During atrial pacing in a control subject (Control) for higher heart rates the ESPVR is shifted leftward, and Ees increases: contractility increases as heart rate increases. Upper right panel. A patient with severe LV hypertrophy (Hypertensive cardiomyopathy) displays a decrease in the ESPVR slope for heart rate increases: from 70 to 100 bpm and at further increases in heart rate (from 100 to 120 and to 150 bpm): contractility decreases at higher heart rates. Lower panel. For each study group end-systolic elastance (Ees, mean value ± SD) is plotted at different heart rates during rapid atrial pacing; for the 8 control (controls, non-LVH) patients, the Ees increased with each increment in heart rate. In contrast, Ees fell at faster rates in hypertensive (HYP) subjects. (Modified from: Liu C. Diminished contractile response to increased heart rate in intact human left ventricular hypertrophy. Circulation 1993; 88:1893)
But measuring Ees for increasing heart rates is impractical: increasing heart rates obtained with temporarily pacing has to be adjunct to the LV conductance catheter, the LV pressure catheter, the vena cava balloon, and to afterload changes. Proof of this is that only Liu [21] adopted this method in humans. (Table 1).
Table 1 Force-frequency relationship from the experimental lab to clinical applications
Author Feldman Bhargava Hasenfusss Liu Inagaki Schuler Dehmer Lavie
Journal J Clin Invest Am J Cardiol Eur Heart J Circ Circ Am j Cardiol Am J Cardiol Chest
Year 1988 1988 1994 1993 1999 1982 1981 1989
Method cath lab cath lab cath lab cath lab cath lab nuc nuc nuc
FORCE SP/ESV dP/dt dP/dt Ees dP/dt SP/ESV SP/ESV SP/ESV
TREPPE Yes Yes Yes Yes Yes Base- peak Base-peak Base-peak
HR increase PM PM PM PM PM
EX
ISO EX EX EX
PTS# Disease DC 7 DC 5 DC 9 HYP 10 HYP 17 AR 14 AR 17 MR 11
FFR Upsloping 3 - - - 7 7 11 7
Flat-Biph - 2 - 10 10 7 2 2
Neg 4 3 9 - - - 4 2
Control # 6 3 8 8 10 9 15 -
FFR Upsloping 6 3 8 8 10 9 15 -
Flat-Biph - - - - - - - -
Neg - - - - - - - -
PM = atrial pacing; EX = exercise; ISO = isoproterenol ; DC = dilated cardiomyopathy; HYP = hypertensive cardiomyopathy; CHD = coronary artery disease; AR = aortic regurgitation; MR = mitral regurgitation; FFR = force-frequency relation
Several attempts have been made to transfer the force-frequency relationship from the experimental lab to clinical applications. Such attempts have been based on invasive evaluation in cath lab (Feldman 1988, Bhargava 1988, Hasenfuss 1994, Liu 1993, Inagaki 1999), or noninvasive evaluation with radionuclide scintigraphy (Schuler 1982, Dehmer 1981, Lavie 1989). The extensively adopted maximum rate of pressure rise (max dP/dt) for force measurement is largely preload and afterload dependent. Since End-systolic elastance (Ees), is almost insensitive to changes in preload and afterload, Ees should be measured at each heart rate step increase, as done by Liu and coworkers.
But measuring Ees for increasing heart rates is impractical: increasing heart rates obtained with atrial pacing has to be adjunct to the LV conductance catheter, the LV pressure catheter, the vena cava balloon, and the afterload changes. Proof of this is that only Liu adopted this method in humans.
The scintigraphic approach is noninvasive, but requires exposure to ionizing radiations and – due to limited temporal resolution – allows the measurement of SP/ESV only at baseline at peak exercise. The pattern of the force-frequency relationship over a spectrum of different heart rates cannot be assessed.
If assessment of Ees is difficult under clinical conditions at fixed heart rates, assessment of Ees for increasing heart rates is much more difficult.
The Suga index (SP/ESV ratio) for increasing heart rates: the link toward the stress echo lab FFR measurement in a practical clinical method
A simpler approach was utilized by Feldman and co-workers [26] in DCM pts vs. normal hearts, by measuring the SUGA index (SP/ESV ratio, instead of Ees) at baseline, and for pacing induced heart rate increase to 25 and 50 bpm beyond basal heart rate. SP/ESV ratio measurement is simpler than Ees measurement, and equally provides knowledge of an up-sloping, flat, or biphasic Bowditch treppe (Fig. 11).
Figure 11 The Suga (SP/ESV) index instead of end-systolic elastance for FFR measurement. Since End-systolic elastance (Ees), expressing the slope of the in-vivo, end-systolic ventricular pressure vs. chamber volume relation, is the most "foolproof' window into in vivo myocardial contractility, Ees should be measured at each heart rate step increase. A simpler approach was utilized by Feldman and co-workers by measuring SP/ESV ratio at baseline, and for pacing induced heart rate increase to 25 and 50 bpm beyond basal heart rate. Feldman showed that 7 patients with dilated cardiomyopathy (DCM) demonstrated little or no significant enhancement in SP/ESV ratio during atrial pacing tachycardia. The lack of improvement in cardiomyopathy patients has been contrasted to patients with normal ventricular function (Control) who demonstrated significant increase in SP/ESV ratio. SP/ESV ratio is simpler than Ees measurement, and equally provides knowledge of up-sloping vs flat-biphasic force-frequency relationship. (Modified from: Feldman MD, Alderman JD, Aroesty JM, Royal HD, Ferguson JJ, Owen RM, et al. Depression of systolic and diastolic myocardial reserve during atrial pacing tachycardia in patients with dilated cardiomyopathy. J Clin Invest 1988; 11:1661–9)
Force-frequency relationship in the stress echo lab: a practical, noninvasive, modern approach to contractility
Non-invasive methods [27-30] have been proposed to assess the rest-peak stress change in inotropic state, based upon the assumption that positive inotropic interventions are mirrored by smaller end-systolic volumes and higher end-systolic pressures (Table 1). During bicycle stress echocardiography, dobutamine or pacing stress, continuous 2D echo monitoring is performed by protocol and blood pressure, ECG and left ventricular volumes are obtained at each step, providing the basic information required to build a force-frequency relationship over a wide range of frequencies (Fig. 12). A totally noninvasive estimation of force-frequency relation during stress in the echo lab is theoretically appealing for the identification of limited contractile reserve and latent global left ventricular dysfunction.
Figure 12 Stress echo lab: contractility me too? Blood pressure analysis. One investigator records all blood pressures at rest and during exercise during the study. The blood pressure recording is made using a manometer sphygmomanometer and the diaphragm of a standard stethoscope. Echocardiography is performed using conventional two-dimensional echocardiography and tissue harmonic imaging and digitized on-line into a quad screen, cineloop format. Left ventricular end-systolic volumes are measured from apical four and two chamber view, using the biplane discs-method. To build the force-frequency relationship, the force is determined at each step as the ratio of the systolic pressure (cuff sphygmomanometer)/end-systolic volume index (biplane Simpson rule/body surface area).
This method is similar to the previously proposed ones but is totally noninvasive, with echocardiography used to assess LV volumes during exercise and cuff blood pressure to estimate peak systolic pressure as an index of end-systolic pressure [31-33].
Bowditch treppe and stress echo. Methodology
During (exercise, DOB or pacing) stress echocardiography continuous 2D echo monitoring is performed by protocol and blood pressure, ECG and left ventricular volumes are obtained at each step, providing the basic information required to build a force-frequency relation over a wide range of frequencies [34].
This approach is based on serial assessment of these variables at different exercise steps so that the force-frequency pattern (up sloping, flat, and biphasic) can be assessed (Fig 13).
Figure 13 FFR, from myocardial strips to the echo lab. Time sequence during stress echo (upper panel). The force frequency relation is built off line. The force-frequency relationship is defined up-sloping when the peak exercise SP/ESV index is higher than baseline and intermediate stress values; biphasic, with an initial up-sloping followed by a later down-sloping trend, when the peak exercise systolic pressure/end-systolic volume index is lower than intermediate stress values; flat or negative, when the peak exercise systolic pressure/end-systolic volume index is equal to or lower than baseline stress values. The critical heart rate (or optimum stimulation frequency) is defined as the heart rate at which systolic pressure/end-systolic volume index reaches the maximum value during progressive increase in heart rate; in biphasic pattern, the critical heart rate is the heart rate beyond which the systolic pressure/end-systolic volume index has declined by 5%; in negative pattern the critical heart rate is the starting heart rate. The critical heart rate (or optimum stimulation frequency) is the human counterpart of the treppe phenomenon in isolated myocardial strips; the optimal heart rate is not only the rate that would give maximal mechanical performance of an isolated muscle twitch, but also is determined by the need for diastolic filling. ASD = atrial septal defect; CAD = coronary artery disease; IDDM = diabetic myopathy; MR = mitral regurgitation; DCM = dilated cardiomyopathy.
Baseline and stress echocardiography
The patient undergoes transthoracic echocardiography at baseline and at each 10 beat frequency increase during stress. This is performed using conventional two-dimensional echocardiography and tissue harmonic imaging, and digitized on-line into a quad screen, cineloop format. Images are also recorded on half-inch S-VHS videotape. Left ventricular end-diastolic and end-systolic volumes are measured from apical four and two chamber view, by an experienced observer using the biplane discs-method [35,36] (Fig. 12). Only representative cycles are measured and the average of three measurements is taken. The endocardial border is traced, excluding the papillary muscles. The frame captured at the R wave of the ECG is considered to be the end diastolic frame, and the frame with the smallest left ventricular cavity the end systolic frame.
Blood pressure analysis
One investigator records all blood pressures at rest and during exercise during the study. The blood pressure recording is made using a manometer sphygmomanometer and the diaphragm of a standard stethoscope (Fig. 12).
End-systolic pressure-volume determination
To build the force-frequency relationship, the force is determined at each step as the ratio of the systolic pressure (cuff sphygmomanometer)/end-systolic volume index (biplane Simpson rule/body surface area). The force frequency relation is built off line (Fig. 13). The slope of the relationship is calculated as the ratio between SP/ESV (Systolic Pressure/End-Systolic Volume) index increase (from baseline to peak exercise)/heart rate increase (from baseline to peak exercise). The force-frequency relationship is defined up-sloping when peak exercise SP/ESV index is higher than baseline and intermediate stress values (Fig. 14); biphasic, with an initial up-sloping followed by a later down-sloping trend, when peak exercise systolic pressure/end-systolic volume index is lower than intermediate stress values [6,25] (Fig. 15); flat or negative, when peak exercise systolic pressure/end-systolic volume index is equal to or lower than baseline stress values (Fig. 16). The critical heart rate (or optimum stimulation frequency) is defined as the heart rate at which systolic pressure/end-systolic volume index reaches the maximum value during progressive increase in heart rate; in biphasic pattern, the critical heart rate is the heart rate beyond which systolic pressure/end-systolic volume index has declined by 5%; in a negative pattern the critical heart rate is the starting heart rate [25].
Figure 14 Force-frequency curve with stress echo in a normal subject. Upper panel: On the left, systolic blood pressure by cuff sphygmomanometer (SP, first row); left ventricular end-systolic volumes calculated with biplane Simpson method (ESV, second row); heart rate increase during stress (bpm, third row); in the lowest row, the force-frequency curve built off-line with the values recorded at baseline (second column), and at different steps (third, fourth, fifth column) up to peak stress (sixth column). An increased heart rate is accompanied by an increased systolic pressure with smaller end-systolic volumes (normal up sloping force-frequency relation). Lower panel: molecular basis (first row), action potential (second row) and calcium transient (third row) of myocytes at baseline (first column), intermediate stress (second column) and peak stress (third column). In the normal heart increase in heart rate is accompanied by an increase in myocardial contractile performance (up-sloping FFR). At higher heart rates more and faster "cascade" calcium is released from the SR: more calcium is available in the cytoplasm for C troponin interaction and contraction. Equally calcium reuptake is more and faster in diastole. Both action potential and calcium transient are rapidly peaking in systole at each stress step.
Figure 15 Force-frequency curve with stress echo in a subject with latent LV dysfunction without dilation. Upper panel. On the left, systolic blood pressure by cuff sphygmomanometer (SP, first row); left ventricular end-systolic volumes calculated with biplane Simpson method (ESV, second row); heart rate increase during stress (bpm, third row); in the lowest row, the force-frequency curve built off-line with the values recorded at baseline (second column), and at different steps (third, fourth, fifth column) up to peak stress (sixth column). The force-frequency relation is biphasic, with an initial up-sloping trend followed by a later down-sloping trend. Lower panel: hypothetical molecular basis (first row), action potential (second row) and calcium transient (third row) of myocytes at baseline (first column), intermediate stress (second column) and peak stress (third column). In latent failing myocytes calcium transient can be normal at baseline, but abnormal at higher heart contraction rates: compared with a normal baseline pattern (first column), at intermediate stress (second column) delayed cytosolic Ca2+ diastolic removal occurs; further dysfunction (cytolsolic Ca2+ attenuated rise with depolarization and a markedly delayed return to baseline) occurs at higher heart rates. When the heart beats at frequencies beyond the CHR, when calcium is extruded from the myocyte instead of re-entry in the SR, the O2 consumption for each unit of force developed is doubled; the combination of decreased cardiac force development and increased oxygen uptake indicates decreased efficiency of cardiac work.
Figure 16 Force-frequency curve with stress echo in a subject with dilated cardiomyopathy and depressed baseline left ventricular function (EF% = 30%). On the left: systolic blood pressure by cuff sphygmomanometer (SP, first row); left ventricular end-systolic volumes calculated with biplane Simpson method (ESV, second row); heart rate increase during stress (bpm, third row); in the lowest row, the force-frequency curve built off-line with the values recorded at baseline (second column), and at different steps (third, fourth, fifth column) up to peak stress (sixth column). An increased heart rate at peak exercise is accompanied by no changes in end-systolic volumes (abnormal flat force-frequency relation). Lower panel: molecular basis (first row), action potential (second row) and calcium transient (third row) of myocytes at baseline (first column), intermediate stress (second column) and peak stress (third column). The action potentials are markedly prolonged at baseline and during stress in patients with advanced heart failure; calcium cycling is slow at basal heart rates and even more at higher heart rates. These abnormal patterns are related to a profound derangement of the contractile machinery in the failing myocyte: fewer calcium membrane channels, fewer RNA levels encoding contractile proteins, fewer and dysfunctioning SERCA. A critical alteration of force-frequency relationship occurs, with an inversion of the normal positive to a flat or negative slope.
The critical heart rate (or optimum stimulation frequency) is the human counterpart of the treppe phenomenon; "in situ, the optimal heart rate is not only the rate that would give maximal mechanical performance of an isolated muscle twitch, but also is determined by the need for diastolic filling [6]" (Fig. 13)
Conclusion
This proposed approach allows the assessment of a theoretically robust and sophisticated index of left ventricular contractility with an absolute minimum extra-burden of data acquisition and analysis, since all the basic parameters (heart rate, blood pressure and left ventricular volumes) are routinely acquired during exercise stress echo testing [34]. The extra measurements consist of serial evaluation of ventricular volumes and linear interpolation of the force-frequency relationship. This approach is simple, not time-consuming, and highly feasible [31-33]. This index of global contractility is theoretically appealing for the identification of limited contractile reserve and latent global left ventricular dysfunction.
These are all prerequisites for a larger scale testing in the clinical subsets in which the contractility information can be more important – such as patients with latent ventricular dysfunction [37] or advanced chronic heart failure [33].
Noninvasive measurement of pressure/volume relation (the Suga index) [11,26,31] for increasing heart rates during stress in the echo lab could be the practical answer to this new clinical demand in recent years of a dramatic increase in the number of heart failure patients.
Acknowledgements
TB is funded by a PhD program on Cardiovascular Pathophysiology of the Scuola Superiore S. Anna, Pisa
==== Refs
Franciolini F Petris A Evolution of ionic Channels of biological membranes Mol Biol Evol 1989 6 503 513 2477661
Alpert NR Blanchard EM Mulieri LA Nagai R Zarain-Herberg A Periasamy M Genetic and non-genetic control of myocardial calcium Basic Res Cardiol 1989 84 55 56 2530977 10.1007/BF01907003
Reuter H Stevens CF Tsien RW Yellen G Properties of single channels in cardiac cell culture Nature 1982 279 501 504 6283360 10.1038/297501a0
Bowditch HP Über die Eigenthüm-lichkeiten der Reizbarkeit, welche die Muskelfasern des Herzens zeigen Ber Sächs Akad Wiss 1871 23 652 689
Piot C Lemaire S Albat B Seguin J Nargeot J Richard S High frequency-induced upregulation of human cardiac calcium currents Circulation 1996 93 120 128 8616918
Opie LH Braunwald E, Zipes DP, Libby P, Bonow RO Mechanisms of cardiac contraction and relaxation Heart Disease 2005 Chap 19 7 WB Saunders Company 457 489 page 480
Starling EH Linacre lecture on the law of the heart (1915) 1918 London, Longmans
Sonnenblick EH Downing SE Afterload as a primary determinant of ventricular performance Am J Physiol 1962 204 604 610 13978233
Braunwald E On the difference between the heart's output and its contractile state (Editorial) Circulation 1971 43 171 5540702
Ross J Jr Cardiac function and myocardial contractility: a perspective J Am Coll Cardiol 1983 1 52 62 6826945
Suga H Sagawa K Shoukas AA Load independence of the instantaneous pressure/volume ratio of the canine left ventricle and effects of epinephrine and heart rate on the ratio Circulation Research 1973 32 314 322 4691336
Suga H Sagawa K Instantaneous pressure/volume relationships and their ratio in the excised, supported canine left ventricle Circulation Research 1974 35 117 126 4841253
Kass DA Midei M Graves W Brinker JA Maughan WL Use of a conductance (volume) catheter and transient inferior vena cava occlusion for rapid determination of pressure-volume relationship in man Cathet Cardiovasc Diagn 1988 15 192 202 3197110
Grossman W Braunwald E Mann T McLaurin LP Green LH Contractile state of the left ventricle in man as evaluated from end-systolic pressure-volume relations Circulation 1977 56 845 852 71960
Mulieri LA Hasenfuss G Leavitt B Allen PD Alpert NR Altered myocardial force-frequency relation in human heart failure Circulation 1992 85 1743 1750 1572031
Mulieri LA Leavitt BJ Martin BJ Haeberle JR Alpert NR Myocardial force-frequency defect in mitral regurgitation heart failure is reversed by forskolin Circulation 1993 88 2700 2704 8252681
Layland J Kentish JC Positive force- and [Ca2+] i-frequency relationships in rat ventricular trabeculae at physiological frequencies Am J Physiol Heart Circ Physiol 1999 276 H9 H18
Lewartowski B Pytkowski B Cellular mechanism of the relationship between myocardial force and frequency of contractions Prog Biophys Mol Biol 1987 50 97 120 3331452 10.1016/0079-6107(87)90005-8
O'Rourke B Kass DA Tomaselli GF Kaab S Tunin R Marban E Mechanisms of altered excitation-contraction coupling in canine tachycardia-induced heart failure I Circ Res 1999 84 562 570 10082478
Ross J Miura T Kambayashi M Eising GP Ryu KH Adrenergic control of the force-frequency relation Circulation 1995 92 2327 2332 7554218
Liu CP Ting CT Lawrence W Maughan WL Chang MS Kass DA Diminished contractile response to increased heart rate in intact human left ventricular hypertrophy: systolic versus diastolic determinants Circulation 1993 88 1893 1906 8403335
Hasenfuss G Holubarsch C Hermann HP Astheimer K Pieske B Just H Influence of the force-frequency relationship on haemodynamics and left ventricular function in patients with non-failing hearts and in patients with dilated cardiomyopathy Eur Heart J 1994 15 164 170 8005115
Bhargava V Shabetai R Mathiasen RA Dalton N Hunter JJ Ross J Jr Loss of adrenergic control of the force-frequency relation in heart failure secondary to idiopathic or ischemic cardiomyopathy Am J Cardiol 1998 81 1130 1137 9605055 10.1016/S0002-9149(98)00133-7
Izawa H Yokota M Takeichi Y Inagaki M Nagata K Iwase M Sobue T Adrenergic control of the force-frequency and relaxation-frequency relations in patients with hypertrophic cardiomyopathy Circulation 1997 96 2959 2968 9386163
Inagaki M Yokota M Izawa H Ishiki R Nagata K Iwase M Yamada Y Koide M Sobue T Impaired force-frequency relations in patients with hypertensive left ventricular hypertrophy Circulation 1999 14 1822 1830 10199878
Feldman MD Alderman JD Aroesty JM Royal HD Ferguson JJ Owen RM Grossman W McKay RG Depression of systolic and diastolic myocardial reserve during atrial pacing tachycardia in patients with dilated cardiomyopathy J Clin Invest 1988 11 1661 1669 3183060
Schuller G Olshausen K Schwarz F Mehmel H Hofmann M Hermann HJ Lange D Kubler W Noninvasive assessment of myocardial contractility in asymptomatic patients with severe aortic regurgitation and normal left ventricular ejection fraction at rest Am J Cardiol 1982 50 45 52 7091005 10.1016/0002-9149(82)90007-8
Lavie CJ Lam JB Gibbons RJ Effects of exercise on left ventricular volume and output changes in severe mitral regurgitation. A radionuclide angiographic study Chest 1989 96 1086 1091 2805840
Dehmer GJ Firth BG Hillis LD Corbett JR Lewis SE Parkey RW Willerson JT Alterations in left ventricular volumes and ejection fraction at rest and during exercise in patients with aortic regurgitation Am J Cardiol 1981 48 17 27 7246439 10.1016/0002-9149(81)90567-1
Ginzton LE Lasks MM Brizenidine M Conant R Mena I Noninvasive measurement of the rest and exercise peak systolic pressure/end systolic volume ratio: a sensitive two-dimensional echocardiographic indicator of left ventricular function J Am Coll Cardiol 1984 4 509 516 6470330
Bombardini T Correia MJ Cicerone C Agricola E Ripoli A Picano E Force-frequency Relationship in the Echocardiography Laboratory: A Noninvasive Assessment of Bowditch Treppe? J Am Soc Echocardiogr 2003 16 646 655 12778025 10.1016/S0894-7317(03)00221-9
Bombardini T Agrusta M Natsvlishvili N Solimene F Pap R Coltorti F Varga A Mottola G Picano E Noninvasive assessment of left ventricular contractility by pacemaker stress echocardiography Eur J Heart Failure 2005 2 173 181 10.1016/j.ejheart.2004.04.019
Grosu A Bombardini T Senni M Duino V Gori M Picano E End-systolic Pressure/Volume relations during dobutamine stress echo: a prognostically useful noninvasive index of left ventricular contractility Eur Heart J 2005 16105848
Picano E Stress echocardiography. From pathophysiological toy to diagnostic tool Circulation 1992 85 1604 1612 1555297
Armstrong WF Pellikka PA Ryan T Crouse L Zoghbi WA Stress echocardiography: recommendations for performance and interpretations of stress echocardiography. Stress Echocardiography Task Force of the Nomenclature and Standards Committee of the American Society of Echocardiography J Am Soc Echocardiogr 1998 11 97 104 9487482
Cerqueira MD Weissman NJ Dilsizian V Jacobs AK Kaul S Laskey WK Pennell DJ Rumberger JA Ryan T Verani MS American Heart Association Writing Group on Myocardial Segmentation and Registration for Cardiac Imaging Standardized myocardial segmentation and nomenclature for tomographic imaging of the heart: a statement for healthcare professionals from the Cardiac Imaging Committee of the Council on Clinical Cardiology of the American Heart Association Circulation 2002 105 539 542 11815441 10.1161/hc0402.102975
Agricola E Bombardini T Oppizzi M Margonato A Pisani M Melisurgo G Picano E Usefulness of latent left ventricular dysfunction assessed by Bowditch Treppe to predict stress-induced pulmonary hypertension in minimally symptomatic severe mitral regurgitation secondary to mitral valve prolapse Am J Cardiol 2005 95 41 47 10.1016/j.amjcard.2004.09.048
|
16150150
|
PMC1242240
|
CC BY
|
2021-01-04 16:38:31
|
no
|
Cardiovasc Ultrasound. 2005 Sep 8; 3:27
|
utf-8
|
Cardiovasc Ultrasound
| 2,005 |
10.1186/1476-7120-3-27
|
oa_comm
|
==== Front
Clin Pract Epidemiol Ment HealthClinical Practice and Epidemiology in Mental Health : CP & EMH1745-0179BioMed Central 1745-0179-1-141614304210.1186/1745-0179-1-14ResearchThe distribution of the common mental disorders: social inequalities in Europe Fryers Tom [email protected] David [email protected] Rachel [email protected] Traolach [email protected] Psychiatry, University of Leicester, Leicester, UK2 Epidemiology and Public Health, University of Exeter, Exeter, UK3 WHO Collaborating Centre, Institute of Psychiatry, London, UK4 Psychiatry, University of Leicester, Leicester, UK5 International and Public Health, New York Medical College, Valhalla, USA2005 5 9 2005 1 14 14 26 4 2005 5 9 2005 Copyright ©2005 Fryers et al; licensee BioMed Central Ltd.2005Fryers et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Background
The social class distribution of the common mental disorders (mostly anxiety and/or depression) has been in doubt until recently. This paper reviews the evidence of associations between the prevalence of the common mental disorders in adults of working age and markers of socio-economic disadvantage.
Methods
Work is reviewed which brings together major population surveys from the last 25 years, together with work trawling for all European population studies. Data from more recent studies is examined, analysed and discussed. Because of differences in methods, instruments and analyses, little can be compared precsiely, but internal associations can be examined.
Findings
People of lower socio-economic status, however measured, are disadvantaged, and this includes higher frequencies of the conditions now called the 'common mental disorders' (mostly non-psychotic depression and anxiety, either separately or together). In European and similar developed populations, relatively high frequencies are associated with poor education, material disadvantage and unemployment.
Conclusion
The large contribution of the common mental disorders to morbidity and disability, and the social consequences in working age adults would justify substantial priority being given to addressing mental health inequalities, and deprivation in general, within national and European social and economic policy.
==== Body
Introduction
This paper seeks to explore what is known about the associations of psychiatric disorders with indicators of social disadvantage, and therefore about social risk factors in individuals and populations, and the potential for targetting with additional resources to preventive or ameliorative ends.
The recent European Mental Health Status Project [1], commissioned by the European Commission, reviewed the data available on prevalence of mental illness in European populations in relation to social, economic and service factors. In this context, a 'Survey of Surveys' identified and collated over 200 population studies, but the methods of data collection, instruments, analytical methods, and presentation of results varied so much, that very few data were strictly comparable, and very limited meta-analysis proved possible [2].
However, in respect of social disadvantage and the distribution of psychiatric disorder, we were fortunate in having a recently completed review of the world literature, together with an extended analysis of the First British National Psychiatric Household Survey undertaken for the Government of the United Kingdom [3-6]. This paper first briefly summarises these two studies, adds data from a major German study published more recently, and from a number of smaller studies identified by the Survey of Surveys, and then considers some of the major issues arising from the results.
In all western countries, most physical diseases, and severe, 'psychotic' psychiatric disorders are well known to be distributed unequally by social position [7,8]. Psychotic disorders severely affect individual patients and their families, but are relatively rare. A far more extensive burden of mental illness in the community arises from less severe but more numerous 'common mental disorders', (often called 'neurotic illness'; mostly anxiety and depression, separately or together) for which associations with social position have been unclear in the scientific literature [9]. That, then is the focus of this paper.
A systematic literature review; large-scale population studies
(Fryers, Melzer & Jenkins, 2003 [3]; and Melzer, Fryers & Jenkins, 2004 [6]
Before about 1980, population surveys had no validated, systematic instruments to identify psychiatric disorders, but several have been developed since. Measurement and classification of both mental disorder and social position carry inherent ambiguities; confidence in analysis and interpretation therefore require individual linked data in large populations. The following criteria were used to identify studies for inclusion in the review:
• community based studies (general household populations)
• populations encompassing a broad spectrum of social class variation
• samples of 3,000 or more adults of working age
• methods of identification of mental illness by validated standard instruments
• social position identified by explicit, standard markers
• a diagnostic range encompassing the common mental disorders
• individual data linking mental health measures and social indicators; i.e. not area studies
• relevance to UK policy development; studies from established market economies
• fieldwork undertaken since 1980
• published output on the key areas of interest
Computer-accessed research-literature data-bases were exhaustively searched, but they are often ineffective for broad or ambiguous categories, and proved so in this case. Moreover, they do not include books, or reports from research institutes or government departments, a necessary source of detailed information on large-scale population surveys. Most information came from cross references and direct enquiry of researchers and units, to create a unique database of almost 1,000 references.
Nine large-scale studies were identified which fulfilled the criteria (Table 1). For these, the published work was examined independently by two researchers, with regard to the validity and reliability of their methods, and their findings relating to the prevalence of the common mental disorders and differentials in social position. Of five European studies, four were from the UK and one from The Netherlands. One each was from Canada and Australia; two were from the USA. Since this work was finished, data have become available from the German National Health Survey of 1999 which appears to fulfil the inclusion criteria. This is described later.
Table 1 Surveys included in the Inequalities Review [6]
Annual Health Surveys for England (HSE), annually from 1993
National Psychiatric Morbidity Survey of Great Britain (household sample), 1993
Health and Life-style Survey (HLS), 1984–85 and follow-up, 1991–92
British Household Panel Survey (BHPS),1991–92
Netherlands Mental Health Survey and Incidence Study (NEMESIS), 1996
Edmonton Survey of Psychiatric Disorders (Canada), 1983–86
Australian National Survey, 1997
USA Epidemiologic Catchment Area Program (ECA), 1980–83
USA National Co-morbidity Study (NCS), 1990–92
Although all 9 studies used recognised instruments with at least some published validation, several different instruments were in use, and even the same instruments were applied in different ways. Categories of disorder, indicators of social position, and presentation of statistics were so diverse that no numerical meta-analysis was possible. Response rates, not always high, (54% – 80%) also prejudiced interpretation.
Poverty, education, housing, occupation, employment, social status and social engagement are relatively tangible measures, for which 'Social Class' or 'Socio-Economic Status' are merely proxies, but these markers of social disadvantage are not independent of each other. Other factors are known to be important – childhood experience, physical illness, life events, working situations, and social networks – but they were barely acknowledged by these large-scale cross-sectional studies. If we wish to have evidence of the direction of causation for associations discovered, we need longitudinal studies. The evidence available from the UK birth cohort studies is briefly summarised below and is available in more detail in the source documents. [6]
Nevertheless, some comparison of the cross-sectional studies was possible. In each study, the categories which most nearly approximated to the 'common mental disorders' were examined; usually this meant 'all affective disorders', 'all depressive disorders', 'dysthymia', and 'all anxiety disorders'. Similarly, in most studies, three indicators of social disadvantage could be compared: education, employment, and material circumstances, as well as occupational social status. Although the studies used different taxonomies, differentials within the taxonomy could be recorded for each one.
For education, the highest and lowest groups were compared, whether measured by years of education or qualifications achieved. For employment the 'unemployed and seeking work' were compared with either 'all others of working age', or 'all employed'. Material circumstances were measured in many ways, but the lowest and highest in each hierarchy could be compared. The associations detected were subjected to statistical tests of significance, and odds ratios for each relationship quoted wherever possible.
Taking higher prevalence of disorder in less privileged groups as a 'positive' association, of the nine population-based studies with adequate measures of mental health and indicators of social disadvantage, eight provided evidence of an association between less privileged social position and higher prevalence of the common mental disorders, on at least one of the available indicators (Table 2). The one study showing no clear relationships had the lowest response rate (54%), which may have limited its capacity to demonstrate associations. Less education was 'positive' in four out of five studies. Unemployment showed positive associations in six out of seven studies, though in one study the association was positive only for men. Low income, wealth, assets, or other markers of material standard of living were positive in all six studies. Less privileged occupational social class was positive in three studies out of six. Perhaps most importantly, no study showed a contrary trend with any indicator.
Table 2 Number of included studies reporting associations with higher rates of the common mental disorders, by indicators of less privileged social position [3;6]
Less education Unemployment Lower income or material circumstances Low social status
Number of studies reporting associations Total reporting 5 7 6 6
Positive Men & women separately 2 3* 2 2
association Men & women combined (separate data not given) 2 3 4 1
Total positive 4 6 6 3
No clear association 1 1 0 3
Inverse association 0 0 0 0
Note: *one study, positive only for men; women equivocal.
These statistically significant positive associations do not reveal the degree of difference; compared to the most privileged groups, the most deprived groups seldom had more than a doubling in prevalence, that is odds ratios were almost always less than 2.
This simple overview suggests some robustness of findings despite the serious methodological limitations in reviewing such diverse studies. Education, employment and material circumstances provided better indicators than occupational social class, but there is remarkable consistency in the broad evidence from these nine large-scale population-based studies; the common mental disorders are significantly more frequent in socially disadvantaged populations.
Limiting & disabling neurotic illness and markers of social disadvantage
(Melzer, Fryers T, Jenkins R, Brugha T, & McWilliams B, 2003 [4]; and Melzer, Fryers & Jenkins, 2004 [6]
Data from the 1993 National Psychiatric Survey of Great Britain [5] (supplied by the Data Archive, University of Essex) were subjected to detailed analysis:
• to clarify if markers of social position were independent of each other,
• to incorporate measures of disability into case identification,
• to indicate priority groups,
• to estimate effect sizes.
The 1993 Survey interviewed a representative household sample of over 10,000 people aged 16 to 64 using the Clinical Interview Schedule (CIS Revised) to record 'neurotic' symptoms or illness during the previous week. 15.5% had 'neurotic illness' which would justify clinical monitoring or active treatment in primary care; 63% of these had symptoms with an average duration of six months or more. Those reporting that "their mental symptoms stopped them doing things" were defined as having 'limiting neurotic disorder'; those reporting also that they "had difficulty in doing at least one activity of daily living" were defined as having 'disabling neurotic disorder'. In all groups most people had anxiety and/or depressive disorders.
Of the whole survey population, 8.3% had 'limiting neurotic disorder' and 3.4% had 'disabling neurotic disorder'. Consistent with the WHO Global Burden of Disease estimates [10], neurotic illness made a large contribution to all disability reported in the British survey. For example, of those with difficulties in three or more activities of daily living, 38% had a 'limiting neurotic disorder'. Women had more neurotic illness than men, but risks were equal for 'disabling neurotic disorder'.
Higher prevalence rates of the common mental disorders were associated with every marker of less privileged social position incorporated into the interview schedules, but multivariate analysis adjusting for gender, age and competing markers, left three as 'surviving independent markers':
• being unemployed or economically inactive
• poorer material circumstances (housing tenure and lack of car ownership)
• less education (having left full-time schooling before age 16)
Occupational social class was not a significant marker after adjustment.
'Disabling neurotic disorder' was associated with being economically inactive or unemployed (OR >2). In other analyses, 'disabling neurotic disorder' was associated with having two or more physical illnesses (OR >6) and having two or more adverse life events (OR >3). Using other data from the survey, lone parents, those with physical diseases involving two or more disease systems, and those who were unemployed, together made up 20% of the population, but contributed 51% of those with 'disabling neurotic disorder'.
Cross sectional data cannot clarify the direction of causation, though wider evidence provides some support for deprived circumstances causing the disorders [6]. Clarification needs longitudinal studies, which should include other potential risks, such as carer status, known to be associated with high rates of depression, and history of abuse. The lone parent group should receive special attention because of effects on the children.
The European Survey of Surveys, 2002
The 'Survey of Surveys' of the European Mental Health Status Project identified more than 200 population surveys across Europe, but few provided comparable data because of differences in methods, instruments, analysis and presentation, or because they were small-scale community studies. A very restricted meta-analysis proved possible with surveys using a GHQ or CIDI instrument [2], including four of the five European studies in the review summarised above. These five are listed in Table 3 and briefly described below. Added to them is the German Health Survey of 1999 which appears to fulfil the same inclusion criteria as the studies reviewed, but has not yet published many results [11,12]
Table 3 Characteristics of European studies included in Maudsley review [6] with the German Health Survey, 1999 [11]. (Adapted from [6])
European Surveys Year Type of study Population sampled Size of sample (achieved) Response rate Mental health instrument
1 Annual Health Surveys for England 1993, repeated annually population survey All adults in England, children from 1995 16,569 (1993) 76% for full interview, 66% for nurse tests (1993) GHQ-12, cut-off 4+
2 National Psychiatric Morbidity Survey of Great Britain (household sample) 1993 population survey All adults in England, Wales and Scotland (excluding Highland and Islands) 10,108 80% Clinical Interview Schedule (CIS revised)
3a Health and Life-style Survey 1984–85 population survey Adults 18+, England, Wales, Scotland 9,003 73% for interview, 54% for self-completed questionnaire GHQ-30 (+ a malaise measure)
3b Health and Life-style Survey – follow-up 1991–92 follow-up of 84/85 respondents Adults 18+, England, Wales, Scotland 5,352 59% of those interviewed in 1984/5 were re-interviewed GHQ-30 (+ a malaise measure)
4 British Household Panel Survey 1991–92 population survey, with follow-up after one year Adults aged 16+, households in Great Britain, south of Caledonian Canal 10,264 74% of 7,488 households GHQ-12, cut-off 3+
5 Netherlands Mental Health Survey and Incidence Study (NEMESIS) 1996 population survey with follow-up at one and three years Adults 18–64 resident in The Netherlands 7,147 64% Composite International Diagnostic Interview (CIDI); GHQ-12
6 National German Health Survey (GHS) 1999 population survey Adults 18–65 resident in Germany 4181 ? Composite International Diagnostic Interview (CIDI – Munich version)
Health Survey for England, annually from 1993
Annually since 1991, adults aged 16 and over in England have been sample surveyed using structured interviews and clinical tests. Since 1993 most years have included the General Health Questionnaire (GHQ -12) [13], two questions about stress, and questions on perceived social support, occupation, income, material standard of living, and employment. Completed interviews have been approximately 16,000, a response rate of 74% of sampled households, and 92% of adults within these households [14,15].
A 'positive' score (4 or more on the GHQ -12), was considered to indicate a psychiatric disorder diagnosable by a clinician. Year by year there has been little variation in results; in 1998 for example, 13% of men and 18% of women were recorded as 'positive', correlated highly with perceived lack of social support, recent acute sickness, and long-standing illness. There were weak associations with occupational social class, but significant and progressive associations with low 'equivalised household income', especially among men (9% in the highest income quintile to 20% in the lowest income quintile, in 1998).
The First UK National Household Psychiatric Survey, 1993
12,000 adults aged 16–64 were selected from a representative sample of 15,000 households in Great Britain, and over 10,000 interviews achieved, a response rate of 80% [5]. Trained lay interviewers used the Clinical Interview Schedule – Revised (CIS-R); scores were converted into ICD-10 diagnoses; 12 or more was taken to indicate 'likely to have a neurotic disorder'. A separate alcohol and drug schedule was used, and people with 'possible psychosis' were identified for a SCAN interview with a clinician. Occupational social class, income, material standard of living, housing status, education and employment were recorded [16].
An occupational social class gradient for women largely disappeared with adjustment for more precise indicators of social disadvantage. For men, the highest social class had about half the positive scores of other classes, unchanged by adjustment [17]. Unemployment was associated with higher positive scores, and was the factor most strongly associated with symptom prevalence in men and women, while low material standard of living and poor education had the highest rates of probable neurosis. However, the association with education disappeared when adjusted for other socio-demographic variables.
The Health and Life-style Survey (HLS), 1984–85 and 1991–92
9,003 residents of Great Britain aged 18 years or over, were interviewed, and 82.4% of these examined by a nurse. 6,572 GHQ-30 questionnaires were completed, a score of 5 or more being considered positive; scores were continuously varied for both men and women [18]. Data on occupation of head of household, income, housing tenure and education were recorded. Though not designed as a cohort study, after 7 years 5,352 people (59% of the original sample), were traced and re-interviewed [18]. GHQ scores related to occupational social class showed no consistent pattern. Unemployment was clearly related to high scores in 1984/85, but not in 1991/92.
Of special interest was the finding that positive scores in 1984/85 were associated with significantly increased all-cause mortality after seven years, even after adjusting for age, sex, social class, smoking behaviour, and limiting long-standing illness, and after removing 'un-natural' deaths which might have been specifically related to psychiatric disorder. There was an approximately linear relationship between the risk of dying and the number of symptoms on the GHQ-30, especially for men [19].
The British Household Panel Survey (BHPS), 1991–92
Of 7,488 British households selected, 5,511 were contacted, involving 10,264 individuals aged 16 and over, of which 9,064, 88% of subjects, completed the GHQ-12. They were followed up a year later. A score of 3 or more was considered 'positive'. Occupation of subject, parents, and head of household were recorded, together with employment data [20]. An indicator of material standard of living combined income, and elements of housing and possessions.
The results gave a gradient with occupational social class (subject or head of household, but not parents), which disappeared in men up to age 55 after adjusting for material standard of living, but was still true for women of all ages [21]. Material standard of living was strongly associated with high frequency of GHQ positives (3+), but possibly only maintainance, not onset of common mental disorders. 'Subjective financial strain', (one question with three possible answers), was correlated with onset of symptoms [22]. Physical illness was associated with GHQ-12 positives. Using also one-year follow-up data, unemployment was also associated with maintenance but not onset of symptoms, which diminished in those gaining employment in the year, and increased in those losing employment in the year, unless for looking after the family or retirement. Scores also decreased during the year for those marrying, and increased for those divorcing or separating.
The Netherlands Mental Health Survey & Incidence Study (NEMESIS), 1996
7,147 individuals aged 18–64 (64.2%) were interviewed from 11,140 eligible households using the CIDI (and SCID if psychosis was indicated). 43.6% of those refusing the CIDI completed the GHQ-12. Refusers proved to have similar mental health profiles to responders. Family income, average net income per person, employment status, and years of education were recorded [23]. 5,618 adults were interviewed after one year, 79.4% of the cohort [24].
The three commonest disorders, anxiety, depression and alcohol were often present together. Men had more alcohol and other drug disorders; women had more anxiety and depressive disorders. Very poor education, low income, and 'non-employment' were associated with both mood and anxiety disorders [23]. The one-year follow-up showed unemployment associated with the common mental disorders [25].
The National German Health Survey, 1999
The German Health Survey of 1999 used the CIDI and DSM IV to identify 'cases' in a realised sample of 4,181. Data for 12-month prevalence of 'any mood disorder' and 'any anxiety disorder', relating fairly closely to the 'common mental disorders' are available [11,12]. Prevalence rates are very similar to similar surveys elsewhere. 12-month prevalence was analysed for level of school achievement, employment, and an index of 'social class' combining education, income and job status. In each case high prevalences were found in the more disadvantaged groups (Table 4)
Table 4 German National Health Survey 1999, Mental Health Supplement [12].
12 month prevalence:
Any mood disorder Any anxiety disorder
%w OR CI %w OR CI
Education
Hauptschule (2nd y school) 13.2 1.0 15.4 1.0
Mittlere Reife (= 'GCSE') 12.2 0.9 0.7–1.1 14.8 0.9 0.7–1.1
Abitur (= 'A levels') 9.5 0.7 0.5–0.9 11.3 0.7 0.5–0.9
Employment
FT employed 1.0 1.0
Unemployed 20.0 2.3 1.6–3.2 23.2 2.2 1.6–3.0
Social Class (an index combining education, income, and current job status)
Low 16.4 1.0 18.6 1.0
Medium 12.0 0.7 0.6–0.9 14.4 0.8 0.6–0.9
High 8.8 0.5 0.4–0.7 11.3 0.6 0.4–0.8
People with Abitur level education had less illness, just significant at the 0.05 level. The unemployed had significantly more illness than the full-time employed. People in medium and high class groups had significantly less illness than those in the low class group (this indicator incorporates education and income with occupation). These results harmonise closely with the overall results of the Inequalities Review described above.
Summary of the six major European studies (Table 5)
Table 5 'Positive' associations with less privileged social status and the common mental disorders in European surveys (adapted from [6])
European Surveys Education Employment status Income and material standard of living Occupational social status
1 HSE 1993+ - - positive association for income progressive for both men and women 1998 No clear distribution for either men or women
2 UK Psych Survey 1993 Positive for no qualifications or least years of education for both men and women Positive for unemployed in both men and women Positive for income, housing type/tenure, and car ownership Positive for women (SC I+II compared to SC IV+V); positive for men (SC I compared to all other classes)
3a HLS 1984/85 - Positive for unemployment in men in both age groups - No clear social class distribution
3b HLS 1991/92 - No clear relationship - No clear social class distribution
4 BHPS 1991/92 - Unemployment associated with maintenance, not onset in 1-year follow-up; symptoms reduced on gaining employment (men and women combined) Positive for low income, 'poverty index', and index of material standard of living (men and women combined) Positive association for both men and women
5 NEMESIS 1996 Positive for least education (men and women combined) Positive for unemployment (men and women combined) Positive for income (men and women combined) -
6 GHS 1999 Just positive for lowest qualifications (men and women combined) Positive for unemployment (men and women combined) - Positive for SC index combining education, income & job status (men and women combined)
Other surveys
Although the Survey of Surveys found few directly comparable studies, some results can be compared in their internal relationships, in the same manner as the literature review described above, and three studies provide data on markers of social disadvantage. These studies would not have fulfilled the strict inclusion criteria for that review, neither have they been subject to the validating processes undertaken in that review. The results should, therefore, be treated with caution.
The Northern Ireland Survey of 1997 [26] used the GHQ-12 with a cut-off point of 3 or more indicating a 'possible case'. Using the UK Occupational Social Classification, lower groups (classes III manual – V) showed higher prevalences than higher groups (I – III non-manual) for both men and women except in the youngest age group. The largest difference was in women aged 45–64. Over age 65, social class differences were very small. Using more detailed Socio-Economic Groups (SEGs), prevalences were progressively higher with lower SEG. Using an education marker of 'some formal qualification' compared with 'no formal qualification', the former had markedly lower prevalence in women, especially young women, but differences were very small in men. People who owned their own house had lower prevalence than those who rented, and those who had access to a car had less than those who did not. Those on lower incomes had higher prevalence than those on higher incomes: twice the rate at age 16–44; twice the rate at age 45–64 in men, three times the rate in women.
These results, though from a smaller survey, are similar to the general results from the large-scale British surveys.
A survey in Belgium in 1997 [27] used the GHQ-12 with a cut-off point of 2 or more. There was no clear detailed pattern in relation to educational level, but 'primary school only' had higher results than all those 'more than primary' combined. A separately recorded 'depression score' (for the previous 12 months) did show markedly less positive scores with better education.
A study of two regions in France, Basse Normandy and Ile de France [28], found that being unemployed was associated with significantly more depression than other employment groups, but education was a mixed and equivocal picture.
If we add to Table 2 the results of these studies and the available findings of the German Health Survey of 1999, (acknowledging the provisional nature of some of the data) we get an expanded Table 6:
Table 6 Expanded number of studies reporting associations with higher rates of the common mental disorders, by indicators of less privileged social position.
Less education Unemployment Lower income or material circumstances Low social status
Number of studies reporting associations Total reporting 9 9 7 8***
Positive Men & women separately 5** 5* 3 4
association Men & women combined (separate data not given) 2 3 4 1
Total positive 7** 8* 7 5
No clear association 2 1 0 3
Inverse association 0 0 0 0
Note: *one study positive only for men; women equivocal; **one study positive only for women; equivocal for men; *** the German 'social class' incorporated education and income as well as occupation.
This adds a little extra weight to the major review without altering the general picture. It is still most notable that no study has given an inverse association between the three markers of social disadvantage and the prevalence of the common mental disorders.
Initial results from ESEMeD
The European Study of the Epidemiology of Mental Disorders (ESEMeD/MHEDEA 2000) was a comparison of cross-sectional samples of the non-institutionalised population aged 18 years or more, in six countries: Belgium, France, Germany, Italy, the Netherlands and Spain [29]. Different private companies were contracted to undertake the survey in each country. Trained interviewers used a computer-assisted personal interview (CAPI) including the most recent version of the Composite International Diagnostic Interview (CIDI 2000) to assess the presence of mental disorders in face-to face interviews in people's own homes. The total combined sample chosen was 38,015 people, of which 19,706 were interviewed. Response rates varied from 42.1% in France to 71.9% in Spain, giving an overall response rate of 55% [30,31].
We have examined data made available from the ESEMeD study. These are in the form of distributions of odds ratios (ORs) for associations in individual subjects between various social indicators and psychiatric disorder in the 12 months previous to interview. Unemployment data (having a job against not having a job) are the most relevant to inequality analyses; living alone (against not living alone) could possibly have a bearing; receiving Government Assistance (against receiving none) could be very relevant, but the data are not considered reliable by the researchers. Interpretations of data are generally prejudiced by low response rates.
As regards unemployment, all ORs were positive for 'any psychiatric disorder in the previous 12-months', but two of them were not significant; the highest OR (2.49) was Germany. For 'any mood disorder in the previous 12 months', the OR for The Netherlands was negative but not significant; all others were positive and significant at the 5% level; the highest being Germany (OR 5.42). For 'any anxiety disorder in the previous 12 months', all ORs were positive but only two were significant – Germany (OR 1.72) and Italy (1.70). For 'any alcohol disorder in the previous 12 months'. five countries gave positive ORs but only Germany was significant (OR 4.47).
In general, these unemployment results indicate the expected association in individual subjects with psychiatric disorder, but interpretation of the difference between results for different country samples will have to await further analysis. In particular, it will be interesting to see if any light can be thrown upon the tendency of the figures for Germany to be consistently much higher than other countries. The explanation may lie in differences in sampling and interviewing, as each country organised these through different agencies.
The results for living alone offer no particular interest. Of 4 analyses (any disorder; mood disorder; anxiety disorder; alcohol disorder, in the 12 months prior to interview) including all 6 countries, only one result was statistically significant at the 5% level – Germany (OR 1.71 for any 12-month disorder)
For receiving Government Assistance, most analyses were not signifcant. For 'any 12-month disorder', only Germany (OR 1.36) and Italy (OR 1.37) gave significant results. For 'any 12-month mood disorder', only Italy (OR 1.26) and The Netherlands (OR 1.15) gave significant results. For 'any 12-month anxiety disorder', no result was significant. Especially in the light of the doubts of the researchers about the reliability of these data, no interpretation can be offered.
Estimates of size of effect and of relative risk
While there is clearly broad consistency in the findings, these analyses tell us nothing of the size of effect. Examination of the odds ratios available from the studies under consideration shows that only rarely did they exceed 2, which indicates a doubling of the risk in less privileged groups for the common mental disorders, compared with more privileged groups [6]. ORs and 95% confidence intervals can be summarised for the different markers.
Education:
• In the 1993 British Psychiatric Morbidity Survey (sample aged 16–64), men with no educational qualifications had an OR of 1.29 (1.03–1.62), and women had an OR of 1.26 (1.06–1.49) for recent neurotic disorder, compared to those with A level qualifications (university entrance);
• In the 1996 Netherlands national survey, (sample aged 18–64), people with 0–11 years of education had an OR of 1.55 (1.22–1.98) for mood disorders, compared to people with 16 or more years of education.
Because data are so few from European studies, it is worth adding:
• In the 1990–92 USA NCS (sample aged 15–54), people with 0–11 years of education had an OR of 1.79 (1.31–2.43) for 'any affective disorder', and an OR of 2.82 (2.26–3.51) for 'any anxiety disorder' in the previous 12 months, compared to those with 16 or more years of education.
• In the 1997 Australian National Survey, people who did not complete secondary school had an OR of 1.53 for affective disorders compared to people with post-school qualifications. The sample included all aged 18 and over, so these results will be confounded by age.
Employment.
• In the 1991–92 UK BHPS, the unemployed had an OR of 1,54 (1.13–2.10) for the maintainance of GHQ-12 'case-ness' (scores of 3 or more at both base-line and one-year follow-up), compared to employed people. The sample was all aged 16 and over, so the results will be confounded by age.
• In the 1993 British Psychiatric Morbidity Survey (sample aged 16–64), people who were unemployed had an OR of 2.59 (2.17–3.10) for recent neurotic disorder, compared to people in full-time employment.
• In the 1996 Netherlands national survey, (sample aged 18–64), ORs of 4.3 (3.24–5.72) for mood disorders, and 2.23 (1.70–2.91) for anxiety disorders were reported for a mixed group of 'disabled and unemployed', compared with people who were employed. This strange grouping prejudices interpretation; disabled people may well have higher rates of mood and anxiety disorders unrelated to employment per se.
Also worth noting:
• In the 1990–92 USA NCS (sample aged 15–54), ORs of 2.2 (1.6–2.9) for 'any affective disorder', and 2.1 (1.6–2.8) for 'any anxiety disorder' (both life-time prevalence), were reported for people not working and neither 'home-makers' or 'students', compared with people who were working.
• In the 1997 Australian National Survey (sample 18 and over), the long-term unemployed (12 months or more) had ORs of 2.4 (1.4–4.3) for 'any affective disorder', and 2.8 (1.6–5.0) for 'any anxiety disorder' compared to employed people. Analysis excluded those 'not in the labour force'.
Income or material standard of living.
• In the 1991–92 UK BHPS (sample16 and over), comparing household income in quintiles, the middle three fifths had more GHQ-12 'positives' (scores of 3 or more), OR 1.16 (1.0–1.34), and the lowest fifth had far more GHQ-12 'positives', OR 1.45 (1.21–1.74) than the highest fifth.
• In the 1993 British Psychiatric Morbidity Survey (sample aged 16–64), people renting their homes had an OR of 2.17 (1.79–2.64) for men, and 1.71 (1.48–1.98) for women, for recent neurotic disorder, compared to people who owned their own homes.
• In the 1998 Health Survey of England (sample aged 16 and over), ORs for the lowest quintile of equivalised household income was 1.53 (1.12–2.09) for men and 1.11 (0.87–1.41 – not significant) for women, for GHQ 'caseness' (scores of 4 or more), compared to the highest quintile.
• In the 1996 Netherlands national survey (sample aged 18–64), ORs of 1.56 (1.20–2.03) for mood disorders, and 1.77 (1.43–2.21) for anxiety disorders, were reported for the lowest income quartile compared to the highest income quartile.
In addition, we might note that:
• In the 1990–92 USA NCS (sample aged 15–54), ORs of 1.73 (1.29–2.32) for 'any affective disorder', and 2.12 (1.63–2.77) for 'any anxiety disorder' (both 12-month prevalence), were reported for people earning $0–$19000 a year compared to people earning $70,000 or more.
Discussion
In general we might say that the odds against disadvantaged groups are undoubted, but modest, the increased risk being generally between one and a half and two times that for the least disadvantaged groups. However, it should be remembered that these are all rather crude measures of ambiguous phenomena; there is nothing subtle or precise in population surveys of psychiatric illness, although the situation is much better than twenty years ago and is improving still.
We must make do with what we have, and recognise that the conclusions of the recent work described in this paper, whilst undramatic, represent a real advance on previous knowledge. There can be no doubt now that disadvantaged groups in European populations experience more anxiety and depression, measurable on standard instruments and representing significant suffering for individuals, and serious loss of production and social function, with important consequences for children, communities and work-places. We can begin to define populations at risk, though this will still be rather generalised.
The scientific literature from major population studies currently permits very little detailed comparitive analysis of risk factors other than the three presented above, education, employment, and income/material standard of living, which can be measured in fairly similar ways in all western societies. Social Class or Socio-Economic Group can only be a proxy for these, and, no doubt, other more precise and tangible markers of social position and social experience. We now need focussed investigations into causative factors and possible means of prevention, and evaluations of means of relieving suffering and improving function.
The evidence drawn from cross-sectional studies, however large, cannot determine the direction of causation, which is, no doubt, complex, and not all one way. Cohort evidence has the potential to help here, but the little available evidence is fragmentary and supports only tentative conclusions [5]. In general it appears that higher rates of disorder in adulthood are associated with multiple disadvantage in childhood, including parental divorce and economic hardship, and parental psychiatric illness. These causative factors are generally also associated with social disadvantage.
The excess of the common mental disorders in disadvantaged people is well enough established to justify health policy initiatives to ensure that access to effective diagnosis and treatment is improved, especially at the primary health care level, and especially in communities with high levels of social disadvantage. A wide range of treatment strategies should be available, including, where appropriate, drugs, counselling and other therapies, and social interventions to improve disadvantageous situations. Concurrent physical illness must also be addressed in the total treatment package. Interventions need to be properly evaluated.
Research relating mental ill health to social disadvantage has already produced a wealth of useful evidence, but general conclusions useful to policy makers are to some extent prejudiced by incompatible methods, measures and analyses. Standardising and validating a small range of instruments and indicators, and closer collaboration between researchers, especially across the EU, would both facilitate and economise on future studies. But, in the reality of many studies already performed, there is also need for better methods of synthesising disparate findings of this kind [2].
Large scale longitudinal studies are especially needed if the complexities of cause are ever to be teased out. There are also continuing opportunities for exploiting already existing large scale data bases. Little in the literature considered above addresses issues of cultures and sub-cultures, and their impact on the mental health and mental health risks of individuals. Communal, societal influences on experience, behaviour and health, as well as individual actions and attitudes need to inform both research and political action.
Conclusion
People of lower socio-economic status, however measured, are disadvantaged, and this includes higher frequencies of the conditions now called the 'common mental disorders' (mostly non-psychotic depression and anxiety, either separately or together). In European and similar developed populations, relatively high frequencies are associated with poor education, material disadvantage and unemployment. Their large contribution to morbidity and disability, and the social consequences in working age adults would justify substantial priority being given to addressing mental health inequalities, and deprivation in general, within national and European social and economic policy.
But disadvantaged people also tend to live in communities and cultures that are disadvantaged by noxious environments, poor human services, high levels of smoking, drinking, drug taking, and violence. These are almost certainly causally associated with high levels of psychiatric morbidity also found, possibly mediated or enhanced by poor education, low incomes and low status work. These factors may affect duration as well as onset and thus increase prevalence in populations.
However, there are well known policy implications relating to social exclusion and deleterious social environments; it does not need population surveys to show that serious poverty, deprivation, environmental degradation and social stress should be high on the political agenda; it is a matter of equity, justice and human rights.
==== Refs
EU Commission The State of Mental Health in the European Union. Brussels 2004
Fryers T Brugha T Morgan Z Smith J Hill T Carta M Kovess V Lehtinen V Prevalence of psychiatric disorder in Europe; the potential and reality of meta-analysis Social Psychiatry & Psychiatric Epidemiology 2005 39 899 905 10.1007/s00127-004-0875-9
Fryers T Melzer D Jenkins R Social inequalities and the common mental disorders: a systematic review of the evidence Social Psychiatry & Psychiatric Epidemiology 2003 38 229 37 10.1007/s00127-003-0627-2
Melzer D Fryers T Jenkins R Brugha T McWilliams B 'Social position and the common mental disorders with disability: estimates from the National Psychiatric Survey of Great Britain Social Psychiatry & Psychiatric Epidemiology 2003 38 238 43 10.1007/s00127-003-0628-1
Meltzer H Gill B Petticrew M Hinds K The Prevalence of Psychiatric Morbidity Among Adults Living in Private Households. ONS Report 1, London 1995
Melzer D Fryers T Jenkins R Social Inequalities and the Distribution of the Common Mental Disorders. Maudsley Monograph 44 2004 Psychology Press, Hove & New York
Acheson D Independent Inquiry into Inequalities in Health 1998 The Stationery Office, London
Marmot M Wilkinson R Social Determinants of Health 1999 Oxford University Press, Oxford
Dohrenwend BP Socio-economic status (SES) and psychiatric disorders: are the issues still compelling? Social Psychiatry and Psychiatric Epidemiology 1990 25 41 7
Murray C Lopez AD The Global Burden of Disease 1996 Harvard School of Public Health, Harvard
Wittchen H-U Carter RM Pfister H Montgomery SA Kessler RC Disabilities and quality of life in pure and co-morbid generalized anxiety disorder and major depression in a national survey International Clinical Psycho-pharmacology 2000 15 319 328
Jacobi F Wittchen H-U Holting C Hofler M Pfister H Muller N Lieb R Prevalence, comorbidity and correlates of Mental disorders in the general population: results from the German Health Interview and Examination Survey (GHS) Psychological Medicine 2004 34 597 611 10.1017/S0033291703001399 15099415
Goldberg DP Gater R Sartorius N Ustun TB Piccinelli M Gureje O The validity of two versions of the GHQ in the WHO study of mental illness in general health care Psychological Medicine 1997 27 191 7 10.1017/S0033291796004242 9122299
Erens B Primatesta P Health Survey for England: Cardiovascular Disease 1998 1999 The Stationery Office, London
Bennett N Dodd T Flatley J Freeth S Bolling K Health Survey for England 1993 1995 HMSO, London
Jenkins R Bebbington PE Brugha T Farrell M Gill B Lewis G The National Psychiatric Morbidity Surveys of Great Britain – strategy and methods Psychological Medicine 1997 27 765 74 10.1017/S003329179700531X 9234455
Lewis G Bebbington PE Brugha T Farrell M Gill B Jenkins R Socio-economic status, standard of living and neurotic disorder Lancet 1998 352 605 9 10.1016/S0140-6736(98)04494-8 9746021
Huppert FA Whittington JE Longitudinal changes in mental state and personality measures The Health and Life-style Survey: Seven Years On., Dartmouth, Aldershot 1993 133 54
Huppert FA Whittington JE Symptoms of psychological distress predict 7-year mortality Psychological Medicine 1995 25 1073 86 8588004
Buck N Gershuny J Rose D Scott J Changing Households: The British Household Panel Survey, 1990–1992 1994 ESRC Research Centre on Micro-Social Change, Colchester
Weich S Lewis G Material standard of living, social class, and the prevalence of the common mental disorders in Great Britain Journal of Epidemiology and Community Health 1998 52 8 14
Weich S Lewis G Poverty, unemployment and common mental disorders: population based cohort study British Medical Journal 1998 317 115 19 9657786
Bijl RV Ravelli A van Zessen G Prevalence of psychiatric disorder in the general population: results of The Netherlands Mental Health Survey and Incidence Study (NEMESIS) Social Psychiatry and Psychiatric Epidemiology 1998 33 587 95 10.1007/s001270050098
de Graaf R Bijl RV Smit F Ravelli A Vollebergh WA Psychiatric and sociodemographic predictors of attrition in a longitudinal study: The Netherlands Mental Health Survey and Incidence Study (NEMESIS) American Journal of Epidemiology 2000 152 1039 47 10.1093/aje/152.11.1039 11117613
Spijker J Bijl RV de Graaf R Nolen WA Determinants of poor 1-year outcome of DSM-III-R major depression in the general population: results of The Netherlands Mental Health Survey and Incidence Study (NEMESIS) Acta Psychiatrica Scandinavic 2001 103 122 30 10.1034/j.1600-0447.2001.103002122.x
NISRA The 1997 Northern Ireland Health and Social Well-being Survey; Summary Report, 1999 Northern Ireland Statistics and Research Agency, DHSS (NI), Belfast
Kittel F Ribourdouille M Dramaix M Mental health data analysis from the National Health Survey, Belgium 1997 Archives of Public Health 2001 50 347 357
Kovess-Masfety V Lecoutour X Delavelle S Urbanicity and mood disorders; comparisons between French Regions Social Psychiatry & Psychiatric Epidemiology 2004 39 11
Alonso J Ferrer B Romera B Vilagut G Angermeyer MC Bernert S Brugha TS Taub N McColgan Z The European study of the Epidemiology of mental disorders (ESEMeD/MHEDEA 2000) project: rationale and methods International Journal of Methods in Psychiatric Research 2003 11 55 67
The ESEMeD/MHEDEA 2000 investigators Sample and methods of the European Study of the Epidemiology of Mental Disorders (ESEMeD) Supplement to Acta Psychiatrica 2004
The ESEMeD/MHEDEA 2000 investigators Lifetime and 12-month prevalence of mental disorders in Six European countries: results from The European Study of the Epidemiology of Mental Disorders (ESEMeD) Supplement to Acta Psychiatrica 2004
|
16143042
|
PMC1242241
|
CC BY
|
2021-01-04 18:01:01
|
no
|
Clin Pract Epidemiol Ment Health. 2005 Sep 5; 1:14
|
utf-8
|
Clin Pract Epidemiol Ment Health
| 2,005 |
10.1186/1745-0179-1-14
|
oa_comm
|
==== Front
CytojournalCytoJournal1742-6413BioMed Central London 1742-6413-2-141615329610.1186/1742-6413-2-14ResearchIs an increase in CD4/CD8 T-cell ratio in lymph node fine needle aspiration helpful for diagnosing Hodgkin lymphoma? A study of 85 lymph node FNAs with increased CD4/CD8 ratio Hernandez Osvaldo [email protected] Thaira [email protected] Sherif [email protected] New York University Medical Center, Department of Pathology, New York, New York, USA2005 9 9 2005 2 14 14 14 2 2005 9 9 2005 Copyright © 2005 Hernandez et al; licensee BioMed Central Ltd.2005Hernandez et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
An elevated CD4/CD8 T-cell ratio on flow cytometry (FCM) analysis has been reported in the literature to be associated with Hodgkin lymphoma (HL). The purpose of our study was to determine the diagnostic significance of an elevated CD4/CD8 ratio in lymph node fine needle aspiration (FNA) specimens.
Design
Between 1996 and 2002, out of 837 lymph node FNAs submitted for flow cytometry analysis, 85 cases showed an elevated CD4/CD8 ratio, defined as greater than or equal to 4, without definitive evidence of a lymphoproliferative disorder. The cytologic diagnoses of these 85 cases were grouped into four categories: reactive, atypical, Hodgkin lymphoma (HL), and non-Hodgkin lymphoma (NHL). Histologic follow-up was available in 17/85 (20%) of the cases.
Results
5 of the 64 cases in which FCM and cytology did not reveal evidence of a lymphoproliferative disease had tissue follow-up because of persistent lymphadenopathy and high clinical suspicion. 3/5 (60%) confirmed the diagnosis of reactive lymphadenopathy. The two remaining cases (40%) were positive for lymphoma (1HL, 1NHL). 8/15 cases called atypical on cytology had histologic follow-up. 7/8 (87.5%) cases were positive for lymphoma (3HL, 4NHL). 3/4 cases called HL on cytology had tissue follow-up and all 3 (100%) confirmed the diagnosis of HL. One case diagnosed as NHL on cytology was found to be a diffuse large B-cell lymphoma. In summary, out of 17 cases with histologic follow-up 4/17 (24%) were reactive with CD4/CD8 T-cell ratio of 4.1–29, 7/17 (41%) were HLs with CD4/CD8 T-cell ratio of 5.3 – 11, and 6/17 (35%) were NHLs with CD4/CD8 T-cell ratio of 4.2 – 14.
Conclusion
An elevated CD4/CD8 ratio on FCM is a nonspecific finding which may be seen in both reactive and lymphoproliferative disorders. The cytomorphologic features of the smear are more relevant than the sole flow cytometric finding of an elevated CD4/CD8 ratio.
lymph nodeFNAcytologyHodgkin lymphoma
==== Body
Introduction
Performing an excisional biopsy of an enlarged lymph node has been considered standard practice for the evaluation of lymphadenopathy, especially in-patients without prior history of lymphoma [1]. An excised lymph node provides sufficient histologic material to assess cytologic and architectural details. In many centers including ours, it is now becoming more common to forego excisional biopsy and instead perform fine-needle aspiration (FNA) as the initial work-up for lymphadenopathy [2-5]. In addition to being a safe and rapid procedure, FNAs are effective because current lymphoma classification systems place more emphasis on immunophenotype and genotype [6]. Sufficient material can be aspirated for cytomorphology, flow cytometry (FCM), and molecular studies. Even though FNAs have been shown to be effective in the evaluation of lymphomas, especially when used in conjunction with FCM, certain disorders may be difficult to diagnose.
Diagnosis of Hodgkin lymphoma by FNA biopsy is difficult [7-9]. It requires the presence of diagnostic Reed-Sternberg (R-S) cells or their variants (classic, lacunar, popcorn (lymphocytic histiocytic, L&H) and mummified) in a mixed cellular background composed of small lymphocytes, eosinophils, plasma cells, and neutrophils. The difficulty arises from the fact that the number of diagnostic cells may be limited to a few R-S cells or variants. Also, R-S cells may not be seen because the aspirated lymph node is only partially involved by the neoplastic process, which decreases the chance of obtaining the diagnostic cells, or because the neoplastic cells are hidden by fibrosis, granulomas or necrosis, which also changes the cellular background composition, and further complicates the diagnosis. Furthermore, cases were neutrophils are prominent may be mistaken for suppurative lymphadenitis [10]. In addition, this characteristic mixed cellular background is absent in cases of nodular lymphocyte predominant Hodgkin lymphoma and lymphocyte rich classical Hodgkin lymphoma, which in these conditions makes the diagnosis dependent on detection of L&H or R-S cells. On the other hand, in classical HL while the background mixed infiltrate is almost always seen, it is non-specific and may be seen in a variety of other reactive conditions.
Earlier studies examined the lymphocytic component of the background mixed infiltrates to help in diagnosing Hodgkin lymphoma and understanding the biology of this disease. Using immunohistochemical techniques the majority of lymphocytes in the background were shown to be T cells with an increase in helper/suppressor (CD4/CD8) ratio [11,12]. More recent studies showed that the mixed cellular infiltrate characteristic of HL contains an abundant amount of CD4 positiveTh2-lymphocytes [13]. Several cytokines are released by R-S cells leading to stimulation of the influx of CD4+ lymphocytes. Poppema et al. have demonstrated that CC chemokines TARC and MDC attract CD4+ T lymphocytes by binding to a CC-chemokine receptor (CCR4) found specifically on Th2 cells [14]. Fibroblasts in HL have also been shown to produce a CC-chemokine, eotaxin, which has been shown to increase the influx of Th2 lymphocytes and eosinophils [15]. This Th2 milieu generates a comfortable environment for the neoplastic R-S cells to survive and expand.
In this study we investigated the value of an increased CD4/CD8 ratio (≥4), in otherwise non-diagnostic flow cytometry findings, as a diagnostic marker for Hodgkin lymphoma in FNA biopsies. In 85 cases with increased CD4/CD8 T-cell ratio, only 7 cases were proven, using tissue sections, to be Hodgkin lymphoma with one additional case called by cytology (9%). 6 cases were proven to be non-Hodgkin lymphoma (7%). The CD4/CD8 T-cell ratio was not significantly different in these populations 5.27–13 vs. 4.2–14 with average of 7.7 vs. 7.3.
Materials and methods
A review of the cytology records at New York University Medical Center and Bellevue Hospital revealed that between 1996 and 2002, 837 lymph node FNAs were performed in which material was submitted for FCM. In 119 cases (14%), the material submitted was insufficient for flow cytometric evaluation. Our sole selection criterion for including cases in our study was an elevated CD4/CD8 ratio by flow cytometry, defined as greater than or equal to 4. 85/718 (12%) cases fit our criteria. In all the cases, the B lymphocyte component was polyclonal and the T cells showed no aberrant immunophenotype.
All 85 aspirates were obtained on palpable lymph nodes by cytopathologists using 25 to 27 gauge needles. Aspirated material was immediately smeared on slides, air-dried, and stained with Diff-Quik and ultra-fast Papanicolaou stains. Additional material obtained for FCM was placed in RPMI solution. The cytologic diagnoses were grouped into four categories: reactive, atypical, HL, and NHL. Histologic follow-up was available in 17/85 (20%) of the cases.
Flow cytometric studies were performed using FACscan flow cytometer (Becton Dickinson, San Jose, CA). The specimens, suspended in RPMI solution, were centrifuged at 1500 rpm for 5 minutes, the supernatant was discarded, and red blood cells were lysed using Becton Dickinson FACS Lysing Solution. The cell pellets were then washed and resuspended in phosphate buffered saline. Aliquots of the cell suspension were then incubated with different combinations of monoclonal antibodies (three or four in each tube) including CD2, CD3, CD4, CD5, CD7, CD8, CD10, CD19, CD20, CD23, HLA-DR, and Kappa/Lambda light chain (all antibodies from Becton Dickinson, San Jose, CA). An average of 10000 lymphocytes was collected in the lymphocyte gate using forward and side scatter.
Results
As shown in Table 1, the 85 lymph node aspirates were obtained from 85 patients ranging in age from 19 to 87 years. 29 patients were male and 56 were female. Eleven patients had a previous diagnosis of a lymphoproliferative disorder (1 HL, 6 mycosis fungoides, 2 NHL, 1 acute lymphoblastic leukemia, 1 Castleman disease). Lymph node sites were as follows: 14 inguinal, 43 cervical, 15 axillary, 2 epitrochlear, 5 submental, 2 subauricular, 2 intra-parotid, 1 occipital, and 1 mediastinal. The lymph nodes ranged in diameter from 0.5 to 5 cm. CD4/CD8 T cell ratios ranged from 4 to 29. A CD4/CD8 T-cell ratio of 4 was chosen as the cut-off point because it represents twice the normal value (2), and a cut-off point of 3.9 was shown to represent the lower limit of CD4/CD8 ratio is cases of HL in a recently published report [13].
Table 1 Summary of patient's characteristics and clinical history
Patients No.
Total no. of FNA 85
No. of patients 85
Gender (M:F) 29:56 (1/1.9)
Age (yr.) 19–87
Site of FNA
Cervical 43
Axillary 15
Inguinal 14
Submental 5
Epitrochlear 2
Subauricular 2
Inta-parotid 2
Occipital 1
Mediastinal 1
Overall, 64/85 (75%) lymph node aspirates were read as reactive on cytomorphology (Table 2). The cytologic findings in these cases consisted of a heterogeneous population of small and large lymphocytes with no cytologic atypicality mixed with tingible body macrophages with no definitive R-S cells or variants (Figure 1). 5 of the 64 reactive cases in which cytomorphology and FCM did not reveal evidence of a lymphoproliferative disorder had tissue follow-up because of persistent lymphadenopathy and high clinical suspicion. 3/5 cases (60%) confirmed the diagnosis of reactive lymphadenopathy.
Table 2 Summary of cytologic and histologic evaluation
FNA Diagnosis # of cases (%) Biopsy Follow up CD4/CD8
Reactive 64 (75%) 5 (3R, 1HL, 1NHL) 4.1–29 (7.5)
Atypical 15 (18%) 8 (1R, 3HL, 4NHL) 4.6–14 (7.9)
HL 4 (5%) 3 (3HL) 5.3–13 (7.4)
NHL 2 (2%) 1 NHL 4–5.6 (4.8)
FNA: fine needle aspiration; R: reactive; HL: Hodgkin lymphoma; NHL: non-Hodgkin lymphoma. CD4/CD8 ratio range and mean.
Figure 1 A case of reactive lymphadenopathy, cytology preparation shows a mixture of small and large lymphocytes, plasma cells and macrophages.
The two remaining cases (40%) were positive for lymphoma. The first was diagnosed as nodular lymphocyte predominant Hodgkin lymphoma. Histologic evaluation revealed focal effacement of the nodal architecture by vague nodules of small lymphocytes admixed with occasional large atypical cells. Immunohistochemical stains performed on snap-frozen tissue showed that the small cells were mostly B-lymphocytes. The large atypical cells were positive for LCA, CD20 and EMA, and negative for CD15 and CD30. CD57 positive lymphocytes formed rosettes around the large cells. The second case was diagnosed as peripheral T cell lymphoma. The patient had a history of breast cancer treated by chemotherapy and radiation. The lymph node was enlarged due to an expansion of the interfollicular area by medium-sized atypical lymphocytes and increased vascularity with no significant increase in large cells. Immunohistochemistry confirmed that the atypical cells were CD3 and CD4 positive. CD8 positive T cells represented a minority of the cells. Flow cytometry failed to diagnose this case because a limited panel of antibodies against T-cell surface markers was used. The increased CD4/CD8 ratio (4.20) was thought to be a reactive change secondary to chemotherapy for breast cancer.
15 out of 85 cases were read as atypical. These cases consisted mostly of a heterogeneous population of small lymphocytes admixed with a population of intermediate or large cells, some showing angulated nuclei and prominent nucleoli with no definitive R-S cells or variants. 8/15 cases called atypical on cytology had histologic follow-up. 7/8 (87.5%) cases were positive for lymphoma (3 HL, 4 NHL). The remaining case was found to be a reactive lymph node on tissue follow-up. Three of the lymphoma cases were classical HL, nodular sclerosis type I. These lymph nodes were effaced by a mixture of small lymphocytes, plasma cells, and histiocytes forming nodules separated by collagenous bands. Within the mixed cellular infiltrate many classic R-S cells and variants, which stained strongly for CD15 and CD30, were found.
Two of the non-Hodgkin lymphoma cases were diagnosed as diffuse large B cell lymphomas (Figure 2). These cases show lymph nodes with sheets of large lymphocytes having vesicular nuclei and prominent nucleoli. The large cells were LCA and CD79a positive. In one case the large cells were CD30, but not CD15, positive. In these two cases flow cytometry reports noted the presence of a small population of large B cells with no surface immunoglobulin expression. A lymph node biopsy was suggested. The two remaining lymphoma cases were follicular lymphoma (Figure 3). Tissue sections showed ill-defined coalescing enlarged follicles with lost polarity, composed of sheets of large lymphocytes with no tingible body macrophages. The follicular cells were positive for B cell markers and bcl-2. In these cases flow cytometry results suggested the presence of a small population of large B cells with very dim expression of CD10 and surface immunoglobulin, called atypical and lymph node biopsy was suggested.
Figure 2 A case of diffuse large B cell lymphoma with limited amount of recovered cells and no diagnostic flow cytometry findings. The cytology preparation shows predominantly large cells with high nuclear/cytoplasmic ratio, irregular nuclei and occasional prominent nucleoli.
Figure 3 A case of follicular lymphoma, cytology preparation shows a mixture of small lymphocytes with scanty cytoplasm and irregular nuclei (centrocytes) mixed with a population of larger lymphocytes with scanty cytoplasm with rounded nuclei and single or multiple small nucleoli (centroblasts).
Three of 4 cases called HL on cytology had tissue follow-up and in all 3 (100%) the diagnosis of HL was confirmed. Two were classic HL, nodular sclerosis type with abundant lacunar R-S cells (Figure 4). The third case was a syncytial variant of HL, nodular sclerosis type. This case contained solid cohesive clusters of atypical cells that stained for CD30 and CD15. Flow cytometry for these three cases showed an increase in CD4/CD8 T cell ratio with polyclonal B cells.
Figure 4 A case of classical Hodgkin lymphoma, nodular sclerosis, cytology preparation shows binucleated Reed-Sternberg cells in a mixed cellular background.
Two of the 85 cases were diagnosed as NHL on cytology. The first case showed a population of large pleomorphic lymphocytes with significant cytologic atypia. No tissue follow-up was available for this case. The second case showed a population of large atypical lymphocytes in a background of necrosis. Tissue follow-up revealed a diffuse large B cell lymphoma. Flow cytometry study for these two cases failed to detect any large cell component and was called negative.
In summary, out of 17 cases with histologic follow-up 4/17 (24%) were reactive with CD4/CD8 ratio of 4.1–29, 7/17 (41%) were HLs with CD4/CD8 ratio of 5.3 – 11, and 6/17 (35%) were NHLs with CD4/CD8 T-cell ratio of 4.2 – 14.
Discussion
Cytomorphology and FCM are complementary and effective in differentiating between reactive and lymphoproliferative processes, specifically non-Hodgkin lymphomas (NHLs) [3-5,16-18]. The accuracy and reliability of these techniques allows for a diagnosis without the need to perform an excisional biopsy. However, there are instances where the FCM results are nondiagnostic, but they may instead imply a certain disorder. An increased CD4/CD8 ratio in the presence of a polyclonal B lymphocyte population is one such condition. This finding has been reported in the literature to be associated with, but not diagnostic of, Hodgkin lymphoma (HL) [11-13,19,20]. Using flow cytometry, a recent study shown an increase in CD4/CD8 T-cell ratio (range 3.9 to 28 with average of 11.2) in lymph nodes of patients with classical Hodgkin lymphoma [11].
In this study, out of 85 cases read as reactive/atypical by flow cytometry, 13 were proven to be lymphoma using confirmatory biopsy (7 HD and 6 NHL). Cases of HL were not accurately diagnosed by FCM because of the paucity of neoplastic cells relative to reactive cells. CD4/CD8 T-cell ratio was not helpful in the diagnosis of HL. The average value of CD4/CD8 T-cell ratio was essentially the same for reactive, HL, and NHL cases. Overall, using the currently available antibodies, HL is not a diagnosis to be made by FCM. In cases of suspected HL, it is better to perform immunocytochemical stains on cytospin preparations or cellblock sections using the appropriate markers such as CD15 and CD30.
Six NHL cases were not accurately interpreted by FCM. One case was T-cell lymphoma and 5 cases were large B-cell lymphoma. The T-cell lymphoma case was missed by FCM because an insufficient panel of monoclonal antibodies was used and the lymphoma diagnosis was not clinically suspected. Four out of the five cases of large cell lymphoma were misinterpreted because FCM failed to confirm B-cell monoclonality. In these four cases the presence of a small population of large cells with no surface immunoglobulin restriction was mentioned and a confirmatory lymph node biopsy was suggested. Absence of immunoglobulin in these large cells is abnormal and in itself is a proof of malignancy [22]. Several recent studies highlighted the difficulty in diagnosing DLBL by flow cytometry due to lack of surface immunoglobulin expression or limited cell recovery [5,12,21,22]. These studies concluded that, in such cases diagnosis of lymphoma could be made in confidence if the cytology features and the clinical settings were reviewed.
The fifth NHL case was missed by flow cytometry because of the limited number of cells recovered from the FNA sample, which was largely necrotic. The presence of necrotic cells in FN aspiration is often an indication of malignancy. In these cases FCM may be helpful in detecting small population of clonal viable cells. However, lack of clonality does not exclude a neoplastic process and the limitation of the study has to be mentioned clearly in the FC diagnostic report.
In summary, using an increased CD4/CD8 T-cell ratio as the sole diagnostic abnormality did not help in differentiating HL from reactive or NHL cases. Cytology diagnosis and confirmation by immunohistochemical studies performed on cytospin or cellblock preparations is probably more relevant than FCM in cases of HL. In working up cases of large cell lymphoma or those with significant necrosis, communication between cytologist and hematopathologist, coordination of cytomorphology with immunophenotypic data, and knowledge of the clinical setting are essential to reach accurate diagnoses.
Note
Corresponding article: Beaty MW, Geisinger KR: Hodgkin lymphoma flow me? Cytojournal 2005, 2:13 [23]
Acknowledgements
"Co-editors of CytoJournal Vinod B. Shidham, MD, FRCPath, FIAC and Barbara F. Atkinson, MD thank the academic editor: Nancy Young, M.D., Department of Pathology, Fox Chase Cancer Center, 7701 Burholme Avenue Room C427, Philadelphia, PA 19111 (Email: [email protected]) for organizing and completing the peer-review process for this manuscript."
==== Refs
Habermann TM Steensma DP Lymphadenopathy Mayo Clin Proc 2000 75 723 732 10907389
Buley ID Fine needle aspiration of lymph nodes J Clin Pathol 1998 51 881 885 10070327
Cartagena N Katz RL Hirsch-Ginsberg C Childs CC Ordonez NG Cabanillas F Accuracy of diagnosis of malignant lymphoma by combining fine-needle aspiration cytomorphology with immunocytometry and in selected cases, Southern blotting of aspirated cells: a tissue-controlled study of 86 patients Diagn Cytopathol 1992 8 456 464 1396024
Steel BL Schwartz MR Ramzy I Fine needle aspiration biopsy in the diagnosis of lymphadenopathy in 1,103 patients: role, limitations and analysis of diagnostic pitfalls Acta Cytol 1995 39 76 81 7847013
Meda BA Buss DH Woodruff RD Diagnosis and subclassification of primary and recurrent lymphoma: the usefulness and limitations of combined fine-needle aspiration cytomorphology and flow cytometry Am J Clin Pathol 2000 113 688 699 10800402 10.1309/0Q7F-QTGM-6DPD-TLGY
Harris NL Jaffe ES Stein H World Health Organization classification of neoplastic diseases of the hematopoietic and lymphoid tissues: report of the Clinical Advisory Committee meeting J Clin Oncol 1999 17 3835 3849 10577857
Tarantino DR McHenry CR Strickland T Khiyami A The role of fine-needle aspiration biopsy and flow cytometry in the evaluation of persistent neck adenopathy Am J Surg 1998 176 413 417 9874424 10.1016/S0002-9610(98)00233-5
Jennings CD Foon KA Recent advances in flow cytometry: application to the diagnosis of hematologic malignancy Blood 1997 90 2863 2892 9376567
Moreland WS Geisinger KR Utility and outcomes of fine-needle aspiration biopsy in Hodgkin's disease Dign Cytopathol 2002 26 278 282 10.1002/dc.10104
Vicandi B Jimenez-Heffernan JA Lopez-Ferrer P Hodgkin's disease mimicking suppurative lymphadenitis: a fine-needle aspiration report of five cases Dign Cytopathol 1999 20 302 306 10.1002/(SICI)1097-0339(199905)20:5<302::AID-DC11>3.0.CO;2-X
Borowitz MJ Croker BP Metzgar RS Immunohistochemical analysis of the distribution of lymphocyte subpopulation in Hodgkin's disease Cancer Treat Rep 1982 66 667 674 6210433
Gupta S Subpopulation of human T lymphocytes. Maldistribution of T cell subsets associated with abnormal locomotion of T cells in untreated adult patients with Hodgkin's disease Clin Exp Immunol 1980 42 186 195 6970098
Gorczyca W Weisberger J Liu Z An approach to diagnosis of T-cell lymphoproliferative disorders by flow cytometry Cytometry 2002 50 177 190 12116341 10.1002/cyto.10003
Poppema S van den Berg A Interaction between host T cells and Reed-Sternberg cells in Hodgkin lymphomas Sem in Cancer Biol 2000 10 345 350 10.1006/scbi.2000.0327
Jundt F Anagnostopoulos I Bommert K Hodgkin/Reed-Sternberg cells induce fibroblasts to secrete eotaxin, a potent chemoattractant for T cells and eosinophils Blood 1999 94 2065 2071 10477736
Dong HY Harris NL Preffer FI Pitman MB Fine-needle aspiration biopsy in the diagnosis and classification of primary and recurrent lymphoma: a retrospective analysis of the utility of cytomorphology and flow cytometry Mod Pathol 2001 14 472 481 11353059 10.1038/modpathol.3880336
Ravinsky E Morales C Kutryk E Chrobak A Paraskevas F Cytodiagnosis of lymphoid proliferations by fine needle aspiration biopsy. Adjunctive value of flow cytometry Acta Cytol 1999 43 1070 1078 10578980
Liu K Stern RC Rogers RT Dodd LG Mann KP Diagnosis of hematopoietic processes by fine-needle aspiration in conjunction with flow cytometry: a review of 127 cases Diagn Cytopathol 2001 24 1 10 11135461 10.1002/1097-0339(200101)24:1<1::AID-DC1000>3.0.CO;2-J
Poppema S Immunology of Hodgkin's disease Baillieres Clin Haematol 1996 9 447 457 8922239
Knowles DM Immunophenotyping and immunogenotyping approaches useful in distinguishing benign and malignant lymphoid proliferations Semin Oncol 1993 20 583 610 8296197
Mayall F Dray M Stanley D Harrison B Allen R Immunoflow cytometry and cell block immunohistochemistry in the FNA diagnosis of lymphoma: a review of 73 consecutive cases J Clin Pathol 2000 53 451 457 10911803 10.1136/jcp.53.6.451
Young NA Al-Saleem TI Ehya H Smith MR Utilization of fine-needle aspiration cytology and flow cytometry in the diagnosis and subclassification of primary and recurrent lymphoma Cancer Cytopathol 1998 84 252 261
Beaty MW Geisinger KR Hodgkin lymphoma flow me? Cytojournal 2005 2 13 16150141 10.1186/1742-6413-2-13
|
16153296
|
PMC1242242
|
CC BY
|
2021-01-04 16:36:22
|
no
|
Cytojournal. 2005 Sep 9; 2:14
|
utf-8
|
Cytojournal
| 2,005 |
10.1186/1742-6413-2-14
|
oa_comm
|
==== Front
Harm Reduct JHarm Reduction Journal1477-7517BioMed Central London 1477-7517-2-151616475110.1186/1477-7517-2-15Book ReviewReview of "In the Eye of the Needle: Diary of a Medically Supervised Injecting Centre" by Ingrid van Beek Allen & Unwin 2004 Clear Allan [email protected] Executive Director, Harm Reduction Coalition, 22 West 27th Street, 5th Fl, New York, NY 10001, USA2005 15 9 2005 2 15 15 12 7 2005 15 9 2005 Copyright © 2005 Clear; licensee BioMed Central Ltd.2005Clear; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
==== Body
However they are labeled, there are a couple of dozen safe injection facilities, safe injection rooms, safe injection spaces, drug consumption rooms, or medically supervised injecting centres around the globe. The most recent have appeared in Vancouver, Canada, and the most scrutinized is in Sydney, Australia. Dr. Ingrid Van Beek has diarized the early history of Sydney's centre in her book, "In the Eye of the Needle". This book is much more than the story of the medically supervised injection center. The lessons imparted here are invaluable for everyone working with drug users and have universal application. Van Beek skillfully weaves several different strands throughout the book including the mechanics of opening and running the center; the emotional toll it takes on staff; the humanity of the drug users the center serves; the need of the clients; and the scrutiny that an institution comes under for being deemed "controversial". Ultimately this book is about how compassion, healthcare, dignity and human rights can be obtained for drug users.
Medically supervised injection centres are controlled environments aimed at reducing the negative consequences of injecting in public places and are usually staffed by medical professionals and social workers. Injecting in uncontrolled spaces leads to missed and hurried shots, overdose risk and "offending" the general public. Supervised injection centres emerged in Western Europe in the 1990s, principally in Switzerland and Germany, and evaluations of the centres have shown them to be an effective intervention and a practical tool for health promotion among drug users. The centres are particularly appropriate in locations with thriving street based drug markets. As experienced in Vancouver and Sydney, it is not uncommon for injection rooms to be foreshadowed by activists who first set up illegal spaces and then gain legal status as a result of government support.
Despite the corny title, "In the Eye of the Needle" is the best book yet written on the experience of working in the field of harm reduction. It sounds like I'm damning with faint praise because, as far as I know, there are no other books that explore the worker's experience. But no, this is really a great book. You do not have to open a safe injection room to relate to everything that occurs herein. Involved in drug services, needle exchange, housing, mental health, drug treatment? The book covers, all too familiarly, the issues – site location, back stabbing by colleagues, under appreciation of hard working staff, tears, sweat and blood everywhere. More blood than you can possibly imagine. Blood on the walls and blood on floor. We're talking about people shooting up. Favorite drug users die and colleagues who you don't know are using, overdose. Opposing politicians and the media are scurrilous, self-serving, immoral hypocrites. At the same time, politicians and media come through with the support when needed, always by a hair's breadth and often without getting any substance behind the facts. It's a lonely business sometimes. The only people who seem to know what goes on are the workers. And the drug users. Read this book and plot out a media strategy.
Dr. Ingrid van Beek was already running primary health care services for drug users when she was contacted by a local police chief concerned about the volume of emergency calls for overdose situations. She took on the oversight of the medically supervised injecting centre on top of her day job and spent the next couple of years running from centre to centre. The organizing work was intense and all encompassing. Van Beek worked with everyone from the local residents and business groups to government representatives and law enforcement. At the same time, the United Nations International Narcotics Control Board decided to criticize the centre, overlooking for arcane reasons the pre-existing European injection spaces, causing more political fall out for the centre.
As a clinician, Dr. van Beek is the perfect foil for taking on controversial services. She handles the burdensome site visits from the police and health authorities with a stellar resignation. "Although I pointed out that no other health facility in the land is routinely inspected by licensing authorities without notice, it was argued that some in the community might expect this and we must be seen to be absolutely squeaky clean in all respects." Who in the harm reduction field does not feel that we are held to a different standard than other services? The medically supervised injecting centre, as described in the book, is clinical in design as opposed to the more relaxed community oriented European model of injection rooms. Smoking is not allowed. Users cannot hang out. In 18-months, the centre handled 554 overdoses without loss of life and thousands of injection episodes. The most poignant moment in the book is the frustration of receiving an evaluation that minimized the success of the centre by giving it a marginal passing grade. Rigor is fine in research but researcher rigor mortis is sad. The number of deaths the opening of the centre prevented cannot be quantified. Prevention is hard to prove conclusively. Too often research is cautious and verges on the ridiculously conservative. It seems this was the case here.
Aside from describing the creation of the centre, this book is a much-needed primer for overdose prevention. Overdose deaths can be largely prevented with appropriate and timely care. When the centre first opened, Sydney experienced a heroin drought and most users in distress were revived with oxygen. Only as the drought was ending and heroin purity increasing did the centre use naloxone, and then only infrequently. "Increasingly," van Beek writes, "I appreciate that the injecting centre provides a unique setting in which health care workers actually see the overdose occurring from the very outset, identifying symptoms of heroin (or whatever drug) overdose and administering appropriate treatment very soon thereafter. This can't and doesn't occur in any other circumstance. By treating an overdose so early in its course, the damage already done and its natural progression is reversed so that Narcan (used to start breathing) will no longer be needed in most cases. This is how injecting centres potentially reduce the morbidity (damage to vital organs, especially the brain) and the mortality otherwise associated with overdose in unattended situations, even when there is a very prompt and efficient ambulance service on hand."
Globally, overdose has not had the attention that HIV prevention for drug users has received. It is an issue, however, that deserves equal attention. Whereas syringe exchange programs emerged rapidly as an HIV prevention strategy in the United States at the beginning of the nineties – seven programs in the late 1980s to 90 programs five years later – overdose prevention and education, however, have not expanded in the same fashion. 5 campaigns in five cities or states in five years is inadequate considering, west of the Rockies, drug users are more at risk of dying of an opiate overdose than from HIV. In France, scaling up of buprenorphine (Subutex) has greatly reduced mortality from opiate overdose. In Russia overdose is common, syringe exchanges are stagnant, and substitution therapy is illegal. In such a milieu, it is hard to envision a supervised injection center. Van Beek's detailed account of such a center can help the uninformed understand the mechanics of overdose first hand. One of the strengths of the book is that it is detailed enough to be of universal help in guiding one to develop overdose interventions for drug users without having to open a centre.
Aside from overdose prevention, safe injection rooms are primarily thought of as nuisance abatement strategies. Although not touched upon in the book, they are good venues for safe injection education on reducing soft tissue infections and disease prevention. At the recent International Harm Reduction Conference in Belfast, staff from the Sydney medically supervised injecting centre talked about the lack of blood awareness and the poor injection techniques of injectors, despite years of education by providers. Investigation by staff at the centre revealed that education such as rotating veins or releasing tourniquets before injection weren't necessarily practical for the user. This poses new challenges for providers and users alike, but it is a path we need to travel together.
It has been less of a challenge to incorporate harm reduction as a national approach to drugs in countries with health care systems that help people when they get sick, as opposed to systems that benefit insurance and pharmaceutical companies when a consumer needs services. It is also less challenging in countries that recognize that drug users are citizens with rights. This book, however, points up that even within a country, such as Australia, which prides itself on pragmatism, it is still difficult to implement an intervention that can benefit the lives of drug users. The story of the development of the medically supervised injection centre is a classic case of harm reduction struggling to assert its worth. Despite the success of the centre, there is no intention of replicating the facility in other places in Australia. It has not inspired other countries to follow suit (except for Vancouver, Canada) and injection centres are taking the path of other interventions for drug users in that they will be adopted gradually at a glacial pace.
I would love to see a spate of books inspired by "In the Eye of the Needle: Diary of a Medically Supervised Injecting Centre". We could come up with our own "Catch 22" or "One Flew Over the Cuckoo's Nest" to match the Kafkaesque world we inhabit. In the meantime we have with "In the Eye of the Needle", a readable, intelligent, quality account of the creation of a harm reduction program.
|
0
|
PMC1242243
|
CC BY
|
2021-01-04 16:36:49
|
no
|
Harm Reduct J. 2005 Sep 15; 2:15
|
utf-8
|
Harm Reduct J
| 2,005 |
10.1186/1477-7517-2-15
|
oa_comm
|
==== Front
Int J Health GeogrInternational Journal of Health Geographics1476-072XBioMed Central London 1476-072X-4-221617657710.1186/1476-072X-4-22EditorialWeb GIS in practice III: creating a simple interactive map of England's Strategic Health Authorities using Google Maps API, Google Earth KML, and MSN Virtual Earth Map Control Boulos Maged N Kamel [email protected] School for Health, University of Bath, Claverton Down, Bath BA2 7AY, UK2005 21 9 2005 4 22 22 19 9 2005 21 9 2005 Copyright © 2005 Boulos; licensee BioMed Central Ltd.2005Boulos; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
This eye-opener article aims at introducing the health GIS community to the emerging online consumer geoinformatics services from Google and Microsoft (MSN), and their potential utility in creating custom online interactive health maps. Using the programmable interfaces provided by Google and MSN, we created three interactive demonstrator maps of England's Strategic Health Authorities. These can be browsed online at – Google Maps API (Application Programming Interface) version, – Google Earth KML (Keyhole Markup Language) version, and – MSN Virtual Earth Map Control version. Google and MSN's worldwide distribution of "free" geospatial tools, imagery, and maps is to be commended as a significant step towards the ultimate "wikification" of maps and GIS. A discussion is provided of these emerging online mapping trends, their expected future implications and development directions, and associated individual privacy, national security and copyrights issues. Although ESRI have announced their planned response to Google (and MSN), it remains to be seen how their envisaged plans will materialize and compare to the offerings from Google and MSN, and also how Google and MSN mapping tools will further evolve in the near future.
==== Body
Background
The emergence of online consumer geoinformatics services
Mainstream Web search engines like Google and MSN Search have recently joined the geographic search bandwagon by releasing their own dedicated geographic interfaces, which run in standard Web browsers and also provide the general public with detailed satellite imagery/aerial photography map layers that were once only available to experts and select user communities.
Google Maps and Google Earth
Google released Google Maps ( – a localized UK version is also available; see, for example, Paddington Station, London, W2 1RH (Satellite): ). Google also released Google Earth , a fat client, standalone 3D (three-dimensional) desktop application that offers anyone with Internet access a planet's worth of imagery and other geographic information, allowing users to virtually sightsee exotic locales like Paris, France, and Maui (Hawaii), Grand Canyon and Niagara Falls in the USA, as well as viewing points of interest such as local restaurants, hospitals, schools, and more. Google Earth uses KML (Keyhole Markup Language) to store data [1].
(It is noteworthy that NASA is also offering its own World Wind 3D application that lets users zoom from satellite altitude into any place on Earth.)
Thanks to Google Maps API (Application Programming Interface – ), many third party applications and custom annotated maps have begun to appear [2]. Two good UK examples of such applications/custom annotated maps are the Health QOF (Quality and Outcomes Framework) Database map and London July 2005 Terrorist Attacks map . (But because Google Maps has its roots in XML (eXtensible Markup Language), users were also able to produce their own custom annotated Google maps, e.g., based on their own GPS (Global Positioning System) locational data, and to even tie in images and video to create interactive multimedia maps, well before the API was publicly documented [3].)
Smugmug Maps is another good example of a third-party Google Maps application in action. Smugmug, a photo hosting Web site, plots geocoded photographs to their actual locations on Google Maps (or Google Earth via a KML Google Earth feed: ), and allows location-based searching of photographs all over the world [4]. All smugmug RSS (Really Simple Syndication) feeds are now geo-enabled. If a photo has latitude, longitude, and altitude information (geographic metadata), it will show up in all feeds (see ).
MSN Virtual Earth
Microsoft's response to Google Maps and Google Earth comes in the form of MSN Virtual Earth . A distinguishing feature of MSN Virtual Earth is its 'Locate Me' tool. Wired users can be located via their IP (Internet Protocol) address (this has been done for some time – see, for example, [5] and ). Wireless users can download a small application that does locating based on connection to a Wi-Fi access point. MSN Virtual Earth also features aerial oblique imagery (45 degree angle views or 'Eagle Eye Views') of major US metropolitan areas, provided by Pictometry International Corp. .
An MSN Virtual Earth Map Control/API (see and ) allows users to create their own custom online maps, and add their own data to MSN Virtual Earth.
Yahoo! Maps
Corresponding offerings from Yahoo! search engine have been modest by comparison, and include Yahoo! Maps and an associated API . As at the time of writing, the Yahoo! Maps service does not offer any satellite imagery/aerial photography, but this might change in the near future. The latest traffic status/incidents, as well as Wi-Fi hotspots can be visualized on Yahoo! Maps. Gottipati [6] provides a useful comparison of Google Maps API and Yahoo! Maps API.
Web browser toolbars and other developments
Dedicated Web browser toolbars and extensions have also started to appear, e.g., MutantMaps , a Mozilla Firefox toolbar that allows navigation between five popular mapping sites (Google Maps, MSN Virtual Earth, MultiMap.co.uk, TerraServer.com and 192.com) while preserving user's longitude, latitude and zoom levels, and gMapIt, another Mozilla Firefox extension that allows users to find directions from Google Maps based on publicly listed US phone numbers .
Along the same vein, Amazon.com is now also providing A9 Block View , an online Yellow Pages/map service that offers US maps with street-level photos.
ESRI's response
Some commentators have recently wondered if users will soon eschew ArcGIS and ArcIMS (see and ) in favour of using Google Maps API and MSN Virtual Earth API to quickly create Web map applications. ESRI's response to all of the recent online consumer geoinformatics services described in this article was to announce (in 2005) its new partnership with National Geographic , GlobeXplorer , and TeleAtlas , plus Geospatial One Stop (GOS – ) and a few other partners like MDA (MacDonald, Dettwiler and Associates Ltd. – ), to upgrade the National Geographic MapMachine , a map service/online atlas that provides global map coverage for an extensive set of Earth science themes. MapMachine was first launched in November 1999, and is powered by ESRI's ArcWeb Services . The planned upgrade aims at bringing satellite imagery, aerial photos, and street-level data to MapMachine users. Users will be able to access the service through a new viewer that is aimed at a mass audience, and appears to be ESRI's direct response to Google Earth and Microsoft Virtual Earth. However, one important difference from those services is that the ArcGIS back end will also allow users of the new service to accomplish much more sophisticated tasks, such as service area analysis. The next generation of MapMachine will also provide a link to GOS data and metadata to help users discover information about their area of interest or study. MapMachine will include capabilities for 3D globe services, allowing GIS users to "pull in" their own map services to overlay onto a globe. Also planned is the addition of ESRI's MapStudio , an ArcWeb Services application used by many daily newspapers to create maps for printing, to enable users to create customised maps [7].
Methods
We wanted to experiment with the programmable interfaces provided by Google and MSN by mapping the headquarters locations of England's 28 Strategic Health Authorities (SHA), and associating these locations on the resultant maps with relevant online information about the corresponding SHAs from England's National Health Service (NHS) Web site .
Geocoding
The programmable mapping interfaces provided by Google and MSN currently only accept longitude and latitude coordinates and do not provide their own geocoding services. Geocoding in our case was only based on the first part of SHA headquarters postcodes (e.g., PL12 for the South West Peninsula SHA, PL12 6LE), and was done using jibble.org list (see and download at ). The worldKit geocoder , a free online worldwide city geocoder, also uses jibble.org list for UK postcode geocoding.
Google Maps API version of our SHA map
Guided by Google Maps API online documentation and Gottipati's online tutorial [6], we produced the Google Maps API version of our interactive SHA map. Google Maps API lets developers embed Google Maps in their own Web pages with JavaScript.
We had to first visit Google's sign-up page to get a free API key for the Web site where our maps were to be published. API keys are site-specific. We included all SHA coordinates and Internet addresses (for accessing further information) in a separate XML file. An XSLT (eXtensible Stylesheet Language Transformation) stylesheet is used to display SHA information taken from this XML document.
We used J. Shirley's GxMarker (see and download at ) instead of Google's GMarker to have marker tooltips (see inset in Figure 1).
Figure 1 Screenshot of Google Maps API version of England's SHA Locator. Screenshot of our Google Maps API version of England's SHA Locator showing the shadowed "info window" for Cumbria and Lancashire SHA, with a clickable external link to access further information about this SHA . The map features all the standard Google Maps controls for zooming, panning (also possible by dragging the map), and displaying/switching satellite and hybrid views. The inset shows an example of the tooltips that appear when the mouse hovers over the markers (or pins) on the map.
Readers wanting to further explore Google Maps API might be interested in Google Mapki Knowledge Base and list of developer tools . Google Mapki is a forum for sharing ideas, implementations, and help for the Google Maps API.
Google Earth KML version of our SHA map
The development of the Google Earth KML version of our maps was again guided by Google's online documentation available at . KML is an XML-based language. A new MIME (Multipurpose Internet Mail Extensions) type (application/vnd.google-earth.kml+xml kml) must be added to the server hosting the KML feed file to help client Web browsers like Internet Explorer associate the file with the appropriate client application (Google Earth) rather than opening it as a plain XML file.
MSN Virtual Earth Map Control version of our SHA map
Finally, we created a third version of our maps using MSN Virtual Earth Map Control (download control at ), and guided by Part 1 of Roodyn's excellent tutorial available from Via Virtual Earth, Virtual Earth developer resource centre [8].
It should be noted that not all possible features of the programmable mapping interfaces provided by Google and MSN have been explored or demonstrated in our exercise and its outputs (see 'Results' below). For example, it is also possible to add VML (Vector Markup Language) polyline overlays to maps created using Google Maps API. Also, Part 2 of Roodyn's tutorial describes additional controls and widgets that can be used with MSN Virtual Earth Map Control [9].
Results
The three interactive SHA map demonstrators we have created can be browsed online at (Google Maps API version – Figure 1), (Google Earth KML version – Figure 2), and (MSN Virtual Earth Map Control version – Figure 3). The maps have been successfully tested in both Internet Explorer 6-SP2 and Mozilla Firefox 1.0.6 Web browsers.
Figure 2 Screenshot of Google Earth KML version of England's SHA Locator. Screenshot of our Google Earth KML version of England's SHA Locator (see instructions at ) showing our KML SHA feed in Google Earth, with an "info window" for West Yorkshire SHA. "Info windows" allow users to access external Summary Information about the corresponding SHA ( in this example, displayed in the lower pane of Google Earth), as well as Google Earth-generated driving directions to or from the selected SHA. The KML feed featured in this screenshot is available at and is intended to be opened by Google Earth desktop application, which can be downloaded at .
Figure 3 Screenshot of MSN Virtual Earth Map Control version of England's SHA Locator. Screenshot of our MSN Virtual Earth Map Control version of England's SHA Locator in combined street and aerial (satellite) style modes. Like the main MSN Virtual Earth service, our custom map allows users to move the map around by dragging it, to zoom in and out with the mouse wheel, and to zoom in by double-clicking on a location. Clicking a Strategic Health Authority 'S' marker (or pin) on the map will display the corresponding Summary Information page from England's National Health Service (NHS) Web site (e.g., for Leicestershire, Northamptonshire and Rutland SHA).
Discussion
The geodata-rich society
ESRI president Jack Dangermond recently predicted that the supply of satellite and aerial imagery will increase by two folds in the next few years. Availability will also increase greatly, via Web portals and online GIS services. This is all part of what Dangermond describes as a "geodata-rich society" that will have access to more geospatial information of all kinds, including, in addition to imagery, GPS/location data, geo-demographic data, and data from real-time monitoring [10]. The Internet is already the 'foundation medium' to access, link and use all these data.
Satellite imagery and remote sensing are quickly entering the mainstream. Today, satellite imagery data are abundantly available from multiple sources, including companies such as Space Imaging , Orbimage , DigitalGlobe , GlobeXplorer , Spot Image , ImageSat International , and EarthSat ( – an MDA company), and are used in hundred of applications. But thanks to online consumer services like Terraserver , Google Earth, and Microsoft Virtual Earth (see 'Background' section above), satellite imagery has also been made familiar and accessible to millions of people.
The wikification of GIS, maps and satellite imagery/aerial photography: imaging and geospatial information for the wide masses
There is no doubt the different online consumer geoinformatics services that have been presented in the 'Background' section of this article, including the different geographic search interfaces from major Web search engine providers, have significantly contributed (in record time) to raising the general public interest in geography and satellite imagery. As millions of people start "playing" with these new online "gadgets" or "toys" from Google and Microsoft, many of them will soon start thinking about becoming active participants, sharing information and collaborating online (notions that have been rightly associated with the Web for quite a long time), rather than just being satisfied with a passive information consumer/viewer role. (The reader should note that it has been estimated that about 800 million persons are online today worldwide [11].)
However, although Google Maps API (and similar API offerings from other providers) enables users to deeply customize the standard provider's interface (Google Maps), and to create their own custom annotated maps (custom applications based on Google Maps), such APIs remain difficult for the non-expert, average user to exploit. This author expects the technology to further evolve to enable the average Web user to share geospatial information, to customize, annotate and publish his/her own online maps and related Web applications, and to collaborate with other users/online communities within an online customizable and collaborative mapping environment, all without the need for any prior programming knowledge or expertise. (Such user-friendly applications that do not require end-users to have any programming expertise to use them can also be built using the existing APIs.)
The current 'wiki' concept is not far from this vision. A wiki (from Hawaiian wiki, to hurry, swift) is a collaborative Web site whose content can be edited by anyone who has access to it [12]. Perhaps the best example of a wiki in action today is 'Wikipedia – The Free Encyclopedia' (see the 'wiki' entry in Wikipedia at ). A related Web information sharing technology is the 'blog'. A blog (WeBLOG) is a Web site that contains dated entries in reverse chronological order (most recent first) about a particular topic. Functioning as an online journal, blogs can be written by one person or a group of contributors. Entries contain commentary and links to other Web sites, and images as well as a search facility may also be included ([13] – see the 'blog' entry in Wikipedia at ).
Wikis, and in particular Wikipedia, have grown very popular in recent months and years [14]. Wikis represent a promising principle that can significantly transform the Internet information age. Special conferences have been and are being organized to discuss this interesting Web phenomenon of wikis; for example, Wikimania 2005, the First International Wikimedia Conference, 4–8 August 2005, Frankfurt am Main, Germany , and the ACM (Association for Computing Machinery)-sponsored WikiSym 2005, the 2005 International Symposium on Wikis, 17–18 October 2005, San Diego, California, USA .
Along the same lines, it is not difficult to imagine the development in the very near future of 'geowikis', 'mapwikis', geo-enabled blogs, 'mapblogs' (imagine, for example, people with an Internet-connected, GPS-enabled mobile device wanting to blog their movements, and share their activity spaces and geo-referenced news with other online users for various purposes), and even geo-enabled, mappable Web/RSS feeds and map feeds (see the Smugmug KML photo feed example mentioned in the 'Background' section above). In fact some early geowiki examples have already found their way on the Web; see, for example, , , and also worldKit GeoWiki, a publicly editable map application (a simple online demo of worldKit GeoWiki to which anyone can add their own data is available at ).
Another example is the Katrina Information Map , which was built using Google Maps [15]. Katrina Information Map was conceived for use by people affected by Hurricane Katrina (August 2005) and their relatives who have, or are trying to find, information about the status of specific locations affected by the storm and its aftermath. Users having information about the status of an area that is not yet on the map can easily contribute to the map by adding/appending their information to it. (Readers interested in Hurricane Katrina's online maps and imagery in general might also find the following two sites useful: , , and .)
The possibilities and potentials are endless. This is what this author calls the ultimate "wikification" of GIS, maps and satellite imagery/aerial photography. If the majestic Tate Museum in London is currently posting captions from its visitors next to its greatest works of art [16], why shouldn't online maps (even those from very reputable sources like the National Geographic Society) allow a similar approach?!
Associated individual privacy, national security, data confidentiality, and copyrights/digital rights management issues
As geospatial technology progresses and becomes more readily available to the wide masses around the world who are connected to the Internet, the interrelated issues of GIS and map data confidentiality/individual privacy, and even national security start to surface, calling for further examination of, and research into these delicate aspects of Internet GIS and Web maps [5,17-22].
For example, in public health worldwide, any public identification of an individual's health status and residence, regardless of level of contagion or risk, is usually prohibited with very few exceptions, e.g., Megan's Law in the US, which allows the release of residential information on registered child sex offenders to the public by local government [17,23]. In fact, thanks to the latter law, we have a service like the Georgia Sex Offender Maps , which was built using Google Maps API. SARS (Severe Acute Respiratory Syndrome) mapping in Hong Kong in 2003 using disaggregate case data at individual building level in near real time was another noticeable exception to this well-established public health confidentiality rule, and also a unique and rare GIS opportunity that resulted in some very comprehensive public Internet mapping services [24].
It is noteworthy that Google Maps API terms and conditions state, "There are some uses of the API that we just don't want to see. For instance, we do not want to see maps that identify the places to buy illegal drugs in a city, or any similar illegal activity. We also want to respect people's privacy, so the API should not be used to identify private information about private individuals."
On another level, following the September 2001 terrorist events in the US, many federal and local spatial databases, e.g., "critical infrastructure" spatial data, were assessed by their holding agencies as a potential liability to national security and withdrawn from the Internet or public dissemination. The current concern is to find an appropriate balance between public access to spatial information and protection of information considered a priority for national security [17,23].
But despite all these undeniable, legitimate and real concerns about Internet GIS and map data privacy and confidentiality, many of the doubts and misgivings that are raised concerning these aspects of Internet GIS seem to be ill founded, or at least exaggerated. Entchev [19] has wisely stated, "Let us not cripple the GIS system to meet some vague privacy perceptions".
Another thorny Internet GIS issue that needs to be addressed is that of data and map copyrights. Conner [14] has rightly described online maps as a copyright minefield. Copyrighted geo-data and maps are usually more difficult and expensive to acquire and use.
But as geo-data become more important in everything from blogs through mobile phones to finding lost people, free maps could make more and more of a difference [14]. However, someone needs to pay the bill for such "free" maps, and so finding sustainable commercial models for adoption by online geo-data and Web map providers is becoming of prime importance these days [25]. Examples of such commercial models include ad-sponsored map services, and low-cost, added-value paid services supporting the free service like Google Earth plus and Google Earth Pro . Microsoft also provides an alternative ad-supported, but still free, "commercialized" version of their MSN Virtual Earth Map Control for commercial Web sites [26].
The Open Geospatial Consortium's (OGC) work on Geospatial Digital Rights Management (GeoDRM) is also poised to become an important enabler in the context of geo-data and map copyrights [27]. A great deal of work has already been done in the area of data ownership and rights management for the online e-book, video and music industries, with some mature working solutions already in existence from companies like Macrovision , Microsoft ( and ) and Adobe . Such developments are of interest to the geospatial community in that many geospatial data providers need to control or track who has access to their data and how the data are used. The lack of a GeoDRM capability has been identified as a major barrier to the broader adoption of Web-based geospatial technologies. The mission of OGC GeoDRM Working Group is to coordinate and mature the development and validation of work being done on digital rights management for the geospatial community [27].
Conclusion
Google and MSN's worldwide distribution of "free" geospatial tools, imagery, maps and, eventually, in future versions of their products, analysis capabilities, is to be commended. Building on the powerful and universal visual language of geography, they succeeded in making their customizable multi-purpose maps and imagery of the world familiar and accessible to millions of ordinary Web users around the globe from outside the fields of specialized geosciences. Although ESRI have announced their planned response to Google (and MSN), it remains to be seen how their envisaged plans will materialize and compare to the offerings from Google and MSN, and also how Google and MSN mapping tools will further evolve in the near future.
==== Refs
Schutzberg A KML Gets Two Thumbs Up from File Format Experts Directions Magazine 2005 Jul 23
Blankenhorn D Google map API transforms the Web (ZDNet.com Open Source Blog – 12 Jul 2005)
Dybwad B HOW-TO: Make your own annotated multimedia Google map (Engadget.com – 8 Mar 2005)
SmugMug Maps goes live (O'Reilly Radar – 18 Aug 2005)
Boulos MN Location-based health information services: a new paradigm in personalised information delivery Int J Health Geogr 2003 2 2 12556243 10.1186/1476-072X-2-2
Gottipati H Hacking Maps with the Google Maps API (O'Reilly XML.com – 10 Aug 2005)
Luccio M Partnership To Expand MapMachine GIS Monitor 2005 Jul 28
Roodyn N Creating your first Virtual Earth Web page – Part 1 (Via Virtual Earth – 25 Jul 2005)
Roodyn N Creating your first Virtual Earth Web page – Part 2 (Via Virtual Earth – 26 Jul 2005)
Luccio M Editor's Introduction GIS Monitor 2005 Aug 18
Global Internet Statistics (by Language) (Global Reach – 30 Mar 2004)
Fourth Edition Definition of 'wiki' from the American Heritage® Dictionary of the English Language 2004 Houghton Mifflin Company
Definition of 'blog' from Freedman's Computer Desktop Encyclopedia 2005 Computer Language Company Inc.
Connor A Rewriting the rule books (BBC News WEBLOG WATCH – The Magazine's review of weblogs – 15 Aug 2005)
Thompson B Net offers map help after the flood (BBC News (Technology) – 2 Sep 2005)
Wreden N Case Studies: Wikification in Action (blog entry – 13 Jul 2005)
Boulos MN Towards evidence-based, GIS-driven national spatial health information infrastructure and surveillance services in the United Kingdom Int J Health Geogr 2004 3 1 14748927 10.1186/1476-072X-3-1
Boulos MN Cai Q Padget JA Rushton G Using Software Agents to Preserve Individual Health Data Confidentiality in Micro-scale Geographical Analyses Journal of Biomedical Informatics 2005
Entchev A GIS and privacy Directions Magazine 2005 Mar 24
Francica J The Flow of Free, Location-Aware News in the Age of Terrorism (Directions Media All Points Blog – 18 Jul 2005)
South Korea discusses security concerns with U.S. over Google Earth (Yahoo News/Associated Press – 31 Aug 2005)
Barlow K (for The World Today) Google Earth prompts security fears (ABC (Australian Broadcasting Corporation) News Online – 8 Aug 2005)
Croner CM Public health, GIS, and the Internet Annu Rev Public Health 2003 24 57 82 12543872
Boulos MN Descriptive review of geographic mapping of severe acute respiratory syndrome (SARS) on the Internet Int J Health Geogr 2004 3 2 14748926 10.1186/1476-072X-3-2
Francica J The End of Free Maps (Directions Media All Points Blog – 9 Jun 2005)
Roodyn N Building Your First Commercial Website using Virtual Earth (Via Virtual Earth – 15 Sep 2005)
OGC (Open Geospatial Consortium Inc.) Geo Digital Rights Management (GeoDRM) WG (GeoDRM WG, 2004)
|
16176577
|
PMC1242244
|
CC BY
|
2021-01-04 16:39:06
|
no
|
Int J Health Geogr. 2005 Sep 21; 4:22
|
utf-8
|
Int J Health Geogr
| 2,005 |
10.1186/1476-072X-4-22
|
oa_comm
|
==== Front
J Circadian RhythmsJournal of Circadian Rhythms1740-3391BioMed Central London 1740-3391-3-121616229210.1186/1740-3391-3-12ResearchFailure to respond to endogenous or exogenous melatonin may cause nonphotoresponsiveness in Harlan Sprague Dawley rats Price Matthew Rocco [email protected] Julie Anita Marie [email protected] M Eric [email protected] Annaka M [email protected] Mauricio [email protected] Paul D [email protected] Department of Biology, College of William and Mary, Williamsburg, VA 23187, USA2005 14 9 2005 3 12 12 8 7 2005 14 9 2005 Copyright © 2005 Price et al; licensee BioMed Central Ltd.2005Price et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Responsiveness to changing photoperiods from summer to winter seasons is an important but variable physiological trait in most temperate-zone mammals. Variation may be due to disorders of melatonin secretion or excretion, or to differences in physiological responses to similar patterns of melatonin secretion and excretion. One potential cause of nonphotoresponsiveness is a failure to secrete or metabolize melatonin in a pattern that reflects photoperiod length.
Methods
This study was performed to test whether a strongly photoresponsive rat strain (F344) and strongly nonphotoresponsive rat strain (HSD) have similar circadian urinary excretion profiles of the major metabolite of melatonin, 6-sulfatoxymelatonin (aMT6s), in long-day (L:D 16:8) and short-day (L:D 8:16) photoperiods. The question of whether young male HSD rats would have reproductive responses to constant dark or to supplemental melatonin injections was also tested. Urinary 24-hour aMT6s profiles were measured under L:D 8:16 and L:D 16:8 in young male laboratory rats of a strain known to be reproductively responsive to the short-day photoperiod (F344) and another known to be nonresponsive (HSD).
Results
Both strains exhibited nocturnal rises and diurnal falls in aMT6s excretion during both photoperiods, and the duration of the both strains' nocturnal rise was longer in short photoperiod treatments. In other experiments, young HSD rats failed to suppress reproduction or reduce body weight in response to either constant dark or twice-daily supplemental melatonin injections.
Conclusion
The results suggest that HSD rats may be nonphotoresponsive because their reproductive system and regulatory system for body mass are unresponsive to melatonin.
==== Body
Introduction
Responsiveness of the reproductive system, metabolic rate, and other traits to changing photoperiods from summer to winter seasons is an important physiological trait in most temperate-zone mammals [1]. Seasonal changes in photoperiod, or day length, modify reproductive timing in many temperate-zone mammals including sheep, hamsters, rodents, horses, and ferrets by acting through the photoperiod pathway [2-4]. The photoperiod pathway transduces the photoperiod into a physiological signal beginning with the transduction of light or dark input from specialized photoreceptors and ganglion cells in the eye through the retinohypothalamic tract into two regions of the hypothalamus, the suprachiasmatic nucleus (SCN), and later the paraventricular nucleus. A sympathetic norepinephrine signal from the SCN then passes to the hindbrain, the superior cervical ganglion in the spinal cord, and eventually to the pineal gland, which releases the indoleamine hormone melatonin [5]. Pinealoctyes within the pineal gland convert tryptophan into 5-hydroxytryptamine (serotonin), acetylate serotonin into N-acetylserotonin (NAT), and finally methylate NAT with the enzyme hydroxyindole-O-methyltransferase to form melatonin (N-acetyl-5-methoxytryptamine) [5]. In the presence of light, inhibition of NAT enzyme activity reduces melatonin synthesis, and thus melatonin is secreted from pinealocytes primarily in darkness. The duration of elevated melatonin provides a physiological signal for photoperiods [6]. Melatonin binds to one or more receptor types, MT1 or MT2, initiating cellular responses that apparently produce the physiological effects of this hormone [7,8].
The photoperiod pathway is crucial for regulation of seasonal function in most temperate zone animals [1] However, there is genetic variation in photoresponsiveness within and among species of rodents [9]. It has been proposed that this variation is likely to be important in animal function and evolution [9,10]. With respect to humans, there is debate over the function of the photoperiod pathway [11,12]. Recent reviews suggest that genetic variation in the pathway may have functional and medical significance in humans [13,14]. Thus, identifying the physiological basis of and consequences from genetic variation in photoperiodic responses may be useful in understanding mammalian variation in this trait, with potential relevance to humans as well. A potential cause of nonphotoresponsiveness is a failure to secrete or metabolize melatonin in a pattern that reflects photoperiod length. Such variation occurs in humans, and the clinical significance of atypically elevated or depressed melatonin levels is widely recognized in human sleep disturbances and clinical conditions [reviewed in [14,15]]. Reduced amplitude and duration of nocturnally elevated melatonin is characteristic of a wide range of psychiatric disorders, including major depression and bipolar affective disorder [16].
Patterns of melatonin secretion can be estimated by the pattern of excretion of the primary metabolite of melatonin, 6-sulfatoxymelatonin (aMT6s) [17-19]. After synthesis, melatonin is rapidly metabolized in the liver and kidney by hydroxylation and subsequent sulfonation to produce aMT6s for later excretion in urine [18,19]. Because of the relatively rapid conversion of melatonin, it has been argued that melatonin secretion patterns are related to the amount of aMT6s present in urine, and aMT6s has been used as an indirect estimator of periods of elevated circulating melatonin [5]. However, this estimate can be imprecise because some melatonin is metabolized by other pathways, the conversion rate to aMT6s may vary genetically, and urine may be held in the bladder for some time before micturition.
Laboratory rats vary genetically in their responses to short-day photoperiods (eight hours light, 16 hours dark; SD). Some strains are functionally non-photoperiodic [2,20], including Sprague Dawley rats from Harlan USA (HSD) [21], though such strains are sometimes reproductively photoresponsive if a short photoperiod is combined with secondary cues such as food restriction, testosterone treatment, or olfactory bulbectomy [2,22]. In contrast, many other strains, including Fisher 344 (F344), Brown Norway (BN), ACI, BUF, and PVG inbred rat strains, are robustly reproductively photoresponsive, thereby demonstrating the presence of rat inter-strain variation in physiological and reproductive responses to short photoperiods [23-25]. Exposure to short photoperiods alone causes changes in F344 and BN reproductive organ size, food intake, and body weight [26]. Even stronger responses occur when food restriction or neonatal testosterone treatment is combined with short photoperiod treatment [21,24].
In the present study, tests were performed to find out whether the aMT6s urinary excretion pattern would vary between short and long photoperiods in young photoresponsive F344 and nonphotoresponsive HSD rats. We chose these two strains because young F344 rats have the greatest response to short photoperiod reported in rats, and HSD rats are the only strain for which there is clear evidence for a lack of response to short photoperiod [21,23,27]. In order to further examine the effects of photoperiod on melatonin, the question of whether young HSD rats would exhibit inhibition of reproductive development in response to constant dark or supplemental timed injections of melatonin was tested. As a photoperiodic strain, it was predicted that young F344 rats would have nocturnally elevated aMT6s, and that the duration of elevation would be longer in short photoperiods. Because non-manipulated young HSD rats are not photoperiodic [23], it was hypothesized that young HSD rats might lack nocturnally elevated aMT6s as an underlying cause of their nonphotoresponsiveness, or that any rise in aMT6s would not differ between long and short photoperiods in non-manipulated individuals. It was also hypothesized that if melatonin secretion was inadequate, low, or absent in young HSD rats, supplemental melatonin or constant dark might suppress reproductive development. An alternative hypothesis is that young HSD rats are normally nonresponsive not because of deficiencies in the pattern of nocturnally elevated melatonin, but because of a lack of response to short-day patterns of elevated melatonin. Under the alternative hypothesis, it was predicted that both strains would produce a nocturnal rise in aMT6s excretion and differences between long and short photoperiods in aMT6s excretion.
Methods
Experiment 1. aMT6s Excretion Patterns in F344 and HSD rats
This experiment used a 2 × 2 design with HSD and F344 rats in short-day (L8:D16; lights on at 0900; SD) and long-day (L16:D8; lights on at 0500; LD) photoperiods (n = 12 rats/treatment group). Breeder rats of the inbred Fischer F344 NHsd and outbred HSD strains from Harlan Sprague Dawley (Indianapolis, IN) were bred in polypropylene cages in LD photoperiod (40 × 23 × 23 cm) with stainless-steel wire tops and bedding of pine shavings. Harlan Teklad rodent diet (Indianapolis, IN) and tap water were provided ad libitum. Relative humidity was 40–65%, and temperature was maintained at 23 ± 3°C. Due to bright light's ability to cause retinal damage to albino rats, light intensity was maintained between 100 and 300 lux, as measured five cm above the cage floor. After weaning at age 21–24 days in LD, twelve young rats from each strain were transferred to SD, while twelve rats from each strain remained in LD. All were housed individually in polypropylene cages (33 × 20 × 20 cm). To avoid inconsistencies in aMT6s secretion due to the estrus cycles of female rats [28], only male rats were used in this study.
At age 7 to 8 weeks (± 3 days), when F344 rats are highly photoperiodic but HSD rats are not [23], rats were transferred to hanging cages (27 × 20 × 20 cm) with wire mesh bottoms and funnels to collect urine. Rats were given ad libitum tap water and fed a liquid diet reported to be complete for rats (Osmolite HN, Ross Laboratories, Columbus, OH) to stimulate urine secretion [17]. Lighting remained as above. Rats were then given 3 to 4 days to acclimate to cage and diet changes. At 15-minute sampling intervals over two consecutive 24-hour periods, urine was automatically collected (Eldex Universal Fraction Collector, Eldex Laboratories, Inc., Napa, CA). After each of the two 24-hour collection periods, each sample was weighed to determine urinary output volume, and samples were stored at -20°C. Concentration and volume changes due to evaporation over the collection period were corrected against a water evaporation control for each day of collection. Groups of eight successive 15-minute samples were combined to create two-hour sample periods, covering periods beginning at 0100, 0300, 0500, 0700, 0900, 1100, 1300, 1500, 1700, 1900, 2100, and 2300 hours. Finally, because pilot studies indicated that single 24-hour periods were missing urine samples from some two-hour collection periods from some animals, corresponding samples from the same time periods in the first and second days of collection were combined. The result produced urine samples from periods two hours in duration on successive nights from the same time period, with 12 such two-hour sample periods per individual. Urine samples were assayed for aMT6s with a 6-sulfatoxymelatonin ELISA kit (Buhlmann Laboratories, Allschwil, Switzerland) according to the manufacturer's protocol. Inter-assay coefficient of variation (CV) was 17% and intra-assay CV was 10% for standards near the midrange of values in this study. Data analysis treated each two-hour sampling interval as a single data point.
Experiment 2. Effects of Constant Dark on HSD Rats
This experiment tested whether constant dark might provide a physiological signal that would suppress reproduction (as measured by gonad or seminal vesicle size) or body mass in HSD rats. HSD rats were raised until weaning at age 21 days in LD. At that time, one group of rats was transferred to SD (n = 13), and another group to constant dark (n = 11). After four weeks of treatment, rats were euthanized and body mass, paired testis mass, and paired seminal vesicle mass (emptied of fluid contents) were recorded.
Experiment 3. Effects of Supplemental Melatonin on HSD Rats
This experiment tested whether supplemental melatonin might provide a physiological signal that would suppress reproduction (as measured by gonad or seminal vesicle size) or body mass in HSD rats. HSD rats were raised until weaning at age 21 days in LD. At that time, all rats were transferred to SD (lights on at 0900 h and lights out at 1700 h). For the following four weeks, one group (n = 24) was given S.C. injections of melatonin twice daily (100 μg of melatonin dissolved in 0.1 ml of 10% ethanol and 90% physiological saline), and a control group (n = 23) was injected with ethanolic saline vehicle. Injections were given twice daily at 1230 and 1500 hours. Single injections of this amount of melatonin at 1500 hours in SD suppressed reproduction and inhibited growth in F344 rats [27]. The injection at 1230 hours was included in this experiment because pilot data suggested a single injection did not affect young HSD rats. After four weeks of treatment, rats were euthanized and body mass, paired testis mass, and paired seminal vesicle mass (emptied of fluid contents) was recorded.
Data Analysis
In statistical testing of data on aMT6s, the data from each strain was analyzed independently for nocturnally elevated aMT6s excretion and for differences in excretion between SD and LD. Variation in mean aMT6S was assessed with ANOVA (Statview 4.5), with photoperiod as the factor. Comparisons for equality of variance indicated no significant differences in variance between photoperiods or between strains. The researchers conducted a final set of analyses comparing the two strains, with both photoperiod and strain as factors, to test for clear differences between strains that might be related to photoresponsiveness. The strain comparison was considered statistically appropriate because this experiment was testing a prediction derived from other information that HSD rats would be different in an estimator of melatonin rhythms.
Unpaired t-tests were used to compare effects of constant dark or supplemental melatonin on body mass, testis mass, and seminal vesicle mass in experiments two and three.
All procedures were conducted in accordance with the Guide for Care and Use of Laboratory Animals and approved by the Research on Animal Subjects Committee (RASC) of the College of William and Mary.
Results
Experiment 1. aMT6s Excretion Patterns in F344 and HSD rats
F344 rats excreted significantly more total aMT6s than HSD rats (F = 4.22, P < 0.05, n = 24 for each strain; Fig. 1). Because F344 rats at these ages are 30% lighter in weight than HSD rats at the ages tested in this experiment [unpublished data and [23]], differences in excretion would be even more pronounced if expressed as excretion per unit body mass, with F344 rats excreting approximately 40% more aMT6s per unit body weight than HSD rats. Total aMT6s excretion did not differ significantly between SD and LD (F = 1.63, P = 0.21). There was a diurnal pattern of aMT6s excretion in both strains, with the lowest levels near the middle of the light period and the highest levels near the middle of the dark period (Fig. 2).
Figure 1 Total urinary 6-sulfatoxymelatonin production in ng per 24 h for F344 and HSD rats. Asterisk indicates P < 0.05. For each strain, n = 24 rats.
Figure 2 24-hour urinary 6-sulfatoxymelatonin excretion rhythms (ng/2 h) for F344 and HSD rats in SD (upper panel) and LD (lower panel). Values shown are means +/- SEM. Bars at the top of the figure indicate periods of light and dark for the SD and LD treatments, respectively. For each treatment group, n = 12 rats.
In SD, the pattern of excretion of aMT6s was very similar for the two strains of rats (Fig. 2). Excretion of aMT6s began rising in the collection period beginning at 9:00 pm, four hours after the onset of dark. Levels of aMT6s remained elevated, relative to the light period, through the remaining five collection periods of the dark period. The duration of excretion of aMT6s did not differ significantly between the two strains during the SD dark period (Repeated Measures ANOVA, F = 0.89, P = 0.35).
In LD, the pattern of excretion of aMT6s differed between the strains of rats (Fig. 2). Across the total dark period, there was an insignificant statistical trend (Repeated Measures ANOVA, F = 2.79, P = 0.098) for a higher level of excretion of aMT6s in F344 rats than in HSD rats. In the two collection periods immediately after the end of the dark period, aMT6s excretion was significantly higher in F344 rats than in HSD rats (Repeated Measures ANOVA, F = 12.22, P < 0.001).
In both strains of rats, aMT6s excretion was elevated for a longer duration in SD than in LD, but this difference between photoperiods was more pronounced in HSD rats (Fig. 2). In both strains, aMT6s excretion was significantly higher in SD than in LD in the 0500 and 0700 collection periods (F344: Repeated Measures ANOVA, F = 4.20, P < 0.05; HSD: Repeated Measures ANOVA, F = 11.64, P < 0.001). In the collection periods beginning at 5:00 pm or 7:00 pm for each strain, aMT6s excretion was low in both SD and LD (Fig. 2). Finally, unlike the case for F344 rats, HSD rats in the collection period beginning at 9:00 pm excreted lower levels of aMT6s in LD than in SD (Fig. 2; F = 5.02, P = 0.03).
Some HSD rats either lacked a clear diurnal pattern of aMT6s excretion or had a very low amplitude nocturnal rise. In contrast, all F344 rats had a clear diurnal pattern of aMT6s with a robust nocturnal rise in aMT6s excretion. For example, the two F344 rats in SD and LD with the lowest total aMT6s excretion for their treatment groups nonetheless had a robust nocturnal rise in aMT6s excretion (Fig. 3). In contrast, the two HSD rats in SD and LD with the lowest total aMT6s excretion for their treatment groups had poorly developed rhythms of aMT6s excretion (Fig. 3).
Figure 3 24-hour urinary 6-sulfatoxymelatonin excretion rhythms (ng/2 h) for two individual F344 rats (one in SD and one in LD) and two individual HSD rats (one in SD and one in LD). Bars at the top of the figure indicate periods of light and dark for the SD (upper panel) and LD (lower panel) treatments. The four rats selected for presentation were those with the lowest total 6-sulfatoxymelatonin excretion in their respective treatment groups.
Experiment 2. Effects of Constant Dark on HSD Rats
HSD rats held in constant darkness for four weeks following weaning did not differ from SD controls in body mass, testis mass, or seminal vesicle mass (Fig. 4, P > 0.10 for all).
Figure 4 Mean (+/- SEM) of body mass (a), paired testis mass (b), and paired seminal vesicle mass (c) of young HSD rats held in SD or constant dark (24D). Sample sizes: n = 13 in SD and 11 in 24D. NS indicates a lack of significant differences.
Experiment 3. Effects of Supplemental Melatonin on HSD Rats
HSD rats given twice daily injections of melatonin for four weeks did not differ from saline controls in body mass, testis mass, or seminal vesicle mass (Fig. 5, P > 0.10 for all).
Figure 5 Mean (+/- SEM) of body mass (a), paired testis mass (b), and paired seminal vesicle mass (c) of young HSD rats in SD treated with saline injections (Sal) or melatonin (Mel). Sample sizes: n = 23 in Sal and 24 in Mel. NS indicates a lack of significant differences.
Discussion
Both strains of rats were found to have generally higher levels of excretion in the dark period than in the light period (Fig. 2), and both had a longer duration of nocturnally elevated aMT6s excretion in SD than in LD. These differences between SD and LD were as apparent in HSD rats as in F344 rats (Fig. 2). This suggests that, based on the pattern of aMT6s excretion, both HSD rats and F344 rats should be able to use melatonin secretion as a physiological signal to distinguish SD and LD. The data for young HSD rats on the nocturnal rise of aMT6s excretion and approximate amounts of aMT6s excreted per hour are consistent with nocturnal rises in L12:D12 reported by Usui and colleagues [29] on older Sprague Dawley rats from a different source (Clea Japan, Tokyo). However, in a few HSD rats in this study the pattern of aMT6s excretion lacked a clear nocturnal rise or had only a slight nocturnal rise (Fig. 3). In HSD rats, neither four weeks of constant darkness nor four weeks of supplemental melatonin affected body mass or suppressed reproduction (Figs. 4 and 5). In previous tests on young F344 rats at the same age, four weeks of short photoperiod treatment suppressed reproductive development and somatic growth. Relative to rats in LD, testis mass in SD was lower by about 50%, seminal vesicle mass in SD was lower by 80%, and body mass in SD was lower by 10–20% [21,23,30]. Pinealectomy blocked effects of SD [23], and four weeks of melatonin injections in LD caused reproductive suppression, reduced body mass, and also enhanced the suppressive effects of SD in short days [27].
These results suggest that there is a nocturnal rise in nocturnal melatonin in both young HSD and young F344 rats and a difference in both strains between SD and LD (Fig. 2), but only young HSD rats fail to respond to changes in photoperiod and to exogenous melatonin. While there were significant statistical differences between strains, the differences were small and may not reflect differences in serum melatonin levels. In contrast, there is previous evidence that in F344 rats, the normal endogenous melatonin signal does not produce a maximal response to short photoperiods. Exogenous melatonin delivered to young F344 rats in SD as S.C. injections before the dark period resulted in greater reproductive inhibition and lower body weight than SD alone [27]. In this study, the presence of nocturnal rises in excretion of aMT6s and differences between SD and LD patterns for both strains, along with evidence for a failure of HSD rats to respond to supplemental melatonin, is consistent with the alternative hypothesis, which says that differences in photoresponsiveness arise from inter-strain differences in physiological mechanisms responsible for processing the melatonin signal, rather than from inadequate melatonin secretion. In a previous comparison of young rats of these two strains [31], there was an up to 2.5-fold higher specific binding of iodomelatonin in the brains of young F344 rats than young HSD rats. Significant differences between HSD and F344 rats were found in the thalamic paraventricular nucleus and reunions nucleus, but not in some other brain areas, including the SCN. This suggests that the response to melatonin signals might be different in HSD and F344 rats, even if those melatonin signals were identical.
Young F344 rats excreted 25% more aMT6s than same-age HSD rats over two-day collection periods (Fig. 1), despite body weights that are approximately 30% lower at this age. This suggests that young HSD rats either secrete less melatonin than F344 rats or excrete a higher amount of melatonin and its metabolites through an alternative pathway (e.g., via the feces). The biological significance of this difference is not clear. However, it is possible that the small number of HSD rats that had little diurnal change in aMT6s (Fig. 3) may have too small a nocturnal rise in melatonin secretion for consistent responses to melatonin.
As in previous aMT6s studies in laboratory species [9,17,32-34] and human populations exposed to different photoperiods [35], substantial differences among individuals in amplitude and total excretion amount were observed within all four groups (e.g., Fig. 3). While some of this variation might be due to variation in urination pattern, the variation in total amount of aMT6s excreted should be only slightly affected by variation in urination of rats on a liquid diet. Due to the fact that inbred F344 rats are highly genetically similar, this suggests substantial environmental influences on melatonin secretion patterns, even in a highly controlled laboratory environment.
Variation in melatonin receptor number, density, or location have been implicated as potential sources of variation in this pathway in other species [31,36]. Differences in photoresponsiveness might also be attributable to variation in neurotransmitter systems mediating reproductive responses to melatonin, including negative feedback sensitivity to sex steroids or the influence of additional cues, such as food intake [9,27]. This is consistent with the suggestion that clinically significant circadian dysfunction in humans may occur downstream of melatonin production, or that both downstream as well as upstream processing dysfunction could occur concurrently with melatonin production dysfunction [37].
Conclusion
Both strains of rats in both photoperiods exhibited nocturnal rises and diurnal falls in aMT6s excretion, and the duration of the nocturnal rise was longer in short photoperiod treatments in both. In addition, young HSD rats failed to suppress reproduction or reduce body weight in response to either constant darkness or twice-daily supplemental melatonin injections. In combination, these results suggest that HSD rats may be nonphotoresponsive because their reproductive system and the regulatory system for body mass are unresponsive to melatonin.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MEG designed and conducted pilot experiments on sulfatoxymelatonin excretion and contributed to text on Experiment 1.
MRP and JAMK designed and conducted experiment 1; MRP carried out the assays and final analysis, and had the lead role in writing and revising the manuscript.
AML and MA designed and conducted Experiments 2 and 3, and AML conducted the data analyses and wrote text for Experiments 2 and 3.
PDH supervised the experiments and analysis, and finalized figures and text.
Acknowledgements
We thank E. Hartman, K. Johal, P. Lowman, and M. Park for assistance with data collection, L. Moore for assistance with animal care, and C. D. Jenkins for suggestions and comments. Research was supported by NIH Grant R15 MH62402-01 to PDH, by a Beckman Fellowship to AML, and by a Minor Research Grant and Summer Research Fellowship to MEG from a Howard Hughes Medical Institute Undergraduate Biological Sciences Education Program grant to the College of William & Mary.
==== Refs
Bronson FH Heideman PD Knobil E and Neill JD Seasonal regulation of reproduction in mammals The Physiology of Reproduction 2nd Edition 1994 2 Second New York, Raven Press 541 584
Wallen EP DeRosch MA Thebert A Losee-Olson S Turek FW Photoperiodic response in the male laboratory rat Biol Reprod 1987 37 22 27 3651546
Guerin MV Deed JR Kennaway DJ Mathews CD Plasma melatonin in the horse: measurements in natural photoperiod and in acutely extended darkness throughout the year J Pineal Res 1995 19 7 15 8609596
Goldman BD Mammalian photoperiodic system: formal properties and neuroendocrine mechanisms of photoperiodic time measurement J Biol Rhythms 2001 16 283 301 11506375 10.1177/074873001129001980
Arendt J Melatonin and the pineal gland: influence on mammalian seasonal and circadian physiology Rev Reprod 1998 3 13 22 9509985 10.1530/ror.0.0030013
Bartness TJ Powers JB Hastings MH Bittman EL Goldman BD The timed infusion paradigm for melatonin delivery: What has it taught us about the melatonin signal, its reception, and the photoperiodic control of seasonal responses? J Pineal Res 1993 15 161 190 8120796
Vanecek J Cellular mechanisms of melatonin action Physiol Rev 1998 78 687 721 9674691
von Gall C Stehle JH Weaver DR Mammalian melatonin receptors: molecular biology and signal transduction Cell Tissue Res 2002 309 151 162 12111545 10.1007/s00441-002-0581-4
Prendergast BJ Kriegsfeld LJ Nelson RJ Photoperiodic polyphenisms in rodents: Neuroendocrine mechanisms, costs and functions Quart Rev Biol 2001 76 293 325 11561508 10.1086/393989
Bittner GD Friedman BX Evolution of brain structures and adaptive behaviors in humans and other animals: Role of polymorphic genetic variations The Neuroscientist 2000 6 241 251
Bronson FH Seasonal variation in human reproduction: Environmental factors Quart Rev Biol 1995 70 141 164 7610233 10.1086/418980
Arendt J Is melatonin a photoperiodic signal in humans Adv Exp Med Biol 1999 460 417 424 10810543
Bronson FH Are humans seasonally photoperiodic? J Biol Rhythms 2004 19 180 192 15155003 10.1177/0748730404264658
Wehr TA Melatonin and Seasonal Rhythms J Biol Rhythms 1997 12 518 527 9406025
Arendt J Importance and relevance of melatonin to human biological rhythms J Neuroendocrinol 2003 15 427 431 12622845 10.1046/j.1365-2826.2003.00987.x
Kennedy SH Kutcher SP Ralevski E Brown GM Nocturnal melatonin and 24-hour 6-sulphatoxymelatonin levels in various phases of bipolar affective disorder Psychiatry Res 1996 63 219 222 8878318 10.1016/0165-1781(96)02910-1
Kennaway DJ Urinary 6-sulphatoxymelatonin excretory rhythms in laboratory rats--Effects of photoperiod and light Brain Res 1993 603 338 342 8461986 10.1016/0006-8993(93)91259-U
Yellon SM Daily melatonin treatments regulate the circadian melatonin rhythm in the adult Djungarian hamster J Biol Rhythms 1996 11 4 13 8695891
Karasek M Szuflet A Chrzanowski W Zylinska K Swietoslawski J Circadian serum melatonin profiles in patients suffering from chronic renal failure Neuroendocrinology Letters Supplement 1 2002 23 97 102
Nelson RJ Moffatt CA Goldman BD Reproductive and nonreproductive responsiveness to photoperiod in laboratory rats J Pineal Res 1994 17 123 131 7897584
Heideman PD Deibler RW York LM Food and neonatal androgen interact with photoperiod to inhibit reproductive maturation in Fischer 344 rats Biol Reprod 1998 59 358 363 9687308
Wallen EP Turek FW Photoperiodicity in the male albino laboratory rat Nature 1981 289 402 404 6780913 10.1038/289402a0
Heideman PD Sylvester CJ Reproductive photoresponsiveness in unmanipulated Fischer 344 laboratory rats Biol Reprod 1997 57 134 138 9209091
Lorincz AM Shoemaker MB Heideman PD Genetic variation in photoperiodism among naturally photoperiodic rat strains Am J Physiol Regul Integr Comp Physiol 2001 281 R1817 R1824 11705766
Francisco NR Raymond CM Heideman PD Short photoperiod inhibition of growth in body mass and reproduction in ACI, BUF, and PVG inbred rats Reproduction 2004 128 857 862 15579603 10.1530/rep.1.00390
Shoemaker MB Heideman PD Reduced body mass, food intake, and testis size in response to short photoperiod in adult F344 rats BMC Physiology 2002 2 11 12135532 10.1186/1472-6793-2-11
Heideman PD Bierl CK Sylvester CJ Photoresponsive Fischer 344 rats are reproductively inhibited by melatonin and differ in 2-[125I]iodomelatonin binding from non-photoresponsive Sprague Dawley rats J Neuroendocrinol 2001 13 223 232 11207936 10.1046/j.1365-2826.2001.00617.x
White RM Kennaway DJ Seamark RF Estrogenic effects on urinary 6-sulphatoxymelatonin excretion in the female rat J Pineal Res 1997 22 124 129 9213265
Usui S Okazaki T Honda Y Interruption of the rat circadian clock by short light-dark cycles Am J Physiol Regul Integr Comp Physiol 2003 284 R1255 R1259 12676747
Heideman PD Bierl CK Galvez ME Inhibition of reproductive maturation and somatic growth of Fischer 344 rats by photoperiods shorter than L14:D10 and by gradually decreasing photoperiod Biol Reprod 2000 63 1525 1530 11058561
Heideman PD Kane SL Goodnight AL Differences in hypothalamic 2-[125I]iodomelatonin binding in photoresponsive and non-photoresponsive white-footed mice, Peromyscus leucopus Brain Res 1999 840 56 64 10517952 10.1016/S0006-8993(99)01746-1
Stieglitz A Spiegelhalter F Klante G Heldmaier G Urinary 6-sulphatoxymelatonin excretion reflects pineal melatonin secretion in the Djungarian hamster (Phodopus sungorus) J Pineal Res 1995 18 69 76 7629693
Niehaus M Lerchl A Urinary 6-sulfatoxymelatonin profiles in male Djungarian hamsters (Phodopus sungorus) responding and not responding to short-day photoperiods; possible role of elevated daytime levels J Pineal Res 1998 25 167 171 9745985
Allain D Malpaux B Puechal F Thebault RG De Rochambeau H Chemineau P Genetic variability of the pattern of night melatonin blood levels in relation to coat changes development in rabbits Genet Sel Evol 2004 36 207 216 15040899 10.1051/gse:2003059
Vangelova KK Dalbokova DL Variations in 6-sulphatoxymelatonin excretion and oral temperature under a 12-hour shiftwork environment Rev Environ Health 1998 13 221 226 9987817
Weaver DR Carlson LL Reppert SM Melatonin receptors and signal transduction in melatonin-sensitive and melatonin-insensitive populations of white-footed mice (Peromyscus leucopus) Brain Res 1990 506 353 357 2154292 10.1016/0006-8993(90)91280-T
Weaver DR Stehle JH Stopa EG Reppert SM Melatonin receptors in human hypothalamus and pituitary: implications for circadian and reproductive responses to melatonin J Clin Endocrinol Metab 1993 76 295 301 8381796 10.1210/jc.76.2.295
|
16162292
|
PMC1242245
|
CC BY
|
2021-01-04 16:39:13
|
no
|
J Circadian Rhythms. 2005 Sep 14; 3:12
|
utf-8
|
J Circadian Rhythms
| 2,005 |
10.1186/1740-3391-3-12
|
oa_comm
|
==== Front
J NeuroinflammationJournal of Neuroinflammation1742-2094BioMed Central London 1742-2094-2-191614455210.1186/1742-2094-2-19ResearchSignaling pathways mediating a selective induction of nitric oxide synthase II by tumor necrosis factor alpha in nerve growth factor-responsive cells Thomas Michael S [email protected] WenRu [email protected] Paivi M [email protected] H Uri [email protected] Giulio [email protected] Department of Neuroscience and Cell Biology, the University of Texas Medical Branch at Galveston, Texas - USA2 Department of Pharmacology and Therapeutics, McGill University, Montreal, QC, Canada2005 6 9 2005 2 19 19 10 3 2005 6 9 2005 Copyright © 2005 Thomas et al; licensee BioMed Central Ltd.2005Thomas et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Inflammation and oxidative stress play a critical role in neurodegeneration associated with acute and chronic insults of the nervous system. Notably, affected neurons are often responsive to and dependent on trophic factors such as nerve growth factor (NGF). We previously showed in NGF-responsive PC12 cells that tumor necrosis factor alpha (TNFα) and NGF synergistically induce the expression of the free-radical producing enzyme inducible nitric oxide synthase (iNOS). We proposed that NGF-responsive neurons might be selectively exposed to iNOS-mediated oxidative damage as a consequence of elevated TNFα levels. With the aim of identifying possible therapeutic targets, in the present study we investigated the signaling pathways involved in NGF/TNFα-promoted iNOS induction.
Methods
Western blotting, RT-PCR, transcription factor-specific reporter gene systems, mutant cells lacking the low affinity p75NTR NGF receptor and transfections of TNFα/NGF chimeric receptors were used to investigate signalling events associated with NGF/TNFα-promoted iNOS induction in PC12 cells.
Results
Our results show that iNOS expression resulting from NGF/TNFα combined treatment can be elicited in PC12 cells. Mutant PC12 cells lacking p75NTR did not respond, suggesting that p75NTR is required to mediate iNOS expression. Furthermore, cells transfected with chimeric TNFα/NGF receptors demonstrated that the simultaneous presence of both p75NTR and TrkA signaling is necessary to synergize with TNFα to mediate iNOS expression. Lastly, our data show that NGF/TNFα-promoted iNOS induction requires activation of the transcription factor nuclear factor kappa B (NF-κB).
Conclusion
Collectively, our in vitro model suggests that cells bearing both the high and low affinity NGF receptors may display increased sensitivity to TNFα in terms of iNOS expression and therefore be selectively at risk during acute (e.g. neurotrauma) or chronic (e.g. neurodegenerative diseases) conditions where high levels of pro-inflammatory cytokines in the nervous system occur pathologically. Our results also suggest that modulation of NFκB-promoted transcription of selective genes could serve as a potential therapeutic target to prevent neuroinflammation-induced neuronal damage.
==== Body
Background
Neuroinflammation is thought to play a prominent role in neurodegeneration associated with a variety of acute and chronic insults in both the central (CNS) and peripheral (PNS) nervous system [1,2]. Examples of neurotraumatic or neurodegenerative conditions where the occurrence or role of neuroinflammation has been documented include peripheral nerve injury [3-6], acute and chronic spinal cord injury [7-11], traumatic brain injury [12-14], stroke [15-17], amyotrophic lateral sclerosis (ALS, [18-20] and Alzheimer Disease (AD, [21-24].
Neurons susceptible to neuroinflammatory insults are often dependent for their survival on target derived neurotrophic factors such as nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF) or glia-derived neurotrophic factor (GDNF). The same neurodegenerative conditions have also been associated with the presence of damaging high levels of free radical species leading to pathological oxidative stress [25]. For example, inflammatory involvement in AD pathogenesis has been proposed partly based on observations of increased levels of the pro-inflammatory cytokines tumor necrosis factor alpha (TNFα) and interleukin-1 beta (IL-1β) in cerebrospinal fluid and brain cortex of AD patients [26,27]. Additionally, among the most affected neurons in AD are the basal forebrain cholinergic neurons (BFCN, [28-30]), which rely upon trophic support by target-derived NGF [31,32]. Furthermore, there is strong evidence for the presence of oxidative damage in the AD brain [33-36]. Similarly, neuronal damage following acute spinal cord injury or peripheral nerve injury has been shown to involve a neuroinflammatory as well as oxidative stress component [1,8,10,11,37-39], and traumatic head injury is also known to be associated with increased circulating concentrations of inflammatory cytokines and reduced numbers of basal forebrain cholinergic neurons [13,40-42].
Thus, there seems to be an intimate relationship between pro-inflammatory cytokines, oxidative stress and trophic factors that underscores the neuropathological consequences of extrinsic (e.g. traumatic) or intrinsic (e.g. disease-related) injury to the nervous system. Our previous work has shown that in NGF-responsive rat pheochromocytoma (PC12) cells TNFα induces expression of the free radical nitric oxide (NO) synthesizing enzyme NOS II (iNOS) only in the presence of NGF acting through its high affinity receptor TrkA [43]. Indeed, perturbed levels of NOS and NO-derived oxidative damage have been reported in both acute and chronic neurodegenerative conditions [25], including spinal cord injury [44-46], stroke [47,48] and AD [49-53]. However, TNFα alone has not been shown to be an effective inducer of human iNOS promoter activity [54] or of rat cortical iNOS expression when administered intracerebroventricularly [55]. Nonetheless, TNFα has been shown to contribute to the death of NGF-dependent neurons in vitro [56] and in vivo [57,58]. Therefore, our previous results suggest the attractive idea that one mechanism through which increased levels of TNFα affect certain trophic factor-responsive neurons may involve NO-derived oxidative damage brought about by a synergistic induction of iNOS. Understanding the molecular mechanisms mediating the synergistic NGF/TNFα-promoted induction of iNOS may thus provide novel therapeutic targets for the prevention of certain neurodegenerative events associated with acute or chronic injury of the nervous system.
Here we report that a reversible expression of iNOS, produced in PC12 cells by simultaneous exposure to NGF and TNFα, requires the simultaneous presence of both the low-affinity p75NTR and the high-affinity TrkA NGF receptors. Furthermore, using specific inhibitors and a reporter gene assay, we show that such synergistic effect of the combined NGF/TNFα treatment is mediated by the transcription factor nuclear factor kappa B (NF-κB).
Methods
Materials
All routine reagents and chemicals were obtained from Sigma-Aldrich (St Louis, MO, USA), except where noted otherwise. Recombinant human and rat TNF and rat IGF were obtained from R&D Systems, Minneapolis, MN, USA, purified mouse NGF from Harlan Bioproducts, Indianapolis, IN, USA, and pyrrolidine dithiocarmbamate (PDTC), the octapeptide proteasome inhibitor (PSI), PD98059, K252a and 1400 W from Calbiochem, San Diego, CA, USA.
Clonal cell lines
Stock cultures of rat pheochromocytoma cells (PC12; a kind gift of Dr. Lloyd Greene, Columbia University, New York, NY, USA) and PC12 cells lacking the low affinity p75NTR NGF receptor were maintained in 75 cm2 tissue culture flasks in 10 ml RPMI-1640 culture medium supplemented with 5% heat inactivated fetal bovine serum in a humidified cell incubator at 37°C kept at a 5% CO2 atmosphere. Half of the medium was replaced every other day and the cells were split once a week to maintain cell viability.
Expression vectors
Transient transfection of cells was performed by a liposomal packaging system. Briefly, 1.2 pmol of expression vector were mixed with DMRIE-C (Life Technologies, Carlsbad, CA, USA) in a 1:3 DNA to liposome ratio. The DNA/liposomes were diluted in 400 μl serum free transfection medium (Optimem) and then added to approximately 100,000 cells in a 12 well cell culture plate. The cells were allowed to take up the liposomal DNA for 3 hours before being washed and returned to cell culture medium. Cells were allowed to recover for 24 hours before any treatments. The cDNA coding for chimeric proteins bearing the extracellular domain of the TNFR1 receptor and the transmembrane and cytosolic domains of the NGF receptors (either p75NTR or TrkA) was a kind gift from Dr. Eric Shooter and prepared as described [77], (Stanford University, Palo Alto, Ca, USA). The p-SEAP expression vector, containing the SEAP gene under NF-kB, AP1 or CRE enhancer control, was purchased from Clontech (Palo Alto, CA, USA). Conditioned medium from cells transfected with the SEAP reporter vectors was assayed for alkaline phosphatase by sampling the medium and using the chemiluminescent Great EscAPe SEAP assay (Clontech, Palo Alto, CA, USA), according to manufacturer's instructions.
Western blot analysis
Cells were lysed using an SDS-based lysis buffer (2% SDS, 5 mM EDTA, 50 mM Tris, 1 mM each of DTT, PMSF and protease inhibitor cocktail). Following an ice-cold PBS wash, cells were lysed with SDS lysis buffer and the sonicated briefly before clarifying by centrifugation at 20,000 g for 20 minutes at 4°C. After centrifugation the supernatant was collected and protein content was measured using the standard BCA protein assay (Pierce, Rockford, IL, USA). Protein extracts (40 μg) were diluted in 6X sample buffer and loaded onto a 6% SDS-polyacrylamide gel. Gels were run for one hour at 100 V and then were transferred to a nitrocellulose membrane overnight at 25 V. All incubations were at room temperature in 0.5% Tween in Tris buffered saline (TTBS). The membranes were blocked for one hour in 5% milk in TTBS. Primary monoclonal anti-iNOS (Signal Transduction Laboratories, San Diego, CA, USA) or polyclonal anti-TNFR1 (Santa Cruz Biotechnology, Santa Cruz, CA, USA) were diluted in 2.5% milk in TTBS at 1:1000 and membranes were incubated with the antibody for one hour at room temperature. Membranes were washed three times for ten minutes each in TTBS before incubating for one hour with a horseradish-peroxidase secondary antibody (BioRad, Hercules, CA, USA) at 1:7500 in 2.5% milk in TTBS. Finally, membranes were washed again in TTBS three times for ten minutes each. Immunoreactive bands were visualized by a chemiluminescent western blot detection kit (Amersham Biosciences, Piscatay, NJ, USA) according to manufacturer's instructions. Images were captured using a 12 bit monochrome camera (UVP, Upland, CA, USA).
Reverse transcriptase polymerase chain reaction assay
Total RNA was extracted with Trizol Extraction Kit (Gibco BRL, San Diego, CA, USA) according to manufacturer's instructions. One μg of total RNA from each sample was applied to Ready-to-go RT-PCR Beads (Amersham Biosciences, Piscatay, NJ, USA) and used to complete the amplification protocol according to manufacturer's instructions. Primer sequences for rat iNOS were as follows; forward 5'-CAC GGA GAA CAG AGT TGG-3' and reverse 5'-GGA ACA CAG TAA TGG CCG ACC-3'. Amplified samples were run on agarose gels and stained with ethidium bromide. Images were captured using a 12 bit monochrome camera (UVP, Upland, CA, USA).
Flow cytometry
One μg of antibody against TrkA or p75NTR (Santa Cruz Biotechnology, Santa Cruz, CA, USA) was labeled with Zenon Rabbit IgG labeling kit from Molecular Probes (Eugene, OR) according to manufacturer's instructions and incubated for 1 hr with the cells in suspension. After incubation, labeled cells were visualized and quantified using a Becton Dickinson FACS Vantage Flow Cytometer set at appropriate instrument parameters.
Statistical analysis
Where appropriate, data were expressed as mean +/- standard error of the mean (S.E.M.), and analyzed by student unpaired two-tailed t test with significance set at p < 0.05.
Results
Combined NGF and TNFα induce iNOS message and protein
The upper panel of figure 1 shows a western blot detecting iNOS in PC12 cells treated simultaneously with 10 ng/ml NGF and 10 ng/ml TNFα in the presence or absence of 50 nM K252a, an inhibitor of phosphorylative events associated with tyrosine kinase receptor activation that has been shown to block the function of the high affinity NGF receptor TrkA [61]. There was a marked induction of iNOS expression only in cells simultaneously treated with NGF and TNFα, while neither treatment alone elicited any effect. Furthermore, K252a completely abolished NGF/TNFα-promoted iNOS induction, suggesting that TrkA function is essential to mediate it. As shown in the lower panel of figure 1, along with increased protein levels there was also an induction of iNOS mRNA in PC12 cells treated with NGF and TNFα but not in cells treated with either factor alone.
Figure 1 A: (Top) Western blot analysis detecting the presence of iNOS in 40 μg total protein extracts from PC12 cells treated for 24 hr with 10 ng/ml NGF and 10 ng/ml TNF, individually or combined (Both), in the presence of 50 nM of the receptor tyrosine kinase inhibitor K252a. Positive control (Pos) is 4 μg of total protein extracts from mouse macrophages. (Bottom) RT-PCR detecting iNOS mRNA in PC12 cells treated for 24 hr with 10 ng/ml NGF and 10 ng/ml TNF, individually or combined (Both) compared to untreated cells (Cont). Internal PCR controls lacking reverse transcriptase (RT-) were performed on each sample as shown. Results shown are representative of 3 replicate experiments.
NGF and TNFα are both required for sustained iNOS expression
Figure 2A shows western blots detecting iNOS in cells treated with increasing concentrations of NGF (top panel) or TNFα (bottom panel), in the presence or absence of a fixed amount of TNFα or NGF, respectively. Either factor was ineffective when added alone at any of the concentrations tested. However, there was a marked dose-response increase in iNOS expression when increasing concentrations of NGF or TNFα were added in the presence of a fixed amount of TNFα or NGF, respectively. Figure 2B shows a representative western blot detecting iNOS expression in cells continuously treated with NGF and TNFα as compared to cells in which the combined treatment was withdrawn after 24 hr. The expression of iNOS returned to basal, undetectable, levels between 24 and 48 hr after withdrawal of both TNFα and NGF. Furthermore, as shown in figure 2C, withdrawal of either NGF or TNFα alone was sufficient to abolish iNOS expression induced by the combined treatment, both at the protein (top panel) and mRNA level (bottom panel). To exclude the involvement of unknown serum factors, NGF/TNFα-promoted induction of iNOS was determined in cells cultured for 24 hr in serum free or in defined medium N2 (Figure 2D). There was a detectable iNOS induction in both serum free- and defined medium-cultured cells, although much reduced in serum free conditions, which is predictable as PC12 cells do not survive for longer periods of time (24–48 hrs) in the absence of serum or N2 supplements. Since insulin is present in both serum and the N2 supplement, and can activate the insulin-like growth factor (IGF) receptor, we asked whether TNFα may synergize with IGF, which is also present in serum, to induce iNOS expression. The results shown in Figure 2E indicate that this is not the case.
Figure 2 A: Western blots detecting iNOS in total protein extracts from PC12 cells treated for 24 hr with increasing concentrations of NGF in the presence or absence of 10 ng/ml TNFα (Top) or treated with increasing concentrations of TNFα in the presence or absence of 25 ng/ml NGF (Bottom). Positive control (Pos) is 4 μg of total protein extracts from mouse macrophages. Results shown are representative of 2 replicate experiments. B: Western blot analysis detecting iNOS in total protein extracts from PC12 cells simultaneously pre-treated with 10 ng/ml NGF and 10 ng/ml TNFα. At 24 hr treatment was withdrawn and the presence of iNOS was determined 24, 48, and 72 hr thereafter. Results shown are representative of 3 replicate experiments. C: Western blot analysis (Top) and RT-PCR (Bottom) detecting iNOS protein and mRNA in total protein extracts and total RNA from PC12 cells simultaneously pre-treated for 24 hr with 10 ng/ml NGF and 10 ng/ml TNFα (Both). After 24 hr, treatment was withdrawn and replaced with either NGF or TNFα alone or with both and iNOS expression determined 24 hr thereafter. Results shown are representative of 2 replicate experiments. D: Western blot detecting iNOS in total protein extracts from PC12 cells simultaneously treated for 24 hr with 10 ng/ml NGF and 10 ng/ml TNFα in medium containing serum, in serum free medium (SF) or in defined medium (N2). Results shown are representative of 3 replicate experiments. E: Western blot analysis detecting the presence of iNOS in total protein extracts from PC12 cells treated for 72 hr with 100 ng/ml IGF and 10 ng/ml TNF, individually or combined, as compared to cells simultaneously treated with 10 ng/ml NGF and 10 ng/ml TNFα or untreated controls (Cont). Results shown are representative of 4 replicate experiments.
TNFα/NGF-mediated iNOS expression is independent of NOS enzymatic activity
In order to determine whether the enzymatic activity of iNOS may play a role in sustaining TNFα/NGF-promoted signaling we pretreated PC12 cells with two NOS inhibitors prior to TNFα/NGF tretament. Pretreatment with N(G)-nitro-L-arginine methyl ester (L-NAME) did not affect expression of iNOS induced by the NGF/ TNFα combined treatment (Figure 3A). The same result was observed if a more specific inhibitor of iNOS (1400 W) was used instead of L-NAME (Figure 3B). Concentrations of 1400 W used here have been previously shown to be effective in inhibiting selectively iNOS activity in PC12 cells by others [78]. These results suggest that sustained iNOS expression in response of the combined NGF/TNFα treatment is independent of NOS enzymatic activity.
Figure 3 A: Western blot detecting iNOS in total protein extracts from PC12 cells treated for 24 hr with 10 ng/ml NGF and 10 ng/ml TNFα, either individually or simultaneously (Both). Cells were pretreated with vehicle or 0.5 μM of the generic NOS inhibitor L-NAME. Positive control (Pos) is 4 μg of total protein extracts from mouse macrophages. Results shown are representative of 3 replicate experiments. B: Western blot detecting iNOS in total protein extracts from PC12 cells simultaneously treated with 10 ng/ml NGF and 10 ng/ml TNFα (Both), in the presence or absence of a pre-treatment with varying concentrations of the iNOS-specific inhibitor 1400 W. Results shown are representative of 4 replicate experiments.
NGF/TNFα promoted iNOS induction requires the transcription factor NF-κB
Figure 4 shows results from PC12 cells transiently transfected with a secreted alkaline phosphatase reporter gene construct (SEAP) promoted by enhancer sequences specific for nuclear factor kappa B (NF-κB), activator protein 1 (AP-1), cAMP-responsive element (CRE) or Tal (non-inducible control). Twenty-four hr after transfection cells were treated with 10 ng/ml each of TNFα and NGF (alone or combined) and SEAP released in the culture medium (an index of endogenous transcription factor activation) was assayed 3 hr and 12 hr later. At 3 hr, cells treated with TNFα showed a significant increase in NF-κB activity but not AP-1 or CRE. Cells treated with NGF alone showed at 3 hr no significant increase in NF-κB, AP1 or CRE activity. When cells were exposed to the combined NGF/ TNFα treatment, there was a robust increase in NF-κB activity that was significantly higher than the response induced by the individual treatment with TNFα. On the other hand, neither AP-1 nor CRE activity were significantly affected by the combined NGF/ TNFα treatment. At 12 hr, both TNFα and NGF/TNFα combined treatments significantly increased NF-κB activity, but were not statistically significantly different. NGF-treated cells showed a significant increase in AP-1 and CRE activity at 12 hr, while NF-κB activity was not affected. As a result, there was also a significant increase in AP-1 and CRE activity elicited by the NGF/TNFα combined treatment at 12 hr. Neither NGF nor TNFα (alone or combined) elicited any effect on the control reporter construct Tal, either at 3 or 12 hr.
Figure 4 Detection of SEAP in the culture medium of PC12 cells transfected with a SEAP reporter gene construct under the transcriptional control of enhancers specific for NF-κB, AP-1 or CRE. pTal is the non-enhanced control SEAP reporter vector. Twenty-four hr after transfection, cells were treated with vehicle (Control), 10 ng/ml NGF, 10 ng/ml TNFα or NGF plus TNFα (Both) and the presence of SEAP in the culture medium assayed 3 hr (Top) or 12 hr (Bottom) thereafter. Results are normalized to control cells in each transfection group (N = 3). * and #: p < 0.05 vs. control and TNFα-alone, respectively (two-tailed unpaired Student's t-test). Results shown are representative of 3 replicate experiments.
Involvement of NF-κB was further explored by determining the extent to which pharmacological inhibition of NF-κB would block NGF/TNFα-promoted iNOS induction in PC12 cells. As shown in figure 5A, treatment of PC12 cells with either pyrrolidine di-thio-carbamate (PDTC) or the octapeptide proteasome inhibitor PSI (two effective NF-κB inhibitors that have distinct mechanisms of action [8,63-65], completely abolished NGF/ TNFα-promoted iNOS induction. In this experiment, PD98059, a selective MAPK inhibitor, was used as a negative control. Both NF-κB inhibitors effectively blocked NF-κB-mediated transcriptional activity as determined by SEAP reporter gene assay (Figure 5B), whereas PD98059 had no effect. However, PD98059 completely blocked NGF-promoted neurite outgrowth (Figure 5C), an event that in PC12 cells is dependent on MAPK activation [66]. Furthermore, consistent with the results reported in Figure 4, inhibition of NOS activity by L-NAME did not affect NFκB activation by NGF/TNFα combined treatment (Figure 5D).
Figure 5 A: Western blot detecting iNOS in PC12 cells simultaneously treated with 10 ng/ml NGF and 10 ng/ml TNFα for 24 hr. Thirty minutes before NGF/ TNFα treatment cells were pre-treated with 10 μM pyrrolidinedithyocarbamate (PDTC), 2 μM of a oligopeptide proteosome inhibitor (PSI) or 10 μM of a MAPK inhibitor (PD98059). Results shown are representative of 2 replicate experiments. B: SEAP release in the culture medium of PC12 cells transfected for 24 hr with an NF-κB-sensitive SEAP reporter gene construct and treated for 12 hr with vehicle (Control), 10 ng/ml NGF, 10 ng/ml TNFα or NGF plus TNFα in the presence of 10 μM PD98059, 10 μM PDTC or 2 μM PSI. Data are shown as mean ± S.E.M. from 3 independent replicate experiments. * and #: p < 0.05 vs. control or TNFα-alone cells, respectively (two-tailed unpaired Student's t-test). C: Representative photomicrographs of PC12 cells treated for 48 hr with 10 ng/ml NGF in the presence or absence of 10 μM PD98059 or 2 μM PDTC. D: NFκB transcriptional activity (as measured by a transiently transfected SEAP reporter vector) in PC12 cells treated for 24 hr with 10 ng/ml NGF, 10 ng/ml TNFα or NGF plus TNFα (Both) in the presence of 0.5 μM L-NAME. Data are shown as mean ± S.E.M. from 3 independent replicate experiments. * and #: p < 0.05 vs. control or TNFα-alone cells, respectively (two-tailed unpaired Student's t-test).
NGF/TNFα-promoted iNOS induction requires the simultaneous presence of both the p75NTR and TrkA NGF receptors
Next, we subcloned a PC12 mutant cell line (PC12p75NTR (-)) that lacks p75NTR expression while retaining TrkA at levels comparable with wild type PC12 cells (Figure 6A). NF-κB activity was not significantly increased by the NGF/TNFα combined treatment over the levels induced by TNFα alone in PC12p75NTR (-) (Figure 6B). Consistent with this finding, PC12p75NTR (-) cells exposed to the combined NGF/TNFα treatment did not show any induction of iNOS expression as compared to the parent cell line (Figure 6C). It is important to note that the PC12p75NTR (-) cells used here express TNFα receptor type 1 (TNFR1) at levels comparable (or even higher) than wild type PC12 cells (Figure 6D). Therefore lack of iNOS induction by the NGF/TNFα combined treatment in these cells cannot be ascribed to lack of TNFα responsiveness (as can also be appreciated by the NFκB response induced by TNFα alone shown in figure 6B).
Figure 6 A: Graph depicting the percentage of TrkA- or p75NTR- immunopositive cells in wild type (wt)PC12 cells and PC12 cell mutants lacking the low affinity NGF receptor (PC12p75NTR(-)) from flow cytometry data. Results shown are representative of 3 replicate flow cytometry experiments on the same cell line. B: SEAP release in the culture medium of PC12p75NTR (-) cells transfected for 24 hr with an NF-κB-sensitive SEAP reporter gene construct and treated for 12 hr with vehicle (Cont), 10 ng/ml NGF, 10 ng/ml TNFα or NGF plus TNFα (Both). Data are shown as mean ± S.E.M. from 3 independent replicate experiments. * : p < 0.05 vs. control or NGF-alone cells (two-tailed unpaired Student's t-test). C: Western blot detecting the presence of iNOS in wtPC12 cells and PC12p75NTR (-) cells treated for 24 hr with vehicle (Cont), 10 ng/ml NGF, 10 ng/ml TNFα or NGF plus TNFα (Both). Membrane was re-probed for β-actin (lower panel) to control for equal protein loading. Positive control (Pos) is 4 μg of total protein extracts from mouse macrophages. Results shown are representative of 4 replicate experiments. D: Western blot detecting the presence of TNFR-I in total protein extracts from wtPC12 cells and PC12p75NTR (-) cells. Twenty μg of total protein extracts from rat dorsal root ganglia (DRG) were used as a positive control.
The results obtained in PC12p75NTR(-) would suggest that p75NTR is essential to mediate iNOS induction by the combined TNFα/NGF treatment while the results obtained using K252a (Figure 1) would suggest a prominent role for TrkA. In order to ultimately ascertain the relative role of the two NGF receptors in mediating TNFα/NGF-promoted iNOS induction we made use of PC12 cells transiently transfected with expression vectors coding for chimeric TNFα/NGF receptors constructed as described by Rovelli et al. [77]. These constructs bear the ligand binding domain from the human TNFR1 and the signal transduction domain from rat NGF receptors, either TrkA or p75NTR. Previously, it has been shown that transfection with these chimeras allows for TNF-promoted NGF signaling [77]. Figure 7 shows a western blot detecting iNOS in PC12 cells individually or simultaneously transfected with chimeric TNFα receptors bearing the intracellular domain of p75NTR (p55/p75NTR) or TrkA (p55/TrkA). Transfected cells were then treated either with TNFα and NGF alone, or with both TNFα and NGF. As expected, the combined TNFα/NGF treatment induced a robust expression of iNOS in these PC12 cells, regardless of the presence of any transfected expression vector. As also expected, NGF alone did not elicit iNOS expression in any of the transfected cells. Similarly, TNFα alone did not induce iNOS in cells transfected with either p55/p75NTR or p55/TrkA chimeric receptors. However, TNFα promptly induced iNOS expression in cells transfected with both p55/p75NTR and p55/TrkA chimeric receptors.
Figure 7 Western blot detecting iNOS in 40 μg total protein extracts from PC12 cells treated for 24 hr with 10 ng/ml human TNFα, 10 ng/ml NGF, or both. Twenty-four hr before treatment, cells were transfected with either an empty vector or expression vectors for chimeric receptor proteins bearing the human TNFR1 ligand binding domains and the intracellular domain of either rat p75NTR or TrkA NGF receptors (p75NTR, TrkA or p75NTR+TrkA). Positive control (Pos) is 40 μg of total protein extract from wild type PC12 cells treated with both rat TNFα and NGF. Membrane was re-probed for β-actin (lower panel) to control for equal protein loading and is representative from 3 independent transfections and treatments.
Discussion
The work presented here stems from our original observation that iNOS expression and subsequent NO production can be synergistically induced by NGF and TNFα in a TrkA-dependent manner in PC12 cells [43]. Our present results investigated the signalling pathways involved. Since we consistently observed a higher iNOS expression if NGF is added simultaneously to TNFα, we propose that iNOS expression was induced selectively in NGF-responsive cells. These results do not allow us to rule out the possibility that intermediate factors induced by TNFα or NGF may play a role in sensitizing indirectly cells to NGF or TNFα, respectively. However, the results shown in Figure 2 seem to exclude such a possibility. Indeed, while withdrawal of NGF and/or TNFα allows for a prompt ablation of iNOS expression (Figure 2B), neither NGF nor TNFα alone is sufficient to sustain iNOS expression following withdrawal of TNFα or NGF (Figure 2C). These observations suggest that the simultaneous and continuous presence of both factors is required to sustain iNOS induction/expression and that cell sensitization through a priming mechanism seems unlikely. Nonetheless, other researchers have attributed increased TNFα toxicity in PC12 cells to NGF-induced differentiation [67]. However, our results seem to exclude that differentiation of PC12 cells may have played a role. First, in our experimental conditions iNOS expression occurs as early as 3 hr after the exposure to the combined NGF/TNFα treatment [43], earlier than any morphological differentiation induced by NGF. Second, while blockade of NGF-induced differentiation by the MAPK inhibitor PD98059 (Figure 5C, [68]) had no effect on NGF/TNFα-promoted iNOS expression (Figure 5A), blockade of NFκB did not affect NGF-induced differentiation (Figure 5C) but completely inhibited iNOS expression.
In the present study we also report that induction and maintenance of iNOS expression by the combined NGF/TNFα treatment requires continuous de novo iNOS mRNA synthesis, presumably due to transcription factor regulation. Indeed, abolishing iNOS enzymatic activity had no effect on NGF/TNFα-promoted iNOS induction (Figure 4A,B). Therefore, the involvement of positive feedback due to NO seems unlikely. On the other hand, analysis of transcriptional activity of NF-κB, AP-1 and CRE revealed that NF-κB most likely mediates synergistic iNOS induction by TNFα and NGF. Since iNOS induction can be observed as early as 3 hr after NGF/TNFα combined treatment in PC12 cells [43], the results shown in figure 5 suggest that NF-κB is the only transcription factor among those tested here that is responsive to the simultaneous treatment with TNFα and NGF in a fashion consistent with induction of iNOS expression. In fact, while TNFα alone induced NFκB at 3 hr, this induction was significantly lower than the one promoted by the combined NGF/TNFα treatment. Whether the extent to which NFκB is activated or whether qualitative differences in NFκB subunit composition in response to TNFα as compared to NGF/TNFα treatment may play a role in inducing iNOS expression remains to be established. Nonetheless, inhibition of NF-κB completely inhibited iNOS induction while inhibition of MAPK was ineffective (Figure 5A). Lastly, inhibition of NOS activity failed to block NGF/TNFα-promoted NFκB activation, thus further supporting the idea that targeting NO may acutely ameliorate associated oxidative stress, but could not represent the most comprehensive approach to achieve a long term correction of these events.
Previous studies indicated that NGF can induce NF-κB by acting through the low affinity p75NTR receptor [70]. Thus, involvement of NF-κB in mediating NGF/TNFα combined effects would suggest a role for p75NTR. Indeed, we found that mutant PC12 cells that lack expression of the p75NTR receptor failed to respond in terms of iNOS expression when simultaneously treated with NGF and TNFα. Consistent with this finding, in PC12 cell mutants lacking p75NTR expression NF-κB activity was not induced by the combined NGF/TNFα treatment above the levels observed in cells treated with TNFα alone (Figure 6B).
That PC12 cells bearing only the TrkA receptor failed to respond the combined NGF/TNFα treatment suggests that signaling from p75NTR in combination with TNFα is necessary to induce iNOS expression. On the other hand, our previous work illustrated the importance of TrkA-associated signaling in mediating NGF/TNFα-promoted induction of iNOS [43] (see also figure 1). These results are only apparently in contrast. Indeed, in an admittedly artificial system making use of chimeric constructs we observed that only in the presence of both TNFα-responsive NGF receptor signaling can TNFα promote iNOS expression when added alone. Whether this is a consequence of simultaneous but independent signaling of both types of NGF receptors [79] or recruitment of intracellular signalling elements uniquely driven by the simultaneous activation of both NGF receptors' signaling domains remains to be investigated. On the other hand, these results exclude the possibility that the combined action of TNFα and NGF may derive from yet undescribed interaction(s) of the extracellular domains of their respective receptors following ligand binding.
Thus, our combined results would indicate that there exists a specific pathway involving NF-κB and requiring the simultaneous expression or both types of NGF receptors that is synergistically induced by TNFα and NGF to promote expression of iNOS. This is of particular interest given that neuron types expressing both TrkA and p75NTR receptors are limited and known to be affected in neurodegenerative conditions where neuroinflammation and pro-inflammatory cytokines have been shown to play a significant role. Notably, simultaneous expression of TrkA and p75NTR in the CNS is mostly restricted to the BFCN that are known to be particularly affected in AD. Indeed, others have also described signaling pathways that require the simultaneous expression of both TrkA and p75NTR [71,72] as well as the convergence of TrkA and p75NTR-mediated signaling impinging upon NF-κB [73]. Recent reports in neurons of TNF-promoted signaling occurring selectively in the presence of the glutamate agonist NMDA [4] illustrate the importance of considering the signaling "context" when studying the effects of cytokine treatment.
Overall, our data indicate the possibility that a convergence between NGF-promoted trophic signaling and TNFα could selectively endanger NGF-responsive neurons under conditions of neuroinflammation because of a synergistic action between TNFα and NGF to induce iNOS expression. For example, TNFα overexpressing transgenic mice show selective neurodegeneration of NGF-responsive basal forebrain cholinergic neurons [57] and direct TNFα administration in the brain of mice results in an impairment of basal forebrain cholinergic function [58]. However, whether induction of iNOS and subsequent oxidative damage may play a role in these two models remains to be determined [80].
Conclusion
TNFα and NGF, via concerted signaling events involving NFκB transcriptional activity and targeting NGF-responsive cells bearing both the high and low affinity NGF receptors, converge to stimulate de novo transcription of iNOS. Our present results are relevant to neurodegenerative conditions such as AD [22,74], stroke [17,75], ALS [20,76] and spinal chord injury [8,10] where neuroinflammation and high levels of pro-inflammatory cytokines have been shown to play a significant role and proposed as therapeutic targets.
List of Abbreviations
AraC, cytosine β-D-arabinofuranoside; AD, Alzheimers disease; BDNF, brain derived neurotrophic factor; BFCN, basal forebrain cholinergic neurons; CNS, central nervous system; CRE, cyclic-AMP response element; GDNF, glial derived neurotrophic factor; IGF, insulin-like growth factor; IL-1β, interleukin-1beta; iNOS, inducible nitric oxide synthase; MAPK, mitogen activated protein kinase; NF-κB, nuclear factor kappa B; NGF, nerve growth factor; NO, nitric oxide; nNOS, neuronal nitric oxide synthase; NTR, neurotrophin receptor; PC12, pheochromocytoma; PCN, penicillin; PDTC, pyrrolidinedithyocarbamate; PSI, proteosome inhibitor; SDS, sodium dodecylsulfate; SEAP, secreted alkaline phosphatase; S.E.M, standard error of the mean; Strep, streptomycin; TNFα, tumor necrosis factor alpha; TrkA, troponin-like receptor kinase A; TTBS, tris-buffered saline with tween 20;
Competing interests
The author(s) declare they have no competing interests.
Authors' contributions
MST participated in the conception and design of the study, carried out the bulk of experiments, performed data analysis, and drafted the manuscript. PMJ participated in study design especially with regards to the IGF experiments. WZ participated in study design and coordination and provided the expertise for RTPCR and withdrawal experiments. HUS sub-cloned the PC12p75NTR(-) cells and participated in study design and result interpretation of experiments involving these cells. GT participated in conception, study design, coordination and helped to draft and review the manuscript. All authors read and approved the final manuscript.
Acknowledgements
This work was supported in part by a research development grant by the UTMB Sealy Endowed Fund for Biomedical Research. Michael Thomas is supported by an NIEHS training grant pre-doctoral fellowship from T32 ES007254 and the UTMB Sealy Center for Aging pre-doctoral fellowship.
==== Refs
Floyd RA Neuroinflammatory processes are important in neurodegenerative diseases: an hypothesis to explain the increased formation of reactive oxygen and nitrogen species as major factors involved in neurodegenerative disease development Free Radic Biol Med 1999 26 1346 55 10381209 10.1016/S0891-5849(98)00293-7
McGeer PL McGeer EG Innate immunity, local inflammation, and degenerative disease Sci Aging Knowledge Environ 2002 2002 re3 14602998
Creange A Lefaucheur JP Authier FJ Gherardi RK Cytokines and peripheral neuropathies Revue Neurologique 1998 154 208 216 9773044
Sung YJ Ambron RT Pathways that elicit long-term changes in gene expression in nociceptive neurons following nerve injury: contributions to neuropathic pain Neurol Res 2004 26 195 203 15072639 10.1179/016164104225013761
Chandross KJ Nerve injury and inflammatory cytokines modulate gap junctions in the peripheral nervous system Glia 1998 24 21 31 9700486 10.1002/(SICI)1098-1136(199809)24:1<21::AID-GLIA3>3.0.CO;2-3
Stoll G Jander S Myers RR Degeneration and regeneration of the peripheral nervous system: from Augustus Waller's observations to neuroinflammation J Peripher Nerv Syst 2002 7 13 27 11939348 10.1046/j.1529-8027.2002.02002.x
Norenberg MD Smith J Marcillo A The pathology of human spinal cord injury: defining the problems J Neurotrauma 2004 21 429 40 15115592 10.1089/089771504323004575
La Rosa G Cardali S Genovese T Conti A Di Paola R La Torre D Cacciola F Cuzzocrea S Inhibition of the nuclear factor-kappaB activation with pyrrolidine dithiocarbamate attenuating inflammation and oxidative stress after experimental spinal cord trauma in rats J Neurosurg Spine 2004 1 311 21 15478370
Popovich PG Jones TB Manipulating neuroinflammatory reactions in the injured spinal cord: back to basics Trends Pharmacol Sci 2003 24 13 7 12498725 10.1016/S0165-6147(02)00006-8
Hausmann ON Post-traumatic inflammation following spinal cord injury Spinal Cord 2003 41 369 78 12815368 10.1038/sj.sc.3101483
Bareyre FM Schwab ME Inflammation, degeneration and regeneration in the injured spinal cord: insights from DNA microarrays Trends Neurosci 2003 26 555 63 14522149 10.1016/j.tins.2003.08.004
Bayir H Kochanek PM Clark RS Traumatic brain injury in infants and children: mechanisms of secondary damage and treatment in the intensive care unit Crit Care Clin 2003 19 529 49 12848319 10.1016/S0749-0704(03)00014-9
Morganti-Kossmann MC Rancan M Stahel PF Kossmann T Inflammatory response in acute traumatic brain injury: a double-edged sword Curr Opin Crit Care 2002 8 101 5 12386508 10.1097/00075198-200204000-00002
Lenzlinger PM Morganti-Kossmann MC Laurer HL McIntosh TK The duality of the inflammatory response to traumatic brain injury Mol Neurobiol 2001 24 169 81 11831551 10.1385/MN:24:1-3:169
Sundararajan S Landreth GE Antiinflammatory properties of PPARgamma agonists following ischemia Drug News Perspect 2004 17 229 36 15334171 10.1358/dnp.2004.17.4.829049
Dirnagl U Inflammation in stroke: the good, the bad, and the unknown Ernst Schering Res Found Workshop 2004 87 99 15032055
Danton GH Dietrich WD Inflammatory mechanisms after ischemia and stroke J Neuropathol Exp Neurol 2003 62 127 36 12578222
Consilvio C Vincent AM Feldman EL Neuroinflammation, COX-2, and ALS – a dual role? Exp Neurol 2004 187 1 10 15081582 10.1016/j.expneurol.2003.12.009
Pompl PN Ho L Bianchi M McManus T Qin W Pasinetti GM A therapeutic role for cyclooxygenase-2 inhibitors in a transgenic mouse model of amyotrophic lateral sclerosis Faseb J 2003 17 725 7 12586733
McGeer PL McGeer EG Inflammatory processes in amyotrophic lateral sclerosis Muscle Nerve 2002 26 459 70 12362410 10.1002/mus.10191
Cacquevel M Lebeurrier N Cheenne S Vivien D Cytokines in neuroinflammation and Alzheimer's disease Curr Drug Targets 2004 5 529 34 15270199 10.2174/1389450043345308
McGeer EG McGeer PL Inflammatory processes in Alzheimer's disease Prog Neuropsychopharmacol Biol Psychiatry 2003 27 741 9 12921904 10.1016/S0278-5846(03)00124-6
Gupta A Pansari K Inflammation and Alzheimer's disease Int J Clin Pract 2003 57 36 9 12587940
McGeer PL McGeer EG Local neuroinflammation and the progression of Alzheimer's disease J Neurovirol 2002 8 529 38 12476347 10.1080/13550280290100969
Emerit J Edeas M Bricaire F Neurodegenerative diseases and oxidative stress Biomed Pharmacother 2004 58 39 46 14739060 10.1016/j.biopha.2003.11.004
Tarkowski E Liljeroth AM Minthon L Tarkowski A Wallin A Blennow K Cerebral pattern of pro- and anti-inflammatory cytokines in dementias Brain Res Bull 2003 61 255 60 12909295 10.1016/S0361-9230(03)00088-1
Tarkowski E Andreasen N Tarkowski A Blennow K Intrathecal inflammation precedes development of Alzheimer's disease J Neurol Neurosurg Psychiatry 2003 74 1200 5 12933918 10.1136/jnnp.74.9.1200
McGeer PL McGeer EG Suzuki J Dolman CE Nagai T Aging, Alzheimer's disease, and the cholinergic system of the basal forebrain Neurology 1984 34 741 5 6539435
Rinne JO Paljarvi L Rinne UK Neuronal size and density in the nucleus basalis of Meynert in Alzheimer's disease J Neurol Sci 1987 79 67 76 3612173 10.1016/0022-510X(87)90260-7
Vogels OJ Broere CA ter Laak HJ ten Donkelaar HJ Nieuwenhuys R Schulte BP Cell loss and shrinkage in the nucleus basalis Meynert complex in Alzheimer's disease Neurobiology of Aging 1990 11 3 13 2183081 10.1016/0197-4580(90)90056-6
Hefti F Nerve growth factor promotes survival of septal cholinergic neurons after fimbrial transections J Neurosci 1986 6 2155 62 3746405
Hartikka J Hefti F Development of septal cholinergic neurons in culture: plating density and glial cells modulate effects of NGF on survival, fiber growth, and expression of transmitter-specific enzymes J Neurosci 1988 8 2967 85 2842468
Perry G Castellani RJ Smith MA Harris PL Kubat Z Ghanbari K Jones PK Cordone G Tabaton M Wolozin B Oxidative damage in the olfactory system in Alzheimer's disease Acta Neuropathol (Berl) 2003 106 552 6 12955399 10.1007/s00401-003-0761-7
Butterfield DA Boyd-Kimball D Castegna A Proteomics in Alzheimer's disease: insights into potential mechanisms of neurodegeneration J Neurochem 2003 86 1313 27 12950441 10.1046/j.1471-4159.2003.01948.x
Giasson BI Ischiropoulos H Lee VM Trojanowski JQ The relationship between oxidative/nitrative stress and pathological inclusions in Alzheimer's and Parkinson's diseases Free Radic Biol Med 2002 32 1264 75 12057764 10.1016/S0891-5849(02)00804-3
Perry G Nunomura A Hirai K Takeda A Aliev G Smith MA Oxidative damage in Alzheimer's disease: the metabolic dimension Int J Dev Neurosci 2000 18 417 21 10817925 10.1016/S0736-5748(00)00006-X
Gold BG Udina E Bourdette D Navarro X Neuroregenerative and neuroprotective actions of neuroimmunophilin compounds in traumatic and inflammatory neuropathies Neurol Res 2004 26 371 80 15198862 10.1179/016164104225013734
Rogerio F Teixeira SA de Rezende AC de Sa RC de Souza Queiroz L De Nucci G Muscara MN Langone F Superoxide dismutase isoforms 1 and 2 in lumbar spinal cord of neonatal rats after sciatic nerve transection and melatonin treatment Brain Res Dev Brain Res 2005 154 217 25 15707675 10.1016/j.devbrainres.2004.10.017
Kim HK Park SK Zhou JL Taglialatela G Chung K Coggeshall RE Chung JM Reactive oxygen species (ROS) play an important role in a rat model of neuropathic pain Pain 2004 111 116 24 15327815 10.1016/j.pain.2004.06.008
van Beek J Elward K Gasque P Activation of complement in the central nervous system: roles in neurodegeneration and neuroprotection Ann N Y Acad Sci 2003 992 56 71 12794047
Murdoch I Perry EK Court JA Graham DI Dewar D Cortical cholinergic dysfunction after human head injury J Neurotrauma 1998 15 295 305 9605345
Murdoch I Nicoll JA Graham DI Dewar D Nucleus basalis of Meynert pathology in the human brain after fatal head injury J Neurotrauma 2002 19 279 84 11898797 10.1089/08977150252807018
Macdonald NJ Taglialatela G Tumor necrosis factor alpha and nerve growth factor synergistically induce iNOS in pheochromocytoma cells Neuroreport 2000 11 3453 3456 11095498
Isaksson J Farooque M Olsson Y Improved functional outcome after spinal cord injury in iNOS-deficient mice Spinal Cord 2004
Urushitani M Shimohama S Kihara T Sawada H Akaike A Ibi M Inoue R Kitamura Y Taniguchi T Kimura J Mechanism of selective motor neuronal death after exposure of spinal cord to glutamate: Involvement of glutamate induced nitric oxide in motor neuron toxicity and nonmotor neuron protection Annals of Neurology 1998 44 796 807 9818936 10.1002/ana.410440514
Diaz-Ruiz A Ibarra A Perez-Severiano F Guizar-Sahagun G Grijalva I Rios C Constitutive and inducible nitric oxide synthase activities after spinal cord contusion in rats Neurosci Lett 2002 319 129 32 11834310 10.1016/S0304-3940(01)02540-X
Parmentier-Batteur S Bohme GA Lerouet D Zhou-Ding L Beray V Margaill I Plotkine M Antisense oligodeoxynucleotide to inducible nitric oxide synthase protects against transient focal cerebral ischemia-induced brain injury Journal of Cerebral Blood Flow & Metabolism 2001 21 15 21 11149663 10.1097/00004647-200101000-00003
Sarchielli P Galli F Floridi A Gallai V Relevance of protein nitration in brain injury: a key pathophysiological mechanism in neurodegenerative, autoimmune, or inflammatory CNS diseases and stroke Amino Acids 2003 25 427 36 14661102 10.1007/s00726-003-0028-6
Luth HJ Munch G Arendt T Aberrant expression of NOS isoforms in Alzheimer's disease is structurally related to nitrotyrosine formation Brain Res 2002 953 135 43 12384247 10.1016/S0006-8993(02)03280-8
Luth HJ Expression of endothelial and inducible NOS-isoforms is increased in Alzheimer's disease, in APP23 transgenic mice and after experimental brain lesion in rat: evidence for an induction by amyloid pathology Brain Research 2001 913 57 67 11532247 10.1016/S0006-8993(01)02758-5
Law A Gauthier S Quirion R Say NO to Alzheimer's disease: the putative links between nitric oxide and dementia of the Alzheimer's type Brain Res Brain Res Rev 2001 35 73 96 11245887 10.1016/S0165-0173(00)00051-5
de la Monte SM Lu BX Sohn YK Etienne D Kraft J Ganju N Wands JR Aberrant expression of nitric oxide synthase III in Alzheimer's disease: relevance to cerebral vasculopathy and neurodegeneration Neurobiology of Aging 2000 21 309 19 10867216 10.1016/S0197-4580(99)00108-6
Lee S Zhao ML Hirano A Dickson D Inducible nitric oxide synthase immunoreactivity in the Alzheimer disease hippocampus: association with Hirano bodies, neurofibrillary tangles, and senile plaques Journal of Neuropathology and Experimental Neurology 1999 58 1163 1169 10560659
Darville MI Eizirik DL Regulation by cytokines of the inducible nitric oxide synthase promoter in insulin-producing cells Diabetologia 1998 41 1101 8 9754830 10.1007/s001250051036
Kong GY Peng ZC Costanzo C Kristensson K Bentivoglio M Inducible nitric oxide synthase expression elicited in the mouse brain by inflammatory mediators circulating in the cerebrospinal fluid Brain Res 2000 878 105 18 10996141 10.1016/S0006-8993(00)02716-5
Barker V Middleton G Davey F Davies AM TNFalpha contributes to the death of NGF-dependent neurons during development Nature Neuroscience 2001 4 1194 8 11685224 10.1038/nn755
Aloe L Fiore M Probert L Turrini P Tirassa P Overexpression of tumour necrosis factor alpha in the brain of transgenic mice differentially alters nerve growth factor levels and choline acetyltransferase activity Cytokine 1999 11 45 54 10080878 10.1006/cyto.1998.0397
Wenk GL McGann K Hauss-Wegrzyniak B Rosi S The toxicity of tumor necrosis factor-alpha upon cholinergic neurons within the nucleus basalis and the role of norepinephrine in the regulation of inflammation: implications for alzheimer's disease Neuroscience 2003 121 719 29 14568031 10.1016/S0306-4522(03)00545-1
Kiss JZ Wang C Olive S Rougon G Lang J Baetens D Harry D Pralong WF Activity-dependent mobilization of the adhesion molecule polysialic NCAM to the cell surface of neurons and endocrine cells Embo J 1994 13 5284 92 7957094
Stoppini L Buchs PA Muller D A simple method for organotypic cultures of nervous tissue Journal of Neuroscience Methods 1991 37 173 82 1715499 10.1016/0165-0270(91)90128-M
Ohmichi M Decker SJ Pang L Saltiel AR Inhibition of the cellular actions of nerve growth factor by staurosporine and K252A results from the attenuation of the activity of the trk tyrosine kinase Biochemistry 1992 31 4034 9 1314657 10.1021/bi00131a019
Pappas TC Decorti F Macdonald NJ Neet KE Taglialatela G Tumour necrosis factor-alpha- vs. growth factor deprivation-promoted cell death: different receptor requirements for mediating nerve growth factor-promoted rescue Aging Cell 2003 2 83 92 12882321 10.1046/j.1474-9728.2003.00039.x
Taglialatela G Robinson R Perez-Polo JR Inhibition of nuclear factor kappa B (NFkappaB) activity induces nerve growth factor-resistant apoptosis in PC12 cells Journal of Neuroscience Research 1997 47 155 62 9008146 10.1002/(SICI)1097-4547(19970115)47:2<155::AID-JNR4>3.0.CO;2-E
Taglialatela G Kaufmann JA Trevino A Perez-Polo JR Central nervous system DNA fragmentation induced by the inhibition of nuclear factor kappa B Neuroreport 1998 9 489 93 9512395
Traenckner EB Wilk S Baeuerle PA A proteasome inhibitor prevents activation of NF-kappa B and stabilizes a newly phosphorylated form of I kappa B-alpha that is still bound to NF-kappa B Embo J 1994 13 5433 41 7957109
Pang L Sawada T Decker SJ Saltiel AR Inhibition of MAP kinase kinase blocks the differentiation of PC-12 cells induced by nerve growth factor Journal of Biological Chemistry 1995 270 13585 8 7775407 10.1074/jbc.270.36.21040
Mielke K Herdegen T Fatal shift of signal transduction is an integral part of neuronal differentiation: JNKs realize TNFalpha-mediated apoptosis in neuronlike, but not naive, PC12 cells Mol Cell Neurosci 2002 20 211 24 12093155 10.1006/mcne.2002.1132
Morooka T Nishida E Requirement of P38 Mitogen-Activated Protein Kinase For Neuronal Differentiation in Pc12 Cells Journal of Biological Chemistry 1998 273 24285 24288 9733710 10.1074/jbc.273.38.24285
Perez-Polo JR Foreman PJ Jackson GR Shan D Taglialatela G Thorpe LW Werrbach-Perez K Nerve growth factor and neuronal cell death Molecular Neurobiology 1990 4 57 91 2076219
Carter BD Kaltschmidt C Kaltschmidt B Offenhauser N Bohm-Matthaei R Baeuerle PA Barde YA Selective activation of NF-kappa B by nerve growth factor through the neurotrophin receptor p75 [see comments] Science 1996 272 542 5 8614802
Lad SP Peterson DA Bradshaw RA Neet KE Individual and combined effects of TrkA and p75NTR nerve growth factor receptors. A role for the high affinity receptor site J Biol Chem 2003 278 24808 17 12702729 10.1074/jbc.M212270200
Szutowicz A Madziar B Pawelczyk T Tomaszewicz M Bielarczyk H Effects of NGF on acetylcholine, acetyl-CoA metabolism, and viability of differentiated and non-differentiated cholinergic neuroblastoma cells J Neurochem 2004 90 952 61 15287901 10.1111/j.1471-4159.2004.02556.x
Foehr ED Lin X O'Mahony A Geleziunas R Bradshaw RA Greene WC NF-kappa B signaling promotes both cell survival and neurite process formation in nerve growth factor-stimulated PC12 cells J Neurosci 2000 20 7556 63 11027214
Hoozemans JJ Veerhuis R Rozemuller AJ Eikelenboom P Non-steroidal anti-inflammatory drugs and cyclooxygenase in Alzheimer's disease Curr Drug Targets 2003 4 461 8 12866660 10.2174/1389450033490902
Barone FC Parsons AA Therapeutic potential of anti-inflammatory drugs in focal stroke Expert Opin Investig Drugs 2000 9 2281 306 11060807 10.1517/13543784.9.10.2281
Weydt P Weiss MD Moller T Carter GT Neuro-inflammation as a therapeutic target in amyotrophic lateral sclerosis Curr Opin Investig Drugs 2002 3 1720 4 12528305
Rovelli G Heller RA Canossa M Shooter EM Chimeric tumor necrosis factor-TrkA receptors reveal that ligand-dependent activation of the TrkA tyrosine kinase is sufficient for differentiation and survival of PC12 cells Proc Natl Acad Sci U S A 1993 90 8717 8721 7690970
Jiang H Koubi D Zhang L Kuo J Rodriguez AI Jackson Hunter T Inhibitors of iNOS protects PC12 cells against the apoptosis induced by oxygen and glucose deprivation Neuroscience Letters 2005 375 59 63 15664123 10.1016/j.neulet.2004.10.067
Woo SB Page J Saragovi HU Neet KE Binding of nerve growth factor (NGF) to Trk receptor chimeras Faseb Journal 2004 18 C273
Floden AM Li SS Combs CK beta-Amyloid-stimulated microglia induce neuron death via synergistic stimulation of tumor necrosis factor alpha and NMDA receptors Journal of Neuroscience 2005 25 2566 2575 15758166 10.1523/JNEUROSCI.4998-04.2005
|
16144552
|
PMC1242246
|
CC BY
|
2021-01-04 16:38:22
|
no
|
J Neuroinflammation. 2005 Sep 6; 2:19
|
utf-8
|
J Neuroinflammation
| 2,005 |
10.1186/1742-2094-2-19
|
oa_comm
|
==== Front
Malar JMalaria Journal1475-2875BioMed Central London 1475-2875-4-411616228410.1186/1475-2875-4-41ResearchArtesunate plus sulfadoxine-pyrimethamine for treatment of uncomplicated Plasmodium falciparum malaria in Sudan Elamin Sakina B [email protected] Elfatih M [email protected] Tarig [email protected] Ammar H [email protected] Mamoun M [email protected] Elderderi S [email protected] Ishag [email protected] NationalMalaria Control, Ministry of Health, Khartoum, Sudan2 Albayan College for Science, Sudan University for Science and Technology, Sudan3 University of Kassala, Sudan4 Faculty of Medicine University of Khartoum, The Academy of Medical Sciences and Technology, Department of Obstetrics & Gynecology, Faculty of Medicine University of Khartoum, P. O. Box 102, Khartoum, Sudan2005 14 9 2005 4 41 41 19 6 2005 14 9 2005 Copyright © 2005 Elamin et al; licensee BioMed Central Ltd.2005Elamin et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Early diagnosis and effective treatment with an appropriate drug form the main components of the World Health Organization's strategy to reduce malaria related mortality. The few available drugs might be safeguarded if combined with artesunate. The addition of artesunate to a standard antimalarial treatment substantially reduces treatment failure, recrudescence and gametocyte carriage.
Methods
During late 2004, the efficacy of artesunate (4 mg/kg. day, on days 0–2) plus sulfadoxine-pyrimethamine (25 mg/kg, on day 0) for the treatment of uncomplicated Plasmodium falciparum malaria was investigated in four sentinel areas in Sudan, with different malaria transmission (Damazin, Kassala, Kosti, and Malakal).
Results
Two hundreds and sixty-nine patients completed the 28-day follow-up. On day one, 60 (22.3%) patients were febrile and 15 (5.5%) patients were parasitaemic. On day three, all the patients were afebrile and aparasitaemic. While two patients (0.7%, Kassala) showed late Clinical and Parasitological Failures, the rest (99.3%) of the patients demonstrated Adequate Clinical and Parasitological Response. A gametocytaemia were detected during the follow-up in one patient (0.37%, Kassala). Adverse drug effects were detected in 32 (11.9%) patients
Conclusion
The study showed that AS plus SP is an effective, safe drug in the treatment of uncomplicated P. falciparum malaria in Sudan.
==== Body
Background
There are almost 515 (range 300–660) million episodes of clinical Plasmodium falciparum malaria infections [1]. Drug-resistant malaria is spreading in Africa and countries with high levels of resistance have witnessed increased morbidity and mortality [2]. Early diagnosis and effective treatment with an appropriate drug form the main components of the World Health Organization's strategy to reduce malaria-related mortality [3].
The few available drugs might be safeguarded if combined with artesunate. The addition of artesunate to standard antimalarial treatments substantially reduces treatment failure, recrudescence and gametocyte carriage, preventing the emergence and spread of drug resistance and interrupting the transmission of P. falciparum. Coupled with early detection and confirmed diagnosis, this strategy represents the only way forward in the chemotherapy of malaria [4-8].
Malaria causes between 7.5 to 10 million cases and 35,000 deaths every year in Sudan [9]. Due to the spread of multidrug-resistant P. falciparum malaria in Sudan [10,11], artesunate plus sulfadoxine-pyrimethamine is recommended as the first-line treatment for uncomplicated P. falciparum malaria. The study aimed to investigate the efficacy of AS plus SP, as there is little published data in Sudan [8,12].
Patients and methods
Data collection
The study was conducted in October and November, 2004 at four health centres in different regions of Sudan (Damazin, Kassala, Kosti, and Malakal) (Figure 1). Three of these areas were characterized by low malaria transmission and the fourth (Malakal) was characterized by stable transmission [13]. Febrile (temperature ≥ 37.5°C) patients with uncomplicated P. falciparum malaria [14], who had no history of antimalarial drug use during the preceding two weeks, were recruited for the study. Pregnant women and patients with mixed infections were excluded.
After obtaining informed consent from the patient or the child's parents, a fixed questionnaire including relevant socio-demographic characteristics, medical history, physical findings and investigations conducted was completed for each patient.
Figure 1 A sketch map of the Sudan, showing the main rivers, Khartoum and the four sites where the study was conducted.
Laboratory methods
Blood films were prepared, stained with Giemsa and 100x oil immersion fields were examined. The parasite density was counted against 200 leucocytes, assuming 6,000 leucocytes/μl. All the slides were double-checked blindly and only considered negative if no parasites were detected in 100 oil immersion fields. If gametocytes were seen, then the count was extended to 500 leucocytes.
Treatment and follow up
The patients were given the AS plus SP combination, with artesunate (4 mg/kg. day) given on days 0–2 and a single dose of SP (25 mg/kg) (Dafra Pharma, Beerse, Belgium) given on day 0. The tablets were crushed and dissolved in water for children who were not able to swallow them. Subjects were observed for vomiting for one hour; the full dose was repeated for those who vomited within 30 min and half of the dose was repeated if vomiting occurred between 30 and 60 minutes.
Follow-up and re-treatment
Patients were requested to come on days 1, 2, 3, 7, 14, 21 and 28 and at any time if they felt unwell. At each visit, body temperature was measured and blood films were prepared. During the follow-up the patients were asked if they suffered from side effects which can be expected from antimalarial treatment (nausea, vomiting, abdominal pain, dizziness and rash); these symptoms were considered to be drug related if they had not been reported at the patient's first presentation in the clinic.
Quinine was given for treatment failures. Early Treatment Failures (ETF) in case of significant parasitaemia at day 2 or 3 or parasites and fever at day 3. Late Clinical Failures (LCF) for cases with parasites and fever during follow-up after day 3 and Late Parasitological Failures (LPF) for parasite infections with/without fever during the follow-up. Cases which remained negative during follow-up were considered Adequate Clinical and Parasitological Responses (ACPR). These were modified WHO guidelines [14,15].
Statistics
Data were entered into a computer database and SPSS software (SPSS Inc., Chicago, IL, USA) was used for statistical analysis. The means (age, weight, temperature and parasite count) were calculated for all the patients and were compared between the patients in the different locations using one way analysis of the variance (ANOVA), when the data is normally distributed and by the Kruskal Wallis test if the data was not normally distributed. Percentages were calculated and compared for the patients in the four locations by an χ2 test. P < 0.05 was regarded significant.
Ethical clearance
The study received ethical clearance from the Sudanese National Malaria Administration.
Results
Two hundred and ninety (32.5%) out of 890 screened patients fulfilled the criteria and were enrolled in the study. Twenty-one (7.2%) of these were lost in the follow-up and 269 patients (72, 50, 70, and 77 from Damazin, Kassala, Kosti, and Malakal, respectively) completed the 28-day follow-up. Their different characteristics are shown in Table 1. The mean age and weight were significantly higher in the Kassala group. The parasite count was significantly higher in the Malakal area. 37.0% (100 patients) were children less than five years old; this proportion was significantly higher in Malakal group (92.2%, see Table 1). One hundred and twenty seven subjects (47.2%) were females; their percentages were not significantly different within the groups (see Table 1).
Table 1 The base line (day 0) characteristics of the 269 patients who completed the 28 days of follow-up after the treatment with artesunate plus sulfadoxine-pyrimethamine*.
Variable Total (N = 269) Damazin (N = 72) Kassala (N = 50) Kosti (N = 70) Malakal (N = 77) Significance
Age, years 12.1 (12.4) 10.6 (9.0) 24.3 (15.4) 14.7 (12.08) 3.4 (1.2) P < 0.05
Weight, Kg 28.1 (19.3) 25.6 (14.5) 47.4 (21.5) 33.2 (18.5) 3.2 (0.36) P < 0.05
Temperature, °C 38.2 (0.76) 38.1 (0.6) 38.2 (0.6) 37.9 (0.5) 38.6 (0.9) P > 0.05
Parasite count, rings/μ 25532.3 (24196.2) 21924.3 (17748.7) 32240.3 (28372.9) 11509.8 (11135.2) 37297.5 (27844.5) P < 0.05
Children < 5 years 100 (37.0) 12 (16.7) 5 (10) 12 (17.1) 71 (92.2) P < 0.05
Female 127 (47.2) 37 (51.4) 18 (36) 32 (45.7) 40 (51.9) P > 0.05
*Data were shown as mean (SD) or numbers (%) as appropriate
On day one, 60 (22.3%) patients were febrile and 15 (5.5%) patients were parasitaemic. By day three all the patients were afebrile and aparasitaemic. There were two (0.7%) Late Clinical and Parasitological Failures (days 7 and 22) from Kassala, there was no Clinical and Parasitological Failures from other locations (Table 2). Only one patient (Kassala) showed gametocytaemia on day 14 of the follow-up. Thirty two (11.9%) patients suffered expected adverse effects (nausea, itching and dizziness), but these were mild and resolved spontaneously.
Table 2 Trail profile, showing number of patients enrolled, treated and completing the 28 days of follow-up after the treatment with artesunate plus sulfadoxine-pyrimethamine*.
Variable Total Damazin Kassala Kosti Malakal
The recruited patients 290 77 53 76 84
Lost to follow-up 21 (7.2) 5 (6.5) 3 (5.6) 6 (7.9) 7 (8.3)
Patients completed the 28-days of follow-up 269 (94.8) 72 (93.5) 50 (94.4) 70 (92.1) 77 (91.7)
Accurate clinical and parasitological response 267 (99.3) 72 (100) 48 (96) 70 (100) 77 (100)
Late clinical and parasitological failure 2 (0.7) 0 (0) 2 (4) 0 (0) 0 (0)
*Data were shown as numbers (%).
Discussion
The study investigated the efficacy of AS plus SP for the treatment of uncomplicated P. falciparum malaria at four sites in the Sudan. This is probably the largest study reporting AS plus SP efficacy in Sudan until now. Although, the baseline characteristics (age and parasite count) were significantly different between the four locations, the study showed that two (0.7%, Kassala) out of 269 patients were found to have Late Clinical and Parasitological Failures. Since the parasite genotyping (PCR) was not conducted, the possibility of re-infection/recrudescence is still there. Hundred percent efficacy of AS plus SP was recently reported from eastern Sudan [8] and 99% from southern Sudan [12]. The high cure rate in this study is comparable to that reported from neighbouring African countries [5,6]. However, the highest drug resistant P. falciparum strains were reported from eastern Sudan [11,16]. The expected adverse effects (nausea, itching and dizziness) were reported in (11.9%) of the patients in this study. These results were in line with, observations of others, where the adverse effects were not significantly different, if compared with those of SP alone [5,8].
The adverse effects (nausea, vomiting) might influence the adherence to AS plus SP, especially science this therapy is only available in the oral form, which is not the medication preferred by Sudanese patients [17]. Furthermore, adherence may be influenced by the multiple doses of the combination, rather the previous single dose of SP, which was reported to be the most important single factor for the best adherence of SP among Sudanese patients [17].
A post- treatment gametocytaemia was detected in one patient in Kassala area. High (20%) levels of gametocytaemia had been reported in the eastern Sudan following SP, quinine and mefloquine treatment [18-20]. However, it has not been reported during the follow-up of patients in the eastern Sudan treated with artemether, artesunate plus mefloquine or AS plus SP [8,19,21]. The ability of artesunate to reduce the post-treatment gametocytaemia is important, as it may reduce transmission [4].
Conclusion
The study showed that, As plus SP is an effective, safe drug in the treatment of uncomplicated P. falciparum malaria in Sudan.
Authors' contributions
SBE, EMM, TA, MTM, ESA carried out the study in the different sites and participated in the statistical analysis and procedures, AHK participated in the statistical analysis, IA coordinated and participated in the design of the study, statistical analysis and the drafting of the manuscript. All the authors read and approved the final version.
==== Refs
Snow RW Guerra CA Noor AM Myint HY Hay SI The global distribution of clinical episodes of Plasmodium falciparum malaria Nature 2005 434 214 217 15759000 10.1038/nature03342
Zucker JR Ruebush TK Obonyo C Otieno J Campbell CC The mortality consequences of the continued use of chloroquine in Africa: experience in Siaya, western Kenya Am J Trop Med Hyg 2003 68 386 390 12879851
World Health Organization A global strategy for malaria control 1993 Geneva WHO
Nosten F Brasseur P Combination therapy for malaria: the way forward? Drugs 2002 62 1315 1329 12076181
Obonyo CO Ochieng F Taylor WR Ochola SA Mugitu K Olliaro P ter Kuile F Oloo AJ Artesunate plus sulfadoxine-pyrimethamine for uncomplicated malaria in Kenyan children: a randomized, double-blind, placebo-controlled trial Trans R Soc Trop Med Hyg 2003 97 585 591 15307433 10.1016/S0035-9203(03)80038-X
Priotto G Kabakyenga J Pinoges L Ruiz A Eriksson T Coussement F Ngambe T Taylor WR Perea W Guthmann JP Olliaro P Legros D Artesunate and sulfadoxine-pyrimethamine combinations for the treatment of uncomplicated Plasmodium falciparum malaria in Uganda: a randomized, double-blind, placebo-controlled trial Trans R Soc Trop Med Hyg 2003 97 325 330 15228253 10.1016/S0035-9203(03)90161-1
Adjuik M Babiker A Garner P Olliaro P Taylor W White N Artesunate combinations for treatment of malaria: meta-analysis Lancet 2004 363 9 17 14723987 10.1016/S0140-6736(03)15162-8
Adam I A/Elbasit I Idris SM Malik EM Elbashir MI A comparison of the efficacies of artesunate plus sulfadoxine-pyrimethamine versus sulfadoxine-pyrimethamine alone, in the treatment of uncomplicated Plasmodium falciparum malaria in eastern Sudan Ann Trop Med Parasitol 2005 449 455 16004704 10.1179/136485905X36299
Malik EM Khalafalla OM Malaria in Sudan: past, present and the future Gezira Journal of Health Sciences 2004 1 47 51
Adam I Osman ME ElGhazali G Ahmed GI Gustafson LL Elbashir MI Efficacies of chloroquine, sulphadoxine-pyrimethamine and quinine for the treatment of uncomplicated Plasmodium falciparum malaria in Eastern Sudan Ann Trop Med Parasitol 2004 98 661 666 15509420 10.1179/000349804225021514
Adam I Elbashir MI Resistance of P. falciparum to antimalarial drugs in Sudan: Review Khartoum Med J 2005
van den Broek I Amsalu R Balasegaram M Hepple P Alemu E Hussein B Al-Faith M Montgomery J Checchi F Efficacy of two artemisinin combination therapies for uncomplicated falciparum malaria in children under 5 years, Malakal, Upper Nile, Sudan Malar J 2005 4 14 15730557 10.1186/1475-2875-4-14
Al Gadal AA Buck AA Malaria in the Sudan Proceedings of the Conference on Malaria in Africa 1986 Washington, DC: American Institute of Biological Sciences 156 159
World Health Organization Assessment of therapeutic efficacy of antimalarial drugs for uncomplicated falciparum malaria in areas with intense transmission Document WHO/MAL/961077 1996 Geneva: WHO
WHO Assessment and Monitoring of Antimalarial Drug Efficacy for the treatment of uncomplicated Falciparum malaria WHO/HTM/RBM/200350 2003
Salah MT Mohammed MM Himeidan YE Malik EM Elbashir MI Adam I A randomized comparison of sulphadoxine-pyrimethamine and combination of sulphadoxine-pyrimethamine with chloroquine in the treatment of uncomplicated falciparum malaria in Eastern Sudan Saudi Med J 2005 26 147 148 15756376
Abdel-Humeed AA Malaria case management at the community level in Gezira Sudan Afr J Med Sc 2001 30 43 46
Adam I Ibrahim MH A/Elbasit I Elbashir MI Efficacy of sulphadoxine-pyrimethamine in the treatment of uncomplicated Plasmodium falciparum malaria in a small sample of Sudanese children East Mediterr Health J 2004
Adam I A/Elbasit I Elbashir MI Efficacies of mefloquine alone and artesunate followed by mefloquine for the treatment of uncomplicated Plasmodium falciparum malaria in eastern Sudan Ann Trop Med Parasitol 2005 99 111 117 15814029 10.1179/136485905X19801
Adam I Idris H Mohamed-Ali AA A/Elbasit IA Elbashir MI Quinine versus artemether in the treatment of severe falciparum malaria in Sudanese children East Afr Med J 2002 79 621 625 12683344
Adam I Salih I Elbashir MI Quinine for the treatment of uncomplicated Plasmodium falciparum malaria in eastern Sudan Trans R Soc Trop Med Hyg 2005 90 736 738 16084548 10.1016/j.trstmh.2005.05.005
|
16162284
|
PMC1242247
|
CC BY
|
2021-01-04 16:24:13
|
no
|
Malar J. 2005 Sep 14; 4:41
|
utf-8
|
Malar J
| 2,005 |
10.1186/1475-2875-4-41
|
oa_comm
|
==== Front
Malar JMalaria Journal1475-2875BioMed Central London 1475-2875-4-441617429510.1186/1475-2875-4-44ResearchFitness consequences of Anopheles gambiae population hybridization Menge David M [email protected] Tom [email protected] Daibin [email protected] Aditi [email protected] Goufa [email protected] John C [email protected] Louis [email protected] Guiyun [email protected] Department of Biological Sciences, State University of New York at Buffalo, Buffalo, NY 14260, USA2 Mbita Point Field and Training Station, International Center of Insect Physiology and Ecology, P.O. Box 30772, Nairobi, Kenya3 Department of Epidemiology and Public Health, University of Miami, Miami, FL 33177, USA4 Institut de Recherche pour le Développement (IRD) UR 016/ LIN, 911 avenue Agropolis, BP 64501 34394 Montpellier Cedex 5, France2005 20 9 2005 4 44 44 29 5 2005 20 9 2005 Copyright © 2005 Menge et al; licensee BioMed Central Ltd.2005Menge et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The use of transgenic mosquitoes with parasite inhibiting genes has been proposed as an integral strategy to control malaria transmission. However, release of exotic transgenic mosquitoes will bring in novel alleles along with parasite-inhibiting genes that may have unknown effects on native populations. Thus it is necessary to study the effects and dynamics of fitness traits in native mosquito populations in response to the introduction of novel genes. This study was designed to evaluate the dynamics of fitness traits in a simulation of introduction of novel alleles under laboratory conditions using two strains of Anopheles gambiae: Mbita strain from western Kenya and Ifakara strain from Tanzania.
Methods
The dynamics of fitness traits were evaluated under laboratory conditions using the two An. gambiae strains. These two geographically different strains were cross-bred and monitored for 20 generations to score fecundity, body size, blood-meal size, larval survival, and adult longevity, all of which are important determinants of the vector's potential in malaria transmission. Traits were analysed using pair-wise analysis of variance (ANOVA) for fecundity, body size, and blood-meal size while survival analysis was performed for larval survival and adult longevity.
Results
Fecundity and body size were significantly higher in the progeny up to the 20th generation compared to founder strains. Adult longevity had a significantly higher mean up to the 10th generation and average blood-meal size was significantly larger up to the 5th generation, indicating that hybrids fitness is enhanced over that of the founder strains.
Conclusion
Hybridization of the two mosquito populations used in this study led to increased performance in the fitness traits studied. Given that the studied traits are important determinants of the vector's potential to transmit malaria, these results suggest the need to release genetically modified mosquitoes that have the same or very similar backgrounds to the native populations.
==== Body
Background
Malaria is one of the most fatal infectious diseases in the tropics despite continued efforts to contain it [1-3]. Manipulating vector competence to lower transmission efficiency has been proposed as a possible integral component in the control of malaria transmission [4-6]. It is expected that manipulation, such as the introduction of fertile mosquitoes transformed with anti-parasite molecules, will result in a population of mosquitoes with compromised biological ability to facilitate development and transmission of the malaria parasites [5]. Over the past several years, remarkable progress has been made in mosquito germ-line transformation and in the identification of parasite-inhibiting molecules [5,7] that can be incorporated into a genetically modified vector to render it incapable of transmitting the malaria parasite. For example, Anopheles gambiae cell lines have been successfully transformed with the Hermes element [8,9] and the Minos transposable element, bearing an exogenous gene, has been efficiently integrated into the genome of Anopheles stephensi [10]. A number of genes inhibiting malaria parasite development in the mosquito vectors have been identified, including single-chain monoclonal antibodies [11], salivary-midgut peptide [12], phospholipase A2 [13] and mosquito innate immune genes [14,15]. It has been shown that genetically-modified mosquitoes with parasite-inhibiting genes can efficiently suppress parasite infection intensity in mosquitoes in the laboratory [5,14]. The availability of the complete genome sequence of An. gambiae provides an unprecedented opportunity to study the genetic aspects of malaria transmission and to identify new targets for transmission blocking [17].
The success of the transgenic mosquito approach will depend on the spread and even the fixation of parasite-inhibiting genes into natural vector populations [4,6]. Despite the progress in mosquito genetic transformation and the isolation of parasite-inhibiting genes, little has been done to determine how the fertile genetically-transformed mosquitoes should be released or how fast the introduced genes will spread in natural populations. Previous studies suggest that 'gene-driving' mechanisms are needed to accelerate the spread and fixation of introduced genes in natural populations of competent vectors [18-20]. These genes can be driven to fixation either using transposable elements or through Wolbachia-induced cytoplasmic incompatibility [21]. However, no matter which gene-driving mechanism is used, releasing transgenic mosquitoes in nature will require careful laboratory work and controlled field studies to determine the rate of gene-spread and to assess biosafety issues [19]. Thus, before any transgenic mosquitoes are released to the environment, the dynamics of released genes should be well understood. The mosquitoes to be released are likely to be transgenes of recent anopheline colonies from native populations; alternatively, exotic transgenic mosquitoes may be released. In either scenario, novel alleles in addition to parasite-inhibiting genes may be introduced to the environment, and it is necessary to evaluate and understand the dynamics of these novel alleles. The fitness and feeding behaviours of hybrids between the introduced and native mosquito populations should be evaluated since these hybrids may become an increased nuisance, if they bite humans more vigorously or survive longer than native mosquitoes.
This paper reports the results of an experiment designed to evaluate the dynamics of genes under laboratory conditions using An. gambiae Mbita strain from western Kenya and Ifakara strain from Tanzania. These two strains were crossbred and monitored up to the 20th generation to determine: a) if the dynamics of gene frequency from either strain can be predicted based on selected fitness traits and b) at what generation the fitness traits stabilize. Since the two founder populations are geographically different, it is likely that the hybrids' fitness will be enhanced in the initial crosses. The fitness traits studied were blood-meal size, adult body size (measured by wing length), fecundity, larval survivorship, and adult longevity. These traits are important indicators of vector fitness as discussed in Yan et al [18].
Methods
Mosquito rearing and maintenance
Two strains (Mbita and Ifakara) of laboratory-reared An. gambiae mosquitoes were used in this study. The Mbita strain was originally collected at Mbita Point (000 25'S, 340 13'E) in west Kenya and has been maintained in the laboratory since 1999. The Ifakara strain was originally collected in Njage village, 70 km from Ifakara, southeastern Tanzania, and has been maintained under laboratory conditions since 1996. The colonized populations did not exhibit a significant reduction in the observed heterozygosity in six microsatellite markers, including AGXH1D1 and AGXH131 of chromosome X, AG2H46 and AG2H79 of chromosome 2, and AG3H29C and AG3H33C of chromosome 3 (D. Zhong and G. Yan, unpublished data).
The populations of the two parent strains used in this study were raised from mosquito eggs obtained from the existing colonies maintained at the International Centre for Insect Physiology and Ecology (ICIPE), Mbita Point Field Station in the mosquito-rearing insectary. Post- mated females were allowed to engorge from a volunteer's arm for 15 minutes to facilitate oviposition. Fully engorged females were isolated from non-engorged ones and placed in single oviposition cups. Each oviposition cup contained a wet Whatman filter paper disc at the bottom and a strip of the same Whatman filter paper leaning against the walls of the oviposition cup for the mosquito to perch on. During the oviposition period, the mosquitoes were maintained on a diet of glucose solution (6%) provided by a wick of cotton placed on top of the mesh covering the oviposition cup.
The oviposited eggs were collected from the wet filter paper disks in the oviposition cups and transferred to plastic containers of distilled water. Upon hatching, the first in-star larvae were transferred to plastic trays measuring 21 cm in diameter. Larval densities were maintained at 300 larvae per tray in distilled water maintained at a depth of 8 cm. Trays containing larvae were fed daily on 30 mg of Tetramin® fish food and were subjected to natural ambient light. On pupation, the pupae were moved to standard 30 × 30 × 30 cm netting cages. After emergence, adult mosquitoes were held in cages and were offered 6% glucose and distilled water by means of moistened cotton wick placed on top of the mesh on the cages. The cages were kept in the insectary at ambient tropical temperature and under artificial light provided by fluorescent tubes. Relative humidity was maintained at ambient by basins of water placed in the insectary.
Hybrids of the F1 generation from the two strains were obtained mixing 100 Ifakara females, 100 Ifakara males, 100 Mbita females, and 100 Mbita males in a single 30 × 30 × 30 cm cage. The F1 hybrid larvae were reared to adults under similar conditions. Subsequent generations from F2 to F20 were obtained by mass mating and reared under similar conditions with a population size maintained at approximately 1,000 individuals.
Measurement of fitness components
The fitness of mosquitoes sampled from cage populations of the two single strains, Mbita and Ifakara (F0), and the hybrids between the two strains at the F1, F5, F10, F15, and F20 generations, was measured. In order to confirm the direction and effect of heterosis and also to control any assortative mating of mixed populations, reciprocal matings of Mbita male and Ifakara female and vice versa were was carried out, and their F1 and F5 progeny were subjected to measurement of fitness traits. The fitness components studied included mosquito fecundity, body size (measured by wing length), blood-meal size, larval survival, and adult longevity. All of these fitness traits were measured in two replicate experiments. Because the traits measured in this study could be sensitive to environmental parameters, the populations and filial generations used were reared in the same manner as previously described. The differences observed among the two founder populations and their different filial generations are therefore likely to be due to their genetic differences.
(i) Fecundity
Fecundity is an important fitness trait as it indicates the efficiency of conversion of the blood meal to eggs and it also influences the number of offspring from a single female mosquito that are likely to survive to become adults [22,23]. Fecundity was scored as the number of eggs laid plus any eggs retained in a single gonotrophic cycle. Females were fed as previously described, and gravid females (n = 50 in two replicates) were then transferred to individual glass tubes for oviposition. The number of eggs laid by each individual was recorded. Any female who died before or after oviposition was dissected, and any retained eggs were counted. In most cases, no eggs were retained. When eggs were retained, those eggs contributed less than 1% of the total number of eggs in that gonotrophic cycle.
(ii) Wing length
Wing length is a reliable correlate of mosquito body size [22,23]. As a fitness trait, mosquito size may influence survivorship, developmental time, and the ability to acquire a blood meal; it also has a positive correlation to fecundity [23]. Wing length was measured from the same individuals (n = 50 for two replicates) used to measure fecundity. One wing was removed at the time of dissection, and wing length was measured from the axial incision of the apical margin as described by Kelly and Edman [23]. Each wing was mounted on a glass microscope slide in a small drop of distilled water. Wing length was measured to the nearest 0.01 mm using a compound microscope.
(iii) Blood-meal size
Blood-meal size influences both mosquito fecundity and the potential to acquire pathogens [24,25]. Indirect quantification of blood-meal size was conducted based on the principle of haemoglobinometry using the HiCN method [26]. Starved three-day-old females were allowed to feed from a volunteer's arm for 15 minutes. Blood-fed females (n = 50 for two replicates) were immediately frozen and thoroughly ground in individual tubes; Drabkins reagent was then added. This reagent converts all haemoglobin (Hb) to a cyano-derivative whose color intensity is proportional to total Hb concentration in the blood-fed mosquito. This mixture was then vortexed and centrifuged at 1,500 rpm for five minutes. The supernatant was transferred to a test tube with 800 μl distilled water and vortexed briefly. The absorbance was read at 415 nm using a Chemlab Spectronic 20D spectrophotometer.
A standard curve was prepared for each of the assays done on the Mbita, the Ifakara, and their filial generations using known volumes of human blood (0.8, 1.6, 2.4, 3.2, and 4.0 μl) and measuring the corresponding absorbance. The absorbance of unfed female mosquitoes (n = 5) was also measured, and the average of their absorbance reading was used as a correction factor for the absorbance obtained from blood-fed females.
(iv) Larval and adult survivorship
Larval survivorship has a significant influence on the number of pupae and adult mosquitoes that emerge. Adult longevity, on the other hand, influences the number of gonotrophic cycles and the potential for pathogen transmission. For determining larval survivorship, 50 first in-star larvae of the founder populations and their filial generations were transferred into distilled water maintained at a depth of 8 cm in plastic trays measuring 21 cm in diameter. Trays containing larvae were fed daily with 30 mg of Tetramin® fish food, and were subjected to natural ambient light, and monitored daily. Dead larvae were recorded and removed; larvae that pupated were also recorded, removed, and then transferred to emergence cages. Mosquito adult survivorship of the two founder populations and their respective filial generations were examined. In each case, 50 individuals were put in a cage in the insectary and maintained on a glucose diet, as previously described. Mosquitoes were examined daily and dead individuals were counted and removed until the day the last individual died.
Data analysis
The original fitness traits data were analysed using analysis of variance (ANOVA) for fecundity, body size, and blood-meal size. Larval and adult survival were subjected to survival analysis to compare the performance of phenotype traits between the founder parents, reciprocal crosses and each filial generation. Comparisons were also made between successive filial generations to infer the level at which maximum genotype hybridization was achieved. The significance of all statistical tests was set at 0.05. Statistical analyses were performed in MINITAB (Minitab Inc.) and SPSS (SPSS Inc.) computer software. The dynamics of population hybridization were also simulated in random-mating generations founded by Mbita strain (A) and Ifakara strain (B), using p(ABt+1) = 1 - p(At+1) - p(Bt+1), where p(ABt+1) is the proportion of hybrids between the two strains at generation t, p(At+1) is the proportion of Mbita homozygotes at generation t, and p(Bt+1) is the proportion of Ifakara homozygotes at generation t.
Results
Fecundity
There was a significant difference in fecundity (p < 0.001) between the Mbita and Ifakara strains, with the Ifakara strain having at least 20% more eggs than the Mbita strain in the two replicate measurements (Table 1). There was also a significant difference (p < 0.001) in fecundity between the parental strains and their hybrid progeny at F1, F5, F10, F15, and F20 in the two replicates studied in this experiment. The hybrid progeny had a higher mean, suggesting a high degree of heterosis for this trait. A pair-wise analysis of variance between F1 and F5, F5 and F10, F10 and F15, and F15 and F20 showed no significant difference in fecundity (results not shown). There was a significant difference in the fecundity of F1 and F10 (p < 0.05), but no significant difference between F1 and F15 or F15 and F20 (results not shown). Likewise, there was no significant difference in fecundity between F5 and F20 or F10 and F20.
Table 1 Summary of fitness traits for Mbita and Ifakara strains, and filial generations of hybridization between the two strains (100 Ifakara females, 100 Ifakara males, 100 Mbita females, and 100 Mbita males).
Trait Replicate Mbita Ifakara F1 F5 F10 F15 F20
Wing length (mm) 1 2.92 (0.02) 2.96 (0.02) 3.10 (0.02) 3.09 (0.03) 3.11(0.02) 3.14(0.03) 3.03(0.16)
2 2.87 (0.03) 2.89 (0.03) 3.17 (0.03) 3.34 (0.03) 2.98(0.02) 3.07(0.03) 2.95(0.02)
Fecundity* 1 47.17(2.66) 58.25(2.29) 63.37(2.28) 64.58(2.18) 70.26(2.05) 65.08(2.02) 68.90 (1.83)
2 35.96(2.56) 49.28(2.67) 62.60(2.95) 65.42(2.32) 76.97(2.80) 63.60(2.23) 61.44(2.38)
Blood-meal size (μl) 1 3.15(0.14) 2.54(0.08) 3.96(0.26) 2.78(0.15) 2.53(0.23) 3.10(0.15) 3.13(0.16)
2 2.93 (0.18) 2.61 (0.11) 4.51 (0.21) 4.02 (0.16) 2.97 (0.20) 3.19 (0.17) 3.21 (0.13)
Larval survivorship (%) 1 91.78 94.44 98.00 95.22 97.00 95.78 95.88
2 96.44 92.54 98.34 93.34 96.66 93.44 95.78
Mean adult longevity (days) 1 22.31 14.52 20.03 29.00 31.81 20.71 16.95
2 19.61 15.66 19.97 21.66 33.72 16.38 18.64
Note: The values in parenthesis are standard errors. * Fecundity is the number of eggs laid plus the number of eggs retained.
Wing length
No significant difference in wing length was found between the Mbita and Ifakara strains (Table 1). However, there was a consistent significant difference between the two founder strains and their F1, F5, F10, F15, and F20 progeny in the two replicates (p < 0.05). This can be attributed to heterosis for this trait. The inter-progeny difference was significant between F1 and F5 (p < 0.05) but not between F5 and F10, F10 and F15, and F15 and F20(results not shown).
Blood-meal size
An analysis of variance showed a significant difference in blood-meal size between the Mbita and Ifakara strains (Table 1; p < 0.001) with the Mbita strain taking a larger blood meal. There was also a significant difference in blood-meal size between the founder populations – Mbita and Ifakara – and the F1 progeny (p < 0.05). All of the other filial generations had a significantly larger blood-meal size than the Ifakara strain (p < 0.001), whereas the F5, F10, F15, and F20 generations' blood-meal size did not differ significantly from the Mbita strain (Table 1). A pair-wise analysis of variance showed significant difference between F1 and F5 (p < 0.001), but not between F5 and F10, F10 and F15, and F15 and F20. Mean blood-meal size showed significant difference between F1 and F10, F1 and F15, and F1 and F20, whereas there was no significant difference between the means of F5 and F15, and F5 and F20. The mean blood-meal sizes of F10 and F20 were also not significantly different.
Larval survivorship and adult longevity
Adult longevity was significantly greater in the Mbita strain compared to the Ifakara strain. The mean longevity of the filial generations showed an increasing trend from F1, F5, and F10, which had the greatest longevity due to heterosis (Table 1). From F10, the mean longevity showed a decreasing trend in F15 and F20 (Table 1). Data for larval survival did not show significant consistent differences between the founder parents and the filial generations – unlike with the other traits – because more than 90% of the larvae in all experiments survived to become pupae (Table 1).
Reciprocal crosses
Data on reciprocal crosses of males and females of either Mbita or Ifakara strains showed significantly higher means of body size, fecundity, and blood-meal size at both the F1 and F5 generations than either founder strain (Table 2). This confirms that enhanced fitness in the progeny of Mbita and Ifakara strains is due to heterosis between the two genotypes.
Table 2 Means of fitness traits for Mbita and Ifakara strains and their respective female/male crosses
Strain/Population Wing length in mm (standard error) Fecundity* (standard error) Blood-meal size (μl) (standard error)
Mbita 2.86 (0.03) 35.97 (2.56) 3.15 (0.14)
Ifakara 2.89 (0.03) 49.27 (2.67) 2.54 (0.08)
F1 from Mbita female × Ifakara male 3.13 (0.03) 74.87 (2.63) 4.04 (0.16)
F1 from Ifakara female × Mbita male 3.03 (0.02) 83.20 (5.36) 4.51 (0.21)
F5 from Mbita female × Ifakara male 2.94 (0.02) 68.93 (2.61) 4.02 (0.16)
F5 from Ifakara female × Mbita male 3.14 (0.02) 68.77 2.69) 3.38 (0.18)
* Fecundity is the number of eggs laid plus the number of eggs retained.
Population hybridization
Computer simulation indicates the proportion of hybrids between Mbita and Ifakara strains increases continuously and reaches to 100% after 10 generations (Figure 1).
Figure 1 Plot of expected proportion of hybrids against time (Generation). The simulation assumed equal sex ratio and equal number of Anopheles gambiae Mbita and Ifakara strains in random mating, as in the experimental set up.
Discussion
In this study, the fitness consequences of population hybridization were examined in 20 filial generations from crosses between two geographically different An. gambiae strains. Four fitness components were examined: fecundity, body size, blood-meal size and adult survival. These fitness traits are important determinants of the vector's potential to transmit the malaria parasites (Plasmodium species) and should, therefore, be considered in strategies aimed at controlling malaria transmission through the introduction of exotic and genetically modified mosquitoes. The results of this study show that fitness traits are subject to the effects of heterosis, i.e., hybrids have increased values for fitness traits than do the parent populations. Mosquito body size (measured as wing length) and fecundity showed heterosis up to the 20th generation. In both replicates, the 20th generation had significantly higher values (at least 5% longer wing length and 20% more eggs) than either parent. Blood-meal size also showed heterosis, with the F1 and F5 generations imbibing at least 50% more blood than the Ifakara strain. Subsequent generations were not significantly different from the Mbita strain, but showed significantly greater blood-meal size than the Ifakara strain. Adult longevity showed equally robust heterosis in the hybrid generations. The F1, F5, and F10 generations had, respectively, 40%, 66% and 100% higher mean longevity than the founding Ikafara strain; the 15th and 20th generations had mean longevity that was between the two founder strains but was higher than the Ifakara strain. Further analysis indicated that the maximum heterosis for body size, fecundity, and adult longevity occurred at the 10th generation, after which there was no significant pair-wise difference.
These results can be attributed to increasing hybridization of the Ifakara and Mbita strains. In the experimental design used in this study, 100 adult mosquitoes of each sex from the two strains were put together to mate. In the first generation, there were more (Mbita × Mbita) and (Ifakara × Ifakara) matings. In subsequent generations, the genotypes mating were Mbita, Ifakara, and intercrosses of the founder populations, and therefore the probability of within-strain mating (Mbita × Mbita; or Ifakara × Ifakara) reduced from generation to generation. The increasing trends of heterosis in the studied traits were due to increasing hybridization of the Mbita and Ifakara strains, as demonstrated in Figure 1. On attaining maximum hybridization, heterosis is maximally expressed, as reflected by the performance of the fitness traits. As would be expected [27], this is then followed by eventual decline in performance due to the waning effect of heterosis.
Conclusion
The results of this study show that hybridisation of different Anopheles gambiae populations lead to enhanced fitness than the parental populations. In the proposed use of transgenic mosquitoes as a strategy to control malaria transmission, it is expected that transgenic mosquitoes with the inherent ability to resist malaria parasite development will be released into the field [4-6]. By cross-mating with native wild populations, transgenes are expected to spread out and be fixed in the native wild population to render it incapable of supporting parasite development and transmission [20]. If the transgenic mosquitoes are exotic to the region, the possibility exists that hybrids will have greater fitness than native mosquitoes. As shown by the experiments done in this study, the progeny of the Mbita and Ifakara strains have consistently higher fecundity and body size than either of the founder parents up to the 20th generation. If exotic genetically modified mosquitoes are introduced, it is, therefore, probable that the hybrids may live longer, exhibit higher fecundity, have larger body size and will likely feed more. Increased hybrid fitness would lead to stabilizing selection, rendering the refractory genes more difficult to be fixed. Given that greater mosquito density, fecundity, human biting habits and longevity of anopheline mosquito vectors are positively correlated with increased vectorial capacity [28,29], these results strongly indicate the need to release genetically modified mosquitoes that have the same or very similar genetic makeup to that of the native populations.
Authors' contributions
DM and TG conducted fitness studies and prepared the manuscript. DZ, AP and GZ analysed the data and assisted with manuscript preparation. GY, LG, and JCB conceived the design of the study. All authors read and approved the final manuscript.
Acknowledgements
We thank B. Knols, G. Killeen, and J. Githure for facilitating the work. This research was supported by NIH grant D43 TW01505 and the UNDP/WORLD BANK/WHO Special Programme for Research and Training in Tropical Diseases (TDR) grant A10429.
==== Refs
Breman JG The ears of the hippopotamus: manifestations, determinants, and estimates of the malaria burden Am J Trop Med Hyg 2001 64 1 11 11425172
Gallup JL Sachs JD The economic burden of malaria Am J Trop Med Hyg 2001 6 84 96
Breman JG Alilio M Mills A Conquering the intolerable burden of malaria: What's new, What's needed: A summary Am J Trop Med Hyg 2004 71 1 15 15331814
Clarke T Mosquitoes minus malaria Nature 2002 419 429 430 12368830 10.1038/419429a
Ito J Ghosh A Moreira LA Wimmer EA Jacobs-Lorena M Transgenic anopheline mosquitoes impaired in transmission of a malaria parasite Nature 2002 417 452 455 12024215 10.1038/417452a
Moreira LA Jing W Frank HC Jacobs-Lorena M Fitness of anopheline mosquitoes expressing transgenes that inhibit Plasmodium development Genetics 2004 166 1337 1341 15082552 10.1534/genetics.166.3.1337
Catteruccia F Godfray HC Crisanti A Impact of genetic manipulation on the fitness of Anopheles stephensi mosquitoes Science 2003 299 1225 1227 12595691 10.1126/science.1081453
Jasinskiene N Coates CJ Benedicts MQ Cornel AJ Rafferty CS James AA Collins FH Stable transposon-mediated transformation of yellow fever mosquito Aedes aegypti, using the Hermes element from the housefly Proc Natl Acad Sci USA 1998 95 3743 3747 9520437 10.1073/pnas.95.7.3743
Zhao Y Eggleston P Stable transformation of Anopheles gambiae cell line mediated by the Hermes mobile genetic element Insect Biochem Mol Biol 1998 28 213 219 10.1016/S0965-1748(98)00020-4
Catterucia F Nolan T Loukeris TS Blass C Savakis C Kafatos FC Crisanti A Stable germline transformation of the malaria mosquito Anopheles stephensi Nature 2000 405 959 962 10879538 10.1038/35016096
de L Capurro L Coleman J Beernstsen BT Myles KM Olson KE Rocha E Krettli AU James AA Virus expressed, recombinant single-chain antibody blocks sporozoite infection of salivary glands in Plasmodium gallinaceum-infected Aedes aegypti Am J Trop Med Hyg 2000 4 427 433
Ghosh AK Ribolla PE Jacobs-Lorena M Targeting Plasmodium ligands on mosquito salivary glands and midgut with a phage display peptide library Proc Natl Acad Sci USA 2001 98 13278 13281 11687659 10.1073/pnas.241491198
Zieler H Keister DB Dvorak JA Ribeiro JM A snake venom phospholipase A(2) blocks malaria parasite development in the mosquito midgut by inhibiting ookinete association with the midgut surface J Exp Biol 2001 204 4157 4167 11809789
Christophides GK Zdobnov E Barillas-Mury C Birney E Blandin S Blass C Brey PT Collins FH Danielli A Dimopoulos G Hetru C Hoa NT Hoffman JA Kanzok SM Letunic I Levashina EA Loukeris TG Lycett G Meister S Michel K Moita LF Muller M Osta MA Paskewitz SM Reichhart JM Rzhetsky A Troxler L Vernick KD Vlachou D Volz J Mering C Xu J Zheng L Bork P Kafatos FC Immunity-related genes and gene families in Anopheles gambiae Science 2002 298 159 165 12364793 10.1126/science.1077136
Osta MA Christophides GK Kafatos FC Effects of mosquito genes on Plasmodium development Science 2004 303 2030 2032 15044804 10.1126/science.1091789
James AA Blocking malaria parasite invasion of mosquito salivary glands J Exp Biol 2003 206 3817 3821 14506217 10.1242/jeb.00616
Holt RA Subramanian GM Halpern A Sutton GG Charlab R Nusskern DR Wincker P Clark AG Ribeiro JM Wides R Salzberg SL Loftus B Yandell M Majoros WH Rusch DB Lai Z Kraft CL Abril JF Anthouard V Arensburger P Atkinson PW Baden H de Berardinis V Baldwin D Benes V Biedler J Blass C Bolanos R Boscus D Barnstead M Cai S Center A Chaturverdi K Christophides GK Chrystal MA Clamp M Cravchik A Curwen V Dana A Delcher A Dew I Evans CA Flanigan M Grundschober-Freimoser A Friedli L Gu Z Guan P Guigo R Hillenmeyer ME Hladun SL Hogan JR Hong YS Hoover J Jaillon O Ke Z Kodira C Kokoza E Koutsos A Letunic I Levitsky A Liang Y Lin JJ Lobo NF Lopez JR Malek JA McIntosh TC Meister S Miller J Mobarry C Mongin E Murphy SD O'Brochta DA Pfannkoch C Qi R Regier MA Remington K Shao H Sharakhova MV Sitter CD Shetty J Smith TJ Strong R Sun J Thomasova D Ton LQ Topalis P Tu Z Unger MF Walenz B Wang A Wang J Wang M Wang X Woodford KJ Wortman JR Wu M Yao A Zdobnov EM Zhang H Zhao Q Zhao S Zhu SC Zhimulev I Coluzzi M della Torre A Roth CW Louis C Kalush F Mural RJ Myers EW Adams MD Smith HO Broder S Gardner MJ Fraser CM Birney E Bork P Brey PT Venter JC Weissenbach J Kafatos FC Collins FH Hoffman SL The genome sequence of the malaria mosquito Anopheles gambiae Science 2002 298 129 149 12364791 10.1126/science.1076181
Yan G Serverson DW Christensen BM Costs and benefits of mosquito refractoriness to malaria parasites: Implications for the genetic variability of mosquitoes and genetic control of malaria Evolution 1997 5 441 450
Braig HR Yan G The spread of genetic constructs in natural insect populations. Genetically engineered organisms: Assessing environmental and human health effects 2001 Letourneau CRS Press 251 314
Yan G Severson DW Dynamics of molecular markers linked to the resistance loci in a mosquito-Plasmodium system Genetics 2003 164 511 519 12807772
Dobson SL Marsland EJ Rattanadechakul W Wolbachia induced cytoplasmic incompatibility in single and superinfected Aedes albopictus (Diptera: Culicidae) J Med Entomol 2001 38 382 387 11372962
Ward RA Genetics aspects of the susceptibility of mosquitoes to malaria infection Exp Parasitol 1963 13 328 341 10.1016/0014-4894(63)90084-5
Kelly R Edman JD Mosquito size and multiple transmission of avian malaria in the laboratory J Am Mosq Control Assoc 1992 8 386 388 1474384
Woke PA Ally MS Rosenberger CR The number of eggs developed related to the quantities of human blood ingested in Aedes aegypti (L) Ann Entomol Soc Am 1956 49 435 441
Kershaw WE Lavoipierre JMM Beesley WN Studies on the intake of microfilariae by their insect vectors. VII. Further observations on the intake of microfilariae of Dirofilaria immitis by Aedes aegypti in laboratory conditions: The pattern of the intake of a group of flies Ann Trop Med Parasitol 1955 49 203 211 13239083
Briegel H Lea AO Klowden MJ Hemoglobinometry as a method for measuring blood meal sizes of mosquitoes J Med Entomol 1979 15 235 238
Shull GH John Gowan Beginnings of the heterosis concept 1952 Heterosis Ames: Iowa State College Press 14 48
Bruce CLJ Essentials of malariology 1985 Second New York: John Wiley and Sons
Molineaux L The epidemiology of human malaria as an expression of its distribution including some implications for its control Malaria, principles and practice of malariology 1988 2 London: Churchill Livingston 913 998
|
16174295
|
PMC1242248
|
CC BY
|
2021-01-04 16:26:57
|
no
|
Malar J. 2005 Sep 20; 4:44
|
utf-8
|
Malar J
| 2,005 |
10.1186/1475-2875-4-44
|
oa_comm
|
==== Front
Mol CancerMolecular Cancer1476-4598BioMed Central London 1476-4598-4-341615689810.1186/1476-4598-4-34ResearchGlobal gene expression profiling of cells overexpressing SMC3 Ghiselli Giancarlo [email protected] Chang-Gong [email protected] Department of Pathology and Cell Biology, Thomas Jefferson University, 1020 Locust Street, Philadelphia, PA 19107, USA2 Kimmel Cancer Center, Thomas Jefferson University, 1020 Locust Street, Philadelphia, PA 19107, USA3 Department of Microbiology and Immunology, Thomas Jefferson University, 1020 Locust Street, Philadelphia, PA 19107, USA2005 12 9 2005 4 34 34 20 10 2004 12 9 2005 Copyright © 2005 Ghiselli and Liu; licensee BioMed Central Ltd.2005Ghiselli and Liu; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The Structural Maintenance of Chromosome 3 protein (SMC3) plays an essential role during the sister chromatid separation, is involved in DNA repair and recombination and participates in microtubule-mediated intracellular transport. SMC3 is frequently elevated in human colon carcinoma and overexpression of the protein transforms murine NIH3T3 fibroblasts. In order to gain insight into the mechanism of SMC3-mediated tumorigenesis a gene expression profiling was performed on human 293 cells line stably overexpressing SMC3.
Results
Biotinylated complementary RNA (cRNA) was used for hybridization of a cDNAmicroarray chip harboring 18,861 65-mer oligos derived from the published dEST sequences. After filtering, the hybridization data were normalized and statistically analyzed. Sixty-five genes for which a putative function could be assigned displayed at least two-fold change in their expression level. Eighteen of the affected genes is either a transcriptional factor or is involved in DNA and chromatin related mechanisms whereas most of those involved in signal transduction are members or modulators of the ras-rho/GTPase and cAMP signaling pathways. In particular the expression of RhoB and CRE-BPa, two mediators of cellular transformation, was significantly enhanced. This association was confirmed by analyzing the RhoB and CRE-BPa transcript levels in cells transiently transfected with an SMC3 expression vector. Consistent with the idea that the activation of ras-rho/GTPase and cAMP pathways is relevant in the context of the cellular changes following SMC3 overexpression, gene transactivation through the related serum (SRE) and cAMP (CRE) cis-acting response elements was significantly increased.
Conclusion
We have documented a selective effect of the ectopic expression of SMC3 on a set of genes and transcriptional signaling pathways that are relevant for tumorigenesis. The results lead to postulate that RhoB and CRE-BPa two known oncogenic mediators whose expression is significantly increased following SMC3 overexpression play a significant role in mediating SMC3 tumorigenesis.
==== Body
Introduction
The Structural Maintenance of Chromosome 3 protein (SMC3) is a key component of the nuclear multimeric protein complex named cohesin. This complex, which also includes SMC1, scc1 and scc3, forms joints between the replicating DNA strands and holds together the sister chromatids throughout G2 phase while opposing the splitting force exerted by the spindle microtubules [1]. In addition to its essential role in mitotic and meiotic chromosome segregation, SMC3 plays an important role in DNA recombination [2], is a component of the DNA damage repair mechanism [3] and is involved in the microtubule-based intracellular transport [4]. SMC3 expression is elevated in a large fraction of human colon carcinoma and in the intestinal tumors of mice genetically prone to develop polyps [5]. SMC3 expression level is controlled in intestinal epithelial cells through the APC/β-catenin/TCF4 transactivation pathway a signaling system that is almost invariably altered in colon carcinomas [6]. Furthermore NIH3T3 fibroblasts overexpressing SMC3 lose cell-cell contact inhibition, display anchorage-independent growth and form foci of transformation [5]. These findings support the idea that up-regulation of SMC3 expression is either permissive or sufficient to trigger cell transformation. The mechanism of SMC3-mediated cell transformation has however remained speculative.
In order to identify genes whose expression is affected by SMC3 overexpression, high-density oligonucleotide microarray chip harboring 18,861 human gene-specific oligonucleotides were hybridized with cRNA derived from 293 cells with different expression level of SMC3. The 293 cells are human embryonic kidney cells that have become immortalized following transformation by adenovirus type 5 [7] and display latent tumorigenicity [8]. This represent a well characterized model for human tumorigenesis that has been frequently utilized for in vitro and in vivo assessment of the oncogenic or tumor suppressor potential of a number of genes [9-13]. Statistical analysis of the microarray data has revealed that many of the genes affected by SMC3 overexpression in 293 cells are members or modulators of the ras-rho/GTPase family of proteins and of the cAMP signaling pathway. The analysis of the activity of a panel of reporter vectors monitoring different transactivation pathways further corroborates the idea that ras-rho/GTPase and cAMP response element binding proteins play a predominant role in orchestrating the cell changes subsequent to SMC3 overexpression. In particular RhoB and CRE-BPa, two major modulators of cellular transformation and response to genotoxic stress and whose level is significantly increased following SMC3 overexpression, may act as important mediators of SMC3 activity at cellular level.
Results and Discussion
A microarray analysis of the genome-wide effect of SMC3 overexpression identifies candidate genes mediating SMC3 tumorigenicity
The identification of genes that are affected by SMC3 up-regulation may provide important clues regarding the biology of this cohesin protein and shed light on the mechanism at the basis of the SMC3-induced tumorigenesis. Toward this end the changes in gene expression caused by sustained SMC3 overexpression were analyzed in fetal kidney 293 cells using a large microarray of human gene-specific oligonucleotides. Stable overexpression of SMC3 in these cells was assessed by semi-quantitative RT-PCR and the result confirmed in cells at a later division stage by Western immunoblotting (fig. 1). On the average we detected ~3-fold elevation of the SMC3 mRNA and protein levels. These values compare well with the 2-folds elevation of SMC3 transcript level measured in the microarray analysis (Table I). Of relevance is the fact that these changes in expression level are similar to those previously detected in a series of human colon carcinomas and in the intestinal polyps of APC+/min mice or in NIH3T3 cells genetically engineered to ectopically express SMC3 [5]. A comparable increase in SMC3 expression has also been detected in liver metastatic cancer cells [14], in marrow stem cells exiting quiescence [15], in vascular endothelial cells following angiogenic stimulus [16], and in human T-cells infected with Varicella-Zoster [17] suggesting that maximal achievable SMC3 expression level is similar in different cell and tissues contexts. Interestingly these changes in transcriptional activity are not accompanied by significant changes in the expression of the other components of the cohesin complex consistent with the idea that SMC3 level is regulated independently from that of the other partner proteins.
Figure 1 Identification and characterization of 293 cells stably overexpressing human SMC3. a): 293 cells were transfected with SMC3-pcDNA3.1 expression vector or the empty vector alone. Cell clones with stably integrated vector were selected in medium containing 500 μg/ml of G418 for three weeks. The expression of SMC3 and of the housekeeping gene G3DPH in drug-resistant clones was evaluated by semi-quantitative RT-PCR. b): SMC3 expression analysis was repeated on cells at later passages to confirm that clones stably overexpressing SMC3 were used. This time SMC3 expression was examined by Western immunoblotting. For this purpose cells were solubilized in lysis buffer and analyzed by 8% SDS-PAGE. After transfer of the proteins to nitrocellulose, SMC3 was immunodetected using a goat polyclonal antibody followed by ECL detection of the immunocomplex. The expression of α-tubulin was used to assess the inter-sample variability using nitrocellulose filters stripped of the SMC3 immunocomplexes.
Table 1 Significantly regulated genes in SMC3 overexpressing 293 cells. Human expressed sequence tags and genes with no known function are not included. Complete results of the array have been submitted to the EBI microarray database.
Genes SAM Score Fold changes
Bamacan/SMC3 2.06 2.04
Extracellular Matrix
GPC3, glypican 3 2.55 2.49
lectin, mannose-binding, 1 like 2.05 2.13
HAS2, hyaluronan synthase 2 1.89 2.03
LAMB3, lamin, beta 3 -1.99 -3.76
Transporter and ion channels
ATP6V1B1, ATPase, H+ transporting V1 subunit B, kidney isoform 3.15 2.40
KCTD12, potassium channel tetramerisation domain 12 1.85 2.02
SLC7A10, solute carrier family 7 member 10, neutral amino acid transporter -2.00 -2.00
Metabolism
MIG12, MID1 interacting G12-like protein 2.58 2.65
CKB, creatine kinase, B chain 2.00 2.28
ECHDC3, enoyl Coenzyme A hydratase domain containing 3 2.25 2.20
ALDH1A3, aldehyde dehydrogenase 1 family, member A3 1.94 2.16
MOXD1, monooxygenase, DBH-like 1 2.53 2.05
TXNRD3, thioredoxin reductase 3 1.81 2.00
Growth factors and receptors
LTBP2, latent transforming growth factor beta binding protein 2 1.95 2.68
GABRE, gamma-aminobutyric acid A receptor, epsilon 2.04 2.54
DLL1, delta-like 1, notch ligand 1.83 2.23
GDF9, growth differentiation factor 9 1.88 2.19
IGFBP7, insulin-like growth factor binding protein 7 1.81 2.15
GNRH1, gonadotropin-releasing hormone 1 2.23 2.04
IRAK1, interleukin-1 receptor associated kinase 1 -2.17 -2.13
EGFL3, EGF-like-domain, multiple 3 -2.20 -2.16
IL22RA1, interleukin 22 receptor, alpha 1 -2.22 -2.48
OR4D1, olfactory receptor, family 4, subfamily D, member 1 -1.91 -3.03
GRB7, growth factor receptor-bound protein 7 -2.20 -4.18
Signal Transduction
ARHGEF4, Rho guanine nucleotide exchange factor 4 2.36 2.85
RGS14, regulator of G-protein signaling 14 2.17 2.57
RPS6KA5, ribosomal protein S6 kinase, polypeptide 5 2.09 2.46
PAK6, p21-activated kinase 6 2.16 2.38
RIN3, Ras and Rab interactor 3 1.86 2.19
RHOB, ras homolog gene family, member B 1.81 2.18
SCFD1, sec1 family domain containing 1 2.31 2.08
LRRK1, leucine-rich repeat kinase 1 2.16 2.06
RAB40B, GTP-binding protein 40B 1.91 2.15
PDE6B, phosphodiesterase 6B, cGMP-specific 2.02 2.09
MCF2L, MCF.2 cell line derived transforming sequence -1.92 -2.16
S100A8, S 100 calcium binding protein A8, calgranulin A -1.92 -2.86
ADCY2, adenylate cyclase 2 -1.92 -3.03
Transcriptional factors
MEOX2, mesenchyme homeobox 2, growth arrest specific homeobox 2.14 3.96
IRF4, interferon regulatory factor 4 2.20 3.28
KHDRBS3, KH domain, RNA binding, signal transduction associated 3 2.24 2.62
PPARA, peroxisome proliferative activated receptor, alpha 2.22 2.50
CRE-BPa, cAMP responsive element binding protein, ATF2-like 1.94 2.37
NEK9, never-in-mitosis-gene a-related kinase 9 1.83 2.33
CREM, cAMP responsive element modulator -1.87 -2.11
NCOA6, nuclear receptor coactivator 6 -1.88 -2.32
TXB21, T-box 21 -1.88 -2.42
POU3F1, POU domain, class 3, octamer-binding TF6 -2.21 -2.44
NKX6-1, NK6 transcriptional factor related, locus 1, homeobox 6A -1.90 -2.53
CTNNA1, catenin alpha 1 -1.90 -2.57
BCL6B, B-cell CLL/lymphoma 6, member B zinc finger protein -2.72 -2.79
ZNF236, zinc finger protein 236 -1.90 -3.11
DNA repair, gene transcription
MHL3, mutL homolog 3 2.76 2.81
DNMT2, DNA cytosine-5-methyltransferase 2 1.88 2.07
ADPRTL2, ADP-ribosyltransferase 2 2.11 2.00
SNRPN, small nuclear ribonuclear polypeptide N -2.25 -2.64
Various
MYOM2, myomesin 2 2.77 5.10
NOPE, neighbor of Punc E11 2.53 2.56
C5, complement component 5 2.09 2.46
COCH, coagulation factor C homolog, cochlin 2.15 2.09
VMD2L1, vitelliform macular dystrophy 2-like 1 -2.17 -2.09
GRN, granulin -1.94 -2.14
HPN, hepsin serine protease -2.02 -2.27
LPHN1, latrophilin 1 -1.90 -2.38
PRND, prion protein 2 -2.01 -2.50
HYPM, huntingtin interacting protein M -2.18 -3.98
To ensure the accurate identification of genes whose regulation is altered in response to SMC3 overexpression, the array data were normalized and filtered to exclude gene displaying high inter-array variability [18] and then statistically analyzed with SAM [19] (fig. 2). The algorithm uncovered 114 genes that either increase (n = 70) or decreased (n = 44) their expression level by at least two-folds. Differentially expressed genes with an entry in the Unigene database were classified according to their putative main function and listed in Table I. About one-fifth of the affected genes encode secreted or cell surface proteins such as components of the extracellular matrix or cell surface receptors and growth factors. This group of genes includes Glypican 3 a heparan sulfate proteoglycan that is a cell surface co-receptor for heparin-binding growth factors and is elevated in several forms of cancer [20]. About two-third of the differentially expressed genes is engaged in signal transduction and gene transactivation. In this group there are genes involved in cAMP signal transduction such the adenylate cyclase 2, CREM and CRE-BPa [21,22]. Further examination of the list of the affected genes reveals a rather specific effect of SMC3 on the expression of genes that are effectors (RhoB and RAB40B) or modulators (ARHGEF4, RGS14, RIN3, PDE6B, ADP-rybosyltransferase) of the ras-rho/GTPase signaling pathway [23-25]. Rho is a family of small GTPases that is not mutated in cancer and therefore their involvement in tumorigenesis is dependent upon the activation status and the expression level. The up-regulation of RhoB expression may be of particular relevance in the context of SMC3-mediated tumorigenesis. RhoB is an early response gene that is transcriptionally activated following DNA damage [26] through a mechanism requiring ATF2-mediated gene transactivation [27]. ATF2 is a member of the CRE-binding protein gene family whose transcriptional activity is in turn directly regulated by ATM in response to genotoxic stress [28]. Interestingly an ATM-dependent pathway is also responsible for the phosphorylation of SMC3 in response to irradiation [3] raising the possibility that RhoB and SMC3 are members of the same DNA integrity surveillance pathway. Other findings suggest that SMC3 and RhoB expression may be linked. The expression of these genes is increased following activation of the Wnt/β-catenin/TCF4 transactivation [6,29]. On the other hand SMC3 and RhoB expression is negatively affected by Ras [30,31]. RhoB has both a positive and negative role in cell growth and it has been postulated it operates in a contextual manner. For example RhoB is essentially required for Ras-mediated transformation but at the same time it is necessary for the apoptotic response of transformed cells to DNA-damaging agents [25-27], suppresses tumorigenesis and restrains the transformed characteristics of neoplastic cells [32]. SMC3-transformed cells however do not display sign of apoptosis as revealed by Annexin V binding or propidinium iodine nuclear staining (data not shown). Furthermore, as revealed by the microarray analysis, the expression of apoptotic agents and mediators was not affected by SMC3 upregulation (Table I). Given that RhoB and SMC3 may exert opposite effect on cell transformation, we postulate that SMC3 oncogenic activity prevails by overcoming the pro-apoptotic action of RhoB.
Figure 2 Two-class analysis of the microarray data set. The six 293 cells arrays (three from the control clones, and three from the SMC3-overexpressing clones) were subjected to SAM two-class unpaired analysis where the software created a field of observed versus expected gene regulation values from the array data. A delta parameter (0.409) was chosen to limit the field and provide the optimal output of significantly to falsely significantly regulated genes (16 in our analysis). The threshold was set at 1.00 corresponding to a twofold difference in regulation from the control cells when data are entered as log2 values. Dashed lines: delta parameter which defines the significance field; dots above the upper line: probable significantly up-regulated genes; dots below the lower line: probable significantly down-regulated genes.
SMC3 acts as an oncogene in human cells and activates the expression of a set of early-response genes
In order to assesses whether changes in the gene expression observed in cell stably overexpressing SMC3 are part of an early response to SMC3 elevation and may thus mediate the tumorigenic potential of this cohesin protein, 293 cells were transiently transfected with SMC3-pcDNA3.1 or alternatively with the empty expression vector (fig. 3a). Ectopic expression of SMC3 in 293 cells enhanced cell proliferation (fig. 3b) and promoted cell aggregation, rounding and piling as observed when the cells were cultured in serum-deficient medium (fig. 3c,d). As anchorage-independent growth is considered to be in vitro test for tumorigenesis, we examined the growth of SMC3-transfected 293 cells in a semi-soft agarose medium. Scoring of the number and the size of the colonies formed revealed that SMC3 acts as an inducer of cell transformation. Not only the percent of colonies formed was increased (~95% compared to ~50% originating from pcDNA3.1-transfected cells) but the colonies formed after three weeks where significantly larger (12 ± 3 colonies of diameter of >100 μm/9 mm2 vs. none generated by pcDNA3.1-transfected cells) (fig. 3e,f). These results demonstrate that SMC3 has the ability to enhance the tumorigenic potential of immortalized human cells. SMC3 does not require the ectopic co-expression of other oncogenes to achieve its tumorigenic function.
Figure 3 Effect of SMC3 overexpression on 293 cells colony growth. a): 293 cells at 70% confluence were transfected with 1 μg/ml of pcDNA3.1 (control) or SMC3-pcDNA3.1 expression vectors and cell RNA collected after 48 hrs in 1 ml TriReagent. SMC3, and G3PDH transcript levels were assayed by semiquantitative RT-PCR and the products analyzed by agarose electrophoresis. The result of a control and two independent samples of cells transfected with SMC3-pcDNA3.1 are shown. b) Growth rate of cells overexpressing SMC3. Cells were seeded in 96-well plates (2,500 cells/well in a final volume of 150 μl) in DMEM medium supplemented with 1.5% FCS. Cell proliferation was assessed by using a CellTiter 96 colorimetric assay. Each point represent the mean ± SD of four independent determinations. c,d) Transfected cells were seeded in 35 mm plates and grown for 10 days in DMEM supplemented with 1.5% FCS. To detect foci of transformation, cells were fixed in 70% ethanol followed by staining with 0.1% methylene-blue. e,f) Colony formation in semisoft agarose. Twenty-four h after transfection, cells were trypsinized and resuspended in 0.2% agarose in DMEM containing 10% fetal bovine serum and plated on top of solidified agarose (0.4%) dissolved in the same medium in 35 mm dishes. After 3 weeks of culture cell colonies were examined under a light microscope.
Cells transiently transfected with SMC3-pcDNA3.1 displayed at 3-folds increase in SMC3 transcript level. This was matched by a 5-fold elevation of RhoB transcript level (fig 4a). The effect of SMC3 was confirmed when RhoB protein level was assessed by immunoblotting (fig. 4b). Because SMC3 has the capability to transform NIH3T3 cells [5] we examined whether the ectopic expression of SMC3 enhances RhoB expression in the murine fibroblasts. The Western immunoblotting results confirmed this hypothesis (fig. 4b). The finding corroborates the idea that a causal link exist between SMC3 and RhoB expression that is not restricted to a specific cell context. The mechanism whereby SMC3 elevation leads to RhoB upregulation can be only presently speculated. It has been reported that interference with the cohesin complex turnover, such as it may occur when SMC3 level is elevated, causes derangement of the cell cycle progression and increases the number of chromosomal segregation errors leading to DNA breakage and aneuploidy [33]. As discussed previously, the ensuing activation of the ATM pathway may lead to RhoB transactivation via a CREB-dependent mechanism [27,28]. The evidence that p53- and CRE-mediated gene transactivation are enhanced in cells constitutively overexpressing SMC3 (fig. 5), is consistent with an activation of the ATM-dependent pathway [34] and corroborates this scenario. The upregulation of MLH3 and DNMT2 (see Table I) – two genes involved in DNA base mismatch repair, further supports this hypothesis. DNA damage would normally result in cell cycle arrest. Cells constitutively overexpressing SMC3 however display enhanced growth rate suggesting that they can escape the surveillance of the cell cycle gatekeepers. Recently SMC3 has been identified as a component of a complex including SMC1 and BRCA1, that operates as an effector in the ATM-dependent S-phase checkpoint [3,35]. It is possible that when present in excess SMC3 acts in dominant-negative fashion with respect to the activity of the SMC1/SMC3/BRCA1 complex affecting the efficacy of the ATM-dependent S-phase checkpoint.
Figure 4 Effect of SMC3 on the expression of a set of early-response genes. a) Cells were transfected with 1 μg/ml of pcDNA3.1 (control) or SMC3-pcDNA3.1 and RNA collected after 48 hrs in 1 ml TriReagent. Gene transcripts were amplified by RT-PCR and the products analyzed by agarose electrophoresis. b) Analysis of RhoB level in NIH-3T3 and 293 cells overexpressing SMC3. Cells were transfected as in a) and 48 h later solubilized in lysis buffer. Twenty and 50 μg respectively of NIH-3T3 and 293 cell lysate proteins were analyzed by 12.5% SDS-PAGE. After transfer to nitrocellulose, RhoB was immunodetected using a rabbit polyclonal antibody followed by ECL detection of the immunocomplex. Nitrocellulose-bound proteins were stained with Ponceau-red to evidence the amount of proteins loaded.
Figure 5 Transcriptional activity in control and SMC3 overexpressing cells. Cells were seeded in 12 well plates and used at 70% confluence. Treatment groups, each as triplicate samples, were transfected with 0.1 μg/ml of the indicated reporter vector together with 0.01 μg/ml of PH-RL transfection control vector using 5 μl/ml Lipofectamine. The reporter vectors harbor multiple repeats of the consensus sequence for different transcriptional binding sites which drive the expression of a firefly luciferase. Transactivation activity was calculated based on the firefly luciferase level correcting for the transfection efficiency using the renilla luciferase level in a dual luciferase assay. Data shown are the mean ± SD from three independent determinations. * p < 0.01 SMC3-overexpressing vs. control.
Further analysis of SMC3-transfected 293 cells and the parent cells provided evidence that in addition to RhoB also CRE-BPa, RGS14 and ARHGEF4 are part of the set of genes activated following transient elevation of SMC3. RGS14 is a target of the p53 tumor suppressor and its overexpression inhibits both Gi- and Gq-coupled growth factor receptor mediated activation of the mitogen-activated protein kinase signaling pathway in mammalian cells [36]. Because p53 transactivation pathway is activated following SMC3 overexpression (see below) we postulate that RGS14 may be involved in counteracting the SMC3 mitogenic activity. ARHGEF4 (also known as Asef) is a Rac-specific guanine nucleotide exchange factor that is activated following binding to APC [37,38]. APC-ARHGEF4 complex plays an important role in the regulation of the actin cytoskeleton, cell morphology and migration and affects E-cadherin-mediated cell-cell adhesion. Agents such as ARHGEF4 may thus act as mediators of the effect of SMC3 overexpression on cell growth and morphogenesis.
SMC3 overexpression specifically activates the SRE and CRE transactivation pathways
To further examine the cellular response to SMC3 overexpression, the activation level of a number of transactivation pathways was investigated. We reason that selective changes in transcriptional activity could further sort out the key players mediating the SMC3 biological activity. For this purpose 293 cells were transfected with SMC3 expression vector or pcDNA3.1 vector alone (control group) together with a series of luciferase reporters whose expression is driven by target-specific cis-acting elements present in multiple copies in the vector promoter. The serum responsive element (SRE) is a known target of RhoB as well as of other ras-rho/GTPase [25]. Consistent with the idea that elevation of RhoB is functionally significant a 2.8-fold elevation of SRE-transactivation activity was detected in SMC3-overexpressing cells (fig. 5). On the contrary AP1-dependent gene transactivation which is mediated by c-jun, c-fos and ATF2 homo or heterodimers, was significantly suppressed. Given that RhoB, but not other ras-rho/GTPase, suppresses AP1-mediated gene transactivation [30] this result support the idea that following SMC3 elevation, RhoB plays a central role in mediating gene transactivation. Consistent with a major role of the CREB in the SMC3 mediated cell events, the response of the CRE-dependent reporter was greatly increased (2.6-fold) in cells stably overexpressing SMC3. This cAMP responsive element is a bona-fide target for the CRE-BPa transcriptional factor whose expression is significantly increased following SMC3 elevation (Table I, fig. 4a). CRE-BPa is a member of the ATF2 cAMP-binding proteins family that is activated by a variety of kinases including protein kinase A, JNK/SAPK, p38-MAPK, AKT, and calcium-calmodulin-dependent kinases and is involved in tumorigenesis of endocrine tissues and different forms of leukemia [21]. To the activation of the CREB transactivation pathway may also contribute the downregulation of CREM, a gene encoding several spliced products some of which are CRE-transactivation repressors [22]. The glucocorticoid, TGFβ-activin and of NF-kB transactivation pathways are potential target of RhoB or CRE-BPa, and their activity was also tested [26,39,40]. NFkB-mediated transactivation which has been reported to be downregulated by RhoB in NIH3T3 cells [26] was not significantly affected in our experiments suggesting that this regulatory mechanism is either cell context-specific or that SMC3 elicits other changes that counterbalance the RhoB-dependent NF-kB loss of activity. Likewise gene transactivation from the GRE and TARE cis-acting elements was not affected. Taken together the results are consistent with the idea that following SMC3 elevation, SRE- and CRE- mediated gene transactivation is specifically engaged.
Conclusion
The results presented provide a molecular signature of the changes that occurs in epithelial cells following SMC3 overexpression. In particular we document a selective effect of the ectopic expression of SMC3 on a set of early-response genes such as RhoB and CRE-BPa and on the related transcriptional pathways SRE and CRE that play key roles in tumorigenesis. SMC3 acts as an oncogene in human cells and we show that RhoB and CRE-BPa are part of a set of genes activated following transient SMC3 transfection. These findings provide important initial information on the chain of events occurring following SMC3 overexpression that will allow in future studies to focus on the underlying mechanism of the association between SMC3 deregulation and specific oncogenic pathways.
Methods
Establishment of cell lines stably overexpressing SMC3
293 cells were grown to 70% confluence in 10 cm plates and transfected with 3 μg of pcDNA3.1 expression vector harboring the entire human SMC3 coding sequence (SMC3-pcDNA3.1) [5] and using Lipofectamine as transfecting agent. To generate a control cell line a second batch of cells was transfected instead with the empty pcDNA3.1 vector. After 48 h stably transfected cells were selected in medium containing 500 μg/ml of G418. Clones of the surviving cells were expanded and SMC3 expression examined by semi-quantitative RT-PCR. For this purpose, 1 μg of cell RNA was reverse transcribed with Sensiscript (Qiagen) reverse transcriptase priming with oligo-dT. An aliquot of the RT reaction product was amplified using ExTaq (Takara) DNA polymerase and SMC3-specific primers of sequence: 5'-GAGTAGAAGAACTGGACAGA-3' and 5'-GATTGTACCTCAGTTTGCTG-3'. To ensure that the amplification reaction had not reached saturation, DNA production was monitored after 25 and 30 cycles by analysis on 1% agarose and by staining with ethidium bromide. Gels were photographed, the picture scanned and the band intensity quantified by densitometry with an image scanner. SMC3 protein expression was assessed by Western immunoblotting. Briefly, cell lysates in 150 mM NaCl 1% Nonidet P40 0.5% Na-deoxycholate 50 mM Tris-HCl pH 7.4 were electrophoresed on 8% SDS-PAGE slab gel and the separated proteins transferred onto a nitrocellulose filter. Immunoblotting was performed with goat anti-human SMC3 (1:1,000) antibody (Santa Cruz Biotech) at 25°C for 1 h, followed by incubation in anti-goat IgG horseradish peroxidase conjugated (1:10,000) secondary antibody. Immunocomplexes were identified using an enhanced chemiluminescence (ECL) kit (Pierce) followed by autoradiograph. Three SMC3 overexpressing clones and three control clones were selected for the gene expression profiling.
Microarrays and data analysis
For the target preparation, 5 μg of human untransfected and transfected 293 cell line total RNA were reverse transcribed with SuperscriptT-II/RNaseH- priming with T7-(dT)24 oligonucleotides and the second-strand cDNA synthesized using E. coli DNA polymerase I [18]. Biotinylated cRNA was generated using T7 RNA polymerase and Biotin 11-UTP. Ten μg of purified unfragmented target cRNA was used for hybridization of each KCC/TJU human 18.5 K Expression Bioarray (Compugen Human Oligo Set 1.0) chip containing 18,861 oligos (65-mer) corresponding to 17,260 unique clusters and 18 bacterial control probes. The microarrays were hybridized, washed, and processed using a direct detection method of the biotin-containing transcripts by a Streptavidin-Alexa647 conjugate. Processed slides were scanned using a Perkin Elmer ScanArray XL5K scanner and the spot intensity quantitated using the ScanArray Scanning and QuantArray programs (PerkinElmer) [18]. The chips analyzed presented consistent staining over the entire microarray and had appropriate data distribution. Spots with raw intensity value comprised within one SD from the average background value were excluded from the analysis. The remaining values – about 90% of the whole arrayed genes, were normalized by dividing each spot's intensity (after background subtraction) by the median signal intensity of all test probes. Genes that displayed inter-array variability exceeding one SD unit were furthermore excluded from the final statistical analysis. The data and protocols have been submitted to the EBI ArrayExpress database (samples 171479SUB800 through 171484SUB800).
Statistical analysis of the microarray data
We input the log2 of the gene expression measurements from three sets of microarray experiments each including a control and an SMC3 overexpressing cell sample. Through a series of permutation the program computes a statistic score di for each gene i measuring the strength of the relationship between gene expression and the response variable and creates a profile of observed versus expected values. The values which lie outside a user-defined region that can be adjusted to achieve an optimum of positive vs. false positive values, are considered significantly related to the response and thus regarded as significantly regulated genes (see fig. 2). Since each experiment consisted of the data from two independent chips hybridized with the control and the SMC3-overexpressing cells cDNA, an unpaired two-class analysis was carried out to discover significant changes in gene regulation compared to the control cell line.
Gene transactivation activity assay
Reporter vectors harboring multiple copies of the consensus sequences for the AP1, cAMP (CRE), serum (SRE), p53, TGFβ (TARE), NF-kB and glucocorticoid (GRE) response elements, were obtained from Stratagene or Clontech. Cells cultures at 70% confluence in 12 wells plates were used in all the experiments. The transfection mix contained 10 ng/ml of phRL-SV40 plasmid to monitor the transfection efficiency, 100 ng/ml of the designed plasmid, and 1 μg/ml of SMC3-pcDNA3.1 expression vector. Plasmids were mixed in medium 199 followed by the addition of 10 mg/mg DNA of Tfx-50 (Promega) transfection agent according to the manufacturer directions and finally added to the cultures. After 1 h incubation at 37°C in humidified incubator the cells were supplemented with 2 ml of growth medium and the luciferase activity assayed 24 h later using a Promega dual-luciferase kit. All experiments were carried out with triplicate samples. The statistical difference between groups of data was analyzed by Student's t-test.
SMC3 transient transfection and gene transcript level analysis
Cells (either 293 or NIH3T3) at 70% confluence were transiently transfected with 1 μg/ml of SMC3-pcDNA3.1 expression vector using Lipofectamine as transfection agent. Control cells were transfected instead with 1 μg/ml of pcDNA3.1 empty vector. After 48 h, the cells were washed in PBS and the total RNA extracted with TriReagent. Gene transcripts were amplified by RT-PCR and the products quantified by gel electrophoresis. The primers used had the following sequence: RhoB: 5'-CCTGCTGATCGTGTTCAGTAA-3' and 5'-TCATAGCACCTTGCAGCAGTT-3'; CRE-BPa: 5'-ATGATTTATGAGGAATCCAAGAT G-3' and 5'-TTAAAGAATCGGATTCAGGTCTGT-3'; RGS14: 5'-CTGGTGGGCAATGAACAGAAGGCC-3' and 5'-GGGCTGAGTCGGTGGTGGAGTTCA-3'; ARHGEF4: 5'-AGCCTCAAGCCAAAAGCCAGCAGC-3' and 5'-CTCACTTGCTGGCAGAGGAAGGCCA-3'; G3PDH: 5'-TGAAGGTCGGAGTCAACGGATTTGGT-3' and 5'-CATGTGGGCCATGAGGTCCACCAC-3'. RhoB protein level in cells was examined by Western immunoblotting as described previously using a rabbit polyclonal antibody (1:1,000) from Bethyl.
Cell proliferation assay
Cell proliferation was examined using a CellTiter 96 (Promega) assay kit according to the manufacturer's instructions. Cells growing in log phase were trypsinized and seeded in 96-well plates (2,500 cells/well in a final volume of 150 μl) in replicates of 4 and incubated at 37°C in 5% CO2 and 95% air in DMEM medium containing 1.5% FCS. At 24 h, 72 h or 96 h, 20 μl of the kit dye solution was added to each well and the plates incubated at 37°C for an additional 1 h. The absorbance of the formazan product generated was measured at 490 nm using a 96-well Dynatech MR600 plate reader.
Cell overgrown assay
In order to examine the ability of cells to form foci of transformation, cells were seeded at 30% confluence in 35 mm plates and cultured in DMEM supplemented with 1.5% medium. After 7 days the medium was removed, the cell washed with PBS, and fixed with 70% ethanol on ice. Foci of cell aggregation were evidenced by staining with 0.1% methylene-blue dissolved in water followed by 3 washing in water to decrease background staining.
Anchorage-independent cell growth in soft agar
Anchorage-independent colony formation of cells was assayed as described [5]. Briefly, cells growing in log-phase were trypsinized and resuspended at 37°C in 0.2% agarose in DMEM containing 10% fetal bovine serum and plated on top of solidified agarose (0.4%) dissolved in the same medium in 35 mm dishes. After 3 weeks of culture at 37°C in CO2 humidified incubator, the number of cell aggregates over that of single cells and the number of colonies of diameter >100 μm found in randomly selected areas of 9 mm2, was recorded.
Authors' contributions
The experiments with cells were conducted in G. G. laboratory with the technical help of Mr. Amit Agrawal and Mr. Chirag Patel. G.G. is also responsible for the drafting of the manuscript. The microarray analysis was conducted by C-G.L at the Microarray Facility of the Kimmel Cancer Center.
Acknowledgements
This work was supported by grant RO1-CA82290 to GG.
==== Refs
Koshland DE Guacci V Sister chromatid cohesion: the beginning of a long and beautiful relationship Curr Opin Cell Biol 2000 12 297 301 10801457 10.1016/S0955-0674(00)00092-2
Uhlmann F Chromosome cohesion and segregation in mitosis and meiosis Curr Opin Cell Biol 2001 13 754 761 11698193 10.1016/S0955-0674(00)00279-9
Yazdi PT Wang Y Zhao S Patel N Lee EY Qin J SMC1 is a downstream effector in the ATM/NBS1 branch of the human S-phase checkpoint Genes Dev 2002 16 571 582 11877377 10.1101/gad.970702
Shimizu K Shirataki H Honda T Minami S Takai Y Complex formation of SMAP/KAP3, a KIF3A/B ATPase motor-associated protein, with a human chromosome-associated polypeptide J Biol Chem 1998 273 6591 6594 9506951 10.1074/jbc.273.12.6591
Ghiselli G Iozzo RV Overexpression of bamacan/SMC3 causes transformation J Biol Chem 2000 275 20235 20238 10801778 10.1074/jbc.C000213200
Ghiselli G Coffee N Munnery CE Koratkar R Siracusa LD The cohesin SMC3 is a target the for beta-catenin/TCF4 transactivation pathway J Biol Chem 2003 278 20259 20267 12651860 10.1074/jbc.M209511200
Graham FL Smiley J Russell WC Nairn R Characteristics of a human cell line transformed by DNA from human adenovirus type 5 J Gen Virol 1977 36 59 74 886304
Shaw G Morse S Ararat M Graham FL Preferential transformation of human neuronal cells by human adenoviruses and the origin of HEK 293 cells FASEB J 2002 16 869 871 11967234
Guan LS Li GC Chen CC Liu LQ Wang ZY Rb-associated protein 46 (RbAp46) suppresses the tumorigenicity of adenovirus-transformed human embryonic kidney 293 cells Int J Cancer 2001 93 333 338 11433396 10.1002/ijc.1338
Kessler O Shraga-Heled N Lange T Gutmann-Raviv N Sabo E Baruch L Machluf M Neufeld G Semaphorin-3F is an inhibitor of tumor angiogenesis Cancer Res 2004 64 1008 1015 14871832 10.1158/0008-5472.CAN-03-3090
Cheng JD Dunbrack RLJ Valianou M Rogatko A Alpaugh RK Weiner LM Promotion of tumor growth by murine fibroblast activation protein, a serine protease, in an animal model Cancer Res 2002 62 4767 4772 12183436
Kamei D Murakami M Nakatani Y Ishikawa Y Ishii T Kudo I Potential role of microsomal prostaglandin E synthase-1 in tumorigenesis J Biol Chem 2003 278 19396 19405 12626523 10.1074/jbc.M213290200
Hamid T Malik MT Kakar SS Ectopic expression of PTTG1/securin promotes tumorigenesis in human embryonic kidney cells Mol Cancer 2005 4 3 15649325 10.1186/1476-4598-4-3
Kakiuchi S Daigo Y Tsunoda T Yano S Sone S Nakamura Y Genome-wide analysis of organ-preferential metastasis of human small cell lung cancer in mice Mol Cancer Res 2003 1 485 499 12754296
Lambert JF Liu M Colvin GA Dooner M McAuliffe CI Becker PS Forget BG Weissman SM Quesenberry PJ Marrow stem cells shift gene expression and engraftment phenotype with cell cycle transit J Exp Med 2003 197 1563 1572 12782718 10.1084/jem.20030031
Glienke J Schmitt AO Pilarsky C Hinzmann B Weiss B Rosenthal A Thierauch KH Differential gene expression by endothelial cells in distinct angiogenic states Eur J Biochem 2000 267 2820 2830 10785405 10.1046/j.1432-1327.2000.01325.x
Jones JO Arvin AM Microarray analysis of host cell gene transcription in response to varicella-zoster virus infection of human T cells and fibroblasts in vitro and SCIDhu skin xenografts in vivo J Virol 2003 77 1268 1280 12502844 10.1128/JVI.77.2.1268-1280.2003
Liu CG Calin GA Meloon B Gamliel N Sevignani C Ferracin M Dumitru CD Shimizu M Zupo S Dono M Alder H Bullrich F Negrini M Croce CM An oligonucleotide microchip for genome-wide microRNA profiling in human and mouse tissues Proc Natl Acad Sci U S A 2004 101 9740 9744 15210942 10.1073/pnas.0403293101
Tusher VG Tibshirani R Chu G Significance analysis of microarrays applied to the ionizing radiation response Proc Natl Acad Sci U S A 2001 98 5116 5121 11309499 10.1073/pnas.091062498
Filmus J Glypicans in growth control and cancer Glycobiology 2001 11 19R 23R 11320054 10.1093/glycob/11.3.19R
Mayr B Montminy M Transcriptional regulation by the phosphorylation-dependent factor CREB Nat Rev Mol Cell Biol 2001 2 599 609 11483993 10.1038/35085068
Foulkes NS Borrelli E Sassone-Corsi P CREM gene: use of alternative DNA-binding domains generates multiple antagonists of cAMP-induced transcription Cell 1991 64 739 749 1847666 10.1016/0092-8674(91)90503-Q
Coleman ML Marshall CJ Olson MF RAS and RHO GTPases in G1-phase cell-cycle regulation Nat Rev Mol Cell Biol 2004 5 355 366 15122349 10.1038/nrm1365
Malliri A Collard JG Role of Rho-family proteins in cell adhesion and cancer Curr Opin Cell Biol 2003 15 583 589 14519393 10.1016/S0955-0674(03)00098-X
Prendergast GC Actin' up: RhoB in cancer and apoptosis Nat Rev Cancer 2001 1 162 168 11905808 10.1038/35101096
Fritz G Kaina B Aktories K The ras-related small GTP-binding protein RhoB is immediate-early inducible by DNA damaging treatments J Biol Chem 1995 270 25172 25177 7559652 10.1074/jbc.270.50.29998
Fritz G Kaina B Transcriptional activation of the small GTPase gene rhoB by genotoxic stress is regulated via a CCAAT element Nucleic Acids Res 2001 29 792 798 11160903 10.1093/nar/29.3.792
Shi Y Venkataraman SL Dodson GE Mabb AM LeBlanc S Tibbetts RS Direct regulation of CREB transcriptional activity by ATM in response to genotoxic stress Proc Natl Acad Sci U S A 2004 101 5898 5903 15073328 10.1073/pnas.0307718101
Taneyhill L Pennica D Identification of Wnt responsive genes using a murine epithelial cell line model system BMC Dev Biol 2004 4 6 12 15140269 10.1186/1471-213X-4-6
Jiang K Delarue FL Sebti SM EGFR, ErbB2 and Ras but not Src suppress RhoB expression while ectopic expression of RhoB antagonizes oncogene-mediated transformation Oncogene 2004 23 1136 1145 14647415 10.1038/sj.onc.1207236
Yoon JW Kita Y Frank DJ Majewski RR Konicek BA Nobrega MA Jacob H Walterhouse D Iannaccone P Gene expression profiling leads to identification of GLI1-binding elements in target genes and a role for multiple downstream pathways in GLI1-induced cell transformation J Biol Chem 2002 277 5548 5555 11719506 10.1074/jbc.M105708200
Liu AX Rane N Liu JP Prendergast GC RhoB is dispensable for mouse development, but it modifies susceptibility to tumor formation as well as cell adhesion and growth factor signaling in transformed cells Mol Cell Biol 2001 21 6906 6912 11564874 10.1128/MCB.21.20.6906-6912.2001
Jallepalli PV Waizenegger IC Bunz F Langer S Speicher MR Peters JM Kinzler KW Vogelstein B Lengauer C Securin is required for chromosomal stability in human cells Cell 2001 105 445 457 11371342 10.1016/S0092-8674(01)00340-3
Banin S Moyal L Shieh S Taya Y Anderson CW LeBlanc S Tibbetts RS Enhanced phosphorylation of p53 by ATM in response to DNA damage Science 1998 281 1674 1677 9733514 10.1126/science.281.5383.1674
Kitagawa R Bakkenist CJ McKinnon PJ Kastan MB Phosphorylation of SMC1 is a critical downstream event in the ATM-NBS1-BRCA1 pathway Genes Dev 2004 18 1423 1438 15175241 10.1101/gad.1200304
Buckbinder L Velasco-Miguel S Chen Y Xu N Talbott R Gelbert L Gao J Seizinger BR Gutkind JS Kley N The p53 tumor suppressor targets a novel regulator of G protein signaling Proc Natl Acad Sci U S A 1997 94 7868 7862 9223279 10.1073/pnas.94.15.7868
Kawasaki Y Senda T Ishidate T Koyoma R Morishita T Iwayama Y Higuchi O Akiyama T Asef, a link between the tumor suppressor APC and G-protein signaling Science 2000 289 1194 1197 10947987 10.1126/science.289.5482.1194
Kawasaki Y Sato R Akiyama T Mutated APC and Asef are involved in the migration of colorectal tomour cells Nat Cell Biol 2003 5 211 215 12598901 10.1038/ncb937
Moustakas A Stournaras C Regulation of actin organisation by TGF-beta in H-ras-transformed fibroblasts J Cell Sci 1999 112 ( Pt 8) 1169 1179 10085252
Su LF Knoblauch R Garabedian MJ Rho GTPases as modulators of the estrogen receptor transcriptional response J Biol Chem 2001 276 3231 3237 11060289 10.1074/jbc.M005547200
|
16156898
|
PMC1242249
|
CC BY
|
2021-01-04 16:36:35
|
no
|
Mol Cancer. 2005 Sep 12; 4:34
|
utf-8
|
Mol Cancer
| 2,005 |
10.1186/1476-4598-4-34
|
oa_comm
|
==== Front
Mol CancerMolecular Cancer1476-4598BioMed Central London 1476-4598-4-351615689910.1186/1476-4598-4-35ResearchAttenuated Expression of DFFB is a Hallmark of Oligodendrogliomas with 1p-Allelic Loss McDonald J Matthew [email protected] Valerie [email protected] Ellen [email protected] Raymond [email protected] Janet [email protected] Gregory N [email protected] Kenneth [email protected] Wei [email protected] Departments of Pathology and Neurosurgery, the University of Texas M. D. Anderson Cancer Center, Houston, Texas, USA2005 12 9 2005 4 35 35 18 7 2005 12 9 2005 Copyright © 2005 McDonald et al; licensee BioMed Central Ltd.2005McDonald et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Allelic loss of chromosome 1p is frequently observed in oligodendroglioma. We screened 177 oligodendroglial tumors for 1p deletions and found 6 tumors with localized 1p36 deletions. Several apoptosis regulation genes have been mapped to this region, including Tumor Protein 73 (p73), DNA Fragmentation Factor subunits alpha (DFFA) and beta (DFFB), and Tumor Necrosis Factor Receptor Superfamily Members 9 and 25 (TNFRSF9, TNFRSF25). We compared expression levels of these 5 genes in pairs of 1p-loss and 1p-intact tumors using quantitative reverse-transcriptase PCR (QRTPCR) to test if 1p deletions had an effect on expression. Only the DFFB gene demonstrated decreased expression in all tumor pairs tested. Mutational analysis did not reveal DFFB mutations in 12 tested samples. However, it is possible that DFFB haploinsufficiency from 1p allelic loss is a contributing factor in oligodendroglioma development.
==== Body
Introduction
Oligodendroglial tumors with allelic losses on 1p usually display loss of relatively long regions, a phenomenon that has made the identification of putative 1p tumor suppressor genes difficult [1-6]. However, the vast majority of reported oligodendroglioma cases with 1p-loss have involved the 1p36 region, with several breakpoints within the region observed [5-9]. It is important to note that several apoptotic genes have been mapped to 1p36. Diminished apoptosis has been recognized as one of the hallmarks of most types of cancer, representing one of the major ways known for a tumor-cell population to expand [10]. Therefore, we tested TP73, TNFRSF9, TNFRSF25, DFFA, and DFFB, all of which are 1p genes involved in apoptosis, for differential expression in 1p-status subsets of oligodendroglioma. QRTPCR analysis of match-paired samples demonstrated that levels of DFFB were decreased in all 1p-allelic loss cases. In contrast, the other tested genes showed heterogeneous patterns of expression. This result suggests DFFB to be a key molecule affected by 1p-deletion in oligodendroglioma.
Materials and methods
Samples
The records of 177 patients who underwent treatment for oligodendroglial tumors at the University of Texas M.D. Anderson Cancer Center (UTMDACC) between 1981 and 2002 were collected and reviewed. These patients were initially diagnosed as having low-grade oligodendroglioma or mixed oligoastrocytoma, anaplastic oligodendroglioma or mixed oligoastrocytoma, or glioblastoma multiforme with significant oligodendroglial component by neuropathologists from UTMDACC and later confirmed by two of the authors (KA and GF). Mixed tumors were included in this study since clear pathologic discrimination between glioma subtypes is sometimes difficult, and as a group, oligoastrocytomas often have 1p deletions [4]. In fact, both the oligodendroglial and astrocytic components of mixed tumors have been observed to have this genetic signature [4].
Tissue for DNA isolation was obtained from paraffin-embedded samples. Each tissue block was histologically assessed for tumor by a neuropathologist (KA). Sections were directly cut from the block for DNA isolation if at least 90% of the tissue was determined to be tumor. If the proportion of tumor was <90%, 10 to 20 unstained slides were prepared from the block and tumor tissue was dissected from normal tissue. DNA was isolated by digesting deparaffinized tumor sections for 3 to 5 days with proteinase K at 55°C (0.5 mg/ml in 100 mmol/L NaCl, 10 mmol/L Tris-HCl, pH 8.0, 25 mmol/L ethylenediaminetetraacetic acid, 0.5% sodium dodecyl sulfate), followed by a phenol:chloroform:isoamyl alcohol extraction and isopropanol precipitation.
Tissue for RNA isolation was obtained from fresh/frozen samples. Each frozen section was histologically assessed for tumor by a neuropathologist (GF or KA) and used only if at least 90% of the tissue was determined to be tumor. For RNA isolation, up to 50 mg of tissue was frozen in liquid nitrogen, crushed into powder using a mortar and pestle, and dissolved in 1 ml of Trizol® Reagent. 200 μl chloroform was added to the sample, vortexed at high speed for 15 seconds, and centrifuged at 12,000 × g for 15 minutes at 4°C. After transfer of the aqueous phase to fresh 1.5 ml Eppendorf tube, an equal volume of 70% ethanol was added and mixed by tube inversion. The sample was then loaded onto a QIAGEN RNeasy® mini column and centrifuged at 16,000 × g for 20 seconds (QIAGEN Sciences, Germantown, MD). The column was washed twice with 500 μl of QIAGEN's RPE buffer. RNA was eluated off the column in 50 μl of nuclease-free water. RNA was quantified using a spectrophotometer and qualitated with an Agilent BioAnalyzer 2100 (Agilent Technologies, Palo Alto, CA).
Detection of 1p allelic loss
Quantitative Microsatellite Analysis (QuMA) was used as previously described to determine 1p allelic loss in 177 tumors in this study [11-13].
Quantitative Reverse Transcription-Polymerase Chain Reaction (QRTPCR) Analysis
Initial experiments were performed to determine the valid range of RNA concentrations and to demonstrate the similarity of PCR efficiencies for each gene of interest compared to the endogenous control gene cyclophilin. To determine fold-changes in each gene, QRTPCR was performed on the ABI Prism 7700 using the commercially available gene expression assay for p73, DFFA, and DFFB (Hs00232088_m1, Hs00189336_m1, Hs00237077_m1, respectively) and the cylophilin Vic-labeled Pre-Developed Assay Reagent (Applied Biosystems, Foster City, CA) without multiplexing. In triplicate, we amplified 50 ng cDNA for each sample for each assay in a reaction containing 1× TaqMan® Universal PCR Master Mix without AmpErase UNG and 1× gene expression assay with the following cycling conditions: 10 minutes at 95°C, then 45 cycles of 95°C for 15 seconds and 60°C for 1 minute. Calculations were performed using the δδCt method to determine fold-difference in 1p-loss cases relative to the matched 1p-intact cases. Fold changes for TNFSF5, TNFRSF9, TNFRSF11a, and TNFRSF25 were determined in a similar fashion, using commercially available gene expression assays (Hs00374176_m1, Hs00155512_m1, Hs00187189_m1, and Hs00237054_m1, respectively) and the 18S rRNA TaqMan® Endogenous Control (Hs99999901_s1).
Mutation screening
PCR amplifications of exons 1–6 were carried out using 100-μL reaction volumes with 1.5 mmol/L MgCl2; 200 μmol/L each of deoxy (d)-ATP, dGTP, dTTP, and dCTP; 2 pmol of each primer; 100 ng template DNA; and 1 U AmpliTaq Gold polymerase (Applied Biosystems, Foster City, CA). Amplifications of exon 7 were the same with the exception that the reaction mix had a concentration of 7% dimethylsulphoxide (DMSO). PCR cycling conditions were 10 min at 94°C, followed by 40 cycles of 94°C for 1 min, 65°C for 1 min, and 72°C for 1 min, followed by 15 min at 72°C. Sequencing reactions were setup using the BigDye Terminator Cycle Sequencing Reaction Kit with AmpliTaq DNA polymerase FS (Applied Biosystems, Foster City, CA) according to the manufacturer's specifications, and were subjected to gel electrophoresis on an ABI PRISM 3700 (Applied Biosystems, Foster City, CA). Sequencing data were aligned with the Sequencer program using DFFB sequence as reported by the Human Genome Database [14]. Forward and reverse PCR primer sequences are listed in Table 1.
Table 1 Primer sets used for amplifying and sequencing the coding regions of DFFB.
Name F primer seq R primer seq
DFFBamp1 gcttgcagagctcaccaggtgc cggctgaggcgaacgaaaactacc
DFFBseq1 acggatctgagcagctgg ctcctattctccccacacgc
DFFBamp2 aagcacagctcattccggtcg tgatgggcacctggagctaagc
DFFBseq2 gccctcgtcttgagacc aggacctcggagagtgc
DFFBamp3 gggggaagatgtggtcagaggctc ccacctgagtccttgctgggtacc
DFFBseq3 cttgtgaccggggcag atccaacttcttctggcacc
DFFBamp4 gctgtagtaagctgtgttcgtgccactg gcgctagcttccctcaccagagc
DFFBseq4 ggaggacagagcaagacc ccagatccacgcaagc
DFFBamp5 gggtctcagagggccatggag cctgtgtgcactgcagcttgagag
DFFBseq5 atggatcgagagccagtg ggcaagggctgaaggtc
DFFBamp6 cgggaggcggaggttgtagtaagc ctgggctgtaacacgggtgcag
DFFBseq6 gccactgcactccagc ccatggcagggacagg
DFFBamp7 gggaatttgtgaagagctgtgactgc ccccaacaattcagaaatgtaatgaaatcag
DFFBseq7 gctatgacctgttgcctgtg ggcacctgttaaaatgatgc
Results and Discussions
We evaluated 177 oligodendroglial tumors using QuMA for 1p-allelic loss in an attempt to determine a consensus region of deletion [11-13]. Loss was observed in 92 tumors, which in most cases involved the entire chromosomal arm. However, six tumors demonstrated localized loss involving the 1p36 region, defining a consensus region of deletion (Figure 1). These results were similar to those observed for other oligodendroglial 1p-deletion mapping studies, in which consensus regions of deletion involved 1p36 [5,6,8,9].
Figure 1 Common region of allelic loss on the short arm of chromosome 1 in oligodendrogliomas. Markers used for screening 1p-allelic loss and their placement on the genetic and cytogenetic maps of 1p. Black squares indicate where tumors retained allelic balance, whereas gray squares indicate allelic loss.
The second part of our strategy included the identification of abrogated cellular pathways in oligodendrogliomas with 1p/19q allelic loss. The transcriptsomes of eight pairs of gender- and age-matched tumors were measured using a Pathway microarray consisting of 1,500 functionally characterized genes constructed in our Cancer Genomics Core Laboratory. We used paired sample tests (the Sign Rank test and the paired t-test) to identify differentially expressed genes. While the sign rank test uses the null hypothesis that the medians for the two classes are same, the paired t-test uses the null hypothesis that the means of the two classes are same. We recognize a gene as significant when the sample data for the gene gives a p-value less than 0.01 (99% confidence). This analysis revealed a number of genes that demonstrated robust differential expression (McDonald and Zhang, unpublished results).
We evaluated three of these genes with QRTPCR either because of location in our consensus region of deletion (p73), or due to their relationship to genes in our region of interest (Tumor Necrosis Factor Super Family Ligand 5 [TNFSF5] and Tumor Necrosis Factor Receptor Super Family 11a [TNFRSF11a]). Both TNFSF5 and TNFRSF11a are involved in apoptotic pathways that include several genes located in our region of interest: TNFRSF9, TNFRSF25, DFFA, and DFFB. Therefore, these genes were also tested via QRTPCR to determine if they had differential gene expression. Based on fresh/frozen tissue availability of the original 170 cases, total RNA samples from thirteen age- and gender-matched pairs of 1p/19q loss and intact tumors were evaluated for differential gene expression of p73 and the six TNF pathway genes. Of the seven genes, only DFFB was differentially expressed in all 13 pairs of tumor samples (Figure 2). Figure 2 also displays the differential expression levels for DFFA and TP73 for the tested tumor pairs. In contrast to DFFB, there were 3 pairs (25%) in which the 1p/19q loss tumors had higher DFFA expression. Likewise, 5 of the 13 pairs (38%) had higher TP73 expression in the 1p/19q loss tumors. Similarly, 33%, 50%, 50%, and 83% of tested pairs had higher expression of TNFSF5, TNFRSF9, TNFRSF11a, and TNFRSF25 in the 1p/19q loss tumors, respectively (data not shown). Since DFFB was the only tested gene that was differentially expressed in the same direction by all 13 pairs of tumors, we viewed DFFB as the best tumor suppressor gene candidate in our study.
Figure 2 QRTPCR results in oligodendroglioma subsets. Black bar, 1p-intact samples; light gray bars, 1p-loss samples. A) Expression of DFFB was lower in all 1p-loss samples as compared to their matched 1p-intact samples. Ten of the thirteen 1p-loss tumors had lower expression of DFFB compared to normal brain, with three tumors demonstrating 1–2× the amount of normal brain DFFB expression. In contrast, all 1p-intact tumors had DFFB expression greater than or equal to that seen in normal brain. B) Expression of DFFA was lower in most 1p-loss samples as compared to their matched 1p-intact samples. Only three 1p-loss tumors had higher DFFA expression. Differential expression was detected to a degree; most 1p-loss tumors (10 of 12) demonstrated 1–2× the amount of normal brain DFFA expression, whereas most 1p-intact tumors (9of 12) demonstrated ≥ 2× the amount of normal brain DFFA expression. C). Differential p73 expression was not detected. In five of the pairs, 1p-loss tumors had higher expression, whereas in seven other pairs, 1p-intact tumors had higher expression.
We addressed the candidacy of DFFB as an oligodendroglioma tumor suppressor gene by mutation analysis. Twelve tumors with 1p-allelic loss were screened for mutations by sequencing the 1.2 kb coding region of DFFB. No coding region mutations were detected in any of the samples, which may indicate that haploinsufficiency of DFFB is enough of a genetic insult to contribute to tumorigenesis. In order to thoroughly test this hypothesis, it will be necessary to further investigate DFFB, perhaps by determining if the DFFB promoter has been hypermethylaed and/or if intronic sequence has been mutated in tumor samples.
DFFB-null mouse lines have been established via gene targeting [15]. Resultant mice developed normally but their lymphocytes were more susceptible to DNA damage. These animal model experiments suggest that DFFB is a weak tumor suppressor, which may only manifest its function in the presence of stress and DNA damage. Brain tissue samples from three six-month-old specimens revealed neuropil spongiosis, but no tumor development was observed (data not shown). We are not clear at present the implication of the neuropil spongiosis phenotype. Consistent with our data, a sequencing effort in a neuroblastoma study did not reveal a tumor-specific mutation in DFFB [16]. The gene expression level of DFFB was not analyzed in that study.
Thus, this study revealed that attenuated expression of the DFFB gene is a signature of oligodendrogliomas with 1p-allelic loss. Since DFFB contributes to both chromosomal condensation and DNA degradation during apoptosis, decreased expression of DFFB may subject cells to DNA damage stresses, which in turn may contribute to both tumorigenesis and better response to DNA damaging chemotherapy. Further studies are needed to investigate the role of the DFFB gene in the etiology of oligodendroglioma.
Acknowledgements
This work was partially supported by a gift from the "Anthony Bullock III Research fund".
==== Refs
Reifenberger J Reifenberger G Liu L James CD Wechsler W Collins VP Molecular genetic analysis of oligodendroglial tumors shows preferential allelic deletions on 19q and 1p Am J Pathol 1994 145 1175 1190 7977648
Bello MJ Vaquero J de Campos JM Kusak ME Sarasa JL Saez-Castresana J Pestana A Rey JA Molecular analysis of chromosome 1 abnormalities in human gliomas reveals frequent loss of 1p in oligodendroglial tumors Int J Cancer 1994 57 172 175 8157354
Bello MJ Leone PE Vaquero J de Campos JM Kusak ME Sarasa JL Pestana A Rey JA Allelic loss at 1p and 19q frequently occurs in association and may represent early oncogenic events in oligodendroglial tumors Int J Cancer 1995 64 207 210 7622310
Kraus JA Koopmann J Kaskel P Maintz D Brandner S Schramm J Louis DN Wiestler OD von Deimling A Shared allelic losses on chromosomes 1p and 19q suggest a common origin of oligodendroglioma and oligoastrocytoma J Neuropathol Exp Neurol 1995 54 91 95 7815084
Smith JS Alderete B Minn Y Borell TJ Perry A Mohapatra G Hosek SM Kimmel D O'Fallon J Yates A Feuerstein BG Burger PC Scheithauer BW Jenkins RB Localization of common deletion regions on 1p and 19q in human gliomas and their association with histological subtype Oncogene 1999 18 4144 4152 10435596 10.1038/sj.onc.1202759
Husemann K Wolter M Buschges R Bostrom J Sabel M Reifenberger G Identification of two distinct deleted regions on the short arm of chromosome 1 and rare mutation of the CDKN2C gene from 1p32 in oligodendroglial tumors J Neuropathol Exp Neurol 1999 58 1041 1050 10515227
Dong SM Pang JC Poon WS Hu J To KF Chang AR Ng HK Concurrent hypermethylation of multiple genes is associated with grade of oligodendroglial tumors J Neuropathol Exp Neurol 2001 60 808 816 11487055
Dong Z Pang JS Ng MH Poon WS Zhou L Ng HK Identification of two contiguous minimally deleted regions on chromosome 1p36.31-p36.32 in oligodendroglial tumours Br J Cancer 2004 91 1105 1111 15475940 10.1038/sj.bjc.6602093
Barbashina V Salazar P Holland EC Rosenblum MK Ladanyi M Allelic losses at 1p36 and 19q13 in gliomas: correlation with histologic classification, definition of a 150-kb minimal deleted region on 1p36, and evaluation of CAMTA1 as a candidate tumor suppressor gene Clin Cancer Res 2005 11 1119 1128 15709179
Hanahan D Weinberg RA The hallmarks of cancer Cell 2000 100 57 70 10647931 10.1016/S0092-8674(00)81683-9
Suzuki S Moore DH Ginzinger DG Godfrey TE Barclay J Powell B Pinkel D Zaloudek C Lu K Mills G Berchuck A Gray JW An approach to analysis of large-scale correlations between genome changes and clinical endpoints in ovarian cancer Cancer Res 2000 60 5382 5385 11034075
Ginzinger DG Godfrey TE Nigro J Moore DH Suzuki S Pallavicini MG Gray JW Jensen RH Measurement of DNA copy number at microsatellite loci using quantitative PCR analysis Cancer Res 2000 60 5405 5409 11034080
Nigro JM Takahashi MA Ginzinger DG Law M Passe S Jenkins RB Aldape K Detection of 1p and 19q loss in oligodendroglioma by quantitative microsatellite analysis, a real-time quantitative polymerase chain reaction assay Am J Pathol 2001 158 1253 1262 11290543
Wheeler DL Church DM Edgar R Federhen S Helmberg W Madden TL Pontius JU Schuler GD Schriml LM Sequeira E Suzek TO Tatusova TA Wagner L Database resources of the National Center for Biotechnology Information: update Nucleic Acids Res 2004 32 (Database issue):D35-40 14681353 10.1093/nar/gkh073
Kawane K Fukuyama H Yoshida H Nagase H Ohsawa Y Uchiyama Y Okada K Iida T Nagata S Impaired thymic development in mouse embryos deficient in apoptotic DNA degradation Nat Immunol 2003 4 138 144 12524536 10.1038/ni881
Judson H van Roy N Strain L Vandesompele J Van Gele M Speleman F Bonthron DT Structure and mutation analysis of the gene encoding DNA fragmentation factor 40 (caspase-activated nuclease), a candidate neuroblastoma tumour suppressor gene Hum Genet 2000 106 406 413 10830907 10.1007/s004390000257
|
16156899
|
PMC1242250
|
CC BY
|
2021-01-04 16:36:35
|
no
|
Mol Cancer. 2005 Sep 12; 4:35
|
utf-8
|
Mol Cancer
| 2,005 |
10.1186/1476-4598-4-35
|
oa_comm
|
==== Front
Mol PainMolecular Pain1744-8069BioMed Central London 1744-8069-1-271617908810.1186/1744-8069-1-27ResearchNeonatal local noxious insult affects gene expression in the spinal dorsal horn of adult rats Ren Ke [email protected] Svetlana I [email protected] Fang [email protected] Ronald [email protected] Michael S [email protected] Department of Biomedical Sciences, and Program in Neuroscience, University of Maryland, Baltimore, MD 21201; USA2005 22 9 2005 1 27 27 25 7 2005 22 9 2005 Copyright © 2005 Ren et al; licensee BioMed Central Ltd.2005Ren et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Neonatal noxious insult produces a long-term effect on pain processing in adults. Rats subjected to carrageenan (CAR) injection in one hindpaw within the sensitive period develop bilateral hypoalgesia as adults. In the same rats, inflammation of the hindpaw, which was the site of the neonatal injury, induces a localized enhanced hyperalgesia limited to this paw. To gain an insight into the long-term molecular changes involved in the above-described long-term nociceptive effects of neonatal noxious insult at the spinal level, we performed DNA microarray analysis (using microarrays containing oligo-probes for 205 genes encoding receptors and transporters for glutamate, GABA, and amine neurotransmitters, precursors and receptors for neuropeptides, and neurotrophins, cytokines and their receptors) to compare gene expression profiles in the lumbar spinal dorsal horn (LDH) of adult (P60) male rats that received neonatal CAR treatment within (at postnatal day 3; P3) and outside (at postnatal 12; P12) of the sensitive period. The data were obtained both without inflammation (at baseline) and during complete Freund's adjuvant induced inflammation of the neonatally injured paw. The observed changes were verified by real-time RT-PCR. This study revealed significant basal and inflammation-associated aberrations in the expression of multiple genes in the LDH of adult animals receiving CAR injection at P3 as compared to their expression levels in the LDH of animals receiving either no injections or CAR injection at P12. In particular, at baseline, twelve genes (representing GABA, serotonin, adenosine, neuropeptide Y, cholecystokinin, opioid, tachykinin and interleukin systems) were up-regulated in the bilateral LDH of the former animals. The baseline condition in these animals was also characterized by up-regulation of seven genes (encoding members of GABA, cholecystokinin, histamine, serotonin, and neurotensin systems) in the LDH ipsilateral to the neonatally-injured paw. The largest aberration in gene expression, however, was observed during inflammation of the neonatally injured hindpaws in the ipsilateral LDH, which included thirty-six genes (encoding numerous members of glutamate, serotonin, GABA, calcitonin gene-related peptide, neurotrophin, and interleukin systems). These findings suggest that changes in gene expression may be involved in the long-term nociceptive effects of neonatal noxious insult at the spinal level.
microarrayreal-time RT-PCRneonatal injurycarrageenanpaindevelopment
==== Body
Introduction
Emerging evidence indicates that neonatal noxious insults result in changes in pain responsivity in adults (reviewed in: [1-3]). It has been demonstrated that these changes may, at least in part, be associated with permanent alterations in pain processing in the spinal cord [4-7].
Recently, we have developed a carrageenan (CAR) injection-based rat model of short-lasting (24 – 48 h) locally-induced noxious insult. In this model, rats subjected to a single subcutaneous injection of CAR into a hindpaw after birth, develop a long-term widespread bilateral increase in baseline nociceptive threshold along with a long-term enhanced localized ipsilateral increase in nocifensive responses (enhanced hyperalgesia) to adult inflammation of the neonatally-injured paw [8,9]. We also determined that the sensitive period for generating these effects is within the first postnatal week [9]. The present study was aimed at screening for changes in gene expression, which may be involved in the above-described long-term nociceptive effects of neonatal noxious insult at the spinal level. Here, we largely focused on three aspects of gene expression in the lumbar region of the spinal dorsal horn (LDH) of adult rats receiving CAR injection during the neonatal sensitive period. First, in expectation of gaining insight into the spinal contribution to the bilateral baseline hypoalgesia characteristic of our model, we identified changes in gene expression encompassing LDH on both ipsilateral and contralateral sides to the neonatally injured hindpaw. Second, we identified baseline changes in gene expression unique to the LDH ipsilateral to the site of neonatal injury. The interest in the latter changes was prompted by findings in several studies [4,5,7] that the apparent similarity in the latency of withdrawal responses to acute noxious stimulation in both hindpaws of adult animals, which received neonatal injections of inflammatory agents, occurs in the presence of long-term alterations in the receptive field sizes, background activity, and evoked responses of the LDH neurons receiving information from the neonatally-injured paw. Therefore, we hoped that our observations might aid in explaining why these diverse structural/physiological changes do not result in locally-altered behavioral pain responsivity. Third, we investigated abnormalities in gene regulation, which may be associated with hyperalgesia seen in our model upon adult inflammation of the neonatally-injured paw.
To achieve our goal we employed microarray profiling of gene expression in the ipsilateral and contralateral LDH of adult rats receiving neonatal CAR hindpaw injections either within or outside of the sensitive period, both at baseline and during complete Freund's adjuvant (CFA)-induced inflammation of neonatally CAR-injected paws. The analysis employed custom-designed glass microarrays containing 205 sixty-mer oligo-probes for all known genes for: glutamate, gamma-aminobutyric acid (GABA), adrenaline and serotonin transporters; glutamic acid decarboxylase (GAD); N-methyl-D-aspartate (NMDA), alpha-amino-3-hydroxy-5-methylisoxazole-4-propionic acid (AMPA), kainate, GABAA receptor subunits, glutamate metabotropic, GABAB, adenosine, adrenergic, serotonergic, histamine, and σ receptors; and peptide precursors and receptors for opioids, somatostatin (SOM), neuropeptide Y (NPY), tachykinins, neurotensin, calcitonin gene-related peptide (CGRP), cholecystokinin (CCK), and vasoactive intestinal peptide (VIP). Probes for a range of neurotrophins and cytokines and their receptors were also included. Each microarray-identified change in gene expression was verified by quantitative real-time reverse transcription-polymerase chain reactions (real-time RT-PCRs).
Results
Microarray-based Gene Expression Profiling
In assessing the microarray-generated data, we began by looking at the results of comparison of gene expressions in the adult (60-days-old; P60) male rats receiving a single CAR injection within the sensitive period (on P3 [9]; CAR-P3 group) and rats receiving CAR injection outside the sensitive period (on P12 [9]; CAR-P12 group) with those in the animals receiving no such injections (CONT group). Then, we turned to comparison of gene expressions between CAR-P3 and CAR-P12 rats to identify changes that may be specifically related to the CAR injection at P3. The latter comparison was aimed at discarding changes that might be common to both CAR treatments within and outside of the sensitive period.
Baseline Levels of Gene Expression
The search for bilateral effects identified seven genes (coding for GABAB1d receptor, 5-HT5a serotonergic receptor, A1 adenosine receptor, σ receptor, NPYR6 receptor, preproenkephalin, and γ-tachykinin) with significantly increased expression in the ipsilateral and contralateral LDH of CAR-P3 rats as compared to those in CONT animals (Table 1). These increases ranged from 1.5 to 3.5 fold. No significant alterations in the expression of these genes were detected in the CAR-P12 group (Table 1), which made the observed changes specific for CAR-P3 animals. Six more genes (coding for GAD65, α2b and β3 adrenergic receptors, δ-preprotachykinin, CCKR-1 receptor, and interleukin IL-10) showed higher bilateral levels of gene expression in both CAR-3 and CAR-P12 groups (Table 1). However, in CAR-P3 rats these genes were expressed at significantly higher levels (by 1.2 – 2.6 fold) than in CAR-P12 animals (Table 1). Finally, two genes (NR2B subunit of NMDA receptor and SstR2 somatostatin receptor), while being up-regulated in both CAR-P3 and CAR-P12 groups relative to CONT group, showed significantly lower levels of expression (by 0.5 and 0.7 fold respectively) in CAR-P3 rats as compared to those in CAR-P12 animals (Table 1).
Table 1 Genes with significantly altered basal bilateral levels in the LDH of CAR-P3 as compared to their bilateral expressions in the LDH of CONT and CAR-P12 animals [one-way ANOVAs, (with the Benjamini and Hochberg's FDR correction procedure set at the P-value cutoff of 0.05) followed by Tukey's post-hoc tests (with the same P-cutoff value) in the implementation of GeneSpring GX software (Agilent, Palo Alto, CA).)]. Differences in gene expression are presented as ratios between CAR-P3 and CONT, CARP-12 and CONT, and CAR-P3 and CAR-P12 groups. Statistically insignificant differences are marked as 'ND' The comparisons are based on the microarray profiling that employed 5 slides for comparison of CONT and CAR-P3 samples and another 5 slides for comparison of CONT and CAR-P12 samples.
Genes Left LDH (ipsilateral to the site of neonatal injury) Right LDH (contralateral to the site of neonatal injury)
CAR-P3/CONT CAR-P12/CONT CAR-P3/CAR-P12 CAR-P3/CONT CAR-P12/CONT CAR-P3/CAR-P12
GABAB1d 1.6 ND 2.1 2.2 ND 2.2
GAD65 2.1 1.8 1.2 2.5 1.4 1.8
A1 aden. rec. 2.2 ND 2.0 1.55 ND 1.64
Pre-proenkephalin 2.0 ND 1.5 1.5 ND 1.9
σ rec. 3.2 ND 2.9 2.3 ND 2.5
NPYR6 3.5 ND 2.7 2.8 ND 2.8
γ-tachikinin 2.1 ND 2.3 2.7 ND 3.1
δ-preporotachykinin 2.2 1.5 1.5 3.1 1.8 1.7
α2b adr. rec. 2.0 1.6 1.2 3.0 1.7 1.8
β3 adr. rec. 2.2 1.7 1.3 3.0 1.7 1.8
5-HT5a 2.0 ND 1.8 1.6 ND 1.2
CCKR-1 2.2 1.8 1.2 3.4 2.3 1.5
IL-10 4.3 1.6 2.6 2.0 1.4 1.4
NR2B 2.2 3.7 0.6 2.4 5.0 0.5
SstR2 2.2 3.2 0.7 3.9 5.3 0.7
In addition to bilateral changes, comparison of the CONT and CAR-P3 groups revealed neonatal CAR injection-induced up-regulation (by 1.9–3.2 fold) of seven genes (coding for GABAB1f, GABAB1b, 5-HT1b serotonergic, NtsR1 neurotensin and H1 histamine receptors, GAG67, and preprocholecystokinin) confined to the left LDH ipsilateral to the neonatal injury (Table 2). These changes were undetectable in CAR-P12 rats, indicating their CAR-P3 specificity.
Table 2 Genes with significantly altered basal levels in the left LDH ipsilateral to the neonatally-injured hindpaw in CAR-P3 as compared to their expressions in the corresponding LDH of CONT and CAR-P12 animals [one-way ANOVAs, (with the Benjamini and Hochberg's FDR correction procedure set at the P-value cutoff of 0.05) followed by Tukey's post-hoc tests (with the same P-cutoff value) in the implementation of GeneSpring GX software (Agilent, Palo Alto, CA).)]. Differences in gene expression are presented as ratios between CAR-P3 and CONT, CAR-P12 and CONT, and CAR-P3 and CAR-P12 groups. Statistically insignificant differences are marked as 'ND' The comparisons are based on the microarray profiling that employed 5 slides for comparison of CONT and CAR-P3 samples and another 5 slides for comparison of CONT and CAR-P12 samples.
Left LDH (ipsilateral to the site of neonatal injury)
Genes CAR-P3/CONT CAR-P12/CONT CAR-P3/CAR-P12 Genes CAR-P3/CONT CAR-P12/CONT CAR-P3/CAR-P12
GABAB1b 1.9 ND 1.7 NtsR1 2.1 ND 1.6
GABAB1f 3.2 ND 2.0 5-HT1b 2.0 ND 1.8
GAD67 2.0 ND 1.4 Preprocholecystokinin 2.2 ND 2.0
H1 hist. rec. 3.2 ND 1.7
Adult inflammation-induced Changes in Gene Expression in the LDH Ipsilateral to the Neonatally-injured Hindpaw
In the left LDH ipsilateral to the site of neonatal injury in CAR-P3 rats, twenty-six genes showed side-selective adult inflammation-induced up-regulation (ranging from 1.4 to 6.1 fold) compared to their expressions in the corresponding LDH of CONT animals with similarly inflamed left hindpaw (Table 3). These included genes coding for: NR1-4a and NR3B subunits of NMDA receptor, GluRD and GluR6 subunits of AMPA receptor, GABAAα2 and GABAAα4, subunits of GABAA receptor, GABAB1b receptor, 5-HT1a, 5-HT1d, 5-HT1e, 5-HT1f, 5-HT2c, 5-HT4, 5-HT5a, and 5-HT6 serotonergic receptors, calcitonin gene-related polypeptide-2 (CALK-2), neurotrophin 3 (NT-3), TrkA and TrkC neurotrophin receptors, IL-1α, IL-6, IL-12A, IL-15, IL-20, and IL-24 interleukins, and IL-12 β1 interleukin receptor. No significant changes in these genes were observed in the CAR-P12 group (Table 3), indicating that the aforementioned alterations in gene expression were CAR-P3-specific. Additionally, thirteen genes exhibited similarly ipsilateral up-regulations in both CAR-P3 and CAR-P12 rats, but the increases in the CAR-P3 group were significantly greater (by 1.1 – 2.4 fold) than those in the CAR-P12 group (Table 3). These genes encode: NR1-1b, NR2A, and NR2D subunits of NMDA receptor, mGluR1a metabotropic receptor, GABAAα5 and GABAAα6, subunits of GABAA receptor, GABAB1c receptor, 5-HT5b serotonergic receptor, calcitonin gene-related peptide receptor (CGRPR), NT-4 neurotrophin, and IL-5 interleukin. Finally, one gene (encoding NK2 tachykinin receptor), while also being up-regulated in the CAR-P3 and CAR-12 groups, showed a significantly weaker increase in CAR-P3 rats as compared to that in CAR-P12 animals (by 0.7 fold).
Table 3 Genes with significant adult inflammation-induced alterations in their expressions in the left LDH (ipsilateral to the CFA-inflamed neonatally-injured hindpaw) in CAR-P3 rats as compared to those in the left LDH of CONT and CAR-P12 animals after inflammation of the corresponding paw [one-way ANOVAs, (with the Benjamini and Hochberg's FDR correction procedure set at the P-value cutoff of 0.05) followed by Tukey's post-hoc tests (with the same P-cutoff value) in the implementation of GeneSpring GX software (Agilent, Palo Alto, CA).)]. Differences in gene expression are presented as ratios between CAR-P3 and CONT, CAR-P12 and CONT, and CAR-P3 and CAR-P12 groups. Statistically insignificant differences are marked as 'ND' The comparisons are based on the microarray profiling that employed 5 slides for comparison of CONT and CAR-P3 samples and another 5 slides for comparison of CONT and CAR-P12 samples.
Left LDH (ipsilateral to the site of neonatal injury)
Genes CAR-P3/CONT CAR-P12/CONT CAR-P3/CAR-P12 Genes CAR-P3/CONT CAR-P12/CONT CAR-P3/CAR-P12
5-HT1a 3.3 ND 2.4 GABAAα2 2.0 ND 2.0
5-HT1d 1.9 ND 2.1 GABAAα4 1.8 ND 2.0
5-HT1e 1.6 ND 1.6 GABAAα5 1.9 1.4 1.4
5-HT1f 1.8 ND 2.0 GABAAα6 2.9 1.7 1.7
5-HT2c 1.4 ND 1.7 GABAB1b 2.6 ND 2.4
5-HT4 1.9 ND 2.4 GABAB1c 2.2 1.5 1.5
5-HT5a 3.0 ND 2.3 NR1-1b 3.2 2.4 1.3
5-HT5b 3.1 2.1 1.5 NR1-4a 2.2 ND 2.0
5-HT6 1.7 ND 1.7 NR2A 2.1 1.5 1.4
IL-1α 1.6 ND 1.6 NR2D 2.3 1.5 1.5
IL-5 2.5 1.6 1.6 NR3B 1.6 ND 2.0
IL-6 1.8 ND 2.2 GluRD 2.0 ND 2.0
IL-12A 1.7 ND 1.3 GluR6 2.1 ND 1.7
IL-12R β1 6.1 ND 4.3 mGluR1a 2.5 1.4 1.8
IL-15 2.4 ND 2.0 CGRPR 2.6 1.5 2.4
IL-20 3.4 ND 2.4 NT-4 2.0 1.2 1.6
IL-24 3.5 ND 3.0 CALK2 2.5 ND 1.9
TrkA 2.3 ND 2.0 VGF 2.6 ND 1.4
TrkC 1.5 ND 1.4 NK2 1.5 2.2 0.7
Among the genes examined in this study, only one (encoding σ receptors) showed significant bilateral changes in its expression in CAR-P3 rats (but not in CAR-P12 animals) after adult inflammation of neonatally-injured hindpaw (left side: CAR-P3/CONT = 3.7 and CAR-P3/CAR-P12 = 2.6; right side: CAR-P3/CONT = 2.8 and CAR-P3/CAR-P12 = 1.9).
Verification of the Microarray-identified Alterations in Gene Expression by Real-time RT-PCR
The real-time RT-PCR analysis supported the findings of the microarray gene profiling, with two exceptions (Figs. 1, 2, 3). First, at the baseline, the expressions of the gene coding for 5-HT5a receptor, although upregulated vs. CONT rats were similar in CAR-P3 and CAR-P12 animals on the side ipsilateral to the neonatally-injured paw. There were also no differences in the expression of this gene between all three groups on the contralateral side (Fig. 1). Second, no differential expression of the gene for NT4 factor was detected between our animal groups after adult inflammation (Fig. 3). Consequently, these effects identified by microarrays were judged to be false.
Figure 1 Real-time RT-PCR-determined levels of expression (calculated as R0 using β-actin as internal control) for genes identified by microarray profiling as having significantly altered basal bilateral levels in the LDH of CAR-P3 as compared to their bilateral expressions in the LDH of CONT and CAR-P12 animals (Table 1). A, genes, which were identified by microarrays as having similar expressions in CONT and CARP-12 groups, but changed (up-regulated) in CAR-P3 group. B, genes, which were identified by microarrays as having differences in expression between CONT and CAR-P3 and CAR-P12 groups as well as between CARP-3 and CAR-P12 groups (both when CAR-P3 > CAR-P12 and CAR-P3 < CAR-P12). The data for the LDH ipsilareal to the neonatally-injured paw and the LDH contralateral to that paw are shown separately. Each data-point represents a mean of 5 animals in a group. The data were analyzed using one-way ANOVAs, (with the Benjamini and Hochberg's FDR correction procedure set at the P-value cutoff of 0.05) followed by Tukey's post-hoc tests (with the same P-cutoff value) in the implementation of GeneSpring GX software (Agilent, Palo Alto, CA). The presented real-time RT-PCR data support the microarray-based findings except when marked by '$' sign (microarrays suggested CARP-3 expression > CARP12 expression, while RT-PCR showed CAR-P3 expression = CAR-P12 expression) and '&' sign (in contrast to microarrays, no inter-group differences were detected by RT-PCR).
Figure 2 Real-time RT-PCR-determined levels of expression (calculated as R0 using β-actin as internal control) for genes identified by having significantly altered basal levels in the left LDH ipsilateral to the neonatally-injured hindpaw in CAR-P3 as compared to their expressions in the corresponding LDH of CONT and CAR-P12 animals (Table 2). The data for the LDH ipsilareal to the neonatally-injured paw and the LDH contralateral to that paw are shown separately. Each data-point represents a mean of 5 animals in a group. The data were analyzed using one-way ANOVAs, (with the Benjamini and Hochberg's FDR correction procedure set at the P-value cutoff of 0.05) followed by Tukey's post-hoc tests (with the same P-cutoff value) in the implementation of GeneSpring GX software (Agilent, Palo Alto, CA). The presented real-time RT-PCR data support the microarray-based findings.
Figure 3 Real-time RT-PCR-determined levels of expression (calculated as R0 using β-actin as internal control) for genes identified by microarray profiling as showing significant adult inflammation-induced alterations in their expressions in the left LDH (ipsilateral to the CFA-inflammed neonatally-injured hindpaw) in CAR-P3 rats as compared to those in the left LDH of CONT and CAR-P12 animals after inflammation of the corresponding paw (Table 3). A, genes, which were identified by microarrays as having similar expressions in CONT and CARP-12 groups, but changed (up-regulated) in CAR-P3 group. B, genes, which were identified by microarrays as having differences in expression between CONT and CAR-P3 and CAR-P12 groups as well as between CARP-3 and CAR-P12 groups (both when CAR-P3 > CAR-P12 and CAR-P3 < CAR-P12). Each data-point represents a mean of 5 animals in a group. The data were analyzed using one-way ANOVAs, (with the Benjamini and Hochberg's FDR correction procedure set at the P-value cutoff of 0.05) followed by Tukey's post-hoc tests (with the same P-cutoff value) in the implementation of GeneSpring GX software (Agilent, Palo Alto, CA). The presented real-time RT-PCR data support the microarray-based findings except when marked by '$' sign (microarrays suggested CARP-3 expression > CARP12 expression, while RT-PCR showed CAR-P3 expression = CAR-P12 expression).
Discussion
This study has revealed that CAR injection of a hindpaw in rat pups during neonatal sensitive period leads to abnormal expression levels of multiple genes in the LDH of adult animals, both at the basal and CFA-induced inflammation states. All of the affected genes showed up-regulation, which parallels the earlier observation that peripheral nerve injury largely causes up-regulation of genes encoding signal carrying and modulating proteins in the LDH [10]. Our findings support the notion that regulatory mechanisms of gene expression in the pain-processing circuitry of the spinal cord are vulnerable to long-term alterations by early local noxious insult.
Bilateral Alterations in Gene Expression in the LDH at Baseline
Among 205 genes profiled in this study, the expression of twelve genes were up-regulated in a bilateral fashion in the LDH of CAR-P3 rats either exclusively or much stronger than that in CAR-P12 animals. These genes encode members of several protein groups, such as GABA synthesis enzymes and receptors, interleukins and their receptors, serotonin, adenosine, σ, NPY, and CCK receptors, and opioid, and tachykinin peptides; yet, each of these groups is represented only by one or two genes suggesting that, at least for the examined genes, the observed changes in expression are diffused, without any group being perceived as a dominant carrier of these changes. However, it is interesting that the translational products of the majority of these genes are most likely to enhance inhibitory processing of nociceptive input in the LDH. These include genes encoding GAD65 GABA synthesis enzyme, anti-inflammatory cytokine, IL-10, and GABAB1d, A1 adenosine, α2b and β3 adrenergic, and NPYR6 receptors [11-23]. The remaining genes encode largely pro-nociceptive products, CCKR-1 and σ-receptors, γ-tachykinin, and δ-preprotachikinin [24-26]. Overall, these data are consistent with the previous findings of bimodal inhibitory and facilitatory changes in descending spinal modulation of nociception after tissue injury [27]. Nevertheless, this prevalence of bilateral changes in gene expression promoting anti-nociception points to a possible net increase in the spinal nociception inhibitory drive, which would be a contributing factor to the bilateral hypoalgesia in CAR-P3 rats. An enhanced spinal anti-nociceptive drive is further supported by our recent microarray gene profiling studies in the periaqueductal gray (PAG), which is a major brain center involved in endogenous pain control. These studies showed that the PAG in CAR-P3 animals is characterized by up-regulation of several genes involved in promoting descending inhibition in these animals [28].
It should also be noted that not all alterations in gene expression were stronger in CAR-P3 then in CARP-12 animals. We found two genes that were up-regulated in CAR-P3 group less than in CARP-12 group; one (encoding NR2B subunit of MNDA receptor) is likely pro-nociceptive, while another (encoding SstR2 receptor) is anti-nociceptive [29,30].
Changes in Gene Expression in the LDH Ipsilateral to Neonatally-Injured Hindpaw at the Baseline and During Adult Inflammation
Without adult inflammation, seven genes showed significant CAR-P3-specific up-regulation in their expression in the LDH ipsilateral to the neonatally-injured hindpaw. Three of these genes code for anti-nociceptive products involved in GABA neurotransmission: GAD 67, and GABAB1b and GABAB1f receptors [12-14], while the other four genes generate potentially pro-nociceptive products, preprocholecystokinin, and NtsR1 neurotensin, H1 histamine and 5-HT1b serotonergic receptors [26,31-34]. These findings do not support our expectation of strong gene-expression driven baseline inhibition of pain processing in the LDH on the neonatally-injured side. Clearly, more studies are needed to reconcile long-term changes induced specifically in the ipsilateral LDH by neonatal inflammation of a hindpaw with the apparent similarity in baseline pain responsiveness of both hindpaws.
Previously, we suggested [9] that specific localized early noxious insult-induced long-term deficits in the molecular organization of the spinal circuitry processing nociceptive information from the neonatally-injured hindpaw are likely to be most notable during a strong local challenge, such as CFA-induced inflammation of the neonatally-injured paw. Indeed, the present study noted that adult inflammation of the left hindpaw (which had received CAR injections in CAR-P3 and CAR-P12 animals) induced significant up-regulation in expression of thirty-six genes in the ipsilateral LDH of CAR-P3 rats as compared to that in similarly-inflamed CONT animals. These differences were either absent or were significantly less pronounced in CAR-P12 animals. Furthermore, in contrast to the baseline conditions, the adult inflammation-induced changes in multiple members of several large gene groups in our arrays. In particular, eight of the affected genes encode interleukins (IL-1α, IL-5, IL-6, IL-12A, IL-15, IL-20, IL-24) and the interleukin receptor, IL-12 β1, which are pro-inflammatory molecules [35-39]. Another eight genes encode glutamate neurotransmission-related proteins, including metabotropic mGluR1a receptor, subunits of NMDA receptor (NR1-4a, NR1-1b, NR2A, NR2D, NR1-4a) and AMPA glutamate (GluRD and GluR6) receptors, which play mostly pro-nociceptive roles in the LDH [13,40]. The up-regulated subunits of NMDA and AMPA receptors are known to have multiple roles in modulating functionality of these receptors, nevertheless, the overall effect of their up-regulation is likely to increase NMDA and AMPA receptor ligand sensitivity and intracellular signaling [41-46]. The next group of eight genes encodes eight 5-HT serotonergic receptors (5-HT1a, 5-HT1d, 5-HT1e, 5-HT1f, 5-HT2c, 5-HT4, 5-HT5a, and 5-HT6). Unfortunately, the specific outcome of the observed up-regulation of these receptors is difficult to predict, since the actions of most of these receptors have not been tested in spinal cord neurons, and those receptors that have been examined are capable of exerting both pro- and anti-nociceptive actions depending on their cellular position within the spinal dorsal horn [13]. Finally, a group of six genes encodes either subunits of the GABAA receptor (GABAAα2, GABAAα4, GABAAα5, and GABAAα6 – all subunits enhancing activity of GABAA ionic channels; [47]) or GABAB receptors (GABAB1b and GABAB1c). As mentioned earlier, these receptors inhibit neuronal facilitation in the spinal dorsal horn [12-14]. The remaining genes code for CGRP, CGRPR receptor, and CALK-2, which have been suggested to have a pro-nociceptive influence in the LDH [13], and neurotrophin-3, and TrkA and TrkC neurotrophin receptors, which reportedly participate in the neurotrophin-associated maintenance of the central spinal sensitization associated with persistent pain [48-53]. Based on the aforementioned observations, it is reasonable to assume that the abnormal regulation of expression of numerous genes involved in processing nociceptive information in the LDH ipsilateral to the neonatally CAR-injected hindpaw may play a significant role in the enhanced hyperalgesia during inflammation of this paw in adult rats.
Among the examined genes, only one (encoding NK2 receptor) showed stronger adult inflammation-induced expression in animals subjected to neonatal CAR injection outside of the sensitive period as compare to those receiving CAR within this period. Activation of NK2 is known to promote analgesia [54].
Finally, it is interesting that very few genes expressing significant inter-group differences in their levels prior to inflammation showed such differences after inflammation was re-introduced in adult animals. Only GABAB1b and σ receptors continued to display significant up-regulation upon adult inflammation of the neonatally-injured paw in the LDH of CAR-P3 rats in the ipsilateral and bilateral manner respectively. Unfortunately, the microarray analysis as performed in this study, does not allow us to discern the reasons why the majority of genes displaying baseline CAR-P3-specific changes failed to show alterations during adult inflammation.
The limited screening of gene expressions described in this paper is just the first step toward understanding the molecular mechanisms involved in long-term alteration in nociception produced by early local noxious insult. Both much wider high throughput profiling studies, including those of proteins, and more focused and detailed investigations, addressing the sites of molecular, alterations on the cellular level and experimental testing of their functional consequences, are needed to fully address this question.
Methods
Animals
Adult female and male Sprague-Dawley breeders were purchased from Harlan (Indianapolis, IN) and bred at the University of Maryland Animal Facility. Litters with ≥10 pups were selected for this study. They were divided into three groups. In the CAR-P3 and CAR-P12 groups, the pups received a single injection of 0.25% CAR in sterile 0.9% saline (1 μl/g) in to the plantar surface of their left hindpaw at 9:00 pm on either P3 or P12 respectively. The injection was conducted with a microliter syringe with a 31-gauge needle. Previously, we demonstrated that, at both ages, such an injection induces short-lasting (~48 h) increase in the diameter of the injected paw, accompanied by reduced withdrawal latencies to thermal and mechanical noxious stimulation [8]. Pups in the CONT group received no injections. Except for the hindpaw injection procedure, all rats were subject to the same human intervention involved in routine animal care such as weekly cage cleaning and daily food and water changes, etc. We did not include a brief handling control group since brief handling after birth has been shown not to produce long-term alterations in pain responsivity [8]. A saline-injected control group also was not included since saline injection by itself is a noxious experience and it is beyond the scope of the present study to distinguish between long-term consequences of different types of early noxious insults. After weaning at P21, male pups were housed in groups of three with animals in any given cage receiving the same neonatal treatment. The females were used in different experiments (we understand the importance of extending the present studies to females, which will be done in the future). At P59, the left hindpaws of half of the rats in CONT, CAR-P3 and CAR-P12 groups received an inflammation-inducing injection of CFA (0.05 ml, 1:1 oil/saline emulsion). The rest of the animals received no injections. Twenty-four hours later (at P60), all rats were anesthetized by i.p. injections of Nembutal. The left and right LDH of the L4, L5 levels were dissected out and snap-frozen in liquid nitrogen. All experiments were approved by the University Institutional Animal Care and Use Committee.
Microarrays
In this part of the study, we used left and right sides of the LDH from ten CAR-P3, ten CAR-12, and ten CONT animals, one half at baseline and another half after CFA-induced inflammation of the neonatally-injured paw per group. For both baseline and inflammation states, each animal belonged to a different litter. For each state, the collected samples were processed in randomly selected CAR-P3/CONT and CAR-P12/CONT pairs per microarray slide [5 pairs each; 10 pairs for both comparison groups × 2 (left and right) = 20 microarray slides total/inflammation state].
For the analysis, the total sample RNA was isolated with TRIzol protocol (Invitrogen, Carsbad, CA). The RNA yield was determined using a DU 640 Spectrophotometer (Beckman, Coulter, Fullerton, CA). The 260/280 nm ratios of the samples were 1.8. The RNA isolated from a specific pairs of samples (5 μg RNA per sample) was reverse transcribed utilizing a 3DNA Array Detection Kit (Genesphere, Hartfield, PA); with primers containing different capture sequences for CONT and experimental (CAR-P3 and CARP-12] tissue. The resultant cDNA was then hybridized to oligonucleotide microarrays custom designed and printed for us by TeleChem International (Sunnyvale, CA). The sequences of the chosen mRNAs were obtained from the Gene Bank with their unique regions identified with the help of the 'BLAST-2 Sequences' and 'Multiple Alignment' websites ( and respectively). The same two websites were used for the verification of the uniqueness of the oligoprobe sequences selected by TeleChem International. Probes for four internal standards, glyceradehyde-3-phosphate-dehydrogenase (GAPDH), β-actin, ubiquitin, and βIII-tubulin, were also included in the microarrays. The oligonicleotide probes were printed with the TeleChem's advanced Stealth technology on SuperAldehyde substrate coated-glass slides in quadruplicates, with each of the four sets of probes occupying a different quadrant of the slide microarray matrix. The cDNA hybridization was performed at 55°C for 16 h in 2 × SDS-Based Hybridization Buffer (3DNA Array Detection Kit, Genesphere, Hartfield, PA). The hybridization was followed by washing with 2 × SSC and 0.2% SDS at 42°C, 2 × SSC at room temperature, and 0.2 × SSC at room temperature, 10 min each. Washed slides were air-dried for 30 sec. For fluorescent labeling of microarray-bound cDNA, the slides were incubated for 3 h at 65°C with a Capture Reagent Hybridization Mixture from a 3DNA Array Detection Kit (Genesphere Inc, Hatfield, PA). In this mixture, Alexa 546 fluorochrome-incorporating dendrimers contained single-stranded arms complimentary to the 'capture' sequences used in the reverse transcription of RNA from CONT samples, while Alexa 647 fluorochrome-incorporating dendrimers contained arms complimentary to the capture sequences used in reverse transcription of RNA from CAR-P3 and CARP-12 samples. The labeling reaction was terminated by washing the microarray slides at room temperature for 10 min with 0.005% Triton in 2 × SSC and for another 10 min in 0.2 × SSC. After that, the slides were air-dried for 30 sec. The processed microarray slides were scanned on a GenePix 4100A scanner (Axon Instrument, Union City, CA) with the laser excitation at 532 nm [emission filter 575DF35 (green); photomultiplier voltage 550] for Alexa 546 (control samples) and the laser excitation at 635 nm [emission filter 670DF40 (red); photomultiplier voltage 695] for Alexa 647 (experimental samples). The densitometry was performed with GenePix Pro 4.1 (Axon Instruments, Union City, CA). Background was subtracted using the 'Local Background Correction' procedure and quality control was achieved by 'Quality Control' feature of the GenePix software. Since our microarray slides generated four spots per gene, the median intensity of these quadruplicates was calculated. For each array, the data were normalized with Acuity 3.1 (Axon Instruments, Union City, CA) by applying locally weighted scatterplot smoothing (LOWESS) transformation that utilizes all genes in the microarray [55] and by the internal standards, GAPDH, beta-actin, ubiquitin and βIII-tubulin [56]. LOWESS normalization eliminated non-linear die distortion of the microarray data [55]. The statistical analysis was conducted separately for each inflammation state of the adult animals and the left and right sides as described in [28]. For this analysis, the data from the CONT samples were adjusted to produce a uniform value of 1. Then, the levels of the gene expressions between CAR-P3, CAR-P12, and CONT groups were compared employing one-way ANOVAs (with the false discovery rate (FDR) multiple testing error correction procedure of Benjamini and Hochberg [57] set at the P-value cutoff of 0.05) followed by Tukey's post-hoc tests (with the same P-cutoff value) in the implementation of GeneSpring GX software (Agilent, Palo Alto, CA).
Real-Time RT-PCR
For the verification of the changes in gene expression suggested by the microarray analysis, real-time RT-PCR was conducted in 5 samples per relevant group, each obtained from an animal belonging to a different litter. Also, none of these litters were used in the above-described microarray analysis. All the baseline changes were re-examined bilaterally. The changes induced by CFA-based inflammation were re-examined only ipsilaterally. β-actin was employed as the endogenous control [28]. For each assay, the total RNA was isolated as described for microarray analysis, and then treated with Amplification Grade Deoxyribonuclease I (1 u/μg RNA; Invitrogen, Carlsbad, CA) for 15 min at 25°C. Reverse transcription was performed using Omniscript RT kit (Qiagen, Vanencia, CA) using random hexamers as primers. Real-time PCR utilized TaqMan assay, which requires two primers and a fluorescence-labeled probe for cDNA amplification and visualization. All TaqMan PCR primers and probes were designed using Primer Express 1.5a (Applied Biosystems, Foster City, CA) and custom-synthesized by Applied Biosystems (Foster City, CA). The PCR reactions were carried out on an ABI Prism 7000 Sequence Detector (Applied Biosystems, Foster City, CA). All PCR reactions were run in a monoplex mode – amplification of one gene per reaction well in a PCR plate. Each reaction included sample cDNA (1 μl), the reverse primers (100–500 nM each) and probes (50–100 nM) for the gene of interest, TaqMan Universal Master Mix (25 μl, Applied Biosystems) and deionized water that brought the total volume to 50 μl. The cycles were run at 95°C/10 min, 95°C/15 sec × 45, and 60°C/1 min. All assays were performed in triplicates. The normalization to internal standards and expression of the resultant data was calculated using amplification plot method of Pierson et al. [58] implemented in DART-PCR Excel workbook . In this method amplification efficiency for both target and control genes were calculated from raw data around the midpoint of the transformed signal range, which is more accurate than when derived from an external standard curve [59]. The average efficiency from all (experimental and control) runs of a given gene was then calculated (since this improves accuracy as compared to the use of individual run efficiencies; [58]). The obtained efficiency is used in generating Starting Fluorescence Value for a given gene in a sample (R0), which is proportional to the starting template quantity. Finally, R0 for the target gene is normalized to R0 of the internal control from the same sample [R0(sample) = R0(target)/R0(control)]. To allow inter-plate comparison, the R0(sample) is further normalized to R0(standard) generated for the same target gene and internal control in the wells with the "standard" cDNA run on the same plate with the experimental samples [R0(sample 2 × normalized) = R0(sample)/R0(standard)]. The 'standard' cDNA for the entire study was generated from a 'standard tissue homogenate' that was pooled from the LDH tissue of 10 rats dedicated for this purpose. To assess possible contamination with chromosomal DNA, several RNA samples were processed for PCR with β-actin probe and primers without the RT step. No amplification products were detected.
The statistical analyses of samples from the left and right sides and from different inflammation states of adult animals were performed separately. These analyses include one-way ANOVAs (with the Benjamini and Hochberg's FDR correction procedure set at the P-value cutoff of 0.05) followed by Tukey's post-hoc tests (with the same P-cutoff value) in the implementation of GeneSpring GX software (Agilent, Palo Alto, CA).
Competing interests
The author(s) declare that they have no competing interests.
Acknowledgements
This work was supported by NIH grant NS41384.
==== Refs
Johnston CC Walker CD Boyer K Animal models of long-term consequences of early exposure to repetitive pain Clin Perinatol 2002 29 395 414 12380465 10.1016/S0095-5108(02)00020-9
Lidow MS Long-term effects of neonatal pain on nociceptive systems Pain 2002 99 377 383 12406512 10.1016/S0304-3959(02)00258-0
Fitzgerald M Painful beginnings Pain 2004 110 508 509 15288390 10.1016/j.pain.2004.05.022
Rahman W Fitzgerald M Aynsley-Green A Dickenson A Jensen TS, Turner JA, Weisenfeld-Hallin Z The effects of neonatal exposure to inflammation and/or morphine on neuronal responses and morphine analgesia in adult rats Proceedings of the 8th World Congress on Pain 1997 Seattle: IASP Press 783 794
Ruda MA Ling QD Hohmann AG Peng YB Tachibana T Altered nociceptive neuronal circuits after neonatal peripheral inflammation Science 2000 289 628 631 10915627 10.1126/science.289.5479.628
Tachibana T Ling QD Ruda MA Increased Fos induction in adult rats that experienced neonatal peripheral inflammation Neuroreport 2001 12 925 927 11303761 10.1097/00001756-200104170-00012
Peng YB Ling QD Ruda MA Kenshalo DR Electrophysiological changes in adult rat dorsal horn neurons after neonatal peripheral inflammation J Neurophysiol 2003 90 73 80 12634281
Lidow MS Song ZM Ren K Long-term effects of short-lasting early local inflammatory insult Neuroreport 2001 12 399 403 11209957 10.1097/00001756-200102120-00042
Ren K Anseloni V Zou SP Wade EB Novikova SI Ennis M Traub RJ Gold MS Dubner R Lidow MS Characterization of basal and re-inflammation-associated long-term alteration in pain responsivity following short-lasting neonatal local inflammatory insult Pain 2004 110 588 596 15288399 10.1016/j.pain.2004.04.006
Yang L Zhang FX Huang F Lu YJ Li GD Bao L Xiao HS Zhang X Peripheral nerve injury induces trans-synaptic modification of channels, receptors and signal pathways in rat dorsal spinal cord Eur J Neurosci 2004 19 871 883 15009134 10.1111/j.0953-816X.2004.03121.x
Cooper JR Bloom FE Roth RH The Biochemical Basis of Neuropharmacology 1996 New York: Oxford University Press
Green GM Dickenson A GABA-receptor control of the amplitude and duration of the neuronal responses to formalin in the rat spinal cord Eur J Pain 1997 1 95 104 15102410 10.1016/S1090-3801(97)90067-7
Millan MJ Descending control of pain Prog in Neurobiol 2002 66 355 474 10.1016/S0301-0082(02)00009-6
Malan TP Mata HP Porreca F Spinal GABA(A) and GABA(B) receptor pharmacology in a rat model of neuropathic pain Anesth 2002 96 1161 1167 10.1097/00000542-200205000-00020
Fields HL Basbaum AI Wall PD, Melzack Central nervous system mechanisms of pain Textbook of Pain 1999 England: Churchill livingstone 309 330
Sawynok J Adenosine receptor activation and nociception Eur J Pharmacol 1998 347 1 11 9650842 10.1016/S0014-2999(97)01605-1
Furst S Transmitters involved in antinociception in the spinal cord Brain Res Bull 1999 48 129 141 10230704 10.1016/S0361-9230(98)00159-2
Peng YI Liu HJ Guo L Fu TC Mechanisms of adrenaline-induced antinociception in mice Chin J Physiol 1992 35 205 210 1295718
Hua XY Boublik JH Spicer MA Rivier JE Brown MR Yaksh TL The anti-nociceptive effects of spinally administered neuropeptide Y in the rat: systematic studies on structure-activity relationship J Pharmacol Exp Ther 1991 258 243 248 1677040
Wettstein JG Earley B Junien JL Central nervous system pharmacology of neuropeptide Y Pharmacol Ther 1995 65 397 414 7644568 10.1016/0163-7258(95)98598-K
Silva AP Cavadas C Grouzmann E Neuropeptide Y and its receptors as potential therapeutic drug targets Clin Chim Acta 2002 326 3 25 12417094 10.1016/S0009-8981(02)00301-7
Boddeke EW Involvement of chemokines in pain Eur J Pharmacol 2001 429 115 119 11698032 10.1016/S0014-2999(01)01311-5
DeLeo JA Yezierski RP The role of neuroinflammation and neuroimmune activation in persistent pain Pain 2001 90 1 6 11166964 10.1016/S0304-3959(00)00490-5
Chien CC Pasternak GW Sigma antagonists potentiate opioid analgesia in rats Neurosci Lett 1995 190 137 139 7644123 10.1016/0304-3940(95)11504-P
Honore P Menning PM Rogers SD Nichols ML Mantyh PW Neurochemical plasticity in persistent inflammatory pain Prog Brain Res 2000 129 357 363 11098703
Wiesenfeld-Hallin Z Xu XJ Hokfelt T The role of spinal cholecystokinin in chronic pain states Pharmacol Toxicol 2002 91 398 403 12688385 10.1034/j.1600-0773.2002.910619.x
Ren K Dubner R Descending modulation in persistent pain: An update Pain 2002 100 1 6 12435453 10.1016/S0304-3959(02)00368-8
Anseloni VCZ He F Novikova SI Robbins MT Lidow IA Ennis M Lidow MS Alterations in stress-associated behaviors and neurochemical markers in adult rats after neonatal short-lasting local inflammatory insult Neuroscience 2005 131 635 645 15730869 10.1016/j.neuroscience.2004.11.039
Willis WD Coggeshall RE Sensory mechanisms of the spinal cord 2004 New York: Plenum
Tan PH Yang LC Shih HC Lan KC Cheng JT Gene knockdown with intrathecal siRNA of NMDA receptor NR2B subunit reduces formalin-induced nociception in the rat Gene Ther 2005 12 59 66 15470478 10.1038/sj.gt.3302376
Maeno H Yamada K Santo-Yamada Y Aoki K Sun YJ Sato E Fukushima T Ogura H Araki T Kamichi S Kimura I Yamano M Maeno-Hikichi Y Watase K Aoki S Kiyama H Wada E Wada K Comparison of mice deficient in the high- or low-affinity neurotensin receptors, Ntsr1 or Ntsr2, reveals a novel function for Ntsr2 in thermal nociception Brain Res 2004 998 122 129 14725975 10.1016/j.brainres.2003.11.039
Sakurada S Watanabe H Mizoguchi H Yonezawa A Orito T Katsuyama S Kuramasu A Sakurada C Yanai K Sakurada T Involvement of the histaminergic system in the nociceptin-induced pain-related behaviors in the mouse spinal cord Pain 2004 112 171 182 15494198 10.1016/j.pain.2004.08.018
Vachon P Masse R Gibbs BF Substance P and neurotensin are up-regulated in the lumbar spinal cord of animals with neuropathic pain Can J Vet Res 2004 68 86 92 15188951
Linde M Elam M Lundblad L Olausson H Dahlof CG Sumatriptan (5-HT1B/1D-agonist) causes a transient allodynia Cephalalgia 2004 24 1057 1066 15566420 10.1111/j.1468-2982.2004.00782.x
Dinarello CA The IL-1 family and inflammatory diseases Clin Exp Rheumatol 2002 S1 13 14989423
Conti P Kempuraj D Frydas S Kandere K Boucher W Letourneau R Madhappan B Sagimoto K Christodoulou S Theoharides TC IL-10 subfamily members: IL-19, IL-20, IL-22, IL-24 and IL-26 Immunol Lett 2003 88 171 174 12941475 10.1016/S0165-2478(03)00087-7
Basu A Krady JK Levison SW Interleukin-1: a master regulator of neuroinflammation J Neurosci Res 2004 78 151 156 15378607 10.1002/jnr.20266
McInnes IB Gracie JA Interleukin-15: a new cytokine target for the treatment of inflammatory diseases Curr Opin Pharmacol 2004 4 392 397 15251134 10.1016/j.coph.2004.04.003
Nishimoto N Kishimoto T Inhibition of IL-6 for the treatment of inflammatory diseases Curr Opin Pharmacol 2004 4 386 391 15251133 10.1016/j.coph.2004.03.005
Dubner R Ruda MA Activity-dependent neuronal plasticity following tissue injury and inflammation Trends Neurosci 1992 15 96 103 1373925 10.1016/0166-2236(92)90019-5
Abele R Lampinen M Keinanen K Madden DR Disulfide bonding and cysteine accessibility in the alpha-amino-3-hydroxy-5-methylisoxazole-4-propionic acid receptor subunit GluRD. Implications for redox modulation of glutamate receptors J Biol Chem 1998 273 25132 25138 9737972 10.1074/jbc.273.39.25132
Hatt H Modification of glutamate receptor channels: molecular mechanisms and functional consequences Naturwissenschaften 1999 86 177 186 10331215 10.1007/s001140050593
Ziff EB Recent excitement in the ionotropic glutamate receptor field Ann N Y Acad Sci 1999 868 465 473 10414323
Matsuda K Fletcher M Kamiya Y Yuzaki M Specific assembly with the NMDA receptor 3B subunit controls surface expression and calcium permeability of NMDA receptors J Neurosci 2003 23 10064 10073 14602821
Erreger K Dravid SM Banke TG Wyllie DJ Traynelis SF Subunit-specific gating controls rat NR1/NR2A and NR1/NR2B NMDA channel kinetics and synaptic signaling profiles J Physiol 2005
Huggins DJ Grant GH The function of the amino terminal domain in NMDA receptor modulation J Mol Graph Model 2005 23 381 388 15670959 10.1016/j.jmgm.2004.11.006
Rudolph U Mohler H Analysis of GABAA receptor function and dissection of the pharmacology of benzodiazepines and general anesthetics through mouse genetics Annu Rev Pharmacol Toxicol 2004 44 475 498 14744255 10.1146/annurev.pharmtox.44.101802.121429
Klein R Role of neurotrophins in mouse neuronal development FASEB J 1994 8 738 744 8050673
Malcangio M Garrett NE Cruwys S Tomlinson DR Nerve growth factor- and neurotrophin-3-induced changes in nociceptive threshold and the release of substance P from the rat isolated spinal cord J Neurosci 1997 17 8459 8467 9334418
Mendell LM Albers KM Davis BM Neurotrophins, nociceptors, and pain Microsc Res Tech 1999 45 252 261 10383118 10.1002/(SICI)1097-0029(19990515/01)45:4/5<252::AID-JEMT9>3.0.CO;2-N
Pezet S Onteniente B Grannec G Calvino B Chronic pain is associated with increased TrkA immunoreactivity in spinoreticular neurons J Neurosci 1999 19 5482 5492 10377357
Bennett DL Neurotrophic factors: important regulators of nociceptive function Neuroscientist 2001 7 13 17 11486340
Yajima Y Narita M Narita M Matsumoto N Suzuki T Involvement of a spinal brain-derived neurotrophic factor/full-length TrkB pathway in the development of nerve injury-induced thermal hyperalgesia in mice Brain Res 2002 958 338 346 12470870 10.1016/S0006-8993(02)03666-1
Logfen O Qi Y Lundeberg T Inhibitory effects of tachikinin receptor antagonists on thermally induced inflammatory reactions in a rat model Burns 1999 25 125 129 10208386 10.1016/S0305-4179(98)00125-9
Novikova SI He F Bai J Badan I Lidow IA Lidow MS Cocaine-induced changes in expression of apoptosis-related genes in the fetal mouse cerebral wall Neurotoxicol Teratol 2005 27 3 14 15681117 10.1016/j.ntt.2004.08.004
Draghichi S Data Analysis Tools for DNA Microarray 2003 Boca Raton: Chapman & Hall/CRC
Benjiamini Y Hochberg Y Controlling the false discovery rate; a practical powerful approach to multiple testing J Royal Stat Soc B 1995 57 289 300
Pierson SN Butler JN Foster RG Experimental validation of novel and conventional approaches to quantitative real-time PCR data analysis Nucleic Acid Res 2003 31 e73 12853650 10.1093/nar/gng073
Tichopad A Dilger M Schwartz G Pfaffl MW Standardized determination of real-time PCR efficiency from a single reaction set-up Nucleic Acids Res 2003 31 e122 14530455 10.1093/nar/gng122
|
16179088
|
PMC1242251
|
CC BY
|
2021-01-04 16:40:08
|
no
|
Mol Pain. 2005 Sep 22; 1:27
|
utf-8
|
Mol Pain
| 2,005 |
10.1186/1744-8069-1-27
|
oa_comm
|
==== Front
Nutr JNutrition Journal1475-2891BioMed Central London 1475-2891-4-251615015210.1186/1475-2891-4-25ResearchCactus pear: a natural product in cancer chemoprevention Zou Da-ming [email protected] Molly [email protected] Francisco [email protected] Jean M [email protected] Jian [email protected] Roungyu [email protected] Huaguang [email protected] Changping [email protected] Department of Obstetrics and Gynecology, Arizona Health Sciences Center, University of Arizona, Tucson, Arizona 85724, USA2 Division of Gynecologic Oncology, Arizona Cancer Center, Tucson, Arizona 85724, USA3 Department of Gynecologic Oncology, Fudan Univeristy, Shanghai, 200032, China4 Guangxi Medical University, Guangxi, 532021, China2005 8 9 2005 4 25 25 6 3 2005 8 9 2005 Copyright © 2005 Zou et al; licensee BioMed Central Ltd.2005Zou et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Cancer chemoprevention is a new approach in cancer prevention, in which chemical agents are used to prevent cancer in normal and/or high-risk populations. Although chemoprevention has shown promise in some epithelial cancers, currently available preventive agents are limited and the agents are costly, generally with side effects. Natural products, such as grape seed, green tea, and certain herbs have demonstrated anti-cancer effects. To find a natural product that can be used in chemoprevention of cancer, we tested Arizona cactus fruit solution, the aqueous extracts of cactus pear, for its anti-cancer effects in cultured cells and in an animal model.
Method
Aqueous extracts of cactus pear were used to treat immortalized ovarian and cervical epithelial cells, as well as ovarian, cervical, and bladder cancer cells. Aqueous extracts of cactus pear were used at six concentrations (0, 0.5, 1, 5, 10 or 25%) to treat cells for 1, 3, or 5 days. Growth inhibition, apoptosis induction, and cell cycle changes were analyzed in the cultured cells; the suppression of tumor growth in nude mice was evaluated and compared with the effect of a synthetic retinoid N-(4-hydroxyphernyl) retinamide (4-HPR), which is currently used as a chemoprevention agent. Immunohistochemistry staining of tissue samples from animal tumors was performed to examine the gene expression.
Results
Cells exposed to cactus pear extracts had a significant increase in apoptosis and growth inhibition in both immortalized epithelial cells and cancer cells in a dose- and time-dependent manner. It also affected cell cycle of cancer cells by increasing G1 and decreasing G2 and S phases. Both 4-HPR and cactus pear extracts significantly suppressed tumor growth in nude mice, increased annexin IV expression, and decreased VEGF expression.
Conclusion
Arizona cactus pear extracts effectively inhibited cell growth in several different immortalized and cancer cell cultures, suppressed tumor growth in nude mice, and modulated expression of tumor-related genes. These effects were comparable with those caused by a synthetic retinoid currently used in chemoprevention trials. The mechanism of the anti-cancer effects of cactus pear extracts needs to be further studied.
==== Body
Background
The goal of cancer prevention is to delay or block the processes of initiation and progression from pre-cancerous cells into cancer. Cancer chemoprevention, which targets normal and high risk populations, involves the use of drugs or other chemical agents to inhibit, delay, or reverse cancer development [1,2]. There has been significant success in the study of cancer prevention and chemoprevention in the last 20 years [1,2] and, as a result, the incidences of certain types of cancer have decreased due to prevention techniques and improved screening technology [1,2]. However, the incidence and mortality rates of ovarian cancer have remained essentially unchanged [3], partially because early detection methods (primary prevention) have not been developed and prevention of recurrence (secondary prevention) has not been achieved. Furthermore, only a limited number of potentially useful chemopreventive agent(s) have been tested [2,4-6]. Discovery and development of dietary agents for cancer prevention first began at the National Cancer Institute in 1987 [7]. Although hundreds of agents have been developed in the United States during the past decade, only a few new drugs have actually been approved [7,8]. The development of chemopreventive agents is slow and inefficient. More effective and less toxic agents, including natural products, are needed if we are to reach the goal of cancer prevention, both primary and secondary.
A synthetic retinoid, N-(4-hydroxyphernyl) retinamide (4-HPR), was found to decrease the risk of ovarian cancer in an Italian breast cancer chemoprevention trial [9-11]. Women receiving 4-HPR demonstrated a decreased incidence of ovarian cancer [9,10]; however, after cessation of the treatment, ovarian cancer did develop in the treatment group [10,11]. Other studies have also reported that the response of retinoids was not durable in pre-cancer and cancer treatments for either oral leukoplakia or cervical cancer [13-16]. These reports suggest that long-term administration of agents with lower toxicity will be the most important aspect in chemopreventive agents, especially for normal and high risk populations.
Medical benefits from plant forms have been recognized for centuries. Herbs have been used in Chinese medicine for thousands of years to cure diseases and heal wounds. Recently, it has been found that components in green tea and grape seeds have anticancer effects [17,18]. Also, as a rule, herbs and natural products lack much of the toxicity that is present in synthetic chemicals, thus, enhancing their appeal for long term preventive strategies.
Cactus (Opuntia) has been used for many years as a common vegetable and as medicine by the Native Americans and Mexicans [19-22]. Cactus contains a fruit known as cactus pear (Opuntia ficus-indica) and the plant is referred to as nopale (pad). Cactus pear contains pectin, carotenes, betalains, ascorbic acid, quercetina and quercetin derivatives all of which have antioxidant activity [21-24]. In Chinese medicine, cactus fruit is considered a weak poison and used as medicine for treatment of inflammation and pain [23,24]. It has also been used as a detoxification agent for snake bite [23,24].
In this study, we tested aqueous extracts of cactus pear for its anti-cancer effects in ovarian, cervix, and bladder cancer cells, and in the nude mice ovarian animal model. These results were compared to the effect of 4-HPR, demonstrated the anti-cancer effect of the cactus pear.
Methods
Cell lines
The immortalized ovarian epithelium cells (IOSE), the ovarian cancer cell lines OVCA420, SKOV3; the HPVE6 immortalized cervical epithelium cell line TCL-1; cervical cancer cell lines, HeLa and Me180; and bladder cancer cells UM-UC-6, T24, were all used in this study. Cells were grown in a 1:1 (v/v) mixture of Dulbecco's modified Eagle's medium (DMEM) and Ham's F12 with 10% fetal bovine serum at 37°C in a humidified atmosphere of 95% air and 5% CO2.
Animals
Athymic 4 to 6 weeks old nu/nu BALB/c female mice were purchased from the Animal Production Area at the National Cancer Institute, Frederick Cancer Research Facility (Frederick, MD). The mice were housed in laminar flow cabinets under pathogen-free conditions and maintained at the University of Arizona's Animal Care Facility in the College of Medicine, according to institutional regulations approved by the Animal Welfare Committee as well as current regulations and standards of the Department of Agriculture and the Department of Health and Human Services.
Cactus product
The cactus pear extract was purified from mature cactus fruit by blending. The cactus pear solution contained both the fruit of the cactus and the seeds, and were centrifuged at 4,000 RPM for 30 min and filtered using a 0.45 μM Nalgene filter (Rochester, NY), then aliquoted to 15 ml and stored at -20°C. We used pure extracts which were diluted in cell culture medium to achieve concentrations of 0, 0.5, 1, 5, 10 and 25% (v/v), before being used in cell culture. The osmolality of the solution was 358 m Osm/kg for 25% solution, 342 m Osm/kg for 10%, and 326 m Osm/kg for 5%. The pH was between 7.26–7.28. Animals were treated with pure cactus pear fruit intraperitoneally (i.p.) at 0.4 ml per day.
Effects of cactus products on cell proliferation in monolayer cultures
Cells were plated in 96-well plates at a concentration of 104 cells per well and grown for 24 hours. The cells were then incubated in cactus pear solution at different concentrations for 1, 3, or 5 days. Growth inhibition was determined using the crystal violet method, as described [25]. Briefly, after 5 days of treatment, cells were fixed by 5% glutartaldehyde in phosphate-buffered saline (PBS), rinsed with distilled water, and dried completely. Cells were incubated in a 1:1 (v/v) mixture of 200 mM 3-(cyclohexylamino)-1-propanesulfonic acid (CAPS; pH 9.5) and 0.2% crystal violet at 25°C for 30 min, and then were washed and dried. The fixed and stained cells were solubilized with 10% glacial acetic acid, and absorbance at A590 nm was determined using a plate reader. Growth inhibition was calculated according to the equation: inhibition = (1-Nt/Nc) × 100, where Nt and Nc are the numbers of cells in treated and control cultures, respectively.
All experiments were performed in triplicate and the mean ± standard deviations were calculated. IC50 were also determined at 50% of cell growth rate in each.
Cell cycle analysis by propidium iodide (PI) staining
Cells were treated with 0, 5 and 25% of cactus pear solution for 2 days, were collected by centrifugation and fixed in 4% paraformaldehyde pH 7.4 at room temperature for 30 min, and washed and incubated in 70% ethanol containing 1% HCl at -20°C for 10 minutes. Cells were then stained with 500 μl of propidium iodide/RNase A solution in the dark for 30 min at room temperature, analyzed by flow cytometry using a FACScan flow cytometer (BD Biosciences, San Jose, CA) with a 15 mW Argon laser used for excitation at 488 nm. Fluorescence was measured at 585 nm. Computer analysis was completed using BD Biosciences Cellquest Pro and ModFit LT by Verity Software data processing to provide information on the percentage of apoptotic cells as well as the proportion of cells in G1, S, and G2 phases of the cell cycle.
Analysis of apoptosis induced by cactus product by terminal deoxynucleotidyl transferase (TdT)-mediated fluorescein-deoxyuridine-triphosphate (dUTP) nick-end labeling (TUNEL) assay [25]
Following incubation with 0, 5, and 25% cactus pear solution for 2 days, cells were fixed in 1% formaldehyde in PBS (pH 7.4) for 15 min at 4°C. The cells were then washed twice with PBS, resuspended in 70% ice-cold ethanol and stored in a -20°C freezer until use. For the assay, cells were first suspended in 1 ml wash buffer containing cacodylic acid, Tris-HCl buffered solution and sodium azide (Phoenix flow cytometry kit, Phoenix Flow Systems, San Diego, CA). Approximately 106 cells were resuspended in 50 μl staining buffer containing Tris-HCl buffer, TdT, and fluorescein-12-dUTP (Phoenix flow cytometry kit). Cells were incubated at 37°C for 60 min, and then rinsed twice with PBS. Cells were stained with 500 μl of propidium iodide/RNase A solution in the dark for 30 min at room temperature and then analyzed by flow cytometry using a FACScan flow cytometer (Epics Profile, Coulter Corp., Hialeah, FL) with a 15 mW argon laser used for excitation at 488 nm. Fluorescence was measured at excitation 520 nm and 570 nm. The Phoenix flow cytometry kit included suspensions of cells that served as negative and positive controls for apoptosis. Computer analysis of the data provided information on the percentage of apoptotic cells as well as the proportion of cells in the hypodiploid, G1, S, and G2 phases of the cell cycle.
Human tumor xenografts
Ovarian cancer cells SKOV3 were grown to sub-confluence and harvested using 0.1% trypsin and 1 mM EDTA. The cells were washed with serum containing medium to quench the trypsin and then with serum-free medium. Cell viability was determined by Trypan blue exclusion and only cultures with more than 90% viability were used for the in vivo experiments. The cells were resuspended in medium at 5 × 106 cells. Cactus pear solution, as well as the chemopreventive agent 4-HPR (0.43 mg i.p twice/week, which equivalent to 200 mg/kg human dose) were injected one day prior to tumor cell injection (day 1) (Fig. 1). Control animals received H2O. Tumor cells were injected subcutaneously (day 2). The tumors appeared on day 10–14, and their size was measured twice a week using a caliper. The larger (A) and smaller (B) diameters were used to calculate the tumor volume (V) by using the equation V = 0.4 × A × B2 [27]. The treatment regimen of cactus pear solution was as follows: 0.4 ml of solution injected i.p. twice a week for the first two weeks, then five times a week from the third week to the sixth week (Fig. 1).
Figure 1 Cactus pear extracts treatment schedule in animal. Numbers of injection times/week were represented by arrow bars. Four groups of animal were examined in the study: SKOV3 alone, SKOV3 + H2O, SKOV3 + Cactus extracts, and SKOV3 + 4HPR. 4-HPR concentration was used at 0.43 mg/kg, which equivalent to human 200 mg/kg.
Immunostaining
Paraffin-embedded sections were deparaffinized in xylene, rehydrated through graded alcohols to water, then incubated for 10 min in PBS. The sections were blocked for 30 min with 3% normal horse serum (NHS) diluted in PBS; the sections were then blotted and incubated with p53, annexin IV and VEGF antibodies (Santa Cruz Biotech, Santa Cruz, CA, and Zymed Lab Inc, San Francisco, CA) for 1 hr at room temperature. The endogenous peroxidase was inactivated by incubation for 30 min in 0.015% peroxide in methanol and rehydrated for 10 min in PBS. The slides were incubated with biotinylated horse antibody for 1 hr and washed in PBS, followed by the avidin-biotin-peroxidase complex (ABC, Vector Laboratories, Burlingame, CA). The slides were washed and the peroxidase reaction developed with diaminobenzidine and peroxide, then counterstained with hemotoxylin, mounted in aqua-mount, and evaluated on a light microscopy. Positive and negative antibodies and bladder and ovarian cancer cells were used as controls in each assay.
Statistical analysis
Student's t test was performed to compare two means. One-way ANOVA, followed by the Fisher's Least Square Difference (LSD) test, was used to analyze tumor size in different treatment groups or multiple means. Two-sided P values were determined in all analyses. P < 0.05 is considered as statistically significant.
Results
Growth inhibitory effect of cactus pear solution on human ovarian cell lines
Cactus pear extracts were used at different concentrations (see Methods) to compare the inhibitory effect on a growth of 3 different types of human cancer cells in monolayer cultures. The sensitivity of cancer cells to cactus treatment differed among cell types. Cervical cancer cells were the most sensitive compared with ovarian and bladder cancer cells (Fig. 2a,b,c). One percent (1%) cactus pear solution inhibited 40–60% of immortalized cervical epithelium cells and cervical cancer cells (Fig. 2a). For ovarian cancer cells, 5% cactus pear solution was effective on growth inhibition in IOSE and OVCA420 cells, however, 10% solution was required to inhibit growth in SKOV3 cells (Fig. 2b). The concentration of cactus pear extracts effect on 50% of bladder cancer cell growth was greater than 1% (Fig. 2c). The effect of the cactus pear solution was dose-and time-dependent (Fig. 2). The IC50 (the concentration causing 50% cell death) in cervical and bladder cancer cells after 5-day treatment with cactus pear solution was less than 2 percent. For cervical cells, the IC50 for TCL-1 was 1.5%; HeLa was 1.8%; and ME180 was 0.8%. For bladder cancer cells, IC50 was 0.9% and 1.3% for UM-UC-6 and T24 cells, respectively. However, the IC50 for ovarian cells was varied, IC50 for IOSE, OVCA420, and SKOV3 cells were 2%, 0.8%, and 8%, respectively. Morphological changes were induced by cactus pear extracts 3 days after treatment and were in concordance with the agent's effect on cell growth of cervical cells (Fig. 3), ovarian cells (Fig. 4), and bladder cancer cells (Fig. 5).
Figure 2 Effect of cactus pear extracts on growth of human cervical, ovarian, and bladder cancer cells in monolayer cultures. Cells were grown for 1, 3, or 5 days in the absence (control) or presence of 0.5, 1, 5, 10, or 25% of cactus pear extracts in (a.) immortalized cervical cells and cervical cancer cells; (b.) immortalized ovarian cells and ovarian cancer cells; and (c.) bladder cancer cells. Values are means ± SD of triplicate cultures. The percentage of growth inhibition (GI) was calculated using the equation: % GI = (1-Nt/Nc) × 100; where Nt and Nc represent the numbers of cells in treated and control cultures, respectively.
Figure 3 Effect of cactus pear extracts on the morphology of cervical cells. Immortalized cervical cells and cervical cancer cells were grown in the absence (control) or presence of different concentrations of cactus extracts. The photographs were taken on day 3 after the removal of medium containing floating cells.
Figure 4 Effect of cactus pear extracts on the morphology of ovarian cells. Immortalized ovarian cells and ovarian cancer cells were grown in the absence (control) or presence of different concentrations of cactus extracts. The photographs were taken on day 3 after the removal of medium containing floating cells.
Figure 5 Effect of cactus pear extracts on the morphology of bladder cancer cells. Bladder cancer cells were grown in the absence (control) or presence of different concentrations of cactus extracts. The photographs were taken on day 3 after the removal of medium containing floating cells.
Apoptosis induction by cactus extract in different cancer cells
Cactus pear solution induced apoptosis in all three cancer cell lines tested by TUNEL analysis (Fig. 6 and 7). In cancer cell lines, the strongest effect of apoptosis induction was found in cervical cells. The apoptosis cell population increased by more than 50% at the concentration of 25% cactus extract compared with the untreated cells (Fig. 6). This was consistent with cell growth inhibitory effects (Fig. 2 and 6). The immortalized cervical epithelium cells were the most sensitive in which the apoptotic cells increased over 70% after treatment (Fig. 6). Apoptosis induction in ovarian and bladder cancer cells differed: in ovarian cancer cells, cactus extracts increased apoptosis induction from 40% to 50% in OVCA420 and SKOV3 cells (Fig. 7, left and mid-panel). In T24 bladder cancer cells, apoptosis was 30% (Fig. 7, right panel). Apoptosis induction was not significant at 5% concentration.
Figure 6 Apoptosis induction analyzed by TUNEL assay in cervical cells. Cells were treated with 5% and 25% cactus pear solution for 2 days. Cells were harvested and incubated with TdT in the presence of biotin-labeled BrdU and analyzed by flow cytometry. The percentage of apoptotic cells is represented by dark dots (fluorescence of individual cells) above the line in R3 region (R3 is the computer software analysis apoptosis program).
Figure 7 Apoptosis induction analyzed by TUNEL assay in ovarian and bladder cancer cells. TUNEL analysis results showed apoptosis induction by cactus extract in ovarian cancer cells (left and mid-panel) and bladder cancer cells (right panel).
Cell cycle and apoptosis analysis in cancer cells
DNA content and cell cycle analysis were performed after treatment with 0, 5, and 25% concentrations of cactus pear solution. Results demonstrated that cactus pear extracts affected cell cycle in cancer cells starting at a 5% concentration (Fig. 8a, and 8b). In cervical cancer cells, cactus extracts increased cells in G1 and decreased those in the S phase (Fig. 8a). Treatment with higher concentrations of cactus pear extracts increased cells in G1 and decreased cells in G2 and in the S phase in ovarian and bladder cancer cells (Fig. 8b). The effect of cactus on cell cycle was dose-dependent.
Figure 8 Cell cycle analysis. Cells were treated with 5 and 25% cactus pear extract for 2 days. Cells were stained with propidium iodide/RNase A solution for 30 min then analyzed by flow cytometry using a FACScan flow cytometer. (a.) cervical cancer cells HeLa and Me180; (b.) ovarian cancer cells SKOV3 and OVCA420 (upper), and bladder cancer cells T24 (bottom).
Cactus products inhibited tumor growth in a nude mice model
The treatment groups and the schedule of treatment are shown in Fig. 1. Animal body weight was measured twice a week for weight loss, as an indication of toxicity. Cactus pear extracts had no significant effect on weight loss (Fig. 9a) or animal behavior.
Figure 9 a. Animal body weight curve. The body weight was measured twice a week during the experiment. The picture represents the control animal labeled as H2O and treated-animal as SKOV3 only (SKOV3), SKOV3 + cactus pear (pear sol), and SKOV3 + 4HPR (4-HPR). b. Tumor growth curve. Tumor size in cactus pear and 4-HPR treatment groups, compared with control SKOV3 only and SKOV3 plus H2O, was significantly reduced (p < 0.05). The effect of cactus pear solution compared with 4-HPR on inhibiting tumor growth, both agents were able to inhibit SKOV3 inoculated tumor growth, the difference is not statistically significant (p > 0.05).
The cactus pear solution was able to inhibit tumor growth in nude mice compared with that in untreated animals or animals treated with H2O (Fig. 9b). The effect of cactus pear solution on inhibiting tumor growth indicated by tumor size was compared with 4-HPR, which is currently being used as a chemopreventive agent in ovarian, cervical and bladder cancer clinical trials [11-17] (Fig. 9b). We compared the control animal transplanted with SKOV3 cells only and SKOV3 + H2O to treatment group with either cactus pear extracts or 4-HPR. Cactus pear extracts and 4-HPR significantly reduced tumor size (p < 0.05). The inhibitory effect of 4-HPR was not significantly different from that of the cactus pear extract solution (p > 0.05).
Immunohistochemistry staining for p53, annexin IV and VEGF expression
The expression of p53, annexin IV, and VEGF were examined in animal tumor tissues. 4-HPR and cactus extracts treatment increased annexin IV and decreased VEGF expression; also cactus extracts had a stronger effect on suppression of VEGF expression (Fig. 10). Both 4-HPR and cactus extracts slightly changed p53 expression, where more negative nuclei were observed (Fig. 10).
Figure 10 Representative immunohistochemistry patterns of p53, annexin IV and VEGF in animal tumor sections. p53 expression was stained as positive (+) in SKOV3 only and SKOV3 plus H2O groups, treatment of 4-HPR was slightly changes its expression and most of nuclei were stained negative (upper panel). Cactus extract treatment was found in some of nuclei stained negative (weak). Annexin IV expression was detected negatively (-) in SKOV3 only and SKOV3 plus H2O groups, treatment of both 4-HPR and cactus extracts were increased its expression (mid panel). VEGF expression was detected positively (+) in SKOV3 only and SKOV3 plus H2O groups, treatment of both 4-HPR and cactus extracts were decreased its expression (bottom panel).
Discussion
Remarkable progress has been made over the past two decades in understanding the molecular and cellular mechanisms of pre-cancer and cancer progression [2]. Nonetheless, the development of effective and safe agents for prevention and treatment of cancer remains slow, inefficient, and costly [7], with little to offer the high-risk population for primary prevention and cancer survivors to prevent cancer recurrence. The key to effective chemoprevention is the identification of a chemopreventive agent(s) that can effectively inhibit cancer development without toxic side effects. In an Italian 4-HPR trial, retinoids showed the preventive effect on ovarian cancer only during the period while the drug was taken. After cessation of treatment, the incidence of ovarian cancer increased to the level that was observed in the untreated control group [10,11]. Therefore, chemopreventive agents may need to be used for a long period of time to be effective. As a result, identification of agents with little or no toxicity becomes important. We have shown that cactus pear extracts, a natural product, has anti-cancer activity, although the active component(s) have not been clearly identified. Since it has no toxic effects, cactus pear extracts can be easily used, for example, as dietary supplements [19-21] in normal and high risk populations.
It has been noted that Native Americans have a lower cancer rate when compared to white and African Americans [3]. Both cactus pear and nopale which contain multiple antioxidants, have been used as a dietary supplement for centuries by Native Americans. Our results show that the cactus pear inhibited growth of different cancer cells in vitro and in vivo. Cactus products inhibited cancer cell growth with concentrations as low as 5%; cell cycle was also affected at this concentration with an increase in G1 phase (Fig 2 and 8). However, apoptosis was observed at a higher concentration of 10% (data not shown) and 25% (Fig. 6 and 7).
We also compared cactus with the chemopreventive agent 4-HPR in nude mice. Both cactus and 4-HPR inhibited ovarian cancer growth. The anti-carcinogenic properties of natural and synthetic retinoids have been suggested to be due, in part, to the antioxidant effect [28-30], increased consumption of fruit and vegetables is associated with prevention of various human diseases, and the oxidative damage is an important etiologic risk factor for many diseases, including cancer and heart disease. Cactus pear extracts also contain multiple antioxidants that can reduce oxidative damage. The clinical trial on vitamin C and cactus pear demonstrated that supplements of vitamin C at a comparable dosage enhances overall antioxidant defense but does not significantly affect body oxidative stress [21,22]. Components of cactus pear extract, other than antioxidant vitamins, may play a role in anti-oxidant effects [21,22,31-33].
Carcinogenesis may be viewed as a process of progressive disorganization. This process is characterized by the accumulation of genotypic changes and corresponding tissue and cellular abnormalities including loss of proliferation and apoptosis controls. A dietary agent that can increase anti-proliferation pathways and change cell cycle in cancer cells without toxicity would be a potential agent for chemoprevention. Although the mechanism for cactus pear extract in cancer prevention is unclear, our current study shows that cactus pear does alter the expression of certain genes related to cell growth and apoptosis. Cactus pear extracts increased annexin IV and decreased VEGF expression in animal tumors. Annexin IV, a Ca2+-dependent membrane-binding protein, is expressed in many epithelial cancers [34]. Annexin IV played a pivotal role in the early phases of apoptosis [35], it was identified in initiation of apoptosis in human preneoplastic colonocytes [35], and its expression was regulated by quercetin [35]. Quercetin is one of the components of cactus pear extracts. Our results (unpublished data) and other reports [35,36] suggest quercetin might be one of the active compounds responsible for the anti-carcinogenetic and apoptosis-induction effects of cactus pear extracts. In our study, cactus pear extracts decreased VEGF expression, suggesting that cactus pear extracts might have inhibitory effects on angiogenesis, an important factor contributing to tumor growth and metastasis. We did not observe a significant effect on p53 expression caused either by 4-HPR or cactus pear extracts. Mutation of p53 is expected with the SKOV3 cell line, the tumor cells used in this animal model [37,38] but in this study, we observed minimal effect on p53 expression after treated with cactus extract and 4-HPR. However, since both wild-type and mutant p53 could contribute to induction of apoptosis, involvement of p53 pathway by 4-HPR or cactus pear extract cannot be ruled out by these results.
For developing food-derived agents, the NCI has advocated co-development of a single or purified extract of a few putative active compounds that are contained in food-derived agents [7]. The cactus pear extracts tested in this study could be such a candidate in cancer prevention for both normal and high-risk populations and prevention of recurrence in patients with previous cancers. This product holds promise for long-term use because of the safety of food-derived products and the fact that they are not perceived as a "chemical".
Conclusion
Arizona prickly pear cactus effectively inhibited cell growth in several different immortalized and cancer cell cultures in vitro and suppressed tumor growth in a nude mouse of ovarian cancer model. The mechanism of anti-cancer effect of cactus pear extracts is not yet completely understood. Currently, we are investigating the expression of genes related to cell growth and apoptosis which may be altered by treatment with cactus products to elucidate possible pathways through which this natural product exerts its anti-cancer effects.
Acknowledgements
This project was supported in part by a grant from the National Institutes of Health, NCI-CA75966 and by the Ovarian Cancer Research Fund. We wish to thank William Brands, Nathaniel Kirkpatrick, and J. Dominique Jennings, for their help and assistance with the animal work, and Dr. Sui Zhang and Carole Meyer for their careful editing of the manuscript.
==== Refs
Kelloff GJ Sigman CC Greenwald P Cancer chemoprevention: progress and promise Eur J Cancer 1999 35 2031 2038 10711244 10.1016/S0959-8049(99)00299-3
Kelloff GJ Crowell JA Steel V Lubet RA Boone CW Malone WA Hawk ET Lieberma R Lawrence JA Sigman CC Progress in Cancer Chemoprevention Ann NY Acad Sci 1999 889 1 13 10668477
Cancer statistics 2004 2004 American Cancer Society
Kelloff GJ Sigman CC Hawk ET Johnson KM Crowell JA Guyton KZ Surrogate end-point biomarkers in chemopreventive drug development IARC Sci Publ 2001 154 13 26 11220652
Woude GF Kelloff GJ Ruddon RW Koo HM Sigman CC Barrett JC Day RW Dicker AP Kerbel RS Parkinson DR Slichenmyer WJ Reanalysis of Cancer Drugs: Old Drugs, new Tricks Clin Cancer Res 2004 10 3897 3907 15173099
Kelloff GJ Bast RC JrCoffey DS D'Amico AV Kerbel RS Park JW Ruddon RW Rustin GJ Schilsky RL Sigman CC Woude GF Biomarkers, surrogate end points, and the acceleration of drug development for cancer prevention and treatment: an update prologue Clin Cancer Res 2004 10 3881 3884 15173097
Kelloff GJ Crowell JA Steele VE Lubet RA Malone WA Boone CW Kopelovich L Hawk ET Lieberman R Lawrence JA Ali I Viner JL Sigman CC Progress in Cancer Chemoprevention: Development of Diet-Derived Chemopreventive Agents J Nutr 2000 130 467S 471S 10721931
Steele VE Sharma S Mehta R Elmore E Redpath L Rudd C Bagheri D Sigman CC Kelloff GJ Use of in vitro assays to predict the efficacy of chemopreventive agents in whole animals J Cell Biochem 1996 26 29 53 10.1002/jcb.240630704
Supino R Crosti M Clerici M Warlters A Cleris L Zunino F Induction of apoptosis by fenretinide (4-HPR) in human ovarian carcinoma cells and its association with retinoic acid receptor expression Intl J Cancer 1996 65 491 497 10.1002/(SICI)1097-0215(19960208)65:4<491::AID-IJC17>3.0.CO;2-D
Veronesi U De Palo G Marubini E Costa A Formelli F Mariani L Decensi A Camerini T Del Turco MR Di Mauro MG Muraca MG Del Vecchio M Pinto C D'Aiuto G Boni C Campa T Magni A Miceli R Perloff M Malone WF Sporn MB Randomized trial of fenretinide to prevent second breast malignancy in women with early breast cancer J Natl Cancer Inst 1999 91 1847 1856 10547391 10.1093/jnci/91.21.1847
De Palo G Mariani L Camerini T Marubini E Formelli F Pasin B Decensi A Veronesi U Effect of fenretinide on ovarian carcinoma occurrence Gynecol Oncol 2002 86 24 27 12079295 10.1006/gyno.2002.6663
Zhang D Holmes W Wu S Soprano D Soprano K Retinoids and ovarian cancer J Cell Physiol 2000 185 1 20 10942515 10.1002/1097-4652(200010)185:1<1::AID-JCP1>3.0.CO;2-O
Lippman SM Kavanagh JJ Paredes-Espinoza M Delgadillo-Madrueno F Paredes-Castillas P Hong WK Holdener E Krakoff IH 13-cis-retinoic acid plus intert~ron-ct2a: Highly active systemic therapy for squamous cell carcinoma of the cervix J Natl Cancer Inst 1992 84 241 245 1734085
Lippman SM Kavanagh JJ Paredez-Espinosa M Delgadillo-Madrueno F Paredes-Castillas P Hong WK Massimini G Holdener EE Krakoff IH 13-cis-retinoic acid plus interferon-a2a in locally advanced squamous cell carcinoma of the cervix J Natl Cancer Inst 1993 85 499 500 8445678
Studer UE Biedermann C Prevention of recurrent superficial bladder tumors by oral etretinate:preliminary results of a randomized double blind multicenter trial in Switerland J Urol 1984 131 47 51 6361284
Hong WK Endicott J Itri LM 13-cis retinoic acid in the treatment of oral leukoplakia N Engl J Med 1986 315 1501 1505 3537787
Kim H Hall P Smith M Kirk M Prasain JK Barnes S Grubbs C Chemoprevention by grape seed extract and genistein in carcinogen-induced mammary cancer in rats is diet dependent. Supplement: International Research Conference on Food, Nutrition and Cancer J Nutr 2004 134 3445S 3452S 15570052
Xu Y Ho CT Amin SG Han C Chung FL Inhibition of tobacco-specific nitrosamine-induced lung tumorigenesis in A/J mice by green tea and its major polyphenol as antioxidants Cancer Res 1992 52 3875 3879 1617663
Cornett J How Indians used desert plants 2000 Nature Trails Press
Kay MA Healing with plants in the American and Mexican west 1996 The University of Arizona Press
Knishinsky R Prickly pear cactus medicine 1971 Healing Arts Press, Rochester, Vermont
Tesoriere L Butera D Pintaudi M Allegra M Livrea MA Supplementation with cactus pear (Opuntia ficus-indica) fruit decreases oxidative stress in healthy humans: a comparative study with Vit C Am J Clin Nutr 2004 80 391 395 15277160
Wang PZ Chinese Medicine Surgery 1988 Ancient Chinese Medicine Press 164 183
Nanjing Pharmcologic College Zhong Cao Yao Xue 1976 Jiangsu People Press 681 682
Zou CP Kurie JM Lotan D Zou CC Hong WK Lotan R N-(4-Hydroxyphenyl) Retinamide (4HPR) induces apoptosis in human non-small cell lung cancer (NSCLC) cell lines Clin Cancer Res 1988 4 1345 1355 9607596
Kirkpatrick ND Zou CP Brewer MA Brands WR Drezek RA Utzinger U Endogenous fluorescence spectroscopy of cell suspensions for chemopreventive drug monitoring Photochemistry and Photobiology 2005 81 125 134 15535738 10.1562/2004-08-09-RA-267.1
Zang RY Shi DR Lu HJ Adenovirus 5 mediated E1a gene therapy for ovarian cancer in vitro and in vivo Int J Gynecol Cancer 2001 11 18 23 11285029 10.1046/j.1525-1438.2001.011001018.x
De Palo G Veronesi U Camerini T Formelli F Mascotti G Boni C Fosser V Del Vecchio M Campa T Costa A Can fenretinamide protect women against ovarian cancer? J Natl Cancer Inst 1995 87 146 147 7707390
Goto H Takahashi H Fujii H Ikuta K Yokota S N-(4-Hydroxyphenyl) retinamide (4-HPR) induces leukemia cell death via generation of reactive oxygen species Int J Hematol 2003 78 219 225 14604280
Hail N JrLotan R Mitochondrial respiration is uniquely associated with the prooxidant and apoptotic effects of N-(4-hydroxyphenyl) retinamide J Biol Chem 2001 276 45614 45621 11546781 10.1074/jbc.M106559200
Hail N JrLotan R Mitochondrial permeability transition is a central coordinating event in N-(4-hydroxyphenyl)retinamide-induced apoptosis Cancer Epidemiol Biomarkers Prev 2000 9 1293 1301 11142414
Tesoriere L Allegra M Butera D Livrea MA Absorption, excretion, and distribution of dietary antioxidant betalains in LDLs: potential health effects of betalains in humans Am J Clin Nutr 2004 80 941 945 15447903
Collazo-Siques P Valverde ME Paredes-Lopez O Guevara-Lara F Expression of ripening-related genes in prickly pear (Opuntia sp.) fruits Plant Foods Hum Nutr 2003 58 317 326 15354791 10.1023/B:QUAL.0000040286.28359.76
Li B Dedman JR Kaetzel MA Intron disruption of the annexin IV gene reveals novel transcripts J Biol Chem 2003 278 43276 4383 12912993 10.1074/jbc.M306361200
Herzog A Kuntz S Daniel H Wenzel U Identification of biomarkers for the initiation of apoptosis in human preneoplastic colonocytes by proteome analysis Int J Cancer 2004 109 220 229 14750173 10.1002/ijc.11692
Hansen RK Oesterreich S Lemieux P Sarge DK Fuqua SAW Quercetin inhibits heat shock protein induction but not heat shock factor DNA-binding in human breast carcinoma cells Biochem Biophy Res Comm 1997 239 851 856 10.1006/bbrc.1997.7572
Shin KS Sullenger BA Lee SW Ribozyme-mediated induction of apoptosis in human cancer cells by targeted repair of mutant p53 RNA Mol Ther 2004 10 365 372 15294183 10.1016/j.ymthe.2004.05.007
Hwang ES Kim J Kim JS Kao C Ko SC Chung L Lee JH The effects of the adenovirus-mediated wild-type p53 delivery in human epithelial ovarian cancer cell line in vitro and in vivo Int J Gynecol Cancer 1998 8 27 36 11576284 10.1046/j.1525-1438.1998.09772.x
|
16150152
|
PMC1242252
|
CC BY
|
2021-01-04 16:39:31
|
no
|
Nutr J. 2005 Sep 8; 4:25
|
utf-8
|
Nutr J
| 2,005 |
10.1186/1475-2891-4-25
|
oa_comm
|
==== Front
Nutr Metab (Lond)Nutrition & Metabolism1743-7075BioMed Central London 1743-7075-2-231615690310.1186/1743-7075-2-23ResearchComparison of the effects of three different (-)-hydroxycitric acid preparations on food intake in rats Louter-van de Haar Johanna [email protected] Peter Y [email protected] Anton JW [email protected] Arie G [email protected] Department of BioMedical Research, Numico Research, PO Box 7005, 6700 CA Wageningen, the Netherlands2 Department of Neuroendocrinology, University of Groningen, PO Box 14, 9740 AA Haren, the Netherlands2005 13 9 2005 2 23 23 9 3 2005 13 9 2005 Copyright © 2005 de Haar et al; licensee BioMed Central Ltd.2005de Haar et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Studies on the effects of (-)-hydroxycitric acid (HCA) in humans are controversial. As differences in the HCA preparations may contribute to this apparent discrepancy, the aim of the current study is to compare different HCA-containing preparations in adult Wistar rats.
Design
The effects of 3 different HCA-containing preparations (Regulator, Citrin K, Super CitriMax HCA-600-SXS, all used at an effective HCA dose of 150 and 300 mg/kg, administered intragastrically) on food intake and body weight were studied in adult male Wistar rats. The efficacy was tested under 2 different experimental conditions: 1) after a single dose administration and 2) during repeated administration for 4 subsequent days.
Results
Regulator and Citrin K significantly reduced food intake in both experimental setups, while Super CitriMax HCA-600-SXS was less effective. When administered for 4 subsequent days Regulator and Citrin K diminished body weight gain.
Conclusion
Regulator and Citrin K were shown to be potent inhibitors of food intake in rats, whereas Super CitriMax HCA-600-SXS showed only small and more inconsistent effects. The striking differences in efficacy between these 3 preparations indicate that low doses of a relatively low-effective HCA preparation may have contributed to the lack of efficacy as found in several human studies.
==== Body
Background
(-)-Hydroxycitric acid (HCA) is widely used as an ingredient for nutritional supplements aimed at reduction of food intake, appetite and body weight. However, studies on the effects of HCA in humans are controversial. Four placebo-controlled studies support the efficacy of HCA in man. In these studies, HCA administration led to increased loss of body weight and appetite reduction [1,2], decreased energy intake [3] and increased fat oxidation [4]. In addition, one placebo-controlled study reported increased loss of body weight after combined treatment of HCA and chromium [5]. In contrast, several other studies have not confirmed these proposed effects of HCA on gain of body weight [6-10], energy intake [8-10] or substrate utilization [11,12] in man.
Several factors may contribute to these inconclusive results of the human studies on the efficacy of HCA. First, the doses used in the human studies are highly variable, typically ranging from 5 – 40 mg/kg HCA per day whereas in one trial a dose as high as 250 mg/kg was used [12]. Second, differences in HCA preparations or production processes may also contribute to above-mentioned inconsistency in the results. For instance, HCA may occur either in open chain or in a lactone form. Since the lactone form has shown to be a very less effective inhibitor of the citrate cleavage enzyme [13], different preparations attempt to prevent cyclization of HCA into its (ineffective) lactone by using different counter-ions (such as sodium, calcium or potassium).
To obtain some insight into the difference in efficacy of commercially available HCA preparations, we studied the effects of three different HCA preparations on voluntary food intake and body weight in conscious rats. The product names of these preparations were: Regulator, Citrin K and Super CitriMax HCA-600-SXS (abbreviated as CitriMax), respectively.
Methods
Animals and housing
All experimental protocols were approved by the Animal Experiments Ethical Committee DEC-Consult, Bilthoven, the Netherlands. Male Wistar rats (HsdCpb:WU, Harlan, the Netherlands) aged 3 months and weighing 290–320 gram at arrival were used. The rats were kept at 20 ± 1°C, with lights on from 23.00 h (ZT 0.00) until 11.00 h (ZT 12.00), and with water and RMH-B standard lab chow, containing (w/w) 24% protein, 52% carbohydrates and 6% fat (Hope Farms, Woerden, the Netherlands) ad libitum unless mentioned otherwise.
The rats received a permanent silicone cannula (I.D. 0.6 mm, O.D. 1.2 mm) in the stomach under Isoflurane/oxygen/nitrogen oxide anesthesia according to the method described by Strubbe et al. [14]. This was done to allow stress-free intragastric (ig) administration of components to freely moving rats. The animals were allowed to recover for at least one week after surgery.
(-)Hydroxycitric acid preparations
The following preparations were used: (1) Regulator, a synthetic produced product, which contains 97% of a tri-potassium salt of HCA (HOB Ireland Limited, Dublin, Ireland), (2) Citrin K, an extract of Garcinia cambogia, which contains 50% HCA (Sabinsa Corporation, New Jersey, USA), with potassium as its primary mineral (28 g/100 g) and (3) Super CitriMax HCA-600-SXS (abbreviated as CitriMax), an extract of Garcinia cambogia, which contains 60% HCA (Interhealth Nutraceuticals Incorporated, Concord, California), containing K+ (15 g/100 g) and Ca2+ (11 g/100 g). To test whether the effects are specific to HCA, its structural analogue (4) tri-potassium citrate (Merck Eurolab B.V., Darmstadt, Germany) was used for comparison.
At a concentration of 75 mg HCA/ml demineralized water, the osmotic values of all preparations were 0.545 mOsm/l for Regulator, 0.507 mOsm/l for Citrin K, 0.265 mOsm/l for CitriMax and 0.490 mOsm/l for an equimolar solution of tri-potassium citrate in demineralized water.
Experimental design
Two types of experiments were performed to study the potential differences in efficacy between the different HCA preparations. The first series of experiments focused on the effect of one single administration of each preparation on food and water intake for the following 46 hrs in a 4 days placebo-controlled crossover experiment. In the second series of experiments, HCA was given twice a day for 4 days to study effects of repeated doses of the component on food and water intake. In both studies, rats were housed individually in cages in which food and water intake was monitored online (UgoBasile, Comerio, Italy).
Single administration
4 groups of 6 animals were used; within each group 2 single experiments were done. Each single experiment lasted for 4 days (96 hours). On day 1, food was removed 2 hours before dark onset (ZT 10.00). At ZT 11.30 randomly assigned rats received either a single bolus of the test component (dissolved in a total volume of 1 ml water) or a single bolus of isovolumic amount of water alone through the gastric cannula. At ZT 12.00 food was returned, and food and drink intake were monitored at 1, 2, 3, 4, 5, 6, 12, 24 and 46 hrs after administration. This period of 46 hrs was a washout period at the same time. On day 3 the protocol of day 1 was repeated. The rats that received the component on day 1 now receive vehicle and vice versa. Body weight was registered daily. In this way each dose of each HCA preparation was tested with regard to it's own vehicle treatment in a crossover experiment.
Each HCA preparation was tested, in a group of 6 animals, at two doses corresponding to 150 and 300 mg HCA/kg body weight. Thus, three groups of 6 animals were used to test Regulator at doses of 155 and 310 mg/kg, Citrin K at 300 and 600 mg/kg and CitriMax at 250 and 500 mg/kg, respectively. In the fourth group of 6 animals tri-potassium citrate, first, was tested at a dose of 475 mg per kg body weight (corresponding to the molarity of 300 mg/kg HCA) in comparison with vehicle treatment. Second, tri-potassium citrate (475 mg/kg) was tested in comparison with Regulator (310 mg/kg).
Repeated administration
To study the effects of long-term administration of the different HCA preparations, HCA or vehicle was administered intragastric twice daily for 4 subsequent days (at ZT 11.30 and ZT 17.00). Body weight and food intake were monitored during the 4 days of ig HCA-administration and the 3 days thereafter at ZT 10.00.
For logistic reasons, the long term studies were performed in two subsequent series of experiments. In the first set of experiments (n = 15, for each treatment n = 5) vehicle, Regulator and CitriMax and in the second set (n = 27, for each treatment n = 9) vehicle, Regulator and Citrin K were tested. All tested doses corresponded to the dose of 300 mg HCA/kg body weight.
Data analysis
In the single administration experiments, food intake were analyzed for the crossover experiments by a four-way ANOVA test with the factors, group of treatment order, animals within group, period and type of vehicle treatment, all per HCA preparation. First, the difference between time period and beginning was analyzed to test the possibility of a carryover effect. Because of the absence of a carryover effect, the type of vehicle treatment effects can be tested in the crossover experiment.
To compare the food intake results of the different HCA preparations in the single administration experiments, the delta (= difference) between treatment and vehicle on t = 24 hrs was calculated. To make comparisons between the HCA preparations, these deltas were analyzed using a two independent sample t-test with equal variances.
In the single administration experiments, bodyweight and water intake were also analyzed for the crossover experiments by a four-way ANOVA test, per HCA preparation.
In the repeated-administration experiments, the effects of HCA treatment on food intake, water intake and body weight were analyzed by a one-way ANOVA test. P values less than 0.05 were regarded significant. Data are presented as mean ± SEM.
Results
Single administration
Cumulative food intake was significantly reduced after intragastric administration of 310 mg/kg Regulator (figure 1) and 600 mg/kg Citrin K (figure 2). CitriMax (500 mg/kg) tended to decrease food intake (figure 3), reaching significance at t = 2 h. The low doses of Citrin K (300 mg/kg) and Regulator (155 mg/kg) showed no effect on food intake, CitriMax (250 mg/kg) reduced food intake only at t = 2 h (table 1).
Figure 1 Cumulative food intake in rats up to 46 hrs after a single intragastric administration of vehicle or 310 mg/kg Regulator (n = 6). * p < 0.05, **p < 0.01
Figure 2 Cumulative food intake in rats up to 46 hrs after a single intragastric administration of vehicle or 600 mg/kg Citrin K (n = 6). * p < 0.05, **p < 0.01
Figure 3 Cumulative food intake in rats up to 46 hrs after a single intragastric administration of vehicle or 500 mg/kg Super CitriMax HCA-600-SXS (n = 6). **p < 0.01
Table 1 Cumulative food intake in rats after a single intragastric administration of vehicle or different HCA preparations
Regulator (155 mg/kg) Citrin K (300 mg/kg) Super CitriMax HCA-600-SXS (250 mg/kg)
Vehicle Regulator Vehicle Citrin K Vehicle CitriMax
A t = 1 h 2.9 ± 0.6 2.9 ± 0.4 2.9 ± 0.7 2.5 ± 0.6 2.9 ± 0.6 2.7 ± 0.4
t = 2 h 5.7 ± 0.4 4.6 ± 0.5 4.2 ± 0.6 3.5 ± 0.6 4.3 ± 0.6 3.6 ± 0.7*
t = 3 h 6.9 ± 0.6 6.0 ± 0.2 6.0 ± 0.9 5.0 ± 0.8 6.2 ± 0.5 5.2 ± 0.6
t = 6 h 10.4 ± 0.8 10.4 ± 0.5 9.7 ± 0.9 9.3 ± 0.4 9.3 ± 0.7 9.2 ± 0.6
t = 24 h 22.6 ± 0.8 23.6 ± 0.7 23.8 ± 0.7 23.0 ± 1.1 21.1 ± 1.2 21.5 ± 0.7
Regulator (310 mg/kg) Citrin K (600 mg/kg) Super CitriMax HCA-600-SXS (500 mg/kg)
Vehicle Regulator Vehicle Citrin K Vehicle CitriMax
B t = 1 h 3.3 ± 0.5 2.5 ± 0.1 2.8 ± 0.5 2.8 ± 0.4 2.5 ± 0.7 1.9 ± 0.4
t = 2 h 6.4 ± 0.7 3.6 ± 0.3** 4.9 ± 0.8 3.6 ± 0.6 4.1 ± 0.2 2.7 ± 0.7**
t = 3 h 7.8 ± 0.8 4.9 ± 0.3** 6.8 ± 0.8 3.6 ± 0.6** 5.6 ± 0.5 5.0 ± 0.9
t = 6 h 12.1 ± 1.0 8.7 ± 0.9* 11.2 ± 0.6 7.4 ± 0.5** 9.2 ± 0.8 7.9 ± 0.8
t = 24 h 24.8 ± 1.2 20.9 ± 1.1** 25.6 ± 1.0 23.0 ± 0.8* 19.6 ± 1.2 20.0 ± 0.6
For each HCA preparation n = 6, * p < 0.05, ** p < 0.01
To enable comparison between the effects of the different HCA treatments, the difference between 24 hrs cumulative food intake after component treatment and after vehicle treatment was calculated for each HCA source (defined as delta, data shown in table 2). The negative delta's for Regulator (310 mg/kg) and Citrin K (600 mg/kg), implying a decreased food intake after HCA administration when compared to vehicle, were significantly different from the positive delta for CitriMax (500 mg/kg).
Table 2 Effect of intragastric administration of different HCA preparations on cumulative food intake in rats after 24 hrs.
Regulator Citrin K Super CitriMax HCA-600-SXS
(155 mg/kg) (300 mg/kg) (250 mg/kg)
0.6 ± 0.6 -0.8 ± 1.0 0.4 ± 1.1
Regulator Citrin K Super CitriMax HCA-600-SXS
(310 mg/kg) (600 mg/kg) (500 mg/kg)
-3.9 ± 0.6* -2.7 ± 0.7* 0.4 ± 1.1
Data is calculated, for each HCA treatment, as the delta (= difference) between food intake after HCA treatment minus food intake after vehicle treatment. For each HCA preparation n = 6, * p < 0.05 in comparison with Super CitriMax HCA-600-SXS
Figure 4a shows that 475 mg/kg tri-potassium citrate had no effect on cumulative food intake compared to vehicle treatment. Regulator (310 mg/kg) decreased cumulative food intake compared to tri-potassium citrate (figure 4b).
Figure 4 Cumulative food intake in rats up to 46 hrs after a single intragastric administration of A. vehicle or 475 mg/kg tri-potassium citrate (n = 6). B. 475 mg/kg tri-potassium citrate or 310 mg/kg Regulator (n = 6). * p < 0.05, **p < 0.01
No effect on body weight and water intake was observed in any of the experiments (data not shown).
Repeated administration
The effects of repeated administration of the high doses of the three preparations on food intake were comparable to the data of the single bolus experiments. Both Regulator and Citrin K significantly reduced cumulative food intake at days 1–4, whereas CitriMax did not reduce food intake (figure 5 and 6). Three days after the last administration, food intake was still significantly lower after Regulator and Citrin K treatment (figure 5 and 6). Figure 5 also shows that the cumulative food intake during and after Regulator treatment was significantly lower than during and after CitriMax treatment.
Figure 5 Cumulative food intake in rats during intragastric treatment twice daily for 4 subsequent days (day 1–4) and day 3 thereafter (day 7) with vehicle, 310 mg/kg Regulator or 500 mg/kg Super CitriMax HCA-600-SXS (for each treatment n = 5). * p < 0.05, **p < 0.01
Figure 6 Cumulative food intake in rats during intragastric treatment twice daily for 4 subsequent days (day 1–4) and day 3 thereafter (day 7) with vehicle, 310 mg/kg Regulator or 600 mg/kg Citrin K (for each treatment n = 9). * p < 0.05, **p < 0.01
Regulator resulted in a significantly lower gain of body weight compared to vehicle at day 2 and 4. Citrin K significantly reduced gain of body weight compared to vehicle at day 1, 2 and 4 (data not shown).
Discussion
This study shows that three commercially available HCA preparations exert striking differences in efficacy in inhibiting voluntary food intake in rats. Regulator and Citrin K were potent suppressants of food intake, both after a single bolus and after repeated administration, whereas CitriMax exerted much smaller effects on food intake. Accordingly, repeated administration of Regulator and Citrin K, but not CitriMax, reduced gain of body weight. Many of the peer-reviewed human studies on HCA, which report a lack of efficacy, used a CitriMax preparation as their source [9,10,12] at doses that were considerably lower than used in the present animal study, even when corrected for differences in metabolic rate. High doses of (CitriMax) HCA, however, have been shown to be effective in humans [2]. Therefore, the current study indicates that low doses of a relatively low-effective HCA preparation may have contributed to the lack of efficacy in several human studies.
The results of the current study are in line with earlier observations that HCA reduces food intake in rats [15-22]. Most of the early rat studies had not been carried out with the above-mentioned preparations; instead, a tri-sodium salt of HCA was used [15-18]. To our knowledge, no published rat studies used Regulator or Citrin K, while only the recent studies of Leonhardt et al. have been carried out with Super CitriMax HCA-600-SXG [19,20,22]. In these studies, Super CitriMax-600-SXG significantly inhibited food intake in rats, in contrast to the results obtained in our present studies. Differences in the experimental setup may underlie this apparent discrepancy. In their studies a different rat strain and a different experimental setup (10 days supplementation after substantial, fasting-induced body weight loss) were used. Also, a different CitriMax preparation (SXG instead of SXS) was used at a dose that was considerably higher than the maximal dose used in our single-dose administration studies (around 1000 mg/kg versus 500 mg/kg).
The cause of the differences in efficacy between the various HCA preparations remains speculative. In this study, it cannot simply be explained by differences in the relative HCA content of the used preparations. Regulator (97% pure for HCA) was equally effective as Citrin K (50% pure for HCA). Citrin K was more effective than CitriMax (60% pure for HCA). It cannot be excluded that the 40% non-HCA content of CitriMax contains component(s) that interfered with the suppressive effect of HCA on food intake. Alternatively, the extraction method may have resulted in an increased formation of (-)-HCA lactones, which are less potent inhibitors of the citrate cleavage enzyme [13].
It has been suggested that minerals play an important role in regulating the stability, bio-availability or solubility of HCA. In contrast to the two other more effective HCA sources, CitriMax has relatively a high calcium and relatively a low potassium content. Regulator and Citrin K contain negligible amounts of calcium and a relatively high content of potassium. If, as has been suggested, calcium reduces solubility and hinders bio-availability [23,24], the high calcium content in CitriMax may have negatively affected bio-availability or stability of the HCA molecule, resulting in a lower efficacy. It is unlikely that the high potassium content in the two other HCA sources is directly responsible for the observed inhibition of food intake, as tri-potassium citrate did not affect food intake.
The osmolarity of the CitriMax solution was considerably lower than the osmolarity of the more effective Regulator and Citrin K solutions. It has been suggested that osmolarity plays an important role in the regulation of satiety. Ingestion of hypertonic solutions decreased subsequent food intake in pigs [25]. In humans, the non-absorbable fructose stereoisomer D-tagatose inhibited food intake which may have been caused by the osmotic effects of the unabsorbed D-tagatose [26]. However, as the osmolarity of the ineffective tri-potassium citrate was as high as the osmolarity of Regulator and Citrin K, differences in osmolarity cannot solely account for the differences in efficacy of the various preparations used in this study.
It should be noted, however, that a decrease in food intake as a result of induction of malaise or discomfort by any of the used HCA sources cannot be excluded. Still, previous studies on the occurrence of conditioned taste aversion after HCA administration concluded that the food intake reducing effects of HCA cannot be solely explained by the induction of discomfort [27]. Accordingly, no behavioral signs indicating lack of well-being were recognized during these experiments.
In summary, this study shows that different commercially available HCA preparations show differences in their potency: Regulator and Citrin K are potent inhibitors of food intake in rats, whereas similar doses of CitriMax hardly showed any effect on food intake. As most of the (ineffective) human studies used low dose CitriMax preparations, the present study may indicate that low doses of a relatively low-effective HCA preparation may have contributed to the lack of efficacy in several human studies.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
JL: has made substantial contributions to conception and design, acquisition, analysis and interpretation of data and drafted the manuscript.
PW: has made substantial contributions to conception and design, acquisition, analysis and interpretation of data and has been involved in revising the article critically.
AS: has made substantial contributions to conception and design and has been involved in revising the article critically.
AN: has made substantial contributions to conception and design, has been involved in drafting the article and gave final approval of the version to be published.
All authors read and approved the final manuscript.
Acknowledgements
The authors thank the technicians of the analytical laboratory of Nutricia Cuyk for measuring the mineral content of the HCA preparations. We also acknowledge Marloes G. Poelman for her technical assistance in the repeated administration experiments and Rob Verdooren for statistical help.
==== Refs
Thom E Andrews B Short- and long-term efficacy and tolerability of (-)Hydroxcycitrate in the treatment of obesity [abstract] Int Journal of Obesity 1997 21, Suppl. 2 S61(nr 182)
Preuss HG Bagchi D Bagchi M Rao CV Dey DK Satyanarayana S Effects of a natural extract of (-)-hydroxycitric acid (HCA-SX) and a combination of HCA-SX plus niacin-bound chromium and Gymnema sylvestre extract on weight loss Diabetes Obes Metab 2004 6 171 180 15056124 10.1111/j.1462-8902.2004.00328.x
Westerterp-Plantenga MS Kovacs EM The effect of (-)-hydroxycitrate on energy intake and satiety in overweight humans Int J Obes Relat Metab Disord 2002 26 870 872 12037659 10.1038/sj.ijo.0801855
Lim K Ryu S Ohishi Y Watanabe I Tomi H Suh H Lee WK Kwon T Short-term (-)-hydroxycitrate ingestion increases fat oxidation during exercise in athletes J Nutr Sci Vitaminol (Tokyo) 2002 48 128 133 12171433
Conte AA A Non-Presciption Alternative in Weight Reduction Therapy Am J Bariatr Med 1993 17 19
Rothacker D Waitman E Effectiveness of a Garcinia Cambogia and natural Caffeine combination in weight loss - a double blind placebo-controlled pilot study [abstract] Int Journal of Obesity 1997 21, Suppl. 2 S53(nr 148)
Heymsfield SB Allison DB Vasselli JR Pietrobelli A Greenfield D Nunez C Garcinia cambogia (hydroxycitric acid) as a potential antiobesity agent: a randomized controlled trial Jama 1998 280 1596 1600 9820262 10.1001/jama.280.18.1596
Mattes RD Bormann L Effects of (-)-hydroxycitric acid on appetitive variables Physiol Behav 2000 71 87 94 11134690 10.1016/S0031-9384(00)00321-8
Kovacs EM Westerterp-Plantenga MS de Vries M Brouns F Saris WH Effects of 2-week ingestion of (-)-hydroxycitrate and (-)-hydroxycitrate combined with medium-chain triglycerides on satiety and food intake Physiol Behav 2001 74 543 549 11790414 10.1016/S0031-9384(01)00594-7
Kovacs EM Westerterp-Plantenga MS Saris WH The effects of 2-week ingestion of (-)-hydroxycitrate and (-)-hydroxycitrate combined with medium-chain triglycerides on satiety, fat oxidation, energy expenditure and body weight Int J Obes Relat Metab Disord 2001 25 1087 1094 11443511 10.1038/sj.ijo.0801605
Kriketos AD Thompson HR Greene H Hill JO (-)-Hydroxycitric acid does not affect energy expenditure and substrate oxidation in adult males in a post-absorptive state Int J Obes Relat Metab Disord 1999 23 867 873 10490789 10.1038/sj.ijo.0800965
van Loon LJ van Rooijen JJ Niesen B Verhagen H Saris WH Wagenmakers AJ Effects of acute (-)-hydroxycitrate supplementation on substrate metabolism at rest and during exercise in humans Am J Clin Nutr 2000 72 1445 1450 11101469
Cheema-Dhadli S Halperin ML Leznoff CC Inhibition of enzymes which interact with citrate by (-)hydroxycitrate and 1,2,3,-tricarboxybenzene Eur J Biochem 1973 38 98 102 4149431 10.1111/j.1432-1033.1973.tb03038.x
Strubbe JH Keyser J Dijkstra T Prins AJ Interaction between circadian and caloric control of feeding behavior in the rat Physiol Behav 1986 36 489 493 3085115 10.1016/0031-9384(86)90320-3
Sullivan AC Triscari J Hamilton JG Miller ON Effect of (-)-hydroxycitrate upon the accumulation of lipid in the rat. II. Appetite Lipids 1974 9 129 134 4815800
Sullivan C Triscari J Metabolic regulation as a control for lipid disorders. I. Influence of (-)-hydroxycitrate on experimentally induced obesity in the rodent Am J Clin Nutr 1977 30 767 776 324261
Greenwood MR Cleary MP Gruen R Blase D Stern JS Triscari J Sullivan AC Effect of (-)-hydroxycitrate on development of obesity in the Zucker obese rat Am J Physiol 1981 240 E72 8 7457600
Nageswara Rao BD Sakariah K Lipid-lowering and Antiobesity effect of (-)Hydroxycitric acid Nutrition Research 1988 8 209 212
Leonhardt M Hrupka B Langhans W Effect of hydroxycitrate on food intake and body weight regain after a period of restrictive feeding in male rats Physiol Behav 2001 74 191 196 11564468 10.1016/S0031-9384(01)00547-9
Leonhardt M Langhans W Hydroxycitrate has long-term effects on feeding behavior, body weight regain and metabolism after body weight loss in male rats J Nutr 2002 132 1977 1982 12097679
Hellerstein MK Xie Y The indirect pathway of hepatic glycogen synthesis and reduction of food intake by metabolic inhibitors Life Sci 1993 53 1833 1845 8246682 10.1016/0024-3205(93)90491-K
Leonhardt M Balkan B Langhans W Effect of hydroxycitrate on respiratory quotient, energy expenditure, and glucose tolerance in male rats after a period of restrictive feeding Nutrition 2004 20 911 915 15474881 10.1016/j.nut.2004.06.012
Schaller JL Garcinia cambogia for weight loss Jama 1999 282 234; discussion 235. 10422987
Houpt TR Controls of feeding in pigs J Anim Sci 1984 59 1345 1353 6392274
Buemann B Toubro S Raben A Blundell J Astrup A The acute effect of D-tagatose on food intake in human subjects Br J Nutr 2000 84 227 231 11029974
Bremann B Toubro S Raben A Blundell J Astrup A The acute effect of D-tagatose on food intake in human subjects Br J Nutr 2004 84 227 231
Panksepp J Pollack A Meeker RB Sullivan AC (-)-Hydroxycitrate and conditioned aversions Pharmacol Biochem Behav 1977 6 683 687 233711 10.1016/0091-3057(77)90095-8
|
16156903
|
PMC1242253
|
CC BY
|
2021-01-04 16:37:45
|
no
|
Nutr Metab (Lond). 2005 Sep 13; 2:23
|
utf-8
|
Nutr Metab (Lond)
| 2,005 |
10.1186/1743-7075-2-23
|
oa_comm
|
==== Front
RetrovirologyRetrovirology1742-4690BioMed Central London 1742-4690-2-561616806610.1186/1742-4690-2-56CommentaryValproic acid and HIV-1 latency: beyond the sound bite Smith Stephen M [email protected] Section of Infectious Diseases, Department of Medicine, Saint Michael's Medical Center and Department of Preventive Medicine and Community Health, The New Jersey Medical School, Newark New Jersey 07102, USA2005 19 9 2005 2 56 56 12 9 2005 19 9 2005 Copyright © 2005 Smith; licensee BioMed Central Ltd.2005Smith; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
A recent publication in Lancet by Dr. David Margolis and colleagues raised the prospect that HIV infection may be curable. In this pilot study, which received much attention from the press, Dr. Margolis'group found that valproic acid plus enfuvirtide reduces the pool of CD4+ T-cells, which are latently infected with HIV-1, the so-called viral reservoir. This commentary critically addresses current data on this topic.
==== Body
A recent publication in Lancet [1] has created quite a stir among HIV circles (no, not those containing 1 or 2-LTRs) [2] and the lay press. In a study led by Dr. David Margolis, four patients, each of whom had undetectable HIV plasma viral loads, had enfuvirtide and valproic acid (VPA) added to their anti-HIV regimens. The researchers measured several parameters with particular interest in the effect of VPA on resting CD4+ T-cells latently infected with HIV-1. This pool of cells – often referred to as the viral reservoir – is thought to pose an significant obstacle to HIV eradication. Drs. Finzi, Siliciano, and colleagues first described these CD4+ T-cells that produce little or no viral proteins but can be stimulated with mitogens to produce infectious virus [3]. This pool of cells is also thought to serve as a biological library for archival strains of drug-resistant HIV [4]. Even years after discontinuation of a given drug, resistant strains of HIV-1 quickly re-emerge when the drug is re-introduced into a therapeutic regimen. Therefore, in theory, a treatment protocol that reduces or eliminates this pool of infected cells could eliminate drug-resistant strains, and would have the potential to cure HIV infection. However, in the absence of effective drug intervention, there is little chance of eliminating this viral reservoir; the half-life and pool size of latently infected CD4+ T-cells are considered too great to permit eradication under current regimens in most, if not all, patients. Dr. Siliciano and colleagues estimate the mean time to eradication is 51.2 years in the best case scenario [5], e.g., those patients who have undetectable viral loads and no viral "blips."
Dr. Margolis' laboratory previously demonstrated that VPA can stimulate the release of virus from latently infected CD4+ T-cells in vitro [6]. The stimulatory effect of VPA is equal to, or greater than, that of the mitogen, PHA, but VPA has no effect on T-cell activation or virus production from mitogen-activated lymphoblasts. VPA inhibits histone deacetylase (HDAC)-1, which may be involved in suppressing HIV promoter activity in latently infected, resting CD4+ T-cells. In the Lancet study, four patients with long-term, undetectable viremia were given enfuvirtide, an injectable HIV fusion inhibitor, added to their ongoing regimens. After 4–6 weeks, VPA was then started. The VPA dose (500–750 mg twice per day) was adjusted to maintain plasma concentrations within a defined range (50–100 mg/L). The frequency of infection in resting CD4+ T-cells was measured twice at baseline prior to initiation of VPA therapy, and again 12 weeks after the start of VPA treatment. While baseline measurements showed little or no change in the frequency of latently infected CD4+ T-cells, enfuvirtide and VPA therapy decreased this measurement by 29–84% in all four subjects. No changes were observed in the frequency of HIV proviral DNA or immune activation markers. The authors conclude that HDAC inhibitors, such as VPA, could lead to HIV eradication when combined with other anti-HIV drugs.
The results of this pilot study are intriguing, but must be considered cautiously with a clear understanding of their inherent limitations. By design, the study was not controlled and each patient received two new drugs, enfuvirtide and VPA, and the relative contribution of each drug to lowering the frequency of latently infected CD4+ T-cells is unknown. At least one group has demonstrated that intensification of anti-HIV therapy decreases the half-life of this population [7]. Moreover, absolute CD4+ T-cell counts can vary significantly even in stable patients with undetectable viral loads, and this variability may influence quantitative assessments of the latent pool of infected cells. While the reported decrease in latently infected CD4+ T-cells in the four patients receiving VPA is certainly promising, further evaluations of the efficacy of VPA in combination with other anti-HIV drugs will be needed in larger, controlled clinical studies.
This study also raises important issues regarding the use of enfuvirtide in an intensification regimen. As reported, two patients had residual viremia after intensification with enfuvirtide making it unclear whether this intensification is necessary or especially beneficial. Theses data suggest that intensification may not be necessary or helpful, or more importantly, that reduction of the reservoir pool may occur in the presence of on-going, low level viral replication. Presumably this issue will be addressed in future studies.
Of considerable interest is the potential mechanism by which VPA reduces the frequency of latently infected, resting CD4+ T-cells. Presumably, through inhibition of HDAC, VPA allows initiation of viral transcription, which in turn leads to production of viral proteins and virions, and cell death due to virally induced cytotoxicity. Paradoxically, VPA does not activate resting CD4+ T-cells, thus making it unclear how HIV transcription is upregulated and viral promoter activity is increased.
In this regard, many critical questions remain to be answered. Foremost, what happens to the pool of latently infected cells after enfuvirtide and VPA are discontinued? Data by Dr. Margolis and colleagues have shown VPA works quickly in vitro to induce virus production from latently infected cells. In vivo, very few new latently infected cells would be expected to develop over the duration of VPA treatment presented in their Lancet study. And yet a proportion of latently infected cells remained after 3 months of therapy. Do these remaining cells represent a distinct subset from those eliminated by enfuvirtide and VPA? Viral RNA clearance has two phases of decay [8]. Does the population of latently infected CD4+ T-cells have similarly complicated kinetics?
Others have tried to reduce the latent viral reservoir, primarily through activation of resting CD4+ T-cells. In one sobering example, Drs. Fauci, Chun, Lane and colleagues stopped anti-HIV medications in two patients, who had undetectable viral loads for years and had received IL-2 therapy [9,10]. At the time HIV therapy was discontinued virus could not be cultured from resting CD4+ T-cells, either from the blood or lymph nodes, of either patient. More significantly, proviral DNA was below the limit of detection (0.5 copies per 106 PBMC). Despite these impressive laboratory findings, within three weeks, plasma HIV RNA levels became detectable and rose above 10,000 copies per ml.
Finally, the assumption that the population of latently infected, resting CD4+ T-cells is the only reservoir for HIV-1 in vivo is largely untested. While the persistence of these cells likely guarantees chronic infection, their elimination may not result in eradication. Other pools of HIV-infected cells or tissue reservoirs may exist. While current results presented in the Lancet study certainly provide reason to be optimistic, it is critical to balance this optimism with further rigorous clinical evaluations, including larger, controlled studies of VPA and enfuvirtide. As always, the results of any pilot study must be interpreted with caution and placed in the proper context of existing knowledge. It is my hope that Dr. Margolis and colleagues are correct, and that HIV reservoirs can be eliminated through the additional administration of a small molecule, such as VPA. However, the challenge of eradicating HIV remains daunting, and history and science have yet to yield simple answers.
==== Refs
Lehrman G Hogue IB Palmer S Jennings C Spina CA Wiegand A Landay AL Coombs RW Richman DD Mellors JW Coffin JM Bosch RJ Margolis DM Depletion of latent HIV-1 infection in vivo: a proof-of-concept study Lancet 2005 366 549 555 16099290 10.1016/S0140-6736(05)67098-5
Cohen J HIV/AIDS. Report of novel treatment aimed at latent HIV raises the 'c word' Science 2005 309 999 1000 16099956 10.1126/science.309.5737.999a
Finzi D Hermankova M Pierson T Carruth LM Buck C Chaisson RE Quinn TC Chadwick K Margolick J Brookmeyer R Gallant J Markowitz M Ho DD Richman DD Siliciano RF Identification of a reservoir for HIV-1 in patients on highly active antiretroviral therapy Science 1997 278 1295 1300 9360927 10.1126/science.278.5341.1295
Siliciano JD Siliciano RF A long-term latent reservoir for HIV-1: discovery and clinical implications J Antimicrob Chemother 2004 54 6 9 15163657 10.1093/jac/dkh292
Siliciano JD Kajdas J Finzi D Quinn TC Chadwick K Margolick JB Kovacs C Gange SJ Siliciano RF Long-term follow-up studies confirm the stability of the latent reservoir for HIV-1 in resting CD4+ T cells Nat Med 2003 9 727 728 12754504 10.1038/nm880
Ylisastigui L Coull JJ Rucker VC Melander C Bosch RJ Brodie SJ Corey L Sodora DL Dervan PB Margolis DM Polyamides reveal a role for repression in latency within resting T cells of HIV-infected donors J Infect Dis 2004 190 1429 1437 15378435 10.1086/423822
Ramratnam B Ribeiro R He T Chung C Simon V Vanderhoeven J Hurley A Zhang L Perelson AS Ho DD Markowitz M Intensification of antiretroviral therapy accelerates the decay of the HIV-1 latent reservoir and decreases, but does not eliminate, ongoing virus replication J Acquir Immune Defic Syndr 2004 35 33 37 14707789
Simon V Ho DD HIV-1 dynamics in vivo: implications for therapy Nat Rev Microbiol 2003 1 181 190 15035022 10.1038/nrmicro772
Chun TW Davey RTJ Engel D Lane HC Fauci AS Re-emergence of HIV after stopping therapy Nature 1999 401 874 875 10553903 10.1038/44755
Davey RTJ Bhat N Yoder C Chun TW Metcalf JA Dewar R Natarajan V Lempicki RA Adelsberger JW Miller KD Kovacs JA Polis MA Walker RE Falloon J Masur H Gee D Baseler M Dimitrov DS Fauci AS Lane HC HIV-1 and T cell dynamics after interruption of highly active antiretroviral therapy (HAART) in patients with a history of sustained viral suppression Proc Natl Acad Sci U S A 1999 96 15109 15114 10611346 10.1073/pnas.96.26.15109
|
16168066
|
PMC1242254
|
CC BY
|
2021-01-04 16:36:38
|
no
|
Retrovirology. 2005 Sep 19; 2:56
|
utf-8
|
Retrovirology
| 2,005 |
10.1186/1742-4690-2-56
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-1021615329910.1186/1465-9921-6-102ReviewThe pathophysiological function of peroxisome proliferator-activated receptor-γ in lung-related diseases Huang Tom Hsun-Wei [email protected] Valentina [email protected] Bhavani Prasad [email protected] Diana Shu-Hsuan [email protected] Basil D [email protected] Faculty of Pharmacy, A15, University of Sydney, New South Wales, 2006, Australia2005 9 9 2005 6 1 102 102 17 1 2005 9 9 2005 Copyright © 2005 Huang et al; licensee BioMed Central Ltd.2005Huang et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Research into respiratory diseases has reached a critical stage and the introduction of novel therapies is essential in combating these debilitating conditions. With the discovery of the peroxisome proliferator-activated receptor and its involvement in inflammatory responses of cardiovascular disease and diabetes, attention has turned to lung diseases and whether knowledge of this receptor can be applied to therapy of the human airways. In this article, we explore the prospect of peroxisome proliferator-activated receptor-γ as a marker and treatment focal point of lung diseases such as asthma, chronic obstructive pulmonary disorder, lung cancer and cystic fibrosis. It is anticipated that peroxisome proliferator-activated receptor-γ ligands will provide not only useful mechanistic pathway information but also a possible new wave of therapies for sufferers of chronic respiratory diseases.
Peroxisome proliferator-activated receptor-gammarespiratory diseasesasthmachronic obstructive pulmonary diseaselung cancer.
==== Body
Introduction
It would be fair to say that airway diseases place a significant burden on the population in terms of health, social and economic costs. Leading the way are the chronic pulmonary disorders such as asthma and lung cancer, riddled with significant obstacles associated with their various drug treatments, including limited effectiveness, immunity and side effects. Recent studies delve into the role of inflammation in the airways and its associated army of diverse cell types including leukocytes, lymphocytes, neutrophils and eosinophils [1]. Modern treatments have focused on receptor-mediated responses in an attempt to effectively counteract a specific disease state. Recently, peroxisome proliferator-activated receptors (PPAR), in particular, PPAR-γ, have surfaced as novel immunomodulators due to their anti-inflammatory actions, most notably in cardiovascular and diabetes-related diseases [2,3]. This regulation of inflammatory responses by PPAR-γ has been extended to processes within the lung, through actions on both immune and non-immune cells [5]. Widespread clinical use of PPAR-γ agonists has provided a possible new direction in the treatment of airway inflammatory diseases through control of PPAR-γ regulated pathways [4]. This has uncovered the potential of inhaled PPAR-γ agonists in the treatment of airway inflammation via the many cellular targets in the lung such as T lymphocytes, epithelial cells and smooth muscle cells with the possibility of delivering them locally, with minimal side effects, compared to the currently available corticosteroids [5]. Current studies have allowed greater insight into the role of the receptor on the modulation of airway respiratory diseases by interaction with its agonists, 15-deoxy-Δ12,14-prostaglandin J2 (15D-PGJ2) and thiazolidinediones (TZD). This review will summarise the connections between PPAR-γ interactions with agonists and the mechanisms involved in lung cellular processes in chronic diseases such as asthma, lung cancer, cystic fibrosis and chronic obstructive pulmonary disease (COPD).
PPARs: Background
Since the turn of the decade, the science of receptor-mediated responses has progressed rapidly, uncovering many unknown pathways of pharmaceutical drug action and, lately, targeting many diseases where conventional medicine has had limited success. The literature on the PPAR physiology is extensive. Briefly, the PPARs are a family of transcription factors belonging to the nuclear hormone receptor superfamily [6,7]. Three PPAR isoforms, designated PPAR-α (NR1C1), PPAR-β (also called PPAR-δ, FAAR, NuC1 or NR1C2) and PPAR-γ (NR1C3) have been cloned and are differentially expressed in several tissues including liver, kidney, heart and muscle. PPAR-α primarily regulates cellular lipid metabolism and modulates inflammation. PPAR-β participates in embryonic development, implantation and bone formation. PPAR-γ, which is the focus of this review, is a key factor in adipogenesis and is primarily advocated in insulin sensitivity, cell cycle regulation and cell differentiation [6]. A large proportion of PPARs actions are mediated through binding to PPAR-response elements (PPRE) on DNA. PPRE are constituents of direct repeat (DR) hexameric sequences (AGGTCA), which are separated by one or two nucleotides (DR-1 and DR-2 element). Distinct areas such as the DNA binding and the ligand-independent transactivation domains have been identified and these influence the transduction of the PPAR-induced response [8]. PPARs heterodimerise with the 9-cis-retinoic acid receptors (RXR) and the resultant heterodimer subsequently binds to PPRE with the recruitment of cofactors. PPARs regulate numerous genes through ligand-dependent transcriptional activation and repression. This conformational interaction has a profound affect on numerous cellular processes, including lipid metabolism, glucose homeostasis, cell cycle progression, cell differentiation, inflammation and extracellular matrix remodelling [9]. The localisation of a ligand to the ligand-binding domain results in a conformational change of the receptor, thereby allowing transactivation of the appropriate genes [6]. The natural prostaglandin D2 metabolite, 15D-PGJ2 and synthetic anti-diabetic TZDs are principal ligands of PPAR-γ and will be the focus of the review.
Expression and physiological role of PPAR-γ in lung
Expression
Historically, the discovery of PPAR-α led to the subsequent identification of other isoforms such as PPAR β/δ and PPAR-γ [10]. The PPAR-γ gene contains three promoters that yield three sub-isoforms, namely, PPAR-γ1, PPAR-γ2 [11] and PPAR-γ3 [12]. A comparison of the tissue-distribution of PPAR-γ transcripts among different species illustrates the presence of PPAR-γ1 in a broad spectrum of tissues such as heart, skeletal muscle, small and large intestine, kidney, pancreas and spleen, whereas PPAR-γ2 is restricted to adipose tissue [6]. Structurally, PPAR-γ2 contains an additional 30 amino acids at the N-terminal end relative to PPAR-γ1. PPAR-γ3 is abundant in macrophages, the large intestine and white adipose tissue [12]. Specific to the distribution of PPAR-γ in lung, the expression of PPAR-γ1 was exhibited at relatively high levels in bovine lung compared to PPAR-γ2. The cellular expression profile of PPAR-γ in pulmonary tissue has not been well characterised, but studies have uncovered abundant expression of PPAR-γ in airway epithelium [13], in bronchial submucosa [14], in mononuclear phagocytes such as human alveolar macrophages (AM) [3], human T lymphocytes [2], in two different human bronchial epithelial cells, NL20 and BEAS [15] and human airway smooth muscle (HASM) cells [2,16]. In HASM cells, PPAR-α but not PPAR-β was expressed [47]. Primary normal human bronchial epithelial cells and human lung epithelial cell lines BEAS 2B, A549 and NCI-H292 all express PPAR-γ and PPAR-β, but not PPAR-α [28]. Both PPAR-α and PPAR-γ are expressed by eosinophils [29]. Mice, rat and human lung models have been pivotal to the greater understanding of the mechanistic pathways related to PPAR-γ and the various lung diseases (Figure 1).
Figure 1 Expression of PPAR-γ in various tissues and its role in lung and other organs. PPAR-γ ligands implicated in the treatment of chronic inflammatory disorders in lung. Activation of PPAR-γ in heart, intestine, kidney, skeletal muscle, pancreas, macrophages and adipose tissue results in energy homeostasis and this effect also found to be crucial in the pathophysiology of different disorders. Please refer text for more information.
Physiology
Although established for glucose metabolism, target cells for PPAR-γ agonists and the mechanisms by which they hinder inflammation within the airways are not well defined [5]. Culminating evidence suggests that PPAR-γ may act by exerting its influence as a negative immunomodulator regulating inflammatory respiratory responses (Figure 2). Pro-inflammatory cytokines seem to be the first point of call. For example, in adipose tissue, the adipogenic action of the TZD PPAR-γ ligands are opposed by several pro-inflammatory cytokines, including tumour necrosis factor (TNF)-α and interferon (IFN)-γ (Figure 2). In vitro, the TZDs blocked the effects of TNF-α on both adipogenesis and insulin sensitivity and, similarly, 15D-PGJ2 was found to prevent IFN-γ-induced murine macrophage activation [17].
Figure 2 Activation of PPAR-γ by endogenous (15D-PGJ2) and exogenous (TZDs) ligands results in transcription of wide array of genes that can control pathogenesis of acute and chronic disorders in various tissues of lungs. Please refer text for more information. Abbreviations: 15D-PGJ2: 15-deoxy-Δ12,14-prostaglandin J2, Cpla2: cytosolic phospholipase A2, TZDs: Thiozolidinediones NSAIDs: Non-steroidal anti-inflammatory drugs MCP:1monocyte chemoattractant protein, G-CSF: granulocyte-colony-stimulating factor, GM-CSF:granulocyte-macrophage-colony-stimulating factor, KC: keratinocyte-derived chemokine, NOS: Nitric oxide synthases, SP-B: surfactant proteins-B, MMP-9: matrix metalloproteinase 9, TGF-β: Transforming growth factor-β, IgE and IgG1: Immunoglubulin E and Immuno globulin G1, NF-κB: Nuclear factor-κB, EP2: Prostaglandin E2 receptor, PGE2: Prostaglandin E2, aP2: Adipocyte fatty acid binding protein, UCP 1&3: Uncoupling proteins 1 & 3, Acrp30: Adipocyte complement related factor 30, FATP-1: Fatty acid transport protein-1.
In murine macrophages and human lung epithelial cell line A549, expression of PPAR-γ was upregulated by interleukin-4 (IL-4), a cytokine critical for certain subsets of airway inflammation [17,18]. Similarly, IL-4 induced 12/15-lipoxygenase (12/15-LO), an enzyme capable of generating PPAR-γ agonists in vivo. 12/15-LO was also highly expressed in surface airway epithelial cells under basal conditions [17]. Nitric oxide synthases (NOS) are responsible for the in vivo synthesis of NO, a short-lived molecule that is an effective bactericidal agent and may also regulate expression of various pro-inflammatory genes, such as IL-8, a potent chemoattractant and activator of neutrophils. Both NOS and IL-8 play an important role in airway host defence and elevated levels of IL-8 are found in bronchoalveolar lavage fluid from intrinsic asthmatic patients [19]. The two PPAR-γ agonists, 15D-PGJ2 and ciglitazone dose-dependently blocked the cytokine-induced expression of the inducible form of NOS. Ciglitazone alone only slightly affected cytokine-induced IL-8 secretion, however, the agonist significantly reduced IL-8 secretion from cells pre-treated with IL-4 [17]. Therefore, PPAR-γ is expressed and upregulated by IL-4 in airway epithelial cells and through the activation of airway epithelial, PPAR-γ down-regulates expression of inflammatory mediators. In essence, PPAR-γ may act as an anti-inflammatory agent via 12/15-LO-dependent pathways [17].
Certain lung proteins may also be involved. The association of PPAR-γ with the recruitment and activation of peripheral blood monocytes, such as the potent chemokine monocyte chemoattractant protein (MCP)-1, has also been studied [20]. MCP-1 is produced by lung epithelial cells during the course of inflammatory lung diseases. Studies by Momoi's group [20] have demonstrated TZD's ability to inhibit MCP-1 protein and mRNA expression in cytokine-treated A549 lung epithelial cells.
The expression and physiological role of PPAR-γ in pulmonary nonciliated bronchiolar epithelial cells (Clara cells) and alveolar type II (AT II) epithelial cells has also been investigated [21]. These cells are highly lipogenic and are responsible for maintaining pulmonary surfactant homeostasis [22]. Among the surfactant proteins, SP-B is a 79-amino acid amphipathic peptide that is synthesised and produced in Clara cells and AT II epithelial cells. The SP-B facilitates lamellar body formation in AT II epithelial cells and phospholipid spreading during the respiratory cycles. The inhibitory effect of PPAR-γ ligands on SP-B gene expression reveals a novel mechanism in the regulation of pulmonary surfactant homeostasis [21]. In the presence of 15D-PGJ2, the transcriptional level of SP-B was down-regulated in respiratory epithelial cell line and whole lung explant systems. Similarly, 15D-PGJ2 suppressed hSP-B gene activity at the -218 to -41 promoter region in human pulmonary adenocarcinoma H441 cell line transfected with various hSP-B luciferase reporter gene constructs.
The intricate multifactorial coordination of PPAR-γ and CCAAT/enhancer-binding proteins (C/EBP) for lung development during the perinatal period has also been displayed [23,24]. C/EBPs is a family of basic leucine-zipper transcription factors controlling a wide array of genes and have been postulated to serve a central role in normal tissue development and regulation of cell proliferation or differentiation [25]. C/EBPβ and δ are known to act synergistically with PPAR-γ to promote adipocyte differentiation [23]. C/EBPα gene-deficient mice die shortly after birth due to abnormal lung histology, including interstitial thickening and hyperproliferation of AT II cells [26]. In developing foetal rat lungs, the C/EBPα, β, δ, and PPAR-γ1 mRNA expression was increased by 3- to 5-fold from Day 18 of gestation, peaking at 1 to 2 days before birth. However, there was a transient decline of expression during the first postnatal day and a return to prenatal levels on postnatal Day 5. In the AT II cell line, C/EBPα mRNA was not detected throughout the developmental stage; C/EBPβ and δ mRNAs expression was similar to that of whole lung, with a prenatal rise profile, whereas PPAR-γ did not display any developmental increase. The expression of PPAR-γ2 was not detected in whole lung or in AT II cell line [24].
Changes in the metabolism of fatty acids such as arachidonic acid may also have detrimental effects on chronic respiratory diseases including asthma, chronic bronchitis, cystic fibrosis and bronchiectasis, as well as lung injury and sepsis [27] The 85-kDa cytosolic phospholipase A2 (cPLA2) plays an essential role in the control of arachidonic acid metabolism. It has been shown that cPLA2 overexpression significantly increased the PPAR-γ-mediated reporter activity and this activation by cPLA2 may represent a novel mechanism for the control of airway inflammation [28].
Asthma and PPAR-γ
Asthma is a widespread chronic disease, with an increasing incidence among children under 18 years of age [1]. Latest news reports headline the disease and its appearance in the elderly at an alarming rate. Sufferers are plagued with many undesirable pro-inflammatory events in the airway, including narrowing and increased production of mucous, thickening of the wall and thus reduction of the airflow through the lungs. This response is accompanied by the activation of cell types such as T cells and eosinophils and histopathological cellular airway restructuring within the airways [4,5,14]. Airway inflammation and alterations in cellular turnover are histopathologic features of asthma [4] and recently, research has disclosed the involvement of PPARs such as PPAR-γ and PPAR-α in many facets of the disease such as decreasing antigen-induced airway hyperresponsiveness, lung inflammation, eosinophilia, cytokine production and serum levels of antigen-specific IgE [29]. Airway remodelling is characterised by the increase in subepithelial membrane (SBM) and collagen deposition. A recent study displayed a positive correlation between PPAR-γ expression and SBM thickening and collagen deposition in the epithelium [4]. In the submucosa, PPAR-γ expression was related to both SBM thickening and to the number of proliferating cells. Negative correlation was found between the intensity of PPAR-γ expression in the bronchial submucosa, the airway epithelium and the smooth muscle to the forced expiratory volume (FEV1) values. Inhaled steroids (either administered alone or in combination with oral steroids) restrained PPAR-γ expression in all the compartments, cell proliferation, SBM thickness and collagen deposition, enhancing apoptotic death in the epithelium and the submucosa. In this study, T lymphocytes in the bronchial mucosa failed to express PPAR-γ. Therefore, PPAR-γ may be an indicator of airway inflammation and remodelling in asthma (Table 1).
Table 1 This table shows PPAR-γ activators, inflammatory mediators affected by PPAR-γ expression and different disorders which can be controlled by up-regulation of PPAR-γ. Abbreviations: TZDs: Thiozolidinediones, NSAIDs: Non-steroidal anti-inflammatory drugs, 15D-PGJ2: 15-deoxy-Δ12,14-prostaglandin J2, Cpla2: cytosolic phospholipase A2, IL-4: Interleukin-4, MCP:1monocyte chemoattractant protein, G-CSF: granulocyte-colony-stimulating factor, GM-CSF:granulocyte-macrophage-colony-stimulating factor, KC: keratinocyte-derived chemokine, NOS: Nitric oxide synthases, SP-B: surfactant proteins-B, MMP-9: matrix metalloproteinase 9, TGF-β: Transforming growth factor-β, IgE and IgG1: Immunoglubulin E and Immuno globulin G1, NF-κB: Nuclear factor-κB, EP2: Prostaglandin E2 receptor, PGE2: Prostaglandin E2.
LIGANDS DOWN REGULATION IMPLICATION UP REGULATION IMPLICATION
TZDs (Exogenous) Cytokines (IL-8, IL-4, IL-5, IL-6 and IL-13)
NOS
MCP-1 Asthma and other pulmonary inflammatory diseases aP2
UCP1
UCP3
Acrp30 Insulin resistance
Obesity
Hyperlipidaemia
NSAIDs (Exogenous) SP-B
AHR
15D-PGJ2 (Endogenous) TGF-β
GATA-3
IgE and lgG1
IL-4 (Endogenous) T-cell response
MMP-9
G-CSF and KC
azelaoyl-phosphocholine (Endogenous) GM-CSF COPD FATP-1
LPL (Adipose tissue) Atherosclerosis
Eicosenoids (Endogenous) Cyclin D1
NF-κB
PGE2
EP2 Lung cancer (NSCLC, LCC)
In ovalbumin (OVA)-sensitised BALB/c mice (a murine model of human asthma), PPAR-γ activation by ciglitazone treatment inhibited antigen-induced airway hyperresponsiveness (AHR), basement membrane thickness, collagen deposition and transforming growth factor (TGF)-β synthesis, lung inflammation, eosinophilia, cytokine production (IL-4, IL-5, IL-6 and IL-13), GATA-3 expression and serum levels of antigen-specific IgE and IgG1. In vitro chemotaxis and antibody-dependent cellular cytotoxicity in human or rat eosinophils were also prevented. The PPAR-γ antagonist GW9662 reversed the above effects [5,29,30]. Similarly, PPAR-γ selective agonist GI 262570 administered intranasally in OVA-induced BALB/c reduced the elevated allergen-induced bronchoalveolar lavage eosinophil and lymphocyte but not neutrophil influx. In OVA-pulsed dendritic cells (DC), rosiglitazone, a PPAR-γ agonist, averted the migration of antigen-loaded DCs in the mediastinal lymph nodes (MLN) and reduced the T-cell response in the MLNs [30]. Therefore, PPAR-γ stimulation of DCs may have a potential therapeutic role in reducing sensitisation to inhaled allergens.
In similar experiments, PPAR-γ agonist GI 262570, PPAR-α agonist GW 9578 and dual PPAR-α/γ agonist GW 2331 selectively inhibited allergen-induced bronchoalveolar lavage eosinophil and lymphocyte influx in OVA-sensitised BALB/c mice. However, PPAR-δ agonist GW 501516 had no effect. There was no inhibition of LPS-induced bronchoalveolar lavage neutrophil influx or TNF-α and keratinocyte-derived chemokine (KC) production by all agonists administered intranasally before the challenge. In A549 cells, the PPAR agonists did not inhibit intracellular adhesion molecule-1 expression. Thus, in vitro data suggests that PPAR effects on bronchoalveolar lavage eosinophil and lymphocyte influx may not be mediated by the antagonism of the NF-κB pathway [31].
Interleukin-5 (IL-5) is the principal regulatory cytokine mediating eosinophil airway inflammation and extending the cell's survival. Eosinophils liberate cytotoxic products at the site of inflammation, thus triggering AHR. IL-5-stimulated (but not spontaneous) eosinophil survival and eotaxin-directed chemotaxis was dose-dependently reduced by the PPAR-γ agonist troglitazone. The results indicated that upregulation of PPAR-γ in asthma may prevent further activation of pro-inflammatory cells of the airway [14].
Enzymes may also play a part in the PPAR-γ puzzle. Matrix metalloproteinase (MMP)-9 (gelatinase B) is a matrix-degrading enzyme found in human normal bronchial epithelial cells and is involved in airway wall remodelling generated by inflammatory processes. Activation of PPAR-γ by rosiglitazone or pioglitazone in human bronchial epithelial NL20 and BEAS cell lines dose-dependently limited the expression of MMP-9 gelatinolytic activity induced by TNF-α and phorbol myristate acetate. In contrast, the expression of the local inhibitor of MMP-9, tissue inhibitor type 1, was retained. In this study, however, transient transfection and electromobility shift assays affirmed inhibition of nuclear factor (NF)-κB activation by PPAR-γ agonists, resulting in decreased MMP-9 mRNA expression [15]. In untreated atopic asthmatic patients, there was an enhanced expression of PPAR-γ, which suggested signs of airway transformation, including increased density of the SBM and collagen deposition in the epithelium, with no relation to proliferation or apoptosis. In contrast, PPAR-γ-expressing cells in the submucosa were related to both SBM thickening and to the number of Ki67-, but not caspase-3-expressing-, cells. It was proposed that PPAR-γ might not be involved in epithelial cell turnover, but rather may manipulate extracellular matrix accumulation and submucosal cell proliferation [4] (Table 1).
PPAR-γ activation also influences lung survival factors and apotosis. In male BALB/c mice, the initial levels of the cytokines were not affected by the PPAR agonists, rosiglitazone or SB 219994. Aerosolised lipopolysaccharide (LPS) exposure caused a significant increase in neutrophil numbers in both lung lavage and tissue, however, lymphomononuclear (LMN) cell numbers in BAL fluid and lung tissue did not change. On pre-treatment with the PPAR ligands, the increase in pro-inflammatory cytokines granulocyte-colony-stimulating factor (G-CSF) and KC levels was reduced in the lung tissue but not in the lung lavage fluid. At the trial doses, the PPAR-γ agonists did not affect LMN cells numbers in the BAL nor lavage or lung tissue homogenate MMP-9 content. Rosiglitazone, when administered after the LPS insult, reduced the lung tissue G-CSF and neutrophilia levels and had no effect on KC or granulocyte-macrophage (GM-CSF) levels. The results suggested therapeutic similarities between rosiglitazone and the steroid, dexamethasone [2] (Table 1).
AMs are phagocytes involved in the ingestion and degradation of inhaled particles. This activates a variety of inflammatory processes involving enhancement of their cytotoxic capabilities. LPS-induced human AMs treated with 15D-PGJ2 and troglitazone showed a significant reduction of the TNF-α cytokine production. This was coupled with an increase in the expression of the scavenger receptor CD36 (which contains a functional PPAR-γ responsive element) and subsequent augmented apoptotic neutrophil phagocytosis in the ligand-treated AMs [3]. Therefore, administration of PPAR-γ synthetic agonists such as TZDs may contribute as adjunct therapeutic agents for airway diseases of the lung, such as asthma [7,14] (Table 1).
Lung Cancer and PPAR-γ
Lung cancer is the leading cause of cancer-related death in developed countries and currently eludes the available therapies. Consequently, the prognosis of patients with lung cancer is generally poor, with a 10–15% 5 year survival rate [32]. High PPAR-γ expression has been suggested as a potential marker for lung cancer and the degree of PPAR-γ protein appears to correlate with the maturational stage, differentiated phenotype, as well as the tumour histological type and grade in lung adenocarcinoma [33,34]. Studies have indicated that upon addition of PPAR-γ selective agonists, growth of lung cancer cells was prevented through the induction of differentiation and apoptosis [35-38]. Additionally, decreased PPAR-γ expression has been correlated with poor prognosis in patients with lung cancer, suggesting that the gene expression may be further diminished as lung cancer progresses [33]. PPAR-γ-selective agonists such as ciglitazone and 15D-PGJ2 have diminished the growth of non-small cell lung cancer (NSCLC) cells through the induction of apoptosis, promotion of differentiation and the down-regulation of cell cycle proteins such as Cyclin D1 [35,37]. Treatment with troglitazone and pioglitazone significantly reduced the number of lung metastases and restricted NSCLC tumour progression in vivo [34]. Similarly, combination of ciglitizone with trichostatin (an inhibitor of histone deacetylase) demonstrated potent growth-inhibitory and differentiation-inducing activity in NSCLC, prompting the possibility of combinational differentiation therapy for the treatment of lung adenocarcinomas [37]. Likewise, untreated large cell carcinoma (LCC) cells displayed increased NF-κB activity, a pro-survival mechanism for this cancer in preventing apoptosis. Upon treatment with thalidomide, the elevated level of NF-κB activity was constrained in the presence of thalidomide and the PPAR-γ protein expression in LCC was dose-dependently increased [32]. Therefore, as activation of PPAR-γ impedes lung tumour progression, it is feasible that TZDs may serve as potential therapeutic agents for both NSCLC and LCC (Table 1).
Another aspect of carcinogenesis is the role of the inducible enzyme, cyclooxygenase (COX)-2. COX-derived prostaglandins (PG) exhibit modulation of cell proliferation, apoptosis, angiogenesis and immunity [39]. Prostaglandin E2 (PGE2) is a major COX-2 metabolite and plays an important role in tumour biology and its function is mediated through G protein-coupled PGE receptor (EP) [40]. The NSCLC cell expressing EP2 receptors, a key modulator of tumor development, has its mRNA and protein expression significantly attenuated in the presence of PPAR-γ ligands, GW1929, 15D-PGJ2, ciglitazone, troglitazone and rosiglitazone [41]. The effects of non-steroidal anti-inflammatory drugs (NSAIDs) on decreased lung cancer cell growth have also been examined [42,43]. Sulindac sulfide, a COX inhibitor, activated PPAR-γ at higher concentration (50 μM). Together with ciglitazone, sulindac sulfide potently suppressed NSCLC cell growth [42]. Another COX-2 inhibitor, nimesulide (which is known to induce PPAR-γ expression), has also had some success in curbing tumour growth in female nu/nu mice xenografted with subcutaneous A549 lung tumour cell line and significantly reduced intratumour PGE2 levels [43]. Therefore, the potential therapeutic application of NSAIDs and TZDs in the treatment and/or prevention of lung cancer are promising, however more research is still needed in order to evaluate the long-term safety and efficacy of combined NSAIDs and TZDs in lung cancer [44] (Table 1).
On the contrary, PPAR-α was not expressed in human lung cancer cell lines and, thus, respective agonists such as bezafibrate and prostanoids (PGE2 and PGF2α) did not inhibit growth of the cancer cell lines by inducing apoptosis [35].
Other Respiratory Disorders and PPAR-γ
Cystic fibrosis is a genetic disorder characterised by functional deficiencies of the reproductive, digestive and respiratory systems. With the help of genetic mapping and improved, more consistent treatment, patients are enjoying longer and fulfilled lives. Adding to the improved outlook, it is believed that respiratory PPAR-γ expression is altered in tissues deficient in the normal cystic fibrosis transmembrane regulator protein (CFTR). It was found that PPAR-γ expression was decreased significantly in (CFTR)-regulated tissues (colon, ileum and lung) from exon 10 CFTR (cftr_/_) mice compared to wild-type mice. In contrast, no differences were found in fat and liver. In the lung tissue of both mice types, there was a mixed labelling of both nuclei and cytoplasm localised to larger bronchi and a diffuse lighter staining of the remaining tissue [45].
The deficiency of GM-CSF is strongly implicated in the pathogenesis of pulmonary alveolar proteinosis (PAP), a rare interstitial lung disease manifested by surfactant accumulation in alveolar airspaces. In PAP individuals, both PPAR-γ mRNA and the PPAR-γ-regulated lipid scavenger receptor, CD36 were reduced in AMs when compared to healthy subjects. PPAR-γ and CD36 deficiency in PAP was cell type-specific in the lung (i.e. found in AM and not in bronchial epithelial cells). In vitro and in vivo GM-CSF treatment of PAP patients fully restored PPAR-γ to healthy control levels [46].
As for asthma patients, cell-proliferating lesions obstruct the vessel lumen and promote pulmonary arterial pressure and reduced blood flow in COPD patients [16,47] (Table 1). In asthma, the eosinophil survival indicator, GM-CSF, is prominent in bronchoalveolar lavage fluid, serum and lung tissue. On the contrary, COPD is characterised by neutrophilia [48]. It has been confirmed that both GM-CSF and the related survival factor, G-CSF are involved in the survival of the neutrophils. Consequently, these factors may aggravate and extend the inflammatory response in neutrophil-related inflammatory lung diseases such as COPD [49,50].
Activation of PPAR-γ by 15D-PGJ2 and ciglitazone induced apoptosis and impeded serum-induced cell growth more effectively than the steroid dexamethasone in HASM. Moreover, PPAR-γ ligands and dexamethasone hampered the IL-1β-induced release of GM-CSF. However, PPAR-γ ligands, but not dexamethasone, similarly deterred G-CSF release. The above actions of 15D-PGJ2 were not dependent on the activation of a traditional cell surface prostanoid receptor. Agents that obstruct proliferation of HASM cells, as well as CSF release, would represent potential new therapies to treat COPD and steroid-insensitive asthma [16] (Table 1).
Conclusion
It appears that chronic lung disorders are not confined to a particular race, sex or age. Studies delving into respiratory diseases have reached a crucial point and the increasing incidence and potential fatality of these debilitating diseases has emphasised the urgent quest for novel therapeutic avenues vital to the control and ultimate elimination of such disease. The role of PPAR-γ in regulating adipocyte differentiation and glucose homeostasis has been established and, consequently, further research has uncovered its involvement in inflammatory events of cardiac and, more recently, airway diseases. Antagonism of the pro-inflammatory pathways in respiratory diseases is the likely mechanism of action of the PPARs and their respective agonists. Research on the physiological role of PPAR-γ in the lung is still in its infancy, however, continued advancement in this field will unravel the co-existence and interactions of the PPAR-γ gene and related ligands such as 15D-PGJ2 and TZDs in the prevention or treatment of inflammatory respiratory diseases. It is unlikely that the current PPAR-γ agonists will be used as a monotherapy in airway diseases such as asthma and lung cancer. However, with improved comprehension of the full biological and physiological role of PPAR-γ in these diseases, novel and more potent agonists could be designed to include effective administration of anti-inflammatory therapies with minimal side effects. This could also extend to tackling more elusive or less common lung disorders such as cystic fibrosis, PAP and COPD.
It is unanimously agreed that the PPAR-γ anti-inflammatory pathways must be correctly identified for the particular disease state, as this will have important implications for the type of treatment and its effective administration. This would be determined by factors such as the receptor's presence in the particular sections of the lung (lung tissue compartment versus airway lumen), its expression in specific lung cell types and its influence on pro-inflammatory cytokines, enzymes, proteins, fatty acid metabolism and subsequent pathways. Therefore, it is anticipated that PPAR-γ expression will become a potential indicator of many airway inflammatory diseases leading to a possible prevention or treatment therapeutic application.
Abbreviations
Peroxisome proliferator-activated receptors (PPAR); 15-deoxy-Δ12,14-prostaglandin J2 (15D-PGJ2); thiazolidinediones (TZD); chronic obstructive pulmonary disease (COPD); PPAR-response element (PPRE); direct repeat (DR); 9-cis-retinoic acid receptors (RXR); alveolar macrophages (AM); human airway smooth muscle (HASM); tumour necrosis factor (TNF); interferon (IFN); interleukin-4 (IL-4); 12/15-lipoxygenase (12/15-LO); nitric oxide synthases (NOS); monocyte chemoattractant protein (MCP); alveolar type II (AT II); surfactant protein, (SP); CCAAT/enhancer-binding proteins (C/EBP); cytosolic phospholipase A2 (cPLA2); subepithelial membrane (SBM); forced expiratory volume (FEV1); ovalbumin (OVA); antigen-induced airway hyperresponsiveness (AHR); transforming growth factor (TGF); Immunoglubulin E and Immunoglobulin G1 (IgE and IgG1); dendritic cells (DC); mediastinal lymph nodes (MLN); interleukin-5 (IL-5); matrix metalloproteinase (MMP); nuclear factor (NF); lipopolysaccharide (LPS); lymphomononuclear (LMN); granulocyte-colony-stimulating factor (G-CSF); keratinocyte-derived chemokine (KC); non-small cell lung cancer (NSCLC); large cell carcinoma (LCC); cyclooxygenase (COX); prostaglandin E2 (PGE2); G protein-coupled PGE receptor (EP); non-steroidal anti-inflammatory drugs (NSAIDs); ystic fibrosis transmembrane regulator protein (CFTR); pulmonary alveolar proteinosis (PAP)
==== Refs
Serhan CN Devchand PR Novel antiinflammatory targets for asthma. A role for PPARgamma? Am J Respir Cell Mol Biol 2001 24 658 661 11415929
Birrell BA Patel HJ McCluskie K Wong S Leonard T Yacoub MH Belvisi MG PPAR-gamma agonists as therapy for diseases involving airway neutrophilia Eur Respir J 2004 24 18 23 15293600 10.1183/09031936.04.00098303
Asada K Sasaki S Suda T Chida K Nakamura H Antiinflammatory roles of peroxisome proliferator-activated receptor gamma in human alveolar macrophages Am J Respir Crit Care Med 2004 169 195 200 14563653 10.1164/rccm.200207-740OC
Benayoun L Letuve S Druilhe A Boczkowski J Dombret MC Mechighel P Megret J Leseche G Aubier M Pretolani M Regulation of peroxisome proliferator-activated receptor gamma expression in human asthmatic airways: relationship with proliferation, apoptosis, and airway remodeling Am J Respir Crit Care Med 2001 164 1487 1494 11704601
Honda K Marquillies P Capron M Dombrowicz D Peroxisome proliferator-activated receptor gamma is expressed in airways and inhibits features of airway remodeling in a mouse asthma model J Allergy Clin Immunol 2004 113 882 888 15131570 10.1016/j.jaci.2004.02.036
Kota BP Huang TH Roufogalis BD An overview on biological mechanisms of PPARs Pharmacol Res 2004
Mueller C Weaver V Vanden Heuvel JP August A Cantorna MT Peroxisome proliferator-activated receptor gamma ligands attenuate immunological symptoms of experimental allergic asthma Arch Biochem Biophys 2003 418 186 196 14522590 10.1016/j.abb.2003.08.006
Buchan KW Hassall DG PPAR Agonists as Direct Modulators of the Vessel Wall in Cardiovascular Disease Med Res Rev 2000 20 350 366 10934348 10.1002/1098-1128(200009)20:5<350::AID-MED2>3.0.CO;2-D
Ferre P The biology of peroxisome proliferator-activated receptors: relationship with lipid metabolism and insulin sensitivity Diabetes 2004 53 Suppl 1 S43 S50 14749265
Issemann I Green S Activation of a member of the steroid hormone receptor superfamily by peroxisome proliferators Nature 1990 347 645 650 2129546 10.1038/347645a0
Fajas L Auboeuf D Raspe E Schoonjans K Lefebvre AM Saladin R Najib J Laville M Fruchart JC Deeb S Vidal-Puig A Flier J Briggs MR Staels B Vidal H Auwerx J The organization, promoter analysis, and expression of the human PPARgamma gene J Biol Chem 1997 272 18779 18789 9228052 10.1074/jbc.272.30.18779
Fajas L Fruchart JC Auwerx J PPARgamma3 mRNA: a distinct PPARgamma mRNA subtype transcribed from an independent promoter FEBS Lett 1998 438 55 60 9821958 10.1016/S0014-5793(98)01273-3
Sundvold H Brzozowska A Lien S Characterisation of bovine peroxisome proliferator-activated receptors gamma 1 and gamma 2: genetic mapping and differential expression of the two isoforms Biochem Biophys Res Commun 1997 239 857 861 9367859 10.1006/bbrc.1997.7564
Ueki S Matsuwaki Y Kayaba H Oyamada H Kanda A Usami A Saito N Chihara J Peroxisome proliferator-activated receptor gamma regulates eosinophil functions: a new therapeutic target for allergic airway inflammation Int Arch Allergy Immunol 2004 134 30 36 15166481 10.1159/000077790
Hetzel M Walcher D Grub M Bach H Hombach V Marx N Inhibition of MMP-9 expression by PPARgamma activators in human bronchial epithelial cells Thorax 2003 58 778 783 12947137 10.1136/thorax.58.9.778
Patel HJ Belvisi MG Bishop-Bailey D Yacoub MH Mitchell JA Activation of peroxisome proliferator-activated receptors in human airway smooth muscle cells has a superior anti-inflammatory profile to corticosteroids: relevance for chronic obstructive pulmonary disease therapy J Immunol 2003 170 2663 2669 12594295
Wang AC Dai X Luu B Conrad DJ Peroxisome proliferator-activated receptor-gamma regulates airway epithelial cell activation Am J Respir Cell Mol Biol 2001 24 688 693 11415933
Huang JT Welch JS Ricote M Binder CJ Willson TM Kelly C Witztum JL Funk CD Conrad D Glass CK Interleukin-4-dependent production of PPAR-gamma ligands in macrophages by 12/15-lipoxygenase Nature 1999 400 378 382 10432118 10.1038/22572
Folkard SG Westwick J Millar AB Production of interleukin-8, RANTES and MCP-1 in intrinsic and extrinsic asthmatics Eur Respir J 1997 10 2097 2104 9311510 10.1183/09031936.97.10092097
Momoi A Murao K Imachi H Ishida T Cao WM Sato M Takahara J Inhibition of monocyte chemoattractant protein-1 expression in cytokine-treated human lung epithelial cells by thiazolidinedione Chest 2001 120 1293 1300 11591574 10.1378/chest.120.4.1293
Yang L Yan D Yan C Du H Peroxisome proliferator-activated receptor gamma and ligands inhibit surfactant protein B gene expression in the lung J Biol Chem 2003 278 36841 36847 12829715 10.1074/jbc.M304156200
Du H Witte DP Grabowski GA Tissue and cellular specific expression of murine lysosomal acid lipase mRNA and protein J Lipid Res 1996 37 937 949 8725147
Rosen ED Spiegelman BM Molecular regulation of adipogenesis Annu Rev Cell Dev Biol 2000 16 145 171 11031233 10.1146/annurev.cellbio.16.1.145
Barlier-Mur AM Chailley-Heu B Pinteur C Henrion-Caude A Delacourt C Bourbon JR Maturational factors modulate transcription factors CCAAT/enhancer-binding proteins alpha, beta, delta, and peroxisome proliferator-activated receptor-gamma in fetal rat lung epithelial cells Am J Respir Cell Mol Biol 2003 29 620 626 14568882 10.1165/rcmb.4912
Lekstrom-Himes J Xanthopoulos KG Biological role of the CCAAT/enhancer-binding protein family of transcription factors J Biol Chem 1998 273 28545 28548 9786841 10.1074/jbc.273.44.28545
Sugahara K Iyama KI Kimura T Sano K Darlington GJ Akiba T Takiguchi M Mice lacking CCAAt/enhancer-binding protein-alpha show hyperproliferation of alveolar type II cells and increased surfactant protein mRNAs Cell Tissue Res 2001 306 57 63 11683182 10.1007/s004410100420
Nagase T Uozumi N Ishii S Kume K Izumi T Ouchi Y Shimizu T Acute lung injury by sepsis and acid aspiration: a key role for cytosolic phospholipase A2 Nat Immunol 2000 1 42 46 10881173 10.1038/76897
Pawliczak R Han C Huang XL Demetris AJ Shelhamer JH Wu T 85-kDa cytosolic phospholipase A2 mediates peroxisome proliferator-activated receptor gamma activation in human lung epithelial cells J Biol Chem 2002 277 33153 33163 12077117 10.1074/jbc.M200246200
Woerly G Honda K Loyens M Papin JP Auwerx J Staels B Capron M Dombrowicz D Peroxisome proliferator-activated receptors alpha and gamma down-regulate allergic inflammation and eosinophil activation J Exp Med 2003 198 411 421 12900517 10.1084/jem.20021384
Hammad H de Heer HJ Soullie T Angeli V Trottein F Hoogsteden HC Lambrecht BN Activation of peroxisome proliferator-activated receptor-gamma in dendritic cells inhibits the development of eosinophilic airway inflammation in a mouse model of asthma Am J Pathol 2004 164 263 271 14695339
Trifilieff A Bench A Hanley M Bayley D Campbell E Whittaker P PPAR-alpha and -gamma but not -delta agonists inhibit airway inflammation in a murine model of asthma: in vitro evidence for an NF-kappaB-independent effect Br J Pharmacol 2003 139 163 71 12746235 10.1038/sj.bjp.0705232
DeCicco KL Tanaka T Andreola F De Luca LM The effect of thalidomide on non-small cell lung cancer (NSCLC) cell lines: possible involvement in the PPAR{gamma} pathway Carcinogenesis 2004 25 1805 1812 15205358 10.1093/carcin/bgh210
Sasaki H Tanahashi M Yukiue H Moiriyama S Kobayashi Y Nakashima Y Kaji M Kiriyama M Fukai I Yamakawa Y Fujii Y Decreased perioxisome proliferator-activated receptor gamma gene expression was correlated with poor prognosis in patients with lung cancer Lung Cancer 2002 36 71 76 11891036 10.1016/S0169-5002(01)00449-4
Keshamouni VG Reddy RC Arenberg DA Joel B Thannickal VJ Kalemkerian GP Standiford TJ Peroxisome proliferator-activated receptor-gamma activation inhibits tumor progression in non-small-cell lung cancer Oncogene 2004 23 100 108 14712215 10.1038/sj.onc.1206885
Tsubouchi Y Sano H Kawahito Y Mukai S Yamada R Kohno M Inoue K Hla T Kondo M Inhibition of human lung cancer cell growth by the peroxisome proliferator-activated receptor-gamma agonists through induction of apoptosis Biochem Biophys Res Commun 2000 270 400 405 10753637 10.1006/bbrc.2000.2436
Inoue K Kawahito Y Tsubouchi Y Yamada R Kohno M Hosokawa Y Katoh D Bishop-Bailey D Hla T Sano H Expression of peroxisome proliferator-activated receptor (PPAR)-gamma in human lung cancer Anticancer Res 2001 21 2471 2476 11724309
Chang TH Szabo E Enhanced growth inhibition by combination differentiation therapy with ligands of peroxisome proliferator-activated receptor-gamma and inhibitors of histone deacetylase in adenocarcinoma of the lung Clin Cancer Res 2002 8 1206 1212 11948134
Zhang M Zou P Bai M Jin Y Tao X Peroxisome proliferator-activated receptor-gamma activated by ligands can inhibit human lung cancer cell growth through induction of apoptosis J Huazhong Univ Sci Technolog Med Sci 2003 23 138 140 12973930
Stolina M Sharma S Lin Y Dohadwala M Gardner B Luo J Zhu L Kronenberg M Miller PW Portanova J Lee JC Dubinett SM Specific inhibition of cyclooxygenase 2 restores antitumor reactivity by altering the balance of IL-10 and IL-12 synthesis J Immunol 2000 164 361 370 10605031
Han S Sidell N Roser-Page S Roman J Fibronectin stimulates human lung carcinoma cell growth by inducing cyclooxygenase-2 (COX-2) expression Int J Cancer 2004 111 322 331 15221958 10.1002/ijc.20281
Han S Roman J Suppression of prostaglandin E2 receptor subtype EP2 by PPARgamma ligands inhibits human lung carcinoma cell growth Biochem Biophys Res Commun 2004 314 1093 1099 14751245 10.1016/j.bbrc.2004.01.007
Wick M Hurteau G Dessev C Chan D Geraci MW Winn RA Heasley LE Nemenoff RA Peroxisome proliferator-activated receptor-gamma is a target of nonsteroidal anti-inflammatory drugs mediating cyclooxygenase-independent inhibition of lung cancer cell growth Mol Pharmacol 2002 62 1207 1214 12391285 10.1124/mol.62.5.1207
Shaik MS Chatterjee A Singh M Effect of a selective cyclooxygenase-2 inhibitor, nimesulide, on the growth of lung tumors and their expression of cyclooxygenase-2 and peroxisome proliferator-activated receptor-gamma Clin Cancer Res 2004 10 1521 1529 14977856
Pang L Nie M Corbett L Knox AJ Cyclooxygenase-2 expression by nonsteroidal anti-inflammatory drugs in human airway smooth muscle cells: role of peroxisome proliferator-activated receptors J Immunol 2003 170 1043 1051 12517972
Ollero M Junaidi O Zaman MM Tzameli I Ferrando AA Andersson C Blanco PG Bialecki E Freedman SD Decreased expression of peroxisome proliferator activated receptor gamma in cftr-/- mice J Cell Physiol 2004 200 235 244 15174093 10.1002/jcp.20020
Bonfield TL Farver CF Barna BP Malur A Abraham S Raychaudhuri B Kavuru MS Thomassen MJ Peroxisome proliferator-activated receptor-gamma is deficient in alveolar macrophages from patients with alveolar proteinosis Am J Respir Cell Mol Biol 2003 29 6 677 82 12805087 10.1165/rcmb.2003-0148OC
Ameshima S Golpon H Cool CD Chan D Vandivier RW Gardai SJ Wick M Nemenoff RA Geraci MW Voelkel NF Peroxisome proliferator-activated receptor gamma (PPARgamma) expression is decreased in pulmonary hypertension and affects endothelial cell growth Circ Res 2003 92 1162 1169 12714563 10.1161/01.RES.0000073585.50092.14
Sullivan S Broide DH Compartmentalization of eosinophil granulocyte-macrophage colony-stimulating factor expression in patients with asthma J Allergy Clin Immunol 1996 97 966 976 8655893
Aggarwal A Baker CS Evans TW Haslam PL G-CSF and IL-8 but not GM-CSF correlate with severity of pulmonary neutrophilia in acute respiratory distress syndrome Eur Respir J 2000 15 895 901 10853855 10.1034/j.1399-3003.2000.15e14.x
Stanford SJ Pepper JR Burke-Gaffney A Mitchell JA Cytokine-activated human vascular smooth muscle delays apoptosis of neutrophils: relevance of interactions between cyclo-oxygenase-2 and colony-stimulating factors FASEB J 2001 15 1813 1815 11481239
|
16153299
|
PMC1242255
|
CC BY
|
2021-01-04 16:23:26
|
no
|
Respir Res. 2005 Sep 9; 6(1):102
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-102
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-1061616476110.1186/1465-9921-6-106ResearchEffects of nano particles on antigen-related airway inflammation in mice Inoue Ken-ichiro [email protected] Hirohisa [email protected] Rie [email protected] Miho [email protected] Takamichi [email protected] Kaori [email protected] Toshikazu [email protected] Inhalation Toxicology and Pathophysiology Research Team, National Institute for Environmental Studies, Tsukuba, Ibaraki, Japan2 Inflammation and Immunology, Graduate School of Medical Science, Kyoto Prefectural University of Medicine, Kyoto, Japan3 Department of Health Science, Oita University of Nursing and Health Science, Oita, Japan2005 16 9 2005 6 1 106 106 28 1 2005 16 9 2005 Copyright © 2005 Inoue et al; licensee BioMed Central Ltd.2005Inoue et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Particulate matter (PM) can exacerbate allergic airway diseases. Although health effects of PM with a diameter of less than 100 nm have been focused, few studies have elucidated the correlation between the sizes of particles and aggravation of allergic diseases. We investigated the effects of nano particles with a diameter of 14 nm or 56 nm on antigen-related airway inflammation.
Methods
ICR mice were divided into six experimental groups. Vehicle, two sizes of carbon nano particles, ovalbumin (OVA), and OVA + nano particles were administered intratracheally. Cellular profile of bronchoalveolar lavage (BAL) fluid, lung histology, expression of cytokines, chemokines, and 8-hydroxy-2'-deoxyguanosine (8-OHdG), and immunoglobulin production were studied.
Results
Nano particles with a diameter of 14 nm or 56 nm aggravated antigen-related airway inflammation characterized by infiltration of eosinophils, neutrophils, and mononuclear cells, and by an increase in the number of goblet cells in the bronchial epithelium. Nano particles with antigen increased protein levels of interleukin (IL)-5, IL-6, and IL-13, eotaxin, macrophage chemoattractant protein (MCP)-1, and regulated on activation and normal T cells expressed and secreted (RANTES) in the lung as compared with antigen alone. The formation of 8-OHdG, a proper marker of oxidative stress, was moderately induced by nano particles or antigen alone, and was markedly enhanced by antigen plus nano particles as compared with nano particles or antigen alone. The aggravation was more prominent with 14 nm of nano particles than with 56 nm of particles in overall trend. Particles with a diameter of 14 nm exhibited adjuvant activity for total IgE and antigen-specific IgG1 and IgE.
Conclusion
Nano particles can aggravate antigen-related airway inflammation and immunoglobulin production, which is more prominent with smaller particles. The enhancement may be mediated, at least partly, by the increased local expression of IL-5 and eotaxin, and also by the modulated expression of IL-13, RANTES, MCP-1, and IL-6.
==== Body
Introduction
Previous epidemiological studies have indicated that long-term exposure to ambient particulate matter (PM) is linked to increases in mortality and morbidity related to respiratory diseases [1,2]. The concentration of PM of mass median aerodynamic diameter (a density-dependent unit of measure used to describe the diameter of the particle) < or 10 μm (PM10) is related to daily hospital admissions for asthma, acute and chronic bronchiolitis, and lower respiratory tract infections [3]. PM of mass median aerodynamic diameter < or 2.5 μm (PM2.5) are more closely associated with both acute and chronic respiratory effects and subsequent mortality than PM10 [4]. Our laboratory has researched health effects of diesel exhaust particles (DEP), main constituents of PM2.5 in urban areas, especially in vivo. We have reported that DEP exacerbate allergic asthma [5] and acute lung injury related to bacterial infection in murine models [6].
Recently, nano particles, particles less than 0.1 μm in mass median aerodynamic diameter, have been implicated to affect cardiopulmonary systems [4,7]. Indeed, two in vivo studies have demonstrated that nano particles induce prominent airway inflammation as compared with larger particles [8,9]. Nano particles which have a larger surface area than the particles with larger size are able to penetrate deeply into the respiratory tract and cause a greater inflammatory response [10,11].
Bronchial asthma has been recognized as chronic airway inflammation that is characterized by an increase in the number of activated lymphocytes and eosinophils. A number of studies have shown that various particles including carbon black (CB) can enhance allergic sensitization [12-14]. CB has been demonstrated to enhance proliferation of antibody forming cells and both IgE and IgG levels [15,16]. Ultrafine particles (PM and CB) reportedly exaggerate allergic airway inflammation in vivo [17,18]. However, all the studies have not described the size of particles they used. Therefore, no research has been addressed the size effects of particles or nano particles on allergic airway inflammation in vivo.
The aim of the present study was to elucidate the effects of two sizes of carbon nano particles (14 nm or 56 nm) on allergic airway inflammation, local expression of cytokines, chemokines, and 8-hydroxy-2'-deoxyguanosine (8-OHdG), and production of total IgE and antigen-specific IgG1, IgG2a, and IgE.
Materials and methods
Animals
Male ICR mice 6 to 7 wk of age and weighing 29 to 33 g (Japan Clea Co., Tokyo, Japan) were used in all experiments. They were fed a commercial diet (Japan Clea Co.) and given water ad libitum. Mice were housed in an animal facility that was maintained at 24 to 26°C with 55 to 75% humidity and a 12-h light/dark cycle.
Study protocol
Mice were divided into six experimental groups (Fig. 1). The vehicle group received phosphate-buffered saline (PBS) at pH 7.4 (Nissui Pharmaceutical Co., Tokyo, Japan) containing 0.05% Tween 80 (Nakalai Tesque, Kyoto, Japan) once a week for 6 wk. The ovalbumin (OVA) group received 1 μg of OVA (Sigma Chemical, St. Louis, MO) dissolved in the same vehicle every 2 wk for 6 wk. The nano particle groups received 50 μg of nano particles (14 nm: PrinteX 90 or 56 nm: PrinteX 25, degussa, Dusseldorf, Germany) suspended in the same vehicle every week for 6 wk. The OVA + nano particle groups received the combined treatment in the same protocol as the OVA and the nano particle groups, respectively. The surface area of the 14 nm nano particles was 300 m2/g and that of 56 nm nano particles was 45 m2/g. The size of each particle was quantified by JEM-2010 transmission electron microscope (TEM; JEOL, Tokyo, Japan). Nano particles were autoclaved at 250°C for 2 h before use. The suspension was sonicated for 3 min using an Ultrasonic disrupter (UD-201; Tomy Seiko, Tokyo, Japan). In each group, vehicle, OVA, nano particles, or OVA + nano particles was dissolved in 0.1 ml aliquots, and inoculated by the intratracheal route through a polyethylene tube under anesthesia with 4% halothane (Hoechst, Japan, Tokyo, Japan). The animals were studied 24 h after the last intratracheal administration, with lung histology, bronchoalveolar lavage (BAL), protein levels of cytokines and chemokines in the lung tissue supernatants, immunohistochemistry for 8-OHdG, and with Igs. The studies adhered to the National Institutes of Health guidelines for the experimental use of animals. All animal studies were approved by the Institutional Review Board.
Figure 1 Study Protocol.
Blood retrieval and analysis
Mice were anesthetized with diethyl ether. The chest and abdominal walls were opened, and blood was retrieved by cardiac puncture. Serum was prepared and frozen at – 80°C until assayed for total IgE and antigen-specific IgG1, IgG2a, and IgE.
Histologic evaluation
After exsanguinations, the lungs were fixed by intratracheal instillation with 10% neutral phosphate-buffered formalin at a pressure of 20 cm H2O for at least 72 h. Slices 2 to 3 mm thick of all pulmonary lobes were embedded in paraffin. Sections 3 μm thick were stained with Diff-Quik (International Reagents Co., Kobe, Japan) or periodic acid-Schiff (PAS) and examined by two of us (HT and KI) in a blind fashion.
Morphometric analysis for numbers of eosinophils, neutrophils, mononuclear cells, and goblet cells around the airways
Sections were stained with Diff-Quik to quantitate the numbers of infiltrated eosinophils, neutrophils, and mononuclear cells. The length of the basement membrane of the airways was measured by videomicrometer (Olympus, Tokyo, Japan) in each sample slide. The number of eosinophils, neutrophils, and mononuclear cells around the airways were counted with a micrometer under oil immersion. Results were expressed as the number of inflammatory cells per millimeter of basement membrane as described previously [5].
To quantitate goblet cells, sections were stained with PAS. The number of goblet cells in the bronchial epithelium was counted by micrometer. Results were expressed as the number of goblet cells per millimeter of basement membrane as described previously [5].
BAL
The trachea was cannulated after the collection of blood. The lungs were lavaged with 1.2 ml of sterile saline at 37°C, instilled bilaterally by syringe. The lavage fluid was harvested by gentle aspiration. This procedure was conducted two more times. The average volume retrieved was 90 % of the 3.6 ml that was instilled; the amounts did not differ by treatment. The fluid collections were combined and cooled to 4°C. The lavage fluid was centrifuged at 300 g for 10 min, and the total cell count was determined on a fresh fluid specimen using a hemocytometer. Differential cell counts were assessed on cytologic preparations. Slides were prepared using an Autosmear (Sakura Seiki Co., Tokyo, Japan) and were stained with Diff-Quik (International reagents Co.). A total of 500 cells were counted under oil immersion microscopy.
Quantitation of cytokine and chemokine protein levels in the lung tissue supernatants
In a separate series of experiments, the animals were exsanguinated and the lungs were subsequently homogenized with 10 mM potassium phosphate buffer (pH 7.4) containing 0.1 mM ethylenediaminetetraacetic acid (Sigma, St Louis MO), 0.1 mM phenylmethanesulphonyl fluoride (Nacalai Tesque, Kyoto, Japan), 1 μM pepstatin A (Peptide Institute, Osaka, Japan) and 2 μM leupeptin (Peptide Institute) as described previously [5]. The homogenates were then centrifuged at 105,000 g for 1 h. The supernatants were stored at -80°C. Enzyme-linked immunosobent assays (ELISA) for interleukin (IL)-4 (Amersham, Buckinghamshire, UK), IL-5 (Endogen, Cambridge, MA), IL-6 (Biosource, Nivelles, Belgium), IL-13, eotaxin, macrophage chemoattractant protein (MCP)-1, and regulated on activation and normal T cells expressed and secreted (RANTES: R&D systems, Minneapolis, MN) in the lung tissue supernatants were conducted using matching antibody pairs according to the manufacture's instruction. The second antibodies were conjugated to horseradish peroxidase. Subtractive reading of 550 nm from the reading at 450 nm were converted to pg/ml using values obtained from standard curves generated with varying concentrations of recombinant IL-4, IL-5, IL-6, IL-13, eotaxin, MCP-1, and RANTES, with limits of detection of 5 pg/ml, 5 pg/ml, 3 pg/ml, 1.5 pg/ml, 2 pg/ml 2 pg/ml, and 2 pg/ml, respectively.
Immunohistochemistry
The production of 8-OHdG in the lung was detected by immunohistochemical analysis (n = 8 in each group) using anti-8-OHdG polyclonal antibody (Japan Institute for the Control of Aging, Shizuoka, Japan) as described previously [19,20]. Deparaffinized slides were blocked with 10% goat serum for 1 h. After blocking, anti-8-OHdG antibody (0.5 μg/ml) was incubated with the sections for 1 h at room temperature, followed by the incubation of a biotinylated secondary antibody and streptavidin-peroxidase conjugate. Then, the slides were incubated with 3-amino, 9-ethyl-carbazole chromogen, and counterstained with hematoxylin in AutoProbe III kit (Biomeda, Foster City, CA, USA). For each of the lung specimens, the extent and intensity of staining with anti-8-OHdG antibodies were graded on a scale of 0–4+ by two blinded observers on two separate occasions using coded slides as previously described [21]. A 4+ grade implies maximally intense staining, whereas 0 implies no staining.
Antigen-specific IgG determination
Antigen-specific IgG1 or IgG2a antibodies were measured by ELISA with solid-phase antigen [5,22]. In brief, microplate wells (Dynatech, Chantilly, VA) were coated with OVA overnight at 4°C and then incubated at room temperature for 1 h with PBS containing 1% bovine serum albumin (BSA; Sigma) containing 0.01% thimerosal (Nakalai Tesque). After washing, diluted samples were introduced to the microplate and incubated at room temperature for 1 h. After another washing, the wells were incubated at room temperature for 1 h with biotinylated rabbit anti-mouse IgG1 or IgG2a (Zymed Laboratories, San Francisco, CA). After yet another washing, the wells were incubated with horseradish-peroxidase-conjugated streptavidin (Sigma) at room temperature for 1 h. The wells were then washed and incubated with o-phenylenediamine and H2O2 in dark at room temperature for 30 min. The enzyme reaction was stopped with 4 N H2SO4. Absorbance was read at 492 nm. Each plate incubated a previously screened standard plasma that contained a high titer of anti-OVA antibodies. The results were expressed in titers, calculated based on the titers of the standard plasma. Cut off values for antibody-positive plasma were set to hold as the mean value of absorbance of preimmune plasma.
Total IgE and antigen-specific IgE determination
Antigen-specific IgE antibody was measured by IgE-capture ELISA [5,22]. In brief, microplate wells were coated with a rat anti-mouse IgE monoclonal antibody (Yamasa Syoyu Co., Chiba, Japan) at 37°C for 3 h and then incubated at 37°C for 1 h with 1% BSA-PBS and 0.01% thimerosal. After washing with PBS containing 0.05% Tween 20 (PBST; Nacalai Tesque), diluted samples were introduced to the microplate and incubated overnight at 4°C. After washing with PBST, biotinylated OVA was added to each well and incubated for 1 h at room temperature with β-D-galactosidase-conjugated streptavidin (Zymed). After the final washing, the wells were incubated with 4-methylumbelliferyl-β-galactoside (Sigma) as the enzyme substrate at 37°C for 2 h. The enzyme reaction was stopped with 0.1 M glycine-NaOH (pH, 10.3). The fluorescene intensity was read by a microplate fluorescene reader (Fluoroskan Flow Laboratories, Costa Mesa, CA). Each plate included a previously screened standard plasma that contained a high titer of anti-OVA antibodies. The results were expressed in titers, calculated based on the titers of the standard plasma. Cut off values for antibody-positive plasma were set two hold as mean fluorescene units of preimmune plasma. Total IgE was measured by capture ELISA in a manner similar to the detection of antigen-specific IgE. A biotinylated rat anti-mouse IgE (BD Biosciences Pharmingen, San Diego, CA) was used to detect captured IgE in place of biotinylated OVA. A450 readings of the samples were converted to nanograms per milliliter using a standard curve generated with double dilutions of mouse IgE κ isotype standard (BD Biosciences Pharmingen).
Statistical analysis
Data were reported as mean ± SEM. Differences in the numbers of infiltrated inflammatory cells and goblet cells, cytokine protein levels, and immunogloblin concentrations and titers between groups were determined using analysis of variance (Stat view version 4.0; Abacus Concepts, Inc., Berkeley, CA) as described previously [5]. If differences between groups were significant (P < 0.05), Fisher's protected least significant difference test was used to distinguish between pairs of groups.
Results
Effects of nano particles on antigen-related airway inflammation
To evaluate the effect of nano particles on antigen-related airway inflammation, we investigated the cellular profile of BAL fluid and lung histology.
The numbers of total cells and macrophages were significantly greater in the nano particle, OVA, and OVA + nano particle groups than in the vehicle group (P < 0.01: Table 1). Furthermore, the numbers were significantly greater in the OVA + 14 nm nano particle group than in the OVA group or the 14 nm particle group (P < 0.01: Table 1). Although the numbers were greater in the OVA + 56 nm nano particle group than in the OVA group or the 56 nm nano particle group, the difference did not achieve significance. OVA challenge increased the number of eosinophils as compared with vehicle challenge without significance. The numbesr of eosinophils were greater in the OVA + nano particle groups than in the vehicle group (P < 0.05 for OVA + 14 nm nano particle, N. S. for OVA + 56 nm nano particle). The number was significantly greater in the OVA + 14 nm nano particle group than in the OVA group (P < 0.01) or 14 nm nano particle group (P < 0.05). The number was also greater in the OVA + 56 nm nano particle group than in the OVA group or 56 nm nano particle group, but the difference did not achieve significance. Challenge with nano particles significantly elevated the numbers of neutrophils as compared with vehicle challenge (P < 0.01 for 14 nm, P < 0.05 for 56 nm). OVA also elevated the number without significance as compared with vehicle challenge. The number was significantly greater in the OVA + 14 nm nano particle group than in the 14 nm nano particle group (P < 0.05) or the OVA group (P < 0.01). The number was also greater in the OVA + 56 nm nano particle group than in the nano particle group or the OVA group, the difference did not reach significance. Challenge with nano particles elevated the numbers of mononuclear cells as compared with vehicle challenge (P < 0.05 for 14 nm, N. S. for 56 nm). OVA also elevated the number of mononuclear cells without significance as compared with vehicle challenge. The number was significantly greater in the OVA + 14 nm nano particle group than in the vehicle (P < 0.01) or the OVA group (P < 0.05). The number was greater in the OVA + 56 nm nano particle group than in the OVA group, but difference did not achieve significance. There were no significant differences between the nano particle groups and OVA + nano particle groups.
Table 1 Cellular profile in bronchoalveolar lavage fluid.
Group Animals (n) Total Cells (× 104/total BAL) Macrophages (× 104/total BAL) Eosinophils (× 104/total BAL) Neutrophils (× 104/total BAL) Mononuclear Cells (× 104/total BAL)
vehicle 16 36.88 ± 3.56 36.74 ± 3.53 0 ± 0 0.12 ± 0.05 0.015 ± 0.01
14 nm 13 111.69 ± 9.27** 83.79 ± 6.03** 0.332 ± 0.176 27.04 ± 4.98** 0.491 ± 0.201*
56 nm 14 97.36 ± 16.06** 88.64 ± 15.34** 0.331 ± 0.177 8.09 ± 2.49* 0.265 ± 0.093
OVA 16 85.06 ± 12.63** 81.91 ± 12.4** 0.705 ± 0.255 2.2 ± 0.62 0.121 ± 0.059
OVA + 14 nm 16 193.69 ± 18.33** ## $$ 141.86 ± 14.97** ## $ 13.667 ± 4.731** ## $ 36.9 ± 3.67** ## $ 0.878 ± 0.232** #
OVA + 56 nm 17 102.65 ± 11.64** 90.7 ± 10.12** 3.984 ± 2.669 7.79 ± 2.29* 0.204 ± 0.073
Six groups of mice were intratracheally administered with vehicle, ovalbumin (OVA), nano particles, or combinations of OVA and nano particles for 6 wk. Bronchoalveolar lavage (BAL) was conducted 24 h after the last intratracheal instillation. Total cell counts were determined on fresh BAL fluid, and differential cell counts were assessed with Diff-Quik-staining. Results are presented as mean ± SEM. *P < 0.05 versus vehicle, **P < 0.01 versus vehicle, #P < 0.05 versus OVA, ##P < 0.01 versus OVA. $P < 0.05 versus nano particles. $$P < 0.01 versus nano particles.
The magnitude and cellular profiles of airway inflammation were also evaluated in lung specimens stained with Diff-Quik. Intratracheal instillation of nano particles provided diffuse deposition of the particles into the bilateral lungs, including the bronchi and alveolar spaces. The particles were occasionally present within the subepithelial neutrophils and alveolar macrophages. The combined instillation of OVA + 14 nm nano particles for 6 wk led to a marked infiltration of eosinophils and mononuclear cells around the bronchi and bronchioles. OVA + 56 nm nano particles also induced severe airway inflammation, but the severity was less than that of OVA + 14 nm nano particles. Either OVA or nano particles alone resulted in slight recruitment of eosinophils and neutrophils. Vehicle administration caused little infiltration of inflammatory cells.
To quantitate the infiltration of inflammatory cells around the airways, we expressed the number of these cells per length of basement membrane of the airways (Table 2). The number of eosinophils was greater in the OVA group than in the vehicle group without significance. The number of eosinophils was significantly greater in the OVA + 14 nm nano particle group than in the vehicle, the 14 nm nano particle, or the OVA group (P < 0.01). The number was greater also in the OVA + 56 nm nano particle group than in the OVA group or the 56 nm nano particle group, but the difference did not achieve significance. OVA increased the number of neutrophils as compared with vehicle challenge without significance. In the presence of OVA, nano particles with a diameter of 14 nm significantly increased the number as compared with vehicle or OVA challenge (P < 0.01 for vehicle, P < 0.05 for OVA)). In the presence of OVA, nano particles with a diameter of 56 nm increased the number as compared with vehicle (P < 0.05) or OVA (N. S.). Challenge with nano particles increased the numbers as compared with vehicle challenge (P < 0.01 for 14 nm nano particle, N. S. for 56 nm nano particle). There were no significant differences between the OVA + nano particle groups and the nano particle groups. The number of mononuclear cells was significantly greater in the OVA group than in the vehicle group (P < 0.05). 14 nm nano particles significantly increased the number (P < 0.05 versus vehicle). The number was significantly greater in the OVA + 14 nm group than in the OVA group or the 14 nm nano particle group (P < 0.01). The number was also greater in the OVA + 56 nm nano particle group than in the 56 nm nano particle group (P < 0.05) or the OVA group, but the difference did not reach statistical significance.
Table 2 Numbers of inflammatory cells and goblet cells in lung tissue.
Group Animals Eosinophils Neutrophils Mononuclear Cells Goblet Cells
(n) (n/mm)
vehicle 8 0.187 ± 0.064 0.616 ± 0.140 0.495 ± 0.224 0.177 ± 0.070
14 nm 8 1.419 ± 0.466 3.825 ± 1.073 ** 2.431 ± 0.736 * 0.243 ± 0.068
56 nm 7 0.252 ± 0.055 1.710 ± 0.426 0.967 ± 0.418 2.510 ± 2.249
OVA 6 2.442 ± 0.761 1.671 ± 0.222 2.546 ± 0.479 * 5.262 ± 4.150
OVA + 14 nm 6 7.252 ± 2.745 ** ## $$ 4.144 ± 0.795 ** # 5.393 ± 0.560 ** ## $$ 17.141 ± 6.702 ** # $$
OVA + 56 nm 6 3.022 ± 0.830 2.546 ± 0.563 * 2.202 ± 1.086 12.932 ± 3.230 * $
Animals received intratracheal instillation of vehicle, nano particles, OVA, or OVA + nano particles for 6 wk. Lungs were removed and fixed 24 h after the last intratracheal administration. Sections were stained with Diff-Quik for measurement of inflammatory cells around the airways or with PAS for goblet cells in the bronchial epithelium. Results are expressed as numbers of cells per length of basement membrane of airways. Values are mean ± SEM. *P < 0.05 versus vehicle, **P < 0.01 versus vehicle, #P < 0.05 versus OVA, ##P < 0.01 versus OVA. $P < 0.05 versus nano particles. $$P < 0.01 versus nano particles.
Nano particles increase goblet cells after antigen challenge
To evaluate airway epithelial injury and hypersecretion of mucus, lung sections were stained with PAS (Table 2). OVA plus 56 nm nano particles increased the number of goblet cells as compared with vehicle without significance. The number was significantly greater in the OVA + 14 nm nano particle group than in the vehicle (P < 0.01), the OVA (P < 0.05), or the 14 nm nano particle group (P < 0.01). The number was greater also in the OVA + 56 nm nano particle group than in the vehicle (P < 0.05), the OVA (N. S.), or the 56 nm nano particle group (P < 0.05).
Effects of nano particles on local expression of Th2 cytokines in the presence of antigen
To explore the role of local expression of Th2 cytokines in the effects of nano particles on antigen-related airway inflammation, we quantitated protein levels of IL-5, IL-4, and IL-13 in the lung tissue supernatants (Table 3). OVA challenge increased the level of IL-5 as compared with vehicle challenge without significance. In the presence of OVA, nano particles significantly elevated levels of IL-5 as compared with vehicle (P < 0.01) or OVA (P < 0.05 for 56 nm, P < 0.01 for 14 nm). The levels were significantly greater in the OVA + nano particle groups than in the nano particle groups (P < 0.01). The levels of IL-13 were significantly greater in the OVA + 14 nm nano particle group than in the OVA group or 14 nm nano particle group (P < 0.01). The levels were greater also in the OVA + 56 nm nano particle group than in the OVA group (N. S.) or the 56 nm nano particle group (P < 0.05). The level of IL-4 was significantly lower in the OVA + 56 nm nano particle group than in the OVAgroup (P < 0.05). There were no other significant differences among the experimental groups.
Table 3 Protein levels of Th2 cytokines in the lung tissue supernatants.
Group Animals IL-5 IL-13 IL-4
(n) (pg/total lung tissue supernatants)
vehicle 16 5.5 ± 1.1 4.0 ± 1.1 204.3 ± 13.1
14 nm 13 4.6 ± 1.9 7.4 ± 4.3 194.7 ± 12.8
56 nm 14 7.1 ± 1.9 21.1 ± 8.9 204.3 ± 12.5
OVA 16 26.8 ± 11.1 16.2 ± 7.3 216.5 ± 14.8
OVA + 14 nm 16 113.7 ± 32.0** ## $ 120.3 ± 42.4** ## $ 178.5 ± 16.2
OVA + 56 nm 17 88.8 ± 38.0** # $ 61.8 ± 27.5* 170.8 ± 17.6#
Six groups were intratracheally inoculated with vehicle, nano particles, OVA, or the combination of OVA and nano particles for 6 wk. Lungs were removed and frozen 24 h after the last intratracheal administration. Protein levels in the lung tissue supernatants were analyzed using ELISA. Results are shown as mean ± SEM. *P < 0.05 versus vehicle, **P < 0.01 versus vehicle, #P < 0.05 versus OVA, ##P < 0.01 versus OVA. $P < 0.05 versus nano particles. $P < 0.01 versus nano particles.
Effects of nano particles on local expression of eotaxin, MCP-1, RANTES, and IL-6 in the presence of antigen
To investigate the local expression of eotaxin, MCP-1, RANTES, and IL-6, we measured protein levels of these cytokine and chemokines in the lung tissue supernatants (Table 4). OVA challenge increased the levels of eotaxin without significance as compared with vehicle challenge. The levels were significantly greater in the OVA + nano particle groups than in the vehicle (P < 0.01), the nano particle group (P < 0.05 for OVA + 56 nm nano particle, P < 0.01 for OVA + 14 nm nano particle), or the OVA (P < 0.05 for OVA + 56 nm nano particle, P < 0.01 for OVA + 14 nm nano particle) group. Nano particle challenge increased the levels of MCP-1 as compared to vehicle challenge (P < 0.01 for 14 nm, N. S. for 56 nm). OVA challenge slightly increased the levels without significance as compared with vehicle challenge. Nano particles combined with OVA enhanced the level as compared with nano particle alone (P < 0.01 for 14 nm nano particle, N. S. for 56 nm nano particle) or OVA alone (P < 0.01 for OVA + 14 nm nano particle group, P < 0.05 for OVA + 56 nm nano particle group). The levels of RANTES were significantly greater in the OVA + nano particle groups than in the vehicle group (P < 0.01 for OVA + 14 nm nano particle, P < 0.05 for OVA + 56 nm nano particle), or the OVA group (P < 0.01 for OVA + 14 nm nano particle, P < 0.05 for OVA + 56 nm nano particle), or the nano particle groups (P < 0.01 for 14 nm, N. S. for 56 nm). The levels of IL-6 were significantly greater in the OVA + 14 nm and OVA + 56 nm nano particle groups than in the vehicle group (P < 0.01), the OVA group (P < 0.01 for OVA + 14 nm nano particle group, P < 0.05 for OVA + 56 nm nano particle group), or the nano particle groups (P < 0.01).
Table 4 Protein levels of eotaxin, MCP-1, RANTES, and IL-6 in the lung tissuesupernatants.
Group Animals eotaxin MCP-1 RANTES IL-6
(n) (pg/total lung tissue supernatants)
vehicle 16 67.9 ± 2.9 20.8 ± 3.5 174.6 ± 15.8 105.8 ± 3.6
14 nm 13 99.4 ± 7.1 239.0 ± 17.4** 192.3 ± 18.0 151.7 ± 8.2
56 nm 14 89.5 ± 9.2 89.3 ± 7.4 201.6 ± 22.1 106.5 ± 4.9
OVA 16 130.9 ± 32.3 41.9 ± 9.2 160.7 ± 17.1 118.3 ± 5.7
OVA + 14 nm 16 804.8 ± 175.7** ## $$ 542.8 ± 45.4** ## $$ 432.6 ± 27.8** ## $$ 297.7 ± 30.2** ## $$
OVA + 56 nm 17 399.1 ± 86.9** # $ 124.0 ± 14.1** # 235.4 ± 16.8* # 202.4 ± 54.2** # $$
Six groups were intratracheally inoculated with vehicle, nano particles, OVA, or the combination of OVA and nano particles for 6 wk. Lungs were removed and frozen 24 h after the last intratracheal administration. Protein levels in the lung tissue supernatants were analyzed using ELISA. Results are shown as mean ± SEM. *P < 0.05 versus vehicle, **P < 0.01 versus vehicle, #P < 0.05 versus OVA, ##P < 0.01 versus OVA. $P < 0.05 versus nano particles. $$P < 0.01 versus nano particles.
Effects of nano particles on 8-OHdG formations in the presence or absence of antigen
We next studied 8-OHdG formation generated from deoxyguanosine in DNA by oxidative stress in the lung. In the vehicle group, nuclear staining with 8-OHdG was barely detectable (Fig. 2A). Nano particles or OVA challenge induced moderate staining with 8-OHdG (Fig. 2B, C, D). On the other hand, OVA plus nano particles resulted in intense immunoreactive 8-OHdG staining as compared to OVA or nano particles alone (Fig. 2E, F). The intensity and the extent of the immunoreactivity were more prominent in the OVA + 14 nm nano particle group (Fig. 1E) than in the OVA + 56 nm nano particle group (Fig. 2F). As typically shown in the OVA + nano particle groups, we found the expression of 8-OHdG in macrophages phagocyting nano particles as well as polymorphonuclear leukocytes (Fig. 2E, F).
Figure 2 Immunohistological staining for 8-hydroxy-2'-deoxyguanosine (8-OHdG) in the lung obtained from (A) vehicle group, (B) 14 nm nano particle group, (C) 56 nm nano particle group, (D) OVA group, (E) OVA + 14 nm nano particle group, and (F) OVA + 56 nm nano particle group (n = 8 in each group). Lungs were removed twenty-four h after the last intratracheal instillation. Arrows denote positive staining. Original magnification × 300.
We performed morphometric analysis to quantitate the extent and intensity of immunoreactive 8-OHdG among the experimental groups. As compared to vehicle treatment (immunohistochemical score, mean ± SEM: 0.8 ± 0.3), nano particles or OVA treatment revealed increased immunoreactivity (14 nm nano particle: 2.0 ± 0.4, P < 0.05 versus vehicle; 56 nm nano particle: 1.5 ± 0.2; OVA: 1.9 ± 0.5). The scores were greater in OVA + nano particle groups (OVA + 14 nm nano particle group: 2.9 ± 0.6; OVA + 56 nm nano particle group: 2.5 ± 0.3) than in the vehicle (P < 0.01), OVA, or nano particle groups.
Effects of nano particles on adjuvant activity for total IgE and antigen-specific production of IgG and IgE
To exanime whether nano particles have adjuvant activity for total IgE and antigen-specific Ig production, we measured total IgE and antigen-specific IgG1, IgG2a, and IgE (Table 5). Total IgE levels were significantly greater in the OVA + nano particle groups than in the vehicle group (P < 0.01 for OVA + 14 nm nano particle group, P < 0.05 for OVA + 56 nm nano particle group). The levels were also significantly greater in the OVA + 14 nm nano particle group than in the OVA or the 14 nm nano particle group (P < 0.05). The antigen-specific IgG1 was significantly greater in the OVA + 14 nm nano particle group than in the other groups (P < 0.01 versus each other group). The antigen-specific IgG2a was not significantly different among the experimental groups. The combination of OVA plus 14 nm nano particles significantly increased antigen-specific production of IgE as compared with vehicle or OVA alone (P < 0.05).
Table 5 Levels of total IgE and antigen-specific IgG1, IgG2a, and IgE.
Group Animals (n) Total IgE (ng/ml) Antigen-Specific IgG1 (titers) Antigen-Specific IgG2a (titers) Antigen-Specific IgE (fluorescene intensity)
vehicle 12 93.8 ± 27.5 485.3 ± 274.2 152.1 ± 48.8 625.2 ± 49.3
14 nm 9 152.3 ± 36.1 1880.3 ± 1656.2 163.5 ± 49.2 698.7 ± 61.0
56 nm 11 304.1 ± 127.4 1128.2 ± 717.1 182.5 ± 72.9 610.9 ± 54.1
OVA 10 443.4 ± 173.2 2178.5 ± 910.7 256.7 ± 206.2 576.2 ± 89.3
OVA + 14 nm 10 871.9 ± 246.7** # $ 28050.3 ± 13840.4** ## $ 354.9 ± 141.7 860.7 ± 120.5* ##
OVA + 56 nm 10 515.6 ± 122.4* 4150.5 ± 1657.0 126.7 ± 32.2 600.7 ± 62.8
Six groups were intratracheally inoculated with vehicle, nano particles, OVA, or the combination of OVA and nano particles for 6 wk. Serum samples were retrieved 24 h after the last intratracheal instillation. Total IgE and antigen-specific IgG1, IgG2a, and IgE were analyzed using ELISA. Results are expressed as mean ± SEM. *P < 0.05 versus vehicle, **P < 0.01 versus vehicle, #P < 0.05 versus OVA, ##P < 0.01 versus OVA, $P < 0.01 versus nano particles.
Discussion
The present study demonstrated that nano particles administered by the intratracheal route enhanced airway inflammation associated with antigen challenge in mice. The inflammatory component was characterized by increased numbers of eosinophils, neutrophils, and mononuclear cells. Recruitment of these cells was accompanied by an increment in goblet cells in the bronchial epithelium. The airway inflammation induced by the combined administration of nano particles with antigen modulated local expression of IL-5, eotaxin, IL-13, RANTES, MCP-1, and IL-6. The formation of 8-OHdG was moderately induced by nano particles or antigen alone, and was further enhanced by antigen plus nano particles as compared with nano particles or antigen alone. The enhancing effects were more prominent with 14 nm nano particles than with 56 nm nano particles. Furthermore, 14 nm nano particles enhanced total IgE and antigen-specific production of IgG1 and IgE.
DEP exacerbate allergic diseases including allergic asthma [5]. Elementary carbon, which is mainly involved in the nuclei of DEP, can enhance allergic sensitization. We used CB in the present study, since CB is a useful prototypical particle for the research on the effects and their mechanisms of PM including DEP. Because CB is relatively inert, the effects of particle size can be elucidated without confounding factors [23]. Al-Humadi and coworkers have demonstrated that CB exacerbates airway inflammation related to antigen in rats [18]. Last and colleagues have demonstrated that ambient particles with a diameter of less than 2.5 μm partially exacerbated lung inflammation related to antigen [17]. However, the comparative study focusing on the effects of particle size on antigen-related airway inflammation has never been conducted in vivo. In the present study, nano particles aggravated antigen-related airway inflammation, which was confirmed by the counts of inflammatory leukocytes in BAL fluid and by the histological assesment. Furthermore, we showed that nano particles exaggerated goblet cell hyperplasia elicited by antigen. In overall trends, the enhancing effects were more prominent with 14 nm nano particles than with 56 nm nano particles. Furthermore, 14 nm nano particles had obvious adjuvant activity for the antigen-specific production of IgG1 and IgE. These results clearly indicate that nano particles can aggravate antigen-related airway inflammation in vivo. Also, the effects are greater with smaller particles than with larger particles. We have previously examined the effects of DEP on allergic airway inflammation using 100 μg of DEP in vivo [5,24,25]. Based on the previous studies from our laboratory, we chose the dosage of 50 μg/body of nano particles, which can be considered to be involved in 100 μg of DEP as elementary carbon. Indeed, the enhancing effects of 14 nm nano particles on the airway inflammation and cytokine expressions are comparable to those of DEP in the previous study [5]. Another important point in this study is the surface area of the nano particles used. Surface area of particles exposed reportedly correlates magunitude of airway inflammation [26]. In our study, the surface area of the 14 nm nano particles was 6.7 fold larger than that of 56 nm nano particles (300 m2/g versus 45 m2/g). Nano particles with larger surface area are likely to attach more immunoregulative molecules than those with smaller surface area. As a result, smaller nano particles (14 nm) may lead to more prominent aggravation of antigen-related airway inflammation than larger nano particles (56 nm) in the present study. We did not examine the effects of the nano particles with the same particle number in the present study. However, the number of smaller nano particles is larger than that of larger nano particles when the particles make the same weight. Alternatively, our study has demonstrated not only the size effects of nano particles, but also the effects of their surface area and/or the effects of their number on the antigen-related airway inflammation. Future independent studies uniforming the surface area or particle number will provide better widestanding for the effects of the nano particles on antigen-related airway inflammation.
Allergic asthma is often associated with activation of IL-5 gene cluster, a pattern compatible with predominant activation of Th2-like T-lymphocyte population. IL-5 is essential for maturation of eosinophils in the bone marrow and their release into the blood [27,28]. Also, these Th2 cytokines are implicated in the pathogenesis of allergic reactions via their roles in mediating IgG1 and IgE production, and in differentiation, vascular adhesion, recruitment, activation, and survival of eosinophils. In our study, airway inflammation induced by the combined administration of nano particles and antigen were concomitant with the increased protein levels of IL-5. These results provide the first evidence that nano particles can accelerate antigen-related IL-5 expression and subsequent eosinophilic inflammation.
IL-13 is also recognized to regulate eosinophilic inflammation, and mucus secretion [29]. On the other hand, IL-6 is believed to participate in airway remodeling [30]. In the present study, nano particles enhanced the expression of the proteins in the presence of antigen. Therefore, nano particles may aggravate mucus hypersecretion and airway remodeling, at least partly, through the enhanced expression of IL-13 and IL-6. In fact, the OVA + nano particle groups showed enhancement in the mucus hypersecretion as compared with the OVA group or the nano particle groups.
Among chemokines, eotaxin is essential for eosinophil recruitment in antigen-related airway inflammation [31,32]. RANTES is a strong chemotactic and activating factor for eosinophils, and can modulate eosinophil adhesion [33,34]. In fact, our previous studies have confirmed that the exaggerated allergic airway inflammation induced by DEP paralleled the local elevation of the inflammatory proteins [5,22]. In the present study, nano particles enhanced the expression of these proteins in the presence of antigen as compared with antigen alone. The results suggest that nano particles aggravate allergic airway inflammation, at least in part, via the enhancement of the local expression of these proteins.
Interestingly, in our study, nano particles challenge increased the lung level of MCP-1 as compared to vehicle challenge. Also, the levels were significantly greater in the OVA + nano particle groups than in the vehicle or the OVA group. MCP-1 is a CC chemokine, and is chemoattractant for monocytes [34]. It also has a chemoattractant effect of CD4+ and CD8+ T lymphocytes [35]. MCP-1 also plays a role in recruitment of eosinophils to acute and chronic inflammatory sites [36]. Furthermore, some particles such as silica [37] and amosite asbestos [38] reportedly can induce MCP-1 in vitro and in vivo. Thus, our findings indicate that pulmonary exposure to carbon nano particles may induce MCP-1 expression in the airways. In addition, the aggravating effects of nano particles on antigen-related airway inflammation should be mediated, at least in part, via the enhanced expression of this chemokine.
In overall trends, the enhancing effects of nano particles on local expression of cytokines and chemokines related to antigen challenge were more prominent with 14 nm nano particles than with 56 nm nano particles. The differences in the enhanced expression of the proteins between the two sizes of nano particles may contribute, at least partly, to the differences in the magnitude of antigen-related airway inflammation and goblet cell hyperplasia.
Redox imbalance is a critical factor for tissue injury in various pulmonary diseases including inflammation such as asthma [39,40]. Airway macrophages from individuals with asthma produce more reactive oxidative species (ROS) than those from control subjects [41]. On the other hand, nano particles have been implicated to induce and/or enhance oxidative stress [42]. Furthermore, a recent study has demonstrated that the same type of nano particles as we used can induce ROS in the alveolar macrophages [43]. We, therefore, evaluated the contribution of oxidative stress to the deterious effects of nano particles on antigen-related airway inflammation. 8-OHdG is a proper marker of the oxidative stress. In our study, immunoreactivity of 8-OHdG in the lung was more intense in the OVA + nano particle groups than in the nano particle groups or the OVA group. Further, enhanced immunoreactivity for 8-OHdG was detected in alveolar macrophages as well as polymorphonuclear leukocytes. It is suggested that the enhanced oxidative stress is involved, at least partly, in the aggravation of antigen-related airway inflammation caused by nano particles.
Antigen-specific IgE is thought to contribute to inflammatory cell accumulation after antigen challenge via degranulation of mast cells [44]. On the other hand, IgG with antigen is a strong agonist for eosinophil degradation in vitro [45]. Furthermore, late asthmatic reactions are associated with IgG antibody [46]. Previous studies have reported that DEP with antigen demonstrate adjuvant activity for IgE production in vivo [47] and that those without antigen induce nonspecific IgE response in humans [48,49]. In addition, we have reported that DEP enhance antigen-specific production of IgE and IgG induced by intratracheal challenge with antigen in vivo [5,25]. In the present study, the combined intratracheal administration of 14 nm nano particles and antigen showed a significant greater increase in total IgE and antigen-specific IgG1 and IgE than the other administration. The enhancement in the antigen-specific immunogloblin production by 14 nm nano particles can induce the enhanced release of a variety of inflammatory mediators such as histamine and leukotrienes, resulting in aggravated manifestations of allergic asthma.
Finally, in the real world, we inhale nano particles and antigen in ambient air, not particle suspension nor aliquot of antigen. The dose of nano particles injected in the present study can be estimated to be less than a hundred fold than that we inhale in daily life. Further, real PM including DEP are complex mixture of carbon, metals, and organics, which are different from CB used in the present study. Thus, it remains to be elucidated in future whether daily inhalation of nano particles with or without other compounds including organic chemicals than elementary carbon combined with occasional exposure of aerosol antigen lead to the same results as the present study.
Conclusion
The present study has shown evidence that nano particles can aggravate antigen-related airway inflammation. The effect may be mediated, at least partly, through the increased local expression of IL-5 and eotaxin and also by the modulated expression of IL-13, MCP-1, IL-6, and RANTES. Furthermore, 14 nm nano particles enhance total IgE and antigen-specific production of IgG1 and IgE. These results suggest that nano particles can be a risk for exacerbation of allergic asthma. The aggravating effect may be larger with the smaller particles.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
KI participated in the design of the study and collection of the data, performed the statistical analysis and wrote initial drafts of the manuscript. HT participated in the design of the study, helped to organize the data and the results, and to prepare the manuscript. RY, MS, and KS participated in the collection of the data. TI participated in the design and coordination and helped to draft the manuscript. TY helped to prepare the manuscript. All authors read and approved the final manuscript.
Acknowledgements
The authors are grateful to Naoko Ueki and Emiko Shimada for their assistance.
==== Refs
Abbey DE Nishino N McDonnell WF Burchette RJ Knutsen SF Lawrence Beeson W Yang JX Long-term inhalable particles and other air pollutants related to mortality in nonsmokers Am J Respir Crit Care Med 1999 159 373 82 9927346
Cohen AJ Pope CA 3rd Lung cancer and air pollution Environ Health Perspect 1995 103 Suppl 8 219 24 8741787
Dockery DW Pope CA 3rd Acute respiratory effects of particulate air pollution Annu Rev Public Health 1994 15 107 132 8054077 10.1146/annurev.pu.15.050194.000543
Peters A Wichmann HE Tuch T Heinrich J Heyder J Respiratory effects are associated with the number of ultrafine particles Am J Respir Crit Care Med 1997 155 1376 83 9105082
Takano H Yoshikawa T Ichinose T Miyabara Y Imaoka K Sagai M Diesel exhaust particles enhance antigen-induced airway inflammation and local cytokine expression in mice Am J Respir Crit Care Med 1997 156 36 42 9230723
Takano H Yanagisawa R Ichinose T Sadakane K Yoshino S Yoshikawa T Morita M Diesel exhaust particles enhance lung injury related to bacterial endotoxin through expression of proinflammatory cytokines, chemokines, and intercellular adhesion molecule-1 Am J Respir Crit Care Med 2002 165 1329 35 11991888 10.1164/rccm.2108122
Utell MJ Frampton MW Acute health effects of ambient air pollution: the ultrafine particle hypothesis J Aerosol Med 2000 13 355 9 11262441
Ferin J Oberdorster G Penney DP Pulmonary retention of ultrafine and fine particles in rats Am J Respir Cell Mol Biol 1992 6 535 42 1581076
Li XY Brown D Smith S MacNee W Donaldson K Short-term inflammatory responses following intratracheal instillation of fine and ultrafine carbon black in rats Inhal Toxicol 1999 11 709 31 10477444 10.1080/089583799196826
MacNee W Donaldson K How can ultrafine particles be responsible for increased mortality? Monaldi Arch Chest Dis 2000 55 135 9 10949875
Nemmar A Vanbilloen H Hoylaerts MF Hoet PH Verbruggen A Nemery B Passage of intratracheally instilled ultrafine particles from the lung into the systemic circulation in hamster Am J Respir Crit Care Med 2001 164 1665 8 11719307
Maejima K Tamura K Taniguchi Y Nagase S Tanaka H Comparison of the effects of various fine particles on IgE antibody production in mice inhaling Japanese cedar pollen allergens J Toxicol Environ Health 1997 52 231 48 9316645 10.1080/009841097159674
Lambert AL Dong W Winsett DW Selgrade MK Gilmour MI Residual oil fly ash exposure enhances allergic sensitization to house dust mite Toxicol Appl Pharmacol 1999 158 269 77 10438660 10.1006/taap.1999.8709
Lambert AL Dong W Selgrade MK Gilmour MI Enhanced allergic sensitization by residual oil fly ash particles is mediated by soluble metal constituents Toxicol Appl Pharmacol 2000 165 84 93 10814556 10.1006/taap.2000.8932
Lovik M Hogseth AK Gaarder PI Hagemann R Eide I Diesel exhaust particles and carbon black have adjuvant activity on the local lymph node response and systemic IgE production to ovalbumin Toxicology 1997 121 165 78 9230448 10.1016/S0300-483X(97)00075-9
van Zijverden M van der Pijl A Bol M van Pinxteren FA de Haar C Penninks AH van Loveren H Pieters R Diesel exhaust, carbon black, and silica particles display distinct Th1/Th2 modulating activity Toxicol Appl Pharmacol 2000 168 131 9 11032768 10.1006/taap.2000.9013
Last JA Ward R Temple L Pinkerton KE Kenyon NJ Ovalbumin-induced airway inflammation and fibrosis in mice also exposed to ultrafine particles Inhal Toxicol 2004 16 93 102 15204782 10.1080/08958370490265077
Al-Humadi NH Siegel PD Lewis DM Barger MW Ma JY Weissman DN Ma JK The effect of diesel exhaust particles (DEP) and carbon black (CB) on thiol changes in pulmonary ovalbumin allergic sensitized Brown Norway rats Exp Lung Res 2002 28 333 49 12097228 10.1080/01902140290091976
Sanbongi C Takano H Osakabe N Sasa N Natsume M Yanagisawa R Inoue K Kato Y Osawa T Yoshikawa T Rosmarinic acid inhibits lung injury induced by diesel exhaust particles Free Radic Biol Med 2003 34 1060 9 12684091 10.1016/S0891-5849(03)00040-6
Inoue K Takano H Yanagisawa R Sakurai M Ichinose T Sadakane K Hiyoshi K Sato M Shimada A Inoue M Yoshikawa T Role of metallothionein in antigen-related airway inflammation Exp Biol Med (Maywood) 2005 230 75 81 15618129
Sano H Hla T Maier JA Crofford LJ Case JP Maciag T Wilder RL In vivo cyclooxygenase expression in synovial tissues of patients with rheumatoid arthritis and osteoarthritis and rats with adjuvant and streptococcal cell wall arthritis J Clin Invest 1992 89 97 108 1729286
Ichinose T Takano H Miyabara Y Sagai M Long-term exposure to diesel exhaust enhances antigen-induced eosinophilic inflammation and epithelial damage in the murine airway Toxicol Sci 1998 44 70 9 9720143 10.1006/toxs.1998.2459
Lambert AL Mangum JB DeLorme MP Everitt JI Ultrafine carbon black particles enhance respiratory syncytial virus-induced airway reactivity, pulmonary inflammation, and chemokine expression Toxicol Sci 2003 72 339 46 12655033 10.1093/toxsci/kfg032
Takano H Ichinose T Miyabara Y Yoshikawa T Sagai M Diesel exhaust particles enhance airway responsiveness following allergen exposure in mice Immunopharmacol Immunotoxicol 1998 20 329 36 9653676
Miyabara Y Takano H Ichinose T Lim HB Sagai M Diesel exhaust enhances allergic airway inflammation and hyperresponsiveness in mice Am J Respir Crit Care Med 1998 157 1138 44 9563731
Duffin R Clouter A Brown DM Tran CL MacNee W Stone V Donaldson K The importance of surface area and specific reactivity in the acute pulmonary inflammatory response to particles Ann Occup Hyg 2002 46 Suppl 1 242 5
Clutterbuck EJ Hirst EM Sanderson CJ Human interleukin-5 (IL-5) regulates the production of eosinophils in human bone marrow cultures: comparison and interaction with IL-1, IL-3, IL-6, and GMCSF Blood 1989 73 1504 12 2653458
Greenfeder S Umland SP Cuss FM Chapman RW Egan RW Th2 cytokines and asthma. The role of interleukin-5 in allergic eosinophilic disease Respir Res 2001 2 71 9 11686868 10.1186/rr41
Wynn TA IL-13 effector functions Annu Rev Immunol 2003 21 425 56 12615888 10.1146/annurev.immunol.21.120601.141142
DiCosmo BF Geba GP Picarella D Elias JA Rankin JA Stripp BR Whitsett JA Flavell RA Airway epithelial cell expression of interleukin-6 in transgenic mice. Uncoupling of airway inflammation and bronchial hyperreactivity J Clin Invest 1994 94 2028 35 7962549
Quackenbush EJ Wershil BK Aguirre V Gutierrez-Ramos JC Eotaxin modulates myelopoiesis and mast cell development from embryonic hematopoietic progenitors Blood 1998 92 1887 97 9731045
Humbles AA Conroy DM Marleau S Rankin SM Palframan RT Proudfoot AE Wells TN Li D Jeffery PK Griffiths-Johnson DA Williams TJ Jose PJ Kinetics of eotaxin generation and its relationship to eosinophil accumulation in allergic airways disease: analysis in a guinea pig model in vivo J Exp Med 1997 186 601 12 9254658 10.1084/jem.186.4.601
Alam R Stafford S Forsythe P Harrison R Faubion D Lett-Brown MA Grant JA RANTES is a chemotactic and activating factor for human eosinophils J Immunol 1993 150 3442 8 7682241
Kakazu T Chihara J Saito A Nakajima S Effect of RANTES on eosinophil adhesion to plates coated with recombinant soluble intercellular adhesion molecule-1 and expression of beta 2-integrin adhesion molecules on eosinophils Int Arch Allergy Immunol 1995 108 9 11 7549527
Penny LA Monocyte chemoattractant protein 1 in luteolysis Rev Reprod 2000 5 63 6 10864849 10.1530/ror.0.0050063
Conti P DiGioacchino M MCP-1 and RANTES are mediators of acute and chronic inflammation Allergy Asthma Proc 2001 22 133 7 11424873
Barrett EG Johnston C Oberdorster G Finkelstein JN Silica-induced chemokine expression in alveolar type II cells is mediated by TNF-alpha-induced oxidant stress Am J Physiol 1999 276 L979 88 10362723
Hill GD Mangum JB Moss OR Everitt JI Soluble ICAM-1, MCP-1, and MIP-2 protein secretion by rat pleural mesothelial cells following exposure to amosite asbestos Exp Lung Res 2003 29 277 90 12746042 10.1080/01902140303788
Heffner JE Repine JE Pulmonary strategies of antioxidant defense Am Rev Respir Dis 1989 140 531 54 2669581
Caramori G Papi A Oxidants and asthma Thorax 2004 59 170 3 14760161 10.1136/thorax.2002.002477
Calhoun WJ Reed HE Moest DR Stevens CA Enhanced superoxide production by alveolar macrophages and air-space cells, airway inflammation, and alveolar macrophage density changes after segmental antigen bronchoprovocation in allergic subjects Am Rev Respir Dis 1992 145 317 25 1310575
Li N Sioutas C Cho A Schmitz D Misra C Sempf J Wang M Oberley T Froines J Nel A Ultrafine particulate pollutants induce oxidative stress and mitochondrial damage Environ Health Perspect 2003 111 455 60 12676598
Beck-Speier I Dayal N Karg E Maier KL Schumann G Schulz H Semmler M Takenaka S Stettmaier K Bors W Ghio A Samet JM Heyder J Oxidative stress and lipid mediators induced in alveolar macrophages by ultrafine particles Free Radic Biol Med 2005 38 1080 92 15780766 10.1016/j.freeradbiomed.2005.01.004
Zweiman B The late-phase reaction: role of IgE, its receptor and cytokines Curr Opin Immunol 1993 5 950 5 8297529 10.1016/0952-7915(93)90111-5
Kaneko M Swanson MC Gleich GJ Kita H Allergen-specific IgG1 and IgG3 through Fc gamma RII induce eosinophil degranulation J Clin Invest 1995 95 2813 21 7769121
Durham SR Lee TH Cromwell O Shaw RJ Merrett TG Merrett J Cooper P Kay AB Immunologic studies in allergen-induced late-phase asthmatic reactions J Allergy Clin Immunol 1984 74 49 60 6736483 10.1016/0091-6749(84)90086-1
Takafuji S Suzuki S Koizumi K Tadokoro K Miyamoto T Ikemori R Muranaka M Diesel-exhaust particulates inoculated by the intranasal route have an adjuvant activity for IgE production in mice J Allergy Clin Immunol 1987 79 639 45 2435776
Diaz-Sanchez D Dotson AR Takenaka H Saxon A Diesel exhaust particles induce local IgE production in vivo and alter the pattern of IgE messenger RNA isoforms J Clin Invest 1994 94 1417 25 7523450
Diaz-Sanchez D Tsien A Casillas A Dotson AR Saxon A Enhanced nasal cytokine production in human beings after in vivo challenge with diesel exhaust particles J Allergy Clin Immunol 1996 98 114 23 8765825
|
16164761
|
PMC1242256
|
CC BY
|
2021-01-04 16:23:25
|
no
|
Respir Res. 2005 Sep 16; 6(1):106
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-106
|
oa_comm
|
==== Front
Virol JVirology Journal1743-422XBioMed Central London 1743-422X-2-791614304110.1186/1743-422X-2-79ResearchCharacterisation of parapoxviruses isolated from Norwegian semi-domesticated reindeer (Rangifer tarandus tarandus) Klein Joern [email protected] Morten [email protected] Department of Microbiology and Virology, University of Tromsø, Breivika, N-9037 Tromsø, Norway2 Danish Institute for Food and Veterinary Research, Department of Virology, Lindholm, DK-4771 Kalvehave, Denmark3 Section of Arctic Veterinary Medicine, Department of Food Safety and Infection Biology, The Norwegian School of Veterinary Science, PO Box 6204, N-9292 Tromsø, Norway2005 5 9 2005 2 79 79 28 6 2005 5 9 2005 Copyright © 2005 Klein and Tryland; licensee BioMed Central Ltd.2005Klein and Tryland; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Two outbreaks of the disease contagious ecthyma were reported in 1999 and 2000 in Norwegian semi-domesticated reindeer (Rangifer tarandus tarandus). Contagious ecthyma is an epidermal disease of sheep and goats worldwide, which is caused by the zoonotic parapoxvirus orf virus. Characterisation of clinical samples from the two outbreaks in semi-domesticated reindeer in Norway by electron microscopy and PCR (B2L) revealed typical parapoxvirus particles and partial gene sequences corresponding to parapoxvirus, respectively. If contagious ecthyma in reindeer is caused by orf virus, the virus may be transferred from sheep and goats, via people, equipment and common use of pastures and corrals, to reindeer. Another possibility is that contagious ecthyma in reindeer is caused by a hitherto unclassified member of the parapoxvirus genus that circulates among reindeer herds and remains endemic in Norway.
Results
Genomic comparisons of one standard orf strain (orf NZ2) and the reindeer isolates, employing restriction fragment length polymorphism (RFLP) and random amplified polymorphic DNA (RAPD) analysis, demonstrated high similarity between the reindeer viruses and known orf virus strains. Partial DNA sequences of two different viral genes were determined for the different isolates and compared with corresponding parapoxvirus genebank sequences. The comparison/alignment and construction of phylogenetic trees also point to an affiliation of the reindeer viruses to the species orf virus.
Conclusion
The results of this work imply that the parapoxvirus causing contagious ecthyma in Norwegian semi-domesticated reindeer belongs to the species orf virus and that the orf virus crosses the host species barrier from sheep and goat to semi-domesticated reindeer.
==== Body
Background
Parapoxviruses (PPVs) (family Poxviridae) cause dermal diseases most commonly in sheep, goats and cattle [2], but also in semi-domesticated reindeer (Rangifer tarandus tarandus) [3] and wildlife, like seals [4], red deer [5] and squirrels [6]. The genus Parapoxvirus consists of five species and three tentative species [7]: Bovine popular stomatitis virus (BPSV), orf virus (ORFV), parapoxvirus of red deer in New Zealand (PVNZ), pseudocowpox virus (PCPV), and squirrel parapoxvirus (SPPV), as well as the tentative species of the genus: auzdyk disease virus, chamois contagious ecthyma virus and sealpox virus.
The first documented outbreak of a contagious ecthyma-like disease in Norwegian reindeer, took place in 1976 among experimental animals at the National Reindeer Research Station in Lødingen [8]. Also human parapoxvirus infections were associated with this outbreak [9]. In April 1999, the first outbreak in reindeer under regular herding conditions was reported in Troms County, followed by an outbreak one year later, also in April, in Nordland County. The outbreak in 2000 involved at least 30 out of 150 reindeer, which were corralled and supplementary fed during the winter. Seven of the infected animals died, due to massive oral cauliflower-like lesions and secondary bacterial infections [1].
In Norway, there are both wild and semi-domesticated reindeer. The latter is herded over an area of approximately 140.000 km2, which is about 40% of the mainland area of Norway. Most reindeer herding is conducted by the Saami people (indigenous people of Scandinavia), and the most dense reindeer herding area is in Finnmark county, Northern Norway, where the total number of animals is estimated to be around 140.000 animals (estimate for 2003 [10]). Semi-domesticated reindeer have seasonal migrations between winter (usually inland) and summer (usually coastal areas) pastures, and the pastures are organised in districts, reducing the contact between animals of different districts. However, sheep and goats also use the coastal pastures during spring, summer and autumn, and contact between different animal species is possible. Furthermore, it is not uncommon to use the same corrals and transport vehicles for reindeer and sheep, and remaining scabs from sheep with contagious ecthyma may as well be a source of infection for reindeer. Since the disease contagious ecthyma and orf virus is present among sheep and goats in most parts of the country, it is necessary to characterise the causative agent of contagious ecthyma in reindeer, in order to find out whether reindeer have their own parapoxvirus species, like red deer in New Zealand [5], or whether a transmission of virus between sheep and goats and reindeer is more likely.
Due to genetic heterogeneity within the genus, classification of PPV remains problematic and relies mostly on the source of the virus isolate; id est Parapoxviruses isolated from sheep will be classified as orf virus, virus isolated from cattle as pseudocowpox [11,12]. However, several molecular techniques have been used to characterise parapoxviruses. Inoshima et al. [13] reported a polymerase chain reaction (PCR) protocol assumed to be able to amplify all members of the parapoxvirus genus, based on primers from the B2L gene encoding a major envelope protein [14]. The gene region encoding a viral interleukin 10 orthologue (vIL-10) [15] has been used for characterisation, in combination with sequencing of the amplified DNA products. Restriction fragment length polymorphism (RFLP) has also been used to compare parapoxvirus isolates [5,16], and can be conducted both on genomic DNA as well as on smaller DNA fragments in combination with PCR. Also random amplified polymorphic DNA analysis (RAPD) may be used for characterisation purposes, as reported as a useful technique for comparing orthopoxviruses [17].
The aim of the present study was to characterise the causative agent of contagious ecthyma in semi-domesticated reindeer in Norway, and to compare this virus with other parapoxvirus isolates. This information is necessary to be able to sort out whether reindeer are exposed to orf virus, which is commonly affecting sheep and goats and also present among muskoxen in Norway, to bovine papillar stomatitis virus, which is present among cattle in Norway, or whether reindeer hosts a specific parapoxvirus species.
Results
Restriction Fragment Length Polymorphism (RFLP) analysis
RFLP patterns by the three restriction endonucleases, Hind III, Eco R1 and BAM H1 display identical DNA fragment patterns for orf virus and the parapoxvirus isolated from semi-domesticated Norwegian reindeer (Figure 1).
Figure 1 Restriction fragment length polymorphism (RFLP) analysis of the standard orf virus strain NZ2 and the Norwegian reindeer isolate from Troms county 2000 (bands indicated by arrows) displays identical cleavage patterns for the two viruses.
Random Amplification of Polymorphic DNA (RAPD) analysis
The RAPD patterns of the Finnish (Fi94.1Rt) and the Norwegian (N99.1Rt, N00.1Rt) reindeer isolates are similar to that of the orf viruses orf 11 and NZ2 and distinct to the sealpox and pseudocowpox patterns (Figure 2).
Figure 2 Random amplified polymorphic DNA (RAPD) analysis of different parapoxviruses. Lane 1: reindeer Finland 1994, Lane 2: reindeer Norway 1999, Lane 3: reference strain orf 11, Lane 4: reindeer Norway 2000, Lane 5: orf virus NZ2, Lane 6: parapoxvirus from Weddell seal, Lane 7: parapoxvirus from cattle, Norway, Lane 8: mock infected cells, Lane 9 and 10: 1 kb and 100 bp ladder, respectively. Bands indicate similar patterns for orf and reindeer viruses, whereas parapoxvirus from seal and cattle are different.
Polymerase Chain Reaction (PCR)
The vIL-10 PCR and GIF PCR were able to amplify DNA from all 25 isolates, whereas the B2L PCR did only amplify DNA from 17 isolates (Table 1).
Table 1 PCR and sequencing results obtained from parapoxvirus isolates of different species, reference orf virus strain (Orf 11) and an orf virus vaccine strain (NZ2). Successful amplification is indicated by Genebank accession number
Signature Host Country of origin B2L-PCR vIL-10-PCR GIF-PCR
N99.1Rt Rangifer t. tarandus Norway AY605963 AY605995 AY605973
N00.1Rt Rangifer t. tarandus Norway AY605964 AY605994 AY605972
N00.2Rt Rangifer t. tarandus Norway AY605969 AY606005 AY605985
N03.8Rt Rangifer t. tarandus Norway AY605966 AY605992 AY606010
Fi94.1Rt Rangifer t. tarandus Finland AY605965 AY605993 AY605971
Fi92.1Rt Rangifer t. tarandus Finland AY605959 AY606001 AY605979
N79.1Bos Bos spec. Norway AY605960 AY606011 AY605980
N92.1Bos Bos spec. Norway AY605961 AY606002 AY605981
N83.1Bos Bos spec. Norway AY605970 AY606003 AY605982
N85.1Bos Bos spec. Norway - AY963707 AY605984
N71.1Bos Bos spec. Norway - AY606012 AY605983
N02.1Ch Capra hircus Norway - Not sequenced AY606013
N00.1Ch Capra hircus Norway - AY605999 AY605977
N94.1Om Ovibos moschatus Norway AY605962 AY605996 AY605974
N00.1Oa Ovis aries Norway AY605957 AY605998 AY605976
N86.1Oa Ovis aries Norway AY605968 AY606007 Not sequenced
N86.2Oa Ovis aries Norway AY605967 AY606008 AY605990
N03.1Oa Ovis aries Norway - AY963708 AY605991
Sc95.1Hg Halichoerus grypus Scotland U49845AJ622901 AY605997 AY605975
N03.1Lw Leptonychotes weddelli Antarctica AJ622900 AY606015 AY605989AY605989
Orf 11 Cell culture - AY605958 AY606000 AY605978
Orf NZ2 Vaccine - AY963706 AY606006 AY605988
Phylogenetic analysis
The Bayesian tree based on the the partial sequences of the B2L-gene (Figure 3) display species specific clustering of the parapoxvirus-genus.
Figure 3 Bayesian tree based on the partial nucleotide sequences of the B2L gene (379 nt) obtained in this study compared with corresponding DNA sequences from parapoxviruses published in Genebank. Isolates are described by Genebank accession number, parapoxvirus species, source and country of origin. Numbers at major clades indicate clade credibility values in Percent.
Six clusters, representing the squirrel parapoxviruses, seal poxviruses, bovine papular stomatitis virus, pseudocowpoxviruses, parapoxvirus of red deer in New Zealand, and orf viruses were generated.
The Norwegian reindeer isolates clustered together with orf virus isolates from different host species and geographical origins. Also the Finnish reindeer isolates from 1992 and 1994 were in conjunction with the orf virus isolates, whereas more recent Finnish parapoxvirus isolates causing contagious ecthyma in reindeer have been characterised as more related to pseudocowpoxviruses [18].
Phylogenetic analysis of interleukin 10 amino acid sequences from a range of mammalian species and the three translated viral interleukin 10 orthologue nucleotide sequences obtained from the two Norwegian (N00.1Rt and N99.1Rt) and one Finnish (Fi94.1Rt) virus isolates from semi-domesticated reindeer is shown in Figure 4. The positions of the viral IL-10 indicate a high similarity to the corresponding genes (interleukin 10) of the main hosts of orf virus, goat and sheep. (Figure 4).
Figure 4 Maximum Parsimony tree based on the translated nucleotide sequences of the Norwegian and Finnish reindeer vIL-10 gene amplicons obtained in this study, compared with corresponding amino acid sequences from mammals published in Genebank. Black numbers (branch length) describe the genetic distance/number of changes along the branch. Blue numbers (bootstrap values) describe the reliability for each clade in percent.
Discussion
Based on RFLP patterns obtained from the Norwegian reindeer isolates and orf NZ2, a close relationship between these viruses can be assumed (Figure 1). The RAPD patterns of the reindeer isolates and the standard orf virus strains orf 11 and orf NZ2 show high similarity in amplification patterns (Figure 2). This similarity is present in spite of the geographical distance between the Norwegian isolates and the orf strains originating from Great Britain and New Zealand.
Further, this work demonstrates that a virus species characterisation based on the nucleotide sequence of the PCR-product from the B2L-PCR is possible. This characterisation method is easy to perform, because the PCR product is fast to obtain and to proceed.
The phylogenetic analysis of the partial B2L-gene sequence from the Norwegian reindeer isolates shows that these isolates can be allocated to the orf virus species. However, the clade credibility value of 37 % also demonstrate a close relationship between the two different parapoxvirus species, orf and pseudocowpox virus.
Tikkanen et al. [18] demonstrated the affiliation of parapoxvirus isolates obtained from Finnish semi-domesticated reindeer during the early outbreaks (1992–1994) in Finland, to the orf virus species and from later outbreaks to the pseudocowpox viruses, which is congruent with our results.
The close clustering of the translated sequences obtained from virus isolated from reindeer (two Norwegian; 1999, 2000, and one Finnish; 1994) with the amino acid sequences of sheep and goat interleukin 10 demonstrate a high relationship of the parapoxvirus isolated from reindeer to the main hosts of orf virus (sheep and goat). It seems that the viral interleukin 10 orthologue of the reindeer isolates is highly adapted to the immune system of sheep and goats, which also indicates that the reindeer parapoxvirus belongs to the orf virus species.
Our results indicate that neither geographical distance, nor crossing the host barrier from sheep or goat to semi-domesticated reindeer did affect the characteristics of the parapoxvirus orf virus investigated in this study.
The B2L PCR has previously been described as a tool to amplify all species within the parapoxvirus genus [13]. However, we were not able to amplify B2L sequences from nine of the twenty-five isolates included in our study. Many members of the subfamily chordopoxvirinae show genetic rearrangement at the terminal sequences of the genome, which is thought to be an evolutionary mechanism, allowing the virus to adapt to changes of the immune response of the host [19]. Parapoxvirus replicated in vitro show rearrangement of the left terminal end of the viral genome, resulting in the deletion of three genes (E2L, E3L, G1L) and 80 % of a fourth one, the G2L [20-22]. All these four genes play a major role in virulence and host specificity [23]. The B2L gene is localised beside the left terminal end of the genome, so that heterogeneity in the primer binding regions or complete deletion of the B2L gene, through the same rearrangement mechanisms as in the left terminal region itself, can be the reason for the negative results of the B2L-PCR. However, further analyses are needed to evaluate this hypothesis. As compared to the results of the B2L-PCR, the GIF PCR seems to be more sensitive for different members of the genus and can be used for rapid genus-identification. However, the GIF gene is localised in the right terminal gene region [24] and may also undergo genetic rearrangements due to adaptation processes. The GIF gene has so far only been detected in parapoxviruses [23].
The vIL-10-PCR amplified all tested isolates, but rearrangements of the vIL-10 gene have also been demonstrated, in terms of duplications in the inverted terminal repeat [25]. Thus, genetic rearrangements may be a problem when designing a PCR that are supposed to detect parapoxviruses in general. The use of a multiplex PCR, with a combination of one or more of the gene targets desribed above, may thus be a solution to this problem.
Conclusion
The results of this work point out that the parapoxvirus that has caused contagious ecthyma in Norwegian semi-domesticated reindeer belongs to the orf virus species, and it is to assume that the orf virus crosses the host species barrier from sheep and goat to semi-domesticated reindeer.
Sheep and goats in Norway are commonly free-ranging during the snow-free period, sharing pastures with semi-domesticated reindeer. During seasonal migrations and corralling of animals for slaughter etc. they may also share fences, transport vehicles and other equipment.
As far as we know, contagious ecthyma is the single disease that has caused the most serious economical losses for reindeer herders in the Nordic countries in recent times, and especially so in Finland. Consequently, the common use of equipment, pastures, transport vehicles and facilities for semi-domesticated reindeer, sheep and goats should be avoided, to prevent cross infections.
The outcome of parapoxvirus infections in reindeer seems to be dependent on many environmental factors in addition to the exposure to the virus. Certain types of stress is believed to play a key role, and such stress factors may be lack of food, as well as handling, corralling and transport of animals [3,1]. A changing trend in reindeer herding conditions, facing higher animal densities, faster movements of animals using helicopter and snow mobiles, increased use of supplemental feeding and corralling of animals, and increased transport distances to slaughterhouse may represent factors that can predispose for contagious ecthyma as well as other diseases in reindeer.
Methods
Viruses
An overview of the virus isolates included in this study is given in Table 1. Viruses were purified by metrizamide gradient centrifugation from homogenised scab material obtained from Norwegian semi-domesticated reindeer (N99.1Rt, N00.2Rt, N03.8Rt), Finnish semi-domesticated reindeer (Fi94.1Rt, Fi92.1Rt), Norwegian cattle (N79.1Bos, N92.1Bos, N83.1Bos, N71.1Bos, N85.1Bos), Norwegian goat (N02.1Ch, N00.1Ch), Norwegian musk ox (Ovibos moschatus) (N94.1Om, N86.1Om), Norwegian sheep (N00.1Oa, N86.1Oa, N86.2Oa, N03.1Oa, N03.2Oa, N00.2Oa), Scottish grey seal (Halichoerus grypus) (Sc95.1Hg) and Antarctic Weddell seal (Leptonychotes weddellii) (N03.1Lw) as described previously [26].
The orf virus strain orf 11 was provided by the Moredun Research institute (Great Britain) and the orf virus strain NZ2 was derived from a non-attenuated commercial vaccine against contagious ecthyma in sheep (Scabivax®, Shering-Plough A/S Animal Health, Norway).
Cell culture
Orf viruses NZ2 and orf 11 and the parapoxviruses isolated from Norwegian semi-domesticated reindeer (N00.1Rt, N99.1Rt), Finnish semi-domesticated reindeer (Fi94.1Rt) and the Scottish grey seal (Sc95.1Hg) were propagated in Madine-Darby bovine kidney (MDBK; DSMZ No; ACC 174) cells, which are permissive for parapoxviruses [27]. Cells were cultivated in Dulbecco's MEM supplemented with 5 % Fetal Bovine Serum and Penicillin (100 μg/ml)/Streptomycin-solution (100 IU/ml) and incubated at 37°C with 5 % CO2.
DNA extraction
For the purpose of RFLP and RAPD analysis, viral DNA was extracted from the cytoplasma of infected cells, using reducing, non-ionic and proteolytic detergents, as described in detail by Esposito et al. [28].
For PCR, viral DNA was extracted using QIAamp® DNA Mini Kit (QIAGEN, Hilden, Germany).
RFLP
Viral DNA was digested for 4 hours with the restriction enzymes HindIII, EcoRI and BamHI (NEW ENGLAND BioLabs®Inc., UK). DNA fragments were separated on a 0.6 % agarose gels for 20 hours at 0,6 V/cm.
RAPD
RAPD were conducted with the commercial kit Ready.To.Go® RAPD Analysis Beads (Amersham Biosciences AB, Uppsala, Sweden). Five μl of the RAPD analysis primer no. 6 (5'- CCCGTCAGCA-3') were added to 19 μl of dH2O and 1 μl of template DNA and gently mixed. The following low stringency cycling profile was used: initial denaturation at 95°C for 4 min, followed by 45 cycles consisting of denaturation at 95°C for 1 min, annealing at 36°C for 1 min, and elongation at 72°C for 2 min. DNA fragments were separated on a 2 % agarose gel at 7,5 V/cm for 2,5 hours.
PCR
Three different PCR protocols were performed as specified below:
B2L-PCR
Inoshima et.al. [13] describe a PCR specific for the detection of all parapoxviruses, resulting in the amplification of theoretically a 594 bp product. The primers (PPP 1 and PPP 4) are based on the B2L gene sequence of the orf virus strain NZ2. The B2L gene encode a homologue of the vaccinia virus major envelope antigen p37K gene [29]. PCR was carried out as described by Inoshima et al. [16] with the exception that 5 % dimethylsulfoxide (DMSO) was added to the reaction mix as a PCR enhancer.
GIF-PCR
The Granulocyte-macrophage-colony-stimulating factor (GM-CSF) and Interleukin-2 inhibition factor (GIF) is found only in parapoxviruses, and represents an important virulence factor [24]. Amplification of parts of the GIF gene may thus be useful, both for detection of parapoxvirus DNA in tissue samples and for virus species differentiation. The GIF PCR primers (GIF 5 → 5'-gct cta gga aag atg gcg tg-3' GIF 6 → 5'-gta ctc ctg gct gaa gag cg -3'), generating amplicons of approximately 408 bp, were obtained from the published sequence of the orf virus GIF gene (Genebank accession number AF192803.1; Deane et al., 2000) and selected by the online tool "GeneFisher" [30].
Five μl template DNA were added to 45 μl of the PCR reaction mixture containing 0.2 mM primers (GIF 5 and GIF 6), 200 mM each of dATP, dCTP, dGTP and dTTP, 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl2 and 1 U of AmpliTaq® Gold DNA polymerase (Applied Biosystems, Norway). DNA was amplified with a DNA Thermal Cycler PE9700 (Perkin Elmer) by a two-step cycling reaction as follows: 95°C for 15 min, and five cycles of 94°C for 30 sec, 57°C for 2 min and 72°C for 30 sec, and then 35 cycles of 94°C for 30 sec, 57°C for 30 sec and 72°C for 30 sec, followed by a final extension step of 72°C for 10 min. The resulting PCR product was examined by electrophoresis, using a 1,2 % agarose gel, containing 0,005 % ethidium bromide, with a separation time of 1,5 hours at 6,5 V/cm.
vIL-10 PCR
The viral interleukin 10 orthologue [15] need to have a close similarity to the IL-10 of the host for effective virus propagation. PCR targeting this gene may be useful for genus affiliation, and nucleotide sequencing of the PCR amplicon followed by virtual translation to the protein sequence may be suitable for virus characterisation.
Primers (vIL-10-3 → 5'-atg cta ctc aca cag tcg ctc c-3', vIL-10-4 → 5'-tat gtc gaa ctc gct cat ggc c-3') were obtained from consensus sequences previously reported to Genebank (accession numbers OVU82239; [31], OVU60552; [32], AY231116.1; [33], AY186733.1; [34], AY186732; [34]) and selected by the online tool "GeneFisher[30]. The expected length of the resulting amplicon is approximately 300 bp. The reaction mixture, cycling profile and agarose gel analysis was conducted as described for the GIF PCR.
Nucleotide sequencing
The resulting amplicons of the B2L, GIF and vIL-10 PCR were prepared for nucleotide sequencing by enzymatic removal of unused dNTP and primers (ExoSAP-IT™; Amersham Pharmacia Biotech, Sweden). The enzyme preparation (0,5 μl; ExoSAP-IT™) was added directly to 6,5 μl of the PCR product and incubated at 37°C for 1 hour. ExoSAP-IT™ was inactivated by heating to 80°C for 15 minutes. After the clean-up procedure the sequencing protocol for the BigDye®Terminator v3.1 cycle sequencing kit (Applied Biosystems, Norway) was performed. Seven μl of the purified PCR product was mixed with 4 μl "Ready Reaction Premix", 2 μl sequencing-buffer, 3,2μl of 20 μM Primer solution, 1 μl DMSO and 2,8 μl dH2O. This mixture was thermal cycled 25 times at 96°C for 10 seconds, 50°C for 5 seconds and 60°C for 4 minutes. DNA was precipitated with ethanol and the sequence was determined with the "ABI PRISM 377 Genetic Analyser" (Applied Biosystems, Norway).
Phylogenetic analysis
Multiple sequence alignment of the 379 nucleotide long partial B2L-sequence was conducted using CLUSTAL X (version 1.81; [35]) and phylogenetic analysis was performed by Bayesian Analysis using MrBayes [36] with the following settings. The maximum likelihood model employed 2 substitution types ("nst = 2"), with base frequencies set to the empirically observed values ("basefreq = empirical"). Rate variation across sites was modelled using a gamma distribution (rates="gamma"). The Markov chain Monte Carlo search was run with 4 chains for 500000 generations, with trees begin sampled every 100 generations (the first 1000 trees were discarded as "burnin").
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
This work is based on the MPhil thesis of Jörn Klein. Morten Tryland was the main supervisor of this thesis and project leader. The thesis is available online under:
Acknowledgements
Hilde Hansen is greatly acknowledged for her contribution to this characterisation study. The contributors of parapoxvirus isolates are also greatly acknowledged: Antti Oksanen (Finnish reindeer isolates), Johan Krogsrud (isolates from cattle), Terje Josefsen, Karen Sørensen and Torill Mørk (isolates from sheep and goats), Arnoldus S. Blix (isolates from musk oxen), Colin McInnes (orf 11 isolate), and Peter Nettleton (grey seal isolate). This work was supported financially by the Reindeer Husbandry Development Fund, 2002–2004.
==== Refs
Tryland M Josefsen T Oksanen A Aschfalk A Contagious ecthyma in Norwegian semi-domesticated reindeer (Rangifer tarandus tarandus) Vet Rec 2001 149 394 395 11601519
Kuhl JT Huerter CJ Hashish H A case of human orf contracted from a deer Cutis 2003 71 288 290 12729092
Oksanen A Norberg H Smittsom munnskurv Reindriftsnytt 1994 3–4 13 17
Tryland M Klein J Nordoy ES Blix AS Isolation and partial characterization of a parapoxvirus isolated from a skin lesion of a Weddell seal Virus Res 2005 108 83 87 15681058 10.1016/j.virusres.2004.08.005
Robinson AJ Mercer AA Parapoxvirus of red deer: evidence for its inclusion as a new member in the genus Parapoxvirus Virology 1995 208 812 815 7747456 10.1006/viro.1995.1217
Thomas K Tompkins DM Sainsbury AW Wood AR Dalziel R Nettleton PF McInnes CJ A novel poxvirus lethal to red squirrels (Sciurus vulgaris) J Gen Virol 2003 84 3337 3341 14645914 10.1099/vir.0.19464-0
Büchen-Osmond C Editor ICTVdb Index of Viruses: Catalogue of virus taxonomy and nomenclature approved by the International Committee on Taxonomy of Viruses Internet, cited 2003 May 16
Kummeneje K Krogsrud J Contagious ecthyma (orf) in reindeer (Rangifer t. tarandus) Vet Rec 1979 105 60 61 576046
Falk ES Parapoxvirus infections of reindeer and musk ox associated with unusual human infections Br J Dermatol 1978 99 647 54 216383
Hætta JI editor Reindriftsnytt 1978
Robinson AJ Ellis G Balassu T The genome of orf virus: restriction endonuclease analysis of viral DNA isolated from lesions of orf in sheep Arch Virol 1982 71 43 55 6279055 10.1007/BF01315174
Büttner M Rziha H-J Parapoxviruses: From the Lesion to the Viral Genome J Vet Med B 2002 49 7 16 10.1046/j.1439-0450.2002.00539.x
Inoshima Y Morooka A Hiroshi S Detection and diagnosis of parapoxvirus by the polymerase chain reaction J Virol Methods 2000 84 201 208 10680970 10.1016/S0166-0934(99)00144-5
Sullivan J Fraser K Fleming S Robinson A Mercer A Sequence and transcriptional analysis of an orf virus gene encoding ankyrin-like repeat sequences Virus Genes 1994 9 277 282 7597806 10.1007/BF01702883
Haig D Flemming S Immunomodulation by virulence proteins of the Parapoxvirus orf virus Vet Immunol Immunopathol 1999 72 81 86 10614496 10.1016/S0165-2427(99)00119-1
Inoshima Y Murakami K Yokoyama T Sentsui H Genetic heterogeneity among parapoxviruses isolated from sheep, cattle and Japanese serows (Capricornis crispus) J Gen Virol 2001 82 1215 1220 11297696
Stemmler M Neubauer H Meyer H Comparison of closely related orthopoxvirus isolates by random amplified polymorphic DNA and restriction fragment length polymorphism analysis J Vet Med B Infect Dis Vet Public Health 2001 48 647 54 11765800
Tikkanen MK McInnes CJ Mercer AA Buttner M Tuimala J Hirvela-Koski V Neuvonen E Huovilainen A Recent isolates of parapoxvirus of Finnish reindeer (Rangifer tarandus tarandus) are closely related to bovine pseudocowpox virus J Gen Virol 2004 85 1413 1418 15166423 10.1099/vir.0.79781-0
Turner PC Moyer RW The molecular pathogenesis of poxviruses Curr Top Microbiol Immunol 2004 163 125 151 2173654
Fleming S Lyttle D Sullivan J Mercer A Robinson A Genomic analysis of a transposition-deletion variant of orf virus reveals a 3.3 kbp region of non-essential DNA J Gen Virol 2004 76 2969 2978 8847502
Cottone R Büttner M Bauer B Henkel M Hettich E Rziha H-J Analysis of genomic rearrangement and subsequent gene deletion of attenuated Orf virus strain D1701 Virus Res 1998 56 53 67 9784065 10.1016/S0168-1702(98)00056-2
McInnes C Wood A Nettleton P Gilray J Genomic Comparsion of an Avirulent Strain of Orf Virus with that of a Virulent Wild Type Isolate Reveals that the Orf Virus Gene G2L is Non-Essential for Replication Virus Genes 2001 22 141 150 11324750 10.1023/A:1008117127729
Haig D McInnes C Immunity and counter-immunity during infection with the Parapoxvirus orf virus Virus Res 2002 88 3 16 12297324 10.1016/S0168-1702(02)00117-X
Deane D McInnes C Percival A Wood A Thomson J Lear A Gilray J Fleming S Mercer A Haig D Orf virus encodes a novel secreted protein inhibitor of granulocyte-macrophage colony-stimulating factor and interleukin-2 J Virol 2000 74 1313 1320 10627542 10.1128/JVI.74.3.1313-1320.2000
Rzhia H Henkel M Cottone R Meyer M Dehio C Büttner M Parapoxviruses: potential alternative vectors for directing the immune response in permissive and non-permissive hosts J Biotechnol 1999 83 137 145 10.1016/S0168-1656(00)00307-2
Tryland M Sandvik T Hansen H Haukenes G Holtet L Bennett M Mehl R Moens U Olsvik O Traavik T Characteristics of four cowpox virus isolates from Norway and Sweden APMIS 1998 106 623 635 9725795
Kruse N Weber O Selective induction of apoptosis in antigen-presenting cells in mice by Parapoxvirus ovis J Virol 2001 75 4699 4704 11312341 10.1128/JVI.75.10.4699-4704.2001
Esposito J Condit R Obijeski J The preparation of orthopoxvirus DNA J Virol Methods 1981 2 175 179 6268651 10.1016/0166-0934(81)90036-7
Sullivan J Mercer A Fleming S Robinson A Identification and Characterisation of an Orf virus Homologue of the Vaccina Virus Gene Encoding the Major Envelope Antigen p37K Virology 1994 202 968 973 8030257 10.1006/viro.1994.1420
Schleiermacher C Meyer F GeneFisher 2003 April 2004
Fleming SB McCaughan CA Andrews AE Nash AD Mercer A A homolog of interleukin-10 is encoded by the poxvirus orf virus J Virol 1997 71 4857 4861 9151886
Fleming B McCaughan CA Andrews AE Nash AD Mercer AA A homolog of interleukin-10 is encoded by the poxvirus orf virus Journal of Virology 1997 71 4857 4861 9151886
McInnes C Wood A A preliminary transcript map of the early genes of orf virus Unpublished 2003
Rziha HJ Bauer B Adam KH Rottgen M Cottone R Henkel M Dehio C Büttner M Relatedness and heterogeneity at the near-terminal end of the genome of a Parapoxvirus bovis 1 strain (B177) compared with Parapoxvirus ovis (Orf virus) J Gen Virol 2003 84 1111 1116 12692275 10.1099/vir.0.18850-0
Thompson J Gibson T Plewniak F Jeanmougin F Higgins D The ClustalX windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools Nucleic Acids Research 1997 24 4876 4882 9396791 10.1093/nar/25.24.4876
Huelsenbeck J Ronquist F MRBAYES: Bayesian inference of phylogeny Bioinformatics 2001 17 754 755 11524383 10.1093/bioinformatics/17.8.754
|
16143041
|
PMC1242257
|
CC BY
|
2021-01-04 16:38:59
|
no
|
Virol J. 2005 Sep 5; 2:79
|
utf-8
|
Virol J
| 2,005 |
10.1186/1743-422X-2-79
|
oa_comm
|
==== Front
Virol JVirology Journal1743-422XBioMed Central London 1743-422X-2-811615690010.1186/1743-422X-2-81ResearchInvolvement of PKR and RNase L in translational control and induction of apoptosis after Hepatitis C polyprotein expression from a Vaccinia virus recombinant Gómez Carmen E [email protected] Andrée Marie [email protected]ía María Angel [email protected] Elena [email protected] Mariano [email protected] Department of Molecular and Cellular Biology, Centro Nacional de Biotecnología, CSIC, Campus Universidad Autónoma, 28049 Madrid, Spain2005 12 9 2005 2 81 81 28 7 2005 12 9 2005 Copyright © 2005 Gómez et al; licensee BioMed Central Ltd.2005Gómez et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Hepatitis C virus (HCV) infection is of growing concern in public health with around 350 million chronically infected individuals worldwide. Although the IFN-α/rivabirin is the only approved therapy with 10–30% clinical efficacy, the protective molecular mechanism involved during the treatment is still unknown. To analyze the effect of HCV polyprotein expression on the antiviral response of the host, we developed a novel vaccinia virus (VV)-based delivery system (VT7-HCV7.9) where structural and nonstructural (except part of NS5B) proteins of HCV ORF from genotype 1b are efficiently expressed and produced, and timely regulated in mammalian cell lines.
Results
Regulated transcript production and viral polypeptide processing was demonstrated in various cell lines infected with the recombinant VT7-HCV7.9, indicating that the cellular and viral proteolytic machineries are functional within these cells. The inducible expression of the HCV polyprotein by VV inhibits the synthesis of both host and viral proteins over the time and also induces apoptosis in HeLa and HepG2-infected cells. These effects occur accompanying with the phosphorylation of the translation initiation factor eIF-2α. In cells co-infected with VT7-HCV7.9 and a recombinant VV expressing the dominant negative eIF-2α-S51A mutant in the presence of the inductor isopropyl-thiogalactoside (IPTG), protein synthesis is rescued. The IFN-inducible protein kinase PKR is responsible for the translational block, as demonstrated with PKR-/- and PKR+/+ cell lines. However, apoptosis induced by VT7-HCV7.9 is mediated by the RNase L pathway, in a PKR-independent manner.
Conclusion
These findings demonstrate the antiviral relevance of the proteins induced by interferon, PKR and RNase L during expression from a VV recombinant of the HCV polyprotein in human cell lines. HCV polyprotein expression caused a severe cytopathological effect in human cells as a result of inhibition of protein synthesis and apoptosis induction, triggered by the activation of the IFN-induced enzymes PKR and RNase L systems. Thus, the virus-cell system described here highlights the relevance of the IFN system as a protective mechanism against HCV infection.
==== Body
Background
The Hepatitis C virus (HCV) was identified as the causative agent for the majority of posttransfusion and sporadic non-A, and non-B hepatitis cases [1,2]. The World health organization (WHO) estimates that more than 3% of the world's population is infected with the virus. HCV belongs to the genus of Hepacivirus and is a member of the Flaviviridae family, along with Pestivirus and Flavivirus [3]. The HCV genome is a positively charged single stranded RNA molecule that includes two untranslated regions at the 5' and 3' ends, and a large open reading frame (ORF) encoding a 3010–3030 amino acid polyprotein that is co- and posttranslationally cleaved by cellular and viral proteases to produce mature structural (Core, E1, E2 and p7) and nonstructural (NS2, NS3, NS4A, NS4B, NS5A and NS5B) proteins [4,5]. One striking characteristic of HCV is its strong propensity to persist in the infected host, which often leads to severe liver damage, ranging from chronic hepatitis to liver cirrhosis and even hepatocellular carcinoma [6].
The IFN-α monotherapy became the mainstay for treatment of HCV infection until recently, when IFN-α/ribavirin, and pegylated IFN-α/ribavirin combination therapies became available [7]. The IFN-based regimens are still the only approved therapies for HCV [8]. Although the beneficial effect has been documented by numerous studies [9-11], only 10–40% of patients respond to treatment. The molecular mechanisms involved in protection during IFN therapy are not fully understood. Due to the clinical relevance of HCV infection and the differential responses of patients to IFN therapy, it is essential to investigate the molecular mechanisms involved in the sensitivity and resistance patterns of HCV infection in an appropriate model system.
In order to establish a robust in vitro infection model system for HCV, a variety of different approaches, mainly those based on infection with human patient sera of primary human liver cells or diverse cell lines of hepatic or lymphoid origin, have been explored [12,13]. Nonetheless, so far the success of these attempts has been limited due to the extremely low HCV replication levels that prevent detailed studies. The development of subgenomic HCV replicons that generates high-level replication of HCV RNAs in cell culture, has overcome this hurdle [14,15]. In spite of an efficient expression of the structural proteins and high levels of replication, it has not been possible to generate viral particles in cell cultures. Moreover, important information on the potential effect of the structural proteins on the host cell could not be obtained. An alternative approach has been viral delivery systems. In such systems, cells are transfected with a plasmid containing a cDNA clone under the control of a T7 promoter, and then infected with a virus that expresses T7 RNA polymerase. Although this approach has been met with some degree of success [16-18], it is limited by the efficiency with which the plasmid can be transfected into hosts cells. In the case of hepatocyte derived cell lines, the transfection efficiency is often rather low. This inefficiency could be overcome in certain cases, by using recombinant fowlpox viruses to deliver HCV minigenomes under the control of a T7 promoter into cells co-infected with an adenovirus expressing T7 RNA polymerase [19]. Although this system improved the efficiency of delivery, it was not possible to control HCV gene expression. Recently, a virus production system has been developed which is based on the transfection of the human hepatoma cell line Huh-7 with a genomic HCV RNA replicon derived from an individual with fulminant hepatitis [20]. The limited virus yields and virus spread of this cell culture system has been improved using a particular permissive cell line derived from Huh-7 designated Huh-7.5.1 [21]. This provides a significant advance in order to understand the biology of HCV infection in culture systems.
To characterize the antiviral response of the host during expression of the HCV polyprotein, we developed a novel poxvirus-based delivery system (VT7-HCV7.9), that is inducible and able to express structural and nonstructural (except part of NS5B) proteins of HCV ORF from genotype 1b in hepatic and non-hepatic mammalian cell lines. In this virus-cell system, we observed that HCV polyprotein expression controls cellular translation through eIF-2α-S51 phosphorylation, with involvement of the IFN-inducible double-stranded RNA-dependent protein kinase PKR. Moreover, in VT7-HCV7.9 infected cells, we found that HCV polyprotein expression brings about an apoptotic response through the activation of the RNase L pathway.
Results
Generation of a vaccinia virus recombinant expressing the near full-length HCV genome under regulation (VT7-HCV7.9)
In order to study the effect of HCV gene expression on host cellular mechanisms, we developed a novel system based on a poxvirus vector that when induced, expresses the structural and nonstructural (except part of NS5B) proteins of HCV ORF from genotype 1b. Briefly, BSC40 cells infected with the recombinant VT7lacOI virus, that inducibly expresses the T7 RNA polymerase, were transfected with the plasmid transfer vector pVOTE.1-HCV7.9. This transfer vector directs the insertion of the HCV DNA fragment into the viral hemagglutinin (HA) locus under the transcriptional control of the T7 promoter, to generate the recombinant VT7-HCV7.9 (Figure 1A). Upon induction with IPTG, the T7 RNA polymerase is expressed which in turn, allows the transcription of HCV genes in VT7-HCV7.9 infected cells.
Figure 1 Construction and characterization of the recombinant VT7-HCV7.9 virus. A: Generation of recombinant VT7-HCV7.9. A 7.9 Kb DNA fragment containing the structural (C, E1, E2 and p7) and nonstructural (NS3, NS4A, NS4B, NS5A and the amino terminal region of NS5B) proteins of HCV from genotype 1b was cloned into a unique EcoRI restriction site of pVOTE.1 to make the plasmid transfer vector pVOTE.1-HCV7.9. BSC40 cells infected with the recombinant VT7lacOI (VT7), were transfected with the plasmid pVOTE.1-HCV7.9 as described in Materials and Methods to generate the recombinant VT7-HCV7.9. B: Expression of HCV inhibits protein synthesis in mammalian cells. Monolayers of BSC40 cells were infected at 5 PFU/cell with either the parental VT7 or the recombinant VT7-HCV7.9 viruses in the presence (+) or absence (-) of the inductor IPTG. Uninfected (U) and infected cells were metabolically labelled with 35S-Met-Cys Promix (100 μCi/mL) from 4 to 24 h.p.i. as described in Materials and Methods. Approximately 100 μg of total cell protein extracted from uninfected (U) and infected cells, was fractionated by SDS-PAGE followed by autoradiography. (*) represents new additional polypeptides corresponding to the HCV proteins. C: Inducible expression of HCV proteins by recombinant VT7-HCV7.9 virus. BSC40 cells were infected as described above. Total cell protein lysates from uninfected (U) and infected cells at 24 h.p.i. were analysed by Western blot using a human anti-HCV antibody from an infected patient. The protein band migration of Core, E2, NS4B and NS5A, as determined with specific antibodies, is indicated.
To confirm expression of HCV proteins from the VV recombinant, we infected BSC40 cells with VT7-HCV7.9 and employed metabolic labelling, immunoblot and immunofluorescence microscopic analyses. Continuous metabolic labelling of BSC40 cells infected with VT7-HCV7.9 in the presence of IPTG, revealed by SDS-PAGE the synthesis of polypeptides not present in the absence of IPTG (Figure 1B, see new proteins denoted with asteriks). Significantly, in the presence of IPTG, overall protein synthesis was reduced in VT7-HCV7.9 infected cells when compared to protein synthesis in the absence of the inductor. This translational inhibitory effect was specific, since protein synthesis was not affected in cells infected with VT7, with or without IPTG (Figure 1B). The synthesis of HCV proteins in VT7-HCV7.9 infected cells was also documented by Western blot analysis, using sera from an HCV-infected patient. As shown in Figure 1C, HCV proteins of the expected size, for structural and nonstructural polypeptides, were detected only in VT7-HCV7.9 infected cells upon induction with IPTG. The size of specific HCV proteins was confirmed following reactivity with antibodies against Core, E2, NS4B and NS5A (not shown). A heterogeneous pattern of HCV-specific proteins was observed, perhaps as a result of different stages of proteolytic processing of the polyprotein. Confocal microscopy using sera from an infected patient revealed that the HCV proteins expressed in VT7-HCV7.9 infected cells upon induction with IPTG, formed large cytoplasmic aggregates and produced severe disruption of the golgi apparatus, a phenomenon not observed in cells infected in the absence of IPTG (Figure 2). The HCV proteins Core, E2, NS4B and NS5A were individually detected intracellularly with specific antibodies in VT7-HCV7.9 infected HeLa cells upon induction with IPTG (not shown).
Figure 2 Cellular localization of HCV proteins by immunofluorescence microscopy. Subconfluent HeLa cells were infected at 5 PFU/cell with the recombinant VT7-HCV7.9 in the presence (+) or absence (-) of the inductor IPTG. At 16 h.p.i, cells were doubly labelled with polyclonal antibody anti-Gigantine to detect the Golgi complex (red) and a 1/200 dilution of serum from an HCV-infected patient (green) followed by the appropriate fluorescent secondary antibody and ToPro reagent.
The results of Figures 1, 2 reveal that the HCV ORF included in the recombinant VT7-HCV7.9 is efficiently transcribed during infection in the presence of IPTG, generating a viral polyprotein that is processed into mature structural and nonstructural HCV proteins, triggering disruption of the golgi apparatus.
Expression of HCV polyprotein from VV inhibits the production of vaccinia virus
To determine the impact of HCV gene expression on the replication of the recombinant VT7-HCV7.9 virus, we studied the production of infectious VV at 12, 24 and 48 h.p.i, in the presence or absence of the inductor IPTG. As demonstrated in Figure 3 by virus plaque formation and virus titration curves, the production of infectious VV was significantly reduced (over 2 logs) during HCV gene expression. These results reveal that expression of HCV impairs VV replication.
Figure 3 Expression of HCV polyprotein inhibits the production of infectious VV. BSC40 cells were infected at 5 PFU/cell with the recombinant VT7-HCV7.9 in the presence or absence of IPTG. After the indicated times postinfection the cells were collected, centrifuged and resuspended in 300 μL of DMEM. After three freeze-thawing cycles, followed by sonication, the cell extracts were titrated in BSC40 cells. The experiment was performed two times in duplicate. Means and standard deviations are shown.
Expression of HCV polyprotein from VV inhibits cellular and viral protein synthesis through eIF-2α phosphorylation
Next, we determined the nature of the translational block in cells infected with VT7-HCV7.9 in the presence of IPTG. As a control, we included a recombinant VT7-VP3 inducibly expressing the IBDV capsid protein VP3. This virus was constructed similarly to VT7-HCV7.9, and expresses an mRNA encoding VP3 ORF from the vaccinia virus genome via T7 polymerase. Cells infected with VT7-HCV7.9, in the presence or absence of IPTG, were metabolically labelled for 30 min with 35S-Met-Cys Promix at 4, 8, 12 and 16 h.p.i., whole cell lysates fractionated by SDS-PAGE and the protein pattern examined by autoradiography. As shown in Figure 4, a clear reduction in cellular and viral protein synthesis was observed after 4 h.p.i in cells infected with the recombinant VT7-HCV7.9 virus in the presence of IPTG, in contrast with cells infected in the absence of the inductor, or in cells inducibly expressing the VP3 protein (Figure 4A). The protein levels were quantified by densitometry of the bands and are represented in Figure 4B. A strong decrease in protein synthesis becomes apparent by 8 h.p.i.
Figure 4 Time-course analysis of cellular and viral protein synthesis in cells expressing HCV polyprotein. A: BSC40 cells infected with the recombinant VT7-HCV7.9 virus in the presence (+) or absence (-) of IPTG were metabolically labelled with [35S] Met-Cys Promix (50 μCi/mL) at the indicated times (h.p.i) and analysed by SDS-PAGE (12%) and autoradiography. For comparative purposes, we included a similar inducible recombinant virus but expressing the IBDV mature structural capsid protein VP3 (VT7-VP3). B: Inhibition of VV proteins after expression of HCV. The levels of VV proteins were quantitated from autoradiograms using a BioRad GS700 image densitometer and computer software as suggested by the manufacturer. C: Immunoblot analysis of phospho-eIF-2α-S51 protein levels during the time-course of VT7-HCV7.9 infection. The number appearing in each lane represents the ratio of phospho-eIF-2α-S51 levels in infected cells compared to levels in uninfected cells.
Phosphorylation of the α subunit of the eukaryotic translation initiation factor 2 (eIF-2) on serine 51 leads to the downregulation of translation initiation through a well-characterized mechanism involving inhibition of eIF-2B activity [22]. As such, we determined whether HCV polyprotein expression altered this initiation step. Thus, the levels of phospho-eIF-2α-S51 in VT7-HCV7.9 infected cells, in the presence or absence of IPTG, were determined by immunoblot analysis. The results obtained showed that expression of HCV is related to levels of eIF-2α-S51 phosphorylation over time, relative to non-induced VT7-HCV7.9 infected cells (Figure 4C). Similar levels of phosphorylation have been shown to cause growth inhibitory effects in yeast, as well as in mammalian cells [23]. The levels of phospho-eIF-2α-S51 in VT7-VP3 infected cells in the presence of IPTG at the assayed times, were similar to the levels obtained in uninduced VT7-HCV7.9 infected cultures (Figure 4C), and represent the values usually found in VV-infected cells. A shorter time-course analysis of the extent of inhibition of protein synthesis and of eIF-2α-S51 phosphorylation indicates that such effects are clearly observed by 6 h.p.i in VT7-HCV7.9 infected cultures in the presence of IPTG (not shown).
To further assess the role of eIF-2α phosphorylation on the translational arrest, we examined whether expression of the dominant negative non-phosphorylated mutant Ser51-Ala (eIF-2α-S51A) was capable of rescuing the translation inhibitory effects of HCV gene expression. To this end, different combinations of recombinant viruses, VT7-HCV7.9, VT7 and VV-eIF2αNP (inducibly expressing the eIF-2α-S51A mutant), were assayed in the presence or absence of IPTG. The metabolic labelling of infected cells revealed that expression of eIF2α-S51A mutant in cells co-infected with VT7-HCV7.9 in the presence of IPTG, rescues the translational block caused after HCV polyprotein expression (Figure 5A: compare lanes 3, 4 and 6 with lanes 1 and 2). In the absence of IPTG, protein synthesis levels were not affected (Figure 5B).
Figure 5 Expression of the dominant negative eIF-2α-S51A mutant by VV-eIF2αNP rescues the translation inhibition induced by HCV polyprotein. BSC40 cells grown in 12-well plates were infected at a total of 9 PFU/cell with the viruses indicated in the presence or absence of IPTG (1.5 mM). At 18 h.p.i. the cells were metabolically labeled with [35S] Met-Cys Promix (50 μCi/mL) for 30 min. and analysed by SDS-PAGE (12%) and autoradiography.
The above findings demonstrate that the translational block induced after HCV polyprotein expression from VV involves eIF-2α phosphorylation.
HCV polyprotein expression from VV in the hepatic cell line HepG2 inhibits cellular and viral protein synthesis
The HCV is a hepatotropic virus, thus we set out to study the effects of HCV gene expression in a hepatoblast cell line. HepG2 cells were infected with VT7 or VT7-HCV7.9 in the presence or absence of IPTG, metabolically labelled with 35S-Met-Cys Promix from 4 to 24 h.p.i, cell extracts fractionated by SDS-PAGE, and the protein pattern visualized upon autoradiography analysis. As shown in Figure 6A, cells infected with the recombinant VT7-HCV7.9 virus in the presence of IPTG demonstrated the synthesis of new additional polypeptides corresponding to HCV proteins (confirmed by Western blot, not shown), with a marked reduction in protein synthesis, in comparison with cells infected in the absence of the inductor, or in those cells inducibly expressing the T7 RNA polymerase (VT7). Expression of HCV results in decreased levels of VV proteins, as shown by a Western blot using anti-VV antibodies (Figure 6B) and increased phosphorylation levels of eIF-2α-S51 (Figure 6C). These results indicate that HCV polyprotein expression from VV inhibits cellular and viral protein synthesis in hepatoblast cells, which correlates with eIF-2α-S51 phosphorylation.
Figure 6 Expression of HCV polyprotein from VV inhibits cellular and viral protein synthesis in the hepatic cell line HepG2. A: Monolayers of HepG2 cells were infected (5 PFU/cell) with either VT7 or VT7-HCV7.9 recombinant viruses, in the presence (+) or absence (-) of the inductor IPTG. Uninfected (U) and infected cells were metabolically labelled with [35S] Met-Cys Promix (100 μCi/mL) from 4 to 24 h.p.i and treated as described under Materials and Methods. Approximately 100 μg of total cell protein extracted from uninfected and infected cells was fractionated by SDS-PAGE followed by autoradiography. (*) represents new additional polypeptides corresponding to the HCV proteins. B: Immunoblot analysis of total cell protein lysates prepared from uninfected and infected cells at 24 h.p.i. The blot was probed with a rabbit polyclonal anti-serum raised against live VV. C: The blot was stripped and probed again with a polyclonal antibody that recognized phospho-eIF-2α-S51 protein.
Phosphorylation of eIF-2α and translational inhibition induced by HCV polyprotein expression from VV is mediated by PKR
Inhibition of translation through phosphorylation of eIF-2α, is a major stress-responsive checkpoint employed by at least four cellular kinases: PKR, PERK, GCN2, and HRI [24-27]. In particular of these four kinases, PKR has been shown to be the key regulator of cell defence against viral infections, and mediates the antiviral and antiproliferative effects of interferon (IFN) [28]. Activated PKR phosphorylates the α subunit of eIF-2 on serine 51, thus halting initiation of translation of both cellular and viral proteins that eventually leads to inhibition of viral replication [24].
In order to determine if PKR was the kinase responsible for eIF-2α phosphorylation following expression of HCV from VV, we infected PKR knockout cells (PKR-/-) and PKR WT cells (PKR+/+) with VT7 or VT7-HCV7.9 recombinant viruses in the presence of IPTG. As shown in Figure 7A, higher eIF-2α phosphorylation levels were observed in PKR+/+ than in PKR-/- cells after VT7-HCV7.9 infection. The total levels of eIF-2α and β-actin proteins were similar for both cell lines, in uninfected, as well as in VT7 or VT7-HCV7.9 infected cells. To corroborate whether eIF-2α phosphorylation halts translation of cellular and viral proteins, PKR-/- and PKR+/+ cells were infected with VT7-HCV7.9 in the presence or absence of IPTG, metabolically labelled, cell extracts fractionated by SDS-PAGE and proteins pattern visualized employing autoradiography. Only those PKR+/+ VT7-HCV7.9 infected cells in the presence of IPTG, showed a significant reduction of cellular and viral protein synthesis (Figure 7B). As expected, the expression of PKR by VV-PKR when used as a positive control, suppressed protein synthesis in both cell lines. Those data indicates that such cells are responsive to exogenous PKR delivered by VV.
Figure 7 PKR mediates phosphorylation of eIF-2α and inhibition of translation caused by the expression of HCV polyprotein. A: Immunoblot analysis of total cell protein lysates prepared from PKR knockout (PKR-/-) and PKR WT (PKR+/+) cells infected with the parental (VT7) or the recombinant VT7-HCV7.9 viruses in the presence (+) of IPTG for 24 h. The blot was first probed with a polyclonal antibody that recognized phospho-eIF-2α-S51 protein, stripped twice, and reprobed with a polyclonal antibody that recognizes total eIF-2α protein and a monoclonal antibody against β-actin. B: Wild type and PKR-/- cell lines infected with VT7-HCV7.9 in the presence (+) or absence (-) of IPTG were metabolically labelled with 35S-Met-Cys Promix (50 μCi/mL) at 16 h.p.i, fractionated by SDS-PAGE and analysed by autoradiography. The recombinant VV-PKR virus was used as a control. U: uninfected cells.
These findings reveal that PKR is the kinase responsible for eIF-2α phosphorylation as well as for the translational block following HCV polyprotein expression from VV in infected cells.
HCV polyprotein expression from VV induces apoptosis in HeLa and HepG2 cells, an effect that is caspase-dependent
It has been reported that expression in hepatic cells of all structural and nonstructural proteins from HCV cDNA [29] or from full-length RNA [30], can lead to apoptotic cell death, which may be an important event in the pathogenesis of chronic HCV infection in humans. To investigate whether apoptosis occurs in our virus-cell system, HeLa and HepG2 cells were infected with the recombinant VT7-HCV7.9 or coinfected with the recombinant VV-Bcl2 (that inducibly expresses the anti-apoptotic Bcl-2 polypeptide) in the presence or absence of IPTG. The levels of apoptosis were determined at 24 h.p.i (for HeLa cells) or at 48 h.p.i (for HepG2 cells), using an ELISA-based assay that detects the amount of cytoplasmic histone-associated DNA fragments. As shown in Figure 8 (panels A and B), expression of HCV by VT7-HCV7.9 in the presence of IPTG, induces apoptosis to levels similar to those obtained in induced VV-PKR-infected cells, used as a positive control. These apoptosis levels were two fold higher than those found in uninduced VT7-HCV7.9 infected cells. Co-expression from VV of HCV and of Bcl-2 in HeLa and HepG2 cells infected in the presence of IPTG, generates a two-fold reduction in apoptosis levels. A higher reduction in apoptosis was obtained by the Z-VAD-FMK general caspase inhibitor. These results revealed that HCV polyprotein expression from VV induced an apoptotic response, an effect mediated by caspases.
Figure 8 Expression of HCV polyprotein from VV induces apoptosis in HeLa and HepG2 cells that is caspase-dependent. A: HeLa cells were infected at 5 PFU/cell with the recombinant VT7-HCV7.9 individually or in combination (2.5 PFU of each virus/cell) with the recombinant VV-Bcl2 (inducibly expressing the anti-apoptotic Bcl-2 polypeptide) or with a general caspase inhibitor, Z-VAD-FMK (Calbiochem) at 50 μM, in the presence (+) or absence (-) of IPTG. The apoptotic levels were determined at 24 h.p.i by ELISA. B: HepG2 cells were infected at 10 PFU/cell with the recombinant VT7-HCV7.9 individually or in combination (5 PFU of each virus/cell) with the recombinant VV-Bcl2 or with a general caspase inhibitor, Z-VAD-FMK (Calbiochem) at 50 μM, in the presence (+) or absence (-) of IPTG. The apoptotic levels were determined at 48 h.p.i by ELISA. VV-PKR infected cells in the presence (+) of IPTG were used as positive controls.
Apoptosis induced by HCV polyprotein expression from VV is mediated by RNase L in a PKR-independent manner
In addition to PKR, the antiviral effects of IFN are executed through the functions of various proteins, including 2'5 oligoadenylate synthetase (2'-5AS), RNase L and Mx [31-34]. The 2'-5AS/RNase L and PKR pathways respond to dsRNA produced during the course of viral infections, to trigger an antiviral response in cells through RNA degradation and inhibition of protein synthesis. In contrast, Mx proteins obstruct the replicative cycles of particular negative strand RNA viruses by interfering with the intracellular movement and functions of viral proteins [28].
Once it was verified that PKR was the kinase responsible for eIF-2α phosphorylation and for the translational block following expression of HCV from VV, we assayed the activity of RNase L under the same conditions. HeLa cells were infected with VT7 or VT7-HCV7.9 recombinants in the presence or absence of IPTG for 24 h. Total RNA was fractionated in 1% agarose-formaldehyde gel and stained with ethidium bromide. As shown in Figure 9A, cells infected with VT7-HCV7.9 in the presence of IPTG exhibited ribosomal RNA degradation. This effect is mediated by RNase L since a similar pattern of rRNA cleavage products is observed by the co-expression of RNase L and 2-5AS delivered by the recombinant VVs, used as a positive control. In cells infected with either VT7 or VT7-HCV7.9 in the absence of IPTG, ribosomal RNAs were intact. The results of Figure 9A reveal that expression of HCV from VV induces the activation of RNase L.
Figure 9 Expression of HCV polyprotein from VV induces ribosomal RNA degradation mediated by RNaseL and triggers apoptosis through RNase L independently of PKR. A: Monolayers of HeLa cells were either uninfected (U), single-infected with VT7 (5 PFU/cell), single-infected with VT7-HCV7.9 (5 PFU/cell) in the presence (+) or absence (-) of IPTG, or triple-infected with VV-RL + VT7 + VV-25AS (2 PFU of each virus/cell) (C+). Infections proceeded for 24 hours. 2 μg of total RNA was fractionated in 1% agarose-formaldehyde gel and stained with ethidium bromide. Abundant ribosomal RNAs 28S and 18S are indicated. B and C: PKR knockout (PKR-/-) and PKR WT cells (PKR+/+) (panel B), as well as RNase L knockout (RL-/-) and RNase L WT cells (RL+/+) (panel C), were infected at 5 PFU/cell with the recombinant VT7-HCV7.9 virus, in the presence (+) or absence (-) of the inductor IPTG. The apoptotic levels in cell extracts were determined at 24 h.p.i. by ELISA. The recombinant VV-PKR virus was used as a control. U: Uninfected cells.
One interesting parallel between the PKR and 2-5A system is that both pathways contribute to apoptosis [35,36]. In order to compare the role of these pathways in the apoptotic response induced by HCV, we used PKR and RNase L knockout cells. PKR+/+ and PKR-/- as well as RL+/+ and RL-/- cells were infected with VT7 or VT7-HCV7.9 recombinants in the presence of IPTG, and the apoptotic levels were determined by ELISA at 24 h.p.i. As seen in Figure 9, expression of HCV by VT7-HCV7.9 induces apoptosis in PKR+/+ (Figure 9B) and RL+/+ cells (Figure 9C). The levels of apoptosis were similar to those obtained after the expression of PKR from VV-PKR, used as positive control. The levels of apoptosis induced by VT7-HCV7.9 after addition of IPTG, were significantly decreased in RL-/- infected cells (Figure 9C), while in PKR-/- cells, such levels remained similar to those in PKR+/+ cells (Figure 9B). These findings indicate that expression of HCV by VT7-HCV7.9 triggers apoptosis through RNase L, in a PKR-independent pathway.
Finally, we analysed cellular and viral protein synthesis in RNase L knockout cells expressing HCV. Consequently, RL+/+ and RL-/- cells were infected with VT7-HCV7.9 in the presence or absence of IPTG, metabolically labelled, cell extracts fractionated by SDS-PAGE and the pattern of proteins visualized using autoradiography. As shown in Figure 10, the expression of HCV provokes a similar reduction of cellular and viral protein synthesis in RL-/- and RL+/+ infected cells upon induction with IPTG (Figure 10A). This translational block correlates with increased levels of phosphorylation eIF-2α-S51 (Figure 10B) through PKR which is active in both cell lines. This result corroborates that apoptosis induced by HCV through RNase L is independent of the inhibition of protein synthesis caused by PKR.
Figure 10 Expression of HCV polyprotein from VV inhibits cellular and viral protein synthesis in RL+/+ and in RL-/- infected cells. A: RL+/+ and RL-/- cells infected with VT7-HCV7.9 in the presence (+) or absence (-) of IPTG were metabolically labelled with 35S-Met-Cys Promix (50 μCi/mL) at 8 h.p.i, fractionated by SDS-PAGE and analysed by autoradiography. U: uninfected cells. B: Immunoblot analysis of total cell protein lysates prepared from RL+/+ and RL-/- cells infected with VT7-HCV7.9 in the presence (+) or absence (-) of IPTG for 8 h. The blot was first probed with a polyclonal antibody that recognized phospho-eIF-2α-S51 protein, stripped and reprobed with a polyclonal antibody that recognizes total eIF-2α protein.
Discussion
Understanding the molecular mechanisms by which IFN-based therapies decreases HCV viral load, reduces the number of viral quasispecies, improves liver function, and reduces liver fibrosis in 15–30% of patients, is a priority in HCV research. Consequently, both viral and host factors have been implicated during the effective clinical response or resistance phenomenon of patients to IFN treatment [37]. Different in vitro model systems have been developed to study the role of HCV polyprotein on host cell responses [12-21]. The implication of IFN-induced genes and their action in the antiviral response of the host to HCV expression is not yet fully understood.
To further characterize the antiviral response of the host during expression of HCV polyprotein, we developed a novel virus-cell system based on a poxvirus vector, that inducibly expresses the structural and nonstructural (except part of NS5B) proteins of HCV ORF from genotype 1b. The generated recombinant VT7-HCV7.9 virus contains the HCV DNA coding region inserted within the VV HA locus, under the transcriptional control of a T7 promoter, and expresses the T7 RNA polymerase upon induction with IPTG (see Figure 1A). Current systems relying on viral delivery of T7 RNA polymerase are restricted by the efficiency with which HCV cDNAs can be transfected into cells, which in the case of hepatocyte and hepatocyte-derived cell lines, is often low [16-18]. The poxvirus-based system described here permits both the regulated production of the HCV transcripts into cells and the efficient delivery of the HCV genome into a wide variety of primary and continuous cell lines.
In this study, we demonstrate that upon induction with IPTG, HCV proteins are efficiently produced in VT7-HCV7.9 infected cells of various origins. This observation indicates that the DNA fragment of HCV ORF included in the VV genome, is efficiently transcribed and translated into a viral polyprotein precursor that is correctly processed into mature structural and nonstructural HCV proteins, as confirmed with specific antibodies to individual HCV proteins. Significantly, inducible expression of HCV polyprotein in VT7-HCV7.9 infected cells caused a considerable reduction in the production of infectious VV, as well as striking inhibition in total protein synthesis, both viral and cellular. The translational block was observed by 6 h.p.i when all of the HCV proteins were produced. The inhibition of protein synthesis by HCV was highly specific and could not be solely attributed to the induction of HCV RNA transcript since cells infected with VT7-VP3 that expressed the IBDV ORF VP3 mRNA, did not trigger translational inhibition. Furthermore, the HCV ORF included in the VT7-HCV7.9 recombinant virus lacks the 5' UTR, bearing the HCV IRES, and the 3' UTR, both implicated in HCV replication and liver injury [38]. The inhibition of protein synthesis that we have observed in induced VT7-HCV7.9 infected HeLa and HepG2 cells was associated with a significant increase in the phospho-eIF-2α-S51 levels, suggesting that HCV expression might control the cellular translation through eIF-2α-S51 phosphorylation. This translational control was confirmed with a dominant negative non-phosphorylated (NP) mutant Ser51-Ala (eIF-2α-S51A). Expression of the eIF-2α-S51A mutant in cells co-infected with VV-eIF-2α-NP and VT7-HCV7.9 in the presence of IPTG, rescued the translational block induced by HCV (Figure 6). Moreover, we showed that phosphorylation of eIF-2α-S51 was carried out by the cellular kinase PKR, as revealed in knockout PKR-/- cells (Figure 9). The role of PKR and eIF-2α-S51 phosphorylation in HCV infection has been widely studied due to the relevance of this kinase in the cellular antiviral response. As has been previously reported [23,39-41], PKR mediated phosphorylation of eIF-2α-S51 results in inhibition of translation and a blockade of viral protein synthesis, which in turn, inhibits virus replication. For this reason, viruses employ a variety of strategies to inhibit PKR activation and function. Several groups have described the role of certain HCV proteins in cellular translation. HCV NS4A and NS4B proteins mediate translational inhibition and, perhaps, increased degradation of certain cellular proteins [42,43]. In contrast, NS5A and E2 proteins are reported to enhance translation by inhibiting PKR functions [44,45]. Therefore, it seems that during the course of HCV infection, there is a balance between inhibition and enhancement of host cell translation depending on the degree of activation/inhibition of the PKR pathway. Most of these studies have relied on systems that express HCV proteins individually. Nontheless, since all HCV proteins are potentially produced in vivo during virus infection of hepatocytes, it is important to use a full-length genome rather than individual HCV proteins to study the molecular mechanisms involved in virus-host cell interactions and in HCV pathogenesis. In our viral delivery system, the overall expression of structural and nonstructural HCV proteins by recombinant VT7-HCV7.9 virus did not reverse the action of PKR, since host cell translation was inhibited through phosphorylation of eIF-2α-S51 by the kinase. An incapability to prevent PKR activation by HCV polyprotein expression was reported by François and co-workers when they analysed the response to IFN of the human cell line UHCV-11 engineered to inducibly express the entire HCV genotype 1a polyprotein [46]. Although we could not exclude the possibility that a certain level of inhibition of PKR by NS5A or E2 occurs at a much localized level, the resistance to IFN exhibited by some HCV genotypes as a result of viral protein expression, cannot be explained solely by inhibition of the negative control of PKR translation. It is possible that during the course of HCV infection, NS5A plays a role in inhibiting PKR locally at the site of HCV protein synthesis. NS5A may, however, participate in the blockade of IFN's antiviral action through another mechanism, such as the reported interaction with the Ras-associated Grb-2 protein [47]. These results confirm the necessity to re-evaluate all types of interactions between any particular HCV protein and its cellular partner(s) in the context of expression of all of the HCV proteins. Consequently, as shown here by confocal microscopy (Figure 2), the HCV proteins are localized within aggregates in the cell cytoplasm which might influence their interaction with PKR, a protein found surrounding the nucleus, in microsomes and in the nucleolus [24,48].
Several in vitro studies reveal that synthesis of HCV structural proteins or the full-length genome have a direct cytotoxic effect or activate an apoptotic response in osteosarcoma, hepatoma and B cell lines [29,30,49-51]. Furthermore, the alteration of ER membranes [52] and the activation of signalling pathways characteristic of an ER-stress condition, have been found to be associated with the expression of HCV proteins [53-55]. Although these data suggest that HCV may alter intracellular events with possible consequences on liver pathogenesis, the complex mechanism and the role of the viral proteins implicated are currently unknown. As we have shown in this work, expression of most of the HCV genome from VV induces a cell death phenomenon by apoptosis that should contribute to liver pathogenesis. Apoptosis induced by HCV polyprotein expression was prevented by Bcl-2 and by a general caspase inhibitor (Z-VAD-FMK) indicating a caspase-dependent death process. Even though PKR is the main kinase responsible for eIF-2α phosphorylation and for translation inhibition induced by the expression of HCV in VT7-HCV7.9 infected cells, it does not appear to be involved in apoptosis within this system, as revealed from studies performed in PKR+/+ and PKR-/- knockout cells. The extent of apoptosis induction by HCV expression was the same in PKR+/+ and in PKR-/- cells (Figure 9B), suggesting that other pathways may be involved. PKR induces apoptosis in response to activation by different stimuli, such as the accumulation of dsRNA as a by-product during virus replication [36], or when PKR is overexpressed in cells [56]. Several authors, however, have reported that PKR can also be activated through the binding of heparin and other polyanions [57,58], or by the cellular activator protein PACT/RAX [59,60]. The events that mediate induction of apoptosis by PKR have been widely studied and both PKR-induced translational block by phosphorylation of eIF-2α, and NF-kB activation, have been shown to be activated during apoptosis [61]. Since PKR has a number of potential substrates and signalling targets, it is likely that the phosphorylation of eIF-2α by PKR in response to HCV expression is not sufficient to mediate the pro-apoptotic effects of this kinase.
In this study, we also demonstrate the activation of endogenous RNase L and its role in the apoptosis induced by HCV expression (Figure 9A,C). Although it is widely accepted that the IFN-induced proteins PKR and RNase L require the expression of dsRNA for their activation (either directly in the case of PKR or indirectly via 2'-5'-OAS in the case of RNase L), there are several reports that documented the effect of HCV proteins on PKR and 2'-5'-OAS activation. The NS5A and E2 proteins can suppress the PKR pathway [44,45], whereas the Core protein can transcriptionally activate the 2'-5'-OAS gene through an IRES present within IFN-inducible gene promoter [62]. Like PKR, the 2-5AS/RNase L system can control virus growth by inducing apoptosis in response to viral infection [35,36]. Overexpression of RNase L or activation of the endogenous enzyme induces apoptosis by a mitochondrial-caspase dependent pathway that is suppressed by Bcl-2 [63-65]. Similarly, apoptosis induced by HCV polyprotein expression was inhibited by Bcl-2 (Figure 8). Although the apoptotic levels induced by HCV proteins remain invariable in PKR+/+, PKR-/-, and RL+/+ cells, the levels are significantly decreased in RL-/- cells, indicating that inducible expression of HCV proteins by VT7-HCV7.9 triggers apoptosis through RNase L in a PKR-independent pathway. Under physiologic conditions, RNase L activity is tightly regulated by 2'-phosphodiesterase and RNase L inhibitor [66,67] such that only a limited activation of RNase L occurs. The mechanism of the regulation of RNase L inhibitor is unknown, but the reduction of its expression seems to be advantageous for host defence together with the enhanced 2-5 OAS activity. Yu and co-workers [68] described that hepatic overexpression of PKR mRNA, and reduced expression of an RNase L inhibitor mRNA, are parameters that seem to contribute to an anti-HCV response. In agreement with our results, it has been reported that the absence of RNase L has an anti-apoptotic effect in multiple cell types treated with a variety of different agents [69]. The effects that have been observed in this study upon HCV polyprotein expression from VV are likely to have biological significance during HCV infection as there is ample evidence that VV recombinants can be used to study the function of multiple genes and that the assigned function mimics the effects described in non-viral systems [70].
Conclusion
We have developed an efficient viral delivery system expressing the polyprotein of HCV in numerous mammalian cell lines in a faithfully, efficient and time regulated manner, allowing us to analyze the host response to HCV proteins. We demonstrate that two components of the interferon (IFN) system, protein kinase PKR and RNase L, are activated during HCV polyprotein expression and are responsible for translational control and induction of apoptosis. These two pathways are likely to limit the replication capacity of HCV. Thus, the virus-cell system described here highlights the relevance of the IFN system as a protective mechanism against HCV infection.
Methods
Cells and viruses
Cells were maintained in a humidified air 5% CO2 atmosphere at 37°C. African green monkey kidney cells (BSC40) and human cells (HeLa) were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% newborn calf serum (NCS). Human HepG2 hepatocellular carcinoma cells (ATCC HB-8065) were maintained in DMEM supplemented with penicillin (0.6 μg/mL); streptomycin (60 μg/mL); glutamine (2 mM); N-2-hydroxyethylpiperazine-N'-2-ethanosulfonic acid (HEPES) buffer, pH 7.4 (20 mM) and 10% fetal calf serum (FCS). Mouse 3T3-like fibroblasts derived either from homozygous PKR knockout mice (PKR-/-) or PKR wild type mice (PKR+/+) [71] were obtained from C. Weissmann (University of Zurich, Switzerland) and grown in DMEM supplemented with 10% FCS. Wild type mouse embryo fibroblasts (MEFs) derived from C57BL6 mice (RL+/+) and fibroblast lacking the RNase L gene (RL-/- MEFs) derived from mice with the RNase L gene disrupted [35], were propagated in DMEM supplemented with 10% FCS, and were a gift from R. Silverman (Cleveland Clinic, USA)
The recombinant vaccinia virus (VV) that is inducible and expresses the T7 RNA polymerase (VT7lacOI) was previously described [72]. Virus VT7-VP3 expressing the IBDV mature structural capsid protein VP3 [73] was kindly provided by J.F. Rodríguez (CNB, Spain). VVeIF-2α NP was generated through homologous recombination in TK- 143B cells, as previously reported [74]. The recombinant VV-PKR TK- expressing IPTG-inducible PKR was generated by homologous recombination of their respective pPR35-derived plasmid with the WR strain of VV in BSC40 cells, as previously described [56]. VV recombinant expressing Bcl-2 protein (VV-Bcl2) was generated as previously reported [75]. The recombinant vaccinia viruses VV-RL and VV-2-5AS were obtained after introduction of plasmid pTM-RL and pSC-2-5AS respectively into the TK region of wild-type vaccinia virus (WR) DNA by homologous recombination as described [76]. All VV recombinants were grown in BSC40 cells and purified by banding on sucrose gradients [77].
Generation of the recombinant vaccinia VT7-HCV7.9 virus
A 7.9 Kb DNA fragment containing the structural (C, E1, E2 and p7) and nonstructural (NS2, NS3, NS4A, NS4B, NS5A and the amino terminal region of NS5B) proteins of HCV ORF from genotype 1b was excised with EcoRI from the original full-length HCV genome containing plasmid pcDNA-hcv1b (kindly provided by Ilkka Julkunen from National Public Health Institute, Finland). This DNA fragment was cloned into the VV insertion/expression vector pVOTE.1 [72] previously digested with EcoRI and dephosphorylated by incubation with alkaline phosphatase, Calf Intestinal (CIP) as described in Figure 1A. The resulting plasmid, pVOTE.1-HCV7.9 directs the insertion of HCV genes into the HA locus of the VT7lacOI genome under the transcriptional control of the T7 promoter. BSC40 cells were infected with the recombinant vaccinia virus VT7lacOI at a multiplicity of 0.05 PFU/cell, and then transfected with 10 μg of plasmid DNA pVOTE.1-HCV7.9 using lipofectamine reagent according to manufacturer's instructions (Invitrogen). The selection and amplification of the recombinant VT7-HCV7.9 virus was carried out as previously described [78]. The purity of the recombinant virus was confirmed by PCR analysis. The plasmid pVOTE.1 as well as the VV recombinant VT7lacOI, were kindly provided by Bernard Moss (NIH, USA).
Metabolic labelling of proteins
Different cell lines grown in 12 well plates were infected at an infection multiplicity of 5 PFU/cell with the viruses indicated, and maintained either in the presence or absence of the inductor isopropyl-β-D-thiogalactoside (IPTG) (1.5 mM final concentration). For continuous metabolic labelling of proteins, the cells were rinsed three times with Met-Cys-free DMEM at 4 h post-infection (p.i) and incubated with 100 μCi of [35S] Met-Cys Promix (Amersham) per mL in a mixture of Met-Cys-free DMEM and complete DMEM (9:1) for 16–20 h. After three washes with phosphate buffered saline (PBS) cells were resuspended in Laemmli buffer and analysed by sodium-dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) followed by autoradiography. For discontinuous metabolic labelling of proteins, the cells were rinsed three times and incubated with Met-Cys-free DMEM 30 minutes prior to labelling. After incubation, the medium was removed and 50 μCi of [35S] Met-Cys Promix per mL in Met-Cys-free DMEM was added for an additional 30 minutes. The cells were washed with PBS and treated as described above.
Immunoblotting
The HCV-antibody positive human sera used in this study was kindly provided by Dr Rafael Fernández from the Ramón and Cajal Hospital (Spain). The rabbit polyclonal anti-serum against live vaccinia virus was previously described [79]. The rabbit polyclonal anti eIF2α [PS51] phosphospecific antibody was supplied by BIOSOURCE. The monoclonal antibody against β-actin was supplied by SIGMA. Rabbit polyclonal anti-eIF2α antibody was supplied by Santa Cruz, CA.
For immunoblot analyses, total cell extracts were boiled in Laemmli sample buffer, and proteins were fractionated by 12% SDS-PAGE. After electrophoresis, proteins were transferred to nitrocellulose membranes using a semi-dry blotting apparatus (Gelman Sciences). Filters were mixed with antisera in PBS containing non-fat dry milk at 5% (BLOTTO), incubated overnight at 4°C, washed three times with PBS, and further incubated with secondary antibody coupled to horseradish peroxidase in BLOTTO. After the PBS wash, the immunocomplexes were detected by enhanced chemiluminescense Western blotting reagents (ECL) (Amersham).
Immunofluorescence
Specific antibody for Golgi apparatus (anti-Gigantine) was kindly provided by Manfred Renz from the Institute of Immunology and Genetics Karlsruhe (Germany).
HeLa cells cultured on coverslips were infected at 5 PFU/cell with VT7-HCV7.9 in the presence or absence of IPTG (1.5 mM final concentration). At 16 h.p.i, cells were washed with PBS, fixed with 4% paraformaldehyde and permeabilized with 2% Triton X-100 in PBS (room temperature, 5 min). Cells were incubated with a human antibody recognizing HCV proteins together with anti-Gigantine antibody. Coverslips were then extensively washed with PBS, and incubated in darkness for 1 h at 37°C, with secondary antibody conjugated with green fluorochrome Cy2 (Jackson Immunoresearch) and with the DNA staining reagent ToPro (Molecular Probes). Images were obtained by using Bio-Rad Radiance 2100 confocal laser microscope, were collected by using Lasersharp 2000 software and were processed in LaserPix.
Measurement of the extent of apoptosis
The cell death detection enzyme-linked immunosorbent assay (ELISA) kit (Roche) was used according to manufacturer's instructions. This assay is based on the quantitative sandwich enzyme immunoassay principle, and uses mouse monoclonal antibodies directed against DNA and histones to estimate the amount of cytoplasmic histone-associated DNA fragments.
Total RNA isolation
Total RNA from uninfected or infected cells was isolated using Ultraspect-II resin purification system (Biotecx). RNA was denatured and analyzed in 1% formaldehyde agarose gels and stained using ethidium bromide as previously described [76].
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
CEG has generated the vaccinia virus recombinant VT7-HCV7.9 and has analyzed protein expression in culture cells. AMV has performed confocal microscopy and defined apoptosis in infected cells. MAG has performed PKR and RNase L assays with KO cells. EDG has performed rRNA cleavage assays. ME conceived the study, has supervised the work, and provided the tools necessary for the performance of the research.
Acknowledgements
This investigation was supported by research grants BIO2000-0340-P4, BMC2002-03246 and Fundación Marcelino Botin from Spain and QLK22002-00954 from the European Union to ME. CEG was supported by a fellowship from Carolina Foundation and MAG from the Ministry of Science and Technology of Spain. We thank the expert technical assistance of Victoria Jiménez. We also thank JF Rodríguez, R. Bablanian and P. Martinez for critically reviewing the manuscript.
==== Refs
Choo QL Kuo G Weiner AJ Overby LR Bradley DW Houghton M Isolation of a cDNA clone derived from a blood-borne non-A, non-B viral hepatitis genome Science 1989 244 359 362 2523562
Kuo G Choo QL Alter HJ Gitnick GL Redeker AG Purcell RH Miyamura T Dienstag JL Alter MJ Stevens CE An assay for circulating antibodies to a major etiologic virus of human non-A, non-B hepatitis Science 1989 244 362 364 2496467
Robertson B Myers G Howard C Brettin T Bukh J Gaschen B Gojobori T Maertens G Mizokami M Nainan O Netesov S Nishioka K Shin i T Simmonds P Smith D Stuyver L Weiner A Classification, nomenclature, and database development for hepatitis C virus (HCV) and related viruses: proposals for standardization. International Committee on Virus Taxonomy Arch Virol 1998 143 2493 2503 9930205 10.1007/s007050050479
Bartenschlager R The NS3/4A proteinase of the hepatitis C virus: unravelling structure and function of an unusual enzyme and a prime target for antiviral therapy J Viral Hepat 1999 6 165 181 10607229 10.1046/j.1365-2893.1999.00152.x
Reed KE Rice CM Overview of hepatitis C virus genome structure, polyprotein processing, and protein properties Curr Top Microbiol Immunol 2000 242 55 84 10592656
Hoofnagle JH Hepatitis C: the clinical spectrum of disease Hepatology 1997 26 15S 20S 9305658 10.1002/hep.510260703
Fabrizi F Poordad FF Martin P Hepatitis C infection and the patient with end-stage renal disease Hepatology 2002 36 3 10 12085342 10.1053/jhep.2002.34613
Hoofnagle JH Management of hepatitis C: current and future perspectives J Hepatol 1999 31 Suppl 1 264 268 10622600 10.1016/S0168-8278(99)80414-0
Thomas HC Torok ME Forton DM Taylor-Robinson SD Possible mechanisms of action and reasons for failure of antiviral therapy in chronic hepatitis C J Hepatol 1999 31 Suppl 1 152 159 10622579 10.1016/S0168-8278(99)80393-6
Poynard T Moussalli J Ratziu V Regimbeau C Opolon P Effect of interferon therapy on the natural history of hepatitis C virus-related cirrhosis and hepatocellular carcinoma Clin Liver Dis 1999 3 869 881 11291256 10.1016/S1089-3261(05)70244-0
Manns MP McHutchison JG Gordon SC Rustgi VK Shiffman M Reindollar R Goodman ZD Koury K Ling M Albrecht JK Peginterferon alfa-2b plus ribavirin compared with interferon alfa-2b plus ribavirin for initial treatment of chronic hepatitis C: a randomised trial Lancet 2001 358 958 965 11583749 10.1016/S0140-6736(01)06102-5
Bartenschlager R Lohmann V Replication of hepatitis C virus J Gen Virol 2000 81 1631 1648 10859368
Kato N Shimotohno K Systems to culture hepatitis C virus Curr Top Microbiol Immunol 2000 242 261 278 10592664
Lohmann V Korner F Koch J Herian U Theilmann L Bartenschlager R Replication of subgenomic hepatitis C virus RNAs in a hepatoma cell line Science 1999 285 110 113 10390360 10.1126/science.285.5424.110
Ikeda M Yi M Li K Lemon SM Selectable subgenomic and genome-length dicistronic RNAs derived from an infectious molecular clone of the HCV-N strain of hepatitis C virus replicate efficiently in cultured Huh7 cells J Virol 2002 76 2997 3006 11861865 10.1128/JVI.76.6.2997-3006.2002
Chung RT He W Saquib A Contreras AM Xavier RJ Chawla A Wang TC Schmidt EV Hepatitis C virus replication is directly inhibited by IFN-alpha in a full-length binary expression system Proc Natl Acad Sci U S A 2001 98 9847 9852 11493707 10.1073/pnas.171319698
Mizuno M Yamada G Tanaka T Shimotohno K Takatani M Tsuji T Virion-like structures in HeLa G cells transfected with the full-length sequence of the hepatitis C virus genome Gastroenterology 1995 109 1933 1940 7498659 10.1016/0016-5085(95)90761-0
Myung J Khalap N Kalkeri G Garry R Dash S Inducible model to study negative strand RNA synthesis and assembly of hepatitis C virus from a full-length cDNA clone J Virol Methods 2001 94 55 67 11337040 10.1016/S0166-0934(01)00278-6
Yap CC Ishii K Aizaki H Tani H Aoki Y Ueda Y Matsuura Y Miyamura T Expression of target genes by coinfection with replication-deficient viral vectors J Gen Virol 1998 79 ( Pt 8) 1879 1888 9714235
Wakita T Pietschmann T Kato T Date T Miyamoto M Zhao Z Murthy K Habermann A Krausslich HG Mizokami M Bartenschlager R Liang TJ Production of infectious hepatitis C virus in tissue culture from a cloned viral genome Nat Med 2005 11 791 796 15951748 10.1038/nm1268
Zhong J Gastaminza P Cheng G Kapadia S Kato T Burton DR Wieland SF Uprichard SL Wakita T Chisari FV Robust hepatitis C virus infection in vitro Proc Natl Acad Sci U S A 2005 102 9294 9299 15939869 10.1073/pnas.0503596102
Rowlands AG Panniers R Henshaw EC The catalytic mechanism of guanine nucleotide exchange factor action and competitive inhibition by phosphorylated eukaryotic initiation factor 2 J Biol Chem 1988 263 5526 5533 3356695
Clemens MJ Hershey JWMMBSN Protein kinases that phosphorylate eIF2a and eIF2b and their role in eukaryotic cell translational control Translational control 1996 New York, Cold Spring Harbor Laboratory Press 139 172
Kaufman R Sonenberg NHJWMMB The double-stranded RNA-activated protein kinase PKR Translational control of gene expression 2000 New York, Cold Spring Harbor Laboratory Press 503 527
Chen JJ Sonenberg NHJWMMB Heme-regulated eIF2a kinase Translational control of gene expression 2000 New York, Cold Spring Harbor Laboratory Press 529 546
Hinnebusch AG Sonenberg NHJWMMB Mechanism and regulation of initiator methionyl-tRNA binding to ribosomes Translational control of gene expression 2000 New York, Cold Spring Harbor Laboratory Press 185 244
Ron DHHP Sonenberg NHJWMMB PERK and translational control by stress in the endoplasmic reticulum Translational control of gene expression 2000 New York, Cold Spring Harbor Laboratory Press 547 560
Stark GR Kerr IM Williams BR Silverman RH Schreiber RD How cells respond to interferons Annu Rev Biochem 1998 67 227 264 9759489 10.1146/annurev.biochem.67.1.227
Kalkeri G Khalap N Akhter S Garry RF Fermin CD Dash S Hepatitis C viral proteins affect cell viability and membrane permeability Exp Mol Pathol 2001 71 194 208 11733945 10.1006/exmp.2001.2392
Kalkeri G Khalap N Garry RF Fermin CD Dash S Hepatitis C virus protein expression induces apoptosis in HepG2 cells Virology 2001 282 26 37 11259187 10.1006/viro.2000.0835
Staeheli P Pavlovic J Inhibition of vesicular stomatitis virus mRNA synthesis by human MxA protein J Virol 1991 65 4498 4501 1649344
Sen GC Ransohoff RM Interferon-induced antiviral actions and their regulation Adv Virus Res 1993 42 57 102 7679242
Hassel BA Zhou A Sotomayor C Maran A Silverman RH A dominant negative mutant of 2-5A-dependent RNase suppresses antiproliferative and antiviral effects of interferon Embo J 1993 12 3297 3304 7688298
Li XL Blackford JA Hassel BA RNase L mediates the antiviral effect of interferon through a selective reduction in viral RNA during encephalomyocarditis virus infection J Virol 1998 72 2752 2759 9525594
Zhou A Paranjape J Brown TL Nie H Naik S Dong B Chang A Trapp B Fairchild R Colmenares C Silverman RH Interferon action and apoptosis are defective in mice devoid of 2',5'-oligoadenylate-dependent RNase L Embo J 1997 16 6355 6363 9351818 10.1093/emboj/16.21.6355
Der SD Yang YL Weissmann C Williams BR A double-stranded RNA-activated protein kinase-dependent pathway mediating stress-induced apoptosis Proc Natl Acad Sci U S A 1997 94 3279 3283 9096384 10.1073/pnas.94.7.3279
Hu KQ Vierling JM Redeker AG Viral, host and interferon-related factors modulating the effect of interferon therapy for hepatitis C virus infection J Viral Hepat 2001 8 1 18 11155147 10.1046/j.1365-2893.2001.00253.x
Chang M Marquardt AP Wood BL Williams O Cotler SJ Taylor SL Carithers RLJ Gretch DR In situ distribution of hepatitis C virus replicative-intermediate RNA in hepatic tissue and its correlation with liver disease J Virol 2000 74 944 955 10623757 10.1128/JVI.74.2.944-955.2000
Hershey JW Translational control in mammalian cells Annu Rev Biochem 1991 60 717 755 1883206 10.1146/annurev.bi.60.070191.003441
Merrick WCHJW Hershey JWMMBSN The pathway and mechanism of eukaryotic protein syntesis Translational control 1996 New York, Cold Spring Harbor Laboratory Press 31 70
Korth MJ Katze MG Evading the interferon response: hepatitis C virus and the interferon-induced protein kinase, PKR Curr Top Microbiol Immunol 2000 242 197 224 10592662
Florese RH Nagano-Fujii M Iwanaga Y Hidajat R Hotta H Inhibition of protein synthesis by the nonstructural proteins NS4A and NS4B of hepatitis C virus Virus Res 2002 90 119 131 12457968 10.1016/S0168-1702(02)00146-6
Kato J Kato N Yoshida H Ono-Nita SK Shiratori Y Omata M Hepatitis C virus NS4A and NS4B proteins suppress translation in vivo J Med Virol 2002 66 187 199 11782927 10.1002/jmv.2129
Gale MJJ Korth MJ Tang NM Tan SL Hopkins DA Dever TE Polyak SJ Gretch DR Katze MG Evidence that hepatitis C virus resistance to interferon is mediated through repression of the PKR protein kinase by the nonstructural 5A protein Virology 1997 230 217 227 9143277 10.1006/viro.1997.8493
Taylor DR Shi ST Romano PR Barber GN Lai MM Inhibition of the interferon-inducible protein kinase PKR by HCV E2 protein Science 1999 285 107 110 10390359 10.1126/science.285.5424.107
Francois C Duverlie G Rebouillat D Khorsi H Castelain S Blum HE Gatignol A Wychowski C Moradpour D Meurs EF Expression of hepatitis C virus proteins interferes with the antiviral action of interferon independently of PKR-mediated control of protein synthesis J Virol 2000 74 5587 5596 10823866 10.1128/JVI.74.12.5587-5596.2000
Tan SL Nakao H He Y Vijaysri S Neddermann P Jacobs BL Mayer BJ Katze MG NS5A, a nonstructural protein of hepatitis C virus, binds growth factor receptor-bound protein 2 adaptor protein in a Src homology 3 domain/ligand-dependent manner and perturbs mitogenic signaling Proc Natl Acad Sci U S A 1999 96 5533 5538 10318918 10.1073/pnas.96.10.5533
Besse S Rebouillat D Marie I Puvion-Dutilleul F Hovanessian AG Ultrastructural localization of interferon-inducible double-stranded RNA-activated enzymes in human cells Exp Cell Res 1998 239 379 392 9521856 10.1006/excr.1997.3908
Moradpour D Kary P Rice CM Blum HE Continuous human cell lines inducibly expressing hepatitis C virus structural and nonstructural proteins Hepatology 1998 28 192 201 9657112 10.1002/hep.510280125
Moradpour D Wakita T Wands JR Blum HE Tightly regulated expression of the entire hepatitis C virus structural region in continuous human cell lines Biochem Biophys Res Commun 1998 246 920 924 9618312 10.1006/bbrc.1998.8727
Sung VM Shimodaira S Doughty AL Picchio GR Can H Yen TS Lindsay KL Levine AM Lai MM Establishment of B-cell lymphoma cell lines persistently infected with hepatitis C virus in vivo and in vitro: the apoptotic effects of virus infection J Virol 2003 77 2134 2146 12525648 10.1128/JVI.77.3.2134-2146.2003
Egger D Wolk B Gosert R Bianchi L Blum HE Moradpour D Bienz K Expression of hepatitis C virus proteins induces distinct membrane alterations including a candidate viral replication complex J Virol 2002 76 5974 5984 12021330 10.1128/JVI.76.12.5974-5984.2002
Tardif KD Mori K Siddiqui A Hepatitis C virus subgenomic replicons induce endoplasmic reticulum stress activating an intracellular signaling pathway J Virol 2002 76 7453 7459 12097557 10.1128/JVI.76.15.7453-7459.2002
Pavio N Romano PR Graczyk TM Feinstone SM Taylor DR Protein synthesis and endoplasmic reticulum stress can be modulated by the hepatitis C virus envelope protein E2 through the eukaryotic initiation factor 2alpha kinase PERK J Virol 2003 77 3578 3585 12610133 10.1128/JVI.77.6.3578-3585.2003
Liberman E Fong YL Selby MJ Choo QL Cousens L Houghton M Yen TS Activation of the grp78 and grp94 promoters by hepatitis C virus E2 envelope protein J Virol 1999 73 3718 3722 10196264
Lee SB Esteban M The interferon-induced double-stranded RNA-activated protein kinase induces apoptosis Virology 1994 199 491 496 7510087 10.1006/viro.1994.1151
George CX Thomis DC McCormack SJ Svahn CM Samuel CE Characterization of the heparin-mediated activation of PKR, the interferon-inducible RNA-dependent protein kinase Virology 1996 221 180 188 8661426 10.1006/viro.1996.0364
Hovanessian AG Galabru J The double-stranded RNA-dependent protein kinase is also activated by heparin Eur J Biochem 1987 167 467 473 3653103 10.1111/j.1432-1033.1987.tb13360.x
Ito T Yang M May WS RAX, a cellular activator for double-stranded RNA-dependent protein kinase during stress signaling J Biol Chem 1999 274 15427 15432 10336432 10.1074/jbc.274.22.15427
Patel RC Sen GC PACT, a protein activator of the interferon-induced protein kinase, PKR Embo J 1998 17 4379 4390 9687506 10.1093/emboj/17.15.4379
Gil J Alcami J Esteban M Induction of apoptosis by double-stranded-RNA-dependent protein kinase (PKR) involves the alpha subunit of eukaryotic translation initiation factor 2 and NF-kappaB Mol Cell Biol 1999 19 4653 4663 10373514
Naganuma A Nozaki A Tanaka T Sugiyama K Takagi H Mori M Shimotohno K Kato N Activation of the interferon-inducible 2'-5'-oligoadenylate synthetase gene by hepatitis C virus core protein J Virol 2000 74 8744 8750 10954577 10.1128/JVI.74.18.8744-8750.2000
Castelli JC Hassel BA Wood KA Li XL Amemiya K Dalakas MC Torrence PF Youle RJ A study of the interferon antiviral mechanism: apoptosis activation by the 2-5A system J Exp Med 1997 186 967 972 9294150 10.1084/jem.186.6.967
Diaz-Guerra M Rivas C Esteban M Activation of the IFN-inducible enzyme RNase L causes apoptosis of animal cells Virology 1997 236 354 363 9325243 10.1006/viro.1997.8719
Rusch L Zhou A Silverman RH Caspase-dependent apoptosis by 2',5'-oligoadenylate activation of RNase L is enhanced by IFN-beta J Interferon Cytokine Res 2000 20 1091 1100 11152576 10.1089/107999000750053762
Benoit De Coignac A Bisbal C Lebleu B Salehzada T cDNA cloning and expression analysis of the murine ribonuclease L inhibitor Gene 1998 209 149 156 9524254 10.1016/S0378-1119(98)00040-7
Martinand C Salehzada T Silhol M Lebleu B Bisbal C The RNase L inhibitor (RLI) is induced by double-stranded RNA J Interferon Cytokine Res 1998 18 1031 1038 9877446
Yu SH Nagayama K Enomoto N Izumi N Marumo F Sato C Intrahepatic mRNA expression of interferon-inducible antiviral genes in liver diseases: dsRNA-dependent protein kinase overexpression and RNase L inhibitor suppression in chronic hepatitis C Hepatology 2000 32 1089 1095 11050060 10.1053/jhep.2000.19287
Houge G Robaye B Eikhom TS Golstein J Mellgren G Gjertsen BT Lanotte M Doskeland SO Fine mapping of 28S rRNA sites specifically cleaved in cells undergoing apoptosis Mol Cell Biol 1995 15 2051 2062 7891700
Gil J Esteban M Vaccinia virus recombinants as a model system to analyze interferon-induced pathways J Interferon Cytokine Res 2004 24 637 646 15684816
Yang YL Reis LF Pavlovic J Aguzzi A Schafer R Kumar A Williams BR Aguet M Weissmann C Deficient signaling in mice devoid of double-stranded RNA-dependent protein kinase Embo J 1995 14 6095 6106 8557029
Ward GA Stover CK Moss B Fuerst TR Stringent chemical and thermal regulation of recombinant gene expression by vaccinia virus vectors in mammalian cells Proc Natl Acad Sci U S A 1995 92 6773 6777 7624318
Fernandez-Arias A Martinez S Rodriguez JF The major antigenic protein of infectious bursal disease virus, VP2, is an apoptotic inducer J Virol 1997 71 8014 8018 9311897
Lee SB Esteban M The interferon-induced double-stranded RNA-activated human p68 protein kinase inhibits the replication of vaccinia virus Virology 1993 193 1037 1041 8096351 10.1006/viro.1993.1223
Lee SB Rodriguez D Rodriguez JR Esteban M The apoptosis pathway triggered by the interferon-induced protein kinase PKR requires the third basic domain, initiates upstream of Bcl-2, and involves ICE-like proteases Virology 1997 231 81 88 9143305 10.1006/viro.1997.8494
Diaz-Guerra M Rivas C Esteban M Inducible expression of the 2-5A synthetase/RNase L system results in inhibition of vaccinia virus replication Virology 1997 227 220 228 9007077 10.1006/viro.1996.8294
Esteban M Defective vaccinia virus particles in interferon-treated infected cells Virology 1984 133 220 227 6702105 10.1016/0042-6822(84)90443-4
Earl PLMB Ausubel FMBRKREMDSJLSJSK Generation of recombinant vaccinia viruses Current protocols in Molecular Biology 1993 2 New York, John Wiley and Sons 16.17.1 16.18.10
Demkowicz WE Maa JS Esteban M Identification and characterization of vaccinia virus genes encoding proteins that are highly antigenic in animals and are immunodominant in vaccinated humans J Virol 1992 66 386 398 1727494
|
16156900
|
PMC1242258
|
CC BY
|
2021-01-04 16:39:00
|
no
|
Virol J. 2005 Sep 12; 2:81
|
utf-8
|
Virol J
| 2,005 |
10.1186/1743-422X-2-81
|
oa_comm
|
==== Front
World J Surg OncolWorld Journal of Surgical Oncology1477-7819BioMed Central London 1477-7819-3-601615940510.1186/1477-7819-3-60Case ReportIntra-abdominal angiosarcoma developing in a capsule of a foreign body: report of a case with associated hemorrhagic diathesis Joo Young-Tae [email protected] Chi-Young [email protected] Eun-Jung [email protected] Young-Joon [email protected] Soon-Chan [email protected] Sang-Kyung [email protected] Soon-Tae [email protected] Woo-Song [email protected] Department of Surgery, Gyeongsang National University Collage of Medicine, Jinju, South Korea2005 14 9 2005 3 60 60 1 4 2005 14 9 2005 Copyright © 2005 Joo et al; licensee BioMed Central Ltd.2005Joo et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Backgrounds
Angiosarcoma occurs very rarely in the gastrointestinal tract and can present great diagnostic difficulty, especially when it is associated with intraabdominal abscess or granulation tissue.
Case presentation
We report a case where the angiosarcoma was diagnosed after the occurrence of disseminated angiosarcoma and concurrent hemoperitoneum. The tumor developed in the fibrous capsule of a foreign body, which was possibly related to the previous appendectomy twenty years ago, and became a widely disseminated malignant neoplasm in the abdomen. After the operation, the patient's course was dominated by a fatal consumptive coagulapathy. Pathologic examination of the multiple intra-abdominal lesions showed the histological and immunohistological characteristics of the angiosarcoma.
Conclusion
Even though angiosarcoma in the gastrointestinal tract is extremely rare, when dealing with intraabdominal abscess or the gastrointestinal bleeding in patients who have undergone surgery or radiation therapy in the past, the possibility of angiosarcoma should be considered. To make the definite diagnosis of angiosarcoma and to avoid the misdiagnosis of foreign body granuloma, thorough histological examination and immunohistochemical staining may be prerequisite.
==== Body
Background
Angiosarcoma is a rare malignant tumor with the incidence of 1%–2% of all sarcomas. It occurs in the skin and subcutis in most cases, and less commonly, it occurs in the liver [1], spleen [2], adrenal gland, and the ovary. Its occurrence in the gastrointestinal tract is extremely rare [3,4], and the development of hemoperitoneum is even more rare in the worldwide [2]. Due to the diagnostic difficulties, intraperitoneal metastases are already present in many patients at the time of diagnosis. For the differential diagnosis, histological findings including immunohistochemistry are most important. We report a case of intraperitoneal disseminated angiosarcoma presenting as hemoperitoneum, which was initially misdiagnosed as intraabdominal abscess.
Case presentation
A 61 year-old male was admitted for constipation and a palpable mass in the periumbilical area. His medical history revealed the surgery for acute perforative appendicitis 20 years ago and occasional intestinal obstruction afterwards. He denied any history of radiation or occupational exposure to chemicals. The small bowel series showed an external compressive lesion in the distal ileum and the computerized tomographic (CT) scan revealed a 6 cm, well-marginated mass in the anteromedial aspect of the ascending colon (Figure 1). Under the suspicion of foreign body granuloma due to gauze (gossypibioma), an exploratory laparotomy was performed. The whole intestine was found to be covered in severe adhesion; however, ascites or blood in the abdominal cavity was absent. In the anteromedial aspect of ascending colon, there was an encapsulated mass 5 cm in size, and it was adhered to ascending colon and distal ileum (Figure 2). The mass was filled with abscess and granulation tissue, and the frozen biopsy reported the diagnosis of abscess. Only resection of the mass was performed, and the patient was discharged without any complications.
Figure 1 Enhanced abdominal CT scan of upper abdomen showing a 6 cm sized mass in the right lower quadrant of the abdomen at the first operation. The mass contained air bubbles, tiny calcifications and hematoma.
Figure 2 Macroscopic view of the mass in the right paracolic gutter showing a 5 × 3 cm sized, encapsulated mass filled with abscess and granulation tissue.
Forty days after the discharge, he was admitted again with anemia, abdominal distension, and melana. The colonoscopy revealed the hematoma and the stenosis of the lumen directly above the cecum. The abdominal CT scan showed large volume of blood in the abdomen and multiple peritoneal nodules. The angiography and 99 mTc labeled RBC scan showed active bleeding around the ileocecal valve. Emergent laparotomy was performed and multiple nodules were found on the wall of ileum, liver, mesentery and peritoneum. In the distal ileum, two nodular lesions with 3 cm and 2 cm in size were bleeding actively, and they were adhered to each other. The nodules were diagnosed as sarcoma through the frozen biopsy, and the distal ileum including the pathologic lesion was resected. Seven days later, reoperation was performed due to substantial hemorrhage from the peritoneum, the mesentery, and the small intestine wall. Two days after the last surgery, the patient expired of uncontrollable bleeding due to disseminated intravascular coagulopathy.
On the macroscopic examination of resected ileum, there were two ill-defined tan solid tumors with mucosal ulceration, each measured 3 × 1.5 cm and 2 cm in diameter, involving the entire intestinal wall and extending to the subserosa of adhered loop. (Figure 3).
Figure 3 Macroscopic view of the segment of the small intestine showing an ill-defined tan solid mass that involved the entire intestinal layer.
Under microscope, spindle-shaped or epithelioid cells were arranged as a plate and the rudimentary vessel lumen were detected occasionally (Figure 4a). Separated from these two lesions, several small angiosarcomas containing foreign body granulomas were found in the subserosal layer of the intestine. Also, the nodules of liver and mesentery were diagnosed as metastatic angiosarcomas.
Figure 4 Microscopic findings of the angiosarcoma in the small intestine. 4a) Atypical spindle or epithelioid tumor cells were arranged in sheets, and rudimentary lumen formation was rarely noted. (H&E, × 200) 4b) The tumor cells are strongly positive for anti-CD31. (PAP, × 100)
Immunohistochemical staining was performed and tumor cells were positive for CD31, CD34, and vimentin (Figure 4b), whereas negative for factor antigen, CD117, and S-100. The tumor cells were also negative for cytokeratin (AE1/3) and EMA. The foreign body granulomas were surrounded by CD31 positive cells, partially or entirely.
The previously resected mass, which was diagnosed as abscess, was reviewed. In a low magnification field, abscess in the center and fibrosis with vascular proliferation in the periphery were noted (Figure 5). However, when the periphery was examined under the high magnification, spindle-shaped or epithelioid cells were arranged as plate patterns, in some area, well-differentiated vessels were formed as similar to the lesions from the ileum. The additional serial sections revealed more foreign materials surrounded by epithelioid tumor cells and the invasion of tumor cells to the blood vessel. The tumor cells were positive for CD31 and CD34, and negative for cytokeratin. This supported the final diagnosis of a foreign body granuloma associated-angiosarcoma.
Figure 5 Microscopic finding of the angiosarcoma in the anteromedial aspect of ascending colon. Proliferation of malignant blood vessels was seen in the periphery of the abscess(left upper corner) (H&E, × 40). Inlet: Malignant epitheliod tumor cells were proliferating around the calcified foreign material, which was found in the periphery of the abscess. (H&E, × 200)
Discussion
Although the causality has not been clearly elucidated yet, several factors have been reported to be related to the development of angiosarcoma. Vinyl chloride, arsenic, and thorium dioxide have been reported to cause angiosarcoma in the liver [5]. The radiation and chronic inflammation are important predisposing factors for angiosarcoma [6,7], and rarely, angiosarcoma may develop in association with foreign materials, especially metals, Dacron graft or a retained surgical sponge. According to our review of the literature, 19 cases of angiosarcoma caused by a foreign material have been reported with 4 cases developing in the abdominal cavity. All of these 4 cases developed in relation to a gauze retained for a prolonged period after previous abdominal surgery [8-11]. The formation of fibrotic capsule surrounding the vicinity of foreign body has been reported to be an important factor for developing angiosarcomas [8,9]. In present case, when considering the facts that foreign material was identified in the nodules of ileum and ascending colon, and also in the previously resected tumor, and that there was a close topograghic association of the previous appendectomy site with the developing site of the angiosarcoma, it can be speculated that the foreign material from the previous appendectomy was the cause of foreign body associated angiosarcoma, whether the foreign material was gauze or suture material.
Gastrointestinal angiosarcoma usually presents with abdominal pain, bleeding, and obstruction. Especially in cases of small bleeding angiosarcoma in the small intestine, the angiography or 99 mTc-labelled RBC scan may be needed to identify the bleeding focus [12]. In present case, because of the invasion of tumor cells through the serosa, hemorrhage into the abdominal cavity and local metastasis may have developed more readily rather than the gastrointestinal bleeding.
For the definite diagnosis, histological findings are most important and immunohistochemical staining is necessary for the differential diagnosis from undifferentiated malignant tumor, malignant melanoma, and smooth muscle sarcoma [13]. The cells of angiosarcoma are positively stained for vimentin, the endothelial cell markers such as CD31, CD34, and factor, and negatively for epithelial marker such as cytokeratin and EMA, however, in the epithelioid angiosarcoma, it may be stained as positive [14].
Many reported various staining findings in angiosarcoma, and such diverse staining findings imply the diversity for the degree of differentiation seen in angiosarcoma and the variation for the expression of the markers [13,14].
Conclusion
Angiosarcoma in the gastrointestinal tract is extremely rare, and it can present with diverse symptoms. If the intraabdominal abscess or the gastrointestinal bleeding is detected in patients who have undergone surgery or radiation therapy in the past, the possibility of angiosarcoma should be considered. To make the definite diagnosis of angiosarcoma and to avoid the misdiagnosis of foreign body granuloma, thorough histological examination and immunohistochemical staining may be prerequisite.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
YTJ participated in the serial operations of the patient, prepared the manuscript for submission and resubmission
CYJ participated in the serial operations of the patient, prepared the manuscript for submission and resubmission
EJ participated in the serial operations of the patient, prepared the manuscript for submission and resubmission
YJL participated in the serial operations of the patient, prepared the manuscript for submission and resubmission
SCH participated in the serial operations of the patient, prepared the manuscript for submission and resubmission
SKC, participated in the serial operations of the patient, prepared the manuscript for submission and resubmission
WH participated in the serial operations of the patient, prepared the manuscript for submission and resubmission
STP participated in the serial operations of the patient, prepared the manuscript for submission and resubmission, and helped in the final editing process.
All authors approved the final version.
Acknowledgements
Written consent was obtained from the patient's family for publication of the case.
==== Refs
Neshiwat LF Friedland ML Schorr-Lesnick B Feldman S Glucksman WJ Russo RD Jr Hepatic angiosarcoma Am J Med 1992 93 219 222 1497020 10.1016/0002-9343(92)90054-F
Ghani M Coughlin BF Hickey KL Reynolds DR Angiosarcoma: unusual cause of acute abdominal pain and hemoperitoneum Emergency radiology 2000 7 308 311
Brown CJ Falck VG MacLean A Angiosarcoma of the colon and rectum: report of a case and review of the literature Dis Colon Rectum 2004 47 2202 2207 15657674 10.1007/s10350-004-0698-5
Chami TN Ratner LE Henneberry J Smith DP Hill G Katz PO Angiosarcoma of the small intestine: a case report and literature review Am J Gastroenterol 1994 89 797 800 8172159
Lee FI Smith PM Bennett B Williams DM Occupationally related angiosarcoma of the liver in the United Kingdom 1972–1994 Gut 1996 39 312 318 8977349
Aitola P Poutiainen A Nordback I Small-bowel angiosarcoma after pelvic irradiation: a report of two cases Int J Colorectal Dis 1999 14 308 310 10663901 10.1007/s003840050235
Aozasa K Naka N Tomita Y Ohsawa M Kanno H Uchida A Ono K Angiosarcoma developing from chronic pyothorax Mod Pathol 1994 7 906 911 7892158
Ben-Izhak O Kerner H Brenner B Lichtig C Angiosarcoma of the colon developing in a capsule of a foreign body. Report of a case with associated hemorrhagic diathesis Am J Clin Pathol 1992 97 416 420 1543166
Jennings TA Peterson L Axiotis CA Friedlaender GE Cooke RA Rosai J Angiosarcoma associated with foreign body material. A report of three cases Cancer 1988 62 2436 2444 3052791
Cokelaere K Vanvuchelen J Michielsen P Sciot R Epithelioid angiosarcoma of the splenic capsule. Report of a case reiterating the concept of inert foreign body tumorigenesis Virchows Arch 2001 438 398 403 11355176 10.1007/s004280000324
Keymeulen K Dillemans B Epitheloid angiosarcoma of the splenic capsula as a result of foreign body tumorigenesis. A case report Acta Chir Belg 2004 104 217 220 15154584
Allison KH Yoder BJ Bronner MP Goldblum JR Rubin BP Angiosarcoma involving the gastrointestinal tract Am J Surg Pathol 2004 28 298 307 15104292
Ordonez NG del Junco GW Ayala AG Ahmed N Angiosarcoma of the small intestine: an immunoperoxidase study Am J Gastroenterol 1983 78 218 221 6404159
Delvaux V Sciot R Neuville B Moerman P Peeters M Filez L Van Beckevoort D Ectors N Geboes K Multifocal epithelioid angiosarcoma of the small intestine Virchows Arch 2000 437 90 94 10963385 10.1007/s004280000183
|
16159405
|
PMC1242259
|
CC BY
|
2021-01-04 16:39:04
|
no
|
World J Surg Oncol. 2005 Sep 14; 3:60
|
utf-8
|
World J Surg Oncol
| 2,005 |
10.1186/1477-7819-3-60
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2061612022210.1186/1471-2105-6-206Research ArticleGenomes are covered with ubiquitous 11 bp periodic patterns, the "class A flexible patterns" Larsabal Etienne [email protected] Antoine [email protected] Unité de Génétique des Génomes Bactériens, Institut Pasteur, URA CNRS 2171, 28, rue du Docteur Roux, 75724 Paris Cedex 15, France2005 24 8 2005 6 206 206 22 6 2005 24 8 2005 Copyright © 2005 Larsabal and Danchin; licensee BioMed Central Ltd.2005Larsabal and Danchin; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The genomes of prokaryotes and lower eukaryotes display a very strong 11 bp periodic bias in the distribution of their nucleotides. This bias is present throughout a given genome, both in coding and non-coding sequences. Until now this bias remained of unknown origin.
Results
Using a technique for analysis of auto-correlations based on linear projection, we identified the sequences responsible for the bias. Prokaryotic and lower eukaryotic genomes are covered with ubiquitous patterns that we termed "class A flexible patterns". Each pattern is composed of up to ten conserved nucleotides or dinucleotides distributed into a discontinuous motif. Each occurrence spans a region up to 50 bp in length. They belong to what we named the "flexible pattern" type, in that there is some limited fluctuation in the distances between the nucleotides composing each occurrence of a given pattern. When taken together, these patterns cover up to half of the genome in the majority of prokaryotes. They generate the previously recognized 11 bp periodic bias.
Conclusion
Judging from the structure of the patterns, we suggest that they may define a dense network of protein interaction sites in chromosomes.
==== Body
Background
The distribution of nucleotides in genomes is not random, various biases are affecting the genome sequences from organisms spanning the three domains of life. For example, the G+C content affects the genome as a whole.
To visualize the biases in the nucleotides distribution in genomes, investigators have performed a variety of statistical analyses; these operations basically consisted in counting the nucleotides in a variety of subtle ways, while attempting to identify how the counting observed in real examples differed from a random distribution. Relevant statistical methods developed so far include the following: computation of correlations [1], power spectrum analysis [2,3], DNA walking analysis [4], computation of entropy [5,6], Hurst index estimation [7], detrended fluctuation analysis [8], wavelet analysis [9], mutual information function analysis [10], computational linguistics analysis [11].
Among the different biases observed in the nucleotides distribution in genomes, two stood out prominently. Both are short-range biases, i.e. correlating nucleotides over a short distance only, inferior to one thousand base pairs (bp), and both are affecting the genome as a whole. Both are present in many different organisms. This prevalent intensity and ubiquity is a hint that these biases are very likely to be the result of some strong physical constraints and/or biological functions acting on the affected genomes.
The first prevalent bias, the most intense one, is easily visualized in the genomes of all prokaryotes, as well as of lower eukaryotes. It also appears, though very dimly, in the genomes of higher eukaryotes. This bias is periodic with a periodicity of 3 bp (locally, the probability of presence of a given nucleotide depends on its position modulo three). This ubiquitous bias is effectively uncovered by power spectrum analysis [12-17]. Its presence has never been a mystery: it is due to the presence of protein coding genes in genomes. Indeed, the structure of the genetic code strongly affects the distribution of nucleotides within protein coding sequences, biasing the distribution of nucleotide triplets. As the gene density of higher eukaryotes is very small, this bias cannot easily be detected in these organisms. In contrast, for prokaryotes and for lower eukaryotes, in which the gene density is high, this bias is very easily detected. Its association to protein coding proved to be useful to locate exons in higher eukaryotic genomes [18]. This first bias is therefore generated by genomic sequences that are of strong biological significance.
Likewise, the second prevalent bias, also very intense, is visualized in the genomes of most prokaryotes and lower eukaryotes. For a given genome, the bias is encountered throughout the genome. In contrast with the previous 3 bp periodic bias, which spans large distances (typically several hundreds nucleotides) this bias does not involve nucleotides over a distance longer than about one hundred base-pairs: it is a short-range bias. It is also periodic, but this time with a fuzzy periodicity of mean value 11 bp. This signal has been visualized with the straightforward computation of correlations [1,19] or its equivalent, the power spectrum method [17]. The mean value of the periodicity of this bias varies from organism to organism. In the two articles just mentioned, the authors discuss the relation between phylogeny and the distribution of these periods. It turns out that it is generally of 10 bp for Archaea or hyperthermophilic Bacteria and 11 bp or more for the non-hyperthermophilic Bacteria, though there are many exceptions to this rule [19]. In the case of lower eukaryotes, a period of 10 bp for C. elegans and of 11 bp for S. cerevisiae has been observed. In the case of higher eukaryotes, a weak bias of period 10 bp is observed once the many repeated sequences present in these genomes have been removed from the analysis [19]. Moreover, in prokaryotes and lower eukaryotes, the bias is affecting coding sequences as well as non-coding sequences. This general observation is illustrated in Figure 1 with a graphic representation of the correlation function of nucleotide A following itself in the genome of Helicobacter pylori.
Figure 1 Deconvoluted correlation function of A following A in the genome of H. pylori. The correlation function has been treated so as to hide the most intense component of period 3 bp due to the presence of genes in the genome of H. pylori. After treatment, the function reveals a prevalent short-range component of period 11 bp. This component represents the prevalent short-range bias of period 11 bp in the distribution of nucleotides in the genome of H. pylori.
This function measures the probability to get a nucleotide A following another nucleotide A as their distance increases. The correlation function has first been treated by deconvolution so as to hide the overwhelming component of period 3 bp that results from the presence of genes in the genome (see above). The corresponding statistical treatment is described in the Methods section. In the graphic representation of the correlation function shown in Figure 1, there is a prominent component of period 11 bp. It appears as a short-range component as it completely vanishes for nucleotides located more than 70 bp apart. The periodic peaks do not occur every 11 bp exactly but every 10 bp to 12 bp. The strength of the periodic bias is illustrated by their large amplitude.
Although this bias is half as high in intensity as the one created by the presence of genes, and although it is ubiquitous in prokaryotes and lower eukaryotes, the nucleotide sequences generating this bias have not been determined so far. Nonetheless, the biological function that might be at the root of this bias has been proposed. In the case of Archaea, it has been suggested that the positioning of nucleosomes is controlled by some specific sequences, whose nature could however not be identified [1,19].
In the present article, we describe the program we designed, meant to discover the sequences that are generating every short-range bias (excluding the trivial one of period 3 bp generated by the genes) in genomes. Making use of this program, we discovered explicitly the sequences responsible for the bias of period 10–11 bp in the prokaryotic and lower eukaryotic genomes. These sequences, that we named "class A flexible patterns" for reasons that will be clarified in the course of this article, display a new type of organization. We show that the class A flexible patterns are ubiquitous in prokaryotes.
Results
Our aim was to identify the sequences that generate the 11 bp periodic short-range bias. To address this question, we designed a generic program to determine the sequences that generate any short-range bias in genomes nucleotides distribution (see the Methods section): the sequences responsible for the 11 bp periodic bias should belong to the sequences identified by the program.
For each genome of interest, the output of the program is given as a family of patterns. By pattern, we mean any succession of nucleotides with gaps in between (see the Methods section). The family of patterns returned by the program has the following property: the occurrences in the genome of all the patterns belonging to the pattern family match the sequences of the genome supposed to generate its short-range biases (see Methods section). Because of computation time limitations, our program gives an approximate result only: the patterns shape is restricted and the matching may not be exact (see the Methods and Discussion sections).
The program was run with 49 prokaryotic genomes, with four lower eukaryotic genomes and three viruses sequences. We collected the patterns of all the resulting family of patterns and saw that we could class them into two category of patterns. Naming them after their particular structural features, we called them the "rigid patterns" and the "flexible patterns". The rigid patterns are described first, but not discussed in details because they overlap with previously identified repeated sequences. Then we describe the more frequent but elusive flexible patterns. Among those, a great number belongs to a class that we called the "class A flexible patterns", for reasons explained below. The latter patterns are discussed extensively. Finally, we show that the occurrences of the class A flexible patterns define the sequences generating the bias of period 11 bp in genomes.
Rigid patterns
A rigid pattern is a pattern verifying the two following properties: first, the distance between the nucleotides making the pattern is the same for every occurrence of the pattern in the genome. Second, some variability in the nature of the nucleotides composing the pattern is allowed from one occurrence to another one. Most patterns described so far in the literature are rigid patterns. For rigid patterns, the exact distances between the nucleotides and the frequency of occurrence of the nucleotides A,T,G,C composing the pattern account for what is usually termed a "consensus sequence".
As a proof of concept, the program uncovered families of rigid patterns in a few selected genomes. Each family was made of short highly repeated motifs. As could be expected, when present in a genome, highly repeated sequences generate a short-range statistical bias. For example, we found the following rigid pattern in the genome of Escherichia coli (an x represents any nucleotide):
5GCxxxATxxxGCxxxxxxGCxxxATxxxGC-3'
One can recognize in this pattern a consensus for the repeated Bacterial Interspersed Mosaic Elements (BIMEs) sequences of E. coli [20]. It is important to note here that, although these sequences are recognized by our program because they create small but significant biases in the nucleotides distribution of E. coli, they do not contribute to the generation of the bias of period 11 bp. However, the very fact that we uncovered them is an independent validation of our approach.
Flexible patterns
To extend the rigid patterns description, we defined the "flexible patterns". A flexible pattern satisfies the two following properties: first, the nature of the nucleotides composing the pattern is the same for all the occurrences of the pattern in a given genome. Second, the distance between the nucleotides composing the pattern varies in a narrow range between occurrences of the pattern. Hence, a flexible pattern differs from a rigid pattern in that it could not generate a "consensus" by aligning sequences without introducing gaps. As an example, here are different occurrences of a flexible pattern found in the genome of
Pyrococcus furiosus
GxxAxxxTTxxxGxxxT
GxxAxxxTTxxxGxxxT
GAxxxTTxxxxxGxxxT
GxxAxxxTTxxxGxxxxxxT
GxxxAxxxTTxxxGxxxxxxT
GxAxxxTTxxxxGxxxxxxT
5'-xxxxxx-20xxxxxxx--3'
From now on, we will represent a given flexible pattern not by its various spellings but by an average representative, in which the distance between the nucleotides is the mean distance of all the distance observed in all the various spellings. For example, we represent the previous flexible pattern by this average representative:
5'-GxxAxxxTTxxxGxxxxxT-3'
Conversely, in the following, a flexible pattern mentioned by an average representative is defined by the list of similar patterns which are deviating from the average representative by distances varying withing a narrow range between its conserved nucleotides.
The great majority of the patterns that we found by running our program in various genomes turned out to be of the flexible patterns category. We found on average approximately twenty flexible patterns in each genome, be it of a prokaryotic organism or of a lower eukaryotic organism. We observed that the distances between nucleotides composing the flexible patterns we identified vary generally from one to two base pairs. These patterns are composed of five to ten nucleotides spanning a distance of 10 bp to 60 bp. The nucleotides composing these patterns are most of the time either isolated or grouped as dinucleotides.
The description of patterns is limited by our program due to computing time limitations (see the Methods section), for example they cannot be composed of more than six nucleotides. The patterns that we get often seem to be subsets of longer patterns. In the following we mention the longest pattern that can be inferred, but it should be kept in mind that each of its detected variations are composed of only six nucleotides. For example, the following flexible pattern found in H. pylori:
5'-TxxAxGCxTTT-3'
is defined by the following variations:
TxxxGCxxTT
TxxxxGCxxTTxT
TxxxxGCxxTxTT
TxxxxxGCxxTTT
TxxxxxAxGCxTT
AxxGCxTTT
AxGCxTTxT
AxxGCxTTxT
Class A flexible patterns
Among flexible patterns, we observed that a great majority shared a similar structure and were thus easily identifiable. We named "class A flexible patterns" this subset of flexible patterns. We will restrict our study to these patterns, as they account for most, if not all, of the 11 bp period found in the genomes we analyzed.
All class A flexible patterns, though different in spelling, share the same structure, as depicted in Figure 2. The structural features illustrated in this figure are formally defining the class A flexible patterns. The patterns are described here in the standard 5'-3' orientation.
Class A flexible patterns are in total composed of five to ten conserved nucleotides spanning a length of approximately 11 bp to 50 bp. The conserved nucleotides are either isolated or grouped as dinucleotides.
Figure 2 Diagrammatic structure of class A flexible patterns. Class A flexible patterns belong to the category of flexible patterns. Here "flexible" means that there is limited variation in the exact position of their conserved nucleotides. This is shown in the figure by the green arrows, indicating that the position of the conserved nucleotides may vary from one occurrence of the pattern to the next. The particular class of flexible patterns depicted here in the standard 5'-3' orientation is composed of two sets of conserved nucleotides. First, the patterns are shaped by a skeleton of regularly repeated Ts or TTs every 10 bp to 11.5 bp, spanning a maximum of 50 bp. These are called "skeleton nucleotides" and are symbolized by the black and dark grey Ts. The peripheral repeats of the skeleton, in dark grey, are sometimes absent from a given occurrence. The Ts of the central part, spanning 20 bp on average, are always present. Furthermore, class A flexible patterns are composed of a set of "inner nucleotides". These conserved nucleotides are represented here in dark blue. They can be any nucleotide but never Ts. They are located between the Ts of the skeleton and in the central part only.
That these patterns belong to the category of flexible patterns is illustrated in Figure 2 by the green arrows above the nucleotides composing the patterns (always isolated nucleotides or dinucleotides). The distance between any of the isolated nucleotides or dinucleotides varies by 1 bp to 2 bp from one occurrence of the pattern to the next in a given genome. Class A flexible patterns are composed of two subsets of conserved nucleotides: the skeleton nucleotides and the inner nucleotides.
The skeleton nucleotides consist of two to five repeats of the single nucleotide T or of the dinucleotide TT, regularly spaced every 10 bp to 11 bp on average. The central part (nucleotides represented in black in Figure 2) is made of two to three repeats. These repeated nucleotides appear at every occurrence of a given pattern in a given genome. Outlying repeats (nucleotides in dark grey in Figure 2) may extend the skeleton outside the central part. Those are involving single nucleotides Ts exclusively and are not always present: they do not appear in every occurrence of a given pattern. Typically, one or two such peripheral repeats of the single nucleotide T on each side of the central part of the skeleton exist in a given occurrence of a pattern. Note that for a given pattern, the distance (averaged over all the occurrences of the given pattern in a given genome) between two neighboring isolated conserved nucleotides Ts or dinucleotides TTs of the skeleton ranges from 7 bp to 12 bp. Yet, the average of these distances over the two to five repeats of the skeleton of the given pattern remains inside the interval of 10 bp to 11.5 bp. The skeleton structure, spanning up to 50 bp in total, is basically the same for all class A flexible patterns, for only the distances between the Ts and the choice of single or dinucleotides can fluctuate.
The inner nucleotides consist of one to three conserved nucleotides located exclusively in the central part of the skeleton. Most importantly, these conserved nucleotides are found to be either A, G or C (a particular nucleotide specifying the particular kind of pattern identified, see Figure 3) but never T. They are either isolated or grouped as dinucleotides (isolated conserved nucleotides are more frequent than conserved dinucleotides). There can be only one isolated nucleotide or dinucleotide between two neighboring skeleton nucleotides. The position of the inner nucleotides is usually located exactly in the middle of two neighboring Ts of the skeleton. These inner nucleotides play a discriminating role in class A flexible patterns as they differentiate patterns from one another.
Figure 3 A few identified class A flexible patterns. Ten related yet distinct class A flexible patterns common to different genomes have been identified so far. Their structures share common features, which are characteristic of class A flexible patterns. Peripheral repeats of the skeleton nucleotides of the patterns have not been represented here. Skeleton nucleotides are shown in black. Inner nucleotides are shown in dark blue.
The central part of these patterns is composed of three to six skeleton nucleotides and of two to four inner nucleotides (see Figure 2). Altogether, the central part is composed on average of six conserved nucleotides covering from 10 bp to 33 bp. This part of the patterns is the one that varies from one class A flexible pattern to another, both in the choice of single or dinucleotides in the skeleton and in the nature of the inner nucleotides. Therefore, we choose to subsequently identify the patterns using this central part only.
The program we ran is limited to identification of patterns spanning up to a maximum of 60 bp (see the Methods section). This implies that we may have been missing some peripheral repeats of Ts in some occurrences of the patterns, but we did not miss important nucleotides as the latter are located in the central parts of the patterns only.
Distribution of class A flexible patterns in organisms
As a whole, cumulating all the tested genomes, we could identify twenty different types of class A flexible patterns. Some genomes harbor specific class A flexible patterns that are found in no other genome. In contrast, some types of patterns are found in more than one genome. We could identify ten such conserved types of patterns. In Figure 3, we list these ten types of class A flexible patterns.
Patterns numbered 1 to 5 in Figure 3 are present in many genomes, patterns numbered 6 to 10 are present in less than ten different genomes.
In Table 1, we display the organisms in which these patterns were identified, as well as the phylogenetic family to which the organisms belong. It turned out that every one of the 49 prokaryotic genomes tested, two of the four lower eukaryotic genomes tested (Saccharomyces cerevisiae and Caenorhabditis elegans) and the two genomes of bacteriophages analyzed were harboring class A flexible patterns.
Table 1 Distribution of class A flexible patterns in genomes.
1 2 3 4 5 6 7 8 9 10
Aeropyrum pernix X X X Archaea; Crenarchaeota; Thermoprotei; Desulfurococcales
Sulfolobus solfataricus X X X X Archaea; Crenarchaeota; Thermoprotei; Sulfolobales
Sulfolobus tokodaii X X X Archaea; Crenarchaeota; Thermoprotei; Sulfolobales
Pyrobaculum aerophilum X X Archaea; Crenarchaeota; Thermoprotei; Thermoproteales
Archaeoglobus fulgidus X X X Archaea; Euryarchaeota; Archaeoglobi; Archaeoglobales
M. Acetivorans X X X X X X Archaea; Euryarchaeota; Methanosarcinales
Halobacterium sp. X Archaea; Euryarchaeota; Halobacteriales
M. thermoautotrophicum X X Archaea; Euryarchaeota; Methanobacteriales
Methanococcus jannashii X X X Archaea; Euryarchaeota; Methanococcales
Pyrococcus abyssi X X X Archaea; Euryarchaeota; Thermococcales
Pyrococcus furiosus X X X X Archaea; Euryarchaeota; Thermococcales
Pyrococcus horikoshii X X X Archaea; Euryarchaeota; Thermococcales
Thermoplasma acidophilum X X Archaea; Euryarchaeota; Thermoplasmatales
Tropheryma whipplei X X X Bacteria; Actinobacteria; Actinomycetales
Aquifex aeolicus X X X Bacteria; Aquificae; Aquificales
Chlorobium tepidum X X Bacteria; Chlorobi; Chlorobiales
Synechocystis sp. Bacteria; Cyanobacteria; Chroococcales
Deinococcus radiodurans X X Bacteria; Deinococcus-Thermus; Deinococcales
Bacillus subtilis X X Bacteria; Firmicutes; Bacillales
Oceanobacillus iheyensis X Bacteria; Firmicutes; Bacillales
Listeria monocytogenes X X Bacteria; Firmicutes; Bacillales
T. Tengcongensis X X Bacteria; Firmicutes; Clostridia; Thermoanaerobacteriales
Streptococcus pneumoniae X Bacteria; Firmicutes; Lactobacillales
Pirellula sp. X Bacteria; Planctomycetes; Planctomycetales
Magnetactic cocci X X Bacteria; Proteobacteria
Caulobacter vibrioides X Bacteria; Proteobacteria; Alphaproteobacteria; Caulobacteriales
Agrobacterium tumefaciens X X Bacteria; Proteobacteria; Alphaproteobacteria; Rhizobiales
Sinorhizobium meliloti X Bacteria; Proteobacteria; Alphaproteobacteria; Rhizobiales
Rickettsia conorii X X X Bacteria; Proteobacteria; Alphaproteobacteria; Rickettsialles
Rickettsia prowozekii X X X X X Bacteria; Proteobacteria; Alphaproteobacteria; Rickettsialles
Bordetella pertussis X X Bacteria; Proteobacteria; Betaproteobacteria; Burkholderiales
Neisseria meningitidis X Bacteria; Proteobacteria; Betaproteobacteria; Neisseriales
Campylobacter jejuni X X Bacteria; Proteobacteria; Epsilonproteobacteria; Campylobacterales
Helicobacter hepaticus X X X X Bacteria; Proteobacteria; Epsilonproteobacteria; Campylobacterales
Helicobacter pylori X X X Bacteria; Proteobacteria; Epsilonproteobacteria; Campylobacterales
Wolinella succinogenes X X X Bacteria; Proteobacteria; Epsilonproteobacteria; Campylobacterales
P. haloplanktis X X Bacteria; Proteobacteria; Gammaproteobacteria; Alteromonadales
Candidatus bl. floridanus X Bacteria; Proteobacteria; Gammaproteobacteria; Enterobacteriales
Buchnera aphidicola X Bacteria; Proteobacteria; Gammaproteobacteria; Enterobacteriales
Escherichia coli X Bacteria; Proteobacteria; Gammaproteobacteria; Enterobacteriales
Wigglesworthia glossinidia X Bacteria; Proteobacteria; Gammaproteobacteria; Enterobacteriales
Coxiella burnetii X Bacteria; Proteobacteria; Gammaproteobacteria; Legionellales
Haemophilus influenzae X X X X X Bacteria; Proteobacteria; Gammaproteobacteria; Pasteurellales
Pseudomonas aeruginosa X X Bacteria; Proteobacteria; Gammaproteobacteria; Pseudomonadales
Pseudomonas putida X X X X Bacteria; Proteobacteria; Gammaproteobacteria; Pseudomonadales
Vibrio vulnificus X X X Bacteria; Proteobacteria; Gammaproteobacteria; Vibrionales
Xylella fastidiosa X X X X X Bacteria; Proteobacteria; Gammaproteobacteria; Xanthomonadales
Leptospira interrogans X Bacteria; Spirochaetes; Spirochaetales
Thermotoga maritima X X Bacteria; Thermotogae; Thermotogales
Plasmodium falciparum Eukaryota; Alveolata; Apicomplexa
Saccharomyces cerevisiae X X Eukaryota; Fungi; Ascomycota
Encephalitozoon cuniculi Eukaryota; Fungi; Microsporidia
Caenorhabditis elegans X X X Eukaryota; Metazoa; Nematoda
Enterobacteria phage T4 X Virus; Enterobacteria phage T4
S. tengcon.. Vvrus STSV1 X Virus; Fusellovirus
Human herpesvirus 4 Virus; Human herpesvirus 4
1. AxxxxTxxxxAxxxxTTxxxxxAxxxxTxxxxA
2. GxxxxTTxxxCxxxT
3. TTxxxGxxxTTxxxxGxxxxTT
4. TxxxxAGxxxTTxxxxxxxxT
5. TxxxxxxxxxxTxxxGAxxxTT
6. CxxxxxTTxxxCxxxxxxT
7. TxxxGCxGxT
8. TxxCxGxCxTT
9. GxxxxxTxxxxxAxxxxxT
10. TTTxxxCAxxxxxT
First, we found out that class A flexible patterns are ubiquitous in prokaryotes. Indeed, each of 49 genomes of prokaryotes tested harbors one or more different types of class A flexible patterns. The genome of Xylella fastidiosa harbors for instance five different types of patterns. Usually, each genome harbors two to four different types of class A flexible patterns. Second, each of the patterns numbered 1 through 5 in Figure 3 is present in more than 10 different genomes. This makes it possible to discuss the nature of the distribution of these five types of patterns in genomes.
Pattern 1 has been detected in more than 50% of the 56 tested genomes, with no relationship to phylogenetic branches as we found it in Archaea, in Bacteria, in lower eukaryotes and in phages (see Table 1). This pattern alone may be ubiquitous as a low content of this pattern in a given genome would fail to be detected by our approach.
Pattern 2 is present in a total of 19 genomes. Out of these 19 genomes, 16 belong to Proteobacteria. Three further genomes, that do not belong to the Proteobacteria clade, display this type of pattern. Among those, we found first two Bacteria: Deinococcus radiodurans and Tropheryma whipplei. The former lives under highly desiccated or radiation-exposed conditions, with remarkable features in DNA maintenance [21], while the latter is a highly degenerate parasite [22]. The third organism which is not a Proteobacteria and where this type of pattern is present is an Archaeon: Pyrobaculum aerophilum [23]. Overall, the distribution of pattern number 2 in genomes is highly correlated with the Proteobacteria class of organisms. It is present throughout this class of organisms as it has been detected in some genomes of the alpha, beta, epsilon and gamma groups (the delta group has not yet been analyzed). It is also remarkably present in all tested genomes of the epsilon group.
Pattern 3 is present in 18 genomes in total, in Archaea, in Bacteria and in lower eukaryotes. Pattern 4 is present in 13 genomes in all. It has been identified in 11 of the 13 archaeal genomes analyzed (in Crenarcheota as well as in Euryarchaeota). It is also present in two Bacteria (Aquifex aeolicus and Helicobacter hepaticus). Hence, the distribution of this pattern in genomes seems to be somewhat correlated with the archaeal kingdom.
Pattern 5 is present in 14 genomes in total, in Archaea, in Bacteria and in lower eukaryotes. The other identified class A flexible patterns are present in only a few organisms. Moreover, these organisms do not clearly belong to any specific phylogenetic lineage. In Figure 4 are summarized the few parallels that could be drawn between the distribution of class A flexible patterns and phylogeny. Each of these three patterns is present in more than 10 genomes out of the 56 tested.
Figure 4 The distribution of three types of class A flexible patterns is correlated to specific phylogenetic groups of organisms. We identified five class A flexible patterns distributed in many different organisms. Three of them, displayed here, show a distribution which can be related to the phylogeny.
Distribution of class A flexible patterns in a given genome
The occurrences of class A flexible patterns are equally distributed in the two strands of chromosomes. These occurrences cover a considerable part of each genome. The conserved nucleotides of all occurrences of all class A flexible patterns are involving up to one fourth of the total number of nucleotides of a given genome (24% in the case of H. pylori). If we take into consideration the total length that the occurrences of the patterns span in a genome, then it comes up to one half of each genome (51% in the case of H. pylori). In the case of H. pylori, the span of the patterns ranges from 9 bp to 29 bp (Table 2). We observed that the patterns' occurrences can be overlapping. Interestingly, class A flexible patterns occur indifferently in coding and in non-coding regions of genomes. They are neither correlated with the leading nor with the lagging strand of chromosomes. All things considered, there seems to be no obvious bias in the distribution of the occurrences of the patterns.
Table 2 The variations defining the five class A flexible patterns found in the genome of H. pylori
1-5'-xxxxxxxxxxxxxxxTxxxxxxxxxxTxxxxGxxxTTxTxxxxxxxxxxxxxxxxxxxxx-3'
xxxxxxxxxxxxxxxxxxxxTxxxxxxxxxxTxxxxGxxxTTxTxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxTxxxxxxxxxxxTxxxxGxxxTTxTxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxx-20xxxxxxx-10xxxxxxxxx0xxxxxxxxx10xxxxxxxx20xxxxxxxx30
2-5'-xxxxxxxxxxxxxxGGxxTTTxxxxxxxxxxTxxxxxxxxxTxxxxxxxxxxxxxxxxxx-3'
xxxxxxxxxxxxxxxxxxxGxxxTxTTxxxxxxxxxTxxxxxxxxxxTxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxGxxTTTxxxxxxxxxTxxxxxxxxxxTxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxGxxTTTxxxxxxxxxTxxxxxxxxxxxTxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxGxxTTTxxxxxxxxxxTxxxxxxxxxTxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxGGxxTTTxxxxxxxxxxTxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxGGxxTTxTxxxxxxxxxTxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxGxxTTTxxxxxxxxxTTxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxGxGxxTTTxxxxxxxxxxTxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxx-20xxxxxxx-10xxxxxxxxx0xxxxxxxxx10xxxxxxxx20xxxxxxxx30
3-5'-xxxxxxxxxxxxxxxxTxxxxxxxxxTTTxxAAxCxxTxxxxxxxxxxxxxxxxxxxxxx-3'
xxxxxxxxxxxxxxxxxxxxxTxxxxxxxxxxTTxxAxxCxxTxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxTxxxxxxxxxxTxxAAxCxxTxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxTxxxxxxxxxxxTxxAAxCxxTxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxTxxxxxxxxxTTxxAAxxCxxxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxTxAAxCCxTxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxTxxAAxCCxxTxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxTTxxAxCCxxTxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxTTTxxAxxCxxTxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxTTxTxxxAxxCxxTxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxTxTTxxAxxCxxTxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxx-20xxxxxxx-10xxxxxxxxx0xxxxxxxxx10xxxxxxxx20xxxxxxxx30
4-5'-xxxxxxxxxxxxxxxxxxxxxxxxxxGGxxTTTxxxxxCxxxxxxxxxxxxxxxxxxxxx-3'
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxGGxTTTxxxxCxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxGGxxTTxTxxxxCxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxGGxxTxTTxxxxCxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxGxGxTxTTxxxxCxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxGxGxxTTTxxxxCxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxx-20xxxxxxx-10xxxxxxxxx0xxxxxxxxx10xxxxxxxx20xxxxxxxx30
5-5'-xxxxxxxxxxxxxxxxxxxxxxxxxxTxxAxGCxTTTxxxxxxxxxxxxxxxxxxxxxxx-3'
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxTxxxGCxxTTxxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxTxxxxGCxxTTxTxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxTxxxxGCxxTxTTxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxTxxxxxGCxxTTTxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxTxxxxxAxGCxTTxxxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxAxxGCxTTTxxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxAxGCxTTxTxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxAxxGCxTTxTxxxxxxxxxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxx-20xxxxxxx-10xxxxxxxxx0xxxxxxxxx10xxxxxxxx20xxxxxxxx30
|p463pt|The variations defining the five class A flexible patterns found in the genome ofH. pylori
Contribution of class A flexible patterns to the 11 bp periodic bias
The structure of class A flexible patterns is highly reminiscent of the 11 bp periodic bias in genomes of prokaryotes and lower eukaryotes. Indeed, the patterns have a core of repeated Ts or TTs every 10 bp-11 bp on average in all occurrences. It can therefore be expected that because these periodic nucleotides are densely spread, a bias of period 10 bp-11 bp will be generated in the corresponding genome sequences. The length of the patterns when the peripheral repeats are considered (up to 60 bp) is on the same order as the span of the 10 bp-11 bp periodic component in the correlation between nucleotides (see Figure 1). Furthermore, we systematically observed that the component of period 11 bp is somewhat fuzzy (see the blunt shaped peaks in Figure 1). This is consistent with the fact that the distance between neighboring skeleton nucleotides ranges from 7 bp to 12 bp. This is also consistent with the involvement of dinucleotides in class A flexible patterns. Finally, the occurrences of class A flexible patterns distribute throughout a given genome, with no apparent preference for coding or non-coding regions, similarly to the bias of period 10–11 bp. Now we want to show that the class A flexible patterns are indeed the source of the 11 bp periodic bias in genomes. We illustrate this with the genome of H. pylori as the statistical bias of period 11 bp is particularly prominent there. We got the same results for all other genomes analyzed.
The class A flexible patterns discovered in the H. pylori genome are the following:
1-5'-TxxxxxxxxxxTxxxxGxxxTTxT-3'
2-5'-GGxxTTTxxxxxxxxxxTxxxxxxxxxT-3'
3-5'-TxxxxxxxxxTTTxxAAxCxxT-3'
4-5'-GGxxTTTxxxxxC-3'
5-5'-TxxAxGCxTTT-3'
Patterns numbered from 1 to 3 are also found in genomes of other organisms, while patterns 4 and 5 are found only in this genome. Helicobacter pylori is remarkable as the skeleton nucleotides are composed of the trinucleotide TTT. For each of those flexible patterns, Table 2 illustrates the list of their variations. No peripheral repeats are displayed, as we failed to determine any in this particular genome. It is interesting to note that all the variations of these five patterns are indeed over-represented in the genome of H. pylori. We compared the number of occurrences of the patterns in the authentic genome to the number of occurrences in a model genome that keeps only the crude statistical features of the nucleotide distribution in the H. pylori genome (see the Method section). We found that the variations of pattern 1 occur approximately 30% more often in the authentic genome than in the model genome, the variations of pattern 2 approximately 40%, the variations of pattern 3 approximately 30%, the variations of pattern 4 approximately 40%, the variations of pattern 5 approximately 30%. All the nucleotides involved in the occurrences of patterns 1 to 5 and of their reverse complements amount to 24% of the total number of nucleotides contained in the whole genome. To explore whether the bias of period 11 bp in the distribution of the nucleotides is due to these 24% of the genome of H. pylori, we constructed two reference genomes for comparison.
We constructed a first "deconvoluted" genome Gmo(G-) in the following way (see the Methods section): starting from the authentic genome of H. pylori, every nucleotide which belongs to any occurrence of any of the five class A flexible patterns or of their reverse complements is replaced by the nucleotide of a model genome preserving the local composition in hexanucleotides of the authentic genome but not their order (see the Methods section) while every other nucleotide is kept unaltered. We plotted the treated correlation function of Gmo(G-) for the nucleotide A following A (see the Methods section) in Figure 5. The 11 bp periodic bias is now absent from this plot. This means that the 76% of the genome of H. pylori which is not covered by class A flexible patterns does not have any significant 11 bp periodic statistical bias. Hence, we concluded that class A flexible patterns are generating the 11 bp bias in genomes.
Figure 5 The treated correlation function of Gmo(G-). This correlation function of nucleotide A following A reveals biases generated by the part of the genome of H. pylori that do not contain occurrences of class A flexible patterns.
Interestingly, the 11 bp periodic bias disappeared even at correlations over 30 bp, despite the fact that our patterns are never longer that 30 bp for this genome (we have deconvoluted the central parts of the patterns but not the hypothetical peripheral repeats). Deprived of the core sequences of the patterns, the peripheral repeats, even if they exist, can no longer generate much bias. In Figure 5, one can notice a small peak pointing downwards at 11 bp. This probably reflects the fact that we failed to describe accurately the patterns and therefore removed too many sequences, some of which artefactually taken as genuine patterns. Second, we plotted the treated correlation function (see the Methods section) of a complementary model: Gmo(G+), the "convoluted" genome (Figure 6). As in the preceding model, Gmo(G+) is built starting from the authentic genome of H. pylori: all the nucleotides not belonging to occurrences of class A flexible patterns and of their reverse complements are replaced by the nucleotides of a model genome (see the Methods section). The 11 bp statistical bias from the original genome is now visible again (the treated correlation function of the original genome is shown in Figure 1). The correlations over 30 bp are hardly visible, which is consistent with the fact that no peripheral repeats were introduced in the convolution process. In this "realistic" imitation of the H. pylori genome, the correlations below 30 bp are somewhat too intense when compared to the real ones, displayed in Figure 1. This shows again that we removed too many core sequences, as they were not described with enough accuracy. The sum of the treated correlation function of the deconvoluted genome and of the treated correlation function of the convoluted genome fails to be exactly equal to the treated correlation function of the authentic genome. This shows that there exist correlations between occurrences of class A flexible patterns and of neighboring sequences. It can be expected that these correlations involve the undetected peripheral repeats.
Figure 6 The treated correlation function of Gmo(G+). This correlation function of nucleotide A following A reveals biases in the genome of H. pylori which are generated by the occurrences of class A flexible patterns in its genome.
Finally we must note that we chose to illustrate the relationship between class A flexible patterns and the 11 bp bias with the correlation function calculated for an A following an A, as the correlations are specially strong for those two nucleotides. However the results reported are still valid for any combination of two nucleotides.
Discussion
In the present work we focused on class A flexible patterns as they are the source of the 11 bp periodic bias long known to exist in genomes. Because of the technical limitations of our approach we expect that there may still be other classes of flexible patterns in DNA sequences. They must be however relatively less important as genome sequences do not display prominent short-range biases other than the 3 bp and the 11 bp periodic long identified, while deconvolution of authentic genome sequences from the patterns we identified yielded sequences which no longer displayed any outstanding periodicity.
Limitations in the description of class A flexible patterns
As explained in the Methods section, our approach suffers some limitations, mainly due to computational time limitations. First, simply for stochastic reasons (the signal must be significantly higher than the noise), we would not find sequences that are generating weak biases or that are present in a too limited amount in genomes (with a frequency below ). Hence we probably missed the presence of some class A flexible patterns in some genomes. Second, the output of our program may have been somewhat inaccurate. Namely, because of the limitation we had to impose on the correlations order (see the Methods section), we may have identifed some patterns as genuine while they would represent a mix of different patterns present at distinct locations in the genomes. Third, we are bound to miss completely any pattern in which the shorter distance between conserved nucleotides is longer than 14 bp (see the Methods section). Fourth, the patterns spellings are but an approximation. Our program has restrictions in the maximum length and number of conserved nucleotides of patterns it is able to determine. As a consequence, we may have missed peripheral parts of the patterns we identified. Still, these restrictions probably did not affect much our spelling of class A flexible patterns, as these patterns are short enough: the central parts span only 20 bp on average. In contrast, in the identification of rigid patterns, typically made of continuous sequences of conserved nucleotides ("words" or "motifs"), we could not retrieve all conserved nucleotides. This was not, however, the main goal of this work.
Connection to optimal growth temperature
As phylogeny cannot account for the distribution of patterns numbered 3 and 5 in Figure 3, we may wonder whether the distribution of these two class A flexible patterns could be related to physical or biological parameters of the organisms in which they have been identified. We took into account the Gram staining, the cell shape, oxygen dependency, sporulation ability, encapsulation ability, optimal pH and maximum growth temperature, GC content and GC skew. Among those features, the optimal growth temperature somewhat correlates with the distribution of these class A flexible patterns. Indeed, both patterns are present mostly in thermophilic organisms. Still, it remains difficult to draw any firm conclusion in this matter as all tested Archaea but one (Methanosarcina acetivorans) are thermophilic and as these patterns are found mostly in Archaea. The question thus arises to determine whether these patterns are present in archaeal organisms or in thermophilic organisms. It is not yet possible to draw a clear rule from the presently tested genomes.
Class A flexible patterns may define protein interaction sites on the DNA molecule
The very structure of class A flexible patterns offers precious hints to conjecture their biological function. The hypothesis we propose is that the patterns are the signatures of DNA-protein interaction sites. Five arguments tend to support this idea. These are only theoretical arguments and our hypothesis needs to be substantiated by further experiments. First argument: to our knowledge, the length of class A flexible patterns is in a range appropriate for DNA-protein interactions. The total length of the patterns ranges from 11 bp to 60 bp while the length of the central part ranges from 10 bp to 33 bp (see Figure 2). The size of the DNA-protein binding sites usually ranges from 10 bp to 40 bp [24,25]. Hence the central part of the patterns, which is specific and conserved, may be the interacting protein-DNA interface.
Second argument: the number of conserved nucleotides composing the central parts of class A flexible patterns (six on average, see Figure 2) is compatible with the hypothesis. Indeed, if more nucleotides were conserved in the sequence, it is likely that the interaction would be very strong and would therefore have been already identified. Furthermore it would correspond to a stable interaction that would presumably preclude any function of the DNA molecule requiring its opening. In contrast, if there were fewer conserved nucleotides, the interaction would be too weak to create a specific interaction with proteins. Previous studies have established that the average number of conserved nucleotides in DNA-protein interaction sites ranges from five to ten conserved nucleotides [25].
Third argument: the position of the conserved nucleotides of class A flexible patterns is remarkably consistent with the hypothesis of a DNA-protein interaction site. Class A flexible patterns are composed of a skeleton made of regularly repeated Ts or TTs every 10 bp-11.5 bp on average. As the shape of the DNA molecule is helical, with a pitch of average 10.5 bp, varying from 10 bp to 12 bp [26], when unbound, repeated conserved nucleotides of the skeleton always appear at the same side of the helix, in the major groove and in the minor groove respectively (see Figure 7). Inner nucleotides of the patterns, which are always A, G or C depending on the particular pattern considered, are set between the repeated Ts of the skeleton, most often in the middle of two neighboring repeats. Hence, the inner nucleotides also appear on the same two sides of the DNA molecule, through grooves that are opposite to those of the skeleton nucleotides. Note that interactions between proteins and DNA minor grooves are well documented [27,28].
Figure 7 Accessibility of class A flexible patterns. There are two opposed sides from which nucleotides composing this occurrence of this given class A flexible pattern are accessible. The dinucleotides are visible through major grooves only from the upper side and hence fully accessible from this side only. Hence a given occurrence of a given class A flexible pattern in a genome is only accessible from one side of the DNA molecule.
The spatial structure of the DNA molecule of class A flexible patterns is illustrated in Figure 7. The nucleotides composing the example pattern of the figure are accessible from the upper side, with the skeleton nucleotides visible through major grooves and the inner nucleotides visible through minor grooves, or from the lower side, with the skeleton nucleotides visible through minor grooves and the inner nucleotides visible through major grooves.
The skeleton of the patterns is half composed of repeated dinucleotides TTs. In contrast, inner nucleotides are mostly isolated conserved nucleotides. A dinucleotide may be less easily accessed through a minor groove because this groove is too narrow. Conversely, it may be easily accessed through a major groove as the latter is wider. Hence, class A flexible patterns may be actually accessible by only one of the two opposed sides of the DNA double helix, the one where skeleton nucleotides are seen through major grooves, as shown in Figure 7. This gives a very specific argument to think that the function of these patterns may be to define interaction sites with some proteins. Indeed, a protein interacting with the DNA molecule usually comes along one defined side of the molecule and at any rate is never covering the molecule on all sides [29]. The position of the nucleotides composing the patterns is fully consistent with this requirement.
Fourth argument: class A flexible patterns belong to the group of flexible patterns. This means that the exact position of conserved nucleotides of the patterns varies from one occurrence of the patterns in genomes to the next one. This property is fully consistent with the hypothesis that the patterns are signatures of motifs allowing interaction with a geometrically rigid protein, as explained below.
The DNA molecule is a flexible molecule that can be elastically bent, elongated and supercoiled negatively or positively. As a matter of fact, in living cells, the molecule keeps on being constrained by thermal agitation and even more dramatically by the constant action of various molecules. For example, the action of polymerases will induce strong supercoiling ahead and behind where it acts [30]. Finally, the pitch and bending of the DNA helix keeps on varying locally, depending in particular on the local base composition [31].
Under these conditions, the constraint on the precise position in the genome of the conserved nucleotides of an interaction site is low. Indeed, when one conserved nucleotide of a given pattern is shifted from one base pair in the genome, chances are high that one of the probable conformations of the DNA molecule will place this nucleotide at the same spatial position compared to when it is not shifted in the genome and with another conformation of the DNA molecule. This is obviously true only if the shifts are not too important. This tends to confirm that class A flexible patterns define protein interaction sites. Indeed, we observed that from one occurrence to the next, the relative position of nucleotides composing them can vary from one to two base pairs. This is small enough so that there exists a likely conformation of the DNA molecule suitable to make it interact with its associated rigid protein. Alternatively, locally constrained DNA segments (for example through preexisting interaction with particular factors) might interact with proteins with flexible segments. Note that the absence of strong constraints on the position of the conserved nucleotides in class A flexible patterns is not easily compatible with other biological functions.
Fifth argument: the presence of optional peripheral repeats of Ts extending the skeleton at its two sides in class A flexible patterns (see Figure 2), can easily be accounted for under this DNA-protein interaction hypothesis. There are at least two ways to interpret the presence of the peripheral repeats. A first idea is to suppose that they could be used by the DNA molecule to stabilize an interacting protein, as they appear on the same side of the DNA molecule as the rest of the conserved nucleotides of the pattern. These peripheral repeats would not be essential in the interaction, which would be possible only when the central part of class A flexible patterns is involved. A second idea is that the peripheral repeats of Ts in class A flexible patterns may help proteins slide along the DNA molecule in order to reach rapidly the central part of the patterns.
Now we may wonder which interacting proteins could be involved. Here is a few requirements that must be fulfilled by proteins to be good candidates according to the features of class A flexible patterns. First requirement: proteins have to be present in large enough amount in cells in order to be good candidates. Indeed, there are many interaction sites defined by the occurrences of class A flexible patterns in genomes. Alternatively, they may be involved in a dynamic process progressively threading the whole DNA molecule through a ratchet-like mechanism (for example forcing DNA segregation into daughter cells). Second requirement: proteins must not play a role exclusively in the transcription process as the pattern occurrences can be found inside coding regions as well as outside. Third requirement: the interaction sites of proteins with the DNA molecule must not be rigidly defined, as the sites we have uncovered in the present study have never been found previously. The fourth requirement that these proteins must fulfill is related to their presence in the organisms of interest. For each candidate protein, we checked whether its distribution in organisms matched the distribution of class A flexible patterns presented in Table 1. Here are some example of plausible candidates: archaeal histones [32,33], histone-like proteins H-NS and IHF [34-40], two topoisomerases (the reverse gyrase and the topoisomerase IIB-VI) [41-44] and the SMC family of proteins [45-49].
Since the patterns are ideally shaped to display specific but labile interaction with proteins, and since they are densely present in genomes with no relationship to the position of genes, we propose that they may be involved in some biological function such as the shaping of the prokaryotic nucleoid or its segregation before cell division.
Class A flexible patterns could be recognized during homologous recombination
The widespread distribution of flexible patterns of class A along genomes is consistent with selection of the motifs through processes that are fairly ubiquitous and happen sufficiently often in the life of an organism to provide some selective advantage. Until now we have mostly considered structural or regulatory processes involving the DNA molecule as a whole. In the course of evolution the process of recombination plays an essential role as it both permits proof-reading and insertion or deletion of DNA segments. In prokaryotes, recombination involves the formation of long helical filaments of the RecA protein double-stranded DNA [50] and homologs exist in eukaryotes [51]. During the process of recombination, the DNA double helix is distorted, asking for a nucleation process of the first RecA proteins binding, making use of the flexibility of the DNA molecule. The class A flexible patterns, distributed throughout genomes, and insensitive to the origin of the DNA (regions of the genome which are from horizontal gene transfer descent are as likely to harbour the patterns as are the core regions), might play such role. Exchange of base pairs between segments undergoing recombination is essential for recognition of homology, and physical evidence indicates that such an exchange occurs early enough to mediate recognition at A:T base pairs [52]. The conserved skeleton of the class A flexible patterns would provide the required biochemical basis for the process.
Conclusion
In this article, the source of the ubiquitous bias of period 10–11 bp in genomes has been identified. It is generated by specific and ubiquitous sequences that we named "class A flexible patterns". These patterns are flexible patterns whose main property is to display 10 bp-11 bp periodic repeats of Ts. As the patterns are densely spread in genomes, their occurrences naturally generate the bias.
The patterns account for the second largest bias in the nucleotides distribution of prokaryotic genomes, second to the one generated by the use of genetic code in genes, hence their biological function has to be of an essential nature. We discussed what this function could be and suggested that class A flexible patterns could be defining a new category of protein-DNA interaction sites in genomes.
Methods
First we introduce the definition of a correlation function which is used throughout this article. Then we explain the theoretical basis of the program we designed to find the sequences responsible for short-range biases, its actual implementation and its controls.
The correlation function
Definition – a genome G
A genome G of length LG is written with ∀i ∈ [1..LG], xi ∈ {A, T, G, C}. It is taken in the standard 5'-3' orientation.
Definition – a sub-genome S extracted from a genome G
Let be a genome. A sub-genome S of length LS extracted from G is a sub-series of G. We call Esg (G) the set of all the sub-genomes of G. Then, for S ∀ Esg (G) composed of NS nucleotides, ∃σ: [1..NS] → [1..LG] a strictly increasing function so as .
Definition – a pattern m
A pattern m composed of Nm nucleotides and of length Lm is written ; Nm ≥ 1, p1 = 1, = Lm with p a strictly increasing series and ∀i ∈ [1..Nm], xi ∈ {A, T, G, C}. We call Em(N, L) the set of patterns composed of exactly N nucleotides and with a length shorter or equal to L. We call .
Definition – an occurrence of a pattern m in a genome G
Let be a pattern composed of Nm nucleotides and of length Lm.
Given the sub-genome composed of Ns = Nm nucleotides, S is an occurrence of m in G if and only if ∀i ∈ [1..Nm], xi = yσ (i) and pi = σ(i) - σ(1) + p1. We call Eoc(m, G) the set of the occurrences of m in G. # Eoc (m, G) is the number of occurrences of m in G and # Eoc (x, G) is the number of occurrences of the single nucleotide x in G.
Definition – the correlation function f (G)
Given a genome G, an order of correlation Ocor and a length for the computation of the correlation Lana, we define the correlation function f (G) on the space Em (Ocor, Lana): for , .
Our practical calculation of correlation functions is performed as follows: the function is represented by an array of size . For each nucleotide of G, the array cells of all the patterns composed of Ocor nucleotides included in the next Lana bp are increased by one. The number of steps is then proportional to .
The correlation functions of all prokaryotic and lower eukaryotic genomes reveal a strong statistical bias of period 3 bp due to the dense presence of genes in genomes [1]. This bias is of little interest as its source is known. In order to study the other biases in the present work, we always pre-treated the correlation functions so as to hide this trivial bias. This deconvolution step was performed by subtracting the correlation function of a model genome constructed so as to contain only the trivial bias. The concept of model genome has been developed in [53,54]. This is performed here as follows:
Definition – the model genome Gmo(G)
Let us write the genome G as a series of dihexanucleotides: with H = (x1x2x3x4x5x6) representing an hexanucleotide.
The model genome Gmo(G) is a random genome built from G by following these probability rules:
Definition – the treated correlation function ft (G)
The upper line means that ft (G) is the average of correlation functions of several model genomes derived from the same genome G. The treated correlation function is an average of probabilistic functions. Practically, for genomes long enough, after averaging over a few model genomes (usually three) one gets a function that almost completely lost the effects of biases with very short ranges (inferior to 6 bp) and hence lost the effect of the 3 bp periodic bias due to the presence of the genes, but saved most of the effects of other kind of information included in genomes. In the Background section, on Figure 1, we plotted ft (G) restricted on the following set of patterns: (A, A,1,l)l∈[1..100].
Definition – the complementary sub-genome of the sub-genome S
Given a genome G and a sub-genome S, we define naturally as the sub-genome of G which includes in the right order all the nucleotides of G which are not in S.
Definition – the model genome Gmo(S) for a sub-genome S
Let be a genome, be a sub-genome of G, be its complementary sub-genome and be a model genome derived from G.
Definition – the treated correlation function of a sub-genome ft (S)
ft (S) = ft (Gmo (S))
Notation – a pattern family M
A pattern family M is a finite set of patterns. It is noted , ∀i ∈ [1..Nm], mi ∈ Em.
In the Results section, and . On Figure 5, 6, we plotted two correlation functions of those two sub-genomes restricted on the following set of patterns: (A, A,1,l)l∈[1..100].
The rationale of the program
Our goal was to determine which sequences of a given genome G account for the statistical bias of period 11 bp affecting the distribution of its nucleotides. We designed a program meant to find out which sequences were responsible for all short-range non-trivial biases present in a given genome G. Here, "non-trivial" means different from the bias of period 3 bp due to the presence of the genes in genomes. Since the bias of period 11 bp is indeed a short-range bias, the sequences of G generating the bias should be included in the sequences determined by the program. Assuming that the majority of significant statistical biases present in a genome G can be revealed by the correlation function of G, our program does not look directly for the sequences generating the short-range biases but, rather, identifies the sequences generating ft (G) for a given Ocor and Lana (practically four nucleotides and thirty base-pairs). The treated correlation function of a genome that would be biased only by the genes structure is the null function. Our program stands on the approximated formula (1) that we are introducing now.
Definition – a special pattern family for the genome G
A pattern family M will be called "special pattern family" if (Eoc (m, G))m∈M covers exactly, with no overlapping, the sequences of G that generates ft (G) for a given Ocor and Lana and if the positions of the occurrences of the different patterns of M are not correlated. These conditions are written:
We call Espe (G) the set of all special pattern families of G.
Assuming that such families containing only short enough patterns (shorter than one hundred base-pairs) exist, the aim of our program was to determine one of them.
Definition – the simulated genome Gsim (G, m, β)
For a given pattern m, let Gsim (G, m, β) be the simulated genome derived from a genome G and constructed by repeatedly overwriting the pattern m on the original sequence of G (with a frequency β). We call Eocin (m, Gsim (G, m, β)) the set of all the occurrences of m artificially introduced in Gsim (G, m, β).
Property – for M ∈ Espe(G) and ,
Indeed, we have . As , we have .
Considering the way we derived the simulated genomes, it is obvious that the occurrences of the patterns m introduced in Gsim(m, G, β) are not correlated to neighboring sequences. We then assume that natural occurrences of m in G are not too much correlated to neighboring sequences. Hence one gets:
As we introduced the occurrences of the pattern m in a non-correlated manner in Gsim (m, G, β), it results that .
We have because many occurrences of the pattern m have been introduced in Gsim (m, G, β), generating very strong correlations. Hence . As many more occurrences of the pattern m were introduced in Gsim (m, G, β) than there are naturally in G, one has . Finally it comes that .
Hence the treated correlation function of G can be approximated by a linear combination of the correlation functions of the simulated genomes associated to the patterns belonging to a special pattern family. This property gave us a theoretical framework to determine such a special pattern family.
Definition – a positively free family
Let E be a vectorial space and F a family of vectors.
Let us define . The family is positively free in E if and only if
Our idea was to choose a pattern family Minput containing as many patterns as possible that is positively free. If there exists one and only one special pattern family Mspe included in Minput, then there exists a linear decomposition of ft (G) on the with positive coefficients (for any β so as ), i.e. . As this decomposition is unique, by calculating the decomposition of ft (G) on the , one can determine which patterns belong to Mspe. Hence basically our program, for an input of a genome G and a pattern family Minput, chose a suitable β, calculated the and the unique decomposition with positive coefficients of ft (G) on these functions. It gave as an output a pattern family Moutput which consisted in the patterns of Minput for which the treated correlation functions of the associated simulated genomes are involved.
Practical implementation of the program
First of all, we assumed that, for Ocor = 4 and Lana = 30 bp, there exist N > 0 and L > 0 so that there exists one and only one special pattern family included in E (N, L).
Because of computational time limitation, only input pattern families that are not containing too many patterns (less than one thousand patterns) could be tested. To extend the output possibilities of the program, we ran it in a few steps, at the cost of further approximations. First, we entered M0 = E(2,14) ∪ E(3,14) as an input family (this family is positively free). As we did not expect any special pattern family to belong to M0, we did not calculate the decomposition of ft (G) on , but rather a "positive projection" of ft (G) on .
Definition – the positive projection of a vectorial space of finite dimension in the non-void family
Be < > a scalar product in E and || || the associated norm. It is possible to prove that so as , . We call this vector , the positive projection of in F.
We calculated . The coefficients of this positive projection can be assimilated to a frequency of patterns present in G, expressed in bp-1. Then we constructed M1 the output pattern family with all the patterns of M0 for which the coefficient of the treated function of the associated simulated genome is large enough. The selectivity of the program is adjustable at this level. Practically, we kept the patterns for which the coefficients are above , with an average approximately , which makes usually approximately twenty patterns. This is a first approximation in our program. As a second step, we used M2 as an input pattern family. M2 is containing M1 plus all the patterns that can be built by extending the patterns of M1 with one extra nucleotide. The added nucleotide can be placed at any position inside the original patterns or at their sides (as far as 15 bp from the extremities of the original patterns). Again, we calculated a positive projection and got a resulting pattern family M3.
We repeated this step as long as we got patterns that were strictly included in (i.e. all the patterns that are composed of up to six nucleotides and span less than 30 bp). We got usually close to one hundred patterns in this pattern family. Let us call Mfinal this resulting pattern family. It is an approximation of Mspe. Then, by merging the patterns (composed of six nucleotides) that could be identify as subsets of a same longer pattern (composed of more than six nucleotides), we obtained patterns that belonged to while becoming closer to Mspe. Finally, from the patterns contained in Mfinal, we could define approximately twenty flexible patterns per organisms (see the Results section).
Besides the approximation generated by the division of the program into a few steps, a few more approximations were introduced during that process. First, the calculation of the positive projection was performed approximately so as to save calculation time. Second, the correlation functions were calculated on restricted sets, practically on E(4,30), i.e. Ocor = 4 and Lana = 30 bp. This made the description of patterns approximate since we aimed at determining patterns containing more than four nucleotides. The correlation order should be longer than the maximum number of nucleotides we want to find in patterns, otherwise the program may find patterns which are actually artefacts (a mix of genuine patterns present at distinct locations in the genome).
The program was written in C code. Built and operated in this way, the program was run on a genome of 2 Mbp in 3 weeks with a 1.8 Ghz G5 CPU. The most time-consuming step is the calculation of the correlation functions with Ocor = 4 and Lana = 30 bp.
Controls of the program
Different controls were performed to test the selectivity of the program. First, when run on completely random genomes, the coefficients of the first positive projection were below the threshold, so that the resulting pattern family was empty. Second, the program was also tested with artificial genomes built from completely random genomes in which we introduced a given pattern at random locations. The program proved able to extract the pattern back provided that the pattern frequency of introduction was above . Third, the program proved able to identify already known rigid patterns in genomes (see the Results section).
Authors' contributions
EL designed the algorithm and performed the bulk of the outlined study. AD proposed the rationale for the study and outlined its biological implications. Both authors participated in the writing of this article.
Acknowledgements
This work was supported by the BIOSUPPORT program of the Innovation and Technology Fund (ITF) of the Hong Kong's government. We are grateful to Benoît Arcangioli for his suggestion regarding the RecA recognition hypothesis.
==== Refs
Herzel H Weiss O Trifonov EN 10–11 bp periodicities in complete genomes reflect protein structure and DNA folding Bioinformatics 1999 15 187 193 10222405 10.1093/bioinformatics/15.3.187
Fukushima A Ikemura T Oshima T Mori H Detection of period in eukaryotic genomes on the basis of power spectrum analysis Genome Inform Ser Workshop Genome Inform 2002 13 21 29 14571371
Li W Stolovitzky G Bernaola-Galvan P Olivier JL Compositional heterogeneity within, and uniformly between, DNA sequences of yeast chromosomes Genome Res 1998 8 916 918 9750191
Peng CK Buldyrev SV Havlin S Simons M Stanley HE Goldberger AL Mosaic organization of DNA nucleotides Phys Rev E Stat Phys Plasmas Fluids Relat Interdiscip Topics 1994 49 1685 1689 9961383
Herzel H Ebeling W Schmitt AO Entropies of biosequences: the role of repeats Phys Rev E Stat Phys Plasmas Fluids Relat Interdiscip Topics 1994 50 5061 5071 9962591
Schmitt AO Herzel H Estimating the entropy of DNA sequences J Theor Biol 1997 188 369 377 9344742 10.1006/jtbi.1997.0493
Yu ZG Anh VV Wang B Correlation property of the length sequences based on global structure of the complete genome Phys Rev E Stat Nonlin Soft Matter Phys 2001 63 011903 11304283
Buldyrev SV Goldberder AL Havlin S Mantegna RN Matsa ME Peng CK Simons M Stanley HE Long-range correlation properties of coding and non-coding DNA sequences: GenBank analysis Phys Rev E Stat Phys Plasmas Fluids Relat Interdiscip Topics 1995 51 5084 5091 9963221
Audit B Vaillant C Arneodo A d'Aubenton-carafa Y Thermes C Long-range correlations between sites: relation to the structure and dynamics of nucleosomes J Mol Biol 2002 316 903 918 11884131 10.1006/jmbi.2001.5363
Grosse I Herzel H Buldyrev SV Stanley HE Species independence of mutual information in coding and non-coding DNA Phys Rev E Stat Phys Plasmas Fluids Relat Interdiscip Topics 2000 61 5624 5629 11031617
Mantegna RN Buldyrev SV Goldberger AL Havlin S Peng CK Simons M Stanley HE Linguistic features of non-coding DNA sequences Phys Rev Lett 1994 73 3169 3172 10057305 10.1103/PhysRevLett.73.3169
Shepherd JC Method to determine the reading frame of a protein from the purine/pyrimidine genome sequence and its possible evolutionary justification Proc Natl Acad Sci USA 1981 78 1596 600 6940175
Shepherd JC Periodic correlations in DNA sequences and evidence suggesting their evolutionary origin in a comma-less genetic code J Mol Evol 1981 17 94 102 7253039 10.1007/BF01732679
Staden R Finding protein coding regions in genomic sequences Methods Enzymol 1990 183 163 180 2314274
Tsonis AA Elsner JB Tsonis PA Periodicity in DNA coding sequences: implications in gene evolution J Theor Biol 1991 151 323 331 1943144
Gutierrez G Oliver JL Marin A On the origin of the periodicity of three in protein coding DNA sequences J Theor Biol 1994 167 413 414 8207954 10.1006/jtbi.1994.1080
Fukushima A Ikemura T Kinouchi M Oshima T Kudo Y Mori H Kanaya S Periodicity in prokaryotic and eukaryotic genomes identified by power spectrum analysis Gene 2002 300 203 211 12468102 10.1016/S0378-1119(02)00850-8
Fickett JW Tung CS Assessment of protein coding measures Nucleic Acids Res 1992 20 6441 6450 1480466
Schieg P Herzel H Periodicities of 10–11 bp as indicators of the supercoiled state of Genomic DNA J Mol Biol 2004 343 891 901 15476808 10.1016/j.jmb.2004.08.068
Espéli O Moulin L Boccard F Transcription attenuation associated with bacterial repetitive extragenic BIME elements J Mol Biol 2001 314 375 386 11846552 10.1006/jmbi.2001.5150
Makarova KS Aravind L Wolf YI Tatusov RL Minton KW Koonin EV Daly MJ Genome of the Extremely Radiation-Resistant Bacterium Deinococcus radiodurans Viewed from the Perspective of Comparative Genomics Microbiol Mol Biol Rev 2001 65 44 79 11238985 10.1128/MMBR.65.1.44-79.2001
Bentley SD Maiwald M Murphy LD Pallen MJ Yeats CA Dover LG Norberczack HT Besra GS Quail MA Harris DE von Herbay A Goble A Rutter S Squares R Barell BG Parkhill J Relman DA Sequencing and analysis of the genome of the Whipple's disease bacterium Tropheryma whipplei The Lancet 2003 361 637 644 12606174 10.1016/S0140-6736(03)12597-4
Fitz-Gibbon ST Ladner H Kim UJ Stetter KO Simon MI Miller JH Genome sequence of the hyperthermophilic crenarchaeon Pyrobaculum aerophilum Proc Natl Acad Sci USA 2002 99 984 989 11792869 10.1073/pnas.241636498
Berman HM Westbrook J Feng Z Gilliland G Bhat TN Weissig H Shindyalov IN Bourne PE The Protein Data Bank Nucleic Acids Res 2000 28 235 242 10592235 10.1093/nar/28.1.235
Mirny LA Gelfand MS Structural analysis of conserved base pairs in protein-DNA complexes Nucleic Acids Research 2002 30 1704 1711 11917033 10.1093/nar/30.7.1704
Strick TR Allemand JF Bensimon D Croquette V Behavior of supercoiled DNA Biophys J 1998 74 2016 2028 9545060
Dervan PB Burli RW Sequence-specific DNA recognition by polyamides Curr Opin Chem Biol 1999 3 688 693 10600731 10.1016/S1367-5931(99)00027-7
Moravek Z Neidle S Schneider B Protein and drig interactions in the minor groove of DNA Nucleic Acids Res 2002 30 1182 1191 11861910 10.1093/nar/30.5.1182
O'flanagan RA Paillard G Lavery R Sengupta AM Non-additivity in protein-DNA binding Bioinformatics 2005 15746285
Travers A Muskhelishvili G DNA supercoiling – a global transcriptional regulator for enterobacterial growth ? Nat Rev Microbiol 2005 3 157 169 15685225 10.1038/nrmicro1088
Stefl R Wu H Ravindranathan S Sklenar V Feigon J DNA A-tract bending in three dimensions: solving the dA4T4 vs. dT4A4 conundrum Proc Natl Acad Sci USA 2004 101 1177 1182 14739342 10.1073/pnas.0308143100
Malik HS Henikoff S Phylogenomics of the nucleosome Nat Struct Biol 2003 10 882 890 14583738 10.1038/nsb996
Pavlov NA Cherny DI Jovin TM Slesarev AI Nucleosome-like complex of the histone from the hyperthermophile Methanopyrus kandleri (MkaH) with linear DNA J Biomol Struct Dyn 2002 20 207 214 12354072
Nishino K Yamaguchi A Role of Histone-Like Protein H_NS in Multidrug Resistance of Escherichia coli J Bacteriol 2004 186 1423 1429 14973023 10.1128/JB.186.5.1423-1429.2004
Rouquette C Serre MC Lane D Protective role for H_NS protein in IS1 transposition J Bacteriol 2004 186 2091 2098 15028694 10.1128/JB.186.7.2091-2098.2004
Rimsky S Structure of the histone-like protein H-NS and its role in regulation and genome superstructure Curr Opin Microbiol 2004 7 109 114 15063845 10.1016/j.mib.2004.02.001
Tendeng C Bertin PN H-NS in Gram-negative bacteria: a family of multifaceted proteins Trends Microbiol 2003 11 511 517 14607068 10.1016/j.tim.2003.09.005
Murtin C Engelhorn M Geiselmann J Boccard F A quantitative UV laser footprinting analysis of the interaction of IHF with specific binding sites: re-evaluation of the effective concentration of IHF in the cell J Mol Biol 1998 284 949 961 9837718 10.1006/jmbi.1998.2256
Lynch TW Read EK Mattis AN Gardner JF Rice PA Integration host factor: putting a twist on protein-DNA recognition J Mol Biol 2003 330 493 502 12842466 10.1016/S0022-2836(03)00529-1
Swinger KK Rice PA IHF and HU: flexible architects of bent DNA Curr Opin Struct Biol 2004 14 28 35 15102446 10.1016/j.sbi.2003.12.003
Champoux JJ DNA topoisomerases: structure, function, and mechanism Annu Rev Biochem 2001 70 369 413 11395412 10.1146/annurev.biochem.70.1.369
Massé E Drolet M Relaxation of Transcription-induced Negative Supercoiling Is an Essential Function of Escherichia coli DNA Topoisomerase I J Biol Chem 1999 274 16654 16658 10347233 10.1074/jbc.274.23.16654
Massé E Drolet M Escherichia coli DNA Topoisomerase I Inhibits R-loop Formation by Relaxing Transcription-induced Negative Supercoiling J Biol Chem 1999 274 16659 16664 10347234 10.1074/jbc.274.23.16659
Bouthier de la Tour C Portemer C Nadal M Stetter KO Forterre P Duguet M Reverse Gyrase, a Hallmark of the Hyperthermophilic Archaebacteria J Bacteriol 1990 172 6803 6808 2174859
Cobbe N Heck MM Review: SMCs in the world of chromosome biology- from prokaryotes to higher eukaryotes J Struct Biol 2000 129 123 143 10806064 10.1006/jsbi.2000.4255
Cobbe N Heck MM The evolution of SMC proteins: phylogenetic analysis and structural implications Mol Biol Evol 2004 21 332 347 14660695 10.1093/molbev/msh023
Melby TE Ciampaglio CN Briscoe G Erickson HP The symmetrical structure of structural maintenance of chromosomes (SMC) and MukB proteins: long, antiparallel coiled coils, folded at a flexible hinge J Cell Biol 1998 142 1595 1604 9744887 10.1083/jcb.142.6.1595
Löwe J Cordell SC Van den Ent F Crystal structure of the SMC head domain: An ABC ATPase with 900 residues antiparallel coiledcoil inserted J Mol Biol 2001 306 25 35 11178891 10.1006/jmbi.2000.4379
Haering CH Löwe J Hochwagen A Nasmyth K Molecular architecture of SMC proteins and the yeast cohesin complex Mol Cell 2002 9 773 788 11983169 10.1016/S1097-2765(02)00515-4
Prevost C Takahashi M Geometry of the DNA strands whithin the RecA nucleofilaments: role in homologous recombination Q Rev Biophys 2003 36 429 453 15267169 10.1017/S0033583504003956
Krogh BO Symington LS Recombination proteins in yeast Annu Rev Genet 2004 38 233 271 15568977 10.1146/annurev.genet.38.072902.091500
Gupta RC Folta-Stogniew E O'Malley S Takahashi M Radding CM Rapid exchange of A:T base pairs is essential for recognition of DNA homology by human Rad51 recombination protein Mol Cell 1999 4 705 714 10619018 10.1016/S1097-2765(00)80381-0
Karlin S Cardon LR Computational DNA sequence analysis Annu Rev Microbiol 1994 48 619 654 7826021 10.1146/annurev.mi.48.100194.003155
Hénaut A Lisacek F Nitschké P Moszer I Danchin A Global analysis of genomic texts: the distribution of AGCT tetranucleotides in the Escherichia coli and Bacillus subtilis genomes predicts translational frameshifting and ribosomal hopping in several genes Electrophoresis 1998 19 515 527 9588797 10.1002/elps.1150190411
|
16120222
|
PMC1242344
|
CC BY
|
2021-01-04 16:27:28
|
no
|
BMC Bioinformatics. 2005 Aug 24; 6:206
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-206
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2221615329310.1186/1471-2105-6-222DatabaseMicro-Mar: a database for dynamic representation of marine microbial biodiversity Pushker Ravindra [email protected]'Auria Giuseppe [email protected] Jose Carlos [email protected]íguez-Valera Francisco [email protected] Evolutionary Genomics Group, Universidad Miguel Hernández, Apartado 18, 03550 San Juan de Alicante, Alicante, Spain2005 9 9 2005 6 222 222 10 5 2005 9 9 2005 Copyright © 2005 Pushker et al; licensee BioMed Central Ltd.2005Pushker et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The cataloging of marine prokaryotic DNA sequences is a fundamental aspect for bioprospecting and also for the development of evolutionary and speciation models. However, large amount of DNA sequences used to quantify prokaryotic biodiversity requires proper tools for storing, managing and analyzing these data for research purposes.
Description
The Micro-Mar database has been created to collect DNA diversity information from marine prokaryotes for biogeographical and ecological analyses. The database currently includes 11874 sequences corresponding to high resolution taxonomic genes (16S rRNA, ITS and 23S rRNA) and many other genes including CDS of marine prokaryotes together with available biogeographical and ecological information.
Conclusion
The database aims to integrate molecular data and taxonomic affiliation with biogeographical and ecological features that will allow to have a dynamic representation of the marine microbial diversity embedded in a user friendly web interface. It is available online at .
==== Body
Background
The global oceanic ecosystem is highly dependent on the activity of its large population of prokaryotes. However, their small size, relatively diluted environment and reluctance to be grown in pure culture make marine prokaryotes also one of the less known group of microbes. During the last decade, PCR based approaches have produced an enormous amount of information in this field, mostly based on the 16S rRNA genes due to their accepted taxonomic relevance [1]. Other genes such as rpoB [2] and recA [3] have started to be used on the grounds of their evolutionary stability. In recent years the sequencing of large insert libraries such as BACs or Fosmids have produced an important additional source of information. Last year, the Whole Genome Shotgun (WGS) applied to the Sargasso Sea produced a sequence database of 1.045 billion base pairs [4]. Still, many databases deal with only 16S or ITS sequences. For example, the Ribosomal Database Project counts 101632 entries corresponding to 16S sequences [5] and RISSC has more than 1600 entries corresponding to 16S-23S ribosomal spacer sequences [6]. While genomic sequences submission is continuously growing, there is not a clear correspondence in the improvement of analytical power of this enormous amount of information.
Micro-Mar is a novel database storing publicly available marine prokaryotes sequences along with their biogeographical (sampling site, latitude, longitude etc.) and ecological information (depth, temperature, salinity etc.). Each entry represents an individual marine prokaryote with one or more DNA sequences coming from a particular sampling location and depth. The database aims not only to provide a collection of marine prokaryotes data, but also a research tool to relate microbial biodiversity with its environment, opening possibilities for studying adaptations at the level of the microbial community, designing water management strategies, pollution detection or marine productivity prediction.
Construction and content
In order to retrieve marine prokaryotes sequences from the NCBI [7], ad hoc queries were used. Moreover manual sequence searches were also carried out. All the sequences obtained were downloaded in GenBank format. Some of the details, such as geographic origin, depth, temperature etc., were obtained manually (if not available) by searching within the publications or by direct interaction with the authors. Type of entry indicates whether a sequence comes from offshore, inshore or sediments and a PCR product, a cloned DNA product or an isolated strain. A BLAST [8] search against Micro-Mar was performed in order to get the closest marine prokaryotes sequence and also the closest taxonomic unit (generally a pure culture). Top fifty BLAST hits are also available on the webpage to give more idea about the complete similarity profile for a particular sequence. Top fifty BLAST hits to whole NCBI nucleotide sequence database are also reported. The complete dataset was loaded in to MySQL [9] relational tables. Micro-Mar uses LAMP: The Open Source Web Platform [10]. Geographic Information System (GIS) uses JpGraph library [11] to display different sampling locations on a world map. All the web pages follow HTML 4.01 standard and use CSS for consistent styling.
Utility
There are five major options available in the Micro-Mar database: (i) search, (ii) GIS, (iii) local BLAST, (iv) MMSeqUp and (v) forum.
Search
The search option provides an interface for a large number of queries to the database. It can be used to search the database for 5S rRNA, 16S rRNA, ITS, 23S rRNA or CDS sequences along with various biogeographical and ecological parameters. The results can be either in tabular format or in a world map showing different sampling sites through GIS. Each entry is linked back to other databases such as NCBI for more information. A number of entries can be selected to analyze further along with the given sequences by aligning using CLUSTALW [12] and a tree can be created using PHYLIP [13] to see the phylogenetic position of submitted sequences against the Micro-Mar sequences. Alignment files (PHYLIP format) and tree files (Newick tree format, Postscript and PDF) are also available for download.
GIS
The GIS option provides an interface for selecting a particular sampling location on the world map and getting all the sequences from that location and their details. Furthermore, for a selected region on the map, the following information can be obtained, i) taxonomy report: taxonomic details at different levels (domain, phylum, class, order, family and genus); ii) depth report: a plot showing number of sequences vs depth; iii) biodiversity report: a list of organisms found; iv) get all entries and v) advanced search. The reports allow to retrieve sequences corresponding to a particular taxonomy, depth or biodiversity.
Local BLAST
The local BLAST option can be used to do a BLAST search against Micro-Mar database to get the most similar sequences to the submitted sequences. The results can be either in default BLAST format or in a tabular format. All the hits can be selected and displayed on the world map using GIS and all the GIS features discussed above can be used. Selected sequences can be downloaded in FASTA Format and also analyzed further by aligning and creating a tree along with the given sequences. Alignment and tree files can also be downloaded in different formats as described in the search option.
MMSeqUp
MMSeqUp facilitates online sequence submission to the Micro-Mar database. It allows users to upload a file containing new marine prokaryotes sequences. Related biogeographical and ecological parameters can also be submitted on the webpage.
Forum
An online forum powered by PHPBB [14] has been created for the following tasks: (i) FAQs: a compilation of frequently asked questions, (ii) suggestions, (iii) discussion: open discussion and iv) feedback: comments on the Micro-Mar web interface v) What's new: Recent developments in the database.
Discussion
Micro-Mar currently has 8187 entries consisting of 11874 sequences including 5693 16S rDNA, 2177 ITS, 170 23S rDNA, 3448 CDS and one 5S rDNA. Micro-Mar sequences cover 192 different sampling sites widespread on the world oceanic map from Arctic to Antarctic environments representing almost all the oceans (Figure 1). Inshore and offshore representatives are also present demonstrating a wide range of depth, going from surface water (0.5 m) to the deepest of the Mariana Trench (10898 m). The entries fall in two superkingdoms of Archea (959) and Bacteria (6351). There is also a group of 877 unclassified entries. Out of 8187 entries, 7672 entries i.e. more than 93% of entries have complete geographic information available in the database. Similarly more than 85% of entries have depth information available. All the entries have taxonomic details linked back to the NCBI taxonomy database [15]. The details of taxonomic distribution at phylum and class level are summarized in Table 1 [see Additional file 1]. As expected in the marine environment, the biggest taxonomic group is represented by the Proteobacteria with all the 5 classes (α, β, γ, δ and ε) as shown in the Table 2 [see Additional file 1].
Figure 1 World map showing all sampling locations. A figure showing 192 different geographic locations widespread on the world oceanic map from Arctic to Antarctic environment.
Conclusion
The creation of Micro-Mar database is an initiative towards cataloging all the information related to marine prokaryotes collected during the lasts two decades and providing an interface that will help the scientific community to do comparative analyses of marine prokaryotes sequences and make it amenable for biogeographical and ecological analyses. The database is updated every week to include the most recent marine prokaryotic sequences. In near future, more samples from extreme environments will be integrated in the database to improve the analytical power and the biodiversity range. As more and more entries are incorporated, it will be possible to correlate accurately the bacterial biodiversity with biogeographical and ecological parameters giving a global overview of the various aspects of the biodiversity within the oceans. In order to achieve this, it is encouraged that scientists include more information about biogeographical and ecological parameters while submitting their sequences to various public databases.
Availability and requirements
The database is available at . A latest web browser with JavaScript enabled is required to use it.
List of abbreviations
GIS – Geographic Information System
LAMP – Linux + Apache + MySQL + PERL/PHP/Python
Authors' contributions
FRV conceived the study and the general design of Micro-Mar. RP and GD drafted the manuscript. GD and JCA did the specific design, structure development and data input. All the informatics applications were designed and developed by RP. All authors read and approved the final manuscript.
Supplementary Material
Additional file 1
Taxonomic distribution of Micro-Mar sequences Table-1 shows taxonomic distribution of sequences from different domains at "Class" level and Table-2 shows taxonomic distribution of sequences from Proteobacteria class at "Family" level.
Click here for file
Acknowledgements
This work was funded by MIRACLE (EVK3-2002-00087), GEMINI (QLK3-CT-2002-02056) projects of the European Commission and "Mineria Genómica" project of Generalitat Valenciana (GRUPOS03/060). We would like to thank Alex Mira for helping with the manuscript and providing constructive comments. We also thank Boris A. Legault and many other users for giving their feedback on the database.
==== Refs
Woese C Stackebrandt E Macke T Fox G A phylogenetic definition of the major eubacterial taxa Syst Appl Microbiol 1985 6 143 151 11542017
Mollet C Drancourt M Raoult D rpoB sequence analysis as a novel basis for bacterial identification Mol Microbiol 1997 26 1005 1011 9426137 10.1046/j.1365-2958.1997.6382009.x
Karlin S Weinstock G Brendel V Bacterial classifications derived from recA protein sequence comparisons J Bacteriol 1995 177 6881 6893 7592482
Venter J Remington K Heidelberg J Halpern A Rusch D Eisen J Wu D Paulsen I Nelson K Nelson W Fouts D Levy S Knap A Lomas M Nealson K White O Peterson J Hoffman J Parsons R Baden-Tillson H Pfannkoch C Rogers YH Smith H Environmental Genome Shotgun Sequencing of the Sargasso Sea Science 2004 304 66 74 15001713 10.1126/science.1093857
Cole J Chai B Farris R Wang Q Kulam S McGarrell D Garrity G Tiedje J The Ribosomal Database Project (RDP-II): sequences and tools for high-throughput rRNA analysis Nucleic Acids Res 2005 33 D294 D296 15608200 10.1093/nar/gki038
Garcia-Martinez J Bescos I Rodriguez-Sala J Rodriguez-Valera F RISSC: a novel database for ribosomal 16S-23S RNA genes spacer regions Nucleic Acids Res 2001 29 178 180 11125084 10.1093/nar/29.1.178
NCBI
Altschul SF Madden TL Schaffer AA Zhang J Zhang Z Miller W Lipman DJ Gapped BLAST and PSI-BLAST: a new generation of protein database search programs Nucleic Acids Res 1997 25 3389 3402 9254694 10.1093/nar/25.17.3389
MySQL
LAMP
JpGraph
Thompson J Higgins D Gibson T CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice Nucleic Acids Res 1994 22 4673 4680 7984417
Felsenstein J PHYLIP (Phylogeny Inference Package) version 3.6 Distributed by the author Department of Genome Sciences, University of Washington, Seattle; 2004
PHPBB
Wheeler D Chappey C Lash A Leipe D Madden T Schuler G Tatusova T Rapp B Database resources of the National Center for Biotechnology Information Nucleic Acids Res 2000 28 10 14 10592169 10.1093/nar/28.1.10
|
16153293
|
PMC1242345
|
CC BY
|
2021-01-04 16:27:27
|
no
|
BMC Bioinformatics. 2005 Sep 9; 6:222
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-222
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2231615330410.1186/1471-2105-6-223SoftwareSUPERFICIAL – Surface mapping of proteins via structure-based peptide library design Goede Andrean [email protected] Ines S [email protected] Robert [email protected] Berlin Center for Genome Based Bioinformatics, 3D Data Mining Group, Institute of Biochemistry, Charité, Monbijoustr.2, 10117 Berlin, Germany2005 9 9 2005 6 223 223 17 3 2005 9 9 2005 Copyright © 2005 Goede et al; licensee BioMed Central Ltd.2005Goede et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The determination of protein surfaces and the detection of binding sites are essential to our understanding of protein-protein interactions. Such binding sites can be characterised as linear and non-linear, the non-linear sites being prevailant. Conventional mapping techniques with arrays of synthetic peptides have limitations with regard to the location of discontinuous or non-linear binding sites of proteins.
Results
We present a structure-based approach to the design of peptide libraries that mimic the whole surface or a particular region of a protein. Neighbouring sequence segments are linked by short spacers to conserve local conformation. To this end, we have developed SUPERFICIAL, a program that uses protein structures as input and generates library proposals consisting of linear and non-linear peptides. This process can be influenced by a graphical user interface at different stages, from the surface computation up to the definition of spatial regions.
Conclusion
Based on 3D structures, SUPERFICIAL may help to negotiate some of the existing limitations, since binding sites consisting of several linear pieces can now be detected.
==== Body
Background
In order to perform their functions, protein surfaces usually have to interact with each other. However, only accessible parts of a protein can act as binding sites [1]. Since proteins consist of polypeptide chains that fold into complex three-dimensional patterns, binding sites can be divided into two different types: 1. sites that follow the primary amino acid sequence as a continuous or linear interaction site. 2. discontinuous or non-linear binding sites, which are made up of short peptide fragments that are not adjacent in the sequence but are in spatial proximity as a result of folding. Non-linear binding sites predominate in both protein-protein interactions, and in protein binding of small compounds [2]. Their detection is challenging because conventional mapping techniques have limited capabilities [3,4]. The increasing number of structurally-determined proteins often permits a structure-based automated approach to the design of peptide libraries that can mimick particular surface regions. As Atassi et al. [5] and Lee et al [6] proposed, spatially neighbouring sequence segments have to be linked by short (peptidic) linkers to conserve local conformation. To facilitate this process, we have integrated the LIP database containing all peptidic fragments derived from the Brookhaven Protein Data Bank (PDB) up to a length of 15 residues [7]. SUPERFICIAL makes it possible to scan a specific part of the protein or the whole protein. Determination of the peptides and selection of the linkers are automated, and substantial peptide libraries can be generated.
Implementation
The program was implemented in Delphi and is designed for versions of Windows 98 upwards.
Three problems have to be solved:
1. Determination of those parts of the protein surface that provide the basis of the peptide library.
2. Localisation of those peptides that are neighboured in space (but not in sequence) and form a potential non-linear binding site.
3. Detection of linkers to connect the spatially neighbouring peptides in consideration of the local conformation.
Determination of the surface segments
At first, the library should contain only peptides that mimic the surface of the protein, or of the selected protein chain. Therefore, the peptides themselves should consist mainly of amino acids that are solvent-accessible. In general, there are several possibilities of defining an amino acid as surface-exposed. One can estimate the proportion of the surface area of an amino acid that is accessible to water [8] and set a threshold for this value. The threshold, however, can be varied for each type of amino acid. Since the packing of protein structures differs depending on the size, degree of polymerisation, and origin of the structure (NMR, crystal or a model), there is no threshold matching all kinds of structures.
SUPERFICIAL meets that challenge by automatically evaluating the solvent-accessibility for each atom. Depending on the proportion of atoms exposed to the surface (Fig. 2, section C and Table), the accessibility of an amino acid is divided into two states – buried (non-accessible) or exposed (accessible). This option can be used to modify the extension of the protein's surface. If only exposed amino acids are considered for the peptide library, the resulting peptides become very small, notably in scanned semi-exposed helical regions; thus small gaps require filling. For this purpose, a sliding-window technique was used. The user defines a window (Fig. 2, section C) that scrolls down the sequence of the surface to close gaps or eliminate detached amino acids. The resulting solvent-accessible sequence segments represent the surface of the protein and therefore provide the basis for the generation of a peptide library. These segments mimic potential linear binding sites, whereas the non-linear binding sites consist of several segments.
Figure 2 Screenshot of SUPERFICIAL displaying the options. Sections A and B are the same for all submenus/menu items ("load protein", "show", "options" and "peptide"). Section A gives a short description of the options and may act as a guide for the user. In section B the subsequent results of the settings are shown and the user may check the effects on the size of the surface and the peptides. The options in C determine the surface of a protein, whereas the first entry ("percentage of atoms at surface per amino acid") has the greatest influence on the surface extension. Section D gives the definitions for peptide generation. All changes are visualised in B on sequence level. The whole protein is displayed in the submenu "show" (Fig. 3).
Peptide generation
If only linear peptides are of interest, their length can be defined (Fig. 2, section D). The solvent-accessible sequence segments are then tailored accordingly. The procedure to identify and assemble the non-linear peptides is more sophisticated. Starting from one linear peptide-fragment, the surrounding space is scanned in a user-defined diameter (Fig. 2, section D). Peptide-fragments within this diameter are combined to form a single entity.
Search for linkers
To preserve their conformation, the gaps between the peptide-fragments are filled with linkers, short amino acid sequences derived from the LIP (Loops in Proteins) database [7]. The LIP database contains all peptidic fragments from the PDB up to a length of 15 residues. The peptidic fragments obtained from LIP and the peptide-fragments generated by SUPERFICIAL are combined to form the complete non-linear peptides.
The linkers are integrated depending on the distances and angles of the stem atoms, as described in [7]. All possible arrangements of the peptide-fragments of the protein are examined. For each combination the shortest linkers are determined, and the one with the shortest total length is accepted. This procedure may change the order of the peptide-fragments, in case it shortens the linker. Additionally, it minimises the insertion of foreign amino acids.
The current size of the LIP database is approximately 8 Gigabytes, and it contains about 100 million entries. To connect to this database, it is necessary to install this large amount of data. Instead of the whole database, the downloadable version of SUPERFICIAL implements a table that is derived from the LIP database. This table contains a grid of parameters (distances and angles) along with the corresponding number of amino acids necessary to bridge a gap between two peptides. Applying the table instead of the LIP database allows rapid identification of appropriate peptide linkers, though their sequence is arbitrary. Amino acids are represented by the character "X" that can be replaced in praxis by poly-alanine and/or glycine.
Results and discussion
SUPERFICIAL has been tested on Windows 98, NT, 2000 and XP. Additional visualization tools are not required. It can read files in PDB format, which are either derived from the PDB or from modelling. We have successfully tested proteins up to 50,000 atoms, though the maximum size accepted is dependent on computer memory.
SUPERFICIAL automatically defines the protein surface, using preset default values applicable to a range of proteins. To consider the heterogeneity of proteins and for "fine-tuning", the user can choose between various options to specify the surface area (Fig. 2, section A). The user can scan either the entire protein (Fig. 3), selected chains, or a region of specific interest (Fig. 4). The program will only consider the selected part of the protein for scanning and producing a peptide library. All effects of the settings are shown at sequence level in the window above (Fig. 2, section B), and on the annotated 3D structure of the protein (Fig. 3), where the surface is highlighted. When the peptide library is complete, every peptide can be displayed individually and discarded if required. The whole project can be saved and restored at any stage of the process, so different settings can be compared.
Figure 3 Screenshot of SUPERFICIAL showing the 3D view of the protein. The functionality of this tool is exemplified by the crystal structure of a complex between influenza virus neuraminidase and an antibody (PDB-code: 1a14).
Figure 4 Screenshot to illustrate the selection of a region (white ellipse). The peptide library will be generated for this region only.
To avoid problems during peptide synthesis, amino acids can be automatically replaced, e.g. cysteine versus serine. All generated peptides are listed within a saveable table. Such a structure-based peptide library provides the source for chemically-prepared peptide arrays to identify and characterise binding sites, respectively [9,10].
General discussion
Atassi et al. [5,11] and Lee et al. [6] proposed the idea of linking several peptides forming a non-linear binding site with short peptidic linkers. They first identified the amino acids of a non-linear antigenic site in native lysozyme and then linked them into a single peptide by inserting glycine residues. A different approach was used by Casset et al. [12], Franke et al. [13] and Eichler [14]. They used circular scaffolds to present the peptides of a non-linear binding site, and these structures maintained the conformation of the peptides found in the original protein. For all these methods, detailed structural information of the binding site or the interacting amino acids has to be available. These problems are overcome with SUPERFICIAL, since only the structure of the protein is required. Determination of the surface and selection of the peptides can be influenced by the user, while the selection of the linker and the generation of the peptide library are automatic. The whole library provides the basis for a high-throughput synthesis (e.g. the SPOT-synthesis [15]) and the identification of binding peptides.
Methods to connect peptides with linkers are mostly used during homology modelling. Generally, two approaches are applied: ab initio or knowledge-based methods. Ab initio methods usually scan the whole conformational space, while knowledge-based methods search for protein segments with a known three-dimensional structure that fits into a gap. Both methods assess the possible linkers according to potential or scoring functions. For ab initio methods, the complexity, and therefore the time and effort increase with the length of the linker. As shown in [7], detection of suitable linkers by means of LIP is usually performed faster and more accurately than by other methods.
Non-peptidic linkers between peptides can also be applied, but in contrast to the 100 million linkers contained in LIP, their number and availability are limited. Therefore, not all possible conformations of peptide-fragments can be conserved with non-peptidic linkers. Currently, there is no public database of non-peptidic structures that can serve as linkers. Although the combination of peptide fragments and non-peptidic linkers or scaffolds can be advantageous if only a small number of structures is to be synthesised, such a method is not applicable for a high-throughput synthesis.
Predictions concerning the nature of antigenicity and binding sites have a large literature. Determining the antigenicity of different proteins implies that such areas share common properties [16]. Mostly, these involve the hydrophilicity, flexibility and accessibility of a protein. The program BEPITOPE, for example, uses such properties to predict linear protein epitopes and rank them according to their hydrophobicity [17]. SUPERFICIAL follows a different approach: the 3D structure of the whole surface, or parts of it, are considered and transformed into a peptide library representing this surface. Currently, it is the only program that identifies potential non-linear binding sites. Even though information on probable binding sites is not given, SUPERFICIAL includes all potential binding sites by examining the entire protein surface.
Conclusion
SUPERFICIAL is a unique tool for surface mapping, which considers the 3D structure of a protein and translates it into a peptide library. The most novel aspect of this program is its ability to propose peptides that can mimic non-linear binding sites, making it interesting, for instance, in vaccine development.
Availability and requirements
A free version of SUPERFICIAL is available for academic use at :
• Project name: SUPERFICIAL
• Project home page:
• Operating system(s): Windows 98 upwards
• Programming language: Delphi
• Other requirements: none
• Restrictions to use by academics: registration needed
• Restrictions to use by non-academics: licence needed
Authors' contributions
AG created the program, helped to draft the manuscript, web site and demos. ISJ drafted the manuscript, the web site and demos. RP was the coordinator of the project.
Figure 1 Flow chart to illustrate the process from loading a protein to the generation of the peptide library.
Acknowledgements
This work was supported by the BMBF-funded Berlin Center for Genome Based Bioinformatics (BCB).
==== Refs
Ma B Elkayam T Wolfson H Nussinov R Protein-protein interactions: structurally conserved residues distinguish between binding sites and exposed protein surfaces Proc Natl Acad Sci U S A 2003 100 5772 5777 12730379 10.1073/pnas.1030237100
Barlow DJ Edwards MS Thornton JM Continuous and Discontinuous Protein Antigenic Determinants Nature 1986 322 747 748 2427953 10.1038/322747a0
Reineke U Sabat R Volk HD Schneider-Mergener J Mapping of the interleukin-10/interleukin-10 receptor combining site Protein Sci 1998 7 951 960 9568901
Tribbick G Multipin peptide libraries for antibody and receptor epitope screening and characterization J Immunol Methods 2002 267 27 35 12135798 10.1016/S0022-1759(02)00138-2
Atassi MZ Lee CL Pai RC Enzymic and immunochemical properties of lysozyme. XVI. A novel synthetic approach to an antigenic reactive site by direct linkage of the relevant conformationally adjacent residues constituting the site Biochim Biophys Acta 1976 427 745 751 57805
Lee CL Pai RC Atassi MZ Enzymic and immunochemical properties of lysozyme--XV. Delineation of the reactive site around the two central disulfides by immunochemical studies of novel synthetic peptides that contain diglycyl bridges instead of disulfides Immunochemistry 1976 13 681 687 965036 10.1016/0019-2791(76)90209-3
Michalsky E Goede A Preissner R Loops In Proteins (LIP)--a comprehensive loop database for homology modelling Protein Eng 2003 16 979 985 14983078 10.1093/protein/gzg119
Kabsch W Sander C Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features Biopolymers 1983 22 2577 2637 6667333 10.1002/bip.360221211
Wenschuh H Volkmer-Engert R Schmidt M Schulz M Schneider-Mergener J Reineke U Coherent membrane supports for parallel microsynthesis and screening of bioactive peptides Biopolymers 2000 55 188 206 11074414 10.1002/1097-0282(2000)55:3<188::AID-BIP20>3.0.CO;2-T
Reineke U Volkmer-Engert R Schneider-Mergener J Applications of peptide arrays prepared by the SPOT-technology Current Opinion in Biotechnology 2001 12 59 64 11167074 10.1016/S0958-1669(00)00178-6
Atassi MZ The precise and entire antigenic structure of lysozyme: implications of surface-simulation synthesis and the molecular features of protein antigenic sites Adv Exp Med Biol 1978 98 41 99 82389
Casset F Roux F Mouchet P Bes C Chardes T Granier C Mani JC Pugniere M Laune D Pau B Kaczorek M Lahana R Rees A A peptide mimetic of an anti-CD4 monoclonal antibody by rational design Biochem Biophys Res Commun 2003 307 198 205 12850000 10.1016/S0006-291X(03)01131-8
Franke R Doll C Wray V Eichler J Solid-phase synthesis of structurally diverse scaffolded peptides for the mimicry of discontinuous protein binding sites Protein Pept Lett 2003 10 531 539 14683504 10.2174/0929866033478519
Eichler J Rational and random strategies for the mimicry of discontinuous protein binding sites Protein Pept Lett 2004 11 281 290 15327360 10.2174/0929866043406931
Frank R The SPOT-synthesis technique. Synthetic peptide arrays on membrane supports--principles and applications J Immunol Methods 2002 267 13 26 12135797 10.1016/S0022-1759(02)00137-0
Ferrè F Ausiello G Zanzoni A Helmer-Citterich M SURFACE: a database of protein surface regions for functional annotation Nucleic Acids Res 2004 32 Database issue D240 4 14681403 10.1093/nar/gkh054
Odorico M Pellequer JL BEPITOPE: predicting the location of continuous epitopes and patterns in proteins J Mol Recognit 2003 16 20 22 12557235 10.1002/jmr.602
|
16153304
|
PMC1242346
|
CC BY
|
2021-01-04 16:27:27
|
no
|
BMC Bioinformatics. 2005 Sep 9; 6:223
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-223
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2281617152110.1186/1471-2105-6-228SoftwareModeling the emergence of multi-protein dynamic structures by principles of self-organization through the use of 3DSpi, a multi-agent-based software Soula Hédi [email protected] Céline [email protected] François [email protected] Sébastien [email protected] Guillaume [email protected] Olivier [email protected] Laboratoire de Productique et d'Informatique des Systèmes Manufacturiers, Institut National des Sciences Appliquées de Lyon, Villeurbanne, France2 Centre de Génétique Moléculaire et Cellulaire CNRS UMR 5534; Université Claude Bernard Lyon 1, Villeurbanne, France2005 19 9 2005 6 228 228 24 5 2005 19 9 2005 Copyright © 2005 Soula et al; licensee BioMed Central Ltd.2005Soula et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
There is an increasing need for computer-generated models that can be used for explaining the emergence and predicting the behavior of multi-protein dynamic structures in cells. Multi-agent systems (MAS) have been proposed as good candidates to achieve this goal.
Results
We have created 3DSpi, a multi-agent based software that we used to explore the generation of multi-protein dynamic structures. Being based on a very restricted set of parameters, it is perfectly suited for exploring the minimal set of rules needed to generate large multi-protein structures. It can therefore be used to test the hypothesis that such structures are formed and maintained by principles of self-organization. We observed that multi-protein structures emerge and that the system behavior is very robust, in terms of the number and size of the structures generated. Furthermore, the generated structures very closely mimic spatial organization of real life multi-protein structures.
Conclusion
The behavior of 3DSpi confirms the considerable potential of MAS for modeling subcellular structures. It demonstrates that robust multi-protein structures can emerge using a restricted set of parameters and allows the exploration of the dynamics of such structures. A number of easy-to-implement modifications should make 3DSpi the virtual simulator of choice for scientists wishing to explore how topology interacts with time, to regulate the function of interacting proteins in living cells.
==== Body
Background
The possibility of probing the biophysical properties of fluorescent proteins in intact living cells by using confocal microscopy has led to a major step forward in contemporary biology. This technical advance has allowed biologists to obtain a new perception of different biological structures. Most of those structures were shown to be highly dynamic, to an extent that was previously unanticipated. This was shown to be especially important in studies of the nuclear architecture [1]. One important lesson from these studies is that, although various nuclear structures (including speckles, the nucleolus and various nuclear bodies) appear to be stable, their components are permanently engaged in an extraordinarily dynamic process: proteins are exchanged between nuclear structures and the nucleoplasm at a rate that makes the stability of the structures really astonishing. It has therefore been proposed that multi-protein dynamic structures are formed and maintained by principles of self-organization [1]. However, this provocative and speculative model raises the question of the stability of the nuclear structures: how do such structures reconcile the extensive material exchange with their environment and the global stability that we observe at a macroscopic level? Computer-based simulations can help to answer this question. If the self-organization hypothesis is true, then one should be able to virtually reconstruct computer-based model structures using a very restricted set of simple local interaction rules.
Most of the existing "virtual cell" models use an averaging behavior hypothesis. In this case, the overall phenomenon is a consequence of the mean behavior of an "average protein" inferred from those of a large number of single proteins. This assumption is inadequate in one or both of the following cases:
1. First when the number of molecules is too low to be correctly approximated by an average behavior [2]. For example, this is the case for transcriptional events that are increasingly being recognized as intrinsically stochastic events, mostly because the number of transcription factors is low [3].
2. Second, when the structures modeled have a strong spatial component. This will obviously be the case for the nuclear structures described above.
It has been proposed that multi-agent systems (MAS)-based modeling could provide a superior approach in those two contexts [4-9]. The study of MAS focuses on systems in which many "intelligent" agents interact with each other [10]. The agents are considered to be autonomous entities, such as software programs or robots. Their interactions can be either cooperative or selfish. That is, the agents can share a common goal (e.g. an ant colony), or they can pursue their own interests, as in free market models [11]. Most importantly, agents can be purposeless, i.e. they can be endowed with a very limited and simple set of rules. MAS have been successfully applied to various domains, including the popular boids, which mimick the structured motion of a flock of birds [12,13]. Such a use of artificial life to create bottom-up models of the real world follows from the realization from the Conway's life game [14], that simple rules can generate complex patterns.
We therefore decided to explore the MAS potential for modeling simple nuclear structures such as nuclear bodies, or speckles [1]. We demonstrate here that 3DSpi, a MAS-based software, is capable of explaining the emergence and predicting the behavior of cellular multi-protein dynamic structures.
Implementation
Program
3DSpi is built upon two libraries. The first one is called the Open Dynamic Engine (OpenDE or ODE) and is coding for the dynamic interactions. The second is the multi-agent framework called OpenSPEAR (Open Simulation of Physical Environment for Agent Research) and is has been developed specifically by us.
Particle model
The program generates an environment which rules all the interactions between agents. Agents can be either single proteins or groups of proteins clustered together. These agents are considered here as solid 3-dimensional material with dynamic, collision and kinetic rules. All agents are animated by a random motion: at each time step (which corresponds to 1 ms) each agent is endowed with a random force and torque (rotation). The movement is then solved as well as collision (if any) for all agents sequentially.
At the beginning of a simulation, proteins are seeded at the intersection points of a grid (protein types are randomly chosen according to predefined proportions). In the experiments discussed here, this grid is a 6 × 6 × 6 grid and 3DSPi is therefore seeded with 216 particles (one at each intersection point). However, the size of the grid can be modified in order to seed the program with less or more proteins depending on the simulation purpose. Please note that this grid is used only for random seeding purposes and is not used again once the simulation has started.
During the simulations, agents are endowed with a random movement that they will pursue until they hit an obstacle, either the inner face of the nuclear membrane or another protein. In the case where they hit the inner face of the nuclear membrane, their speed is reduced to zero (this is called a soft shock). Since the protein is still animated with a brownian motion, it does not stay indefinitely close to the membrane. In the case where they hit another protein OpenDE is used to compute the resulting movements. Moreover, when a collision occurs, each of the involved agents computes whether it will stick and then at each of the following time steps whether it will stay stuck or not (see Coefficient of Stickiness in the "Parameters" paragraph).
In order to evaluate the dynamic structures that emerge, we needed to introduce a slightly different algorithm. Once two proteins are stuck, we create a meta-structure composed of both proteins. This is done recursively until all proteins composing one structure are within this meta-structure. We would like to stress that the goal of this meta-structure is strictly to compute the number of structures. It has no influence on the behavior of the system whatsoever.
Parameters
In order to compute the brownian motion, we converted all forces parameters into arbitrary units relative to the temporal discretization (the higher the forces, the lower the time discretization). We used a set of parameters that combined both smooth simulation and computation efficiency. These parameters describe the physical environment in which the experiments are conducted. Although they can be modified by the users, they will remain constant for all the experiments. This results in the following parameters: strength max: 500; torque max: 500.
The sizes of the elements are also relative. They are chosen in order to fill sufficiently the cell nucleus while allowing free protein movement. This results in the following parameters in all the simulation shown: cell radius: 50; first protein radius: 1; second protein radius: 3. Note that the arbitrary units used for the forces are related to the size parameters (the movement of the proteins and of the multi-protein structures are related to the force and torque on the basis of their mass and kinetic momentum, i.e. their volume and shape).
Once the movement and collision are set and resolved, each protein in contact with another one will check its sticky position. That is each protein in contact will check the random value for the COS which allows to decide if the pair of proteins considered will stick or not, or remain stuck or not. Thus the sticking period for one given protein follows a geometric law of parameter COS: COSt (1-COS), that gives the probability to stay stuck for t time points.
FLIP-like experiments
Given the random movement of proteins, the "bleaching zone" is modeled through a probability value that a given single protein becomes bleached. This probability was set to 0.01 in the experiments shown. The same probability was applied to all proteins whatever their type.
3DSpi was first run for 20000 time steps using the indicated parameters (see legend to figure 5) in order to reach a "stable state". Then the bleaching was started by applying the 0.01 probability value to become bleached to individual proteins not engaged in any interaction (i.e not stuck to anyone). This is equivalent, since particles move at random, to decide that the bleaching zone would occupy 1% of the overall nuclear surface. Once bleached, the protein remains in this state for the rest of the simulation. It can nevertheless still be engaged in interactions with other proteins, on the very same basis as non-bleached proteins.
Figure 5 A virtual FLIP experiment. In A, the principle of the biological FLIP experiment is shown. One given, labeled protein, that participates in one given biological structure (shown in green, since it is fluorescent at the beginning of the experiment) is bleached whenever it passes through a given region of the nucleus (Bleaching Zone, a region of the nucleus where the laser is turned to the bleaching mode). The overall fluorescence of the structure it then studied as a function of time. If the protein moves freely out of the structure and into the cytoplasm, then it will at some point passes through the bleaching zone, be bleached, and by random movement be incorporated again in the structure,. The overall fluorescence of the structure will therefore decrease with time. B and C: Result of two individual in silico FLIP experiments. The following parameters were used: Protein number: 216; Number ratio: 0.5; Size ratio: 3 and COS = 0.9999 for the experiment shown in B and COS = 0.99999 for the experiment shown in C. The bleaching probability value (see Implementation section) was set to 0.01. Those results were confirmed by 10 independent simulations that gave a very narrow range of output (not shown so that the behavior of one individual simulation can be easily seen).
In all experiments we recorded the number of structures through time. In the FLIP-like experiment we also recorded the number of bleached proteins.
Results
Modeling methodology
3DSpi (3-dimensional Dynamic Simulator of Protein-Protein Interactions) is a multi-agent simulation software that has been developed to model the global structures that can emerge from sets of interacting proteins. Such an approach relies on a strict modeling methodology: the aim of the model is to observe, at a global level, structures that are not explicitly programmed in it. For this, we introduce elementary entities (here the proteins) whose behavior only depends on local interactions. This approach means that the multi-protein structures (here the nuclear bodies) are not explicitly introduced in the model. Thus, if they are observed, we can argue that the self-organizing hypothesis is sufficient to explain their emergence. Moreover, the simulation can also help to characterize the qualitative behavior of the multi-protein structures, thus giving important information to predict the behavior of the original in vivo nuclear structures. It is important to note that our aim is not to use 3DSpi to model detailed protein folding or structure. Therefore, our software is fundamentally different from classical folding software which model the precise structure of a small number of proteins. On the contrary, in 3DSpi, the protein model is very simple (proteins are isotropic spheres, see below), but the purpose of 3DSpi is to predict the spatial structures of assemblies of a very large number of molecules.
Basic functions of 3Dspi
3DSpi enable us to compute the interactions of a large amount of autonomous agents (i.e. there is neither centralized decision process nor high level compartments that exchange materials). All the agents are 3-dimensional solid particles moving in a 3-dimensional space that are able to interact with each other locally. In the simulation proposed here, it is used to simulate two different types of proteins. Each of them is modeled by an "interaction volume" which is considered as an homogeneous, isotropic, sphere. The two protein families differ by their size. Thus they move differently – the larger proteins move slower – and they fill the nucleus space differently. Therefore, the relative proportion of the two protein families may influence the system behavior.
The program requires four values to be defined as input for describing the biological system (plus fixed parameters describing the physical world, see Implementation section):
1. The total number of proteins occupying the "nucleus" space;
2. The ratio in number between the two sorts of proteins. If this value is equal to 0.5, then the same number of each protein species is used.
3. The respective size of the diameter of the two proteins types. If that value is set to 1, then this will results in simulating two proteins with the same size; if that value is set to 3, then this will results in a type of protein 3 times larger in diameter than the other.
4. The Coefficient Of Stickiness (COS), which is the probability, at each time point, that a protein attached to another protein will stay attached to it during the next time step (see Implementation section). The COS can be seen as a very macroscopic consequence of the folding properties of the proteins that sums the affinity of one protein species to another.
The program starts with a random distribution of the proteins. Then every protein is induced to move at random, until it hits another one. The probability that colliding proteins will bind to each other is then calculated as a function of the user-defined COS.
The program runs for a fixed number of one million time steps. Each run can be followed in real-time through a graphic interface that displays a 3-dimensional view of the system (filmed sequences can be obtained by contacting the authors; Figure 4A shows a screen shot). Various numerical values can be recorded during the run including the number of protein structures and their size. In 3DSpi, a "structure" is an observed structure of any size, from one protein to any number of proteins that are bound together. We did not define a structure as being more than one protein since multi-protein structures are not explicitly (i.e. a priori) introduced in the model. Both visual and numerical outputs are computed in real time. On one hand the videos enable the biologist to understand the behavior of the system and to propose hypotheses. On the other hand numerical data are mandatory to statistically validate these hypotheses and to analyze the self-organization behavior of the system precisely (figure 1).
Figure 1 Schematic view of 3DSpi. Starting from local protein parameters (COS, protein size, ...), 3DSpi computes the proteins interactions and provides two different outputs: Video output and numerical data.
Figure 4 Images observed using 3DSpi. A: Screen shot of 3DSpi. The following parameters were used: Protein number: 3375; Number ratio: 0.5; Size ratio: 2; COS: 0.99999. The final state of the system is shown. B: The original image was treated with Adobe® Photoshop®, in order to change the colors. C: A real life picture of speckles (reprinted with permission from SCIENCE [1]).
Generation of structures as a function of COS
We examined how the structures evolved as a function of the COS value (Figure 2A and 2B). The behavior of the system can be predicted easily for two extreme values. With a COS value of 0 no protein binds to any other (i.e. all the "structures" are made of only one protein). Thus the observed number of structures equals the number of seeded proteins (see Implementation section for details). In the other hand, if the COS has a value of 1, all of the proteins will ultimately be bound to each other, resulting in only one very large "structure". These extreme cases are correctly modeled by our system (see Figure 2A).
Figure 2 Number of structures generated using 3DSpi. A and B: Number of structures generated as a function of the COS value. The following parameters were used: Protein number: 216; Number ratio: 0.5; Size ratio: 3. The program was run for 106 time points. In B an enlargement of the right part of the figure in A is shown, on a semi-logarithmic scale. The mean observed during the last 50000 time points in one simulation is shown. The bar indicates the minimum and maximum value observed. C: Number of structures generated as a function of time using three different COS values (see the right part of the picture). The mean observed on 20 independent simulations is shown, and the bar indicates the minimum and maximum value observed in all of those simulations. Protein number: 216; Number ratio: 0.5; Size ratio: 3.
There is a large interval of COS values (between 0 and 0.9) for which nothing happens. In that interval, although proteins collide and bind to each other, these interactions are too transient to generate any large stable structure. However for values comprised between 0.9 and 1.0, an exponentially increasing tendency to form structures is observed (Figure 2B). In this very narrow range the number of structures is a direct function of the COS. We verified that a very similar behavior of the system was observed for two other protein size ratio (size ratio of 1 and 4, not shown) and therefore that this phase transition was a robust behavior of our system.
Moreover, as far as the number of structures is concerned, the behavior of the system is highly reproducible: the extreme values (minimum and maximum number of structures for the different runs) observed during the last 20000 time steps (i.e. after the transient period) are very close to the mean. This suggests that the system is highly robust and generates a predictable dependency on the COS. In order to assess this statement, we ran 20 different simulations, for 3 different COS values, and we recorded the number of structures at each time step (Figure 2C). Two things were readily apparent:
1. The system reaches its equilibrium very quickly. Indeed in the worst case there is no significant change after 2.105 time steps.
2. The number of observed multi-protein structures is a function of the COS value. This means that the number of structures is independent of a particular stochastic run and therefore an invariant of the topology and of local behavior parameters (COS).
We next analyzed the size of the multi-protein structures formed. In order to do this, we plotted the repartition of the proteins according to the size of the structure they belong to, for various values of COS (Figure 3). As expected the majority of proteins move from small structures to large ones when the COS increases. Interestingly, at high COS values, a dynamic equilibrium occurs between two groups of structures, small ones and large ones, without structures of intermediate size (see Figure 3, COS = 0.999999). This is characteristic of a phase transition in which small local differences can have a large impact on the global behavior. This indicates that one observes both the emergence of complex structures and the existence of complex interactions between these structures.
Figure 3 Size of the structures generated using 3DSpi. The percentage of proteins belonging to the various size structures is shown, ranging from 1 to 216, as a function of a COS value ranging from 0.999900 to 0.999999. The program was run for 106 time points. The mean observed for the last 10000 time points on 40 different simulations is shown.
Altogether our data demonstrates that as the COS value increases, the system shifts from a state characterized by numerous small structures toward a system mainly composed of a small number of large structures.
Ability of 3DSpi to mimic biological structures
3DSpi was initially intended to simulate nuclear bodies. Its ability to generate body-like structures was a necessary step toward its validation. We reasoned that in order to generate realistic data, one should approximate the real life observation conditions. For this, we initiated a 3DSpi run with a very large number of agents (of each sort). This resulted in a screen shot of a 3DSpi simulation (Figure 4A). Using an image processing software we modified the initial colors. This generated an image (Figure 4B) that is very strikingly similar to a real-life image of nuclear structures called speckles (Figure 4C). Speckles or splicing factors compartments are known to be dynamic structures, and both their protein and RNA-protein components can cycle continuously between speckles and other nuclear locations [15]. At a very macroscopic level, 3DSpi can therefore generate images of dynamic structures closely resembling those observed using confocal microscopy.
Evidence that the structures generated by 3DSpi are dynamic
The most convincing way to demonstrate that proteins are continuously exchanged between a given nuclear structure and the nucleoplasm is called Fluorescence Loss Induced by Photobleaching (FLIP; [16]). In these experiments, a protein that participates in an observable structure is fluorescently labeled. The resulting fluorescent structure is visualized while a beam set to bleaching mode is used to photobleach a portion of the nucleoplasm (Figure 5A). Photobleaching consists in switching off the fluorescence associated with the protein without destroying the protein itself: the protein is still there and active but it cannot be detected by the confocal microscope anymore.
In the case of the transcriptional complex formed by the glucocorticoid receptor (GR), the FLIP approach demonstrated a rapid decrease in the fluorescence of the GR-containing complexes bound to DNA [16]. This demonstrated that the GR transcriptional complex was continuously exchanging individual GR molecules at a high rate with the nucleoplasm.
We decided to apply this FLIP approach to 3DSpi-generated multi-protein structures. For this, we decided to "bleach" individual proteins not engaged in an interaction, by applying a probabilistic bleaching value for the isolated proteins. It is obvious that this will underestimate the real bleaching since for example proteins dimers can pass through the bleaching zone in a real experiment, but not using our bleaching strategy. We nevertheless feel this is sufficient for probing the extent to which large structures are composed of particles that are continuously exchanged with the nucleoplasm.
We therefore followed the number of structures as well as the number of bleached proteins (Figure 5B and 5C) for two COS values (below and above the phase transition values). It was immediately apparent that for both simulations, the number of bleached proteins increased steadily during the course of the experiment while the number of structures remains stable. This thereby demonstrates that proteins are indeed continuously exchanged to and from the structures and thus confirms the dynamic nature of the multi-protein structures that could intuitively be deduced from the visual system's observation.
The dynamic properties shown by this experiment is a very fundamental result since it shows that, though at a macroscopic level the system behavior is different for different COS values, at a microscopic scale the protein behavior is similar. However, the dynamic of the exchange was clearly influenced by the value of COS. When the lowest COS value was investigated, at the end of the simulation, virtually all of the proteins were bleached, whereas with a larger COS value only about 20% of the proteins have been bleached at the end of the simulation period. In the case where a high COS value is used, it might seems surprising that even deeply buried proteins can be bleached. We explored this issue using the video output of 3DSpi. We observed large structures happen to "break in two" thereby exposing their inner core and exposing the proteins from the inside of the structure. The biological relevance of such a phenomenon needs to be assessed.
Conclusion
We have developed a program called 3DSpi that simulates the behavior of 3-dimensional solid particles, moving at random in a 3-dimensional space, colliding, and binding to each other as a function of a probabilistic value called Coefficient Of Stickiness (COS). Dynamic multi-particles structures appear only for a narrow range of COS values. Within that range, the behavior of 3DSpi, although intrinsically stochastic, and therefore noisy, was shown to be very robust, as assessed by the predictable number of emerging structures, and by the short period of time required to reach a dynamic equilibrium. Moreover, a phase transition occurs to give two distinct distributions of structures. This non-linear feature of the system generated by 3DSpi can be used to model several of the non-linear phenomena found in biological systems. In addition the structures generated by 3DSpi are very realistic as they appear to resemble real life structures. Furthermore, the use of an in silico FLIP-like technique confirmed that the behavior of our model was compatible with the existing data regarding the dynamic nature of cellular substructures [1].
One other published study using MAS to model molecular structures has been conducted in a virtual 2-dimensional space [4]. In our preliminary experiments, we found that 2-dimensional versions of our software were inefficient for generating biologically relevant structures (data not shown). Since biological phenomena occur in a 3-dimensional space, it is therefore essential that studies conducted on spatial structures are performed in a realistic 3-dimensional space. Furthermore, a different modeling strategy was used in [4], in which high-level scenario were explicitly introduced. A more recent 3-dimensional version of a program called HSIM has been proposed [9], that uses rules for encoding the relations between the proteins. Unfortunately, a quantitative analysis of HSIM behavior has not been published, and its availability not publicized, therefore precluding a direct comparison with our 3DSpi model.
In our study, we used a modeling methodology that was strictly designed so as not to encode explicitly the structures we wished to study. As such it generates a surprising consequence (at least in contrast with more classical modeling tools): since the structures are not explicitly programmed-in, the 3DSpi software cannot itself provide their global parameters. In other words, it can not describe the parameters of structures that "don't' exist" at its own modeling level. Therefore we developed an appropriate independent observation tool, just as biologists do while observing real structures.
One of the indirect results of our work is to ask a fundamental – but rarely evoked – question: what is a structure? Multi-Agents System models and 3DSpi can help answering such a question. Indeed, in our system structures are no longer considered as fixed bodies whose components are clearly identified. They are the product of collective protein behaviors that are self-regulated, i.e. do not depend upon a "master" pattern. This self-regulation is the consequence of two opposite (and competing) trends inside the system. On the one hand, the entropy tends to increase disorder and create free proteins. On the other hand the stickiness moves the system toward a more ordered state and ultimately toward a unique and extremely stable large structure. The zone of interest lies around where these two opposite forces attain a dynamic equilibrium – a zone we termed as phase transition. This is in the latter that we found most of the non trivial structures, like a split occurring between two groups of structures, small ones and large ones, without structures of intermediate size.
Our initial aim was to test the possibility that very simple local rules are sufficient, when expressed in a physically relevant model, to generate sophisticated large structures. In particular, we wanted to determine "whether nuclear organization can be reproduced in silico assuming the constraints of self-organizing systems" [1]. We have demonstrated that this does occur. Our modeling results strongly favor the hypothesis that the appearance of large multi-molecular structures does not require sophisticated scaffolding. A transient non-specific protein-protein interaction is sufficient for generating large multi-molecular complexes, provided that affinity or stickiness of protein species lies within the proper range. Of course, the COS value cannot be seen as identical to an affinity or avidity value. It nevertheless remains quite conceivable that a COS-like interaction value can be evaluated using a composite function of biochemically-determined binding parameters. The mathematical approaches to FRAP modeling [17,18] should be helpful in estimating such a parameter from real life observations.
One should stress that we have demonstrated the possibility that such structure emerge by self-organisation, but of course this does not demonstrate that this is the case for real life structures. It is our belief that 3DSpi might be helpful in making predictions about how such a self-organizing system might behave that ultimately could be tested experimentally in living systems. However, predictions could only be made in a more biologically sound version of 3DSpi since the present status of 3DSpi model suffers from a number of limitations. Those limitations include the low number of proteins, the limited number of different proteins species and the uniform interaction pattern (i.e. all interactions are the same, irrespective of the proteins species).
In order to increase the biological relevance of our simulations, proteins should not be considered as homogeneous isotropic spheres. For this we are currently developing an XML-based tool that will allow to construct sophisticated proteins containing multiples domains, with each domain having its own set of interaction rules. To some extent one might envision to derive from real life 3-dimensional protein structures a simplified, but realistic, spatial version that can be modeled using 3DSpi. Using a larger amount of proteins, more different protein species and more sophisticated 3-dimensional versions of proteins, will inevitably require much longer calculation time. This problem might be solved through parallel computation on a computer cluster, a project in progress in our laboratories.
Furthermore, it would be very interesting to simulate protein synthesis and degradation rates, in order to assess the impact in variations of those parameters upon the global behavior of the system.
Another improvement may consist of adding consequences to the binding. For example, one protein could be "activated" (i.e. would now be able to interact with another one) only after having been bound to a third partner. This should result in modeling a signaling pathway, based on probabilistic interactions and random displacement. It would be very interesting to analyze the ability of such a pathway to carry a signal from the outside of a "cell" into its nucleus, as simulated by two encased spheres.
Finally, it is our belief that, as exemplified in the present work for spatially constrained structures. SMA-based modeling will proved to be a precious tool for tomorrow's biologists. Although this early version is restricted in its purposes, we are confident that later versions of 3DSpi will allow to model any type of biological structures, whether modules [19], or hyperstructures [7].
Availability and requirements
OpenDE uses a GPL license and is available at . The OpenSPEAR library is also GPL and is available at . It is distributed freely with no online help or warranty. The 3DSpi program itself (3dspi_code.tar.gz) is also distributed under GPL license. It is available at . The Windows beta-testing implementations of OpenSPEAR and 3DSpi are available at . Linux OS versions of the 3DSpi libraries are available on request by mail to the authors. All further information (including software update) will be available through the following website: . A Windows stand-alone version of 3Dspi (3dspiInstallerv1.0.exe) is also available on the internet
Authors' contributions
H. S., F.P. and S. G. wrote the software. C.R. made most of the experiments described. G.B. and O.G. conceived the study, and participated in its design and coordination. O.G. wrote the initial draft. H.S., C.R., O.G. and G.B. participated in writing the final version of the manuscript. All authors read and approved the final manuscript.
Acknowledgements
We are very much indebted to Céline Becquet, Céline Charavay, Audrey Herr and Tiphaine Martin (INSA students) for enthusiastically generating an early 2-dimensional version (2DSpi) of this software, thereby establishing the proof of concept. We thank François Morlé (CGMC UMR 5534) and Aldo Deandrea (Institut de Physique Nucléaire, UCBL) for helpful discussions during the early steps of that project, and François Morlé and Edmund Derrington (CGMC UMR 5534) for critical reading of the manuscript. We thank all members of the BSMC group for very stimulating discussions. This project was supported by the Région Rhône-Alpes (programme Emergence) and by French funding agencies (ACI IMPBio, projet MOCEME).
==== Refs
Mistelli T Concepts in nuclear architecture BioEssays 2005 27 477 487 15832379 10.1002/bies.20226
Loew LM Schaff JC The Virtual Cell: a software environment for computational cell biology Trends Biotechnol 2001 19 401 406 11587765 10.1016/S0167-7799(01)01740-1
Levsky JM Singer RH Gene expression and the myth of the average cell Trends Cell Biol 2003 13 4 6 12480334 10.1016/S0962-8924(02)00002-8
Ballet P Zemirline A Marcé L The BioDyn language and simulator. Application to an immune response and E. coli and phage interaction J Biol Phys Chem 2004 4 93 101
Becquet C Charavay C Herr A Martin T Morle F Deandrea A Beslon G Gandrillon O Modeling protein-protein interactions by using Multi-Agents Systems JPGD '02: 2002 2002 UCBL Villeurbanne, France
Klein J breve: a 3D simulation environment for the simulation of decentralized systems and artificial life Proceedings of Artificial Life VIII, the 8th International Conference on the Simulation and Synthesis of Living Systems: 2002; Sydney, NSW, Australia 2002 The MIT Press
Amar P Bernot G Bouligand Y Bourguine P Delaplace F Delosme JM Demarty M Fishov I Fourmentin-Guilbert J Fralick J Hyperstructures, genome analysis and I-cells Acta Biotheor 2002 50 357 373 12675536 10.1023/A:1022629004589
Le Sceller L Ripoll C Demarty M Cabin-Flamand A Nystrom T Saier M Norris V Modelling bacterial hyperstructures with cellular automata InterJournal 2000 366
Amar P Bernot G Norris V HSIM: a simulation programme to study large assemblies of proteins J Biol Physics Chem 2004 4 79 84
Weiss G Multiagent Systems. A Modern Approach to Distributed Artificial Intelligence 1999 Cambridge, Massachusetts: The MIT Press
Reynolds CW Flocks, Herds, and Schools: A Distributed Behavioral Model, in Computer Graphics SIGGRAPH '87 Conference Proceedings: 1987 1987 25 34
Berlekamp ER Conway JH Guy RK Winning your ways for your mathematical plays 1982 Academic Press
Lamond AI Spector DL Nuclear speckles: a model for nuclear organelles Nat Rev Mol Cell Biol 2003 4 605 612 12923522 10.1038/nrm1172
McNally JG Muller WG Walker D Wolford R Hager GL The glucocorticoid receptor: rapid exchange with regulatory sites in living cells Science 2000 287 1262 1265 10678832 10.1126/science.287.5456.1262
Carrero G McDonald D Crawford E de Vries G Hendzel MJ Using FRAP and mathematical modeling to determine the in vivo kinetics of nuclear proteins Methods 2003 29 14 28 12543068 10.1016/S1046-2023(02)00288-8
Braeckmans K Peeters L Sanders NN De Smedt SC Demeester J Three-Dimensional Fluorescence Recovery after Photobleaching with the Confocal Scanning Laser Microscope Biophysical J 2003 85 2240 2252
Hartwell LH Hopfield JJ Leibler S Murray AW From molecular to modular cell biology Nature 1999 402 C47 52 10591225 10.1038/35011540
|
16171521
|
PMC1242347
|
CC BY
|
2021-01-04 16:27:26
|
no
|
BMC Bioinformatics. 2005 Sep 19; 6:228
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-228
|
oa_comm
|
==== Front
BMC Med GenetBMC Medical Genetics1471-2350BioMed Central London 1471-2350-6-331617430110.1186/1471-2350-6-33Research ArticleCell cycle and centromere FISH studies in premature centromere division Corona-Rivera Alfredo [email protected] Fabio [email protected] Lucina [email protected] Jorge R [email protected] Cesar [email protected] Teresa A [email protected] Enrique [email protected] Laboratorio de Citogenética Genotoxicidad y Biomonitoreo, Instituto de Genética Humana Dr. Enrique Corona Rivera, Departamento de Fisiología, División de Disciplinas Básicas, Centro Universitario Ciencias de la Salud, Universidad de Guadalajara, Guadalajara, Jalisco, México2 Unidad de Citogenética, OPD Hospital Civil Fray Antonio Alcalde, Guadalajara, Jalisco, México3 Unit of Investigation in Human Genetics, National Medical Center, Instituto Mexicano del Seguro Social, México City, México4 División de Pediatría, OPD Hospital Civil Juan I. Menchaca, Guadalajara, Jalisco, México5 Laboratorio de Genética Humana, Departamento de Fisiología, División de Disciplinas Básicas, Centro Universitario Ciencias de la Salud, Universidad de Guadalajara, Guadalajara, Jalisco, México2005 20 9 2005 6 33 33 9 3 2005 20 9 2005 Copyright © 2005 Corona-Rivera et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Mitotic configurations consistent in split centromeres and splayed chromatids in all or most of the chromosomes or premature centromere division (PCD) have been described in three categories. (1) Low frequency of PCD observed in colchicines-treated lymphocyte cultures from normal individuals. (2) High frequency of PCD with mosaic variegated aneuploidy. (3) High frequency of PCD as a sole chromosome abnormality observed in individuals with no recognizable clinical pattern. We report four members of a family with the third category of PCD.
Methods
Cell cycle duration assessed by average generation time using differential sister chromatid stain analysis and FISH studies of DNA centromere sequences in PCD individuals, are included and compared with previously reported PCD individuals from 9 families.
Results
We observed PCD in colchicine-treated cultures from the propositus, his father, and two paternal aunts but not in his mother and four other paternal and maternal family members, as well as in untreated cultures from the propositus and his father. We observed cytological evidence of active centromeres by Cd stain. Significative cell cycle time reduction in anaphases of PCD individuals (average generation time of 21.8 h;SD 0.4) with respect to individuals without PCD (average generation time of 31.8 h;SD 3.9) was observed (P < 0.005, Student t-test for independent samples). Increased cell proliferation kinetics was observed in anaphasic cells of individuals with PCD, by differential sister chromatid stain analysis. FISH studies revealed the presence of alpha satellite DNA from chromosomes 1, 13, 21/18, X, all centromeres, and CENP-B box sequences in metaphasic and anaphasic cells from PCD individuals.
Conclusion
This report examines evidences of a functional relationship between PCD and cell cycle impairment. It seems that essential centromere integrity is present in these cases.
==== Body
Background
Mitotic configurations consistent in split centromeres and splayed chromatids in all or most of the chromosomes or "premature centromere division" (PCD), have been described in three categories. They are:
(1) Low frequency of PCD (up to 3% of the mitosis) observed in colchicines-treated lymphocyte cultures from normal individuals [1,2].
(2) High frequency of PCD (5% or more) with mosaic aneuploidies involving a variety of chromosomes, called "mosaic variegated aneuploidy", observed in individuals with microcephaly, growth deficiency, severe mental retardation, and risk of malignancy [3-5].
(3) High frequency of PCD (5% or more) as a sole chromosome abnormality [6-10]. Individuals with this condition have no recognizable clinical pattern or occur in healthy individuals. Association of this PCD trait with abortions and infertility has been reported [6,7,9,10], but other authors suggest this PCD trait to be harmless [1,8,11].
We report a family that corresponds to the third category of PCD, in which four individuals showed PCD as a sole chromosome abnormality. Cell cycle duration, assessed by average generation time, and FISH studies of the centromere, are considered. Findings in the family are compared with those on 30 other previously reported individuals with PCD from 9 families. This report examines the evidences of a relationship between PCD and cell cycle impairment of cells bearing main structural centromere components from PCD individuals.
Methods
Family data
A total of nine individuals were available to be studied, five of them from the paternal lineage, four of them from the maternal lineage and the propositus. Infertility was observed in two paternal aunts. Family data are summarized on Fig. 1. The non-consanguineous parents were both aged 27 years at propositus birth. The propositus was born at 36 weeks by abdominal delivery indicated by maternal pre-eclamptic toxemia and showing neonatal hypoxia. Birth weight was 2100 g and height was 46 cm. Early signs included: hypotonia, strabismus, seizures, recurrent respiratory infections and occasional breath-holding spells. Physical examination at 5 years of age showed microcephaly (OFC -4.1 SD), low weight (-4.1 SD) and borderline length (10th percentile); plagiocephaly and brachycephaly, nystagmus, phimosis, right cryptorchidia, adducted thumbs, syndactyly 2–3 of toes, four limbs spasticity and increased deep tendon reflexes. On X-rays, hyperbrachycephaly (cephalic index 91.7), scoliosis, spina bifida occulta (T1–T5), coxa valga, slim bones and osteopenia were found. Cortical and subcortical atrophy was found on cranial CT scan. Brainstem auditory evoked potentials were normal. Psychometric test demonstrated a development age of 3–6 months. The propositus died at 10 years of age by pneumonia. Propositus relatives were phenotypically healthy.
Cytogenetic studies
We performed GTG banded karyotypes from peripheral blood lymphocyte cultures stimulated by phytohemaglutinin and treated one hour with colchicine (SIGMA, 0.104 μg/ml), in the propositus and eight family members indicated in Fig. 1. On harvesting, 15 minutes of hypotonic treatment (0.075 M KCl at 37°C) was performed. Chromosome aberrations in the karyotypes of these individuals were not observed. PCD frequencies in repeated cultures of available individuals were obtained in at least three repeated cultures. Around 100 mitosis were scored per culture giving a total range of 258 to 1142 cells scored per individual (Table 1). To obtain basal frequencies of PCD, ten healthy young individuals were used as controls. Additional simultaneous cultures without mitostatic treatment were performed in the propositus and his parents. Cd staining was performed in the propositus and his parents according to Denton et al. (1977) [12].
Cell cycle studies
Cellular proliferation kinetics was used to determine the cell cycle duration, which is the interval between one mitosis and the subsequent [13]. The method called average generation time (AGT) by differential sister-chromatid stain [14,15], was used to determine cell cycle durations. We compared the results between family members with more than 5% of PCD versus those with less than 3% of anaphase frequencies. The original AGT method [14,15], consider for cell cycle calculations only metaphases, in this family, calculations were also performed considering, prophases, anaphases and total mitotic cells. The procedure was as follows. We obtained AGT's of each individual from accumulated data of 3 simultaneous 72 h cultures with 5 μg/ml of 5'-bromodeoxyuridine (SIGMA), added 24 h after set-up. In each culture, we scored prophases, metaphases and anaphases which had completed either one, two or three cell cycles identified by differential sister chromatid stain pattern. Then, AGT, were calculated per mitotic stage and per individual as follows. (i) Taking into account the percentage of cells at first (M1), second (M2) and third (M3) cell cycle, the replication index (RI) was obtained from this formula: RI = (1X%M1+2X%M2+3%M3)/100. (ii) The RI was used to obtain the AGT following the equation: AGT = time of harvest after exposure to 5'-BrdU/RI. Finally, AGT's were compared in paternal individuals with PCD versus those individuals with low anaphase frequencies from the maternal sibship. Statistical test t-Student for independent samples was used to compare both groups.
FISH studies
We searched for the presence of constitutive structural components of the centromere such as alpha satellite DNA sequences in the propositus and his parents with FISH according to a standard protocol [16], using alpha satellite DNA directly labeled probes to chromosomes 1, 13/21, 18, and X, as well as all centromeres probe. Besides, we searched for the presence of CENP-B box sequence in the propositus using a biotin labeled probe, following the protocol of Matera and Ward (1992) [17].
Results
Cytogenetic studies
PCD frequencies in repeated cultures of available individuals are shown in Table 1. Individuals with PCD of 5% or more corresponded to the propositus, his father and two paternal aunts (II-3, and II-8). Individuals with less of 3% of anaphases were considered as normal and corresponded to mother's propositus, two maternal aunts (II-14, II-15), and two paternal aunts (II-9, II-10). Basal frequencies of PCD, in ten healthy young individuals used as controls, did not differ with respect to PCD individual with less of 3% of anaphases. A PCD image of the propositus is shown in figure 2a. In additional simultaneous cultures without mitostatic treatment PCD was observed in the propositus (5%) and his father (5.5%), but not in his mother. Two centromeric dots in evident primary constrictions were observed by Cd staining in the propositus and his parents (fig. 2b). This finding is considered a cytological evidence of active centromeres [12].
Cell cycle studies
We observed cell cycle duration significatively reduced in anaphasic cells but not in prophasic, metaphasic or total cells (Table 2). Cell cycle reduction was then attributed only to cells in anaphase (PCD cells). Additionally, we observed that in the same culture conditions, 38% of anaphases in the propositus and his father reached the third cell cycle, while only 15% of metaphases did. This was interpreted as increased cell proliferation kinetics of anaphasic cells in individuals with PCD associated to cell cycle shortening in PCD cells. Anaphases from second and third cell cycle are shown in figures 2c and 2d respectively.
FISH studies
We observed in all cases positive fluorescent signals (figure 3a), indicating the presence of all tested sequences. We observed a positive fluorescent signal pattern, indicating the presence of CENP-B box sequence in the propositus (figure 3b).
Discussion
We report four family members with more than 5% of colchicine-anaphase frequencies as a sole chromosome abnormality. Nine families have been reported referring to this trait as PCD [6-10]. Mitosis obtained from colchicine arrested lymphocyte cultures of normal individuals, show rates below 3% [1], 5% [2] or 1% observed in our control individuals. Previous reports of PCD frequencies without mosaic variegated aneuploidy (MVA), ranges 5 to 38% [6-10]. In our family, PCD was observed in the propositus, his father, and two paternal aunts in repeated colchicine-treated cultures in average frequencies of 7 to 22%. It was shown [18], that PCD can be induced through hypotonic increasing time treatment of mitotic cells in peripheral blood lymphocytes of healthy individuals and patients homozygous to PCD trait or PCD and MVA. They found that 0.075 M KCl at 37°C for 20 min, showed 0–2% cells in PCD which fits with our observed frequencies because 15 minutes of hypotonic treatment were used in our peripheral blood cultures. Total mitosis scored in repeated cultures of previous [6-10], and present report did not provide evidences of MVA. Although PCD is considered a rare phenomenon, two studies found in selected population frequencies of 1 of 100 [1] or 1 of 1000 [11].
Only two previous studies tested cell cycle duration in individuals with PCD as a sole chromosome abnormality (Table 3). In both cases they obtained evidences that the cell cycle time can be altered in PCD individuals [6,7]. Rudd et al. (1983) [6], found reduced metaphase duration only in some cultured cells, supposing that such cells could have corresponded to those with PCD. Gabarrón et al. (1986) [7], inferred that only the cells in anaphase or PCD cells, showed accelerated proliferation kinetics and consequently reduced cell cycle duration. We provide evidences to confirm that cell cycle duration is reduced in PCD cells because only in anaphasic cells the cell cycle duration in the family members with PCD was statistically reduced. Additionally, we observed increased cell proliferation kinetics of anaphasic cells in individuals with PCD. These findings are compatible with the co-existence of cellular subpopulations bearing differential proliferation kinetics. Cell cycle reduction, possibly related to premature separation of centromeres and persistence of anaphases, could then be considered a distinctive finding in these cases.
Cell cycle progression requires control mechanisms that could be associated to PCD origin. A basic defect of cell cycle progression or metaphase-anaphase transition in PCD was suggested [19]. Matsuura et al. (2000) [20], demonstrated that cultured fibroblasts from two infants with PCD and mosaic variegated aneuploidy are insensitive to the colcemid-induced mitotic-spindle checkpoint. Mitchel et al. (2001) [21], found that mitotic checkpoint defective MAD2+/- haploinsufficient human colon carcinoma cells showed 20% of precocious anaphases with prematurely separated sister chromatids, compared with 1% in wild-type cells, proposing that PCD is a suitable cytogenetic marker for the identification of mitotic checkpoint defects. In our family PCD was also observed in cultures without colchicine from the propositus and his father as in other PCD reports [6-8] (Table 3). Cultured PCD cells with defective colcemid-induced mitotic-spindle checkpoint reported by Matsuura et al. (2000) [20], were unresponsive to colchicine. Although in our family MVA was not observed as in Matsuura et al. (2000) [20] report, in both cases the mitotic arrest signal was overruled in PCD cells. Hanks et al. (2004) [22], provided the evidence that gene mutations can result in a defective spindle checkpoint in humans. They screened the full coding sequence and intron-exon boundaries of BUB1B, and found truncating and missense mutations inherited from different parents in five of eight families with mosaic variegated aneuploidy providing the first evidence in humans that gene mutations might be responsible for aneuploidy in human cancers. Interestingly, cytogenetic data of families 1, 4 and 5 in Hanks et al. (2004) [22] report showed also PCD, as well as in 6 of those 15 reported cases update [5]. Considering that subjacent genetic cause was demonstrated to MVA, this opens the possibility that BUB1B defects can be involved in PCD origin. Mitotic spindle checkpoint serves as a surveillance mechanism that ensures the faithful transmission of chromosomes from a mother cell to its two daughter cells during mitosis [23], this can be involved also in PCD origin. On the other hand, because at least 6 genes have been involved with such checkpoint [23], the origin of PCD may be heterogeneous and should involve a mechanism that triggers the whole set chromosome segregation at mitotic spindle checkpoint.
Other aspect to be considered in PCD origin is the centromere. A basic defect of centromeric region in PCD was suggested [19]. Two essential DNA sequences of the centromere were evaluated by FISH in this family: alpha satellite and CENPB-Box sequences. Centromere function requires the presence of alpha satellite DNA in all human centromeres [24]. We observed alpha satellite DNA from all centromeres and the centromere of chromosomes 1, 13/21, 18, and X, as well as centromeric 17 bp CENPB-Box sequences in prophasic, metaphasic and anaphasic cells from the propositus and his parents. CENPB-Box sequence interacts with the kinetochore protein CENP-B required for the pairing of sister chromatids as structural support and in the conformation of primary constriction and kinetochore [25]. Cytological evidence support the presence of functional centromeres in PCD cells, by positive Cd stain in the propositus and his parents and in one previous PCD report [8]. Also, primary constrictions are evident in this and previous PCD reports. It seems that essential centromere integrity is present and remains unclear if whether or not is involved in PCD origin. Other mechanisms related to cell cycle regulation and functional components of the centromeric region such as defective centromeric cohesion [26], or kinetochore defective proteins are probable.
The PCD trait as a sole chromosome abnormality occurs in healthy individuals. Some authors suggest this PCD trait to be harmless [1,8,18]. In this report four individuals presented PCD, and three of them were phenotypically normal. Noteworthy, all the 30 PCD individuals from 9 previous families included in Table 3 were also phenotypically normal. Individuals with this category of PCD have no a recognizable clinical pattern [6-10]. In three of such families clinical findings were informed and reported as coincidental [6,8,10]. The abnormal phenotype observed in the propositus shows no concordance to previous PCD without MVA cases. PCD trait observed in healthy paternal relatives and all previous cases, represent the common one dose effect of this autosomal dominant trait (OMIM, *176430) [27]. Such mode of inheritance was concordant with this report because male to male transmission was observed. In autosomal dominant PCD and abnormal phenotype associated to MVA, homozygosity was implicated [28,29]. This statement was confirmed by Plaja et al. (2001) [3] in three patients compared with 8 previous cases exhibiting microcephaly, CNS anomalies, mental retardation, prenatal and postnatal growth retardation and cancer, proposing that in vivo occurrence of random aneuploidies and chromosome or genome instability disorder explained some of the clinical data. It seems that variegated aneuploidy is associated to an abnormal phenotype. Our propositus showed prenatal and postnatal growth retardation, profound developmental delay, hypoplasia of the brain and clonic seizures, coincident with MVA reports [5], but our case did not show MVA nor apparent cancer risk. Alternatively, considering that the inheritance of MVA is recessive [5], and some heterozygotes show levels of PCD without variegated aneuploidy, this can be compatible with those individuals described in present report or in previous reports regarding apparently harmless PCD. However, the relationship between PCD and MVA is uncertain [5]. Also, we considered the possibility that neurogical affectation in the patient studied by us could be associated with neonatal hypoxia. In these cases, genetic heterogeneity may be involved.
Association of this PCD trait with abortions and infertility has been reported [6,7,9,10]. This was observed in 12 of 34 PCD individuals from 8 of 10 previous families including present report (Table 3). We observed infertility in two paternal aunts (II-2 and II-3 in fig. 1); cytogenetical analysis was available only in one of them observing PCD. The estimated abortion frequency in descendents of reported PCD individuals was 37% (22 of 60) which is higher than those observed in general population of 15% [30]. In one report, unexplained recurrent abortion observed in both parents with PCD was considered the consequence of abnormal behavior of the centromeres involving probable homozygous effect [9]. Previous observations are coincidental but remark the occurrence of subfertility in PCD individuals.
Conclusion
Present report represents a new family with PCD as a sole chromosome abnormality. Cell cycle studies revealed that cell cycle reduction could be considered a distinctive finding in these cases. Based in previous reports and the fact that cells of PCD patients were unresponsive to colchicine is probable that a defective colcemid-induced mitotic-spindle checkpoint is involved. Is open the possibility that BUB1B defects or other genes involved in such checkpoint may be involved in PCD origin. In this cases considered DNA centromeric sequences were present. It seems that essential centromere integrity is present and remains unclear if whether or not is involved in PCD origin. Other mechanisms related to cell cycle regulation and functional components of the centromeric region may be involved. Interestingly, the PCD trait as a sole chromosome abnormality occurs in healthy individuals and there is not a characteristic associated abnormal phenotype. Only subfertility seems to be a common finding in these families. Those families deserve further investigation in order to understand possible mechanism of this mitotic trait.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
AR conceived of the study, and participated in its design and coordination, carried out FISH studies, participated in cytogenetic studies, chromosome analysis, statistical analysis and drafted the manuscript. FS participated in the design and coordination of the study and helped to draft the manuscript. LB participated in the design of the study, helped in clinical activities, helped to draft the manuscript and participated in the statistical analysis. JR participated in clinical activities and drafted the manuscript. CP participated in cytogenetic studies, chromosome analysis and statistical analysis. TG participated in cytogenetic studies, chromosome analysis and statistical analysis. EC participated in the design and coordination of the study, carried out clinical activities and drafted the manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We wish to thank Dr. Lisa G. Shaffer for her valuable comments, and laboratory facilities. We wish to thank to Dr. A. Baldini who kindly provided the CENP-B box biotin labeled probe. We are greatly indebted to Rogelio Troyo Sanromán by his statistical assistance and to Venancio Vazquez by his technical support. This work was supported by CONACYT M-5051 and Universidad de Guadalajara funds.
Figures and Tables
Figure 1 Pedigree of the family. Pedigree of the family. PCD frequencies and familial data of investigated individuals are indicated.
Figure 2 PCD figures. Propositus PCD figures are shown with Cd stain (a), giemsa stain (b), and sister chromatid differential stain from second (c), and third (d), cell cycle.
Figure 3 FISH PCD images. FISH PCD images. All centromeres FISH probe red signals (a), and CENP-B box FISH green signals (b) are shown.
Table 1 Percentages of PCD in repetitive colchicine-treated cultures from family members.
Family member No. of cultures Total mitoses scored Percent PCD
Propositus (III-14) 3 399 10.8
Father (II-11) 3 492 22.35
Mother (II-12) 3 1142 0.175
Paternal aunt (II-3) 2 258 7.0
Paternal aunt (II-8) 8 586 8.36
Paternal aunt (II-9) 9 738 2.9
Paternal aunt (II-10) 7 485 2.47
Maternal aunt (II-14) 7 722 2.7
Maternal aunt (II-15) 7 634 1.73
Controls 10 980 0.87
Table 2 Cell cycle durations in paternal sibship with PCD (paternal aunt II-8, propositus III-14, father II-11) versus maternal sibship without PCD (maternal aunts II-14 and II-15, mother II-12).
Groups Average generation time
Prophases Metaphases Anaphases Total cells
Paternal sibship with PCD Mean* (SD)** 29.49 (6.54) 29.81 (3.01) 21.79 (0.41) 27.03 (3.05)
Maternal sibship without PCD Mean* (SD)** 24.39 (3.17) 27.97 (1.18) 31.83 (3.95) 27.62 (1.63)
Comparison between groups t value 3.17 1.18 3.95 1.63
Significance N.S. N.S. P < 0.005 N.S.
* = hours, ** = Standard deviation. Statistical test, t-Student for independent samples.
Table 3 Main features of published families with PCD and present report.
FEATURES Rudd et al. 1983 [6] Gabarrón et al. 1986 [7] Madan et al. 1987 [8] Bajnoczky and Gardó 1993 [9] Keser et al. 1996 [10] Present report
A B C A B C
Individuals with PCD 3 4 3 4 4 4 4 3 1 4
Individuals with abortion or infertility 2/3§§ 0 1/3§ 2/4§ 0 2/4§ 1/4§ 2/4§ 1/1§ 1/4§§
PCD in colchicine treated cultures * 14–15 17–62 10–16 5.2–36 6–12 7–38 6–21 6–28 32 5–55
PCD in untreated cultures * 7–20 10–21 6–6.5 8.5–39 0–3 NI 17 7 NI 5
Controls * 0–0.5 0–0.5 0–0.5 1 0–1 4.1–5.2 0–1 0 0 0.87
Cd stain NI** NI NI NI + NI NI NI NI +
Cell cycle Short NI NI Short NI NI NI NI NI Short
* = range of percentages. ** = Non investigated. § = Individuals with abortion. §§ = Individuals with infertility.
==== Refs
Domínguez MG Rivera H C-Anaphases: a mitotic variant Ann Génét 1992 35 183 185
Chamla Y Saura R C-Anaphases and their relationship with mitotic aggregation Ann Génét 1993 36 206 210
Plaja A Vendrell T Smeets D Sarret E Gili T Català V Mediano C Scheres JMJC Variegated aneuploidy related to premature centromere division (PCD) is expressed in vivo and is a cancer-prone disease Am J Med Genet 2001 98 216 223 11169558 10.1002/1096-8628(20010122)98:3<216::AID-AJMG1091>3.0.CO;2-0
Kajii T Ikeuchi T Yang ZQ Nakamura Y Tsuji Y Yokomori K Kawamura M Fukuda S Horita S Asamoto A Cancer-prone syndrome of mosaic variegated aneuploidy and total premature chromatid separation: report of five infants Am J Med Genet 2001 104 57 64 11746029 10.1002/ajmg.1580
Jacquemont S Boceno M Rival JM Mechinaud F David A High risk of malignancy in mosaic variegated aneuploidy syndrome Am J Med Genet 2002 109 17 21 11932988 10.1002/ajmg.10281
Rudd NL Teshima HE Martin RH Sisken JE Weksberg R A dominantly inherited cytogenetic anomaly: a possible cell division mutant Hum Genet 1983 65 117 121 6654324 10.1007/BF00286646
Gabarrón J Jiménez A Glover G Premature centromere division dominantly inherited in a subfertile family Cytogenet Cell Genet 1986 43 69 71 3780318
Madan K Lindhout D Palan A Premature centromere division (PCD): a dominantly inherited cytogenetic anomaly Hum Genet 1987 77 193 196 3653892 10.1007/BF00272391
Bajnóczky K Gardó S "Premature anaphase" in a couple with recurrent miscarriages Hum Genet 1993 92 388 390 8225320 10.1007/BF01247341
Keser I Lüleci G Gündüz G Premature centromere division in three unrelated families Ann Génét 1996 39 87 90
Chamla Y C-Anaphases in lymphocyte cultures versus premature centromere division syndromes Hum Genet 1988 78 111 114 3276615 10.1007/BF00278177
Denton ET Brooke WR Howell MW A technique for the simultaneous staining of both nucleolar organizer regions and kinetochores of human chromosomes with silver Stain Technol 1977 52 311 313 74868
Mitchinson JM The biology of the Cell Cycle 1971 Cambridge, Cambridge University Press
Ivett JL Tice RR Average generation time: a new method of analysis and quantitation of cellular proliferation kinetics [abstract] Environ Mutagen 1982 4 358
Palma V Tudon H Buentello S Nava S Ostrosky P Salamanca F Methods for the analysis of cellular kinetics in PHA-stimulated blood lymphocytes using BrdU incorporation. A comparative study Mutat Res 1993 286 267 273 7681538
Shaffer LG FISH manual 1996 Baylor College of Medicine, Texas Medical Center, Houston Texas
Matera AG Ward DC Oligonucleotide probes for the analysis of specific repetitive DNA sequences by fluorescence in situ hybridization Hum Mol Genet 1992 1 535 539 1307254
Ikeuchi T Yang ZQ Wakamatsu K Kajii T Induction of premature chromatid separation (PCS) in individuals with PCS trait and in normal controls Am J Med Genet A 2004 127 128 132 15108198 10.1002/ajmg.a.20666
Sullivan KF Glass ChA CENP-B is a highly conserved mammalian centromere protein with homology to the helix-loop-helix family of proteins Chromosoma 1991 100 360 370 1893793 10.1007/BF00337514
Matsuura S Ito E Tauchi H Komatsu K Ikeuchi T Kajii T Chromosomal instability syndrome with total premature chromatid separation with mosaic variegated aneuploidy is defective in mitotic-spindle checkpoint Am J Hum Genet 2000 67 483 486 10877982 10.1086/303022
Mitchel LS Liberal V Chatterjee A Kirchwegger R Pasche B Gerald W Dobles M Sorger P Murty VVVS Benzra R MAD2 haplo-insufficiency causes premature anaphase and chromosome instability in mammalian cells Nature 2001 409 355 359 11201745 10.1038/35053094
Hanks S Coleman K Reid S Plaja A Firth H FitzPatrick D Kidd A Méhes K Nash R Robin N Shannon N Tolmie J Swansbury J Irrthum A Douglas J Rahman N Constitutional aneuploidy and cancer predisposition caused by biallelic mutations in BUB1B Nat Genet 2004 36 1159 1161 15475955 10.1038/ng1449
Lengauer C Wang Z From spindle checkpoint to cancer Nat Genet 2004 36 1144 1145 15514664 10.1038/ng1104-1144
Voullaire LE Slater HR Petrovic V Choo KHA A functional marker centromere with no detectable alpha-satellite, satellite III, or CENP-B protein: activation of a latent centromere? Am J Hum Genet 1993 52 1153 1163 7684888
Yoda K Ando S Okuda A Kikuchi A Okazaki T In vitro assembly of the CENP-B/alpha-satellite DNA/core histone complex: CENP-B causes nucleosome positioning Genes to Cells 1998 3 533 548 9797455 10.1046/j.1365-2443.1998.00210.x
Nasmyth K Peters JM Uhlmann F Splitting the chromosome: cutting the ties that bind sister chromatids Science 2000 288 1379 1384 10827941 10.1126/science.288.5470.1379
On Line Mendelian Inheritance in Man. Premature Centromere Division; PCD*176430
Scheres JMJC Hustinx TWJ Madan K Beltman JD Lindhout D A mitotic mutant causing non disjunction in man Abstracts of the 7th International Congress of Human Genetics Berlín 1986 163
Kajii T Kawai T Takumi T Misu H Mabuchi O Takahashi Y Tachino M Nihei F Ikeuchi T Mosaic variegated aneuploidy with multiple congenital abnormalities: homozygosity for total premature chromatid separation trait Am J Med Genet 1998 78 245 249 9677059 10.1002/(SICI)1096-8628(19980707)78:3<245::AID-AJMG7>3.0.CO;2-O
Warburton D Kline J Stein Z Hutzler M Chin A Hassold T Does the karyotype of a spontaneous abortion predict the karyotype of a subsequent abortion?-Evidence from 273 women with two karyotyped spontaneous abortion Am J Hum Genet 1987 41 465 483 3631080
|
16174301
|
PMC1242348
|
CC BY
|
2021-01-04 16:03:32
|
no
|
BMC Med Genet. 2005 Sep 20; 6:33
|
utf-8
|
BMC Med Genet
| 2,005 |
10.1186/1471-2350-6-33
|
oa_comm
|
==== Front
BMC Med GenetBMC Medical Genetics1471-2350BioMed Central London 1471-2350-6-331617430110.1186/1471-2350-6-33Research ArticleCell cycle and centromere FISH studies in premature centromere division Corona-Rivera Alfredo [email protected] Fabio [email protected] Lucina [email protected] Jorge R [email protected] Cesar [email protected] Teresa A [email protected] Enrique [email protected] Laboratorio de Citogenética Genotoxicidad y Biomonitoreo, Instituto de Genética Humana Dr. Enrique Corona Rivera, Departamento de Fisiología, División de Disciplinas Básicas, Centro Universitario Ciencias de la Salud, Universidad de Guadalajara, Guadalajara, Jalisco, México2 Unidad de Citogenética, OPD Hospital Civil Fray Antonio Alcalde, Guadalajara, Jalisco, México3 Unit of Investigation in Human Genetics, National Medical Center, Instituto Mexicano del Seguro Social, México City, México4 División de Pediatría, OPD Hospital Civil Juan I. Menchaca, Guadalajara, Jalisco, México5 Laboratorio de Genética Humana, Departamento de Fisiología, División de Disciplinas Básicas, Centro Universitario Ciencias de la Salud, Universidad de Guadalajara, Guadalajara, Jalisco, México2005 20 9 2005 6 33 33 9 3 2005 20 9 2005 Copyright © 2005 Corona-Rivera et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Mitotic configurations consistent in split centromeres and splayed chromatids in all or most of the chromosomes or premature centromere division (PCD) have been described in three categories. (1) Low frequency of PCD observed in colchicines-treated lymphocyte cultures from normal individuals. (2) High frequency of PCD with mosaic variegated aneuploidy. (3) High frequency of PCD as a sole chromosome abnormality observed in individuals with no recognizable clinical pattern. We report four members of a family with the third category of PCD.
Methods
Cell cycle duration assessed by average generation time using differential sister chromatid stain analysis and FISH studies of DNA centromere sequences in PCD individuals, are included and compared with previously reported PCD individuals from 9 families.
Results
We observed PCD in colchicine-treated cultures from the propositus, his father, and two paternal aunts but not in his mother and four other paternal and maternal family members, as well as in untreated cultures from the propositus and his father. We observed cytological evidence of active centromeres by Cd stain. Significative cell cycle time reduction in anaphases of PCD individuals (average generation time of 21.8 h;SD 0.4) with respect to individuals without PCD (average generation time of 31.8 h;SD 3.9) was observed (P < 0.005, Student t-test for independent samples). Increased cell proliferation kinetics was observed in anaphasic cells of individuals with PCD, by differential sister chromatid stain analysis. FISH studies revealed the presence of alpha satellite DNA from chromosomes 1, 13, 21/18, X, all centromeres, and CENP-B box sequences in metaphasic and anaphasic cells from PCD individuals.
Conclusion
This report examines evidences of a functional relationship between PCD and cell cycle impairment. It seems that essential centromere integrity is present in these cases.
==== Body
Background
Mitotic configurations consistent in split centromeres and splayed chromatids in all or most of the chromosomes or "premature centromere division" (PCD), have been described in three categories. They are:
(1) Low frequency of PCD (up to 3% of the mitosis) observed in colchicines-treated lymphocyte cultures from normal individuals [1,2].
(2) High frequency of PCD (5% or more) with mosaic aneuploidies involving a variety of chromosomes, called "mosaic variegated aneuploidy", observed in individuals with microcephaly, growth deficiency, severe mental retardation, and risk of malignancy [3-5].
(3) High frequency of PCD (5% or more) as a sole chromosome abnormality [6-10]. Individuals with this condition have no recognizable clinical pattern or occur in healthy individuals. Association of this PCD trait with abortions and infertility has been reported [6,7,9,10], but other authors suggest this PCD trait to be harmless [1,8,11].
We report a family that corresponds to the third category of PCD, in which four individuals showed PCD as a sole chromosome abnormality. Cell cycle duration, assessed by average generation time, and FISH studies of the centromere, are considered. Findings in the family are compared with those on 30 other previously reported individuals with PCD from 9 families. This report examines the evidences of a relationship between PCD and cell cycle impairment of cells bearing main structural centromere components from PCD individuals.
Methods
Family data
A total of nine individuals were available to be studied, five of them from the paternal lineage, four of them from the maternal lineage and the propositus. Infertility was observed in two paternal aunts. Family data are summarized on Fig. 1. The non-consanguineous parents were both aged 27 years at propositus birth. The propositus was born at 36 weeks by abdominal delivery indicated by maternal pre-eclamptic toxemia and showing neonatal hypoxia. Birth weight was 2100 g and height was 46 cm. Early signs included: hypotonia, strabismus, seizures, recurrent respiratory infections and occasional breath-holding spells. Physical examination at 5 years of age showed microcephaly (OFC -4.1 SD), low weight (-4.1 SD) and borderline length (10th percentile); plagiocephaly and brachycephaly, nystagmus, phimosis, right cryptorchidia, adducted thumbs, syndactyly 2–3 of toes, four limbs spasticity and increased deep tendon reflexes. On X-rays, hyperbrachycephaly (cephalic index 91.7), scoliosis, spina bifida occulta (T1–T5), coxa valga, slim bones and osteopenia were found. Cortical and subcortical atrophy was found on cranial CT scan. Brainstem auditory evoked potentials were normal. Psychometric test demonstrated a development age of 3–6 months. The propositus died at 10 years of age by pneumonia. Propositus relatives were phenotypically healthy.
Cytogenetic studies
We performed GTG banded karyotypes from peripheral blood lymphocyte cultures stimulated by phytohemaglutinin and treated one hour with colchicine (SIGMA, 0.104 μg/ml), in the propositus and eight family members indicated in Fig. 1. On harvesting, 15 minutes of hypotonic treatment (0.075 M KCl at 37°C) was performed. Chromosome aberrations in the karyotypes of these individuals were not observed. PCD frequencies in repeated cultures of available individuals were obtained in at least three repeated cultures. Around 100 mitosis were scored per culture giving a total range of 258 to 1142 cells scored per individual (Table 1). To obtain basal frequencies of PCD, ten healthy young individuals were used as controls. Additional simultaneous cultures without mitostatic treatment were performed in the propositus and his parents. Cd staining was performed in the propositus and his parents according to Denton et al. (1977) [12].
Cell cycle studies
Cellular proliferation kinetics was used to determine the cell cycle duration, which is the interval between one mitosis and the subsequent [13]. The method called average generation time (AGT) by differential sister-chromatid stain [14,15], was used to determine cell cycle durations. We compared the results between family members with more than 5% of PCD versus those with less than 3% of anaphase frequencies. The original AGT method [14,15], consider for cell cycle calculations only metaphases, in this family, calculations were also performed considering, prophases, anaphases and total mitotic cells. The procedure was as follows. We obtained AGT's of each individual from accumulated data of 3 simultaneous 72 h cultures with 5 μg/ml of 5'-bromodeoxyuridine (SIGMA), added 24 h after set-up. In each culture, we scored prophases, metaphases and anaphases which had completed either one, two or three cell cycles identified by differential sister chromatid stain pattern. Then, AGT, were calculated per mitotic stage and per individual as follows. (i) Taking into account the percentage of cells at first (M1), second (M2) and third (M3) cell cycle, the replication index (RI) was obtained from this formula: RI = (1X%M1+2X%M2+3%M3)/100. (ii) The RI was used to obtain the AGT following the equation: AGT = time of harvest after exposure to 5'-BrdU/RI. Finally, AGT's were compared in paternal individuals with PCD versus those individuals with low anaphase frequencies from the maternal sibship. Statistical test t-Student for independent samples was used to compare both groups.
FISH studies
We searched for the presence of constitutive structural components of the centromere such as alpha satellite DNA sequences in the propositus and his parents with FISH according to a standard protocol [16], using alpha satellite DNA directly labeled probes to chromosomes 1, 13/21, 18, and X, as well as all centromeres probe. Besides, we searched for the presence of CENP-B box sequence in the propositus using a biotin labeled probe, following the protocol of Matera and Ward (1992) [17].
Results
Cytogenetic studies
PCD frequencies in repeated cultures of available individuals are shown in Table 1. Individuals with PCD of 5% or more corresponded to the propositus, his father and two paternal aunts (II-3, and II-8). Individuals with less of 3% of anaphases were considered as normal and corresponded to mother's propositus, two maternal aunts (II-14, II-15), and two paternal aunts (II-9, II-10). Basal frequencies of PCD, in ten healthy young individuals used as controls, did not differ with respect to PCD individual with less of 3% of anaphases. A PCD image of the propositus is shown in figure 2a. In additional simultaneous cultures without mitostatic treatment PCD was observed in the propositus (5%) and his father (5.5%), but not in his mother. Two centromeric dots in evident primary constrictions were observed by Cd staining in the propositus and his parents (fig. 2b). This finding is considered a cytological evidence of active centromeres [12].
Cell cycle studies
We observed cell cycle duration significatively reduced in anaphasic cells but not in prophasic, metaphasic or total cells (Table 2). Cell cycle reduction was then attributed only to cells in anaphase (PCD cells). Additionally, we observed that in the same culture conditions, 38% of anaphases in the propositus and his father reached the third cell cycle, while only 15% of metaphases did. This was interpreted as increased cell proliferation kinetics of anaphasic cells in individuals with PCD associated to cell cycle shortening in PCD cells. Anaphases from second and third cell cycle are shown in figures 2c and 2d respectively.
FISH studies
We observed in all cases positive fluorescent signals (figure 3a), indicating the presence of all tested sequences. We observed a positive fluorescent signal pattern, indicating the presence of CENP-B box sequence in the propositus (figure 3b).
Discussion
We report four family members with more than 5% of colchicine-anaphase frequencies as a sole chromosome abnormality. Nine families have been reported referring to this trait as PCD [6-10]. Mitosis obtained from colchicine arrested lymphocyte cultures of normal individuals, show rates below 3% [1], 5% [2] or 1% observed in our control individuals. Previous reports of PCD frequencies without mosaic variegated aneuploidy (MVA), ranges 5 to 38% [6-10]. In our family, PCD was observed in the propositus, his father, and two paternal aunts in repeated colchicine-treated cultures in average frequencies of 7 to 22%. It was shown [18], that PCD can be induced through hypotonic increasing time treatment of mitotic cells in peripheral blood lymphocytes of healthy individuals and patients homozygous to PCD trait or PCD and MVA. They found that 0.075 M KCl at 37°C for 20 min, showed 0–2% cells in PCD which fits with our observed frequencies because 15 minutes of hypotonic treatment were used in our peripheral blood cultures. Total mitosis scored in repeated cultures of previous [6-10], and present report did not provide evidences of MVA. Although PCD is considered a rare phenomenon, two studies found in selected population frequencies of 1 of 100 [1] or 1 of 1000 [11].
Only two previous studies tested cell cycle duration in individuals with PCD as a sole chromosome abnormality (Table 3). In both cases they obtained evidences that the cell cycle time can be altered in PCD individuals [6,7]. Rudd et al. (1983) [6], found reduced metaphase duration only in some cultured cells, supposing that such cells could have corresponded to those with PCD. Gabarrón et al. (1986) [7], inferred that only the cells in anaphase or PCD cells, showed accelerated proliferation kinetics and consequently reduced cell cycle duration. We provide evidences to confirm that cell cycle duration is reduced in PCD cells because only in anaphasic cells the cell cycle duration in the family members with PCD was statistically reduced. Additionally, we observed increased cell proliferation kinetics of anaphasic cells in individuals with PCD. These findings are compatible with the co-existence of cellular subpopulations bearing differential proliferation kinetics. Cell cycle reduction, possibly related to premature separation of centromeres and persistence of anaphases, could then be considered a distinctive finding in these cases.
Cell cycle progression requires control mechanisms that could be associated to PCD origin. A basic defect of cell cycle progression or metaphase-anaphase transition in PCD was suggested [19]. Matsuura et al. (2000) [20], demonstrated that cultured fibroblasts from two infants with PCD and mosaic variegated aneuploidy are insensitive to the colcemid-induced mitotic-spindle checkpoint. Mitchel et al. (2001) [21], found that mitotic checkpoint defective MAD2+/- haploinsufficient human colon carcinoma cells showed 20% of precocious anaphases with prematurely separated sister chromatids, compared with 1% in wild-type cells, proposing that PCD is a suitable cytogenetic marker for the identification of mitotic checkpoint defects. In our family PCD was also observed in cultures without colchicine from the propositus and his father as in other PCD reports [6-8] (Table 3). Cultured PCD cells with defective colcemid-induced mitotic-spindle checkpoint reported by Matsuura et al. (2000) [20], were unresponsive to colchicine. Although in our family MVA was not observed as in Matsuura et al. (2000) [20] report, in both cases the mitotic arrest signal was overruled in PCD cells. Hanks et al. (2004) [22], provided the evidence that gene mutations can result in a defective spindle checkpoint in humans. They screened the full coding sequence and intron-exon boundaries of BUB1B, and found truncating and missense mutations inherited from different parents in five of eight families with mosaic variegated aneuploidy providing the first evidence in humans that gene mutations might be responsible for aneuploidy in human cancers. Interestingly, cytogenetic data of families 1, 4 and 5 in Hanks et al. (2004) [22] report showed also PCD, as well as in 6 of those 15 reported cases update [5]. Considering that subjacent genetic cause was demonstrated to MVA, this opens the possibility that BUB1B defects can be involved in PCD origin. Mitotic spindle checkpoint serves as a surveillance mechanism that ensures the faithful transmission of chromosomes from a mother cell to its two daughter cells during mitosis [23], this can be involved also in PCD origin. On the other hand, because at least 6 genes have been involved with such checkpoint [23], the origin of PCD may be heterogeneous and should involve a mechanism that triggers the whole set chromosome segregation at mitotic spindle checkpoint.
Other aspect to be considered in PCD origin is the centromere. A basic defect of centromeric region in PCD was suggested [19]. Two essential DNA sequences of the centromere were evaluated by FISH in this family: alpha satellite and CENPB-Box sequences. Centromere function requires the presence of alpha satellite DNA in all human centromeres [24]. We observed alpha satellite DNA from all centromeres and the centromere of chromosomes 1, 13/21, 18, and X, as well as centromeric 17 bp CENPB-Box sequences in prophasic, metaphasic and anaphasic cells from the propositus and his parents. CENPB-Box sequence interacts with the kinetochore protein CENP-B required for the pairing of sister chromatids as structural support and in the conformation of primary constriction and kinetochore [25]. Cytological evidence support the presence of functional centromeres in PCD cells, by positive Cd stain in the propositus and his parents and in one previous PCD report [8]. Also, primary constrictions are evident in this and previous PCD reports. It seems that essential centromere integrity is present and remains unclear if whether or not is involved in PCD origin. Other mechanisms related to cell cycle regulation and functional components of the centromeric region such as defective centromeric cohesion [26], or kinetochore defective proteins are probable.
The PCD trait as a sole chromosome abnormality occurs in healthy individuals. Some authors suggest this PCD trait to be harmless [1,8,18]. In this report four individuals presented PCD, and three of them were phenotypically normal. Noteworthy, all the 30 PCD individuals from 9 previous families included in Table 3 were also phenotypically normal. Individuals with this category of PCD have no a recognizable clinical pattern [6-10]. In three of such families clinical findings were informed and reported as coincidental [6,8,10]. The abnormal phenotype observed in the propositus shows no concordance to previous PCD without MVA cases. PCD trait observed in healthy paternal relatives and all previous cases, represent the common one dose effect of this autosomal dominant trait (OMIM, *176430) [27]. Such mode of inheritance was concordant with this report because male to male transmission was observed. In autosomal dominant PCD and abnormal phenotype associated to MVA, homozygosity was implicated [28,29]. This statement was confirmed by Plaja et al. (2001) [3] in three patients compared with 8 previous cases exhibiting microcephaly, CNS anomalies, mental retardation, prenatal and postnatal growth retardation and cancer, proposing that in vivo occurrence of random aneuploidies and chromosome or genome instability disorder explained some of the clinical data. It seems that variegated aneuploidy is associated to an abnormal phenotype. Our propositus showed prenatal and postnatal growth retardation, profound developmental delay, hypoplasia of the brain and clonic seizures, coincident with MVA reports [5], but our case did not show MVA nor apparent cancer risk. Alternatively, considering that the inheritance of MVA is recessive [5], and some heterozygotes show levels of PCD without variegated aneuploidy, this can be compatible with those individuals described in present report or in previous reports regarding apparently harmless PCD. However, the relationship between PCD and MVA is uncertain [5]. Also, we considered the possibility that neurogical affectation in the patient studied by us could be associated with neonatal hypoxia. In these cases, genetic heterogeneity may be involved.
Association of this PCD trait with abortions and infertility has been reported [6,7,9,10]. This was observed in 12 of 34 PCD individuals from 8 of 10 previous families including present report (Table 3). We observed infertility in two paternal aunts (II-2 and II-3 in fig. 1); cytogenetical analysis was available only in one of them observing PCD. The estimated abortion frequency in descendents of reported PCD individuals was 37% (22 of 60) which is higher than those observed in general population of 15% [30]. In one report, unexplained recurrent abortion observed in both parents with PCD was considered the consequence of abnormal behavior of the centromeres involving probable homozygous effect [9]. Previous observations are coincidental but remark the occurrence of subfertility in PCD individuals.
Conclusion
Present report represents a new family with PCD as a sole chromosome abnormality. Cell cycle studies revealed that cell cycle reduction could be considered a distinctive finding in these cases. Based in previous reports and the fact that cells of PCD patients were unresponsive to colchicine is probable that a defective colcemid-induced mitotic-spindle checkpoint is involved. Is open the possibility that BUB1B defects or other genes involved in such checkpoint may be involved in PCD origin. In this cases considered DNA centromeric sequences were present. It seems that essential centromere integrity is present and remains unclear if whether or not is involved in PCD origin. Other mechanisms related to cell cycle regulation and functional components of the centromeric region may be involved. Interestingly, the PCD trait as a sole chromosome abnormality occurs in healthy individuals and there is not a characteristic associated abnormal phenotype. Only subfertility seems to be a common finding in these families. Those families deserve further investigation in order to understand possible mechanism of this mitotic trait.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
AR conceived of the study, and participated in its design and coordination, carried out FISH studies, participated in cytogenetic studies, chromosome analysis, statistical analysis and drafted the manuscript. FS participated in the design and coordination of the study and helped to draft the manuscript. LB participated in the design of the study, helped in clinical activities, helped to draft the manuscript and participated in the statistical analysis. JR participated in clinical activities and drafted the manuscript. CP participated in cytogenetic studies, chromosome analysis and statistical analysis. TG participated in cytogenetic studies, chromosome analysis and statistical analysis. EC participated in the design and coordination of the study, carried out clinical activities and drafted the manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We wish to thank Dr. Lisa G. Shaffer for her valuable comments, and laboratory facilities. We wish to thank to Dr. A. Baldini who kindly provided the CENP-B box biotin labeled probe. We are greatly indebted to Rogelio Troyo Sanromán by his statistical assistance and to Venancio Vazquez by his technical support. This work was supported by CONACYT M-5051 and Universidad de Guadalajara funds.
Figures and Tables
Figure 1 Pedigree of the family. Pedigree of the family. PCD frequencies and familial data of investigated individuals are indicated.
Figure 2 PCD figures. Propositus PCD figures are shown with Cd stain (a), giemsa stain (b), and sister chromatid differential stain from second (c), and third (d), cell cycle.
Figure 3 FISH PCD images. FISH PCD images. All centromeres FISH probe red signals (a), and CENP-B box FISH green signals (b) are shown.
Table 1 Percentages of PCD in repetitive colchicine-treated cultures from family members.
Family member No. of cultures Total mitoses scored Percent PCD
Propositus (III-14) 3 399 10.8
Father (II-11) 3 492 22.35
Mother (II-12) 3 1142 0.175
Paternal aunt (II-3) 2 258 7.0
Paternal aunt (II-8) 8 586 8.36
Paternal aunt (II-9) 9 738 2.9
Paternal aunt (II-10) 7 485 2.47
Maternal aunt (II-14) 7 722 2.7
Maternal aunt (II-15) 7 634 1.73
Controls 10 980 0.87
Table 2 Cell cycle durations in paternal sibship with PCD (paternal aunt II-8, propositus III-14, father II-11) versus maternal sibship without PCD (maternal aunts II-14 and II-15, mother II-12).
Groups Average generation time
Prophases Metaphases Anaphases Total cells
Paternal sibship with PCD Mean* (SD)** 29.49 (6.54) 29.81 (3.01) 21.79 (0.41) 27.03 (3.05)
Maternal sibship without PCD Mean* (SD)** 24.39 (3.17) 27.97 (1.18) 31.83 (3.95) 27.62 (1.63)
Comparison between groups t value 3.17 1.18 3.95 1.63
Significance N.S. N.S. P < 0.005 N.S.
* = hours, ** = Standard deviation. Statistical test, t-Student for independent samples.
Table 3 Main features of published families with PCD and present report.
FEATURES Rudd et al. 1983 [6] Gabarrón et al. 1986 [7] Madan et al. 1987 [8] Bajnoczky and Gardó 1993 [9] Keser et al. 1996 [10] Present report
A B C A B C
Individuals with PCD 3 4 3 4 4 4 4 3 1 4
Individuals with abortion or infertility 2/3§§ 0 1/3§ 2/4§ 0 2/4§ 1/4§ 2/4§ 1/1§ 1/4§§
PCD in colchicine treated cultures * 14–15 17–62 10–16 5.2–36 6–12 7–38 6–21 6–28 32 5–55
PCD in untreated cultures * 7–20 10–21 6–6.5 8.5–39 0–3 NI 17 7 NI 5
Controls * 0–0.5 0–0.5 0–0.5 1 0–1 4.1–5.2 0–1 0 0 0.87
Cd stain NI** NI NI NI + NI NI NI NI +
Cell cycle Short NI NI Short NI NI NI NI NI Short
* = range of percentages. ** = Non investigated. § = Individuals with abortion. §§ = Individuals with infertility.
==== Refs
Domínguez MG Rivera H C-Anaphases: a mitotic variant Ann Génét 1992 35 183 185
Chamla Y Saura R C-Anaphases and their relationship with mitotic aggregation Ann Génét 1993 36 206 210
Plaja A Vendrell T Smeets D Sarret E Gili T Català V Mediano C Scheres JMJC Variegated aneuploidy related to premature centromere division (PCD) is expressed in vivo and is a cancer-prone disease Am J Med Genet 2001 98 216 223 11169558 10.1002/1096-8628(20010122)98:3<216::AID-AJMG1091>3.0.CO;2-0
Kajii T Ikeuchi T Yang ZQ Nakamura Y Tsuji Y Yokomori K Kawamura M Fukuda S Horita S Asamoto A Cancer-prone syndrome of mosaic variegated aneuploidy and total premature chromatid separation: report of five infants Am J Med Genet 2001 104 57 64 11746029 10.1002/ajmg.1580
Jacquemont S Boceno M Rival JM Mechinaud F David A High risk of malignancy in mosaic variegated aneuploidy syndrome Am J Med Genet 2002 109 17 21 11932988 10.1002/ajmg.10281
Rudd NL Teshima HE Martin RH Sisken JE Weksberg R A dominantly inherited cytogenetic anomaly: a possible cell division mutant Hum Genet 1983 65 117 121 6654324 10.1007/BF00286646
Gabarrón J Jiménez A Glover G Premature centromere division dominantly inherited in a subfertile family Cytogenet Cell Genet 1986 43 69 71 3780318
Madan K Lindhout D Palan A Premature centromere division (PCD): a dominantly inherited cytogenetic anomaly Hum Genet 1987 77 193 196 3653892 10.1007/BF00272391
Bajnóczky K Gardó S "Premature anaphase" in a couple with recurrent miscarriages Hum Genet 1993 92 388 390 8225320 10.1007/BF01247341
Keser I Lüleci G Gündüz G Premature centromere division in three unrelated families Ann Génét 1996 39 87 90
Chamla Y C-Anaphases in lymphocyte cultures versus premature centromere division syndromes Hum Genet 1988 78 111 114 3276615 10.1007/BF00278177
Denton ET Brooke WR Howell MW A technique for the simultaneous staining of both nucleolar organizer regions and kinetochores of human chromosomes with silver Stain Technol 1977 52 311 313 74868
Mitchinson JM The biology of the Cell Cycle 1971 Cambridge, Cambridge University Press
Ivett JL Tice RR Average generation time: a new method of analysis and quantitation of cellular proliferation kinetics [abstract] Environ Mutagen 1982 4 358
Palma V Tudon H Buentello S Nava S Ostrosky P Salamanca F Methods for the analysis of cellular kinetics in PHA-stimulated blood lymphocytes using BrdU incorporation. A comparative study Mutat Res 1993 286 267 273 7681538
Shaffer LG FISH manual 1996 Baylor College of Medicine, Texas Medical Center, Houston Texas
Matera AG Ward DC Oligonucleotide probes for the analysis of specific repetitive DNA sequences by fluorescence in situ hybridization Hum Mol Genet 1992 1 535 539 1307254
Ikeuchi T Yang ZQ Wakamatsu K Kajii T Induction of premature chromatid separation (PCS) in individuals with PCS trait and in normal controls Am J Med Genet A 2004 127 128 132 15108198 10.1002/ajmg.a.20666
Sullivan KF Glass ChA CENP-B is a highly conserved mammalian centromere protein with homology to the helix-loop-helix family of proteins Chromosoma 1991 100 360 370 1893793 10.1007/BF00337514
Matsuura S Ito E Tauchi H Komatsu K Ikeuchi T Kajii T Chromosomal instability syndrome with total premature chromatid separation with mosaic variegated aneuploidy is defective in mitotic-spindle checkpoint Am J Hum Genet 2000 67 483 486 10877982 10.1086/303022
Mitchel LS Liberal V Chatterjee A Kirchwegger R Pasche B Gerald W Dobles M Sorger P Murty VVVS Benzra R MAD2 haplo-insufficiency causes premature anaphase and chromosome instability in mammalian cells Nature 2001 409 355 359 11201745 10.1038/35053094
Hanks S Coleman K Reid S Plaja A Firth H FitzPatrick D Kidd A Méhes K Nash R Robin N Shannon N Tolmie J Swansbury J Irrthum A Douglas J Rahman N Constitutional aneuploidy and cancer predisposition caused by biallelic mutations in BUB1B Nat Genet 2004 36 1159 1161 15475955 10.1038/ng1449
Lengauer C Wang Z From spindle checkpoint to cancer Nat Genet 2004 36 1144 1145 15514664 10.1038/ng1104-1144
Voullaire LE Slater HR Petrovic V Choo KHA A functional marker centromere with no detectable alpha-satellite, satellite III, or CENP-B protein: activation of a latent centromere? Am J Hum Genet 1993 52 1153 1163 7684888
Yoda K Ando S Okuda A Kikuchi A Okazaki T In vitro assembly of the CENP-B/alpha-satellite DNA/core histone complex: CENP-B causes nucleosome positioning Genes to Cells 1998 3 533 548 9797455 10.1046/j.1365-2443.1998.00210.x
Nasmyth K Peters JM Uhlmann F Splitting the chromosome: cutting the ties that bind sister chromatids Science 2000 288 1379 1384 10827941 10.1126/science.288.5470.1379
On Line Mendelian Inheritance in Man. Premature Centromere Division; PCD*176430
Scheres JMJC Hustinx TWJ Madan K Beltman JD Lindhout D A mitotic mutant causing non disjunction in man Abstracts of the 7th International Congress of Human Genetics Berlín 1986 163
Kajii T Kawai T Takumi T Misu H Mabuchi O Takahashi Y Tachino M Nihei F Ikeuchi T Mosaic variegated aneuploidy with multiple congenital abnormalities: homozygosity for total premature chromatid separation trait Am J Med Genet 1998 78 245 249 9677059 10.1002/(SICI)1096-8628(19980707)78:3<245::AID-AJMG7>3.0.CO;2-O
Warburton D Kline J Stein Z Hutzler M Chin A Hassold T Does the karyotype of a spontaneous abortion predict the karyotype of a subsequent abortion?-Evidence from 273 women with two karyotyped spontaneous abortion Am J Hum Genet 1987 41 465 483 3631080
|
16150144
|
PMC1242349
|
CC BY
|
2021-01-04 16:39:19
|
no
|
Lipids Health Dis. 2005 Sep 8; 4:18
|
latin-1
|
Lipids Health Dis
| 2,005 |
10.1186/1476-511X-4-18
|
oa_comm
|
==== Front
Emerg Themes EpidemiolEmerging Themes in Epidemiology1742-7622BioMed Central London 1742-7622-2-91615330710.1186/1742-7622-2-9Analytic PerspectiveThe epidemiological impact of antiretroviral use predicted by mathematical models: a review Baggaley Rebecca F [email protected] Neil M [email protected] Geoff P [email protected] Department of Infectious Disease Epidemiology, Imperial College London, Norfolk Place, London W2 1PG, UK2005 10 9 2005 2 9 9 13 4 2005 10 9 2005 Copyright © 2005 Baggaley et al; licensee BioMed Central Ltd.2005Baggaley et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
This review summarises theoretical studies attempting to assess the population impact of antiretroviral therapy (ART) use on mortality and HIV incidence. We describe the key parameters that determine the impact of therapy, and argue that mathematical models of disease transmission are the natural framework within which to explore the interaction between antiviral use and the dynamics of an HIV epidemic. Our review focuses on the potential effects of ART in resource-poor settings. We discuss choice of model type and structure, the potential for risk behaviour change following widespread introduction of ART, the importance of the stage of HIV infection at which treatment is initiated, and the potential for spread of drug resistance. These issues are illustrated with results from models of HIV transmission. We demonstrate that HIV transmission models predicting the impact of ART use should incorporate a realistic progression through stages of HIV infection in order to capture the effect of the timing of treatment initiation on disease spread. The realism of existing models falls short of properly reproducing patterns of diagnosis timing, incorporating heterogeneity in sexual behaviour, and describing the evolution and transmission of drug resistance. The uncertainty surrounding certain effects of ART, such as changes in sexual behaviour and transmission of ART-resistant HIV strains, demands exploration of best and worst case scenarios in modelling, but this must be complemented by surveillance and behavioural surveys to quantify such effects in settings where ART is implemented.
==== Body
Introduction
The epidemiological impact of widescale use of (highly active) antiretroviral therapy (HAART, or ART) among HIV patients in industrialised countries has been explored by a number of mathematical modelling studies [1-5]. The consequences of ART use are far from intuitive. Successful ART decreases plasma [6] and seminal viral load [7,8] and so is thought to reduce HIV infectiousness. However, its main function is to increase the life expectancy of infected individuals [9,10], and over time this causes the pool of potential transmitters of infection to grow. These two factors – decreased infectivity but increased duration of infectiousness – have opposing effects on transmission. In addition, increases in risk behaviour could result from increased optimism about HIV prognosis due to the availability of ART. This is an area of uncertainty, with contradictory evidence [11-15].
Mathematical models can be used to address questions regarding the potential impact and effectiveness of various strategies. In terms of ART use, they can be used to investigate:
1) optimising the efficient use of ART;
2) the epidemiological consequences of ART and interaction with behavioural changes/interventions;
3) the likely course of drug resistance evolution
a. Within the individual;
b. Between individuals;
4) achievable levels of coverage and effectiveness;
5) the effective and efficient use of second line treatments; and
6) demographic/health care impact.
This review will briefly describe a range of models investigating the impact of ART use in various settings and evaluate the utility of these dynamic models.
The range of ART models
Mathematical models examining the epidemiological impact of ART broadly fall into two categories; those incorporating HIV transmission dynamics, where incidence of new infections is dependent on HIV prevalence [1,2,4,5], and simpler linear models [16-18]. A summary of ART models is provided in Table 1. Aalen et al [17] constructed a model describing men who have sex with men (MSM) in England and Wales and the use of ART. This Markov multi-stage model represented stages of HIV infection based on CD4 count. The authors considered a variety of treatment scenarios, and incorporated asymptomatic and symptomatic individuals and the concept of eligibility for treatment, making the simulation of treatment uptake and its impact more realistic than previous work. Wood et al [16] constructed a health economic model to predict the future impact of low-level ART use in South Africa from 2000 to 2005. The authors modelled total drug cost, cost per life year gained and the proportion of per person healthcare expenditure required to finance ART in each scenario. The study involved a cost effectiveness analysis comparing the epidemiological impact of ART with other interventions such as prevention of mother-to-child transmission (PMTCT). Freedberg et al [18] also used stages of disease determined by CD4 count and predicted the incremental cost per quality-adjusted year of life gained by ART in the US. Wilson and Blower [19] used a spatial mathematical model to explore ART allocation strategies among health care facilities in the province of KwaZulu-Natal, South Africa, with an emphasis on maximising equity in access to treatment.
Table 1 Summary of existing ART models, by date of publication.
Study Model structure Setting Assumptions Key outcomes (pertaining to ART) and comments
Zaric et al 1998 [74] Dynamic, difference equations MSM and IDU, US cities/regions Infection stratified into HIV and AIDS stages. Allows acquired and transmitted drug resistance. Rate of resistance evolution: 95% per year for non-adherents, 5% per year for adherents. Increased life expectancy due to ART: 1.5-fold for adherents, 1.2-fold for non-adherents, 1.2-fold for adherents infected with drug resistant HIV. The most important factor affecting emergence of drug resistance is adherence to ART.
Aalen et al 1999 [17] Linear MSM, England and Wales, 1990s Markov multi stage model (stage based on CD4 count). Explicit diagnosis and treatment. Estimated HIV incidence in the MSM population, which was fixed at 1200 cases/year for all but one scenario, where incidence was halved from 1995 onwards. Decrease in AIDS incidence due to ART. Number receiving treatment will increase 50–100% by 2001 compared to pre-1996.
Wood et al 2000 [75] Linear model IDU, Vancouver, Canada 1999–2006 Treatment uptake: 80% (scenario 1), 20% (scenario 2). Median increase in life expectancy due to ART: 7 years. No drug resistance or stratification by infection stage or sexual activity class. Prevalence estimates used in DemProj, part of the Spectrum suite of models (information at ) (3 scenarios: prevalence reduced from 7% in 1999 to 5% in 2006, prevalence remains at 7%, prevalence increases to 9% in 2006. Calculated life expectancy and AIDS deaths 1999–2006 for each scenario. Concluded that low level ART use is not sufficient to increase life expectancy in this population and called for expansion in ART coverage.
Wood et al 2000 [16] Linear health economic model (based on previous model [75]) South Africa 2000–2005 Median increase in life expectancy due to ART: 6 years (range: 5–7). Treatment uptake: 25% of infected adults. Spectrum AIDS Impact model was used to adjust the population projections for current and projected HIV-associated mortality. Providing ART for 25% of the infected population could prevent a 3.1 year decline in life expectancy and more than 430,000 incident cases, but with disproportionate expenditure ($19 billion at 2000 prices) compared to preventing mother-to-child transmission.
Blower et al 2000 [5] Dynamic, deterministic MSM, San Francisco, US, up to 2010 Changes in sexual behaviour: no change to doubling of risk. Treatment coverage rates: 50–90% uptake per year. ART reduces infectivity 2- to 100-fold. Acquired drug resistance: 10–60% per year (infections can revert back to ART-sensitive). Resistance is transmitted, but is less fit than wild type. No stratification by stage of infection or sexual activity class. Increased life expectancy due to ART: 1.5- to 3.0-fold. Increasing ART usage would decrease the death rate and substantially reduce HIV incidence (wide range of results due to uncertainty in parameter estimations).
Blower et al 2001 [63] Dynamic, deterministic (used previous model [5]) MSM, San Francisco, US, 1996–2005 Treatment uptake and drug resistance evolution rates as for [5]. No change in risk behaviour. Assumed no resistant strains could arise that were as transmissible as wild type. Transmissibility range: 1–90% as fit as wild type. Implicitly allows superinfection with wild type virus of subjects with primary resistance. Prevalence of ART resistance is already high in San Francisco and will continue to increase substantially through 2005. Transmitted drug resistance will remain low, only increasing gradually, with a doubling time of around 4 years and a predicted median 15.6% (range 0.05–73.21%) new infections resistant to ARVs by 2005.
Freedberg et al 2001 [18] Linear health economic model US Monte Carlo simulation of a hypothetical cohort of infected patients. Disease progression predicted by CD4 count (6 categories) and viral load (5 categories). Detailed description and associated costs of HIV-related morbidity, opportunistic infections and death. Virologic failure represented as 0.5 log increase in viral load for 2 consecutive months. Increased life expectancy due to ART: 2 years. The cost-effectiveness ratio for ART was $13,000–$23,000 per quality-adjusted life year gained. Initial CD4 count and drug costs were the most important determinants of costs, clinical benefits, and cost effectiveness.
Tchetgen et al 2001 [76] Dynamic, deterministic MSM (assumes same population as Blower et al [5]) No stratification by sexual activity or stage of infection (model is for all stages except AIDS; progression to AIDS exits an individual from the model population). Models diagnosis separately from treatment initiation. No sexual behaviour change due to ART. Drug resistance emerges at 1.2–13.5% per year for adherent patients and 67.3–85.9% per year for non-adherent patients. Resistant strains are half as transmissible as wild type. Untreated resistant infections may revert to wild type infections (10% per year). 60% of treated patients adhere. Increased life expectancy due to ART: approximately 3-fold. ART reduces infectivity by 74%. Withdrawal rates also vary by adherence. Although screening for adherence is likely to reduce levels of drug resistance compared to treating all patients, HIV and AIDS incidence rates are likely to increase unless screening accuracy is extremely high.
Dangerfield et al 2001 [77] Dynamic, deterministic MSM, UK 1981–1998 Five stages of infection with varying infectivity broadly corresponding to primary infection, incubation, pre-AIDS and early and late AIDS. Three levels of sexual activity with proportionate mixing. No drug resistance. Proportion initiating ART at each stage (models 1,2 and 3 respectively): incubation = 0%, 0%, 60%; preAIDS = 0%, 10%, 25%; early stage AIDS = 0%, 10%, 35%. No uptake for late stage AIDS, which is defined as the final few months of care – authors assume patients only reach this stage as a result of treatment failure. Infectivity decreases to a constant level for all those treated, which is 35–40-fold less than for pre-AIDS. Three models were designed, differing by prognosis of patients experiencing treatment failure for models 1 and 2. Model 3 stratifies life expectancy on ART by stage of infection at which treatment is initiated.
Law et al 2001 [4] Dynamic, deterministic MSM, Australia 1996 Population-level changes in sexual behaviour: no change to doubling of risk. Decrease in infectivity due to ART: 10-fold (range: 100-fold to none). Proportion of individuals diagnosed and treated increases with progression of disease, as determined by CD4 count. HIV diagnosis modelled separately to ART initiation, median 2-fold decrease in risk behaviour upon diagnosis (range: 25–75% reduction). No stratification by sexual activity group. No incorporation of drug resistance. Stratified by stage of infection in terms of CD4 count (>500 cells/ml, 200–500, <200, AIDS). Proportion treated by disease stage: >500 = 35%, 200–500 = 52%, <200 = 72%, AIDS = 90%. Changes in risk behaviour were linearly associated with increases in incidence, while decreases in infectivity were non-linearly associated with decreases in incidence. Decreases in infectivity of 2-, 5- and 10-fold would be counterbalanced (in terms of incidence) by increases in risk behaviour of 40, 60 and 70%, respectively.
Velasco-Hernandez et al 2002 [1] Dynamic, deterministic (used previous model [5]) MSM, San Francisco, US, Previous model [5] is used to derive an analytical expression for R0. Used assumptions as for the previous model. Changes in risk behaviour: 50% reduction to 100% increase (whole population). Relative fitness of resistant strains: 1% to "approximately as transmissible". Median R0 = 0.90 if risky sex decreased, 1.0 if risky sex remained stable, and 1.16 if risky sex increased. R0 decreased as ART coverage increased. The probability of epidemic eradication is high (p = 0.85) if risky sex decreases (median 25% reduction), moderate (p = 0.5) if it remains stable, and low (p = 0.13) if it increases (median 50% increase). Concluded that ART can function as an effective HIV prevention tool, even with high levels of drug resistance and risky sex, and could eradicate a high prevalence (30%) HIV epidemic.
Law et al 2002 [78] Dynamic, deterministic (extension of previous model [4]) MSM, Australia 1996 Incorporation of other sexually transmitted infections (STI), but not dynamically-assumed 100% increase in prevalence of STI among all MSM regardless of HIV status, due to increased risk behaviour as a result of ART introduction. STI infection increased HIV infectivity 3.5-fold (range: 2–5-fold). Decreases in infectivity of 2-, 5- and 10-fold would be counterbalanced (in terms of incidence) by increases in risk behaviour of 30, 50 and 65%, respectively i.e. even more modest increases than in previous publication [4]. Even small increases in STI as a result of increased risk behaviour could have an important multiplicative effect increasing HIV incidence.
Johnson & Dorrington 2002 [79] ASSA2000 Interventions Model (dynamic, deterministic spreadsheet model) South Africa Stages of infection: stages I to IV of the WHO clinical staging system, with decline in sexual activity at advanced stages. Includes voluntary counselling and testing (VCT) with a corresponding (though transient) decrease in risk behaviour for both infected and uninfected individuals. Reduction in viral load due to ART: 1.76 log10. Reduction in infectivity: 67% per log reduction in viral load. 4 sexual activity classes. Only AIDS patients qualify for treatment. Model assumes a phased roll-out achieving 90% coverage by 2006. First 6 months of ART: death rate = 8.2%, discontinuation rate = 9.1%. Thereafter: death rate = 5.8%/year, discontinuation rate = 5.8%. Resistance not explicitly modelled. ART provision is highly effective at preventing new infections, through reduced infectivity and assumed impact of VCT, and the high coverage level. ART plus VCT reduces incidence of AIDS, but because of increasing numbers starting treatment, the overall number of AIDS cases increases to a peak in 2015. Approximately one million deaths would be averted between 2001 and 2015 if ART is added to a set of AIDS prevention initiatives.
Nagelkerke et al 2002 [80] Dynamic, deterministic Botswana and India Stratified by gender and 2 sexual activity groups (higher group represents CSWs and their clients). No changes in sexual behaviour due to ART, but "effective counselling" of those on ART could decrease infectivity of those developing drug resistance by 50%. ART reduces infectivity to zero. Rate of acquired resistance: 25% per year (range: 5–25%). Transmitted resistance possible (resistant strains appear to be as transmissible as wild type). Rate of treatment uptake: all those infected are recruited at rate 50% per year. No stratification by stage of infection. Compared impacts of an ART programmes to other HIV interventions. Concluded that after transient success, ART would be ineffective within 30 years due to widespread drug resistance. Assumes high treatment uptake rates and pessimistic assumptions regarding transmission of drug resistant strains.
Gray et al 2003 [2] Dynamic, stochastic Rakai, Uganda, 2000–2020 Assumes ART reduces HIV log viral load by 27.0–42.5%, representing decreases in log viral load from 5.32 to 3.06 log10 copies/ml [49], and 5.23 to 3.82 log10 copies/ml [48]. These generate an average decrease in infectivity of 95.7% (0.0023 to 0.0001 per coital act) and 43.5% (0.0023 to 0.0013) respectively. Sexual activity decreases with increasing viral load. Behavioural disinhibition: increased risk by 50–100% (among those on ART only). Treatment uptake: scenario 1: all with viral load >55,000 copies/ml; scenario 2: all subjects, irrespective of viral load (20% of infected persons in Rakai had viral loads >55,000 copies/ml [81]). Range of treatment coverage: 0–100%. Concluded that ART alone cannot control mature HIV epidemics such as that in Rakai. Behavioural disinhibition would counter decreases in HIV infectivity due to ART.
Xiridou et al 2003 [82] Dynamic, deterministic Young MSM, Amsterdam, The Netherlands Steady and casual partnerships. Stages of infection: primary, incubation, AIDS. Sexual activity assumed to cease after development of AIDS. 42% subjects in incubation stage are diagnosed. Diagnosis during incubation results in a 25% (0–50%) reduction in risky behaviour. ART reduces infectivity by 74.5%. Increased life expectancy (before development of AIDS) due to ART: 9.5 years. A 75–99% reduction in infectivity due to ART will be counterbalanced by increases of 50% (range: 30–80%) in risky behaviour with steady partners, but not by increases of up to 100% with casual partners. Increasing HIV testing from 42% to 80% and ART coverage from 70% to 85%, would mean even a 100% increase in risk taking with steady partners would not outweigh the effect of ART on HIV incidence.
Xiridou et al 2004 [3] Dynamic, deterministic (extension of previous model [82]) Young MSM, Amsterdam, The Netherlands Extension of original model [3] to allow for initiation of ART during primary infection. Proportion of men diagnosed during incubation and successfully treated: 60–80%. Proportion initiating ART during primary infection: 1–10%. ART reduces infectivity during incubation by 50–99%. Population-level increase in risk behaviour for both steady and casual partnerships: 0–100%. Mean incubation time to AIDS for initiating ART during incubation: 15–30 years. Decreases in infectivity and increases in life expectancy for those initiating ART during primary infection were forced to be larger than for those initiating ART during incubation. Investigates the role of primary infection in HIV transmission. Estimates that among all new infections only 11% occur during primary infection. The effect of ART during primary infection on transmission is therefore limited. However, in a community with higher risk behaviour among casual partnerships, the fraction of transmission attributed to primary infection increases to 25%.
Clements et al 2004 [38] Dynamic, deterministic (based on previous models [4, 78]) MSM, Australia 1995–2006 Assumed a 10% annual increase in population-level risk behaviour from 1996. A stable proportion receive ART from 1998, which then declines from 70% in 2001 to a median of 50% of diagnosed on ART by 2006. HIV incidence was predicted to have declined during 1996–1998 due to ART, with a slow increase 1998–2001 due to increased risk behaviour while ART usage remained fairly stable. From 2001, a continued increase in risk behaviour coupled with a moderate decline in ART use would lead to a 50% increase in incidence by 2006.
Auvert et al 2004 [26] Linear Township near Johannesburg South Africa, 2002 Used results from cross-sectional study. Under WHO guidelines, all with CD4 counts <200 initiated treatment. Under USDHHS guidelines, all with CD4 counts <350 or viral load >55,000 copies/mL initiated treatment. Reduction in infectivity due to ART calculated using infectivity estimates by viral load category used by Gray et al [2] and comparing change in distribution of viral load in the community with and without ART. Investigated short term impact of ART on incidence. The proportion of infected subjects eligible for ART was 9.5% (95% CI 6.1–14.9%) under WHO guidelines and 56.3% (95% CI 49.1–63.2%) under USDHHS guidelines. The population impact of ART on HIV transmission is small (reduction in annual risk of transmission 11.9% (95% CI 7.1–17.0%)) under WHO guidelines, but higher under USDHHS guidelines (71.8% (95% CI 64.5–77.5%))
Boily et al 2004 [30] Dynamic, deterministic MSM population STI (gonorrhoea) increasing HIV infectiousness is modelled dynamically. Stratified into 6 sexual activity groups with proportionate mixing. Two stages of HIV infection: incubation and AIDS. AIDS patients treated with ART resume the sexual activity of asymptomatic individuals within their activity class. ART reduces infectivity by 25% (pessimistic), 50–90% (moderate), 99% (optimistic). Treatment uptake rates: 10–90% per year, for AIDS patients only, or for all infected subjects Withdrawal rate (due to treatment failure, resistance and toxicity): 0–50% per year. Zero to 55% new bacterial STI could be attributed to widescale ART use, due to more modest increases in risky behaviour (0–25%) at the population level. These increases have a negative impact on HIV if coverage is too low. Increasing ART coverage helps to prevent more HIV infections despite larger increases in risk behaviour and STI that is predicted to ensue. No individual-level increase in risk behaviour; population-level increases in risk behaviour over time are due to ART slowing the depletion of high-risk infected individuals, so these populations are replenished.
Salomon et al 2005 [22] The Goals model (linear spreadsheet model) (based on previous models [20, 21]) Sub-Saharan Africa, calibrated to 3 regions: East, West/Central and Southern, up to 2020. Goals model adjusts UNAIDS/WHO EPP (epidemic projection package) and Spectrum model incidence and prevalence estimates. 5 different risk groups (single and married men and women, and CSWs). Median increase in life expectancy due to ART: 3 years. No drug resistance. Includes STI transmission. 3 stages of infection: primary, incubation and symptomatic. ART reduces infectivity by 99% (optimistic) or 66% (pessimistic). Number of partners reduced by 50% plus 2 times higher condom use (optimistic) or no change (pessimistic) for those treated. Risk behaviour of the general population does not change (optimistic), or condom use declines by 10% (pessimistic). Treatment uptake: 50% ART coverage (of those in need) by 2005, increasing to and remaining at 80% from 2010–2020. This "treatment-centred" response, where little prevention activity occurs, was compared to a "prevention-centred" response where no ART scale-up occurred, and a "combined response", with optimistic and pessimistic assumptions of the effect of ART on prevention efforts being investigated. Explored the potential impact of ART in the context of a broader strategy for HIV/AIDS control, comparing deaths and new infections averted to baseline projections without interventions. A prevention-centred strategy provides greater reductions in incidence and mortality reductions similar to those of treatment-centred strategies by 2020, but more modest mortality benefits over the next 5–10 years. If treatment scale-up leads to reduced effectiveness of prevention efforts, benefits (in terms of infections and HIV/AIDS deaths averted) are considerably smaller than for initiatives which complement each other. The number receiving ART in 2020 ranges from 9.2 million in a pessimistic treatment-only scenario, to 4.2 million in a combined response scenario with positive treatment-prevention synergies.
Wilson & Blower 2005 [19] Spatial model KwaZulu-Natal, South Africa Incorporates heterogeneity in treatment accessibility with distance to health care facilities, and heterogeneous distribution of people infected with HIV. Determining the optimal ART allocation strategy among health care facilities, aiming to maximise equity. Authors' strategy gave more equal access to ART than allocating therapy to the state capital only, or equal allocation to all health care facilities.
In an investigation into the impact of an expanded response (incorporating prevention interventions and care and support activities) on the HIV/AIDS pandemic, Stover et al 2002 did not include the effect of ART because, "there is little empirical data available on the magnitude of the preventive effect of treatment (reduced viral load and hence infectiousness) and care" [20]. However in a later publication, the authors investigated the effects of combining treatment with effective prevention efforts, using the same model (the Goals model [21]), calibrated to sub-Saharan Africa [22]. The Goals model is a Microsoft Excel™ spreadsheet model using linear equations, designed to improve resource allocation for national HIV/AIDS programmes. It feeds into the dynamic epidemic projection package (EPP) and Spectrum, used by the UNAIDS/WHO to produce national HIV/AIDS estimates [23,24], to predict the impact of an intervention. The authors concluded that a prevention-centred strategy provides greater reductions in incidence, but more modest mortality benefits, than treatment-centred scenarios. A combined approach would yield further benefits, but focusing on treatment at the expense of prevention could diminish this effect.
Auvert et al 2004 used a linear model to estimate the proportion of the South African population requiring ART under the then current WHO guidelines (treating all individuals with a CD4 cell count less than 200 cells/mm3 [25]) and to predict the impact of ART on the short term spread of HIV in this setting [26].
Such linear models have generally been used to inform policy makers on issues such as resource allocation, and typically involve only short-term predictions of the effect of ART for health care providers, as estimated by cost-effectiveness analysis [16,18]. The models are relatively straightforward in that they look at the health states of individuals, associated treatments and events that individuals experience, but fail to take account of the non-linear feedback process underlying infectious disease epidemics. Linear models are limited by the accuracy of estimates of HIV incidence used to parameterise the models, which is all the more important because their predictions and conclusions are usually more quantitative in nature than those provided by dynamic models, which have tended to be used to give more qualitative insight. Models incorporating HIV transmission dynamics typically investigate the impact of ART over a longer time frame and are used to address more general questions surrounding ART use, such as whether the benefits of ART provision outweigh the problems and risks, and which approaches to ART provision are most effective. Both types of model are required, to inform policy makers in resource-poor settings about the costs of ART provision in the short term (Wood et al [16], for example), as well as to predict the likely impact of scaling up ART use.
To date, policies designed to ameliorate the HIV/AIDS epidemic in Africa have been heavily based on policies from industrialised countries [27]. However, the epidemiological and economic contexts are so different that there is an urgent requirement to assess whether existing policy options and targets are optimal for resource-poor settings.
Dynamic model structures
Most dynamic models of HIV transmission investigating the impact of ART are deterministic, with a frequency-dependent (density-independent) transmission term. This means that the rate of (sexual) contact between one individual and others within a population does not depend on the density of the population, as it would, for example, in the case of contacts for air-borne infection transmission. HIV transmission models often incorporate relatively complex patterns of sexual behaviour, with model populations stratified into sexual activity groups by rate of partner change, and assuming different degrees of mixing between groups. However, to date most models specifically designed to examine ART impact have assumed homogeneous risk behaviour (although some of these models have investigated changes in risk behaviour of the general population as a result of ART introduction and/or a change upon diagnosis of HIV [4,5]). More realistic incorporation of sexual behaviour is likely to improve the ability of models to capture the observed timescale of African HIV epidemics, namely steady state being reached over decades rather than centuries. Figure 1 shows projections from a homogeneous sexual activity model, illustrating how, with a homogeneous population, realistic prevalence levels (representing epidemics in sub-Saharan Africa) can only be reached over unrealistic timescales (a full description of the model is provided in the Endnote). However, such homogeneous models can simulate HIV epidemics over realistic timescales if they are assumed to represent the 'at-risk proportion' of the total population only. This means that the population is crudely divided into two groups; one group practices no risky behaviour at all, whereas the other has a relatively high rate of (unprotected) sexual partner change. This structure produces an epidemic curve over a realistic timeframe (decades rather than centuries), without producing unreasonably high prevalence levels for the entire population (at-risk and not at-risk).
Figure 1 Model predictions of the effect of ART on a mature epidemic, under various assumptions. Model simulations of the potential impact of ART on a mature epidemic, varied by treatment uptake rate, reduction in infectivity due to treatment and impact on risk behaviour at the population level (see Endnote for model description). The model used only incorporates one stage of HIV infection and so individuals initiate treatment at an earlier stage of infection than is realistic, and there is homogeneous sexual mixing. Scenario A – ART uptake = 50% per year, sexual activity post ART = unchanged (2.5 partners per year), reduction in infectivity due to ART = 50-fold. Scenario B – ART uptake = 50% per year, sexual activity post ART halves (1.25 partners per year), reduction in infectivity due to ART = 50-fold. Scenario C – ART uptake = 50% per year, sexual activity post ART = unchanged (2.5 partners per year), reduction in infectivity due to ART = 1000-fold. Scenario D – ART uptake = 90% per year, sexual activity post ART = unchanged (2.5 partners per year), reduction in infectivity due to ART = 1000-fold. Scenario E – ART uptake = 90% per year, sexual activity post ART = reduced by 20% (2 partners per year), reduction in infectivity due to ART = 1000-fold.
More sophisticated models incorporating sexual behaviour include partner models [28,29] and network models [30,31]. Gray et al [2] use a stochastic simulation incorporating individuals and their contacts, although some assumptions are not clear in the available publication. The need for complexity will depend on the nature of the research question [32]. For example, where changes in sexual behaviour as a result of ART are to be investigated, a more sophisticated description of sexual behaviour is required [30]. Where the effect of ART on transmission is to be investigated, a more realistic pattern of infectivity is required [2,4]. However, while increased complexity can make models more realistic, it also makes them more difficult to parameterise and it more difficult to analyse and interpret model output.
Behaviour change
The possibility of widescale use of ART leading to changes in patterns of risk behaviour, particularly a disinhibition effect, has been of considerable concern. There are competing possible effects; at the individual level, treated patients may increase the frequency of sexual activity due to the severity of their symptoms decreasing, but may receive effective prevention counselling upon treatment initiation, which would decrease the frequency of risky activities. At the population level, in areas with substantial treatment coverage and successful treatment outcomes, there may be an increase in complacency among the general population regarding an HIV diagnosis, leading to increases in risk behaviour. Despite considerable debate [11-15], this relationship has not been convincingly demonstrated in industrialised countries where ART is readily available. A recent paper suggests that recent increases in risk-taking behaviour among MSM may be the result of non-volitional changes at the individual level over time [33]. The depletion of the pool of high-risk individuals in the pre-ART era made it more difficult for the remaining high risk-taking individuals to find partners to engage in risky sex with, but ART has facilitated the differential replenishment of this group. Therefore individuals who previously had to reduce their levels of risky sex could resume their initial high-risk behaviours.
The threat of behavioural disinhibition it is unlikely to be an immediate concern as ART is rolled out in high prevalence, resource-poor settings, where initial coverage is likely to be low and the effectiveness of ART programmes remains to be seen. Furthermore, the behavioural effects resulting from ART use in resource-poor settings are unlikely to follow patterns of industrialised countries. A person's decision to have sex, protected or unprotected, is influenced by a different set of considerations in resource-poor settings than those common in industrialised countries. Key is an individual's ability to negotiate her or his own sexual activity – as defined by fear of stigma, financial need, or the status of women within society. An individual is also less likely to be aware of his or her serostatus, due to lack of testing facilities and/or fears regarding a positive result. The provision of treatment may increase interest in voluntary counselling and testing (VCT) services, which may in turn lead to a decrease in frequency of risk behaviour by those infected. In Cote d'Ivoire for example, individuals reported low sexual activity following an HIV diagnosis, and this was not increased by the offer of ART [34]. Despite the inaccuracies of sexual behaviour data, these results are encouraging.
Given that it is difficult to predict how individuals might change their sexual behaviour as a result of ART introduction in different regions, models are faced with either estimating behavioural parameters from epidemiological data, or exploring pessimistic and optimistic scenarios using parameter values assumed to be at the ends of the spectrum of possible outcomes. Law et al 2001 modelled the effect of ART on the HIV epidemic in Australia in 1996 among the homosexual population [4] and predicted the outcome of the competing effects of increased life expectancy, decreased infectiousness and increases in unsafe sex of uninfected MSM on HIV incidence. Their assumption of a range of no change to a doubling of risky sex was essentially arbitrary, but demonstrated that increases in sexual behaviour (and life expectancy) could negate the beneficial impact of decreased infectiousness on incidence. Blower et al 2000 produced similar results for the homosexual population in San Francisco [5], again using the range of no change to a doubling in sexual risk-taking.
Velasco-Hernandez et al have investigated the conditions under which ART in HIV infected individuals may drive an epidemic to extinction [1]. As can be shown by the model output in Figure 1, for ART to eliminate HIV, an extensive reduction in risk activity at the population level, accompanying ART use (such as a 50% reduction in the partner acquisition rate) is required, together with high levels of treatment uptake and large decreases in infectiousness induced by ART. As behaviour change is notoriously difficult to generate and initial coverage rates for ART in resource-poor settings are likely to be low, this optimistic scenario is highly unlikely.
The early impact of widescale ART use in resource-poor settings where HIV prevalence is currently high will probably not involve substantial population-level increases in risky behaviour. The effectiveness of local ART programmes will likely have to be demonstrated across a broad swath of the population before the perceived threat of AIDS as a disease declines. In lower prevalence regions where high coverage rates are feasible, such changes may occur. Careful monitoring of potential changes in risk behaviour would be very useful, if feasible. Any model designed to explore the impact of sexual behaviour change in resource-poor settings, be it an increase or a decrease, should explicitly model HIV diagnosis separately from treatment initiation, as shown by Law et al [4], because 1) it is knowledge of HIV status and the associated counselling that may change behaviour, 2) the advent of therapy in the sick may change their desire and/or ability for sexual functioning and 3) the attitude of those who know they are infected with HIV may change between not being treated, where they perceive a risk of transmitting to partners, to being treated, where the magnitude of risk may be perceived as smaller. This is one area where the introduction of ART could be used for prevention as well as treatment, through facilitating VCT.
Stage of HIV infection
Some researchers believe that ART could be used as a direct prevention tool due to its effect on viral load leading to a decrease in infectivity and therefore incidence [26,35,36]. However, the competing effects of increasing prevalence due to the effect of ART on life expectancy and potential behavioural disinhibition would make this a risky strategy. Furthermore, models that predict dramatic reductions in incidence due to ART have used unrealistic treatment uptake rates. As described, some have argued that even high prevalence (30%) epidemics can be driven to extinction by ART, when assuming a treatment coverage rate of 50% to 90% [1,5]. The 50% level was estimated from data collected in a telephone sample interview of 462 MSM from four US cities conducted between November 1996 and February 1998 [37]. This was when HAART was in its infancy and treatment was initiated in a large proportion of HIV positive individuals, regardless of infection stage or CD4 count, because a "hit hard, hit early" consensus existed for patient management. Furthermore, the study only included self-identified, HIV-positive MSM, and so individuals unaware of or reluctant to admit their serostatus would have been missed. It is now more common to initiate treatment at a later stage of infection, due to side-effects and the risk of evolution of drug resistance. The proportion of HIV-infected people currently being treated, even in industrialised countries, is likely to be substantially below that required for any prospect of disease elimination.
More realistic patterns of ART use are incorporated in the models of Law et al [4,38] and Gray et al [2], where the proportion of individuals treated increases with severity of HIV disease as determined by CD4 count [4] or plasma viral load [2]. By explicitly modelling changes in infectiousness and sexual activity over time, it has been shown that ART alone cannot be relied upon as a sole prevention tool.
Gray et al [2] and Nagelkerke et al [39] explicitly modelled the impact of ART in resource-poor settings (Uganda, and Botswana and India, respectively). Nagelkerke et al 2002 assumed that those receiving ART and infected with drug-sensitive virus had zero infectivity, which does not reflect the true situation, despite viral load being substantially reduced [39]. Assumed rates of resistance evolution seem optimistically low for ART use in resource-poor settings, only being varied between 5% and 25% of those on ART failing treatment per year, whereas rates as high as 60% have been predicted by others [40-42]. Despite this the model predicted that after transient success, ART would be rendered ineffective within 30 years due to wide-scale emergence of drug resistance, based on resistant virus being as transmissible as sensitive virus.
Gray et al's conclusions were relatively pessimistic [2], contrasting with Blower et al [5]. Gray et al concluded that ART alone cannot control mature HIV epidemics such as that in Rakai, Uganda. This conclusion concurs with Garnett et al 2002 [43], who believe that ART cannot make an impact on a mature epidemic unless treatment is initiated with high coverage and earlier in infection (i.e. with higher CD4 cell counts) than is currently recommended in treatment guidelines. Such early treatment is unfeasible financially and unwarranted clinically, since it would lead to earlier evolution of resistance and treatment failure, leaving individuals running out of treatment options, perhaps even before the onset of AIDS.
The dependence of the epidemiological impact of ART use on the timing of treatment initiation is worth considering in more detail. The progress of HIV infection to AIDS can broadly be divided into four stages: primary infection, incubation, the period preceding AIDS ("pre-AIDS") and AIDS. While there is much between- and within-patient variation, on average, infectiousness is highest during primary infection, pre-AIDS and AIDS. Some experts believe that primary infection carries the highest risk of transmission, because it is associated with high plasma HIV RNA levels and continued sexual activity [44]. However, while some studies are aiming to evaluate the effect of treating individuals in primary infection [3], the vast majority of HIV infections are not diagnosed until well into the incubation period. If primary infection is defined as the period before detectable antibodies against the virus emerge, then testing can only identify those who have completed the primary stage. However, if primary infection is used to describe the high initial viraemia then infection could be diagnosed before this has ended. Treatment could not start earlier than the incubation stage which follows primary infection except in rare circumstances where exposure is known to have occurred. In resource-poor settings, diagnosis is frequently at a very late stage of infection [45,46], partly because of the non-specific nature of symptoms and the difficulty in accessing healthcare. Therefore, initiating treatment at diagnosis or when CD4 counts descend to a certain benchmark, such as 350 or 200 cells/mm3 (as recommended by current guidelines [25,47]), will mean that the highly infectious period of primary infection and the long period of incubation escape the controlling effects of treatment.
In models examining the impact of ART on HIV incidence, inclusion of the variation in infectiousness as a function of infection stage is crucial for producing realistic predictions. As ART can only be initiated upon HIV diagnosis, it will have no effect on transmission from most individuals undergoing primary infection, when risk of transmission is high. By the time an individual has developed AIDS, their sexual activity will have decreased, and so this group of infected individuals will not contribute as much to HIV transmission as the duration of this phase would suggest. Figure 2 shows runs from a four-stage HIV infection model, with various treatment coverage scenarios, determined by stage of infection. Treatment is introduced into a population with a mature HIV epidemic and a high basic reproductive number for the at-risk fraction of the population (R0~5), so it is not surprising that even aggressive implementation of ART to individuals, regardless of stage of infection, cannot lead to elimination. Figure 2 illustrates that ART under more realistic assumptions regarding treatment delivery, in terms of treatment initiation, will have far less impact on incidence. In this model, there is a single treatment regimen and high, but plausible, rates of drug resistance evolution (30% per year), meaning that the effects on transmission are short-lived, coinciding with the effectiveness of the regimen. This illustrates the urgent need for cheap and reliable second-line treatment options to be available for ART roll-out.
Figure 2 Predictions of the impact of ART by stage of infection at which treatment is initiated. Predictions of the impact of the introduction of ART in terms of HIV incidence, by stage of infection at which treatment is initiated (for a brief description of the four stage infection model used, see Endnote). Scenario A – No treatment. Scenario B – ART uptake: AIDS patients only (after a mean of 1 month). Scenario C – ART uptake: AIDS patients (after mean 1 month) and pre-AIDS (after mean 6 months). Scenario D – ART uptake: AIDS patients (after mean 1 month) and pre-AIDS (after mean 6 months) and incubation stage (after mean 4 years). Scenario E – ART uptake: all four stages, after mean 1 month.
Despite our view that Blower et al are over-optimistic [5], Gray et al's assumptions of the effects of ART may similarly be over-pessimistic [2]. The authors assume that ART leads to an average proportional reduction in HIV log viral load of between 26.8% and 43.6%, based on data from the Women's Interagency HIV Study (WIHS) [48] and the John Hopkins Clinic [49] respectively. However, other studies distinguish between patients who respond to a regimen (who typically experience reduction in viral load to undetectable levels (<50 copies/ml)), those who do not respond and those who subsequently experience treatment failure (viral rebound). With these distinctions, an individual responding successfully to ART will have a far greater reduction in viral load than Gray et al assume. Furthermore, the proportion reduction was the value recorded one month and three months after treatment initiation for the WIHS and John Hopkins Clinic patients, respectively. It can take much longer than this for complete reduction of viral load, often to undetectable limits [50] (models generally do not explicitly account for a delay between treatment initiation and effect, but it can be assumed that this is implicitly accounted for in the treatment uptake rate). Gray et al also included the possibility of behavioural disinhibition; the average number of partners for those on treatment was increased by 50% or 100%. Again, the values appear pessimistic and were essentially chosen arbitrarily, probably in order to complement other models [4,5].
Emergence of ART drug resistance
Many models of ART have concentrated on predicting the emergence and spread of ART drug resistance, which has been of concern [51-54]. Once again it is very difficult to make such predictions, as the spread of drug-resistant virus is highly dependent on the replicative fitness of the resistant strains that evolve and their ability to superinfect individuals infected with wild-type strains (i.e. to co-infect someone already infected with wild-type virus, and successfully replicate). Superinfection is perhaps only likely in the successfully treated individual, where suppression of viral load allows the target cell population to recover, hence increasing the chance of successful replication and establishment of a new strain. In the untreated individual, it is unlikely that low frequency resistant virus, typically less fecund than wild type in this environment, would be able to compete against the established viral population sufficiently successfully to allow long-term persistence of the invading strain.
In a context where ART use is common in core groups, the possibility of superinfection of those on ART means that the likely maximum rate of spread of resistance epidemics may be similar to the speed of the initial HIV epidemic. HIV co-infection with different wild-type viruses [55,56], and by wild-type strains re-infecting patients harbouring drug-resistant viruses after a short period of treatment interruption [57,58], have both been documented. Chakraborty et al postulate that it is possible for patients infected with wild type HIV-1 isolates and under successful ART to become exposed to drug-resistant strains that would have significant selective advantage, leading them to outcompete the original wild-type strain and instigate treatment failure [59]. They concede that the probability of an individual undergoing successful treatment of a wild-type strain being exposed to a drug-resistant strain is low, but the large-scale roll-out of ART in high prevalence, resource-poor settings may increase this probability substantially.
The rate at which drug resistance evolves within the individual is likely to become higher in resource-poor settings than industrialised countries; even though there are reports of patient adherence being no lower than in the West [60], potential interruptions in supply due to transport problems and a lack of sophisticated laboratory monitoring systems will limit the success of any ART regimen. However, even with high levels of drug resistance evolving within the individual ("secondary resistance"), transmission of such strains ("primary resistance"), while increasing in many industrialised settings [54,61], is reported to be substantially less frequent than for wild-type HIV [61,62], because there is usually a fitness cost for mutations. Mathematical models may have the ability to predict best- and worst-case scenarios for resistance spread [39,63], but it must be conceded that the degree to which drug resistance and risky behaviour increase as ART use rolls out in Africa and other resource-poor areas cannot yet be quantified.
Blower et al 2001 predicted that acquired resistance will continue to rise, but transmitted resistance is likely to increase only gradually, with a doubling time of around four years and a predicted median of 15.6% of new HIV infections likely to be resistant to antiretroviral drugs by 2005 [63]. This conclusion was due to an assumption that of all possible ART-resistant HIV strains that could possibly evolve, none could be as transmissible as wild-type. The study also assumed that individuals infected with ART-sensitive virus undergoing treatment cannot be co-infected or superinfected by an ART-resistant strain.
Despite the conclusion that transmitted ART resistance will stabilise at low levels, the predicted range around the 15.6% value is very wide (0.05% to 73.21%) [63]. The authors argue that the higher values in the range generated from their sensitivity analysis have a very low probability. However, the choice of parameter distributions in the Monte Carlo sampling of parameter space undertaken in their study was arbitrary (in the sense of not being motivated by prior data) and entirely determines the probability of pessimistic scenarios.
The authors themselves acknowledge that they are "predicting the unpredictable" [63], but argue that their theoretical predictions [5] are in close agreement with empirical data [64]. Both display an increase in primary resistance between 1997 and 2001, but this is a short time period and the increase may reflect the expansion in use of ART over this time. Furthermore, the uncertainty interval around predictions made by the authors is large enough for a wide range of empirical data to fit the model. Blower et al acknowledge that transmission of ART resistance may vary widely by location and that frequent comparison to empirical data is necessary. However, Blower et al in 2005 recommend that large-scale surveillance for detecting transmitted resistance in Africa will be unnecessary for the next decade because transmitted drug resistance will not reach more than 5% during that time [65]. This is due to the assumption that ART use will remain at low levels, although the authors suggest that in urban locations rates of treatment may be higher. They recommend close monitoring of treated patients, but in areas where resources are constrained, this is unlikely to be practicable (WHO guidelines do not consider resistance testing, or even viral load testing, to be a priority in these regions [25,66]). We would argue that surveillance of the prevalence of drug resistance among patients is required in all locations where ART is used, and that the predictive utility of models with high degrees of uncertainty in their input parameters, and hence also their results, is limited.
In this context, it should be noted that increases in levels of acquired resistance are not inevitable – in Switzerland, where more than 80% of prescribed ART is dispensed by one of the highly experienced Swiss HIV Cohort centres, prevalence of drug-resistant HIV in newly infected individuals has been decreasing since 1996 [67]. However, we would argue that such an effect is less likely in resource-poor settings with restricted access to high-quality care and laboratory facilities and potential problems of drug sharing, black market resale of drugs and inappropriate prescribing of mono and dual therapy outside of official ART programmes [53]. The differences between ART programmes in industrialised and developing countries will be so marked that predicting programme impact and patterns of drug resistance from those in former setting is not necessarily informative.
If one relaxes the assumption that no resistant strain can exceed the transmission fitness of wild-type even in the presence of ART use, even greater variation in predicted levels of transmitted drug resistance after 10 years of ART provision is possible (Figure 3). The scenarios illustrated assume a conservative rate of 10% per year for the evolution of resistance in the treated patient and do not allow for the possible enhancement of that rate in individuals suffering viral rebound (an increase in viral load following a previous decrease due to ART) without initial resistance, but who are then maintained on the same regimen (due to a lack of virological testing). Nevertheless, the results show that if a relatively fit variant emerged, the effectiveness of current ART regimens could be compromised after a very short period. There appears to be little change in model results when superinfection alone is allowed to occur, but when heterogeneous sexual activity is incorporated, resistance transmission is predicted to emerge more rapidly. This is due to superinfection allowing resistance to be transmitted through core groups receiving ART.
Figure 3 Model predictions of transmission of ART drug resistance by relative fitness of strains. Predictions of the spread of transmitted (primary) ART resistance under various scenarios, using a simplified ART model (see Endnote). Model output is 10 years after ART introduction. ART is introduced once the epidemic has reached equilibrium. Superinfection refers to the infection with ART-resistant HIV of individuals previously infected with ART-sensitive HIV and successfully undergoing treatment. These are the only individuals without viral outgrowth, and thus will have a pool of target cells rendering them susceptible to infection. Evolution of drug resistance within an individual is at a rate of 10% per year.
These results do not suggest a likelihood of drug resistance transmission but merely demonstrate the potential effects of various scenarios. It is more for biological studies (in particular resistance testing) and within-host models of HIV infection to examine the possibility that such strains could emerge [68-70]. Even in instances where laboratory tests reveal infections with virus deemed "resistant" to more than one drug class (either phenotypically or genotypically), these infections often still respond to treatment. A multidrug-resistant HIV strain would require a large number of compensatory mutations to be of a comparable fitness to wild type strains. However, ongoing treatment pressure in the presence of viral rebound could lead to sequential mutations increasing the fitness of a resistant strain; therefore, there is an argument for close monitoring of patients and implementation of drug resistance surveillance systems as ART is rolled out in resource-poor settings.
Parameterisation
To aid the design of successful ART programmes in resource-poor settings, more information on the impact of ART provision is crucial – data on morbidity and mortality, tolerability, treatment failure and the possible emergence of drug resistant strains are very important. This information also increases the reliability of model predictions by providing more accurate ranges of parameter estimates. Population-level monitoring for changes in risk behaviour and patterns of ARV drug resistance are also required. Information on the performance of ART programmes in these settings is starting to be generated. Pilot programmes such as the Médecins Sans Frontières (MSF) initiatives (established in seven low- and middle-income countries: Malawi, Kenya, South Africa, Cameroon, Cambodia, Thailand and Guatemala) are starting to report back preliminary findings. The six-month outcomes from the MSF projects were positive [71]: the probability of survival at six months was estimated as 89.5% (95% CI 86.8–92.1), with high patient attendance and adherence rates comparable to those in industrialised countries. However, pilot programmes may not be representative of future large-scale ART roll-out, where health-care infrastructure and expertise are likely to be poorer. Other programmes, such as those of the Drug Access Initiative (DAI), formed in 1998 by UNAIDS in collaboration with the Ministries of Health of Chile, Côte d'Ivoire, Uganda and Vietnam, have been running for longer. Reports from the first two years of the initiatives in Côte d'Ivoire have been positive [72], with immunologic and virologic outcomes similar to those reported from industrialised countries. The only resource-poor country to implement ART provision on a large scale is Brazil, which has made ART available free of charge to all eligible patients since 1996, and has produced positive outcomes [73]. The experience of Brazil can give information on the impact of long-term, large-scale ART provision, but is a very different setting to sub-Saharan Africa. As with data from industrialised countries, care must be taken in interpreting and assessing the applicability of results. Modelling provides the tools to predict the consequences of possible activity; by constraining ourselves to examining only those scenarios that are supported by current, gathered data, we can be ignoring other, distinct possibilities, such as the case in which an ART-resistant strain as fit as wild-type could evolve.
Conclusion
We have argued that HIV transmission models predicting the impact of ART use should incorporate a realistic progression through stages of HIV infection in order to realistically capture the timing of treatment initiation. Further elaboration of models is required (depending on the research question being posed), in areas such as time of diagnosis, sexual behaviour and assumptions regarding drug resistance evolution and transmission. All modelling studies are eventually dependent on the availability of setting-specific surveillance and behavioural data, and collection of such data is important for all regions where large scale ART use is introduced.
More investigation is required in order to determine the effect of introducing ART on a substantial scale in resource-poor settings with different stages and magnitudes of HIV epidemic. Models addressing questions of ART implementation in such settings, utilising data from fledgling ART projects where possible, will be of great use in designing cost effective programmes.
List of abbreviations
AIDS Acquired Immunodeficiency Syndrome
ART Antiretroviral therapy
HAART Highly Active Antiretroviral Therapy
HIV Human Immunodeficiency Virus
MSM Men who have Sex with Men
PMTCT Prevention of Mother to Child Transmission
VCT Voluntary Counselling and Testing
Endnote
Model assuming one stage of HIV infection
The model assuming one-stage of HIV infection, used to produce Figure 1, is illustrated in Figure 4, with state variables, parameter symbols and model equations given below. Superinfection with an ART-resistant strain is possible for individuals undergoing successful treatment only (, S = ART-sensitive, T = treated), as these are the only individuals without viral outgrowth, and thus will have a pool of target cells rendering them susceptible to infection. Treatment failure can occur with (κ) or without (f) the evolution of drug resistance. Individuals who have developed treatment failure not accompanied by resistance (, F = treatment failure) are at increased risk of developing drug resistance (κF) because of viral replication in the presence of continued drug pressure. Figure 3 uses a version of this one-stage model, modified to incorporate heterogeneous sexual mixing, with four different sexual activity groups.
Figure 4 Model of HIV transmission and treatment, with one stage of infection only. Schematic illustration of the structure of the one stage HIV transmission model. 1° Res designates those with primary (transmitted) resistance, while 2° Res designates those with secondary (acquired) resistance. ART-Sens denotes people infected with ART-sensitive virus. For clarity, death rates are not shown.
The advantage of one-stage models of infection is that they are relatively simple and analytically tractable; that is, the relationship between each parameter and the outcome of the models, for example in terms of R0[1], can be exactly specified without recourse to simulation and sensitivity analysis. However, while a model should not incorporate complexity for its own sake, stages of HIV infection play a crucial role in the impact of ART, because treatment is only initiated at late stages of infection, and infectivity and sexual activity vary with the course of infection. Similarly, incorporating heterogeneous sexual activity within a model is more important for some research questions than for others. If we want to investigate the potential impact of a transmissible ARV-resistant HIV strain through a population, its spread would appear very different in a model of homogeneous sexual activity, compared to one with heterogeneity and various assumptions regarding mixing between activity classes, where infection would travel through core groups first before spreading into the general population.
State Variables
S = susceptible individuals
= ART-sensitive HIV infected individuals, untreated
= ART-sensitive HIV infected individuals, treated
= ART-sensitive HIV infected individuals, treated but viral rebound
= ART-sensitive HIV infected individuals, suffered viral rebound, withdrawn from treatment
= ART-resistant HIV infected individuals, treated
= ART-resistant HIV infected individuals, untreated
= ART-resistant HIV infected individuals (transmitted resistance), untreated
= ART-resistant HIV infected individuals (transmitted resistance), treated
= ART-resistant HIV infected individuals (transmitted resistance), withdrawn from treatment
λS = force of infection for ART-sensitive virus phenotype
λR = force of infection for ART-resistant virus phenotype (and force of (super)infection for ART-resistant virus phenotype, infecting an HIV wild type infected individual under treatment pressure)
Parameters
For individuals of infection status :
i refers to viral phenotype (S (ART-sensitive) or R (ART-resistant));
j refers to treatment status (U (untreated) or T (treated)).
σN0 entry into model population (rate of recruitment into sexually active class). N0 is the size of the population at time t = 0.
μ death rate due to causes other than HIV infection
excess death rate due to HIV infection for an infected individual in class
γS treatment uptake rate for individuals infected with ART-sensitive HIV
γR treatment uptake rate for individuals infected with ART-resistant HIV
f rate of treatment failure (viral rebound) without ART resistance evolution
κ rate of resistance evolution without previous viral rebound
κF rate of resistance evolution with previous viral rebound
αS rate of treatment withdrawal for individuals initially infected with ART-sensitive HIV
αR rate of treatment withdrawal for individuals initially infected with ART-resistant HIV
HIV transmission probability per partnership for an infected individual in class
c Number of sexual partnerships per year
Probability of transmitting ART-resistant HIV
Probability of transmitting ART-resistant HIV
Probability of transmitting ART-resistant HIV
Probability of transmitting ART-resistant HIV
ι Factor reduction in transmissibility for an infected individual in class
Transmission equations
The forces of infection for ART-sensitive (λS) and ART-resistant (λR) HIV are given below. They are determined by the infectiousness of individuals in each class (β) and the probability that ART-resistant rather than ART-sensitive virus is transmitted in the case of mixed infections (ω).
Model assuming four stages of HIV infection
The model used to produce Figure 2 is essentially the same as for the one stage infection model, but with infection divided into four stages (primary infection, incubation, pre-AIDS and AIDS) as shown in Figure 5. The forces of infection are determined by the infectiousness of individuals in each class, the probability that ART-resistant rather than ART-sensitive virus is transmitted in the case of mixed infections and the rates of sexual partner change, which decrease as individuals reach the final stage (AIDS). The rest of the model structure is as for the one-stage model. Full details of this model, equations and parameter estimates are not shown here but are available on request.
Figure 5 Model of HIV transmission and treatment, with four stages of infection. Schematic illustration of the structure of the stages of HIV infection for the four stage model. Individuals progress from one stage to the next exponentially, with an average duration in each stage as shown in the figure. Individuals have no HIV-related mortality during primary infection or incubation, a very slightly elevated baseline death rate during pre-AIDS and an average one year life expectancy once AIDS has developed.
Competing interests
RB was supported by an unrestricted education grant from GlaxoSmithKline.
Authors' contributions
RB drafted the manuscript and constructed the mathematical models. All authors reviewed articles and read and approved the final manuscript.
Acknowledgements
Supported by an unrestricted educational grant from GlaxoSmithKline.
==== Refs
Velasco-Hernandez JX Gershengorn HB Blower SM Could widespread use of combination antiretroviral therapy eradicate HIV epidemics? Lancet Infect Dis 2002 2 487 493 12150848 10.1016/S1473-3099(02)00346-8
Gray RH Li X Wawer MJ Gange SJ Serwadda D Sewankambo NK Moore R Wabwire-Mangen F Lutalo T Quinn TC Stochastic simulation of the impact of antiretroviral therapy and HIV vaccines on HIV transmission; Rakai, Uganda AIDS 2003 17 1941 1951 12960827 10.1097/00002030-200309050-00013
Xiridou M Geskus R De Wit J Coutinho R Kretzschmar M Primary HIV infection as a source of HIV transmission within steady and casual partnerships among homosexual men AIDS 2004 18 1311 1320 15362664 10.1097/00002030-200406180-00010
Law MG Prestage G Grulich A Van de Ven P Kippax S Modelling the effect of combination antiretroviral treatments on HIV incidence AIDS 2001 15 1287 1294 11426074 10.1097/00002030-200107060-00011
Blower SM Gershengorn HB Grant RM A tale of two futures: HIV and antiretroviral therapy in San Francisco Science 2000 287 650 654 10649998 10.1126/science.287.5453.650
Zhang L Ramratnam B Tenner-Racz K He Y Vesanen M Lewin S Talal A Racz P Perelson AS Korber BT Quantifying residual HIV-1 replication in patients receiving combination antiretroviral therapy N Engl J Med 1999 340 1605 1613 10341272 10.1056/NEJM199905273402101
Vernazza PL Gilliam BL Dyer J Fiscus SA Eron JJ Frank AC Cohen MS Quantification of HIV in semen: correlation with antiviral treatment and immune status AIDS 1997 11 987 993 9223732 10.1097/00002030-199708000-00006
Gupta P Mellors J Kingsley L Riddler S Singh MK Schreiber S Cronin M Rinaldo CR High viral load in semen of human immunodeficiency virus type 1-infected men at all stages of disease and its reduction by therapy with protease and nonnucleoside reverse transcriptase inhibitors J Virol 1997 71 6271 6275 9223532
Mocroft A Vella S Benfield TL Chiesi A Miller V Gargalianos P d'Arminio Monforte A Yust I Bruun JN Phillips AN Changing patterns of mortality across Europe in patients infected with HIV-1. EuroSIDA Study Group The Lancet 1998 352 1725 1730 9848347 10.1016/S0140-6736(98)03201-2
Weidle PJ Holmberg SD DeCock KM Changes in HIV and AIDS epidemiology from new generation antiretroviral therapy Aids 1999 13 S61 68 10885764
HIV incidence among young men who have sex with men – seven U.S. cities, 1994–2000 MMWR 2001 50 440 444 11475380
Dodds J Nardone A Mercey D Johnson A Increase in high risk sexual behaviour among homosexual men, London 1996–8: cross sectional, questionnaire study BMJ 2000 320 1510 1511 10834892 10.1136/bmj.320.7248.1510
Bouhnik AD Moatti JP Vlahov D Gallais H Dellamonica P Obadia Y Highly active antiretroviral treatment does not increase sexual risk behaviour among French HIV infected injecting drug users J Epidemiol Community Health 2002 56 349 353 11964431 10.1136/jech.56.5.349
Murphy G Charlett A Jordan LF Osner N Gill ON Parry JV HIV incidence appears constant in men who have sex with men despite widespread use of effective antiretroviral therapy AIDS 2004 18 265 272 15075544 10.1097/00002030-200401230-00016
Ostrow DE Fox KJ Chmiel JS Silvestre A Visscher BR Vanable PA Jacobson LP Strathdee SA Attitudes towards highly active antiretroviral therapy are associated with sexual risk taking among HIV-infected and uninfected homosexual men AIDS 2002 16 775 780 11964534 10.1097/00002030-200203290-00013
Wood E Braitstein P Montaner JS Schechter MT Tyndall MW O'Shaughnessy MV Hogg RS Extent to which low-level use of antiretroviral treatment could curb the AIDS epidemic in sub-Saharan Africa The Lancet 2000 355 2095 2100 10902622 10.1016/S0140-6736(00)02375-8
Aalen OO Farewell VT De Angelis D Day NE Gill ON New therapy explains the fall in AIDS incidence with a substantial rise in number of persons on treatment expected AIDS 1999 13 103 108 10207551 10.1097/00002030-199901140-00014
Freedberg KA Losina E Weinstein MC Paltiel AD Cohen CJ Seage GR Craven DE Zhang H Kimmel AD Goldie SJ The cost effectiveness of combination antiretroviral therapy for HIV disease N Engl J Med 2001 344 824 831 11248160 10.1056/NEJM200103153441108
Wilson DP Blower SM Designing Equitable Antiretroviral Allocation Strategies in Resource-Constrained Countries PLoS Med 2005 2 132 141 10.1371/journal.pmed.0020050
Stover J Walker N Garnett GP Salomon JA Stanecki KA Ghys PD Grassly NC Anderson RM Schwartlander B Can we reverse the HIV/AIDS pandemic with an expanded response? The Lancet 2002 360 73 77 12114060 10.1016/S0140-6736(02)09339-X
Stover J Bollinger L Cooper-Arnold K Goals model: for estimating the effects of resource allocation decisions on the achievement of goals of the HIV/AIDS strategic plan 2001
Salomon JA Hogan DR Stover J Stanecki KA Walker N Ghys PD Schwartlander B Integrating HIV Prevention and Treatment: From Slogans to Impact PLoS Med 2005 2 e16 15647780 10.1371/journal.pmed.0020016
Improved methods and assumptions for estimation of the HIV/AIDS epidemic and its impact: Recommendations of the UNAIDS Reference Group on Estimates, Modelling and Projections Aids 2002 16 W1 14 12045507 10.1097/00002030-200206140-00024
Ghys PD Brown T Grassly NC Garnett G Stanecki KA Stover J Walker N The UNAIDS Estimation and Projection Package: a software package to estimate and project national HIV epidemics Sex Transm Infect 2004 80 i5 9 15249692 10.1136/sti.2004.010199
WHO Scaling up antiretroviral therapy in resource limited settings: Guidelines for a public health approach WHO 2002 1 115
Auvert B Males S Puren A Taljaard D Carael M Williams B Can Highly Active Antiretroviral Therapy Reduce the Spread of HIV?: A Study in a Township of South Africa J Acquir Immune Defic Syndr 2004 36 613 621 15097305
De Cock K Mbori-Ngacha D Marum E Shadow on the continent: public health and HIV/AIDS in Africa in the 21st century The Lancet 2002 360 67 72 12114058 10.1016/S0140-6736(02)09337-6
Kretzschmar M Morris M Measures of concurrency in networks and the spread of infectious disease Math Biosci 1996 133 165 195 8718707 10.1016/0025-5564(95)00093-3
Bauch CT A versatile ODE approximation to a network model for the spread of sexually transmitted diseases J Math Biol 2002 45 375 395 12424529 10.1007/s002850200153
Boily MC Bastos FI Desai K Masse B Changes in the transmission dynamics of the HIV epidemic after the wide-scale use of antiretroviral therapy could explain increases in sexually transmitted infections: results from mathematical models Sex Transm Dis 2004 31 100 113 14743073
Eames KT Keeling MJ Modeling dynamic and network heterogeneities in the spread of sexually transmitted diseases Proc Natl Acad Sci U S A 2002 99 13330 13335 12271127 10.1073/pnas.202244299
Anderson RM Garnett GP Mathematical models of the transmission and control of sexually transmitted diseases Sex Transm Dis 2000 27 636 643 11099079
Boily MC Godin G Hogben M Sherr L Bastos FI The impact of the transmission dynamics of the HIV/AIDS epidemic on sexual behaviour: a new hypothesis to explain recent increases in risk taking-behaviour among men who have sex with men Med Hypotheses 2005 65 215 226 15922091 10.1016/j.mehy.2005.03.017
Moatti JP Prudhomme J Traore DC Juillet-Amari A Akribi HA Msellati P Access to antiretroviral treatment and sexual behaviours of HIV-infected patients aware of their serostatus in Cote d'Ivoire AIDS 2003 17 S69 77 14565612
Cameron DW Can we reduce HIV transmission by providing healthcare and HIV therapy to commercial sex workers? J Int Assoc Physicians AIDS Care 1998 4 24 26 11366121
Hosseinipour M Cohen MS Vernazza PL Kashuba AD Can antiretroviral therapy be used to prevent sexual transmission of human immunodeficiency virus type 1? Clin Infect Dis 2002 34 1391 1395 11981736 10.1086/340403
Stall R Pollack L Mills TC Martin JN Osmond D Paul J Binson D Coates TJ Catania JA Use of antiretroviral therapies among HIV-infected men who have sex with men: a household-based sample of 4 major American cities Am J Public Health 2001 91 767 773 11344885
Clements MS Prestage G Grulich A Van De Ven P Kippax S Law MG Modeling Trends in HIV Incidence Among Homosexual Men in Australia 1995–2006 J Acquir Immune Defic Syndr 2004 35 401 406 15097157
Nagelkerke N Jha P de Vlas S Korenromp E Moses S Blanchard J Plummer F Modelling HIV/AIDS epidemics in Botswana and India: impact of interventions to prevent transmission Bull World Health Org 2002 80 89 96 11953786
Garnett G Bartley L Cameron D Anderson R Both a 'magic bullet' and good aim are required to link public health interests and health care need in HIV infection Nat Med 2000 6 261 262 10700225 10.1038/73104
Paredes R Mocroft A Kirk O Lazzarin A Barton S van Lunzen J Katzenstein T Antunes F Lundgren J Clotet B Predictors of virological success and ensuing failure in HIV-positive patients starting highly active antiretroviral therapy in Europe: results from the EuroSIDA study Arch Intern Med 2000 160 1123 1132 10789605 10.1001/archinte.160.8.1123
Descamps D Flandre P V C Peytavin G Meiffredy V Collin G Delaugerre C Robert-Delmas S Bazin B Aboulker J Mechanisms of virologic failure in previously untreated HIV-infected patients from a trial of induction-maintenance therapy JAMA 2000 283 205 211 10634336 10.1001/jama.283.2.205
Garnett GP Bartley L Grassly NC Anderson RM Antiretroviral therapy to treat and prevent HIV/AIDS in resource-poor settings Nat Med 2002 8 651 654 12091892 10.1038/nm0702-651
Koopman JS Jacquez JA Welch GW Simon CP Foxman B Pollock SM Barth-Jones D Adams AL Lange K The role of early HIV infection in the spread of HIV through populations J Acquir Immune Defic Syndr Hum Retrovirol 1997 14 249 258 9117458
Sungkanuparph S Vibhagool A Mootsikapun P Chetchotisakd P Tansuphaswaswadikul S Bowonwatanuwong C Opportunistic infections after the initiation of highly active antiretroviral therapy in advanced AIDS patients in an area with a high prevalence of tuberculosis AIDS 2003 17 2129 2131 14502020 10.1097/00002030-200309260-00018
Laurent C Diakhate N Gueye NF Toure MA Sow PS Faye MA Gueye M Laniece I Toure Kane C Liegeois F The Senegalese government's highly active antiretroviral therapy initiative: an 18-month follow-up study AIDS 2002 16 1363 1370 12131213 10.1097/00002030-200207050-00008
USDHHS Guidelines for the use of ARV agents in HIV-infected adults and adolescents. US Department of Health and Human Services 2002 Washington, DC
Barkan SE Melnick SL Preston-Martin S Weber K Kalish LA Miotti P Young M Greenblatt R Sacks H Feldman J The Women's Interagency HIV Study. WIHS Collaborative Study Group Epidemiology 1998 9 117 125 9504278 10.1097/00001648-199803000-00001
Moore RD Understanding the clinical and economic outcomes of HIV therapy: the Johns Hopkins HIV clinical practice cohort J Acquir Immune Defic Syndr Hum Retrovirol 1998 17 S38 41 9586651
Ledergerber B Egger M Opravil M Telenti A Hirschel B Battegay M Vernazza P Sudre P Flepp M Furrer H Clinical progression and virological failure on highly active antiretroviral therapy in HIV-1 patients: a prospective cohort study. Swiss HIV Cohort Study The Lancet 1999 353 863 868 10093977 10.1016/S0140-6736(99)01122-8
Brenner B Turner D Wainberg M HIV-1 drug resistance: can we overcome? Expert Opin Biol Ther 2002 2 751 761 12387674 10.1517/14712598.2.7.751
Cohen OJ Fauci AS Transmission of drug-resistant strains of HIV-1: unfortunate, but inevitable The Lancet 1999 354 697 698 10475176 10.1016/S0140-6736(99)90106-X
Harries AD Nyangulu DS Hargreaves NJ Kaluwa O Salaniponi FM Preventing antiretroviral anarchy in sub-Saharan Africa The Lancet 2001 358 410 414 11502341 10.1016/S0140-6736(01)05551-9
Little S Holte S Routy J Daar E Markowitz M Collier A Koup R Mellors J Connick E Conway B Antiretroviral-drug resistance among patients recently infected with HIV N Engl J Med 2002 347 385 394 12167680 10.1056/NEJMoa013552
Ramos A Hu DJ Nguyen L Phan KO Vanichseni S Promadej N Choopanya K Callahan M Young NL McNicholl J Intersubtype human immunodeficiency virus type 1 superinfection following seroconversion to primary infection in two injection drug users J Virol 2002 76 7444 7452 12097556 10.1128/JVI.76.15.7444-7452.2002
Jost S Bernard MC Kaiser L Yerly S Hirschel B Samri A Autran B Goh LE Perrin L A patient with HIV-1 superinfection N Engl J Med 2002 347 731 736 12213944 10.1056/NEJMoa020263
Altfeld M Allen TM Yu XG Johnston MN Agrawal D Korber BT Montefiori DC O'Connor DH Davis BT Lee PK HIV-1 superinfection despite broad CD8+ T-cell responses containing replication of the primary virus Nature 2002 420 434 439 12459786 10.1038/nature01200
Koelsch KK Smith DM Little SJ Ignacio CC Macaranas TR Brown AJ Petropoulos CJ Richman DD Wong JK Clade B HIV-1 superinfection with wild-type virus after primary infection with drug-resistant clade B virus AIDS 2003 17 F11 16 12700477 10.1097/00002030-200305020-00001
Chakraborty B Kiser P Rangel HR Weber J Mirza M Marotta ML Asaad R Rodriguez B Valdez H Lederman MM Can HIV-1 superinfection compromise antiretroviral therapy? AIDS 2004 18 132 134 15090842
Laniece I Ciss M Desclaux A Diop K Mbodj F Ndiaye B Sylla O Delaporte E Ndoye I Adherence to HAART and its principal determinants in a cohort of Senegalese adults AIDS 2003 17 S103 108 14565616
Leigh Brown AJ Frost SD Mathews WC Dawson K Hellmann NS Daar ES Richman DD Little SJ Transmission fitness of drug-resistant human immunodeficiency virus and the prevalence of resistance in the antiretroviral-treated population J Infect Dis 2003 187 683 686 12599087 10.1086/367989
Yerly S Jost S Telenti A Flepp M Kaiser L Chave JP Vernazza P Battegay M Furrer H Chanzy B Infrequent transmission of HIV-1 drug-resistant variants Antivir Ther 2004 9 375 384 15259900
Blower S Aschenbach A Gershengorn H Kahn J Predicting the unpredictable: transmission of drug-resistant HIV Nat Med 2001 7 1016 1020 11533704 10.1038/nm0901-1016
Grant R Hecht F Warmerdam M Liu L Liegler T Petropoulos C Hellmann N Chesney M Busch M Kahn J Time trends in primary HIV-1 drug resistance among recently infected persons JAMA 2002 288 181 188 12095382 10.1001/jama.288.2.181
Blower S Bodine E Kahn J McFarland W The antiretroviral rollout and drug-resistant HIV in Africa: insights from empirical data and theoretical models AIDS 2005 19 1 14 15627028 10.1097/00002030-200501030-00001
WHO The use of antiretroviral therapy: a simplified approach for resource-constrained countries 2002
Yerly S Kaiser L Race E Bru JP Clavel F Perrin L Transmission of antiretroviral-drug-resistant HIV-1 variants The Lancet 1999 354 729 733 10475184 10.1016/S0140-6736(98)12262-6
Ghani AC Ferguson NM Fraser C Donnelly CA Danner S Reiss P Lange J Goudsmit J Anderson RM De Wolf F Viral replication under combination antiretroviral therapy: a comparison of four different regimens J Acquir Immune Defic Syndr 2002 30 167 176 12045679
Vergu E Mallet A Golmard JL The role of resistance characteristics of viral strains in the prediction of the response to antiretroviral therapy in HIV infection J Acquir Immune Defic Syndr 2002 30 263 270 12131562
Ferguson NM deWolf F Ghani AC Fraser C Donnelly CA Reiss P Lange JM Danner SA Garnett GP Goudsmit J Antigen-driven CD4+ T cell and HIV-1 dynamics: residual viral replication under highly active antiretroviral therapy Proc Natl Acad Sci U S A 1999 96 15167 15172 10611356 10.1073/pnas.96.26.15167
Tassie JM Szumilin E Calmy A Goemaere E Highly active antiretroviral therapy in resource-poor settings: the experience of Medecins Sans Frontieres AIDS 2003 17 1995 1997 12960837 10.1097/00002030-200309050-00023
Diomande FV Bissagnene E Nkengasong JN Maurice C Monga B Laga M Nolan ML The most efficient use of resources to identify those in need of antiretroviral treatment in Africa: empirical data from Cote d'Ivoire's Drug Access Initiative AIDS 2003 17 S87 93 14565614
Bastos FI Kerrigan D Malta M Carneiro-da-Cunha C Strathdee SA Treatment for HIV/AIDS in Brazil: strengths, challenges, and opportunities for operations research AIDScience (Science knowledge environment) 2001 I
Zaric GS Brandeau ML Bayoumi AM Owens DK The effects of protease inhibitors on the spread of HIV and the development of drug-resistance HIV strains: a simulation study Simulation 1998 71 262 275
Wood E Schechter MT Tyndall MW Montaner JS O'Shaughnessy MV Hogg RS Antiretroviral medication use among injection drug users: two potential futures AIDS 2000 14 1229 1235 10894288 10.1097/00002030-200006160-00021
Tchetgen E Kaplan EH Friedland GH Public health consequences of screening patients for adherence to highly active antiretroviral therapy J Acquir Immune Defic Syndr 2001 26 118 129 11242178 10.1097/00042560-200102010-00003
Dangerfield BC Fang Y Roberts CA Model-based scenarios for the epidemiology of HIV/AIDS: the consequences of highly active antiretroviral therapy System Dynamics Review 2001 17 119 150 10.1002/sdr.211
Law MG Prestage G Grulich A Van de Ven P Kippax S Modelling HIV incidence in gay men: increased treatment, unsafe sex and sexually transmissible infections AIDS 2002 16 499 501 11834969 10.1097/00002030-200202150-00029
Johnson L Dorrington R The demographic and epidemiological impact of HIV/AIDS treatment and prevention programmes: an evaluation based on the ASSA2000 model Paper presented at the Demographic Association of Southern Africa Conference, Cape Town, 26–27 September 2002 2002
Nagelkerke NJ Jha P de Vlas SJ Korenromp EL Moses S Blanchard JF Plummer FA Modelling HIV/AIDS epidemics in Botswana and India: impact of interventions to prevent transmission Bull World Health Organ 2002 80 89 96 11953786
Quinn TC Wawer MJ Sewankambo N Serwadda D Li C Wabwire-Mangen F Meehan MO Lutalo T Gray RH Viral load and heterosexual transmission of human immunodeficiency virus type 1. Rakai Project Study Group N Engl J Med 2000 342 921 929 10738050 10.1056/NEJM200003303421303
Xiridou M Geskus R De Wit J Coutinho R Kretzschmar M The contribution of steady and casual partnerships to the incidence of HIV infection among homosexual men in Amsterdam AIDS 2003 17 1029 1038 12700453 10.1097/00002030-200305020-00012
|
16153307
|
PMC1242350
|
CC BY
|
2021-01-04 16:38:27
|
no
|
Emerg Themes Epidemiol. 2005 Sep 10; 2:9
|
utf-8
|
Emerg Themes Epidemiol
| 2,005 |
10.1186/1742-7622-2-9
|
oa_comm
|
==== Front
Environ HealthEnvironmental Health1476-069XBioMed Central London 1476-069X-4-171612487310.1186/1476-069X-4-17ReviewTeratogenicity of depleted uranium aerosols: A review from an epidemiological perspective Hindin Rita [email protected] Doug [email protected] Bindu [email protected] Biostatistics and Epidemiology Concentration, University of Massachusetts School of Public Health and Health Sciences, Amherst, MA, USA 010032 Department of Public Health and Family Medicine, Tufts University School of Medicine, 136 Harrison Ave., Boston, MA, USA 021113 Department of Civil and Environmental Engineering, Tufts School of Engineering, 200 College Avenue, Anderson Hall, Medford, MA, USA 021552005 26 8 2005 4 17 17 19 5 2005 26 8 2005 Copyright © 2005 Hindin et al; licensee BioMed Central Ltd.2005Hindin et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Depleted uranium is being used increasingly often as a component of munitions in military conflicts. Military personnel, civilians and the DU munitions producers are being exposed to the DU aerosols that are generated.
Methods
We reviewed toxicological data on both natural and depleted uranium. We included peer reviewed studies and gray literature on birth malformations due to natural and depleted uranium. Our approach was to assess the "weight of evidence" with respect to teratogenicity of depleted uranium.
Results
Animal studies firmly support the possibility that DU is a teratogen. While the detailed pathways by which environmental DU can be internalized and reach reproductive cells are not yet fully elucidated, again, the evidence supports plausibility. To date, human epidemiological data include case examples, disease registry records, a case-control study and prospective longitudinal studies.
Discussion
The two most significant challenges to establishing a causal pathway between (human) parental DU exposure and the birth of offspring with defects are: i) distinguishing the role of DU from that of exposure to other potential teratogens; ii) documentation on the individual level of extent of parental DU exposure. Studies that use biomarkers, none yet reported, can help address the latter challenge. Thoughtful triangulation of the results of multiple studies (epidemiological and other) of DU teratogenicity contributes to disentangling the roles of various potentially teratogenic parental exposures. This paper is just such an endeavor.
Conclusion
In aggregate the human epidemiological evidence is consistent with increased risk of birth defects in offspring of persons exposed to DU.
==== Body
Background
Depleted uranium (DU) is a man-made, radioactive, heavy metal derived from uranium ore. Naturally occurring uranium ore (rock in which the uranium concentration is approximately 1,000 or more parts per million) is mined and processed to yield a much more concentrated substance, one that is virtually pure uranium. Natural uranium exists in three isotopic forms and contains 99.274% U238, 0.72% U235, and 0.0057% U234 by weight. DU, a byproduct of uranium enrichment, has an isotopic content of 99.75% U238, 0.25% U235, and 0.005% U234. As part of the effort to secure fissile uranium, many thousand tons of DU have been generated.
The chemical and metallic properties of DU do not differ largely from natural uranium ore or uranium oxides. However, U238 has very low specific activity and particles of U238 release radiation very infrequently. DU is commonly reported to retain only 60% of the radioactivity of natural uranium. This takes into consideration only the alpha emissions, but the decay process of DU with its daughter isotopes or decay products, thorium -234 and protactinium -234 adds beta and gamma emissions onto the already existing alpha emissions of U238. Calculations show that considering these beta and gamma emissions, DU has 75% of the radioactivity found in natural uranium [1]. DU has a radioactive half-life measured in billions of years.
DU is a dense metal; its density is 1.7 times that of lead. It reacts with most non-metallic elements; it has pyrophoric properties and may spontaneously ignite at room temperature in air, oxygen and water. These unique properties make it appealing for use in many civilian and military applications. DU is used as X-ray radiation shielding in hospitals, as counter weights for rudders and flaps in commercial aircrafts, in keels of sailing yachts and as ballast in both military and non-military airplanes. DU is used by the military for the production of distinctly powerful projectiles (e.g., bullets/penetrators, missile nose cones) and also as a protective armor for tanks.
As a projectile, a DU penetrator ignites on impact under high temperature; it has a low melting point. Further, DU sharpens as it melts making it easier to pierce heavy armor. As the projectile pierces, it leaves behind its jacket dispersing DU dust into the environment during impact. The quantity of aerosol production is directly proportional to the hardness of the armor. Normally 10–35% and up to 70% of the DU is estimated to be aerosolized on impact or when DU catches fire [2]. Most of the dust particles are reported to be smaller than 5 μm in size, i.e., of a size to be inhaled or ingested by humans [3,4]. They usually remain windborne for an extended time. There is empirical documentation that DU aerosols can travel up to 26 miles and theoretical documentation that they can travel further [5]. Once deposited on the ground the aerosols settle as partially oxidized DU dust. Potential contamination of ground water is another possibility – weathering could mobilize the metal into additional media.
U238 decays primarily by alpha emission. Alpha particles rapidly lose their kinetic energy and have little penetrating power. In the decay process beta and gamma particles are emitted which are more penetrating. Alpha radiation is only hazardous when internalized in the body, but once deposited in living tissue it releases its energy in a concentrated area causing greater damage than beta or gamma radiation.
Still, large quantities of DU and/or radioactive decay products and other radioactive impurities can lead to substantial external exposure. A Geiger counter measurement by a correspondent in the recent Iraq war show that radiation emitting from a DU bullet fragment registered nearly 1000 – 1900 times the normal background radiation level. A three-foot long DU fragment from a 12 mm tank shell registered radiation 1300 times the background level. A DU tank found by the U.S Army radiological team emitted 260 – 270 millirads of radiation per hour compared to the safety limit of 100 millirads per year. A pile of jet-black dust registered a count of 9839 emissions in one minute, a level more than 300 times the average background level [6].
In the United States there are over 50 sites that have been/are engaged in developing, producing, and testing DU munitions [7]. As of 2002 it had been established that a number of other nations, including Britain, also use, produce and/or sell DU-based munitions [8]. The U.S. military deployed DU munitions for the first time during the 1991 Gulf War. DU weapons were also used in the 1994–5 war in Bosnia, the 1999 war in Kosovo, the 2002 U.S. invasion of Afghanistan and in Iraq in 2003.
Trends toward increased use of DU by industry and, more recently, in warfare suggest that there are large and growing numbers of exposed people worldwide, both at production sites and in areas where DU weapons are deployed. While there is no clear basis for estimating the number of people who have been breathing and ingesting food and water in areas contaminated with aerosolized DU particles, the ever-expanding exposure of humans and the environment to DU particles, several micrometer and smaller, mobile and inhalable, necessitates a sense of urgency to better understand this hazard.
In sum, DU bullets are made of almost pure U-238 and DU bullets and projectiles produce largely insoluble ceramic aerosols upon impact. These aerosols, largely respirable, may be a source of toxicity for those exposed. Our specific concern here is whether or not such exposure results in teratogenic outcomes. We present, however, some analysis of the toxicity of natural and non-aerosolized uranium, because the teratogenicty of soluble, natural uranium supports the plausibility of DU being a teratogen, provided it can reach the reproductive organs.
Exposure pathways
DU can enter the body as uranium metal and as uranium oxides from the oxidized DU formed after impact with hard targets and fires. Inhalation of aerosols, ingestion and exposure through contaminated wounds or embedded fragments are all pathways of internal exposure but inhalation is the main route of human exposure both in combat and non-combat situations. Once inhaled, DU particles <5 μm can lodge deep in the lung in alveoli and can be transported by macrophages to the lymph tissues. Thereupon, live tissue immediately adjacent to (or exposed to these) imbedded particles experience infrequent but high LET alpha irradiation along with the potential for chemical toxicity. Because the micro-particles of DU are much larger than individual solubilized molecules, they can create "hot spots" of localized alpha radiation. For this to be relevant to teratogenicty, however, the particles, rather than their dissolution products, would have to reach the reproductive tissue, a phenomenon for which we are unaware of supporting evidence. (Having said this, we hasten to add that the pathways by which radiation exerts its deleterious effects on living beings are by no means fully elucidated. While discussion is beyond the scope of this paper, the "second event" theory [72], and the "bystander effect" exhibited by low level radiation [73,74] are, newly, the basis for serious re-consideration of certain propositions previously much more widely held as true.)
DU is largely non-soluble in ceramic form and is essentially comprised of three primary oxide minerals, UO2, U3O8, and UO3. UO3 is moderately soluble and is converted into soluble uranyl ions, UO2++ through a stepwise reduction. The rate of oxide dissolution will vary depending on crystal structure as well as particle surface area [9]. Uranyl compounds exhibit variable solubility at different pH values, exhibiting lower solubility at circum-neutral pH and greater solubility under acidic conditions (below pH 5) [10]. A study of the dissolution properties of DU under circumstances simulating its inclusion in internal human lung secretions, demonstrated a shortest possible dissolution half-life of slightly less than 4 years [11].
To date studies of military personnel exposed to DU munitions in the 1991 Gulf War have documented the "isotopic signature" of DU in the urine of personnel 8–9 years subsequent to exposure by inhalation [12] and 7 years after exposure among those with embedded fragments [13]. These findings confirm ongoing bioavailability of soluble uranyl ions which, as the dissolution product of internal DU particles, migrate from the initial point of entry into the human body and are eventually, at least partially, excreted in urine.
The health effects related to internal exposure may result from either chemical or radiological toxicity. Solubility determines the kind of toxicity exerted by uranium. The soluble forms of uranium are more associated with toxic chemical effects while insoluble forms are associated with radiological effects. Soluble chemical forms are absorbed within days while insoluble forms generally takes months to years to be absorbed [2]. DU is organotropic and has long-term retention in its target organs, to wit the kidney and the skeletal tissue. The biological retention capability of DU in bones enhances the particulate radiation to the target organs. Though the mechanism of action of DU oxides are not clear, biodistribution studies detail DU accumulation in the bone, kidney, reproductive system, brain and lung with verified nephrotoxic, genotoxic, mutagenic and carcinogenic properties, as well as reproductive and teratogenic alterations [14]. However the subject of inquiry from the vantage of teratogenicity is the potential for DU exposure of egg and embryo and of sperm and cells involved in spermatogenesis. (New information on how radiation affects cells, as noted in the first paragraph of this section, may contribute to elucidation of how reproductive tissue could be damaged indirectly.)
Theoretical Basis for Human Teratogenicity
With a few exceptions, it is only since the late 1980s that uranium's reproductive toxicity has been studied using animal models. Most of the past 15 years of published research on the topic comes from two groups, Domingo, and others working at the University of Barcelona in Spain and McClain, Benson, Miller, Pellmar and others affiliated with the Armed Forces Radiobiology Research Institute (AFRRI) in the United States [15-17]. With at least 6 published studies on topic, Domingo et al. have demonstrated that both oral and subcutaneous administration of UO2++ to female mice engender decreased fertility, embryonic and fetal toxicity including reduced growth and malformations (cleft palate and skeletal defects) and developmental ossification variations. From their maternal animal exposure studies the members of Domingo's group concluded that it was chemical toxicity, not radiation that resulted in teratogenicity [15,18-22]. The chemical reproductive toxicity of DU could act at the molecular level (damaging DNA and RNA), at the cellular level, and/or at the organ level, affecting organs including the testes, placenta, and embryo/fetus.
Two studies of orally dosed male rats that were conducted decades earlier demonstrated substantial degeneration of testes and impact on germ cells; another more recent study provided some similar evidence [21,23,24]. Very recent research suggests that uranium mimics estrogen in mice [25].
The AFRRI studies were funded in 1994 by the military in order to investigate the toxicity of embedded DU fragments [26]. These studies, using a rat model, have demonstrated that DU pellets embedded in male rats led to elevated uranium concentrations in the testes, and that pellets embedded in females led to detectable uranium in the placenta and to "very low levels" of its accumulation in the fetus, though there was no "overt" teratology. There is preliminary evidence of delayed reproductive impact of embedded DU among female rats; the probability of decreased litter size increased in proportion to time since embedding. Several rat studies by the AFRRI group have shown that embedded DU pellets are mutagenic [16,26,27]. In their human studies, McDiarmid et al. found "subtle perturbations" in indices of reproductive health among their shrapnel-exposed human subjects [28].
A Chinese study of reproductive toxicity of enriched uranium noted damage to genetic material, dominant lethality and skeletal abnormalities in fetal rats. Chromosome aberrations in spermatogonia, DNA alterations in spermatocytes and strand breakage in sperm were specifically notified [29]. In vitro experiments documented extensive DNA damage when UO2++ was added to DNA in the presence of an electron donor. Since DNA is particularly dense in sperm-forming cells, such cells may be especially susceptible to UO2++-derived damage. In sum, aerosolized DU is a vehicle for internal delivery of a DNA-tropic substance that is both a heavy metal and an alpha particle emitter.
Chromosomal instabilities have also been documented in humans. In 1997 Zaire et al reported finding increased frequency of sister chromatid exchanges in cells of uranium miners [30]. And in 2001 McDiarmid reported a similar finding among 1991 Gulf War veterans with embedded DU shrapnel [28]. In 2003 Schroder et al. documented chromosomal instability (in the form of increased frequency of dicentric and centric ring chromosomes) among sixteen 1991 Gulf and Balkan war veterans who believe they were exposed to DU via dust inhalation [31]. Mutagenicity of uranium was also observed in residents living near uranium mines. Au et al. looked at non-smokers who resided near uranium mining or milling sites in Texas, but had not worked in the uranium industry [32,33]. They found that residents living near the uranium mining and milling sites had higher frequencies than controls of aberrant cells, chromosome deletions and chromosomal aberrations.
Consideration of evidence regarding the teratogenicity of heavy metals other than uranium is also relevant for estimation of the hazard that DU poses. First, elegant studies of other heavy metals suggest study designs that may be emulated; second, observed associations can inform and contribute to the choice of outcomes for DU studies since various heavy metals may have similar modes of action. Accumulated knowledge of heavy metal teratology is quite extensive; animal teratogenesis by a variety of heavy metals, and human teratogenesis by, at minimum, lead and mercury, are long established [34]. A 1996 study of human anencephaly vis a vis parental and in utero ambient exposure to lead, mercury and vanadium is exemplary. Levels of these three substances were measured in the brain, kidney, liver, and lung of 20 anencephalic and 20 control fetuses, all of which were conceived in an area of Venezuela where these heavy metals were (are?) being constantly released into the environment by "a petroleum empire [that] has grown indiscriminately". In sum, the researchers stated that " [m]ercury and Pb were significantly increased (p less than 0.001) in kidney and liver of anencephalic fetuses. Vanadium was detected exclusively at brain level, being significantly higher in controls (p less than 0.05).... In conclusion, Hg and Pb are toxic elements present in the Eastern coast's environment that should be seriously considered for cause/effect studies when the etiology of anencephaly in this region is considered..." [35].
Epidemiological and other population studies of DU teratogenicity
Investigation of DU teratogenicity in exposed populations is constrained by the rigor with which such populations can be accurately identified [36]. Parental exposure to a possible teratogen can be established on, at least, two different levels. Especially when exposure is from the ambient environment rather than from individual activity (e.g., consumption of a particular substance), documentation at the intra-individual level, through the use of assays measuring biomarkers (as described just above ([35]) enhances clarity. There have been no studies of DU's teratogenicity with concomitant biomarker documentation of individual DU exposure. The first report that assessed, and supported, the feasibility of use of biomarkers for measurement of DU and other environmental chemical exposures during military deployment has recently been reported [37].
On the ecological level, being at-risk for exposure requires documentation of i) the presence of the suspected teratogen in the ambient environment and ii) the location of the allegedly exposed at the time(s) of (peak) exposure. With only ecological-level information regarding exposure, there is a limit to the clarity that can be achieved regarding the biological activity, in a given individual, of the ambient toxin.
Exposure to DU aerosols, most often the "by-product" of war, usually occurs along with other possible, probable and known risky exposures. This poses a particular challenge to inference: distinguishing the role of DU, per se, and requires weighing the evidence as thoughtfully as possible. Indeed, engagement with this challenge is one of the main objectives of this paper. The most important strategy to address this type of challenge is to relate findings in various settings, especially, unusual settings, in which populations received exposure. To the extent that the occurrence of possible risk factors for teratogenesis varies between contexts, the consistency of DU exposure across those contexts increases the probability that it is the (or an) explanatory factor.
Inclusion Criteria for Review
In this article, consideration of DU's reproductive toxicity includes its mutagenic and teratogenic potential, i.e., activity as a pre-conception mutagen affecting the female or male germ line as well as conception-to-birth impact. Studies that consider paternal DU exposure are, uniformly, included. While animal studies pertaining to a variety of endpoints have been referenced above, the human epidemiological studies considered here are limited to those that focus on physical malformations. Epidemiological-type studies of congenital malformations and uranium (though not depleted uranium) exposure are included; studies of populations exposed to other heavy metals and to other sources of low-level radiation are not. (But, there is only one study of congenital malformations among residents of a uranium mining area.) The several studies of reproductive outcomes among military personnel who served in the 1991 Gulf War, though not specifically in areas of Iraq where DU munitions were employed, are included. Findings of studies of all 1991 Gulf War veterans are interpreted accordingly – in these studies, Gulf War deployment, though used as an indicator of DU exposure means only substantially increased probability of exposure compared to non-Gulf War populations.
The Studies
Socorro Case Study
The earliest DU "research" is an instance of preliminary, "shoe leather" epidemiology that was carried out by community activists in the United States [38]. Socorro County, New Mexico is a sparsely populated rural area downwind of a DU-weapons testing site, the New Mexico Institute of Mining and Technology's Terminal Effects Research and Analysis (TERA) division [39]. In a letter addressed to TERA, a community activist enumerated birth defects among infants born in Socorro County between 1979 and 1986. The writer said she was referencing cases reported in the State of New Mexico's passive birth defects registry, i.e. a system that aggregates reports of birth defects though it does not have staff who seek out cases. The writer also reported two infants with birth defects in 1985 that were known to her but not recorded in the registry. In a county with about 250 births/year the writer reported 5 infants born with hydrocephalus (though 1 of 5, she stated, was not recorded in the registry). There was nothing remarkable about the 16 other abnormalities she enumerated.
All the instances of hydrocephalus occurred between 1984 and 1986. In 1998 another community activist requested and received a count of all hydrocephalic births in New Mexico and in Socorro County for the years 1984 – 1988 from the State Department of Health. The registry report documented a total of 19 infants born with hydrocephalus in New Mexico during those 5 years; 3 of them were Socorro County residents. Socorro is less than 1% of the State's population. Though there are several reasons why these findings cannot be considered the results of a methodologically rigorous investigation, the data are provocative, are a cause for concern and should be followed up. It would also be most valuable to seek out and aggregate other "shoe leather" findings.
ABDC Case Series
Hydrocephalus is also a component of the congenital malformation syndrome Goldenhar Syndrome. The Association of Birth Defects Children (ABDC) identified Goldenhar's as apparently occurring in excess among the offspring of male 1991 Gulf War veterans [40]. ABDC generates a birth defects registry by solicitation of reports of children with birth defects from parents. With thousands of parents reporting their children's defects and possible risk factors, including 1991 Gulf War exposures, the data brought the apparent excess to light. Goldenhar Syndrome is a variable cluster of eye, ear, face and vertebral malformations, often includes hydrocephalus.
Araneta et al/Goldenhar Cohort Study
As an attempt to provide a prompt follow-up to the ABDC "alarm", Araneta et al. utilized military hospital records to compare prevalence of Goldenhar Syndrome among offspring of 1991 Gulf War veterans and offspring of a cohort of non-deployed veterans [41]. Though the recorded prevalence of Goldenhar Syndrome was three-fold higher among the 34,000+ offspring of deployed veterans (14.7/100,000) than among the 41,000+ offspring of non-deployed veterans (4.8/100,000), the differential, based on 5 and 2 cases, respectively, of the rare syndrome, was not statistically significant. The utilization of military hospital births is an additional challenge to detection of association. Only active duty military personnel and their wives deliver at military hospitals. It is possible that one or more debilitating wartime exposures led both to termination of active duty status and increased risk of siring an affected child; if so, the choice of study population leads to bias against observing an association. From an epidemiological perspective, this study was designed to test a particular hypothesis and, though sample size obviated statistical association, it affirmed a strong likelihood of increased risk.
Dr. Gunther's Reports of Field Observations
Shortly after the 1991 Gulf War a German physician working in Iraq began to publicize his observations of catastrophic and ongoing ill health and distinctive patterns of abnormality among the Iraqi population. Gunther, the president of the Austrian-based humanitarian and relief organization Yellow Cross International had spent much time in Iraq and has had an appointment as Professor of Infectious Diseases and Epidemiology at the University of Baghdad. Though he has not published systematic data, he is convinced by his clinical experience that DU munitions are the cause of much of the horrific human toll that he saw. As early as 1996 he published a book entitled "Uranium Projectiles: Severely Maimed Soldiers, Deformed Babies, Dying Children" in which he presented his impressions of post 1991 Gulf War ill health, both in words and photographs. A second, tri-lingual (German, English, French) edition was published in 2000 and again the volume is more images than text [42]. Of 28 photographs of diseased or malformed children in the 2000 edition, four pertain to the classical infectious diseases associated with poverty and poor hygiene and two to malnutrition; the captions of the photographs of four malformed children indicate hydrocephalus.
Basra, Iraq Registry Studies
There are reports on the internet of papers delivered at international conferences, held in Iraq, on DU. The reports describe work being carried out by a clinical epidemiology research team in one of the three major maternity hospitals in Basra, Iraq [43,44]. Basra, a city of some 1.6 million, is the second largest Iraqi city and is in the region that was heavily exposed to bombardment with DU munitions during the 1991 Gulf War. Since 1989 the Basra team has kept a congenital malformations registry; each newborn is assessed before discharge from hospital [45]. Both internet-published reports references data for the 1990 birth cohort as the non-exposed group. There were between 9,845 and 13,905 births included in the registry annually between 1990 and 2000. At the World Uranium Weapons Conference held in Hamburg, Germany October 16–19, 2003 a member of the Basra team reported data for total congenital malformations in 2001. Table 1 gives the number and rate of total malformations recorded in the registry by year.
Table 1 Congenital Malformations Surveillance Data from One of Basra's Three Main Maternal and Children's Hospitals, Iraq 1990 – 2001 [43, 44] (Number and Rate of All Malformations Combined)
Year No. of Births No. of congenital malformations Congenital malformation
incidence rate/1000 births
1990 12,161 37 3.04
1991 9,845 28 2.84
1992 11,800 23 1.95
1993 12,416 28 1.31
1994 12,250 36 2.93
1995 10,576 46 4.35
1996 10,470 48 4.56
1997 13,653 32 2.34
1998 10,186 79 7.76
1999 13,905 136 9.78
2000 12,560 221 17.6
2001 11,445 254 22.19
Initially reporting on only the 1990–1998 data, the authors observed "an apparent increase in the incidence rate from 1995 upwards". Therefore, " [t]o improve statistical efficiency of the data collected and overcome small numbers of cases recorded, the pattern and incidence of congenital malformations are grouped into two periods, 1991 to 1994 and 1995 to 1998."
Incidence data, by class of malformation were reported for the time periods 1990, 1991–1994, 1995–1998 and 1999–2000; the data are redacted and presented in Table 2. (In particular, rates have been calculated.) Category names and data for all malformation categories are presented below exactly as they were specified in the internet articles. Malformation categories in which cases were first documented in 1999–2000 are listed only in the Fasy report, which includes data through that later time point. Of particular note, the report of data through 1998 does not include hydrocephalus and the report of data through 2000 indicates no diagnosed cases of hydrocephalus between 1990 and 1998 at the study hospital. However, by personal communication one of the Basra researchers confirmed to the authors that there were infants diagnosed with hydrocephalus born during the years 1990 – 1998, indeed that hydrocephalus was more commonly diagnosed during the latter part of the 90's than previously [45]. No information beyond that given in the table is provided regarding the particular cases included in each grouping.
Table 2 Redacted Presentation of Congenital Malformations Surveillance Data: All Births at One of Basra's Three Main Maternal and Children's Hospitals, Iraq 1990 – 2000 (Malformation Categories Exactly as Given by Basra Clinicians) [43, 44]
Year(s) 1990 1991–94 1995–98 1999–2000
Total number of births 12,161 46,311 44,885 26,465
Class of malformation (Rate per 1000 births (No.))
All malformations combined 3.04 (37) 2.48 (115) 4.57 (205) 13.49 (357)
Cardiovascular
Congenital heart diseases .16 (2) .39 (18) .96 (43) 1.36 (36)
Central Nervous System
Anencephaly .25 (3) .30 (14) .36 (16) 1.74 (46)
Hydrocephalus see text see text see text 1.47 (39)
Meningomyelocele .70 (9) .43 (20) .94 (42) 1.13 (30)
Musculoskeletal
Achondroplasia .25 (3) .06 (3) .20 (9) -- (4)
Arthrogryposis 0 0 0 -- (1)
Phocomelia 0 --(1) .11 (5) 1.21 (32)
Multiple Congenital Malformations
Multiple Congenital Malformations .58 (7) .65 (30) 1.09 (49) 4.23 (112)
Orofacial
Cleft lip & palate -- (1) .15 (7) .20 (9) .79 (21)
Congenital
Chromosomal aberrations .16 (2) .13 (6) .33 (15) .30 (8)
Gastrointestinal
Oesophageal atresia -- (1) -- (3) -- (3) .42 (11)
Omphalocele -- (2) .11 (5) .11 (5) 0
Imperforate anus -- (1) -- (1) 0 .19 (5)
Diaphragmatic hernia -- (4) -- (1) -- (2) -- (3)
Genitourinary
Bladder extrophy -- (2) -- (1) -- (2) -- (1)
Other
Cyclopia 0 0 0 -- (2)
Icthyosis 0 .11 (5) .11 (5) .23 (6)
Diwaniah, Iraq Registry
Research by Al-Shammosy pertains exclusively to neural tube defects and describes the findings of a study that identified all year 2000 births with a neural tube defect (NTD) at the Diwaniah maternal and children's hospital [46]. Diwaniah is a city slightly north and east of Basra and, like Basra, is in the area heavily bombarded with DU munitions during the 1991 Gulf War. The overall 2000 NTD incidence rate in Diwaniah was 8.4/1,000 (73 cases of NTDs among 8,707 births). Table 3 gives the 2000 incidence rates for particular NTDs. Al-Shammosy raises concern about universal maternal exposure to DU while stating, without providing any supportive data, that he investigated other possible risk factors for NTDs but found nothing unusual. Al-Shammosy also cited a previous report on NTDs among Diwaniah newborns, 7/98–2/99. During that 8-month period there were 33 cases of NTDs among 6124 births, 5.4/1000.
Table 3 Prevalence and Type of Neural Tube Defects among Newborns: Diwaniah Maternal and Children's Hospital, Iraq 2000 [46]
Type of Neural Tube Defect Year 2000 Rate/1000 (N)
Anencephaly 3.1 (27)
Meningocele 2.4 (21)
Meningomyelocele 2.3 (20)
Encephalocele 0.6 (5)
Iraqi Congenital Abnormalities Clinic
This clinic-based study of congenital abnormalities among Iraqi stillborns and children <= 2 years of age,[47] provides the number of children or stillborns with specific abnormalities seen among 1038 cases in 1989–90 and among 945 cases in 1992–93 (pre vs. post war). The percent of patients seen whose malformation was anencephaly or hydrocephalus rose from 0.5% to 1.1% and the percent of patients seen with skeletal abnormalities rose from 2.8% to 4.6%. The abstract does not give any information on parents' residence or fathers' occupation.
Mostar Cohorts
Sumanovic-Glamuzina et al. investigated the prevalence and type of congenital malformations among infants born in the West Mostar region of Bosnia and Herzegovina during the years 1995 and 2000 because of concern that DU munitions were used during the 1991–1995 war in that region [48]. Live and stillborns were examined 0 – 3 days postpartum, though a few malformations were noted later. This study is so sorely compromised methodologically that it defies interpretation. For example, the protocol for detecting birth defects was different for the two cohorts. Also, the distinctions between the two-birth cohorts vis a vis exposure is unclear.
US Veterans Cohort Study
There are no studies of birth defects among offspring of American or allied veterans of the 1991 Gulf War which distinguish veterans by the area of Iraq in which they were stationed. Thus, all 1991 Gulf War veterans' studies, from the point of view of assessing the teratogenic capacity of DU munitions, have a "contaminated" exposed group. The war-exposed group is comprised of individuals whose DU exposure would have ranged on a continuum from none to heavy, and they cannot be sorted further.
The 2000 and 2003 studies by Araneta et al of birth defects among offspring born to 1991 Gulf War veterans have their study populations and comparison criteria rigorously defined [49,50]. In the larger, updated study data from all areas in the United States in which there were active birth defects surveillance systems operating during 1989 – 1993 were analyzed. Occurrences of individual birth defects (aggregated into 48 groupings) among offspring of male and female 1991 Gulf War veterans and 1991 non-deployed veterans were ascertained through the first year of life. Those data provide the basis for comparisons regarding birth defects among offspring of gender-specific groups of veterans, including deployed veterans conceiving children pre-war vs. post-war and veterans conceiving children post-war, deployed vs. non-deployed. Analysis relied on statistical tests to identify significantly different prevalence rates for individual groupings of birth defects among cohorts being compared. An infant with multiple anomalies was usually included in several birth defect groups. There were 308 infants conceived post-war to deployed female veterans and 4,648 infants conceived post-war to deployed male veterans. The statistically significant findings are detailed in Table 4.
Table 4 Statistically Significant Findings in the Study by Araneta et al. (2003) [50] of Birth Defects among US Veterans Deployed and Not Deployed in the 1991 Gulf War
Birth defect Comparison groups Relative risk (95% c. i., p-value)
Father was a veteran
Aortic valve stenosis Infants conceived post-war, to deployed fathers vs. non-deployed fathers 6.0 (1.2–31.0, p = 0.026)
Infants of deployed fathers, those conceived post vs. those conceived pre-war ^16.3 (0.09–294, p = 0.011)
Renal agenesis or hypoplasia Infants of deployed fathers, those conceived post vs. those conceived pre-war ^16.3 (0.09–294, p = 0.011)
Tricuspid valve insufficiency Infants conceived post-war, to deployed fathers vs. non-deployed fathers 2.7 (1.1–6.6, p = 0.039)
Mother was a veteran father may or may not have been a veteran
Hypospadias and epispadias Infants conceived post-war, to deployed fathers vs. non-deployed mothers 6.3 (1.5–26.3, p = 0.015)
^5 of 4,648 offspring conceived post-war were so affected vs. 0 of 6,863 conceived pre-war; logit estimator method used for statistical significance testing.
Study of Births to US Active Duty Military
Two earlier records based studies found no association between the (from our perspective) over-expansive exposure criterion (Yes/No deployment in the 1991 Gulf War) and occurrence of birth defects in offspring born post-War to military personnel. Unfortunately, due to design features neither study can contribute to elucidating the nature of the relationship between parental DU exposure and congenital malformations in offspring. Cowan et al. took advantage of the maternity option available to active-duty military personnel and their wives and used military hospital birth records from 1991 – 1993 to assess the occurrence of congenital malformations among offspring of veterans subsequent to deployment in the 1991 Gulf War or elsewhere [51]. Study offspring of male Gulf War veterans numbered over 30,000 and of females Gulf War veterans nearly 4,000. Limiting post-war assessment of (reproductive outcomes among) wartime military personnel to those remaining on active-duty excludes individuals for whom war-induced debility engenders early termination of service. It is surely possible that on the population level wartime DU exposure related to termination of service; but if this is the case, the exposed cohort is also depleted of those who are the hypothesized at-risk population. Cowan et al. found no statistical association between deployment status or length of deployment and all birth defects, birth defects considered frequent and severe enough to pose a public health problem, or several specific categories of those birth defects considered more severe. What remains unknown is the extent to which this finding of no association is related to the a priori exclusion, from the pool of potential study parents, of persons who'd left service.
Study of US Military Offspring, Self-report
There is one other large cohort study that compared reproductive outcomes among 1991 Gulf War veterans and non-Gulf veterans. From an epidemiological perspective, a distinctive feature of the study by Kang et al. is that data on pregnancy outcomes were acquired by self-report and thus are susceptible to recall bias [52]. Beginning in 1996 and using stratified random sampling, Kang et al. assembled two gender and unit (active/reserve/National Guard) diverse study cohorts, each comprised of 15,000 veterans. Seventy-five percent of Gulf War veterans and 65% of non-Gulf veterans completed a study questionnaire or phone interview that provided information about birth defects (and other births outcomes) among the offspring of their first post-deployment pregnancies. There were 3397 offspring of male and female Gulf War veterans and 2646 of non-Gulf War veterans. Prevalence of any birth defect, to re-iterate, by self-report, was elevated among offspring of Gulf War veterans of both genders. The prevalence remained elevated after statistical adjustment for various potential confounders as well as after exclusion of about 1/3 of all reports of defects because reviewers, blind to parents' deployment site, considered the actual (self) report suspect. The authors' classification of the 206 remaining defects was only minimally informative: Data analysis did not separate offspring of exposed fathers and exposed mothers; nor were the sub-categories particularly informative. Among the 3397 offspring of the combined male and female Gulf War veterans there were 111 "isolated anomalies", and 35 other defects grouped in 6 categories, and there were 40 "isolated anomalies" and 20 other defects among the 2646 offspring of the non-Gulf War veterans.
Mississippi National Guard Study
In response to a report in the popular press that there was a cluster of birth defects and other health problems among offspring born to members of two Mississippi National Guard units deployed in the 1991 Gulf War, Penman and Tarver conducted a highly detailed and labor-intensive assessment [53]. They were able to gather data for ninety percent of the 284 members of the two units. At the time of the investigation, the 254 contacted veterans had begotten 67 offspring. There was nothing unusual, statistically or otherwise, about the 3 major and 2 minor congenital malformations observed among the offspring. However, sixty-seven is a small sample population of births to study malformations and assess statistical significance.
British Veterans' Offspring Study
In 2004 Doyle et al. reported the results of a mail survey about post-war reproductive outcomes among 1991 British Gulf War and other veterans [54]. The survey was sent to all UK armed services personnel stationed in the Gulf between 8/90 and 6/91 and a stratum-matched sample of other active duty armed services personnel. Data gathering occurred between 1998 and 2001. The study was designed to gather information on infertility, miscarriage and birth defects, the latter among fetal deaths as well as live births, and to obtain clinical verification of self-reports whenever possible.
24,379 or 53% of the surveyed male1991 Gulf War veterans and 18,439 or 42% of the male veterans deployed elsewhere responded to the survey. Response rates among the 1200+ female personnel in Gulf War-deployed and elsewhere-deployed cohorts were somewhat higher. Because of disappointingly low response rates, an intensive tracing study was undertaken to try to assess whether non-response introduced irremediable bias; its results were not damning. Data were presented that indicated that about half of reported events could be confirmed clinically [55].
Malformations were grouped into 11 classes and 18 sub-classes according to the European Registry of Congenital Anomalies (EUROCAT) and, as well, into two additional classes, malformations of tissues derivative from the embryonic cranial-neural crest and metabolic/single gene defects. The incidence of several classes and sub-classes of malformations were more common among the offspring of Gulf War deployed veterans vs. non-deployed veterans, when the database included all self-reported malformations. Restricting analysis to the 55% of the malformed infants whose abnormalities were clinically verified yielded not only wider confidence intervals around the relative risks (as occurs because of smaller sample sizes) but also "a general shift of the point estimates towards the null" (p. 80). In this latter analysis there were no excesses of malformed infants that were statistically significant.
Canadian 1991 Gulf War Veterans' Health Study
The Goss-Gilroy Inc. consulting firm was hired by the Canadian Department of National Defense to study the health (including reproductive outcomes) of personnel who served in the1991 Gulf War. All 4,262 Canadian veterans of the 1991 Gulf War were mailed a questionnaire and 73% responded/enrolled. Sixty percent of non-deployed veterans selected as controls enrolled. Information about birth defects was sought for all offspring, i.e., those born before, during and after the 1991 war. The study found that, compared to non-deployed veterans, deployed veterans (self-) reported higher rates of congenital malformations among offspring born during each of those three time periods – suggesting the possibility of reporting bias. Confounding the challenge of interpreting the study findings is the fact that the information reported about particular birth defects among children born to deployed and non-deployed veterans was not stratified by time period (before, during, or after military service), rather all classes of defects were compared among all children of deployed and non-deployed veterans.
Australian 1991 Gulf War Veterans' Health Study
This study attempted to gather extensive data (including reproductive histories) on all Australian 1991 Gulf War veterans and on members of a comparison cohort. Just over 80% of Australian's 1876 veterans participated, a somewhat smaller percentage of controls. Reported post-War birth defects were similar among offspring of deployed and non-deployed veterans.
Kuwaiti Congenital Heart Disease Study
Based on clinical suspicion of increased anomalies, Abushaban et al. reviewed the annual frequency and type of congenital heart defects among all Kuwaiti newborns during the years 1986 – 1989 and 1992 – 2000, i.e., pre- and post- the 1991 Gulf War [56]. Abushaban et al. report that during the 1991 Gulf War Iraqi soldiers, before leaving Kuwait, set fire to an astounding 770 Kuwaiti oil fields and the environmental damage, the air, water and land pollution was huge. Also, inter alia, there was a fire at a munitions storage site that housed DU munitions.
The average annual incidence rate of congenital heart disease was 39.5/10,000 for the years 1986 – 1989 and 103.4/10,000 for the years 1992 – 2000, with a suggestion that the rate began to decline in the most recent years. For 13 of 17 sub-categories of congenital heart disease, there was a statistically significantly higher incidence rate in the latter time period.
Shiprock Uranium Mining Area Study
The one other epidemiological study that bears on the issue of DU and birth defects is the 1992 study by Shields et al. that assessed birth outcomes among Navajos working and/or residing in the Shiprock, New Mexico uranium mining area [57]. Utilizing a nested study design, cases and controls were identified among the 13,000+ consecutive Navajo births at the Shiprock Indian Health Service Hospital during the years 1964–1981; cases included infants born with congenital anomalies, developmental disorders, stillbirths and non-injury-related infant deaths. Most children born at Shiprock continued to receive medical care there. Record review identified 320 singletons with "defective congenital conditions"; matched controls were selected. Inclusion in the case-control study was limited to the 266 (83%) dyads where families of both cases and controls were found and interviewed. There were 5 dichotomous measures of uranium exposure: father employed in mining or milling uranium, father living within 0.5 miles of a mine, likewise mother, father living within 0.5 miles of a mine dump or tailings pile, likewise mother. Paternal occupational data measuring exposure to radon daughters and information about grandparents' uranium exposures were gathered in the few instances such information was accessible.
The review of birth charts and subsequent health service records revealed that 140 of the 13,329 (1.1%) children born in the Shiprock uranium mining area 1964 – 1981 were identified as having congenital malformations. The authors grouped the congenital malformations observed among their case infants into five sub-divisions: chromosomal disorders, single gene mutations, multi-factorial conditions with morphological anomalies (excluding hip), hip dysplasia and dislocation, and teratogenic effects and other outcomes of known causes. According to the categories of Shields et al. there were 97 cases with multifactorial conditions, 20 with hip defects and fewer than 10 with each of the other categories of birth defects. (Note that Shields et al. chose to separate out hip dysplasia and dislocation because of the known high prevalence of that congenital malformation among Navajo.) But with 5 sub-divisions of malformations and 5 types of exposures, data analysis through statistical testing revealed only non-statistically significant associations – stymied by small numbers and the confusion of multiple tests.
A limited re-analysis of the available birth defects data is offered in Table 5. Rather than separate statistical analysis of each outcome category by each type of exposure, the data provided for the two more populous defect categories were used to allow for summing (without weighing) the three types of paternal and two types of maternal exposure. For example, an individual father who worked in a mine, resided within a half mile of that mine and resided within a half mile of mine tailings would contribute 3 points worth of exposure. This analysis suggests that there could be associations in the data that were not discerned by the original analytic method and that reanalysis may be warranted.
Table 5 Uranium Exposure and the More Common Categories of Congenital Malformations among Navajo in the Shiprock Area, New Mexico Re-Analysis of the Data of Shields et al. 1992.
Congenital Malformation Count of Exposures
Number of dyads Fathers (max/individual = 3) Mothers (max/individual = 2)
Cases Controls Cases Controls
Multifactorial conditions with morphologic anomalies (excluding hip) 97 42 27 23 20
Hip dysplasia and dislocation 20 16 4 13 2
Other reports
There are numerous references in the news media regarding both i) an excess of birth defects and ii) the occurrence of unusual birth defects among infants born after 1991 to returning Gulf War veterans and to residents in the area of Iraq exposed to DU munitions in the war. Of greatest interest is a 1999 article in the British newspaper The Guardian "Victims of a war they never saw" [58]. The article provides quotations from interviews with several Iraqi researchers that invoke concern. Most strikingly, an Iraqi physician then working in a Basra maternity hospital with 20 – 30 deliveries daily was quoted as saying "August – we had three babies born with no head. Four had abnormally large heads. In September we had six with no heads, none with large heads and two with short limbs...." In October, one with no head four with big heads and four with deformed limbs or other types of deformities." The Western-trained physician-geneticist, author of the previously cited 1994 report on the changing pattern of birth defects observed among genetics clinic patients, pre vs. post 1991 Gulf War [47], is quoted as follows: "We're getting mothers as young as 20 giving birth to Mongol babies... My research shows that the number of children born with Down's syndrome-type defects since the war has tripled."
Possibly the most vivid and widely seen image in this country is the LIFE magazine cover photograph of the child with phocomelia born to a recently returned 1991 male Gulf War veteran [59]. There have been more recent, prominent articles in the popular press as well. Japanese peace activists have also produced a volume of searing photographs of damaged children [60]. The lay press has reported that there has been an increase of birth defects in the region in Holland where an airplane, with DU as its ballast, crashed in 1992 and in the region of Remscheid, Germany where a U.S. army plane with DU ballast crashed in 1988 [61,62].
Framework for assessment of DU teratogenesis from an epidemiological perspective
The most compelling strands of evidence regarding the possibility that DU aerosols are teratogenic are:
i) Findings of the substantial array of DU research undertaken from the vantage of radiation physics, cell biology and animal experimentation;
ii) The documentation of birth defects in southern Iraq, including comparison data from before use of DU munitions in that area.
Documentation of elevated prevalence of the birth defect hydrocephalus among infants born downwind of an American DU munitions testing site is of particular interest in that hydrocephalus has been identified in other contexts as possibly associated with DU exposure.
There are two other sources of information that, with additional detail could be much more informative:
i) The methodologically precise study by Araneta et al. [49] which was designed to assess whether all and particular congenital malformations were more common among offspring of 1991 Gulf War veterans than among other veterans did that well. If it were possible to re-analyze those data and incorporate information on theatre of service of individual veterans, that would provide a truer proxy for DU exposure, rather than the imprecise approximation of Yes/No deployment in the war.
ii) A more data oriented presentation of the basis for his clinical impressions would be most welcome from the physician president of the Yellow Cross International.
There is also a small group of other studies that are methodologically compromised; yet they are relevant in so far as they can be considered in relation to other data. Studies that relied on parental "self-report" of birth defects are given less weight because of the potential for biased recall. Low overall response rates, adds another potential source of error for telephone or mailed surveys – the possibility that sub-groups with differential risk may have had different likelihoods of responding.
A multiplicity of considerations bear upon the establishment of causality and, especially regarding birth defects, epidemiological inference relies on input from a variety of other disciplines. The five now classic epidemiological principles of inference (time order, strength, consistency and specificity of association, and coherence) frame the assessment of teratogenicity. Kline et al. describe coherence as "an ultimate criterion, one in which the observed association is weighed against all previously existing theory and knowledge" [36]. Unusual for the early phase of epidemiological investigation of potential teratogenesis, regarding DU there already exists a body of research on the molecular, cellular and animal model level that indicates plausibility. Indeed, this body of research is substantial enough that of itself it generates concern.
Regarding DU and birth defects, the ability to assess the strength, specificity and consistency of the association, and thus the overall ability to discern causality, are impacted by the absence of i) research that documents exposure on the intra-individual level and ii) research that purposefully attempts to distinguish the role of DU from co-occurring exposures. Nor is there a scale according to which the hazard of a given amount of external exposure has been calibrated. The available high quality bioassay that quantifies an individual's internal DU exposure relies on mass spectrometry of 24-hour urine samples, at a cost of about $1,000 [63]. Efforts to theoretically model dispersal of DU aerosols [3,64] or the amount of DU present in the soil and air some months after use in war [4,65,66] are very recent.
The pattern of secondary dispersal of DU particles is another outstanding question. Yet others include: What is the nature of the relationship between amount of DU inhaled or ingested and subsequent internal DU activity? How do inhalation and ingestion exposure compare? How significant is proximity at the very time that DU is burning? Is risky proximity measured in meters or kilometers? Is the relationship between proximity and hazard linear, on what scale, or is it non-linear? Are wind patterns as important as proximity? Since DU travels by wind and is respirable and ingestible long after the DU fires are out, what risk is attached to ongoing residence in or re-locating to a region where DU munitions were used in the past?
There are also particular challenges to epidemiological investigation posed by the fact that birth defects are the outcome under study. Depending on method of categorization, the prevalence of birth defect categories can be 1 in 1,000, 1 in 5,000 or less. Even with multi-thousand cohorts, unless the actual size of the association is quite large, statistical power to detect an association in a cohort study will be compromised. The implication is that finding statistical association is informative, but finding "no statistical association", absent adequate statistical power, is not definitive. Undertaking case-control studies (which would not be affected by the rarity of the study outcome in the general population) to investigate DU teratogenicity requires the designation of specific birth defects categories as the indicator of case status.
Rarity of outcome is diminished when more specific categories are aggregated into larger groupings; but if aggregation is done inappropriately, rather than aid in accurate assessment, it will add "noise" and diminish the extent to which the data can reflect a true association. Most malformations observed at birth originate in early embryonic development. Therefore, prenatal development processes would provide a meaningful basis for aggregation. Almost none of the available data regarding DU teratogenicity were gathered within such a framework. But studies that used the strategy of "lumping" data by organ system and those that gave detailed data for individual malformations could be "tweaked" for assessment from a developmental embryology perspective.
Discussion of main epidemiological findings and suggestions of next steps
The Basra registry studies [43,44] are the starting point for discussion. The data are both profound and enigmatic. The very existence of the data is remarkable; the data are testimony to the commitment of a clinical research group – to their patients and to the potential of science to promote knowledge that can benefit generations to come.
The data are enigmatic for several reasons. There are jarring differences between the reported data and what would be expected based on malformation registries in the West. The most striking are:
i) The very low incidence of malformations reported pre-war. While the data indicate dramatic increases in incidence of malformations since the 1991 war, they start from such a low level in 1990, pre-war, that they often only reach the baseline Western levels by 1999–2000. Does the low Iraqi baseline reflect a truly lesser incidence of birth defects in Iraq at that time as compared to the West – possibly a reflection of the impact of the types of pollution and stress that are part and parcel of life in the "developed world"? Prior to the 1991 Gulf War, compared to the West, the Iraq environment was surely more pristine and the daily rhythms and challenges different. Or was detection of birth defects at the Basra study hospital less than complete? And, if so, was incomplete detection systematic or random?
ii) The malformation categories in which the data are presented. The categories do not correspond in toto to groupings used in the West. In the absence of detail explaining which abnormalities are included in which categories, inference becomes more difficult.
The hydrocephalus data are an example of the confusion that can result from the lack of clarity regarding which malformations were routinely documented and how they were grouped. It is implausible that there were no infants born with hydrocephalus at the Basra study hospital during the years 1991 – 1998 as that would mean no cases of hydrocephalus among about 100,000 births. It is hoped that this, and other unusual data features, e.g., no reported cases of microcephaly, 1990 – 2000, will eventually be clarified.
iii) The wholesale lumping together of infants identified as having multiple congenital malformations. From an analytic point of view information regarding the composition of the rapidly growing group is sorely missed. This category may be the "repository" for infants with hydrocephalus and microcephaly! Such malformations, and many others, often are not solo conditions.
iv) Also missing from the report is information on pre-and perinatal exposures of affected infants.
Aware of these imperfections, this substantial database remains an important source of information. In comparison to 1990 pre-war data, the data for subsequent years suggest that, in association with DU exposure, there are increased rates of:
Neural tube defects (NTDs),
Births with multiple congenital malformations,
Congenital heart diseases,
Cleft lip and palate,
The unusual skeletal malformation phocomelia
Congenital malformations in toto.
The NTDs reported in the Basra registry are anencephaly and meningomyelocele. Between 1990 and 1999–2000 the incidence of anencephaly rose from 2.5 to 17.4 per 10,000 births and of meningomyelocele from 7.0 to 11.3 per 10,000. The relative risk of these two NTDs among births in the study hospital for the years 1991–94, 1995–98, 1999–2000, in comparison to 1990, were 0.74, 1.31, and 2.91. By 1999–2000, the combined prevalence rate of anencephaly and meningomyelocele in the study hospital had reached 29/10,000. (If meningocele is not subsumed in another of the CNS categories, it is anomalous that there were no observed cases.)
(Anencephaly is a case in point regarding the difficulties relating time trends in frequency of occurrence of particular malformations in the Basra data with population data from the West. On the one hand, the incidence of anencephaly at birth has been declining in the West, 3.5/10,000 births in 1979–1980 vs. 2.3/10,000 in 1986–87 based on annual monitoring of hundreds of thousands of births by the U.S. Birth Defects Monitoring Program [67]. Between 1990 and 1998 the incidence of anencephaly among the Basra study newborns, rose from 2.5 in 1990 to 3.0 during the years 1991 – 94 and 3.6 in 1995 – 98.)
The Diwaniah registry study [46], designed specifically to assess prevalence of NTDs in a city that was bombarded with DU, found higher prevalence rates among births in 2000: 84/10,000 for anencephaly, meningomyelocele, menigocele, and encephalocele, 54/10,000 if the affected are limited to those diagnosed with anencephaly and meningomyelocele. Like the anencephaly rates inferred from the 1999 Guardian article [57], these rates are high in comparison to population rates reported in the West.
Without elaboration, the Diwaniah report mentions that data were gathered for each NTD case on a number of other potentially confounding factors and that those factors were not found to be relevant. With no detail, the rate of all NTDs among the 6124 births during an eight-month period a year previous (7/98–2/99) was also high, 54/10,000, though not quite as high as the 2000 rate. Neither the Basra nor the Diwaniah reports defined their malformation categories; it is unclear which Diwaniah data are more appropriate for comparison with Basra data.
With little detail, and adding the complexity of invoking male-mediated teratogenicity, the citizen-run ABDC case series flagged an excess of Goldenhar Syndrome (a clustering of malformations derivative from the 1st and 2nd brachial arches of the neural crest) among offspring of 1991 male Gulf War veterans [40]. (To re-iterate, only some of the Gulf War veterans were substantially exposed to DU.) A 1997 study by Araneta et al. [41] of two 30,000+ cohorts of veterans was conducted in response to the ABDC "flag". Given the rarity of the syndrome, that the observed 3-fold relative risk associating Goldenhar's with 1991 Gulf War service was not statistically significant, is less informative than the direction of the finding – consonant with other data. It is worth a note that renal agenesis/hypoplasia, an occasional component of Goldenhar Syndrome, was observed by Araneta et al, in other research [50], to be significantly more common among offspring born post-war to American 1991 Gulf War veterans than among their pre-war offspring.
The multiple underlying etiologies of congenital hydrocephalus include neural tube dysmorphology. Hydrocephalus as part of Goldenhar Syndrome is one such manifestation. Citizen activists did identify an excess of hydrocephalus among births downwind of the Socorro, New Mexico DU munitions testing site [38]. The 1999 Guardian article quoted from above also indicated an excess of hydrocephalus among Basra infants exposed in utero to DU aerosols [58]. That perplexing aspect of the Basra registry study – that those data indicate zero cases of hydrocephalus among the 100,000 births during the years 1990 – 98 and an unusually high incidence of the malformation, by any international standard, in 1999–2000 needs to be cited. As previously noted, in a personal communication a member of the Basra team asserted that the frequency of hydrocephalus has been elevated in Basra since 1991 [45].
Multiple congenital malformations is a broad category whose prevalence rose sharply in the data of the Basra registry studies, reaching 42/10,000 by 1999–2000. The relative risk of multiple congenital malformations (no further definition) for the years 1991–94, 1995–98, 1999– 2000 in comparison to 1990, are 1.13, 1.90, and 7.35. For various reasons it would be valuable to learn more about the malformation clusters being observed. For example, this category may be the unfortunately shadowed repository of infants born with congenital hydrocephalus in the years 1990 – 1998.
In the Basra study hospital the relative risk of congenital heart diseases for the years 1991–94, 1995–98, and 1999–2000 in comparison to 1990, was 2.4, 5.8, 8.3. In 1999–2000 the reported prevalence of congenital heart diseases was 14/10,000. By any Western standard even the 1999 – 2000 Basra rate is low. While many congenital heart disorders are not diagnosed immediately at birth and that is the only time point at which the Basra infants were assessed, U.S. data also limited to diagnosis at birth documented overall cardiovascular malformation rates upwards of 35/10,000 [67]. The breadth of cardiac malformations makes it particularly unfortunate that there is no detail about the types of defects recorded in Basra. The extent to which the Basra "congenital heart disease" conditions are congruent with the American "cardiovascular malformations" is not known.
Detail about the Basra cardiac cases would facilitate comparison with data from the Kuwaiti, 1986–1989 vs. 1992–2000, congenital heart defects study and the best of the US veterans studies. The post 1991 Gulf War rates of congenital heart defects overall and of numerous specific defects are high in the Kuwaiti study compared to data from the West. Abushaban et al. [56] found an overall post-war relative risk for congenital heart defects above 2 1/2 and statistically significant differences in incidence (pre vs. post) for 13 of 17 sub-categories. In the US veterans study of Araneta et al. two (of 14) types of congenital heart defects were identified as statistically more common among offspring of 1991 male Gulf War veterans than among offspring of other male veterans [50]. It is of note that certain cardiac malformations occur as part of Goldenhar's.
The Basra registry studies also show a dramatic rise in cleft lip and palate, from one case in a cohort of 12,000+ in 1990 to 21 cases in the 1999–2000 cohort of 26,000+. Compared to Western rates the1990 prevalence is remarkably low and even the 1999–2000 rate is low.
Unlike the categories cleft lip and palate and "congenital heart diseases", phocomelia is a specific and rare birth defect. Its changing prevalence in Basra since the 1991 war is a striking phenomenon. The prevalence of phocomelia rose explosively from 0 cases in 1990 and one in the four-year period 1991–94 to five in the following four years, and 32 in 1999–2000. As such, studies designed to document phocomelia in other contexts where DU munitions have been exploded could contribute significantly to causal inference. Another strategy might be to investigate the residential proximity of women delivering neonates with phocomelia to sites which generate DU-containing pollution. Previously a global rise in phocomelia incidence bespoke thalidomide teratogenesis.
Most striking about the elevated rates and numbers described above is that they continue to rise with time, dramatically so in 1999–2000. The same pattern is true through 2001 for total malformations; there have been annual increases since 1997. This increasing rate of occurrence of malformations could contribute to distinguishing causally related exposures from non-causal associations. The effects of most time-limited exposures diminish as time since exposure increases. With DU, refinement of causal pathways could demonstrate a mechanism that expresses burgeoning internal damage or chronicity and compounding of exposure. The categorical breakdown of the totality of birth defects observed among those born at the Basra study hospital in 2001 will provide further information about trends.
While particular malformations of several of the classes discussed above can be derivative from abnormal neural crest development, it is both premature and not our area of expertise to try to draw conclusions about the possibility that defects derive from common physiologic processes. However, there is a need to more carefully examine the issue; if the biology supports the notion that a single underlying mechanism is possible, it would support the plausibility of DU as a prime candidate. A trans-disciplinary study that provides careful histological comparison of birth defects produced in animals under controlled circumstances and those seen in epidemiological studies is also warranted.
In discussion of their findings through 1998, the Basra researchers stated that their data indicate an increase in birth defects in Basra beginning in 1995, four years after initial exposure. With the data through 2000 at hand, these writers' assessment is that it is difficult to distinguish between a gradual and ongoing increase in rates for various defects vs. a several year lag-to-onset of any elevation of rates. Specifying either of these distinctive temporal patterns for occurrence of birth defects could contribute to efforts to separate the teratogenic roles of DU and other possible antecedent causes, and also to efforts to discern the underlying mechanism that leads to teratogenesis. Both descriptions of the Basra pattern (a gradual rise, a lag-time-to-rise in frequency of occurrence of birth defects) are consonant with the Battelle Laboratory work that measured at minimum a 4-year dissolution and migration half-time for ceramic DU [5].
If it is only according to a multi-month timeframe that DU manifests its teratogenic potential, then the utility of the early post-1991 Gulf War American veterans studies is compromised. The 2003 study by Araneta et al. considered infants born only through 1993. It is because of its longevity that the Basra registry is providing substantial information.
It is noteworthy that the US veterans cohort study does not corroborate several other Basra findings regarding classes of malformations with elevated incidence (neural tube defects – anencephaly and meningomylocele, cleft lip and palate, births with multiple congenital malformations, phocomelia), likely because the designs of the studies were dissimilar. A forward extension of the study of Araneta et al. along with refinement of the exposure measure to allow for classification of Gulf War veterans by whether or not there was DU exposure in their particular theatre of combat and by characteristics of that exposure would be quite informative and is recommended. Inference from the reported statistical association between maternal 1991 Gulf War deployment and hypospadias and epispadias, as a currently completely uncorroborated finding, would also benefit from a refined and expanded analysis.
The issue of how to distinguish the role of DU from that of other suspected teratogens is serious and complex. The response to this challenge is built on the interface of laboratory research and population studies; its glue is the application of epidemiological principles of inference. Laboratory and animal research are proceeding apace and are suggesting plausible pathways by which internalized DU aerosols could be mutagenic and/or teratogenic. As animal studies come to provide more detail about the internal migration of inhaled ceramic DU and its decay particles, inference regarding possible teratogenic pathways for specific birth defects can be refined.
A 1994 U.S. General Accounting Office report identified 21 reproductive toxicants and teratogens, including DU, that were present in the 1991 Gulf War environment [68]. A commonality of excessive occurrence of a particular birth defect among offspring of American veterans, and offspring of Iraqi veterans and resident civilians would decrease the likelihood that certain of those 21 toxins had a causal role in the elevated rate of occurrence of that defect among American veterans' offspring. For example, Iraqis did not receive the "medications and vaccines administered to Gulf War veterans". Therefore, a similar or identical excess of a particular class of birth defects among offspring of DU-exposed Americans and Iraqis could not uniformly be attributed to those medications and vaccines.
It is from this vantage that the Socorro case study is of particular significance. By trans-national standards, the rate of occurrence of hydrocephalus in Basra during the years 1999 and 2000 was very high. (This, notwithstanding the need for clarification of the 1990–1998 registry data regarding occurrence.) If DU is the sole, or one of a small group of, risky exposure(s) shared by residents of the Iraqi war region and residents of the rural Socorro, New Mexico munitions testing region, then the likelihood of a causal role for DU in the genesis of hydrocephalus is increased.
More generally, serious effort needs to be directed toward disentangling the role of DU from that of other potential teratogens in tandem with which DU exposure has frequently occurred. This task becomes less daunting, though more urgent, as the contexts in which DU munitions have been exploded increases. The identities of the "other potential teratogens" disbursed into the environment by the crash of an airplane carrying DU in a civilian area differ, at least somewhat, from those disbursed by DU fires in a combat zone. In response to "widespread distress" about crash-associated risk, a theoretical physics-based model of the 1992 event was developed. While that theoretical study did not include any assessment of the health status of the exposed population (and their offspring), the authors concluded that it was "improbable" that the DU that had burnt and aerosolized as a result of the crash precipitated health problems [69]. Such a purely theoretical approach seems inadequate, especially in light of the popular perception of a post-crash regional increase in malformed births [61]. Furthermore, associations documented in an unexpected context that cohere with findings of planned analyses are highly informative. Conversely, absence of observed associations in small, unexpectedly exposed populations would be less informative.
In addition to Socorro, New Mexico there are 50 other US sites where DU munitions are/have been developed, produced, tested. How many of these sites are located in areas where comprehensive birth defects registries exist? What about other countries? Could assays for DU biomarkers be done on groups of male and female parents of children with and without birth defects resident near such facilities?
For the Iraqi population the 1991 Gulf War was the prelude to various new exposures and circumstances that could be teratogenic – sanctions-induced deprivations such as poverty, malnutrition and degradation of the health care infrastructure. But such circumstances, without specific chemical or radiologic exposures, do not lead to the observed pattern of increasing rates for classes of congenital malformations, notwithstanding the fact that malnutrition does contribute to certain birth defects. If a comparable birth defects registry (1990–2000) were available for births in a section of northern Iraq not exposed to DU bombardment, it could help distinguish between war-induced and post-war exposures.
A cohort study from Kerala, India is a particularly apropos example of a well-executed investigation that was able to detect differences in the occurrence of birth defects (and other untoward pregnancy outcomes) among population groups [70]. In a genetic epidemiological and fertility survey conducted among 700,000 people in regions with normal background radiation (85 to 110 mR/yr) and high background radiation (735 – 563 mR/yr) – from thorium monazite in the soil – Padmanabham et al used personalized, direct contact with families to document a statistically significant increase in congenital malformations and other birth outcomes in the area with higher background exposure. Besides ionizing radiation, consanguinity and nearness of spouse's birthplace were included as additional risk factors for each birth outcome. This study is a model for an investigation of the incidence of birth defects (and other pregnancy outcomes) in regions of Iraq with and without contamination by DU aerosols. Ideally the regions being compared would be as similar as possible on other criteria including distribution of occupations and religion, economic situation, culture, or would allow for "control" of differences, as in the model of the Kerala study. (Of course, this study is also informative because though the radiation exposure in the Kerala region is due to radon, the case for teratogenicity related to increased radiation exposure is made.)
The study of Abushaban et al [56] is an assessment of the impact of DU on one class of birth defects in the absence of sanctions. In Kuwait, where there was DU (and other wartime) exposure(s) but no post-war sanctions, the post-war incidence of cardiac malformations overall and of numerous specific sub-categories was elevated. Kuwaiti trend data regarding prevalence of other classes of birth defects, particularly those elevated in Iraqi and other DU-exposed databases, could be highly informative.
It would be poor science to not acknowledge another potent resource for investigating human DU teratogenicity. Even as shoe leather epidemiology generated the Socorro Case Study, the establishment of opportunities for interaction between scientists and activists might reveal other suspect clusters of birth defects. There are citizen activist groups organized around more than one of the U.S. DU munitions testing sites. Are activists aware of facts-on-the-ground that should inform scientific studies?
From the grassroots to the international context: Resolution of the longstanding jurisdictional conflict regarding representation of the U.N. position on potential health effects of radiation exposure is needed. In 1959 the IAEA and the WHO signed an agreement (WHA 12.40) that has been used by the IAEA to constrain WHO's health-related radiation research and it may also be affecting the UN approach to DU research [71].
There is a serious need for careful epidemiological research that can elucidate the relationship between DU exposure and specific classes of birth defects. High quality, well-funded registries to monitor prevalence of individual classes of birth defects must be the norm in countries where DU munitions have been exploded, in the home countries of veterans who have gone abroad and waged war using DU munitions and in the countries where DU munitions are manufactured. As well, there is a critical need for epidemiological investigations that incorporate companion bioassays to assess internal parental exposure to DU (and other suspected teratogens).
While this article focuses exclusively on congenital malformations there is documentation that in other contexts very early miscarriage is another reproductive endpoint affected by radiation exposure. This implies a radiation dose-response effect such that there is a ceiling in the proportionality between radiation exposure and frequency of congenital malformations. The observed drop in the birth rate in many European countries 7 – 9 months after the Chernobyl disaster presumably relates to an excess of miscarriages induced by that event – obscuring the underlying teratogenic impact. The Basra data report (Table 1) is consistent with a drop in number of births at the study hospital 7 – 9 months subsequent to the January 1991 regional use of DU weaponry. But whether the lower number of births in 1991 is due to increased miscarriage of damaged embryos, to reduced conceptions because of the more general vicissitudes of war or to some other factor cannot be determined with the available information.
There are three broad categories into which epidemiological assessment of data regarding a causal association between a potential risk factor and an outcome distribute: the existence of association (be it positive or negative), lack of association, the conclusion that the extant data are inadequate for inference. Data are never perfect, hence it is incumbent on the epidemiological/public health analyst to distinguish between situations where the data are so imperfect that no valid inference can be drawn and those where valid scientific assessment allows for attribution of risk. Regarding the teratogenicity of parental prenatal exposure to DU aerosols, the evidence, albeit imperfect, indicates a high probability of substantial risk. Good science indicates that depleted uranium weapons should not be manufactured or exploded.
List of abbreviations
AFFRI: Armed Forces Radiobiology Research Institute (given in the text)
DU: depleted uranium
IAEA: International Atomic Energy Agency
no.: number
NTD: neural tube defect
vs.: versus
WHO: World Health Organization
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
RH led the identification of and analysis of the epidemiological studies and took the lead on writing the bulk of the manuscript. DB took the lead on reviewing the toxicological data. He also reviewed the entire manuscript many times and participated in discussions and editing of the entire manuscript. BP researched specific points of detail for the manuscript, participated in discussions and editing, reviewed the entire manuscript and contributed to writing some sections.
Acknowledgements
Sunny Miller, executive director of Traprock Peace Center of Deerfield, MA hosted a presentation by Damacio Lopez (director of IDUST, International Depleted Uranium Study Team) which Rita Hindin attended and that eventually led to the writing of this paper. Our appreciation. Thanks to Len Dietz, Dan Bishop (of IDUST) and Tom Fasy (Mt. Sinai Medical Center, NYC) for their assistance early on explicating DU toxicology, and to the Uranium Weapons Study Team (of Traprock Peace Center) for thoughtful conversations and support to explore leads and deepen understanding of DU. Thanks to the conveners and attendees of the World Uranium Weapons Conference Hamburg Germany, October 16 – 19, 2003. Of greatest importance, Rita's attendance afforded her the opportunity to share thoughtful conversation with and learn from Iraqi researchers, Drs. Jennan Hassan, Jawad Al-Ali and Souad Al-Azzawi. We offer deep thanks, appreciation and respect for the information they shared, and for work that they and their colleagues are doing. We deeply appreciate the reporters and activists who have managed, against great odds, to report bits of information out of Iraq and who, as responsible, thoughtful citizens of many countries, assert their dignity and demand appropriate response to the challenges posed by DU aerosols. Rita also had the opportunity to speak with and learn from Drs. Chris Busby and Michel Fernex at the Hamburg conference. Their contributions to this paper stem from their long-term, on-going, related research as well as, more particularly, to the helpful and thoughtful comments they gave as peer reviewers of the submitted manuscript. Thanks to Tova Neugut for insightful conversations and for reading many early drafts of the manuscript. Jaime DeLemos helped us figure out the chemistry of depleted uranium. We thank Cato Hui for assistance with formatting the manuscript.
==== Refs
Bishop D Just what is Depleted Uranium, Anyway? International Depleted Uranium Study Team (IDUST) 2003
Bleise A Danesi PR Burkart W Properties, use and health effects of depleted uranium (DU): a general overview Journal of Environmental Radioactivity 2003 64 93 112 12500797 10.1016/S0265-931X(02)00041-3
Mitsakou C Eleftheriadis K Housiadas C Lazaridis M Modeling of the Dispersion of depleted uranium aerosol Health Phys 2003 84 538 544 12705453 10.1097/00004032-200304000-00014
Bem H Bou-Rabee F Environmental and health consequences of depleted uranium use in the1991 Gulf War Env Intl 2004 30 123 134 10.1016/S0160-4120(03)00151-X
Dietz LA Contamination of Gulf War Veterans and others 1997 Depleted Uranium Metal of Dishonor, Depleted Uranium education Project 134 152
Peterson S Remains of toxic bullets litter Iraq Christian Science Monitor 2003, May 15
Army Environmental Policy Institute Health and Environmental Consequences of Depleted Uranium Use in the US Army 1995 Technical Report Atlanta, Georgia
Caldicott H The New Nuclear Danger 2002 New York; The New Press
Durakovic A Horan P Dietz LA Zimmerman I Estimate of the time zero lung burden of depleted uranium in Persian Gulf War veterans by the 24-hour urinary excretion and exponential decay analysis Military Medicin 2003 168 600 5
Langmuir D Uranium solution-mineral equalibria at low-temperatures with applications to sedimentary ore-deposits Geochemica et Cosmochimica Acta 1978 42 547 569 10.1016/0016-7037(78)90001-7
Mishima J Parkhurst MA Scherpelz RI Hadlock DE Potential Behavior of Depleted Uranium Penetrators under Shipping and Bulk Storage Accident Conditions Battelle Pacific Northwest Labs, PNL-5415 1985 137
Durakovic A Horan P Dietz L The quantitative analysis of depleted uranium isotopes in British, Canadian, and U.S. Gulf War veterans Military Medicine 2002 167 620 627 12188230
McDiarmid MA Keogh JP Hooper FJ McPhaul K Squibb K Kane R DiPino R Kabat M Kaup B Anderson L Hoover D Brown L Hamilton M Jacobson-Kram D Burrows B Walsh M Health effects of depleted uranium on Gulf War veterans Environ Res 2000 83 168 180 10662531 10.1006/enrs.1999.4012
Durakovic A Undiagnosed illnesses and radioactive warfare Croatian Med J 2003 44 520 532
Domingo J Reproductive and developmental toxicity of natural and depleted uranium: a review Reproductive Toxicology 2001 15 603 609 11738513 10.1016/S0890-6238(01)00181-2
Arfsten DP Still KR Ritchie GD A review of the effects of uranium and depleted uranium exposure on reproduction and fetal development Toxicology and Reproductive Health 2001 17 180 191 10.1191/0748233701th111oa
McClain DE Benson KA Dalton TK Enjik J Emond CA Hodge SJ Kalinich JF Landauer MA Miller AC Pellmar TC Stewart MD Villa V Xu J Biological effects of embedded depleted uranium (DU); summary of Armed Forces Radiobiology Research Institute Research The Science of the Total Environment 2001 274 115 118 11453287 10.1016/S0048-9697(01)00734-3
Domingo JL Chemical toxicity of uranium TEN 1995 2 74 78
Domingo JL Paternain JL Llobet JM Corbella J The developmental toxicity of uranium in mice Toxicology 1989 55 143 152 2711400 10.1016/0300-483X(89)90181-9
Domingo JL Ortega A Paternain JL Corbella J Evaluation of the perinatal and postnatal effects of uranium in mice upon oral administration Archives of Environmental Health 1989 44 395 8 2610529
Llobet JM Sirvent JJ Ortega A Domingo JL Influence of chronic exposure to uranium on male reproduction in mice Fundam Appl Toxicol 1991 16 821 829 1884918 10.1016/0272-0590(91)90167-3
Paternain JL Domingo JL Ortega A Llobet JM The effects of uranium on reproduction, gestation and postnatal survival in mice Ecotoxicology and Environmental Safety 1989 17 291 6 2743917 10.1016/0147-6513(89)90049-3
Maynard EA Downs WL Hodge HC Voegtlin C, Hodge HC Oral toxicity of uranium compounds Pharmacology and Toxicology of Uranium 1953 3 New York: McGraw Hill 1221 1369
Malenchenko AF Barkun NA Guseva GF Effect of uranium on the induction and course of experimental autoimmune orchitis and thyroiditis J Hyg Epidemiol Microbiol Immuno 1978 22 268 277
Whish SR Mayer LP Robinson ZD Layton TM Hoover A Winterton B Begay A Benally D Poyer D Hoyer PB Dyer CA Uranyl Nitrate Mimics Estrogen Activity I n vivo and I n vitro E-hormone conference, the 6th annual symposium on the environment and hormones Center for Bioenvironmental Research at Tulane and Xavier Universities October 27–30,2004
Pellmar TC Stewart MD Villa V Xu J Biological effects of embedded depleted uranium (DU); summary of Armed Forces Radiobiology Research Institute Research The Science of the Total Environment 2001 274 115 118 11453287
Pellmar TC Fuciarelli AF Ejnik JW Hamilton M Hogan JB Strocko S Emond C Landauer M Toxicological evaluation of depleted uranium in rats implanted with depleted uranium pellets Toxicol Sci 1999 49 29 39 as cited in Arfsten DP, Still KR, Ritchie GD, 2001 10367339 10.1093/toxsci/49.1.29
Mc Diarmid MA Squibb K Engelhardt S Oliver M Gucer P Wilson PD Kane R Kabat M Kaup B Anderson L Hoover D Brown L Jacobson-Kram D Surveillance of depleted uranium exposed Gulf War veterans: Health effects observed in an enlarged "friendly fire" cohort J Occup Environ Med 2001 44 991 1000
Zhu SP Hu QY Lun MY Studies on reproductive toxicity of enriched uranium [abstract] Chun Hoa Yu Fanh (Chinese Journal of Preventive Medicine) 1994 28 219 22
Zaire R Notter M Riedel W Thiel E Unexpected rates of chromosomal instabilities and alterations of hormone levels in Namibian uranium miners Radiation Res 1997 147 579 584 9146703
Schroder H Heimers A Frentzel-Beyme R Schott A Hoffmann W Chromosome aberration analysis in peripheral lymphocytes of Gulf War and Balkan War veterans Radiation Prot Dosimetry 2003 103 211 220
Au WW Wilkinson GS Tyring SK Legator MS el Zein R Hallberg L Heo MY Monitoring populations for DNA repair deficiency and for cancer susceptibility Environmental Health Perspectives 1994 104 579 84 8781386
Au WW Lane RG Legator MS Whorton EB Wilkinson GS Gabehart GJ Biomarker monitoring of a population residing near uranium mining activities Environmental Health Perspectives 1995 103 466 70 7656876
Inouye D Teratology of heavy metals: Mercury and other contaminants Teratology 1989 40 652 653
Tahan JE Barrios LC Marcano L Levels of Hg, Pb and V in brain, kidney, liver and lung of anencephalic fetuses from the Eastern coast of Lake Maracaibo, Venezuela Trace Elements and Electrolytes 1996 13 7 13
Kline J Stein Z Susser M Conception to Birth Epidemiology of Human Development 1989 New York NY: Oxford University Press
May LM Heller J Kalinsky V Ejnik J Cordero S Oberbroekling KJ Long TT Meakim KCE Cruess D Lee AP Military deployment human exposure assessment: Urine total and isotopic uranium sampling results J Toxicol Env Hlth Part A 2004 67 697 714 10.1080/15287390490428189
Lopez D thoroughly documented personal communication. Authors have both a copy of the 1980's letter and of the 1998 report issued by the New Mexico Department of Health on hydrocephalus in New Mexico and in Socorro County 1984
Bukowski G Lopez D Uranium Battlefields Home and Abroad: Depleted Uranium Use by the Department of Defense 1993 Citizen Alert & Rural Alliance for Military Accountability, Carson City Nevada: Reno 166
Meyers N Citizen Science- Betty Mekdeci and the ABDC The Newsletter of the Science and Environmental Health Network 1999 3
Araneta MRG Moore CA Olney RS Goldenhar syndrome among infants born in military hospitals to Persian Gulf War veterans Teratology 1997 56 254 261 10.1002/(SICI)1096-9926(199710)56:4<244::AID-TERA3>3.0.CO;2-Z
Gunther S-H Uran-Geschosse: Schwergeschaedigte Soldaten, missgebildete Neugeborene, sterbende Kinder Uranium Projectiles: Severely Maimed Soldiers, Deformed Babies, Dying Children 2000 Projetiles d'uranium: Militaires gravement mutiles, nouveau-nes defformes, enfants mourants AHRIMAN-Verlag GmbH. Freibburg, Germany
Al-Sadoon I Hassan GG Yacoub AA-H Depleted Uranium and health of people in Basrah: Epidemiological evidence. Incidence and pattern of congenital anomalies among births in Basrah during the period 1990 – 1998 Selected Research Works on the Effect of DU on Man & Environment in Iraq – Republic of Iraq Ministry of Higher Education and Scientific Research I 26–27 March, 2002
Fasy T The recent epidemic of pediatric malignancies and congenital malformations in Iraq: The biological plausibility of depleted uranium as a carcinogen and a teratogen Iraqi-American Academics' Symposium for Peace Baghdad University 1/14–16, 2003
Hassan J At World Uranium Conference 2003 Hamburg, Germany
Al-Shammosy MM Neural tube defects in Diwaniah Increasing incidence. Proceedings of the Conference on the Effects of the Use of DU Weaponry on Human and Environment in Iraq Republic of Iraq Ministry of Higher Education and Scientific Research 26–27 March 2002
Al-Taha SAH Al-Kassab A survey of a genetic clinic patients for chromosomal, genetic syndromes and congenital malformations as detected by clinical and chromosomal studies: 1989–1990 vs. 1992–1993 The International Scientific Symposium on Post War Environmental Problems in Iraq Iraqi Society for Environment and Improvement, Baghdad 105 106 December 12–14, 1994
Sumanovic-Glamuzina D Saraga-Karacic V Roncevic Z Milanov A Bozic T Boranic M Incidence of major congenital malformations in a region of Bosnia and Herzegovina allegedly polluted with depleted uranium Croatian Med J 2003 44 579 584
Araneta MRG Destiche DA Schlangen KM Merz RD Forrester MB Gray GC Birth defects prevalence among infants of Persian Gulf War veterans born in Hawaii, 1989–1993 Teratology 2000 62 195 204 10992261 10.1002/1096-9926(200010)62:4<195::AID-TERA5>3.0.CO;2-5
Araneta MRG Schlangen KM Edmonds LD Prevalence of birth defects among infants of Gulf War veterans in Arkansas, Arizona, California, Georgia, Hawaii, and Iowa, 1989 – 1993 Birth Defects Research (Part A) 2003 67 246 260 10.1002/bdra.10033
Cowan DN DeFraites RF Gray GC Goldenbaum MB Wishik SM The risk of birth defects among children of Persian Gulf War veterans New Eng J Med 1997 336 1650 1656 9171068 10.1056/NEJM199706053362306
Kang H Magee C Mahan C Lee K Murphy F Jackson L Matanoski G Pregnancy outcomes among U.S. Gulf War veterans: A population-based survey of 30,000 veterans Ann Epidemiol 2001 11 504 511 11557183 10.1016/S1047-2797(01)00245-9
Penman AD Tarver RS No evidence of increase in birth defects and health problems among children born to Persian Gulf War veterans in Mississippi Mil Med 1996 161 1 6 11082741
Doyle P Maconochie N Davies G Maconochie I Pelerin M Prior S Lewis S Miscarriage, stillbirth and congenital malformation in the offspring of UK veterans of the first Gulf War Intl J Epidemiol 2004 33 74 86 10.1093/ije/dyh049
Maconochie N Doyle P Davies G Lewis S Pelerin M Prior S Sampson P The study of reproductive outcome and the health of offspring of UK veterans of the Gulf war: Methods and description BMC Public Health (on-line) 2003 3 4 10.1186/1471-2458-3-4
Abushaban L Al-Hay A Uthaman B Salama A Selvan J Impact of the Gulf war on congenital heart diseases in Kuwait Intl J Cardiol 2004 93 157 162 10.1016/S0167-5273(03)00193-1
Shields LM Wiese WH Skipper BJ Charley B Benally L Navajo birth outcomes in the Shiprock uranium mining area Health Phys 1992 63 542 551 1399640
O'Kane M Victims of a war they never saw Guardian Weekly Jan 10,1999.
Hudson D Miller K The tiny victims of Desert Storm LIFE 1995 18 46 59
Morizumi T Global association for the banning of depleted uranium weapons (ed) A Different Nuclear War: Children of the Gulf War Photos and text by Morizuni Japan 8/2002
Gunther SH Depleted uranium and the Gulf War Syndrome Covert Action Quarterly 2000 34 35
Tashiro A Discounted Casualties The Human Cost of Depleted Uranium 2001 Chugoku Shimbun.Japan
Guzman M personal communication 2004 Uranium Medical Research Centre, Toronto
Mitchel REJ Sunder S Depleted uranium dust from fired emissions: Physical, chemical and biological properties Health Phys 2004 87 57 67 15194923 10.1097/00004032-200407000-00007
Giannardi C Domenici D Military use of depleted uranium: Assessment of prolonged population exposure J Env Radioactivity 2003 64 227 236 10.1016/S0265-931X(02)00051-6
UUNEP Depleted uranium in Kosovo Post-conflict environmental assessment 2001 UNEP Scientific Team Mission to Kosovo UNEP, Geneva 5th – 19th Nov 2000
Edmonds LD James LM Temporal trends in the prevalence of congenital malformations at birth based on the Birth Defects Monitoring Programs, United States, 1979–1987 MMWR 1990 39 19 23 1979–1987 2124329
United States General Accounting Office Operation Desert Storm Questions remain on possible exposure to reproductive toxicants 1994 Washington DC. GAO-PEMD-94-30
Uijt de Haag Fear of flying J Hazard Mater 2000 76 39 58 10863013 10.1016/S0304-3894(00)00183-7
Padmanabham VT Sugunam AP Brahmaputhran CK Nandini K Pavithran K Heritable anomalies among the inhabitants of regions of normal and high background radiation in Kerala: Results of a cohort study 1988– 1994 Intl J Health Services 2004 34 483 515 10.2190/3XYE-QJPU-01BF-8YKE
International Institute of Concern for Public Health Conflict of interest between IAEA and WHO 1999 Toronto, Canada
Kendall GM Second -event theory reviewed J Radiol Prot 2000 20 79 80 10750958 10.1088/0952-4746/20/3/606
Zhou H Randers-Pehrson G Waldren CA Hei TK Radiation – induced bystander effect and adaptive response in mammalian cells Adv Space Res 2004 34 1368 72 15881649 10.1016/j.asr.2003.10.049
Suzuki M Tsuruoka C Heavy charged particles produce a bystander effect via cell-cell junctions Biol Sci Space 2004 18 241 6 15858391
|
16124873
|
PMC1242351
|
CC BY
|
2021-01-04 16:36:32
|
no
|
Environ Health. 2005 Aug 26; 4:17
|
utf-8
|
Environ Health
| 2,005 |
10.1186/1476-069X-4-17
|
oa_comm
|
==== Front
Environ HealthEnvironmental Health1476-069XBioMed Central London 1476-069X-4-191616475010.1186/1476-069X-4-19MethodologyCluster detection methods applied to the Upper Cape Cod cancer data Ozonoff Al [email protected] Thomas [email protected] Veronica [email protected] Janice [email protected] David [email protected] Ann [email protected] Department of Biostatistics, Boston University School of Public Health, 715 Albany Street, Boston, MA 02118, USA2 Department of Environmental Health, Boston University School of Public Health, 715 Albany Street, Boston, MA 02118, USA3 Department of Epidemiology, Boston University School of Public Health, 715 Albany Street, Boston, MA 02118, USA2005 15 9 2005 4 19 19 14 2 2005 15 9 2005 Copyright © 2005 Ozonoff et al; licensee BioMed Central Ltd.2005Ozonoff et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
A variety of statistical methods have been suggested to assess the degree and/or the location of spatial clustering of disease cases. However, there is relatively little in the literature devoted to comparison and critique of different methods. Most of the available comparative studies rely on simulated data rather than real data sets.
Methods
We have chosen three methods currently used for examining spatial disease patterns: the M-statistic of Bonetti and Pagano; the Generalized Additive Model (GAM) method as applied by Webster; and Kulldorff's spatial scan statistic. We apply these statistics to analyze breast cancer data from the Upper Cape Cancer Incidence Study using three different latency assumptions.
Results
The three different latency assumptions produced three different spatial patterns of cases and controls. For 20 year latency, all three methods generally concur. However, for 15 year latency and no latency assumptions, the methods produce different results when testing for global clustering.
Conclusion
The comparative analyses of real data sets by different statistical methods provides insight into directions for further research. We suggest a research program designed around examining real data sets to guide focused investigation of relevant features using simulated data, for the purpose of understanding how to interpret statistical methods applied to epidemiological data with a spatial component.
==== Body
Background
Unusual geographical patterns of disease may give rise to public concern and explanations are frequently sought. Attention is often directed toward potential environmental and other factors associated with the disease in question. These investigations often have high costs in time and money, and thus it is important to verify objectively that the distribution of cases is indeed "unusual". A number of statistical methods have been suggested to assess the degree and/or the location of spatial clustering of disease cases. A good overview of the general statistical problems of clustering in the area of public health is contained in [1]. For a review with somewhat more depth but narrower scope see [2].
Despite the variety of available statistics, and the importance of understanding the methodology itself, there is relatively little in the literature devoted to comparison and critique of different methods. Most of the available comparative studies rely on simulated data ([3,4] among others) rather than real data sets. Notable exceptions include the leukemia data from upstate New York, which have been extensively analyzed with a variety of methods (see for example [5]). The advantages of using simulated data are clear, namely spatial patterns can be specified in advance and power to detect patterns under specified conditions can be considered. However, the complexity and subtleties of real data sets are frequently beyond our abilities to simulate, and the potentially large number of parameters involved in such simulations make systematic investigation of particular elements a daunting task.
In this paper we compare analytic methods using breast cancer data from the Upper Cape Cod area of Massachusetts. Geographically, the Upper Cape has interesting features that would be difficult to simulate otherwise. Its shape is roughly rectangular, but with uneven edges. Population density is highly heterogeneous, including a large non-residential "hole" in the southwest quadrant (Otis Air Force Base). These geographic features have the potential to affect various spatial methods in different ways and to different extents, making these data rich and complex in a way that simulated data often are not. We chose to compare three methods currently used for examining spatial disease patterns; one is a global test for clustering, one is a local test for clustering, and one combines a global deviance statistic with locally estimated odds ratios. All three methods are relatively simple to implement and none require commercial software. However, only the scan statistic has been implemented in stand-alone software.
We do not attempt to provide a comprehensive comparison of all available methods or to provide a complete analysis of the breast cancer data, and the reader should not interpret the results of our investigation in the context of breast cancer clusters in the Upper Cape Cod region. In contrast to the many published reports on the New York leukemia data, our purpose here is not to infer specific differences between cases and controls in the breast cancer data. Instead we aim to achieve a better understanding of the analytic properties of the methods we have selected, features of the data that may be problematic for each, and which may be most appropriate for particular situations.
It is worth noting that the three methods are not directly comparable, in the sense that one is essentially global (the M-statistic); one is local (the scan statistic); and one calculates local odds ratios along with a global deviance statistic (Webster's Generalized Additive Model (GAM)). Thus there is no reason to expect that the results of hypothesis testing using these very different methods should agree. We argue that instances where the outcome of hypothesis tests using each of these three methods are discordant may reveal important aspects of the data that could not be perceived by using any one method exclusively. In this sense, these methods provide complementary views of the data. The information contained in each approach should be considered as part of a complete and thorough investigation of spatial patterns of disease.
Data
Data are from two population-based case-control studies of breast cancer on Upper Cape Cod, Massachusetts [6-8]. The Massachusetts Cancer Registry was used to identify incident breast cancer cases diagnosed from 1983–1993. Controls were chosen to represent the underlying population that gave rise to cases. Participants were restricted to permanent residents of the upper Cape region with complete residential histories. The case and control populations were frequency matched on age and vital status. Cases and controls were geocoded and locations entered into a Geographic Information System (GIS). For those subjects that moved during the study period, multiple residential locations were included in all analyses as appropriate.
Three latency assumptions were used in this paper. The zero latency analysis included all eligible residences i.e. exposures occurring up to diagnosis were assumed to contribute to the risk of disease. Thus all of the enrolled breast cancer cases (n = 200, representing 321 distinct residential locations) and matched controls (n = 471, representing 756 residential locations) are included in the zero latency analyses.
However, cancers initiated by exposures to environmental carcinogens may take much longer to develop. We therefore performed a 15 year and 20 year latency analysis by restricting inclusion to the residences occupied by participants at least 15 (or 20) years prior to the diagnosis (or index year, for controls). The 15 year latency analyses include 107 cases (170 locations) and 193 controls (389 locations), while the 20 year latency analyses include 248 cases (391 locations) and 341 controls (509 locations). The 20 year latency analysis includes subjects from a follow-up study, thus numbers of cases and controls are higher than would otherwise be expected due to the more restrictive latency assumption.
The latency assumptions thus produce three spatial patterns, giving case and control residences zero, 15, or 20 years prior to diagnosis. These data are described fully, including methodology for selection of cases and controls, demographics, and other features of the study population, in the final report of the full study as well as follow-up papers on the breast cancer data; see [7,8] for further details. For illustration, the spatial distributions of breast cancer cases and controls (with no latency assumption) are shown in Figure 1.
Figure 1 Breast cancer cases and controls. Distribution of breast cancer cases (in red) and controls (in blue). Each point represents the residence of one participant. Locations in this map have been geographically altered to preserve confidentiality. Actual residences were used in the analysis.
Methods
The three statistical methods described here are: Bonetti and Pagano's M-statistic, based on the interpoint distance distribution [5]; Webster's GAM approach, which uses smoothing techniques [9]; and Kulldorff's spatial scan statistic [10]. The M-statistic is a global unfocused test, meaning it is only concerned with departures of the spatial distribution of cases from the distribution of controls, without determining the location of any (possibly multiple) clusters or other differences. The GAM method maps disease odds ratios, provides a global test for deviation from a flat map, and identifies locations with significantly increased or decreased risk (here GAM is the conventional designation for Generalized Additive Model, not the Geographic Analysis Machine of Openshaw [11], also used in cluster investigations). It incorporates a smoothing function for location into a conventional logistic regression which accounts for effects of covariates. Kulldorff's scan statistic, the most widely used method for cluster investigations, scans the entire study region for local excesses and/or reductions of risk. Current implementations of the binary (Bernoulli model) version of the scan statistic allow adjustment for categorical covariates only, and the M-statistic as implemented does not adjust for covariates at all (although allowing for categorical covariates via stratification would seem to be a straightforward extension of the existing method). For simplicity we have chosen to apply all three methods to crude data only, thus avoiding the need to consider the differences in covariate adjustment across the three methods.
M-statistic
Bonetti-Pagano's M-statistic [5] is a non-parametric general test for clustering. It operates by representing and comparing the spatial distributions of two populations (here cases and controls) via the interpoint distance distribution. From any collection of n locations, we can calculate the roughly n2/2 interpoint distances between locations and consider the distribution of these distances. Typically, a resampling procedure on the entire study population is used to generate a baseline (or null) distribution. Both the null distribution (estimated via resampling) and the observed distribution (calculated from the interpoint distances between cases) are binned into histograms, each of which can be represented as a vector. The test statistic is then a Malhalanobis-like distance between the two vectors, weighted by an estimate of the covariance between histogram bins.
More formally, repeated resampling from the entire study population (cases and controls) is used to estimate the distribution of distances under the null hypothesis that both populations are sampled from the same spatial distribution. Binning these distances and taking the mean over all iterations gives expected counts for each bin of the histogram. Experience with this method suggests that the optimal number of bins grows roughly on the order of where n is the number of cases being assessed (see also [12]). Denote by e the vector of expected values in each bin, expressed as a proportion of the total number of distances. Repeated resampling also allows us to estimate the covariance of e, which we will denote by S, a k × k square matrix.
The interpoint distances for the disease cases are calculated, binned, and written as a k-dimensional vector o, the observed bin values (expressed as proportions). Then the M-statistic is:
M = (o - e)'S-(o - e)
where S- is the Moore-Penrose generalized inverse of the sample covariance matrix S. Thus we calculate the difference between the expected (under the null hypothesis of no clustering) bin proportions and the observed bin proportions of the disease cases, inversely weighted by the covariance estimator. As S- is a positive semi-definite matrix, M ≥ 0.
The asymptotic distribution of M is found in [5]. In practice we can use the resampling procedure to calculate the distribution of M empirically under the null hypothesis. Comparing the calculated value of the test statistic to the null distribution gives a p-value that can be interpreted as the probability that the spatial distribution of the disease cases differs from the entire study population by chance alone.
GAM smoothing
Webster et al. [9,13] have used a procedure based on smoothing and generalized additive models (GAMs) to map disease and detect clusters (see [14] for related work). The generalized additive model predicts the log odds of disease (logarithm of the ratio of cases to controls) as a linear function of some covariates and a smooth function of spatial coordinates.
Specifically, the model specifies that for an individual with covariates zi and spatial location (xi, yi), the probability pi of disease is given by:
logit(pi) = S(xi, yi) + βzi
where β denotes the vector of linear regression coefficients for the covariates. S(x, y) is a bivariate smooth function. Webster et al. use a loess (locally-weighted regression smoother) because it is adaptive to changes in data density typically found in population maps. Around each point in the study area, a variable sized window is constructed based on a predetermined number of nearest neighbors; within this window, the data contribute to S(x, y) according to a tricube weighting function. Details are covered thoroughly in [15]. The window size (span) will affect both the bias and the variance (i.e. the amount of smoothing). Reducing the span reduces the bias but also increases the variance (reducing smoothness). Various criteria have been developed to balance these two properties of the smoother. Webster et al. use the Akaike Information Criterion (AIC), which averages the deviance but penalizes the number of degrees of freedom. Minimizing the AIC estimates an "optimal" balance of bias and variance [15] in a computationally feasible manner. The global statistic tests the null hypothesis of a flat map using the deviance of the model with and without the smoothing term. Among the available global test statistics, here we have used the deviance statistic [9]. The distribution of the statistic is estimated using permutation testing, with the case-control status permuted repeatedly. A pointwise test is then used to locate areas with significantly increased or decreased log odds relative to the map as a whole (the overall case-control ratio for crude analyses). The permutations also generate a distribution of the log odds at each location under the null hypothesis. The local p-value is determined by comparing the observed log odds with the null distribution.
After all statistical tests are performed, the log odds are converted to odds ratios using the entire study population as a reference. The odds ratios are mapped and significant "hot" and "cold" spots are delineated by drawing the .025 and .975 quantiles of the pointwise p-value surface. This graphical display is a natural part of the statistic and offers a rapid interpretation of the results of the calculations. The entire procedure can be run with existing software, e.g. S-Plus for the GAM and ArcView for mapping.
We note that care should be taken when interpreting the map of local p-values, because there is no adjustment for multiple testing. Thus under the null hypothesis of identical spatial distributions of cases and controls, we can expect in general that statistically significant local p-values will occur at a higher rate than the Type I error rate specified by the nominal alpha level. In other words, the local p-values are not to be used for hypothesis testing since we do not have adequate control of the Type I error rate. The local p-values do provide information about the measure of effect (in this case the local odds ratio), but inference based on these local p-values alone should be avoided.
Scan statistic
Kulldorff's scan statistic [10] has become the most widely used test for clustering in recent years, both because of its efficacy in detecting single hot (or cold) spots as well as the availability of the free software package SaTScan [16] for implementing the test. The basic idea of the scan statistic is to allow circular windows of various sizes to range across the study region. At each location, the rate of disease inside the window is compared to that outside the window. A hot (respectively cold) spot is characterized by a higher (lower) localized rate of disease.
In a case-control setting, the scan statistic is a likelihood ratio test statistic under a Bernoulli probability model. For a given zone (circular window) Z let pZ, qZ denote the probability of a data point being a case inside or outside the circle, respectively. The likelihood function under this Bernoulli model can be expressed in a straightforward fashion in terms of p, q, and the number of cases and controls inside and outside Z. We can then calculate:
Let denote the zone for which LZ achieves its maximum. This is called the most likely cluster, and we can calculate a test statistic via a likelihood ratio test. Let L0 = supp = q L(Z, pZ, qZ) be the likelihood under the null hypothesis (no clustering) and use
as the statistic of interest. The most likely cold spot is calculated similarly.
As with the other methods, inference is based on permutation of the case-control status. Under repeated permutations, the distribution of λ under the null hypothesis is generated, and we compare the observed value of λ to this distribution to yield a p-value. As noted above, SaTScan provides a relative risk for the most likely hot/cold spot, here an odds ratio inside the circle divided by an odds ratio outside the circle (hence not exactly comparable to the odds ratio computed by the GAM method).
For this study, we used the most recent version of the publicly available software [16] for analyzing binary (case-control) data, searching for either hot or cold spots.
Results
The three statistics in question were calculated for the breast cancer data with each of three latency periods. The results, showing global p-values for the M-statistic and the GAM method, and local p-value (for the identified "most likely cluster") for the scan statistic, are summarized in Table 1.
Table 1 p-values associated with cluster statistics. Results (p-values) of analysis using the scan statistic, the M-statistic, and the GAM method with deviance statistic.
Breast cancer
20 yr lat 15 yr lat No lat
scan stat 0.068 0.241 0.209
M-stat 0.015 0.008 0.539
GAM 0.003 0.006 0.046
The three methods in general are not concordant when considered in a hypothesis testing context. However, all three methods are at least suggestive of significantly different spatial patterns for cases and controls when applied to the 20 year latency data set. The scan statistic result, while not significant at the customary 0.05 level, is nonetheless indicative of an excess of cases in the calculated most likely cluster, and contributes evidence towards a difference between cases and controls when considered in the context of the results of the other two statistics. The smoothed map using the GAM method (Figure 2) shows one hot and one cold spot, a situation in which all three statistics are expected to maintain some reasonable sensitivity. The corresponding "most likely cluster" produced by the scan statistic is also shown (Figure 3). When applied to the breast cancer data set with 15 year latency, both the M-statistic and the GAM indicate differences in the spatial distribution of cases and controls that are very unlikely to be explained by chance. The scan statistic, however, suggests that considered locally, random variation remains a plausible explanation. Examination of the smoothed map (Figure 4) shows two distinct and prominent hot spots in the data, and one cold spot. The presence of multiple clusters in the data may partially explain the divergent results. The associated scan statistic output is also shown (Figure 5).
Figure 2 Breast cancer 20 year latency (GAM). Breast cancer 20 year latency, GAM smoothed rate map. Solid lines delineate areas where the point-wise GAM deviance statistic is less than 0.05.
Figure 3 Breast cancer 20 year latency (scan). Breast cancer 20 year latency, scan statistic most likely cluster. Estimated relative risk for the indicated cluster is 0.823.
Figure 4 Breast cancer 15 year latency (GAM). Breast cancer 15 year latency.
Figure 5 Breast cancer 15 year latency (scan). Breast cancer 15 year latency. RR = 4.629.
When no latency is considered for breast cancer, the M-statistic is no longer statistically significant, making the GAM the only method that offers strong evidence against chance alone explaining the spatial patterns in the data. Figures 6 and 7 show the smoothed map for this data set as produced by the GAM and the cluster identified by the scan statistic, respectively. The GAM map shows a broad, diffuse area of increased risk (odds ratios (ORs) roughly 2.0) along the coast and periphery in the northern Cape Cod area. Kulldorff's likelihood-based method identifies the same area and roughly the same relative risk (RR), but the local excess of cases is not statistically significant. Both methods are detecting a single hot spot, but it is elongated instead of the optimal (circular) configuration for Kulldorff's method. The M-statistic provides no evidence of global differences at the significance level of 0.05, perhaps due to the diffuse nature of the apparent hot spot. Thus the evidence for clustering in this data set is mixed.
Figure 6 Breast cancer no latency (GAM). Breast cancer no latency.
Figure 7 Breast cancer no latency (scan). Breast cancer no latency. RR = 0.453.
Discussion
The discussion of results presented here should not be construed as epidemiologic findings, but rather the output of three statistical methods as applied to real data. The maps produced are for illustration purposes only, and should not be interpreted epidemiologically (one reason being that we have not controlled for covariates).
We remark that the common use of the word "cluster" to describe a disease hot spot represents only one kind of departure of spatial difference between cases and controls. The scan statistic alone restricts itself to this particular kind of spatial difference and further places emphasis on the single most likely circular hot or cold spot. We have chosen to adopt here the broader but more flexible objective of detecting any difference in the spatial distribution of cases compared to the controls. The problem of locating and quantifying local excesses or deficits is clearly important, and both the scan statistic and the GAM address this problem directly. The M-statistic does not, although extensions of distance-based methods to the problem of cluster location are currently being developed [17].
We have presented applications of three well-developed and theoretically-grounded methods to detect spatial differences in the distribution of cases and controls in a real data set. The different patterns seen in this data set, comprising breast cancer with different latency considerations, affect the outcomes of these methods. We have identified at least three features that plausibly are involved (the shape, number, and intensity of areas of inhomogeneity), but there are likely others present here and in other real data sets. For example, methods may have different sensitivity depending on the areal size and/or location of spatial differences. In these cases, the sensitivity of each method may differ depending on location of a particular hot or cold spot, even when size, shape and intensity of the hot/cold spot are comparable (e.g. differing "edge effects" across methods).
Each of these methods would be expected to have certain strengths and weaknesses. The M-statistic has been implemented both in case-control studies [5], and in surveillance settings [18] where there is a large amount of historical data to use as a baseline for the null distribution of distances. Simulations suggest that it has the potential to be sensitive to situations such as multiple hot spots, where other statistics (such as the scan statistic) may lose power [3,4], but these same studies show that the M-statistic will typically underperform other statistics when there is a single hot spot to detect.
Provided there is some historical record, or sufficiently large control population from which to resample, the M-statistic can handle small sample sizes adequately. This is important in a surveillance setting, and is an advantage over rate-based statistics that may have insufficient data in the small sample case to draw proper inferences. In environmental settings these situations may arise in small, neighborhood-sized population studies.
However, as currently implemented the M-statistic does not adjust for covariates, but instead is used on raw spatial data only. The origins of the M-statistic lie in public health surveillance where spatial confounders are implicitly accounted for in the immediate historical record. As noted above, the M-statistic does not locate hot spots, but rather detects a difference between the two populations under comparison. Because these differences are quantified via the interpoint distance distribution and not the geographic locations of cases and controls themselves, results do not have a direct interpretation as do the "most likely cluster" of the scan statistic or the local odds ratios of the GAM.
GAM smoothing is a robust data-based approach that can be run with standard software. The ability to map disease outcomes while adjusting for covariates in a way familiar to epidemiologists is a particular strength. It is semi-parametric, assuming a linear model in the covariates with an additive spatial effect. Ignoring covariates and considering the data on a purely spatial basis there are essentially no statistical assumptions required, although the choice of window size may affect the sensitivity of the smoothing approach. The GAM approach provides global statistics to test the map for overall deviation from flatness as well as a pointwise test to locate areas of significantly elevated and decreased disease risk. Sufficient sample size for stable rates is also important, and results for small sample sizes are difficult to interpret meaningfully.
The scan statistic will certainly excel [3,4] when there is a single hot spot present and that hot spot is roughly circular in shape. The model assumption of a Bernoulli distribution inside and outside a circular region can be suboptimal if either the hot/cold spot is not circular, or if there is more than one spot present. There has been additional work on the scan statistic focusing on examining or improving robustness to the shape of the hotspot [19-21].
The scan statistic is especially appealing because of its immediate identification of the most likely cluster. Public availability of the implementation via the SaTScan software has increased its popularity and visibility. Perhaps most importantly, the method's exceptional power to detect single hot spots deserves consideration in situations where a single hot spot scenario seems plausible, or even possible. As a rate-based approach, the scan statistic is also limited to sample sizes that provide stable rate estimates.
Aggregated data can be handled using a Poisson model, similar in spirit to the Bernoulli model used for case-control data. The currently available software can adjust for covariates in the Poisson case, and adjustments for categorical variables in the Bernoulli model are allowed in the most recent release of the SaTScan software.
Multiple hot/cold spots would seem to be problematic when using the scan statistic, since it uses a likelihood function from a model based on a single hot or cold spot only. Placing restrictions on the underlying probability model clearly results in higher power when the model is correctly specified, but the presence of multiple clusters would imply that the scan statistic has misspecified the model. Thus we should expect that in some of these situations the scan statistic may suffer loss of power. The GAM and M-statistic would be expected to be sensitive to a wider variety of multiple cluster arrangements, but this flexibility is inherent in the global nature of these test statistics in contrast to the essentially local nature of the scan statistic.
The published results cited above indicate that for the benchmark simulated data considered, the scan statistic is quite robust to some multiple cluster arrangements. We note that these comparisons are dependent on the simulated data used for the purposes of the power study. Multiple hot/cold spots may be common occurrences in real data sets, and a more thorough effort to generate realistic simulations for these data is a direction for future research.
Likewise, there is no reason to assume that areas of increased risk will be any particular shape, especially as neither underlying population nor possible exposures are similarly constrained. Several recent papers have continued investigation of the scan statistic and its performance when dealing with non-circular hot spots (as well as extensions of the methodology to improve robustness in these situations); see for example [17]. As with the issue of multiple hot spots, more work may be needed to simulate such data in a realistic manner. We again emphasize the importance of studies that consider real data in addition to synthetic data, and the potential to learn from both types of data as spatial methods continue to develop and improve.
Conclusion
With the variety of approaches to the problem of examining spatial patterns of disease, it is not surprising that some methods are more effective than others for detecting certain patterns. A better understanding of the relative strengths and weaknesses of the various methods is essential to appropriate choices of methodology. Studies of spatial distribution of disease will also benefit from the information available from a variety of statistical methods, and careful consideration of the complementary nature of this information should assist in the interpretation of results of studies with a spatial component.
To this point, much of the work in gaining this understanding has come from analyzing synthetic data, where the underlying model can be controlled and various features superimposed in order to perform a careful study of these strengths and weaknesses. However, some characteristics of real data sets may be hard to simulate with synthetic data, or may not be readily apparent in advance of analysis and further study, and results based on simulated data are at least partially dependent on the particular simulations themselves.
The comparative analyses by different methods of real data sets point to directions for further research of the properties of each of the statistics used in this paper. We suggest a further research program designed around alternately examining real and simulated data sets for these kinds of differences, in order to develop the practical application of statistical methods to epidemiological data with a spatial component.
List of abbreviations
AIC: Akaike Information Criterion
GAM: Generalized Additive Model
GIS: Geographic Information System
OR: Odds Ratio
RR: Relative Risk
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
AO and VV were responsible for statistical programming. All authors contributed to writing and editing.
Acknowledgements
AO's research partially supported by NIH grant RO1-AI28076 and NLM grant RO1-LM007677. TW, VV, DO, JW, and AA are supported by Superfund Basic Research Program 5P42ES 07381.
==== Refs
Lawson A Biggeri A Böhning D Lesaffre E Viel JF Bertollini R Disease Mapping and Risk Assessment for Public Health 1999 Wiley
Forsberg L A review of clustering literature for public health [Preprint]
Kulldorff M Tango T Park PJ Power comparisons for disease clustering tests Comp Stat Data Anal 2003 42 665 684 10.1016/S0167-9473(02)00160-3
Ozonoff A Bonetti M Forsberg L Pagano M Power comparisons for an improved disease clustering test Comp Stat Data Anal 2005 48 679 684 10.1016/j.csda.2004.03.012
Bonetti M Pagano M The interpoint distance distribution as a descriptor of point patterns: An application to cluster detection Stat Med
Aschengrau A Ozonoff D Upper Cape Cancer Incidence Study, Final Report
Aschengrau A Paulu C Ozonoff D Tetrachloroethylene-contaminated drinking water and the risk of breast cancer Environmental Health Perspectives 1998 106 947 953 9703477
Aschengrau A Rogers S Ozonoff D Perchloroethylene-contaminated drinking water and the risk of breast cancer: Additional results from Cape Cod, Massachusetts, USA Environ Health Perspect 2003 111 167 173 12573900
Webster T Vieira V Weinberg J Aschengrau A London JL Spatial analysis of case-control data using generalized additive models Proceedings from EUROHEIS/SAHSU Conference: 30–31 March 2003; Östersund, Sweden 2003 Small Area Health Statistics Unit 56 69
Kulldorff M A spatial scan statistic Commun Statist Theory and Methods 1997 26 1481 1496
Openshaw S Charlton M Wymer C Craft A A mark i geographical analysis machine for the automated analysis of point data sets Intl J Geographic Information Systems 1987 1 335 358
Mann HB Wald A On the choice of the number and width of classes for the chi-square test of goodness of fit Annals of Math Stat 1942 13 306 317
Webster T Vieira V Weinberg J A method for mapping population-based case-control studies using generalized additive models [Preprint]
Kelsall J Diggle P Spatial variation in risk of disease: A nonparametric binary regression approach Journal of the Royal Statistical Society, Series C 1998 47 559 573 10.1111/1467-9876.00128
Hastie TJ Tibshirani RJ Generalized Additive Models 1990 Chapman and Hall
SaTScan v 5.0
Ozonoff A Bonetti M Forsberg L Pagano M The distribution of distances, cluster detection, and syndromic surveillance 2004 Proceedings of the American Statistical Association, Biometrics Section 2004
Ozonoff A Forsberg L Bonetti M Pagano M A bivariate method for spatio-temporal syndromic surveillance MMWR 2004 53S 61 66
Kulldorff M Nagarwalla N Spatial disease clusters: detection and inference Stat Med 1995 14 799 810 7644860
Kulldorff M Zhang Z Hartman J Heffernan R Huang L Mostashari F Benchmark data and power calculations for evaluating disease outbreak detection methods MMWR 2004 53S 144 151 15714644
Tango T Takahashi K A flexibly shaped spatial scan statistic for detecting clusters Intl J Health Geog 2005 4
|
16164750
|
PMC1242352
|
CC BY
|
2021-01-04 16:36:32
|
no
|
Environ Health. 2005 Sep 15; 4:19
|
utf-8
|
Environ Health
| 2,005 |
10.1186/1476-069X-4-19
|
oa_comm
|
==== Front
Int J Behav Nutr Phys ActThe International Journal of Behavioral Nutrition and Physical Activity1479-5868BioMed Central London 1479-5868-2-101612021410.1186/1479-5868-2-10CommentaryThe place of physical activity in the WHO Global Strategy on Diet and Physical Activity Bauman Adrian [email protected] Cora L [email protected] Center for Physical Activity and Health, School of Public Health, University of Sydney, Sydney Australia 20062 Canadian Fitness and Lifestyle Research Institute, Ottawa Canada2005 24 8 2005 2 10 10 8 12 2004 24 8 2005 Copyright © 2005 Bauman and Craig; licensee BioMed Central Ltd.2005Bauman and Craig; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
In an effort to reduce the global burden of non-communicable disease, the World Health Organization released a Global Strategy for Diet and Physical Activity in May 2004. This commentary reports on the development of the strategy and its importance specifically for physical activity-related work of NGOs and researchers interested in increasing global physical activity participation.
Sparked by its work on global efforts to target non-communicable disease prevention in 2000, the World Health Organization commissioned a global strategy on diet and physical activity. The physical activity interest followed efforts that had led to the initial global "Move for Health Day" in 2002. WHO assembled a reference group for the global strategy, and a regional consultation process with countries was undertaken. Underpinning the responses was the need for more physical activity advocacy; partnerships outside of health including urban planning; development of national activity guidelines; and monitoring of the implementation of the strategy.
The consultation process was an important mechanism to confirm the importance and elevate the profile of physical activity within the global strategy. It is suggested that separate implementation strategies for diet and physical activity may be needed to work with partner agencies in disparate sectors (e.g. urban planning for physical activity, agriculture for diet). International professional societies are well situated to make an important contribution to global public health by advocating for the importance of physical activity among risk factors; developing international measures of physical activity and global impacts of inactivity; and developing a global research and intervention agenda.
==== Body
Introduction
Physical inactivity is recognized as a major risk factor for non-communicable diseases (NCDs), and ranks between the second and sixth most important risk factor in contributing to the population burden of disease in westernized countries [1-3]. The increasing global problem of NCDs means that obesity, poor diet and inactivity are increasing problems for countries in the epidemiological transition [4].
From a physical activity standpoint, it is interesting to reflect on the temporal relationship between the accrual of evidence and the time delays to the development of policy frameworks for action. Initial epidemiological studies in the 1950s and 1960s identified, for the first time, population level evidence that inactivity was a risk factor for cardiovascular disease or for all cause mortality [5,6]. This evidence continued to accumulate, such that by 1987 a systematic review reported a consistent relationship between inactivity and cardiovascular disease [7]. Three years later, this was confirmed by a formal meta-analysis [8]. This period and the following half dozen years was characterized by increased interest and advocacy by physical activity researchers and organizations, resulting in consensus statements and a US Surgeon General's Report [9,10]. Gradually, some countries engaged with the physical activity agenda, developed guidelines and started to identify physical activity related health targets [10-12]. However, most countries paid little attention to addressing levels of inactivity in a systematic manner.
Over the past few years the World Health Organization (WHO) has become interested in NCD prevention as a global health concern, fueled by WHO discussion and a resolution to focus on NCD prevention and control in mid 2000 [13]. This document urged countries to "to develop national policy frameworks... to create conducive environment for healthy lifestyles... (largely due to) unhealthy diet, physical inactivity and tobacco use" [13].
Interest in NCD prevention led to reflections on the contributory risk factors, and in a global context, diet and inactivity became issues of concern. Increasing rates of obesity among youth have been recognized since the 1960s [14], and since around 1980 among adults [15], but quite suddenly since the late 1990s, obesity has received increasing political and media interest. This further contributed to increased international interest in inactivity and poor diet by around 2000 or 2001. Furthermore, following the advocacy and efforts stimulated by the Agita programs in South America created interest in developing countries [16]. As a consequence of these efforts, and of the WHA53.17 resolution, the Director General of WHO in 2001 recommended that world Health Day in 2002 should be physical activity focused, and the 'Move for Health' initiative was launched in early 2002 [17].
This commentary reports on the international development of the 2004 WHO Global Strategy for Diet and Physical activity, viewed from the physical activity perspective. The purposes are to report on the development of the strategy, to show how physical activity was positioned during and after the strategy was developed, and to indicate the potential importance of the strategy for the physical activity related work of international organizations, professional societies and researchers interested in the physical inactivity as a global public health problem.
Discussion of the Global strategy – through development to implementation
Physical activity and the development of the Global Strategy
WHO recognized the need and commissioned a global strategy on diet and physical activity at its 56th World Health Assembly [18]. An important influencer was the earlier #916 report by WHO/FAO [19] which indicated the health risks of obesity and overnutrition, and their contribution to global ill-health. The #916 report also indicated the benefits of physical activity on cardiovascular disease, diabetes and for osteoporosis prevention and mentioned the IARC report on the cancer prevention role of physical activity and weight control [20] which had influenced its development. Physical activity was included as an adjunctive idea, obviously contributing to the energy expenditure side of energy balance, but was not an initial impetus for the strategy.
One important contribution to physical activity was the development of the WHO 'Move for Health' day. This had emanated out of the local and national work of Agita!, a community-wide physical activity and advocacy program which started in the 1990s in the San Caetano region of Sao Paulo [16,17]. This initiative started in Brazil, but spread to other parts of South America, and finally led to WHO interest, and to World Health Day in April 2002. Since then, annual 'Move for Health' day work has occurred under the auspices of WHO, as well related efforts through the Agita Mundo NGO in Brazil [21].
Thus the WHO strategy development had a mandate to consider both 'diet and physical activity', and an Expert Reference Group was convened in 2002. This 14 member group primarily consisted of nutrition-oriented experts, but two [the authors of this commentary] had specific physical activity expertise.
The process of developing the strategy comprised several stages. A draft strategy was written by WHO staff and the Expert Reference group by late 2002, and then processes of consulting with individual countries (and through WHO regional consultations), the private sector, NGOs and other UN agencies occurred [22,23].
These discussions and consultations considered diet and physical activity. The feedback from most regions reflected roughly equivalent concern with issues related to diet and physical activity; only one region focused solely on diet related issues. The authors prepared a summary of the physical activity-specific themes emanating from the regional discussions and consultations and submitted this distillation to the Expert Reference group in August 2003 [24]. This was a qualitative review across the regional and country consultations, public web forum, NGO and private sector reports, and UN agencies consultations. It describes 'how the world viewed the important physical activity issues in 2003', from the perspective of the global strategy. The main themes are shown in Figure 1 [adapted from [24]].
Figure 1 How the world characterized the most important physical activity issues relevant to the development of a global strategy, 2003 #.
Underpinning all the responses from the consultations was the need to advocate more widely to governments for coordinated planning and resources for physical activity. In addition, there was widespread recognition that population-wide physical activity efforts need interagency collaboration and partnerships; physical activity promoting efforts require substantial work with other effector agencies outside of health, including departments of transport, urban planning, education and sport, most of which differ from those for nutrition promotion. This is consistent with previous WHO frameworks, dating back to the Ottawa Charter [25]. This intersectoral work will allow the development of health enhancing physical environments, and the development of multi-level policies that could support physical activity efforts [26]. One corollary of this is that, if the effector arms of physical activity programs are different to nutrition, then strategic implementation and partnerships might be kept separate, although the health-related consequences of poor diet and physical inactivity show large degrees of overlap.
Although some countries have physical activity guidelines, these tend to be in developed nations [10-12]. The extent of physical activity policies is even less well documented, and sometimes these are embedded in other policy documents for NCD prevention or obesity [27]. WHO was seen as responsible for supporting guideline development, as well as providing technical support for the evidence base for intervention, particularly in developing countries.
The importance of increasing community awareness was highlighted, but in order to achieve this, consistent physical activity messages are required. The recommendation of 'half an hour of achievable moderate intensity activity on most days of the week' [10] fits well into social marketing and media campaign efforts. Additional physical activity, or activity at a greater intensity may be required for some health outcomes such as cancer prevention or weight loss, but making the message(s) more complicated may confuse efforts to raise community awareness.
Finally, monitoring the implementation of the Strategy was thought to be important – documenting what happens by country, region and at the NGO level would provide a useful framework for assessing the actions undertaken relevant to the Global strategy. This process evaluation should be supplemented by the development of national monitoring systems, to assess and compare epidemiological trends in physical activity behaviors over time.
Physical activity was given emphasis in some regional consultations, especially the WHO Western Pacific region, Europe, and the Pan-American region. The latter region, armed with an already existing physical activity network (RAFA) [17], focused on the need for paradigm shifts, from an emphasis on sport to a new focus on 'active living' [28].
The launch of the Global Strategy and its sequelae
The development of the Global Strategy generated much political interest, media attention and controversy especially around nutrition. Concerns expressed by some Governments and by the private sector influenced the levels of agreement with, and content of the Strategy [29]. Changes made during the preparation of the Global Strategy were mostly confined to nutrition, rather than physical activity [30]. Nutrition appeared to be more controversial, but was sometimes given more emphasis as the "most important" risk factor by some writers [30].
The physical activity elements in the Global Strategy were mostly unaltered by the politics of consultation and revision. One general reference to 'individual responsibility' for physical activity and health was attenuated in the final version by framing individual choices in the context of health promoting environments [18]. The Global Strategy was approved by the World Health Assembly of WHO in May 2004 [18]. It provided a platform for advocacy, and an international 'call to action' to reduce NCD risk factors.
After its release, the Global Strategy received ongoing media attention. The media tended to be overly focused on the nutrition controversy, and sometimes even incorrectly described it as a global "obesity strategy". However, sometimes commercial interests became suddenly interested in funding physical activity promotion programs, perhaps to orient decision-makers and political attention away from the 'overheated' nutrition debate. Thus, large scale partnerships with the private sector around physical activity need to be considered carefully and ethically before rushing into conjoint program development.
Conclusion
The regional consultations were an important mechanism for developing the global strategy. They affirmed the relevance of physical activity in most WHO regions, and added emphasis to physical activity within the overall strategy. The qualitative analysis identified that the regional consultations had played a key advocacy role that helped to drive the physical activity agenda from the periphery, and hence its profile was increased in the final document.
It is suggested that separate implementation of diet and physical activity strategies is needed since there are different effector agencies. Transportation, sport agencies, recreation and urban planning policies are integral to physical activity promotion, whereas agricultural food policy and trade are more central to diet; there is not an automatic overlap of partnerships with these disparate agencies for implementation of the strategy. This is different to a 'health oriented' approach, which sees commonalities and integration only within a non-communicable disease framework, or within approaches to obesity prevention and control. These should be utilized, as the overall objective of the Global Strategy is to reduce NCDs. Nonetheless, engaging with agencies and partnerships outside health is a valuable approach to fostering commitment to developing and resourcing programs.
There clearly needs to be a greater commitment to ongoing population level physical activity measurement. Some efforts at developing global instruments for measuring physical activity have commenced, including the International Physical Activity Questionnaire (IPAQ), which has been shown to be reliable and valid in 12 countries [31]. Other efforts, through the WHO Steps surveillance system, are trialing the Global Physical Activity Questionnaire (GPAQ), a domain-specific short version of the IPAQ instrument [32].
The adoption of the global strategy by the WHO Assembly is a unique opportunity in the history of international physical activity work, as the development of common frameworks, policies and programs would enable greater program opportunities and partnerships at the national level. However, no resources have been earmarked to do this work, and implementation plans remain to be developed. Efforts to engage with countries and move this agenda forward are under way at the regional levels of WHO, with support and advocacy from NGOs and professional groups and societies.
As an international society, ISBNPA has greater potential to contribute to 'big picture participation in global work', compared to national organizations representing obesity or exercise science. The challenges posed by working with the Global strategy are very different to scholarly academic work; this engagement with the Global strategy is unpaid work, requires advocacy with governments and decision makers and is not often rewarded by academic funding or publication. The benefits are in making contributions to real population health efforts, and in improving the underecognised profile of physical activity among risk factors. The International Society of Behavioral Nutrition and Physical Activity (ISBNPA) had supported the development of the Global Strategy through a formal correspondence with WHO; now is a unique time for such organizations and their constituent members to contribute to the great challenges of international population behavior change.
The potential roles of organizations such as ISBNPA, American College of Sport Medicine, International Association for the Study of Obesity and others are in a few key areas of research and policy. First, advocacy for physical activity, to keep it on the political and health agenda of national and regional governments, especially advocating in transitional countries where the burden of NCDs will increase dramatically in the coming decades [34]. Second, it is important to move physical activity to a 'whole of government' agenda, to include a range of agencies, such as sport, transport and urban planning, as well as the private sector and NGOs [27]. Organizations such as ISBNPA can be research and policy brokers in fostering these relationships, and in adopting standard internal policies, which are consistent and could be applied in different contexts. Finally, in terms of science, members of ISBNPA could contribute to the global research and surveillance agenda [33].
The research challenges in the international context are worthy of more urgent efforts. For example, developing international measures of physical activity remains difficult. There are trade offs between capturing several domains of activity [versus one leisure time domain]. Measurement validity is difficult to establish, not only against criterion objective measures of movement or fitness, but in the varied cultural contexts and many different meanings that may be applied to the same self reported physical activity questions in different countries. Other research challenges include the definition of global impacts on inactivity. The globalization research agenda could include studies of trends in occupational physical activity, changes to the domestic and urban environments, factors contributing to the development of pervasive sedentary lifestyles, and monitoring declines in active commuting [35]. The measurement development agenda includes establishing and using standard process and impact indicators to assess the implementation of the global strategy across sectors and populations. Finally, the evaluation designs around a global initiative preclude planned comparison groups. Assessing the impact on physical activity in countries with active policy and resourced programs can be compared to demographically similar countries with limited program development. The international research challenges here are vast, and more complex than controllable smaller sample research in developed countries. Case studies of best practice may be one possible solution, provided the evidence from these is disseminated widely. Nonetheless, for those who are serious about public health approaches to increasing physical activity, global engagement is a necessary component of the work that we have yet to do.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
both authors were the Physical Activity representatives on the Expert Reference group, WHO Global Strategy on diet and physical activity 2002–2004; both contributed to the conceptualizing and writing of this paper. The views and opinions expressed in this commentary are solely those of the authors.
==== Refs
World Health Organization Reducing risks, promoting healthy life World Health Report 2002 2002 Geneva: World Health Organization
Mathers C Vos T Stevenson C Burden of disease and injury in Australia, AIHW Catalogue PHE 17 1999 Canberra: Australian Institute of Health and Welfare
Powell KE Blair SN The public health burdens of sedentary living habits: theoretical but realistic estimates Med Sci Sports Exerc 1994 26 851 6 7934758
Waxman A WHO's global strategy on diet, physical activity and health Response to a worldwide epidemic of non-communicable diseases Scand J Nutrition 2004 48 58 60 10.1080/11026480410033539
Morris JN Heady JA Coronary heart disease and physical activity of work Lancet 1953 2 1053 57 13110049 10.1016/S0140-6736(53)90665-5
Paffenbarger RS Hale WE Work activity and coronary heart disease New England Journal of Medicine 1975 292 545 50 1128551
Powell KE Thompson PD Caspersen CJ Hendrick J Physical activity and the incidence of coronary heart disease Annual Review of Public Health 1987 8 253 87 3555525 10.1146/annurev.pu.08.050187.001345
Berlin J Colditz G A meta-analysis of physical activity in the prevention of coronary heart disease American Journal of Epidemiology 1990 132 612 628 2144946
Pate R Pratt M Physical activity and public health: a recommendation from the Centers for Disease Control and Prevention and the American College of Sports Medicine JAMA 1995 273 402 407 7823386 10.1001/jama.273.5.402
US Department of Health and Human Services Physical Activity and Health: A Report of the Surgeon General 1996 Atlanta, GA: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion S/N 017-023-00196-5
Commonwealth Department of Health and Aged Care (DHAC) National physical activity guidelines for Australians 1999 DHAC: Canberra
Health Canada and Canadian Society for Exercise Physiology Canada's Physical Activity Guide to Healthy Active Living 1998 Cat. No. H39-429/1998-1E
WHA53.17 Fifty-third World Health Assembly. Prevention and control of noncommunicable diseases 2000 World Health Organization, Geneva
Wang Y Monteiro C Popkin BM Trends of obesity and underweight in older children and adolescents in the United States, Brazil, China, and Russia Am J Clin Nutr 2002 75 971 7 12036801
Flegal KM Carroll MD Ogden CL Johnson CL Prevalence and trends in obesity among US adults, 1999–2000 JAMA 2002 288 1723 7 12365955 10.1001/jama.288.14.1723
Matsudo V Matsudo S Promotion of physical activity in a developing country: The Agita Sao Paulo experience Public Health Nutrition 2001 5 1 10
Matsudo S Matsudo V The Agita Sao Paulo Program as a model for using physical activity to promote health Revista Panamericana de Salud Publica 2003 14 265 272 14662077
World Health Assembly 57.17 Global strategy on diet and physical activity 2004 Geneva: World Health Organization
WHO/FAO Diet, nutrition and the prevention of chronic diseases. Report of a joint WHO/FAO Expert consultation WHO Technical report Series #916 World health Organsation/Food and Agriculture Organisation of the United Nations 2002 accessed November 2004
Vanio H Bianchini F IARC handbooks of cancer prevention. Volume 6: weight control and physical activity 2002 Lyon, France: IARC Press
Move for health day. World Health Organization 2002 World Health Organization, Geneva WHO accessed November 2004
Waxman A Prevention of chronic diseases: WHO global strategy on diet, physical activity and health Food Nutr Bull 2003 24 281 4 14564933
Waxman A Norum KR Why a global strategy on diet and physical activity? The global burden of non-communicable diseases Public Health Nutrition 2004 7 381 383 15153269 10.1079/PHN2004623
Bauman A Craig CL unpublished submission to Expert Reference Group, WHO diet and physical activity strategy September 2003 file note WHO Sept 2003
Ottawa charter for health promotion 1986 World Health Organization. Geneva
Bauman A Bellew B Environmental and policy approaches to promoting physical activity. In "Health in the Commonwealth – sharing solutions 1999/2000" Commonwealth Secretariat, London 1999 38 41
Bull FC Bellew B Schoppe S Bauman AE Developments in National Physical Activity Policy: an international review and recommendations towards better practice J Sci Med Sport 2004 7 93 104 15214607
Jacoby E Bull F "Rapid changes in lifestyle make increased physical activity a priority for the Americas." Revista Panamericana de Salud Publica 2003 14 226 228
Cannon G Why the Bush administration and the global sugar industry are determined to demolish the 2004 WHO Strategy on diet, physical activity and health Public Health Nutrition 2004 7 369 380 15153268 10.1079/PHN2004625
Danzon M WHO – Diet and physical activity : the public health debate Europ J Public Health 2004 14 336 10.1093/eurpub/14.3.336
Craig CL Marshall AL Sjostrom M Bauman AE Booth ML Ainsworth BE Pratt M Ekelund U Yngve A Sallis JF Oja P The International Physical Activity Questionnaire (IPAQ): A comprehensive reliability and validity study in twelve countries Med Sci Sports Exerc 2003 35 1381 95 12900694
WHO Steps protocol – non Communicable disease surveillance. Expanded instrument pages 9–10 GPAQ Physical Activity questions 2004 WHO Geneva
Yach D Hawkes C Gould CL Hofman KJ The Global Burden of Chronic Diseases. Overcoming Impediments to Prevention and Control JAMA 2004 291 2616 2622 15173153 10.1001/jama.291.21.2616
Hu G Pekkarinen H Hanninen O Yu Z Tian H Guo Z Nissinen A Physical activity during leisure and commuting in Tianjin, China Bull World Health Organisation 2002 80 933 8
Bell AC Ge K Popkin BM The road to obesity or the path to prevention: motorized transportation and obesity in China Obes Res 2002 10 277 83 11943837
|
16120214
|
PMC1242353
|
CC BY
|
2021-01-04 16:38:04
|
no
|
Int J Behav Nutr Phys Act. 2005 Aug 24; 2:10
|
utf-8
|
Int J Behav Nutr Phys Act
| 2,005 |
10.1186/1479-5868-2-10
|
oa_comm
|
==== Front
Int J Health GeogrInternational Journal of Health Geographics1476-072XBioMed Central London 1476-072X-4-211615940310.1186/1476-072X-4-21MethodologyThe Integrated System for Public Health Monitoring of West Nile Virus (ISPHM-WNV): a real-time GIS for surveillance and decision-making Gosselin Pierre [email protected] Germain [email protected] Sonia [email protected] Monique [email protected] Institut national de santé publique du Québec (INSPQ), 945 Wolfe avenue, Sainte-Foy (Quebec), G1V5B3, Canada2 Centre hospitalier universitaire de Québec (CHUQ), 2705 Laurier boulevard, Sainte-Foy (Quebec), G1V 4G2, Canada3 Centre for Research in Geomatics, Laval University (Quebec), G1K 7P4, Canada2005 13 9 2005 4 21 21 21 6 2005 13 9 2005 Copyright © 2005 Gosselin et al; licensee BioMed Central Ltd.2005Gosselin et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
After its first detection in North America in New York in 1999, West Nile virus was detected for the first time in 2002 in the province of Quebec, Canada. This situation forced the Government of Quebec to adopt a public health protection plan against the virus. The plan comprises several fields of intervention including the monitoring of human cases, Corvidae and mosquitoes in order to ensure the early detection of the presence of the virus in a particular area. To help support the monitoring activities, the Integrated System for Public Health Monitoring of West Nile Virus (ISPHM-WNV) has been developed.
Results
The ISPHM-WNV is a real-time geographic information system for public health surveillance of West Nile virus and includes information on Corvidae, mosquitoes, humans, horses, climate, and preventive larvicide interventions. It has been in operation in the province of Quebec, Canada, since May 2003. The ISPHM-WNV facilitates the collection, localization, management and analysis of monitoring data; it also allows for the display of the results of analyses on maps, tables and statistical diagrams.
Conclusion
The system is very helpful for field workers in all regions of the province, as well as for central authorities. It represents the common authoritative source of data for analysis, exchange and decision-making.
==== Body
Background
He conquered Persia, Greece and Babylon and was revered and feared in East and West alike. At the age of 32, in the height of his power and glory, he was brought down... by a mosquito. According to recent speculations, Alexander the Great may have died as a result of infection by West Nile virus [1].
West Nile virus (WNV) infections may have been occurring in the Middle East for centuries. The virus has spread to new areas of the world and to new populations, causing infections that are characterized by signs and symptoms. Detected in North America for the first time in 1999 in the state of New York, it has since spread throughout the continent. Three years later, in 2002, the number of human cases of infection by this virus had increased dramatically. In the United States, 4156 cases and 284 deaths were confirmed. Moreover, 9862 confirmed cases (with 264 deaths) were reported in 2003 and 2470 (with 88 deaths) in 2004 (as of January 11th, 2005) [2].
The WNV was detected in mosquitoes, Corvidae (American Crows, Blue Jays and Common Ravens in our context) and humans for the first time in the province of Quebec, Canada, in early summer of 2002. A total of 20 confirmed human cases (including 3 deaths) were reported in 2002, 17 confirmed cases in 2003 (no deaths) and 3 cases in 2004 (1 death). This situation forced the Government of Quebec to adopt, in 2003, a public health protection plan against the virus [3] that remains in place to date. The main objectives of this plan are: to prevent complications and human deaths related to WNV infections, to ensure the early detection of the presence of the virus in a geographic area, to identify areas of potential transmission to humans for preventive action and monitoring, and to qualify the level of transmission to humans. The plan comprises several areas of intervention and their performance criteria:
- Monitoring: an integrated monitoring system (human, ornithological, entomological) in real-time;
- Laboratory: speed and provincial autonomy with regards to diagnosis (human, ornithological, entomological);
- Information: a communication plan;
- Intervention: rapid, effective, and flexible to adjust to the evolution of the epidemiologic situation;
- Research and evaluation: of the effectiveness and impacts of the actions taken;
- Decision process: a public health structure to optimize intervention capacity.
Given the particular epidemiologic character of the infection (avian amplifiers, transmission by mosquito vectors), the monitoring component of the plan is comprised of three interrelated elements:
- The monitoring of human cases of infection: the presence of infected persons that have acquired the infection locally confirms an active transmission of the WNV in the concerned area;
- The monitoring of animals: the presence of clustered dead Corvidae infected by the WNV indicates a potential source of amplification; these observations help identify target sites for mosquito monitoring;
- The monitoring of mosquitoes: the presence of a pool of infected mosquitoes indicates that a WNV source exists that presents a potential for the transmission of WNV to humans.
Analysis of the monitoring data makes it possible to target preventive interventions such that the appropriate personal, community, and environmental protection interventions can be considered; however, to be useful, the monitoring data must be available in real-time. The development of an information system to support the monitoring component of the intervention plan for the province of Quebec was entrusted to the Institut national de santé publique du Québec (INSPQ) in 2003 by the Ministère de la Santé et des Services sociaux (MSSS). The mandate of the INSPQ includes the implementation and continual update of an extranet web site dedicated to online data entry, data warehousing and document exchange. Moreover, this site included tables or graphs of relevant surveillance data, an online tool ensuring the validation and geographic localization of relevant events and an online real-time mapping tool for the integration of all data. The INSPQ thus developed the Integrated System for Public Health Monitoring of West Nile Virus (ISPHM-WNV) that allows the different actors involved to rapidly and easily assess the situation and recommend adapted interventions.
Using this system, the evaluation of the epidemiologic situation is carried out by an Expert Group that includes representatives of the appropriate government departments, scientists, and regional health authorities, in order to recommend the optimal interventions against WNV. In the event of a major epidemic situation, a high-level Advisory Committee is consulted on interventions before the Deputy Minister makes a decision.
In Canada, several provincial surveillance systems that include a WNV monitoring component have been implemented in the last few years, and information is shared through the Canadian Network for Public Health Intelligence (CNPHI). The CNPHI is targeted at improving the capacity of the Canadian health system to reduce human illness associated with infectious disease events by supporting intelligence exchange, surveillance activities and outbreak investigations. It is a national framework to collect and process surveillance data, disseminate strategic intelligence, and coordinate response to biological threats [4]. It comprises the West Nile virus Monitor, a surveillance component fed with provincial monitoring data that allows the visualisation of maps and tables. The provinces are responsible for the monitoring of the virus on their respective territory and the implementation of programs and systems to support this task. For example, the British Columbia Centre for Disease Control has implemented the Internet Geographic Information System (GIS) that allows the interactive mapping of various monitoring data related to WNV [5]. In the United States, the Centers for Disease Control and Prevention (CDC) have implemented the National WNV Surveillance System [6]. The objectives of the system are to monitor the geographic and temporal spread of WNV, develop national public health strategies for WNV surveillance, prevention, and control, develop a more complete regional picture of the geographic distribution and incidence of the other clinically important arboviruses, and provide national and regional information to public health officials, elected government officials, and the public. The data entry is done by reporting jurisdictions at the state level through the ArboNET system, the national electronic surveillance system established by CDC to assist states in tracking WNV and other mosquito-borne viruses. The system includes a mapping component and the maps it produces are available on the United States Geological Survey (USGS) web site. On a state or local level, WNV monitoring programs and infrastructures have been implemented. For example, see [7] for information on the infrastructure implemented by the State of New-York. The City of New York has also implemented the Dynamic Continuous-Area Space-Time (DYCAST) GIS system that was developed to identify and prospectively monitor high-risk areas for WNV [8].
In line with those previous examples, this paper outlines the context, architecture and different capabilities of the ISPHM-WNV, including its spatial and cartographic functionalities. In addition, examples of its utility as a tool for the management of risks and the reporting of results on a daily basis are presented. Finally, the paper explores future considerations for the development of such systems.
Results
The Integrated System for Public Health Monitoring of West Nile Virus (ISPHM-WNV)
Various data collection methods have been organized and implemented in order to feed the ISPHM-WNV. The 18 regional health authorities of the province are responsible for the mandatory reporting of all human cases of infection by WNV. The population is invited, by means of media campaigns, to report the presence of dead Corvidae to a reporting center (WNV-Info line). The location of reported Corvidae must be precisely indicated to allow for their collection and subsequent testing. Mosquito monitoring is performed at fixed monitoring stations as well as in potential risk zones identified following the analysis of infected bird clusters.
It is essential that the ISPHM-WNV support the integration of the different data sources and provide the required functionalities to assist the numerous actors with their respective tasks (data entry, data localization, data validation, data visualization, data analysis and decision-making).
The context and the architecture of the ISPHM-WNV
The INSPQ is responsible for:
- The implementation and the continual update, on a daily basis, of a secured extranet web site that is used to allow the exchange of documents and to update the monitoring database;
- The development and deployment, on the extranet network, of an on-line tool for the validation and the geographic localization of the various types of events recorded in the database;
- The development and deployment of a cartographic tool that provides an on-line cartographic representation of the monitoring data;
- The development and the update of charts and graphics of the temporal frequency distributions of the various events recorded in the database;
- The creation of an Expert Group to support the Advisory Committee in the interpretation of the results and in the analysis and the detection of aggregates, in time and space, of reported bird cases, of samples of infected mosquitoes, and of suspected and confirmed human cases.
A team of six people was formed and ensured the development and the management of the ISPHM-WNV. Moreover, the Expert Group was created to assist the development of a systematic analysis protocol for the data contained in the ISPHM-WNV, propose criteria for alerts, increase or withdraw surveillance components, produce the basic relevant analyses (at the provincial and regional health authority levels), detect significant sources of WNV activity, interpret the results, adjust interventions and promptly inform the Advisory Committee if significant expansion of insecticide use is needed. The committee also provides relevant technical opinions in order to improve the monitoring of WNV and the ISPHM-WNV itself.
The context of the system is shown in Figure 1.
Figure 1 The main organizations and groups involved in the Integrated System for Public Health Monitoring of West Nile Virus (ISPHM-WNV).
The extranet site of the ISPHM-WNV constitutes a portal to the different monitoring data collected, and is therefore principally intended for use by the regional public health professionals of the province. The integrated data presented on the extranet site of the ISPHM-WNV are divided into four categories: descriptions and analyses of dead Corvidae, and analyses of mosquitoes, human cases, and animals (horses and others wild animals tested).
The INSPQ purchased an internet-based cartographic representation software and consultation services for the development of a customized georeferencing tool for the geographic localization of the various data. The firm KHEOPS Technologies, with its cartographic software JMap® [9], was chosen to help fulfill the geographic processing needs of the system.
The main implementation steps of the first version of the ISPHM-WNV were carried out in four months in 2003:
- Opening of the reporting center (WNV-Info line) for dead Corvidae, including the implementation of the input forms and the tool for the geographic localization of the reports;
- First implementation of the cartographic representation tool on the extranet site;
- Implementation of a complete version of the cartographic representation tool;
- Implementation of the georeferencing tool for the batch geographic localization of events;
Since then, continuous improvements have been made to the system. The general architecture of the ISPHM-WNV is presented in Figure 2[10]. The different components will be described in more detail in the following sections.
Figure 2 The general architecture of the Integrated System for Public Health Monitoring of West Nile Virus (ISPHM-WNV).
The main components of the ISPHM-WNV – database
MS® SQL Server database
This database supports the entire system and contains not only the monitoring data but also all the data required for the proper functioning and maintenance of the system, including user accounts information, geographic data, the different analysis results and the shared documents.
The main components of the ISPHM-WNV – JMap
Cartographic functionalities
JMap® is an open solution for publishing interactive spatial applications over enterprise, public, and mobile networks [9]. The main characteristics and functionalities of the software are:
- On-line access (Java applets);
- Remote on-line administration via a secured web site;
- Integration of cartographic data of different formats;
- Unlimited number of simultaneous users;
- Cartographic navigation tools and measurement tools;
- Spatial analysis functions;
- Construction of different types of thematic maps;
- Flexible selection of information for display (time periods or qualitative attributes);
- Dynamic links of geographical locations to multimedia documents;
- On-line collaboration (red-lining) and email exchange of annotated information;
- Export data to common office software.
Georeferencing tool
A georeferencing tool based on the street addresses has been developed in JMap® and integrated to the ISPHM-WNV. This tool allows for the precise localization of dead bird reports and infection cases. The base road segments database used is CanMap Streetfiles from DMTI Spatial [11]. The tool includes a phonetic matching algorithm to help correct typographic errors that can occur during manual data entry.
Two versions of the georeferencing tool allow for individual localization of bird reports (for the telephonic reporting center) or batch processing for animal and human cases.
The main components of the ISPHM-WNV – extranet site
The extranet site includes:
- A News section containing the latest scientific information available on the WNV;
- An Overview section allowing for rapid assessment of the situation;
- A Rapid links section containing links to statistical information related to the different types of monitoring;
- A Shared documents section allowing the different users to exchange documents of any type;
- A discussion forum;
- Links to the different monitoring statistics (tables and graphs) sections of the system: human cases, Corvidae, mosquitoes and other animals;
- A link to the cartographic display software (JMap®) allowing for the visualization and the spatial analysis of the monitoring data;
- Links to the administration sections.
Monitoring data – human cases
This component of the system allows the Laboratoire de santé publique du Québec and the representatives of the various regional health authorities to manually enter, validate and visualize information related to human cases of WNV infection. The cases are localized using the georeferencing tool described earlier for representation on a map, along with other pertinent information.
Monitoring data – Corvidae reports
This component of the system allows the operators of the reporting center to manually enter, validate, and visualize the information related to reported dead Corvidae. When they become available, the laboratory tests' results are entered by the laboratory staff. The georeferencing tool allows the operators to immediately validate and geographically position the discovery site and automatically identify and contact the right person to collect the dead bird for laboratory tests. An algorithm was also implemented to help the telephone operator localize the birds when a civic address is not available. Spatial analysis is also performed automatically on each localized bird to determine if collection of the specimen is necessary, according to the number of positive birds already found in the neighbourhood.
Monitoring data – mosquitoes
This component of the system presents a control panel allowing for the transfer of laboratory results and for the calculation of infection rates of the monitored mosquito pools. Statistics are also available on temporal and regional counts of mosquitoes by species, and also on positive pools upon which infection rates are automatically computed using a software developed at the Center for Ecological Entomology, Illinois Natural History Survey (Gu Weidong, personal communication, 2003).
Monitoring data – other animals
This component of the system allows representatives of the Québec Ministère de l'Agriculture, des Pêcheries et de l'Alimentation (MAPAQ) to manually enter, validate, and visualize information related to infection cases in horses and other animals as necessary.
Graphical display – statistics
This component contains, for each monitoring domain, tables and graphics for the spatial (provincial and regional) and temporal distribution of monitoring data. The main statistics page presents the information at the provincial level. The users can then select a particular region for more specific analyses. It is also possible to perform custom analyses for any specified time period. Figure 3 presents an example of a statistical table summarizing information related to the reporting and laboratory tests on dead Corvidae, per region. It comprises a first column containing the regions, a second column containing the number of reports and a series of columns containing the status of the laboratory tests (not analyzed, negative, undetermined, pending or positive).
Figure 3 A tabular display example showing the statistics related to dead bird reporting per region in 2004.
Figure 4 presents an example of a chart representing the number of dead Corvidae reported (blue line) and the number of positive cases of infection for the year 2004 for birds (blue bars) and mosquitoes (yellow bars). There were no human or horse cases in 2004.
Figure 4 Bar chart example presenting the number of dead Corvidae reported and the number of positive cases of infection for the year 2004. This bar chart presents the number of dead Corvidae reported (blue line) and the number of positive cases of infection for the year 2004 for birds (blue bars) and mosquitoes (yellow bars). No cases of horses or humans were reported.
Cartographic display
This component of the ISPHM-WNV contains a cartographic interface allowing for the visualization and analysis of thematic maps including different layers of information that can be displayed according to specific user needs:
- Limits of the regional health authorities;
- Limits of the local health authorities;
- Administrative regions;
- Regional county municipalities;
- Cities and boroughs;
- Population density by dissemination (statistical) area;
- Topography;
- Hydrography;
- Vegetation zones;
- Road network;
- Rail network;
- Hospitals;
- National parks;
- Bird reports (for previous years and year-to-date, with laboratory tests results);
- Infected Corvidae (for previous years and year-to-date);
- Crow principal roost areas in late summer;
- Mosquito pools (for previous years and year-to-date, with laboratory tests results);
- Infected mosquito pools (for previous years and year-to-date);
- Mosquito monitoring stations;
- Human cases of infection (for previous years and year-to-date);
- Equine cases (for previous years and year-to-date);
- Other infected animals (for previous years and year-to-date);
- Larvicide and insecticide sprayed areas (for previous years and year-to-date);
- Planned larvicide and insecticide treatment areas (for previous years and year-to-date);
- Meteorological data.
The different maps can present individual punctual localization of events, or aggregated data according to the different territorial divisions (ex. regional health authorities). A specific interface has been developed to allow the automatic detection of spatio-temporal clusters of Corvidae deaths. This interface uses the SaTScan™ freeware [12] and provides a cartographic representation of the clusters found by SaTScan™. Specific tools have also been implemented in order to filter the data shown on the maps. For all relevant surveillance data, users can also select and download data in MS® Excel on their own workstation.
Figures 5 and 6 present examples of cartographic displays showing the different insecticide treatments and the localization of reported dead Corvidae and mosquito pools (along with their status), and the status of reported dead Corvidae by territory for the local health authorities around the Montreal Island, respectively. In Figure 5, the coral-shaded regions represent zones where the larvicide BTI has been used, or is planned to be used (white = no available information; light coral = checked area without treatment; dark coral = checked and treated area). The pink-shaded regions represent zones where Methoprene has been used, or is planned to be used (white = area not treated; pink = treated area). The green regions represent zones where larval control has taken place. The colored dots represent Corvidae reports (green = negative; white = not collected; grey = collected but not analyzed; red = positive; yellow = undetermined; purple = results pending). The colored triangles represent entomological stations (green = negative; pink = 1 positive pool; red = 2 to 4 positive pools; dark red = 5 to 11 positive pools).
Figure 5 Example of a thematic map showing the different insecticide treatments and the localization of bird reports and mosquito pools, in 2003. This thematic map shows the different insecticide treatments (shaded areas) and the localization of bird reports (colored dots) and mosquito pools (colored triangles) along with their status, in 2003.
Figure 6 Example of a thematic map showing the status of reported dead Corvidae around the Montreal Island, in 2005. This thematic map shows the status (represented by different colors) of reported dead Corvidae for different territorial subdivisions around the Montreal Island, in 2005.
In Figure 6, the pie charts represent test results of collected dead Corvidae (white = not collected; light blue = collected but not analyzed; grey = collection not necessary; red = positive; green = negative; purple = results pending; yellow = undetermined).
Management tools
This component of the system contains different tools for the management of the system. It allows the administrator to control user access to the various components of the system, including access to nominative and confidential data. It also compiles utilization statistics (number of accesses per page) and a log of all operations done by any user on the extranet website. These statistics are used to improve the ergonomics and functionalities of the system.
Use of the ISPHM-WNV for daily management
During the active season of the virus, the Expert Group is responsible for the daily assessment of the situation and meets regularly every week to propose interventions to regional authorities and the Ministry of Health, as necessary. The Advisory Committee only meets to assess major epidemic situations and recommend the needed interventions to the public health authorities.
Interventions are grouped in four categories based on virus activity over the previous three weeks:
- Absence of activity;
- Possible source of infection (infected Corvidae found);
- Probable source of infection (infected mosquitoes or both infected Corvidae and infected mosquitoes found);
- Confirmed source of infection (human cases detected).
A multi-criteria analysis grid is used to plan the different actions to be taken (Table 1).
Table 1 Summary of the 2004 analysis grid containing the possible actions according to the degree of virus activity.
Situation Possible actions
No virus activity - Verify the dead Corvidae reports
- Maintain preventive activities (media)
Possible source of infection (infected Corvidae found) - Target the geographical zone of activity and determine the population density
- Assess clusters of dead Corvidae
- Plan for mobile mosquito monitoring
- Closely monitor equine cases
- Maintain preventive activities (media)
Probable source of infection (infected mosquitoes or both infected Corvidae and infected mosquitoes found) - Target the geographical zone of activity and determine the population density
- Prioritise the analysis of dead Corvidae in the affected region
- Assess clusters of dead Corvidae
- Assess mosquito pool information (species and positivity)
- Closely monitor equine cases
- Integrate meteorological data into the analysis
- Consider insecticide treatments
- Increase preventive activities (media)
Confirmed source of infection (human cases detected) - Associate probabilities with potential acquisition site
- Target the geographical zone of activity and determine the population density
- Prioritise the analysis of dead Corvidae in the affected region
- Assess clusters of dead Corvidae
- Assess mosquito pool information (species and positivity)
- Closely monitor equine cases
- Integrate meteorological data into the analysis
- Consider insecticide treatments
- Increase preventive activities (media)
Results of the monitoring analyses are shared with the public. Every week, a map showing the current situation is produced and is published in the Flash-VNO bulletin by the INSPQ and disseminated to partner organizations and the media. This bulletin is also available within the WNV section of the INSPQ web site for public consultation. The data is also shared within Canada through the Canadian Network for Public Health Intelligence (CMPHI) and also appears in national summaries and websites. This can be especially useful in time of epidemic outbreak and for comparative evaluations.
Evaluation of the ISPHM-WNV
After each active season, an evaluation of the ISPHM-WNV is performed to collect and compile the comments of users with respect to the current version of the system and to collect suggestions for system improvements in order to prepare for the future season. A first formal evaluation based on theoretical models was conducted by interviews in 2003 [13]. 86% of the respondents found that the system was easy to use. All respondents noted that the system is very useful for monitoring of the epidemiologic situation, mainly for entomological, ornithological and human surveillance. 93% of the respondents planned to use the system in the following year (2004). According to the respondents' comments, the system provides many outcomes: it facilitates and speeds data access, it allows for simplified data analysis and for the precise planning of the control interventions, and it facilitates the sharing of information among the different actors [13]. As such, the system contributes to the main objectives of the public health protection plan adopted by the Government of Quebec [3].
Some identified limitations of the 2003 version of the system were the absence of an on-line help document, the absence of a process to notify the users of a significant update (ex. a confirmed human case), the difficulties in extracting data from the system, and the slow display speed for certain types of statistical tables and diagrams. Some of these limitations have already been addressed and others will be addressed in future versions of the system. The evaluation is now based on a simpler 13-question electronic questionnaire.
For 2004, a summary of the use of the system was produced using the access control module. 155 user accounts have been created for public health specialists of the MSSS, the INSPQ, the regional health authorities, and for external partners (ex. Health Canada specialists). 76% of the user accounts have been used at least once during the season. More than 58000 web pages have been accessed from May 30th to November 1st. The most frequently accessed section of the Extranet site was the Corvidae reports monitoring section, followed by the mosquito pools monitoring section and then by the cartographic interface. The cartographic tool was used at least once by 77% of the users.
On the data side, the system managed 2277 bird reports, 866 of which were analyzed. 112 of these reports were positive. Data about 8452 mosquito pools, 21 of which were positive, were also managed. To complete the statistics, 3 confirmed human cases (0 probable cases) and 0 infected horses (2 other infected animals) occurred.
The future of the ISPHM-WNV
Spatial on-line analytical processing (SOLAP)
The system will be enriched, in 2005, with spatial on-line analytical processing (SOLAP) capabilities. A SOLAP tool can be defined as a software that allows rapid and easy navigation within spatial databases and that offers many levels of information granularity, many themes, many epochs and many display modes, synchronized or not: maps, tables and diagrams [14]. SOLAP tools are based on multidimensional analysis. This characteristic allows for rapid and easy analyses of data according to multiple criteria (spatial, temporal and other attributes). SOLAP tools are currently in use in the public health field [15] and their advantages are numerous:
- They aim at supporting, transparently, the way humans think and analyze;
- They have a user interface that hides the complexity of query languages;
- They allow the users to focus on the results of the navigation rather than on the analysis process itself (i.e. focus on "what to obtain" rather than on "how to obtain it");
- Their response time is practically instantaneous;
- They incorporate the cartographic capabilities that most users cannot easily access otherwise.
The multidimensional analysis capabilities will involve a particular data structure that will be updated in real-time as new monitoring data become available.
Future research directions
A research project is currently conducted to incorporate predictive modeling in the ISPHM-WNV. The project is based on the Multi-Agent Geo-Simulation approach and aims at simulating the behaviours of mosquitoes and Corvidae that are linked to the spread and transmission of WNV. This simulation is expected to take place in a virtual mapping environment representing a large territory (the province of Quebec) and according to various climate scenarios and larvicide treatments. A preliminary study carried out in 2004 determined the feasibility of the project [16], and a functional prototype is now running. It will be tested and calibrated with the collaboration of other Canadian jurisdictions in 2005–2006.
Conclusion
This paper presents the Integrated System for Public Health Monitoring of West Nile Virus (ISPHM-WNV), its context, its architecture and its various functionalities as well as examples of its outputs. The ISPHM-WNV is an excellent example of integration of new technologies within the public health field.
This system rapidly became the essential tool for monitoring the activity of the West Nile virus in the province of Quebec. The system is very useful for field workers in all regions of the province, as well as for central authorities [10]. It speeds up the delivery of relevant information to all actors and simplifies the task of data analysis. It represents the common authoritative source of data for analysis, for exchange and for decision-making within the province. The INSPQ is now running the system for the 2005 season, planning to include spatial on-line analytical processing (SOLAP) functionalities to improve the facility and speed of analysis. This first successful implementation in Quebec of a real-time information system on a notifiable disease has had other positive outcomes, including the development of a similar system for other notifiable health events resulting from chemical exposures and other infectious diseases.
We believe our approach can be useful for other jurisdictions planning to implement or improve similar systems, either in Canada or elsewhere. Our system uses recent commercially available technologies and is being further developed to integrate emerging ones, such as spatial OLAP tools and other more recent approaches as they become available. Because of these characteristics, an implementation within another organisation would be done rapidly and the risks would be minimized.
List of abbreviations
BTI: Bacillus thuringiensis israelensis
CDC: Centers for Disease Control and Prevention
CHUQ: Centre hospitalier universitaire de Québec
CNPHI: Canadian network for public health intelligence
DYCAST: Dynamic Continuous-Area Space-Time
GIS: Geographic Information System
INSPQ: Institut national de santé publique du Québec
ISPHM-WNV: Integrated system for public health monitoring of West Nile virus
MAPAQ: Ministère de l'Agriculture, des Pêcheries et de l'Alimentation du Québec
MS: Microsoft
MSSS: Ministère de la Santé et des Services sociaux
SOLAP: Spatial on-line analytical processing
USGS: United States Geological Survey
WNV: West Nile virus
Authors' contributions
PG: Member of the Expert Group and the Advisory Committee in 2003, he was involved in the development and implementation of the system, and in the decision-making process. He participated in the writing and revision of the paper.
GL: He was in charge of the development and implementation of the system, and involved in the production of surveillance reports. Current member of the Expert Group and the Advisory Committee since 2003. He revised the paper.
SR: She was involved in the development of SOLAP and was responsible for the writing of the paper.
MDF: She was involved in the development of the system and has been in charge of the Expert Group and a member of the Advisory Committee since 2003 until now. She revised the paper.
==== Refs
Marr JS Calisher CH Alexander the Great and West Nile Virus Encephalitis Emerging Infectious Diseases 2003 9 1599 1603 14725285
Centers for Disease Control West Nile Virus Statistics, Surveillance and Control 2005
Ministère de la santé et des services sociaux du Québec Plan d'intervention gouvernemental de protection de la santé publique contre le virus du Nil occidental. Québec 2004
CBRN Research and Technology Initiative Canadian Network for Public Health Intelligence 2005
British Columbia Centre for Disease Control West Nile virus information page 2005
Centers for Disease Control and Prevention Epidemic/Epizootic West Nile Virus in the United States: Guidelines for Surveillance, Prevention, and Control, 3 rd revision 2003
Gotham IJ Eidson M White DJ Wallace BJ Chang HG Johnson GS Napoli JP Sottolano DL Birkhead GS Morse DL Smith PF West Nile virus: A Case study in how NY State health information infrastructure facilitates preparation and response to disease outbreaks Journal of Public Health Management and Practice 2001 7 79 89
Theophilides CN Ahearn SC Grady S Merlino M Identifying West Nile virus risk areas: the Dynamic Continuous-Area Space-Time system American Journal of Epidemiology 2003 157 843 854 12727678 10.1093/aje/kwg046
KHEOPS Technologies What is JMap? 2005
Lebel G Real-time GIS for West Nile Virus Surveillance, 2003 season GeoTec Event: Toronto, Canada 28–31 March 2004
DMTI Spatial CanMap Street Files 2005
SatScan SatScan software 2005
Bélanger D Roberge J Gosselin P Surveillance du virus du Nil occidental : Évaluation de l'utilisation du système intégré de données de vigie sanitaire Rapport de recherche INSPQ 2004
Rivest S Bédard Y Marchand P Towards better support for spatial decision-making: Defining the characteristics of Spatial On-Line Analytical Processing (SOLAP) Geomatica, the journal of the Canadian Institute of Geomatics 2001 55 539 555
Bédard Y Gosselin P Rivest S Proulx MJ Nadeau M Lebel G Gagnon MF Integrating GIS Components with Knowledge Discovery Technology for Environmental Health Decision Support International Journal of Medical Informatics 2003 70 79 94 12706184 10.1016/S1386-5056(02)00126-0
Bouden M Moulin B Gosselin P Back C Doyon B Gingras D Lebel G The Geosimulation of West Nile Virus Infection on the Basis Of Climate: A Tool for Risk Management in Public Health C-CIARN Conference: Montreal, Canada 4–7 May 2005
|
16159403
|
PMC1242354
|
CC BY
|
2021-01-04 16:39:06
|
no
|
Int J Health Geogr. 2005 Sep 13; 4:21
|
utf-8
|
Int J Health Geogr
| 2,005 |
10.1186/1476-072X-4-21
|
oa_comm
|
==== Front
J Exp Clin Assist ReprodJournal of Experimental & Clinical Assisted Reproduction1743-1050BioMed Central London 1743-1050-2-111613139810.1186/1743-1050-2-11ResearchPreliminary molecular genetic analysis of the Receptor Interacting Protein 140 (RIP140) in women affected by endometriosis Caballero Virginia [email protected] Rocío [email protected] José Antonio [email protected] Marina [email protected]ópez-Nevot Miguel Angel [email protected]án José Jorge [email protected] Luis Miguel [email protected] Castro Francisco [email protected]ópez-Villaverde Vicente [email protected] Agustín [email protected] Department of Structural Genomics. neoCodex. Averroes N°8. Edf. Acrópolis 110-1. 41020 Seville, Spain2 Unidad de Reproducción. Servicio de Obstetricia y Ginecología. Hospital de Valme, Ctra. Cádiz, s/n 41014 Seville, Spain3 Servicio de Análisis Clínicos. Hospital Universitario Virgen de las Nieves. Avda. Fuerzas Armadas, 2 18014 Granada, Spain4 Unidad de Reproducción Humana Asistida. Hospital Universitario Príncipe de Asturias. Ctra. Alcalá-Meco s/n. 28805 Madrid. Spain2005 30 8 2005 2 11 11 16 6 2005 30 8 2005 Copyright © 2005 Caballero et al; licensee BioMed Central Ltd.2005Caballero et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Endometriosis is a complex disease affecting 10–15% of women at reproductive age. Very few genes are known to be altered in this pathology. RIP140 protein is an important cofactor of oestrogen receptor and many other nuclear receptors. Targeting disruption experiments of nrip1 gene in mice have demonstrated that nuclear receptor interacting protein 1 gene (nrip1), the gene encoding for rip140 protein, is essential for female fertility. Specifically, mice null for nrip1 gene are viable, but females are infertile because of complete failure of mature follicles to release oocytes at ovulation stage. The ovarian phenotype observed in mice devoid of rip140 closely resembles the luteinized unruptured follicle (LUF) syndrome that is observed in a high proportion of women affected of endometriosis or idiopathic infertility. Here we present a preliminary work that analyses the role of NRIP1 gene in humans.
Methods
We have sequenced the complete coding region of NRIP1 gene in 20 unrelated patients affected by endometriosis. We have performed genetic association studies by using the DNA variants identified during the sequencing process.
Results
We identified six DNA variants within the coding sequence of NRIP1 gene, and five of them generated amino acid changes in the protein. We observed that three of twenty sequenced patients have specific combinations of amino-acid variants within the RIP140 protein that are poorly represented in the control population (p = 0.006). Moreover, we found that Arg448Gly, a common polymorphism located within NRIP1 gene, is associated with endometriosis in a case-control study (59 cases and 141 controls, pallele positivity test = 0.027).
Conclusion
Our results suggest that NRIP1 gene variants, separately or in combinations, might act as predisposing factors for human endometriosis.
==== Body
Background
Endometriosis (Online Mendelian Inheritance in Man (OMIM) 131200) is a complex disease affecting to 10–15% of women at reproductive age. The disease consists in pelvic pain and infertility due to the existence of endometrial glands and stroma outside the uterine cavity [1]. Anovulatory cycles and Luteinized Unruptured Follicle syndrome (LUF) are also evident in a great proportion of affected women [2,3]. Moreover, recurrent pregnancy losses, low quality of oocytes and early embryo loss in women with endometriosis have been suggested [4,5].
Receptor Interacting Protein 140 (RIP140) (Swiss-Prot P48552) is a high pleiotropic protein that acts as a co-regulator of multiple members of the nuclear receptor super-family including oestrogen, progesterone, retinoid acid or glucocorticoid receptors. Targeting disruption experiments of this function in mice have demonstrated that nuclear receptor interacting protein 1 (nrip1) gene (GenBank NM_173440), the gene encoding for rip140 protein, is essential for female fertility [3]. Specifically, mice null for nrip1 gene are viable, but females are infertile because of the complete failure of mature follicles to release oocytes at ovulation stage [3]. The ovarian phenotype observed in mice devoid of rip140 closely resembles the LUF syndrome that is observed in a high proportion of women affected by endometriosis or idiopathic infertility [2,3]. In addition, embryo transfer and ovarian transplantation experiments in nrip1 knock-out mice indicate slightly longer pregnancies in nrip1-/- mice and a high rate of foetal and neonatal losses of pups from mothers with nrip1-/- ovaries [6]. These data suggest that rip140 protein may have two functions in mice ovaries: i) an essential role in ovulation; ii) a secondary role in the maintenance of pregnancy [3,6]. More recently, a role for nrip1 gene in fat accumulation has been also proposed [6].
Due to nrip1-/-, female mice have several traits that closely resemble endometriosis findings; we decided to explore the role of the human NRIP1 gene (GenBank NM_003489) in women affected by endometriosis. Direct molecular analysis of endometriotic tissue specimens revealed no de novo mutations in 20 affected tissues. However, different germ-line genetic variants have been detected during our study. The involvement of these germline variants with endometriosis is proposed.
Methods
Patients
Endometriosis was defined according to the endometriosis classification system of the American Society for Reproductive Medicine (1996) [7]. All patients included correspond to stage III or IV of endometriosis. The initial sequencing project analyzed the complete sequence of NRIP1 gene in 20 independent DNA samples obtained from fresh endometriotic tissue derived from peritoneal implants or endometrioma lesions of 20 unrelated women with severe endometriosis. We also obtained fresh blood samples of these patients to test the germ-line or the somatic nature of the DNA variants detected.
To perform association studies between NRIP1 gene and human endometriosis, we genotype three groups of individuals. i) We increase the sample size of the case group (endometriosis group) three-fold using germ-line DNA derived from blood of 39 additional women affected by severe endometriosis (Stage III-IV). Therefore, the sample size of the endometriosis group for the association studies conducted was 59 (118 chromosomes). ii) To estimate population frequencies of mutations or polymorphism detected, 94 unselected and unrelated controls from the same geographical region were genotyped in an anonymous fashion (188 chromosomes). iii) A "super-control" group consisted of 47 healthy and fertile women without any sign or symptom of endometriosis, normal response to gonadotrophins and conserved ovulation (94 chromosomes) was also studied.
The ethnicity background of all probands and controls was Caucasian (white Europid) minimizing the possibility of population stratification in our case-control studies. The referral centers for this study are the Hospital de Valme (Seville), the Hospital Universitario Virgen de las Nieves (Granada), and the Hospital Príncipe de Asturias (Alcalá de Henares, Madrid). Informed consent was obtained from all patients. The institutional review board of referral centers has approved our research.
DNA extraction
We obtained 5 ml of peripheral blood from all patients to isolate germline DNA from leukocytes and about 100 μg of fresh endometriotic tissue during a programmed laparoscopic intervention of the patients. DNA extraction was performed according to standard procedures using Nucleospin Blood Kit (Macherey-Nagel) or alternative protocols. To perform Polymerase Chain Reactions (PCRs), we prepared aliquots of DNA at a concentration of 5 ng/μl. The rest of the stock was cryopreserved at -20°C.
Mutation analysis
NRIP1 cDNA was first cloned by Cavailles et al. [8]. This gene is mono-exonic, spans 7,239 base pairs (bp) and is located at 21q11.2. Genomic sequence containing NRIP1 gene was identified using the blat tool at UCSC Genome Bioinformatics server . Information concerning any Single Nucleotide Polymorphism (SNP) or mutation identified was compared with the UCSC Genome Bioinformatics server and also with the Single Nucleotide Polymorphism Database (dbSNP) at the National Centre for Biotechnology Information (NCBI) . According to standard mutation nomenclature [9], we employed the most frequent allele in the first position and the rarer allele in the last position.
We employed automated DNA sequencing methods to scan the entire coding sequence of NRIP1 gene in selected specimens. Overlapping PCRs covering the entire gene were designed and PCR products were purified and bi-directionally sequenced using the corresponding pair of primers (Table 1). Sequencing reactions were performed using the CEQ Dye Terminator Cycle Sequencing Quick Start Kit (Beckman Coulter, Inc) according to the manufacture's instructions. Fluorograms were analyzed on CEQ™ 8000 Genetic Analysis System following the manufacturer's instructions (Beckman Coulter, Inc).
Table 1 Amplification primer sequences and PCR product size. NRIP1 is a monoexonic gene. We designed eight overlapping amplicons to cover the entire coding sequence of this gene.
PRIMER SEQUENCE 5' → 3' PCR product size (bp)
1F TTCTAGTTCTGCCTCCTTAAC 554
1R ACATTTCTGGCAGTGCATTTC
2F GATCAGGTACTGCCGTTGA 528
2R CGAATCTTCCTGATGTGACT
3F GTGCTATGGTGTTGCATCAAG 572
3R TGCAGGTTATAAGAACTCACTGG
4F CATCATCAAGCAAACTGATGGC 577
4R AGCCCTCAGGGAGTACACAA
5F CTTCAATTGCTACTTGGCCAT 582
5R GTAGTCAACCAACAGGTCCT
6F CTGGAAACACAGATAAACCGATAGG 584
6R TGGCACTTCTAGAATCAAAG
7F AGATAGTTACCTGGCAGATG 572
7R TCCTACTTTCCCTGAGCACT
8F CAGTTGCATGGATAACAGGA 645
8R GTATTGGTTACTGGTGATG
Genotyping
To verify the DNA variants detected during the sequencing process and to perform association studies, we employed Fluorescent Resonance Energy Transfer (FRET) protocols. We designed and synthesized amplification primers and fluorescent detection probes for all the DNA variants identified within the NRIP1 gene. The selected primer pairs and detection probes are summarized in Tables 1 and 2. Real-time PCR was performed in the LightCycler system (Roche Applied Science) using reaction conditions previously published by us [10].
Table 2 Anchor and Sensor probes sequences employed for coding Single Nucleotide Polymorphism (cSNP) analysis using Fluorescence Resonance Energy Transfer (FRET) technology.
Mutation PROBE SEQUENCE 5' → 3
Gly75Gly 75-ANCHOR AGTAATGGTCCAGTTCTCAATACAC – F
75-SENSOR Cy5 – TACATATCAGGGGTCTGGC – Ph
His221Arg 221-ANCHOR Cy5 – AGTGGAACAAAGGTCATGAGTGAAC – Ph
221-SENSOR TCTCCTCATCATGTTGGACA – F
Ile441Val 448-ANCHOR TATTCCAACTGTGTTCCCATAGACT – F
Arg448Gly 448-SENSOR Cy5 – GTCTTGCAAACACCGAACTG – Ph
Ser803Leu 803-ANCHOR GCGCACCTGCCTTACCAGTGTCCCGA – F
803-SENSOR Cy5-GACTTTAAATCGGAGCCTGTT – Ph
Val1079Phe 1079-ANCHOR Cy5 – CGAGAAACACAAGACAAGGACATTT – Ph
1079-SENSOR GGAGGCAATTCTGTTACCAG – F
Nomenclature: F: Fluoresceine, Ph: Phosphate.
The conditions to obtain optimal melting curves for FRET analysis and spectrofluorimetric genotypes were 95°C for 0 s, 63°C for 25 s, 45°C for 0 s and 80°C for 0 s (with a temperature-transfer speed of 20°C/s in each step, except the last step, in which the speed of temperature transfer was 0.1°C/s). In the last step, a continuous fluorometric register was performed fixing the gains of the system at 1, 50, and 50 on channels F1, F2, and F3 respectively. Genotype results using real time-PCR are shown in Figure 1a. To test the specificity of these assays, selected amplicons of different melting patterns were re-sequenced using an automated DNA sequencer (Beckman Coulter CEQ 2000XL, data not shown).
Figure 1 Detection of germ line variants in NRIP1 in patients with severe endometriosis. A) Spectrofluorimetric analysis of NRIP1 gene using real-time PCR. Analysis of the fluorescence measured during melting curve determination in the LightCycler (Roche Applied Science). Each allele has a specific melting point and all alleles are represented by its specific nucleotide change with the exception of Ile441Val and Arg448Gly polymorphisms. Nt c.512 G->A (Gly75Gly, melting points, Allele G: 62°C; Allele A: 57°C). Nt c.949 A->G (His221Arg, melting points, Allele A 61°C; Allele G: 56°C). Nt c.1608 A->G (Ile441Val) and Nt c.1629 C->G (Arg448Gly) (melting points, Allele Val441: 59°C; Allele Gly448: 56°C; wild type: 63°C). Nt c.2695 C->T (Ser803Leu, melting points, Allele C: 61°C; Allele T: 55°C). Nt c.3522 G->T (Val1079Phe, melting points, Allele G: 62°C; Allele T: 54°C). B) Sequence conservation and location of mutations in the RIP-140 protein. Black shading indicates the position of mutations. LXXLL motifs responsible for ligand independent interaction with Retinoid Acid Receptor (RAR) and Retinoid X Receptor (RXR) are in bold and underlyned. Signal peptide is depicted in blue, Carboxyl terminal binding protein (CTBP) and RAR interacting motifs are in red and green respectively. Low complexity regions are shown in grey.
Statistical Analysis
To compare allele and genotype frequencies between patients, control and super-control groups, we performed conventional chi-square tests with Yates correction or Fisher exact test using Statcalc (EpiInfo 5.1, Center for Disease control, Atlanta, GA). For statistical analysis of genotype distribution, test for deviation of Hardy-Weinberg equilibrium or two-point association studies, we employed six different tests adapted from Sasieni (deviation from Hardy-Weinberg equilibrium, allele frequency differences test, heterozygous test, homozygous test, allele positivity test and Armitage's trend test) [11]. These calculations were performed in the online resource at the Institute for Human Genetics, Munich, Germany . Significant thresholds for statistical studies were fixed at p < 0.05.
Results
Looking for somatic mutations within the NRIP1 gene, we determined the complete coding sequence of the candidate gene in 20 selected and unrelated somatic endometriotic tissues using bi-directional automated capillary DNA sequencing. In our primary sequencing project we finished 80,600 bp of DNA. Using our methodology, we identified six single nucleotide DNA variants within the coding sequence of the NRIP1 gene in various unrelated somatic DNA samples. Two of these variants have been previously identified and they are included in the Single Nucleotide Polymorphism Database (dbSNP) at the National Centre for Biotechnology Information (NCBI) (Table 3). Five of these mutations alter the amino acid coding sequence of RIP140 protein generating missense mutations (Fig. 1 and Table 3). Although all mutations were detected in somatic DNA, direct molecular analyses of the corresponding blood samples of mutated tissue also contain the same DNA change. This last result implies that the genetic variants identified are germ-line and, consequently, somatic mutations at NRIP1 locus are not commonly involved in the pathogenesis of human endometriosis.
Table 3 Summary of DNA variants observed within the coding sequence of the NRIP1 gene.
DNA variant* Amino acid Substitution (change in codon) dbSNP** accession number Detection in Endometriotic tissue samples (40 chromosomes) Detection in germline DNA derived from endometriosis patients (118 chromosomes) Detection in controls (282 chromosomes) Status
Nt c.512 G->A None [Gly75] (ggg to gga) rs2229741 15/40 57/118 129/282 Common polymorphism
Nt c.949 A->G His221Arg (cat to cgt) - 1/40 1/118 3/282 Common polymorphism
Nt c.1608 A->G Ile441Val (ata to gta) - 0/20 0/118 4/282 Common polymorphism
Nt c.1629 C->G Arg448Gly (cga to gga) rs2229742 9/40 16/118 19/282 Common polymorphism
Nt c.2695 C->T Ser803Leu (tcg to ttg) - 2/40 3/118 10/282 Common polymorphism
Nt c.3522 G->T Val1079Phe (gtt to ttt) - 1/40 1/118 0/282 Rare Variant/Mutation
*In accordance with genbank number NM_003489.
** dbSNP: the Single Nucleotide Polymorphism Database at the National Centre for Biotechnology Information (NCBI)
To evaluate the polygenic role of NRIP1 gene variants in human endometriosis, we decided to preliminary explore the allelic frequencies and genotypes of these mutations in women affected by endometriosis and unselected controls. To conduct genetic association studies, we developed real-time PCR detection protocols using FRET probes for each DNA mutation identified at NRIP1 locus. Using these techniques, we genotyped the mutations in 200 unrelated women (59 endometriosis patients, 94 unselected controls and 47 super-control women). Overall, 400 different chromosomes have been scored for each DNA variant (Table 4).
Table 4 Association studies of common DNA variants of the NRIP1 gene in relation to human endometriosis.
NRIP1 polymorphism (change in codon) Genotypes Patients (n = 59) Unselected Controls (n = 94) Super Controls (n = 47) All Controls (n = 141) Statistical Analysis*
Gly75Gly (ggg to gga) aa 15 19 9 28 P = 0.34 (Heterozygous test)
ag 27 46 27 73
gg 17 28 11 39
Arg448Gly (cga to gga) cc 44 83 40 123 P = 0.027 (Allele positivity test)
cg 14 10 7 17
gg 1 1 0 1
Ser803Leu (tcg to ttg) cc 56 85 46 130 P = 0.59 (Armitage's trend test)
ct 3 9 1 10
tt 0 0 0 0
*Compares patients versus merged controls. Best p value employing tests for genetic association according to Sasieni (1997). Ile441Val and His221Arg are not analyzed due to small or null sample size in endometriosis samples.
By analyzing the allelic frequencies of DNA variants detected in Spanish population, we classified these variants as common polymorphisms if observed in >1% of chromosomes in controls (Gly75Gly, His221Arg, Ile441Val, Arg448Gly and Ser803Leu) or rare variants if observed with a frequency <1% of chromosomes in controls (Table 3). In contrast, Val1079Phe allele appears only in a single patient in heterozygous state and none of 141 controls. This data could suggest its involvement in the disease. Reinforcing this hypothesis, Val1079Phe is located close to high-conserved domain of the carboxylic end of RIP140 protein that interacts with retinoic acid nuclear receptor (Fig. 1).
Direct inspection of genotypes in patients revealed three genotype patterns within the NRIP1 gene that appear to be over-represented in women affected by endometriosis (p = 0.006, Fisher exact test). The patterns consist of a combination of Arg448Gly together with His221Arg or Val1079Phe variants. We identified three unrelated women affected by endometriosis carrying double heterozygotes (His221Arg/Arg448Gly and Val1079Phe/Arg448Gly) or homozygote (Arg448Gly/Arg448Gly) genotypes for these alleles, respectively. The homozygote (Arg448Gly/Arg448Gly) genotype pattern appeared only in 1 of 94 unselected controls (p = 0.016 Fisher exact test) and none of 47 super-control women (p = 0.023, Fisher Exact test), whereas double heterozygotes genotypes did not appear in any control individual. These results suggest that specific combinations of amino acid changes at NRIP1 locus could be related to endometriosis etiology with a 99.4% of reliability, although given the scarce sample size the presence of polygenes within NRIP1 locus must be proven with a larger and independent re-analysis.
Finally, given the preliminary results, we conducted an small case-control study analyzing all common variants detected within the NRIP1 locus (Gy75Gly, Arg448Gly and Ser803Leu). Table 4 shows the results for those test that maximize the differences between case and control groups for each polymorphism. Genotypic distributions of polymorphisms analyzed are in accordance with the Hardy-Weinberg equilibrium law (p > 0.15), indicating no bias due to technical or stratification problems nor evolution-dependent genetic sweep/selection events (data not shown). Interestingly, our analysis revealed that Arg448Gly polymorphism appears to be weakly associated with endometriosis in our population (Odds ratio = 2.327, pallele positivity test = 0.027). In contrast, no significant association could be achieved when comparing unselected versus super-control women, supporting the accuracy of the selected control panel (p > 0.34 for Gly75Gly, p > 0.41 for Arg448Gly, and p > 0.1 for Ser804Leu).
Overall, our results might support the role of NRIP1 gene in endometriosis, although given the small sample size, we propose an extensive re-analysis by increasing the sample size to confirm our results.
Discussion
Endometriosis is a complex disease affecting 10–15% of women at reproductive age. Very few genes are known to be altered in this pathology. Molecular genetic analyses provide some evidence of genetic association in case-control studies analyzing Estrogen Receptor 1 (ESR1 OMIM 133430) and Cytochrome P450, Family 19, Subfamily A, Polypeptide 1 (CYP19 OMIM 107910) genes. Interestingly, both loci are involved in oestrogen mechanism of production and action [12,13]. In addition, other nuclear receptor genes, such as Progesterone Receptor (PGR OMIM 607311) and Peroxisome Proliferative Activated Receptor, Gamma (PPARG OMIM 601487) gene have been associated with endometriosis in other case-control studies [14,15]. The involvement in endometriosis of loci related to detoxification has been also studied and replicated [16-18].
Given these preliminary findings and the importance of steroid receptors in uterine physiology [19] and endometriosis pathogenesis [1,20], the biochemical pathways involved in steroids production, degradation or mechanisms of action appear to be strong candidates for endometriosis etiology and many other phenotypes related to human fertility.
Following this working hypothesis, targeting disruption of nuclear receptors and their regulators such as nrip1 or CCR4-NOT transcription complex, subunit 7 (cnot7 GenBank AK009561) in animal models have provided direct evidence of the importance of nuclear receptor homeostasis in male and female reproduction [3,21-24].
Here we present the first structural analysis of the human NRIP1 gene in relation to human disease. It is of interest to mention that the dbSNP includes 26 SNPs for NRIP1 gene currently. Eighteen of these variants are located within the 3'untranslated region (3'UTR), this genomic region has not been covered in this study, and the remaining ones are coding SNPs. According to GenBank, only three SNPs in the 3'UTR region and two coding SNPs have been validated in population based studies including more than 150 chromosomes. The rest of the SNPs are the result of the bioinformatic alignment of different cDNA and genomic clones. The allele frequencies here presented for Gly75Gly (dbSNP rs2229741) and Arg448Gly (dbSNP rs2229742) polymorphisms are very similar to those included in dbSNP (data not shown).
Overall, our results are preliminary providingt suggestive, but not definitive, evidence of NRIP1 gene involvement in human endometriosis. We think that conclusive proofs of involvement will be achieved throughout re-analyses of this study in independent cohorts of patients and controls, rather than performing functional analyses of the missense mutations observed. The detection of functionality of DNA variants involved in complex traits such endometriosis, is near to be impossible using conventional technologies because the effect from single genetic variant/mutation is expected to be very small and it is only the joint effect of several susceptibility genes that leads to the disease [25]. In this sense, we are currently recruiting a higher number of patients and controls to perform a proper re-analysis of our results.
Regarding Arg448Gly polymorphism, we propose that the variant could act as a low penetrance allele related to human endometriosis. The molecular mechanism of this mutation is not well understood, although its location and degree of conservation provide some interesting clues. In fact, Arg448 residue of RIP140 protein is completely conserved among humans, rats, mice, gallus and xenopus (Fig. 1b). Moreover, the non-conservative substitution detected (Arg448Gly) might affect the Carboxyl terminal binding protein (CTBP) interacting motif of RIP140 protein that is located close to this amino acid residue (Fig. 1b).
On the basis of genotype analysis in affected women, we propose that Arg448Gly mutation could act in concert with other genetic variants within NRIP1 or other loci. In this way, we found a single woman affected by endometriosis simultaneously carrying Val1079Phe mutation and Arg448Gly polymorphism both in a heterozygous state. Val1079Phe also arises in an inter-specific conserved residue. Moreover, this mutation is located close to (and may disrupt) the retinoid acid receptor interacting motif "LTKTNPILYYMLQK" of RIP140 protein (Fig. 1b). Supporting its involvement in the disease, we have not identified the Val1079Phe mutation in 282-control chromosomes. Intriguingly, retinoid acid receptors alpha, gamma 2 and its regulator cnot7 have been involved in male sterility [24,26,27]. Moreover, the presence of multiple specific functional rare variants in affected patients have been recently proposed and evaluated [28]. This hypothesis is an alternative to explain the genetic component of complex traits in front of the widely accepted common disease common variant hypothesis [29].
Finally, we identified a single patient carrying a unique genotype combination comprising His221Arg rare variant and, again, Arg448Gly polymorphism. The absence of inter-specific amino acidic conservation and the inexistence of known functional domains close to His221Arg variant do not support the functionality of His221Arg allele. However, His221Arg only appears combined with Arg448Gly in a woman affected by endometriosis. This combination never appears in 141 unrelated controls. Functional assays or large cohorts analyses will help to elucidate its involvement in this pathology.
Our results support that NRIP1 gene might contain alleles related to endometriosis in humans. NRIP1 gene encodes a highly pleiotropic nuclear receptor co-regulator (RIP140) [30]. This protein interacts and regulates multiple members of the nuclear receptor super-family. Some of them have been associated with endometriosis. The wide repertoire of RIP140 targets might explain the complex pathological findings observed in human endometriosis. In fact, a recent report revealed a complex mechanism, involved in endometriosis and other oestrogen-related traits, by which ER-mediated oestrogen signaling is modulated by a co-regulatory-like function of activated AhR/Arnt dioxin receptor complex, giving rise to adverse oestrogen-related actions of dioxin-type environmental contaminants [31]. Intriguingly, oestrogen and dioxin nuclear receptor pathways are modulated by RIP140 protein [32,33].
According to previous data, mouse models reports and the present results, NRIP1 gene appears to be an attractive gene for human endometriosis etiology and other related pathologies. In addition, we have found a genetic interaction of NRIP1 gene with Estrogen receptor alpha and beta (ESR1 and ESR2) genes in two estrogen-dependent diseases such as male infertility [34] and osteoporosis (manuscript in preparation). For these reasons, further evaluation of NRIP1 gene in many oestrogen-related phenotypes is warranted.
Conclusion
Our results suggest that NRIP1 gene variants, separately or in combinations, might act as predisposing factors for human endometriosis. According to previous data, mouse models reports and the present results, NRIP1 appears to be an attractive candidate gene for human endometriosis etiology and other estrogen-related pathologies.
Competing interests
The author(s) declare that they have not competing interests.
Authors' contributions
VC, JAS, MC, MAL, FC and VL carried out the recruitment and classification of patients and controls and the biological samples management. RR and JJG carried out the molecular genetic studies and participate in the analysis and interpretation of data. LMR and AR carried out the design of the study, performed the statistical analyses and the interpretation of data.
All authors have been involved in drafting the article or revising it critically for important intellectual content and have given final approval of the version to be published.
Acknowledgements
We are deeply grateful to patients and controls for participation in this study. Funded by the Ministerio de Ciencia y Tecnología (MCYT, Spain, grant numbers FIT-010000-2003-36, FIT-010000-2003-89, FIT-010000-2003-70, PTQ2002-0206) and Organón Española S.A. (BCN, Spain).
==== Refs
Bulum SE Zeitoun KM Takayama K Sasano H Estrogen biosynthesis in endometriosis: molecular basis and clinical relevance J Mol Endocrinol 2000 25 35 42 10915216 10.1677/jme.0.0250035
Donnez J Thomas K Incidence of luteinized unruptured follicle syndrome in fertile women and women with endometriosis Eur J Obstet Gynecol Reprod Biol 1982 14 187 190 7160529 10.1016/0028-2243(82)90095-8
White R Leonardsson G Rosewell I Ann Jacobs M Milligan S Parker M The nuclear receptor co-repressor Nrip1 (RIP140) is essential for female fertility Nat Med 2000 6 1368 1374 11100122 10.1038/82183
García-Velasco JA Arici A Is the endometrium or oocyte/embryo affected in endometriosis? Hum Reprod 1999 2 77 89
Vercammen EE D'Hooghe TM Endometriosis and recurrent pregnancy loss Semin Reprod Med 2000 18 363 368 11355795 10.1055/s-2000-13726
Leonardsson G Steel JH Christian M Pocock V Milligan S Bell J So PW Medina-Gomez G Vidal-Puig A White R Parker MG Nuclear receptor corepressor RIP140 regulates fat accumulation Proc Natl Acad Sci U S A 2004 101 8437 42 15155905 10.1073/pnas.0401013101
American Society for Reproductive Medicine Revised American Society for Reproductive Medicine classification of endometriosis: 1996 Fertil Steril 1997 67 817 821 9130884 10.1016/S0015-0282(97)81391-X
Cavailles V Dauvois S L'Horset F Lopez G Hoare S Kushner PJ Parker MG Nuclear factor RIP140 modulates transcriptional activation by the estrogen receptor EMBO J 1995 14 3741 3751 7641693
den Dunnen JT Antonarakis SE Nomenclature for the description of human sequence variations Hum Genet 2001 109 121 124 11479744 10.1007/s004390100505
Gonzalez A Abril E Roca A Aragon MJ Figueroa MJ Velarde P Royo JL Real LM Ruiz A CAPN10 alleles are associated with polycystic ovary syndrome J Clin Endocrinol Metab 2002 87 3971 3976 12161543 10.1210/jc.87.8.3971
Sasieni PD From genotypes to genes: doubling the sample size Biometrics 1997 53 1253 1261 9423247
Georgiou I Syrrou M Bouba I Dalkalitsis N Paschopoulos M Navrozoglou I Lolis D Association of estrogen receptor gene polymorphisms with endometriosis Fertil Steril 1999 72 64 66 10.1016/S0015-0282(99)00198-3
Kado N Kitawaki J Obayashi H Ishihara H Koshiba H Kusuki I Tsukamoto K Hasegawa G Nakamura N Yoshikawa T Honjo H Association of the CYP17 gene and CYP19 gene polymorphisms with risk of endometriosis in Japanese women Hum Reprod 2002 17 897 902 11925378 10.1093/humrep/17.4.897
Wieser F Schneeberger C Tong D Tempfer C Huber JC Wenzl R PROGINS receptor gene polymorphism is associated with endometriosis Fertil Steril 2002 77 309 312 11821088 10.1016/S0015-0282(01)02984-3
Dogan S Machicao F Wallwiener D Haering HU Diedrich K Hornung D Association of peroxisome proliferator-activated receptor gamma 2 Pro-12-Ala polymorphism with endometriosis Fertil Steril 2004 81 1411 1413 15136115 10.1016/j.fertnstert.2003.10.026
Baranova H Canis M Ivaschenko T Albuisson E Bothorishvilli R Baranov V Malet P Bruhat MA Possible involvement of arylamine N-acetyltransferase 2, glutathione S-transferases M1 and T1 genes in the development of endometriosis Mol Hum Reprod 1999 5 636 641 10381818 10.1093/molehr/5.7.636
Arvanitis DA Goumenou AG Matalliotakis IM Koumantakis EE Spandidos DA Low-penetrance genes are associated with increased susceptibility to endometriosis Fertil Steril 2001 76 1202 1206 11730751 10.1016/S0015-0282(01)02865-5
Arvanitis DA Koumantakis GE Goumenou AG Matalliotakis IM Koumantakis EE Spandidos DA CYP1A1, CYP19, and GSTM1 polymorphisms increase the risk of endometriosis Fertil Steril 2003 702 709 12620480 10.1016/S0015-0282(02)04817-3
Moutsatsou P Sekeris CE Steroid receptors in the uterus: implications in endometriosis Ann N Y Acad Sci 2003 997 209 222 14644828 10.1196/annals.1290.024
Kitawaki J Kado N Ishihara H Koshiba H Kitaoka Y Honjo H Endometriosis: the pathophysiology as an estrogen-dependent disease J Steroid Biochem Mol Biol 2002 83 149 155 12650711 10.1016/S0960-0760(02)00260-1
Kastner P Mark M Leid M Gansmuller A Chin W Grondona JM Decimo D Krezel W Dierich A Chambon P Abnormal spermatogenesis in RXR beta mutant mice Genes Dev 1996 10 80 92 8557197
Dupont S Krust A Gansmuller A Dierich A Chambon P Mark M Effect of single and compound knockouts of estrogen receptors alpha (ERalpha) and beta (ERbeta) on mouse reproductive phenotypes Development 2000 127 4277 4291 10976058
Cui Y Miyoshi K Claudio E Siebenlist UK Gonzalez FJ Flaws J Wagner KU Hennighausen L Loss of the peroxisome proliferation-activated receptor gamma (PPARgamma) does not affect mammary development and propensity for tumor formation but leads to reduced fertility J Biol Chem 2002 17 17830 17835 11884400 10.1074/jbc.M200186200
Nakamura T Yao R Ogawa T Suzuki T Ito C Tsunekawa N Inoue K Ajima R Miyasaka T Yoshida Y Ogura A Toshimori K Noce T Yamamoto T Noda T Oligo-astheno-teratozoospermia in mice lacking Cnot7, a regulator of retinoid X receptor beta Nat Genet 2004 36 528 533 15107851 10.1038/ng1344
Hoh J Ott J Genetic dissection of diseases: design and methods Curr Opin Genet Dev 2004 14 229 232 15172663 10.1016/j.gde.2004.04.006
Lohnes D Kastner P Dierich A Mark M LeMeur M Chambon P Function of retinoic acid receptor gamma in the mouse Cell 1993 21 643 658 8388780 10.1016/0092-8674(93)90246-M
Dufour JM Vo MN Bhattacharya N Okita J Okita R Kim KH Peroxisome proliferators disrupt retinoic acid receptor alpha signaling in the testis Biol Reprod 2003 68 1215 1224 12606456 10.1095/biolreprod.102.010488
Cohen JC Kiss RS Pertsemlidis A Marcel YL McPherson R Hobbs HH Multiple rare alleles contribute to low plasma levels of HDL cholesterol Science 2004 305 869 872 15297675 10.1126/science.1099870
Lohmueller KE Pearce CL Pike M Lander ES Hirschhorn JN Meta-analysis of genetic association studies supports a contribution of common variants to susceptibility to common disease Nat Genet 2003 33 177 182 12524541 10.1038/ng1071
Zilliacus J Holter E Wakui H Tazawa H Treuter E Gustafsson JA Regulation of glucocorticoid receptor activity by 14-3-3-dependent intracellular relocalization of the corepressor RIP140 Mol Endocrinol 2001 15 501 511 11266503 10.1210/me.15.4.501
Ohtake F Takeyama K Matsumoto T Kitagawa H Yamamoto Y Nohara K Tohyama C Krust A Mimura J Chambon P Yanagisawa J Fujii-Kuriyama Y Kato S Modulation of oestrogen receptor signalling by association with the activated dioxin receptor Nature 2003 423 545 550 12774124 10.1038/nature01606
Kumar MB Tarpey RW Perdew GH Differential recruitment of coactivator RIP140 by Ah and estrogen receptors. Absence of a role for LXXLL motifs J Biol Chem 1999 274 22155 22164 10428779 10.1074/jbc.274.32.22155
Peters GA Khan SA Estrogen receptor domains E and F: role in dimerization and interaction with coactivator RIP-140 Mol Endocrinol 1999 13 286 296 9973258 10.1210/me.13.2.286
Galan JJ Buch B Cruz N Segura A Moron FJ Bassas L Martinez-Pineiro L Real LM Ruiz A Multilocus analyses of estrogen-related genes reveal the involvement of ESR1 gene in male infertility and the polygenic nature of the pathology Fertil Steril
|
16131398
|
PMC1242355
|
CC BY
|
2021-01-04 16:40:17
|
no
|
J Exp Clin Assist Reprod. 2005 Aug 30; 2:11
|
utf-8
|
J Exp Clin Assist Reprod
| 2,005 |
10.1186/1743-1050-2-11
|
oa_comm
|
==== Front
J Exp Clin Assist ReprodJournal of Experimental & Clinical Assisted Reproduction1743-1050BioMed Central London 1743-1050-2-111613139810.1186/1743-1050-2-11ResearchPreliminary molecular genetic analysis of the Receptor Interacting Protein 140 (RIP140) in women affected by endometriosis Caballero Virginia [email protected] Rocío [email protected] José Antonio [email protected] Marina [email protected]ópez-Nevot Miguel Angel [email protected]án José Jorge [email protected] Luis Miguel [email protected] Castro Francisco [email protected]ópez-Villaverde Vicente [email protected] Agustín [email protected] Department of Structural Genomics. neoCodex. Averroes N°8. Edf. Acrópolis 110-1. 41020 Seville, Spain2 Unidad de Reproducción. Servicio de Obstetricia y Ginecología. Hospital de Valme, Ctra. Cádiz, s/n 41014 Seville, Spain3 Servicio de Análisis Clínicos. Hospital Universitario Virgen de las Nieves. Avda. Fuerzas Armadas, 2 18014 Granada, Spain4 Unidad de Reproducción Humana Asistida. Hospital Universitario Príncipe de Asturias. Ctra. Alcalá-Meco s/n. 28805 Madrid. Spain2005 30 8 2005 2 11 11 16 6 2005 30 8 2005 Copyright © 2005 Caballero et al; licensee BioMed Central Ltd.2005Caballero et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Endometriosis is a complex disease affecting 10–15% of women at reproductive age. Very few genes are known to be altered in this pathology. RIP140 protein is an important cofactor of oestrogen receptor and many other nuclear receptors. Targeting disruption experiments of nrip1 gene in mice have demonstrated that nuclear receptor interacting protein 1 gene (nrip1), the gene encoding for rip140 protein, is essential for female fertility. Specifically, mice null for nrip1 gene are viable, but females are infertile because of complete failure of mature follicles to release oocytes at ovulation stage. The ovarian phenotype observed in mice devoid of rip140 closely resembles the luteinized unruptured follicle (LUF) syndrome that is observed in a high proportion of women affected of endometriosis or idiopathic infertility. Here we present a preliminary work that analyses the role of NRIP1 gene in humans.
Methods
We have sequenced the complete coding region of NRIP1 gene in 20 unrelated patients affected by endometriosis. We have performed genetic association studies by using the DNA variants identified during the sequencing process.
Results
We identified six DNA variants within the coding sequence of NRIP1 gene, and five of them generated amino acid changes in the protein. We observed that three of twenty sequenced patients have specific combinations of amino-acid variants within the RIP140 protein that are poorly represented in the control population (p = 0.006). Moreover, we found that Arg448Gly, a common polymorphism located within NRIP1 gene, is associated with endometriosis in a case-control study (59 cases and 141 controls, pallele positivity test = 0.027).
Conclusion
Our results suggest that NRIP1 gene variants, separately or in combinations, might act as predisposing factors for human endometriosis.
==== Body
Background
Endometriosis (Online Mendelian Inheritance in Man (OMIM) 131200) is a complex disease affecting to 10–15% of women at reproductive age. The disease consists in pelvic pain and infertility due to the existence of endometrial glands and stroma outside the uterine cavity [1]. Anovulatory cycles and Luteinized Unruptured Follicle syndrome (LUF) are also evident in a great proportion of affected women [2,3]. Moreover, recurrent pregnancy losses, low quality of oocytes and early embryo loss in women with endometriosis have been suggested [4,5].
Receptor Interacting Protein 140 (RIP140) (Swiss-Prot P48552) is a high pleiotropic protein that acts as a co-regulator of multiple members of the nuclear receptor super-family including oestrogen, progesterone, retinoid acid or glucocorticoid receptors. Targeting disruption experiments of this function in mice have demonstrated that nuclear receptor interacting protein 1 (nrip1) gene (GenBank NM_173440), the gene encoding for rip140 protein, is essential for female fertility [3]. Specifically, mice null for nrip1 gene are viable, but females are infertile because of the complete failure of mature follicles to release oocytes at ovulation stage [3]. The ovarian phenotype observed in mice devoid of rip140 closely resembles the LUF syndrome that is observed in a high proportion of women affected by endometriosis or idiopathic infertility [2,3]. In addition, embryo transfer and ovarian transplantation experiments in nrip1 knock-out mice indicate slightly longer pregnancies in nrip1-/- mice and a high rate of foetal and neonatal losses of pups from mothers with nrip1-/- ovaries [6]. These data suggest that rip140 protein may have two functions in mice ovaries: i) an essential role in ovulation; ii) a secondary role in the maintenance of pregnancy [3,6]. More recently, a role for nrip1 gene in fat accumulation has been also proposed [6].
Due to nrip1-/-, female mice have several traits that closely resemble endometriosis findings; we decided to explore the role of the human NRIP1 gene (GenBank NM_003489) in women affected by endometriosis. Direct molecular analysis of endometriotic tissue specimens revealed no de novo mutations in 20 affected tissues. However, different germ-line genetic variants have been detected during our study. The involvement of these germline variants with endometriosis is proposed.
Methods
Patients
Endometriosis was defined according to the endometriosis classification system of the American Society for Reproductive Medicine (1996) [7]. All patients included correspond to stage III or IV of endometriosis. The initial sequencing project analyzed the complete sequence of NRIP1 gene in 20 independent DNA samples obtained from fresh endometriotic tissue derived from peritoneal implants or endometrioma lesions of 20 unrelated women with severe endometriosis. We also obtained fresh blood samples of these patients to test the germ-line or the somatic nature of the DNA variants detected.
To perform association studies between NRIP1 gene and human endometriosis, we genotype three groups of individuals. i) We increase the sample size of the case group (endometriosis group) three-fold using germ-line DNA derived from blood of 39 additional women affected by severe endometriosis (Stage III-IV). Therefore, the sample size of the endometriosis group for the association studies conducted was 59 (118 chromosomes). ii) To estimate population frequencies of mutations or polymorphism detected, 94 unselected and unrelated controls from the same geographical region were genotyped in an anonymous fashion (188 chromosomes). iii) A "super-control" group consisted of 47 healthy and fertile women without any sign or symptom of endometriosis, normal response to gonadotrophins and conserved ovulation (94 chromosomes) was also studied.
The ethnicity background of all probands and controls was Caucasian (white Europid) minimizing the possibility of population stratification in our case-control studies. The referral centers for this study are the Hospital de Valme (Seville), the Hospital Universitario Virgen de las Nieves (Granada), and the Hospital Príncipe de Asturias (Alcalá de Henares, Madrid). Informed consent was obtained from all patients. The institutional review board of referral centers has approved our research.
DNA extraction
We obtained 5 ml of peripheral blood from all patients to isolate germline DNA from leukocytes and about 100 μg of fresh endometriotic tissue during a programmed laparoscopic intervention of the patients. DNA extraction was performed according to standard procedures using Nucleospin Blood Kit (Macherey-Nagel) or alternative protocols. To perform Polymerase Chain Reactions (PCRs), we prepared aliquots of DNA at a concentration of 5 ng/μl. The rest of the stock was cryopreserved at -20°C.
Mutation analysis
NRIP1 cDNA was first cloned by Cavailles et al. [8]. This gene is mono-exonic, spans 7,239 base pairs (bp) and is located at 21q11.2. Genomic sequence containing NRIP1 gene was identified using the blat tool at UCSC Genome Bioinformatics server . Information concerning any Single Nucleotide Polymorphism (SNP) or mutation identified was compared with the UCSC Genome Bioinformatics server and also with the Single Nucleotide Polymorphism Database (dbSNP) at the National Centre for Biotechnology Information (NCBI) . According to standard mutation nomenclature [9], we employed the most frequent allele in the first position and the rarer allele in the last position.
We employed automated DNA sequencing methods to scan the entire coding sequence of NRIP1 gene in selected specimens. Overlapping PCRs covering the entire gene were designed and PCR products were purified and bi-directionally sequenced using the corresponding pair of primers (Table 1). Sequencing reactions were performed using the CEQ Dye Terminator Cycle Sequencing Quick Start Kit (Beckman Coulter, Inc) according to the manufacture's instructions. Fluorograms were analyzed on CEQ™ 8000 Genetic Analysis System following the manufacturer's instructions (Beckman Coulter, Inc).
Table 1 Amplification primer sequences and PCR product size. NRIP1 is a monoexonic gene. We designed eight overlapping amplicons to cover the entire coding sequence of this gene.
PRIMER SEQUENCE 5' → 3' PCR product size (bp)
1F TTCTAGTTCTGCCTCCTTAAC 554
1R ACATTTCTGGCAGTGCATTTC
2F GATCAGGTACTGCCGTTGA 528
2R CGAATCTTCCTGATGTGACT
3F GTGCTATGGTGTTGCATCAAG 572
3R TGCAGGTTATAAGAACTCACTGG
4F CATCATCAAGCAAACTGATGGC 577
4R AGCCCTCAGGGAGTACACAA
5F CTTCAATTGCTACTTGGCCAT 582
5R GTAGTCAACCAACAGGTCCT
6F CTGGAAACACAGATAAACCGATAGG 584
6R TGGCACTTCTAGAATCAAAG
7F AGATAGTTACCTGGCAGATG 572
7R TCCTACTTTCCCTGAGCACT
8F CAGTTGCATGGATAACAGGA 645
8R GTATTGGTTACTGGTGATG
Genotyping
To verify the DNA variants detected during the sequencing process and to perform association studies, we employed Fluorescent Resonance Energy Transfer (FRET) protocols. We designed and synthesized amplification primers and fluorescent detection probes for all the DNA variants identified within the NRIP1 gene. The selected primer pairs and detection probes are summarized in Tables 1 and 2. Real-time PCR was performed in the LightCycler system (Roche Applied Science) using reaction conditions previously published by us [10].
Table 2 Anchor and Sensor probes sequences employed for coding Single Nucleotide Polymorphism (cSNP) analysis using Fluorescence Resonance Energy Transfer (FRET) technology.
Mutation PROBE SEQUENCE 5' → 3
Gly75Gly 75-ANCHOR AGTAATGGTCCAGTTCTCAATACAC – F
75-SENSOR Cy5 – TACATATCAGGGGTCTGGC – Ph
His221Arg 221-ANCHOR Cy5 – AGTGGAACAAAGGTCATGAGTGAAC – Ph
221-SENSOR TCTCCTCATCATGTTGGACA – F
Ile441Val 448-ANCHOR TATTCCAACTGTGTTCCCATAGACT – F
Arg448Gly 448-SENSOR Cy5 – GTCTTGCAAACACCGAACTG – Ph
Ser803Leu 803-ANCHOR GCGCACCTGCCTTACCAGTGTCCCGA – F
803-SENSOR Cy5-GACTTTAAATCGGAGCCTGTT – Ph
Val1079Phe 1079-ANCHOR Cy5 – CGAGAAACACAAGACAAGGACATTT – Ph
1079-SENSOR GGAGGCAATTCTGTTACCAG – F
Nomenclature: F: Fluoresceine, Ph: Phosphate.
The conditions to obtain optimal melting curves for FRET analysis and spectrofluorimetric genotypes were 95°C for 0 s, 63°C for 25 s, 45°C for 0 s and 80°C for 0 s (with a temperature-transfer speed of 20°C/s in each step, except the last step, in which the speed of temperature transfer was 0.1°C/s). In the last step, a continuous fluorometric register was performed fixing the gains of the system at 1, 50, and 50 on channels F1, F2, and F3 respectively. Genotype results using real time-PCR are shown in Figure 1a. To test the specificity of these assays, selected amplicons of different melting patterns were re-sequenced using an automated DNA sequencer (Beckman Coulter CEQ 2000XL, data not shown).
Figure 1 Detection of germ line variants in NRIP1 in patients with severe endometriosis. A) Spectrofluorimetric analysis of NRIP1 gene using real-time PCR. Analysis of the fluorescence measured during melting curve determination in the LightCycler (Roche Applied Science). Each allele has a specific melting point and all alleles are represented by its specific nucleotide change with the exception of Ile441Val and Arg448Gly polymorphisms. Nt c.512 G->A (Gly75Gly, melting points, Allele G: 62°C; Allele A: 57°C). Nt c.949 A->G (His221Arg, melting points, Allele A 61°C; Allele G: 56°C). Nt c.1608 A->G (Ile441Val) and Nt c.1629 C->G (Arg448Gly) (melting points, Allele Val441: 59°C; Allele Gly448: 56°C; wild type: 63°C). Nt c.2695 C->T (Ser803Leu, melting points, Allele C: 61°C; Allele T: 55°C). Nt c.3522 G->T (Val1079Phe, melting points, Allele G: 62°C; Allele T: 54°C). B) Sequence conservation and location of mutations in the RIP-140 protein. Black shading indicates the position of mutations. LXXLL motifs responsible for ligand independent interaction with Retinoid Acid Receptor (RAR) and Retinoid X Receptor (RXR) are in bold and underlyned. Signal peptide is depicted in blue, Carboxyl terminal binding protein (CTBP) and RAR interacting motifs are in red and green respectively. Low complexity regions are shown in grey.
Statistical Analysis
To compare allele and genotype frequencies between patients, control and super-control groups, we performed conventional chi-square tests with Yates correction or Fisher exact test using Statcalc (EpiInfo 5.1, Center for Disease control, Atlanta, GA). For statistical analysis of genotype distribution, test for deviation of Hardy-Weinberg equilibrium or two-point association studies, we employed six different tests adapted from Sasieni (deviation from Hardy-Weinberg equilibrium, allele frequency differences test, heterozygous test, homozygous test, allele positivity test and Armitage's trend test) [11]. These calculations were performed in the online resource at the Institute for Human Genetics, Munich, Germany . Significant thresholds for statistical studies were fixed at p < 0.05.
Results
Looking for somatic mutations within the NRIP1 gene, we determined the complete coding sequence of the candidate gene in 20 selected and unrelated somatic endometriotic tissues using bi-directional automated capillary DNA sequencing. In our primary sequencing project we finished 80,600 bp of DNA. Using our methodology, we identified six single nucleotide DNA variants within the coding sequence of the NRIP1 gene in various unrelated somatic DNA samples. Two of these variants have been previously identified and they are included in the Single Nucleotide Polymorphism Database (dbSNP) at the National Centre for Biotechnology Information (NCBI) (Table 3). Five of these mutations alter the amino acid coding sequence of RIP140 protein generating missense mutations (Fig. 1 and Table 3). Although all mutations were detected in somatic DNA, direct molecular analyses of the corresponding blood samples of mutated tissue also contain the same DNA change. This last result implies that the genetic variants identified are germ-line and, consequently, somatic mutations at NRIP1 locus are not commonly involved in the pathogenesis of human endometriosis.
Table 3 Summary of DNA variants observed within the coding sequence of the NRIP1 gene.
DNA variant* Amino acid Substitution (change in codon) dbSNP** accession number Detection in Endometriotic tissue samples (40 chromosomes) Detection in germline DNA derived from endometriosis patients (118 chromosomes) Detection in controls (282 chromosomes) Status
Nt c.512 G->A None [Gly75] (ggg to gga) rs2229741 15/40 57/118 129/282 Common polymorphism
Nt c.949 A->G His221Arg (cat to cgt) - 1/40 1/118 3/282 Common polymorphism
Nt c.1608 A->G Ile441Val (ata to gta) - 0/20 0/118 4/282 Common polymorphism
Nt c.1629 C->G Arg448Gly (cga to gga) rs2229742 9/40 16/118 19/282 Common polymorphism
Nt c.2695 C->T Ser803Leu (tcg to ttg) - 2/40 3/118 10/282 Common polymorphism
Nt c.3522 G->T Val1079Phe (gtt to ttt) - 1/40 1/118 0/282 Rare Variant/Mutation
*In accordance with genbank number NM_003489.
** dbSNP: the Single Nucleotide Polymorphism Database at the National Centre for Biotechnology Information (NCBI)
To evaluate the polygenic role of NRIP1 gene variants in human endometriosis, we decided to preliminary explore the allelic frequencies and genotypes of these mutations in women affected by endometriosis and unselected controls. To conduct genetic association studies, we developed real-time PCR detection protocols using FRET probes for each DNA mutation identified at NRIP1 locus. Using these techniques, we genotyped the mutations in 200 unrelated women (59 endometriosis patients, 94 unselected controls and 47 super-control women). Overall, 400 different chromosomes have been scored for each DNA variant (Table 4).
Table 4 Association studies of common DNA variants of the NRIP1 gene in relation to human endometriosis.
NRIP1 polymorphism (change in codon) Genotypes Patients (n = 59) Unselected Controls (n = 94) Super Controls (n = 47) All Controls (n = 141) Statistical Analysis*
Gly75Gly (ggg to gga) aa 15 19 9 28 P = 0.34 (Heterozygous test)
ag 27 46 27 73
gg 17 28 11 39
Arg448Gly (cga to gga) cc 44 83 40 123 P = 0.027 (Allele positivity test)
cg 14 10 7 17
gg 1 1 0 1
Ser803Leu (tcg to ttg) cc 56 85 46 130 P = 0.59 (Armitage's trend test)
ct 3 9 1 10
tt 0 0 0 0
*Compares patients versus merged controls. Best p value employing tests for genetic association according to Sasieni (1997). Ile441Val and His221Arg are not analyzed due to small or null sample size in endometriosis samples.
By analyzing the allelic frequencies of DNA variants detected in Spanish population, we classified these variants as common polymorphisms if observed in >1% of chromosomes in controls (Gly75Gly, His221Arg, Ile441Val, Arg448Gly and Ser803Leu) or rare variants if observed with a frequency <1% of chromosomes in controls (Table 3). In contrast, Val1079Phe allele appears only in a single patient in heterozygous state and none of 141 controls. This data could suggest its involvement in the disease. Reinforcing this hypothesis, Val1079Phe is located close to high-conserved domain of the carboxylic end of RIP140 protein that interacts with retinoic acid nuclear receptor (Fig. 1).
Direct inspection of genotypes in patients revealed three genotype patterns within the NRIP1 gene that appear to be over-represented in women affected by endometriosis (p = 0.006, Fisher exact test). The patterns consist of a combination of Arg448Gly together with His221Arg or Val1079Phe variants. We identified three unrelated women affected by endometriosis carrying double heterozygotes (His221Arg/Arg448Gly and Val1079Phe/Arg448Gly) or homozygote (Arg448Gly/Arg448Gly) genotypes for these alleles, respectively. The homozygote (Arg448Gly/Arg448Gly) genotype pattern appeared only in 1 of 94 unselected controls (p = 0.016 Fisher exact test) and none of 47 super-control women (p = 0.023, Fisher Exact test), whereas double heterozygotes genotypes did not appear in any control individual. These results suggest that specific combinations of amino acid changes at NRIP1 locus could be related to endometriosis etiology with a 99.4% of reliability, although given the scarce sample size the presence of polygenes within NRIP1 locus must be proven with a larger and independent re-analysis.
Finally, given the preliminary results, we conducted an small case-control study analyzing all common variants detected within the NRIP1 locus (Gy75Gly, Arg448Gly and Ser803Leu). Table 4 shows the results for those test that maximize the differences between case and control groups for each polymorphism. Genotypic distributions of polymorphisms analyzed are in accordance with the Hardy-Weinberg equilibrium law (p > 0.15), indicating no bias due to technical or stratification problems nor evolution-dependent genetic sweep/selection events (data not shown). Interestingly, our analysis revealed that Arg448Gly polymorphism appears to be weakly associated with endometriosis in our population (Odds ratio = 2.327, pallele positivity test = 0.027). In contrast, no significant association could be achieved when comparing unselected versus super-control women, supporting the accuracy of the selected control panel (p > 0.34 for Gly75Gly, p > 0.41 for Arg448Gly, and p > 0.1 for Ser804Leu).
Overall, our results might support the role of NRIP1 gene in endometriosis, although given the small sample size, we propose an extensive re-analysis by increasing the sample size to confirm our results.
Discussion
Endometriosis is a complex disease affecting 10–15% of women at reproductive age. Very few genes are known to be altered in this pathology. Molecular genetic analyses provide some evidence of genetic association in case-control studies analyzing Estrogen Receptor 1 (ESR1 OMIM 133430) and Cytochrome P450, Family 19, Subfamily A, Polypeptide 1 (CYP19 OMIM 107910) genes. Interestingly, both loci are involved in oestrogen mechanism of production and action [12,13]. In addition, other nuclear receptor genes, such as Progesterone Receptor (PGR OMIM 607311) and Peroxisome Proliferative Activated Receptor, Gamma (PPARG OMIM 601487) gene have been associated with endometriosis in other case-control studies [14,15]. The involvement in endometriosis of loci related to detoxification has been also studied and replicated [16-18].
Given these preliminary findings and the importance of steroid receptors in uterine physiology [19] and endometriosis pathogenesis [1,20], the biochemical pathways involved in steroids production, degradation or mechanisms of action appear to be strong candidates for endometriosis etiology and many other phenotypes related to human fertility.
Following this working hypothesis, targeting disruption of nuclear receptors and their regulators such as nrip1 or CCR4-NOT transcription complex, subunit 7 (cnot7 GenBank AK009561) in animal models have provided direct evidence of the importance of nuclear receptor homeostasis in male and female reproduction [3,21-24].
Here we present the first structural analysis of the human NRIP1 gene in relation to human disease. It is of interest to mention that the dbSNP includes 26 SNPs for NRIP1 gene currently. Eighteen of these variants are located within the 3'untranslated region (3'UTR), this genomic region has not been covered in this study, and the remaining ones are coding SNPs. According to GenBank, only three SNPs in the 3'UTR region and two coding SNPs have been validated in population based studies including more than 150 chromosomes. The rest of the SNPs are the result of the bioinformatic alignment of different cDNA and genomic clones. The allele frequencies here presented for Gly75Gly (dbSNP rs2229741) and Arg448Gly (dbSNP rs2229742) polymorphisms are very similar to those included in dbSNP (data not shown).
Overall, our results are preliminary providingt suggestive, but not definitive, evidence of NRIP1 gene involvement in human endometriosis. We think that conclusive proofs of involvement will be achieved throughout re-analyses of this study in independent cohorts of patients and controls, rather than performing functional analyses of the missense mutations observed. The detection of functionality of DNA variants involved in complex traits such endometriosis, is near to be impossible using conventional technologies because the effect from single genetic variant/mutation is expected to be very small and it is only the joint effect of several susceptibility genes that leads to the disease [25]. In this sense, we are currently recruiting a higher number of patients and controls to perform a proper re-analysis of our results.
Regarding Arg448Gly polymorphism, we propose that the variant could act as a low penetrance allele related to human endometriosis. The molecular mechanism of this mutation is not well understood, although its location and degree of conservation provide some interesting clues. In fact, Arg448 residue of RIP140 protein is completely conserved among humans, rats, mice, gallus and xenopus (Fig. 1b). Moreover, the non-conservative substitution detected (Arg448Gly) might affect the Carboxyl terminal binding protein (CTBP) interacting motif of RIP140 protein that is located close to this amino acid residue (Fig. 1b).
On the basis of genotype analysis in affected women, we propose that Arg448Gly mutation could act in concert with other genetic variants within NRIP1 or other loci. In this way, we found a single woman affected by endometriosis simultaneously carrying Val1079Phe mutation and Arg448Gly polymorphism both in a heterozygous state. Val1079Phe also arises in an inter-specific conserved residue. Moreover, this mutation is located close to (and may disrupt) the retinoid acid receptor interacting motif "LTKTNPILYYMLQK" of RIP140 protein (Fig. 1b). Supporting its involvement in the disease, we have not identified the Val1079Phe mutation in 282-control chromosomes. Intriguingly, retinoid acid receptors alpha, gamma 2 and its regulator cnot7 have been involved in male sterility [24,26,27]. Moreover, the presence of multiple specific functional rare variants in affected patients have been recently proposed and evaluated [28]. This hypothesis is an alternative to explain the genetic component of complex traits in front of the widely accepted common disease common variant hypothesis [29].
Finally, we identified a single patient carrying a unique genotype combination comprising His221Arg rare variant and, again, Arg448Gly polymorphism. The absence of inter-specific amino acidic conservation and the inexistence of known functional domains close to His221Arg variant do not support the functionality of His221Arg allele. However, His221Arg only appears combined with Arg448Gly in a woman affected by endometriosis. This combination never appears in 141 unrelated controls. Functional assays or large cohorts analyses will help to elucidate its involvement in this pathology.
Our results support that NRIP1 gene might contain alleles related to endometriosis in humans. NRIP1 gene encodes a highly pleiotropic nuclear receptor co-regulator (RIP140) [30]. This protein interacts and regulates multiple members of the nuclear receptor super-family. Some of them have been associated with endometriosis. The wide repertoire of RIP140 targets might explain the complex pathological findings observed in human endometriosis. In fact, a recent report revealed a complex mechanism, involved in endometriosis and other oestrogen-related traits, by which ER-mediated oestrogen signaling is modulated by a co-regulatory-like function of activated AhR/Arnt dioxin receptor complex, giving rise to adverse oestrogen-related actions of dioxin-type environmental contaminants [31]. Intriguingly, oestrogen and dioxin nuclear receptor pathways are modulated by RIP140 protein [32,33].
According to previous data, mouse models reports and the present results, NRIP1 gene appears to be an attractive gene for human endometriosis etiology and other related pathologies. In addition, we have found a genetic interaction of NRIP1 gene with Estrogen receptor alpha and beta (ESR1 and ESR2) genes in two estrogen-dependent diseases such as male infertility [34] and osteoporosis (manuscript in preparation). For these reasons, further evaluation of NRIP1 gene in many oestrogen-related phenotypes is warranted.
Conclusion
Our results suggest that NRIP1 gene variants, separately or in combinations, might act as predisposing factors for human endometriosis. According to previous data, mouse models reports and the present results, NRIP1 appears to be an attractive candidate gene for human endometriosis etiology and other estrogen-related pathologies.
Competing interests
The author(s) declare that they have not competing interests.
Authors' contributions
VC, JAS, MC, MAL, FC and VL carried out the recruitment and classification of patients and controls and the biological samples management. RR and JJG carried out the molecular genetic studies and participate in the analysis and interpretation of data. LMR and AR carried out the design of the study, performed the statistical analyses and the interpretation of data.
All authors have been involved in drafting the article or revising it critically for important intellectual content and have given final approval of the version to be published.
Acknowledgements
We are deeply grateful to patients and controls for participation in this study. Funded by the Ministerio de Ciencia y Tecnología (MCYT, Spain, grant numbers FIT-010000-2003-36, FIT-010000-2003-89, FIT-010000-2003-70, PTQ2002-0206) and Organón Española S.A. (BCN, Spain).
==== Refs
Bulum SE Zeitoun KM Takayama K Sasano H Estrogen biosynthesis in endometriosis: molecular basis and clinical relevance J Mol Endocrinol 2000 25 35 42 10915216 10.1677/jme.0.0250035
Donnez J Thomas K Incidence of luteinized unruptured follicle syndrome in fertile women and women with endometriosis Eur J Obstet Gynecol Reprod Biol 1982 14 187 190 7160529 10.1016/0028-2243(82)90095-8
White R Leonardsson G Rosewell I Ann Jacobs M Milligan S Parker M The nuclear receptor co-repressor Nrip1 (RIP140) is essential for female fertility Nat Med 2000 6 1368 1374 11100122 10.1038/82183
García-Velasco JA Arici A Is the endometrium or oocyte/embryo affected in endometriosis? Hum Reprod 1999 2 77 89
Vercammen EE D'Hooghe TM Endometriosis and recurrent pregnancy loss Semin Reprod Med 2000 18 363 368 11355795 10.1055/s-2000-13726
Leonardsson G Steel JH Christian M Pocock V Milligan S Bell J So PW Medina-Gomez G Vidal-Puig A White R Parker MG Nuclear receptor corepressor RIP140 regulates fat accumulation Proc Natl Acad Sci U S A 2004 101 8437 42 15155905 10.1073/pnas.0401013101
American Society for Reproductive Medicine Revised American Society for Reproductive Medicine classification of endometriosis: 1996 Fertil Steril 1997 67 817 821 9130884 10.1016/S0015-0282(97)81391-X
Cavailles V Dauvois S L'Horset F Lopez G Hoare S Kushner PJ Parker MG Nuclear factor RIP140 modulates transcriptional activation by the estrogen receptor EMBO J 1995 14 3741 3751 7641693
den Dunnen JT Antonarakis SE Nomenclature for the description of human sequence variations Hum Genet 2001 109 121 124 11479744 10.1007/s004390100505
Gonzalez A Abril E Roca A Aragon MJ Figueroa MJ Velarde P Royo JL Real LM Ruiz A CAPN10 alleles are associated with polycystic ovary syndrome J Clin Endocrinol Metab 2002 87 3971 3976 12161543 10.1210/jc.87.8.3971
Sasieni PD From genotypes to genes: doubling the sample size Biometrics 1997 53 1253 1261 9423247
Georgiou I Syrrou M Bouba I Dalkalitsis N Paschopoulos M Navrozoglou I Lolis D Association of estrogen receptor gene polymorphisms with endometriosis Fertil Steril 1999 72 64 66 10.1016/S0015-0282(99)00198-3
Kado N Kitawaki J Obayashi H Ishihara H Koshiba H Kusuki I Tsukamoto K Hasegawa G Nakamura N Yoshikawa T Honjo H Association of the CYP17 gene and CYP19 gene polymorphisms with risk of endometriosis in Japanese women Hum Reprod 2002 17 897 902 11925378 10.1093/humrep/17.4.897
Wieser F Schneeberger C Tong D Tempfer C Huber JC Wenzl R PROGINS receptor gene polymorphism is associated with endometriosis Fertil Steril 2002 77 309 312 11821088 10.1016/S0015-0282(01)02984-3
Dogan S Machicao F Wallwiener D Haering HU Diedrich K Hornung D Association of peroxisome proliferator-activated receptor gamma 2 Pro-12-Ala polymorphism with endometriosis Fertil Steril 2004 81 1411 1413 15136115 10.1016/j.fertnstert.2003.10.026
Baranova H Canis M Ivaschenko T Albuisson E Bothorishvilli R Baranov V Malet P Bruhat MA Possible involvement of arylamine N-acetyltransferase 2, glutathione S-transferases M1 and T1 genes in the development of endometriosis Mol Hum Reprod 1999 5 636 641 10381818 10.1093/molehr/5.7.636
Arvanitis DA Goumenou AG Matalliotakis IM Koumantakis EE Spandidos DA Low-penetrance genes are associated with increased susceptibility to endometriosis Fertil Steril 2001 76 1202 1206 11730751 10.1016/S0015-0282(01)02865-5
Arvanitis DA Koumantakis GE Goumenou AG Matalliotakis IM Koumantakis EE Spandidos DA CYP1A1, CYP19, and GSTM1 polymorphisms increase the risk of endometriosis Fertil Steril 2003 702 709 12620480 10.1016/S0015-0282(02)04817-3
Moutsatsou P Sekeris CE Steroid receptors in the uterus: implications in endometriosis Ann N Y Acad Sci 2003 997 209 222 14644828 10.1196/annals.1290.024
Kitawaki J Kado N Ishihara H Koshiba H Kitaoka Y Honjo H Endometriosis: the pathophysiology as an estrogen-dependent disease J Steroid Biochem Mol Biol 2002 83 149 155 12650711 10.1016/S0960-0760(02)00260-1
Kastner P Mark M Leid M Gansmuller A Chin W Grondona JM Decimo D Krezel W Dierich A Chambon P Abnormal spermatogenesis in RXR beta mutant mice Genes Dev 1996 10 80 92 8557197
Dupont S Krust A Gansmuller A Dierich A Chambon P Mark M Effect of single and compound knockouts of estrogen receptors alpha (ERalpha) and beta (ERbeta) on mouse reproductive phenotypes Development 2000 127 4277 4291 10976058
Cui Y Miyoshi K Claudio E Siebenlist UK Gonzalez FJ Flaws J Wagner KU Hennighausen L Loss of the peroxisome proliferation-activated receptor gamma (PPARgamma) does not affect mammary development and propensity for tumor formation but leads to reduced fertility J Biol Chem 2002 17 17830 17835 11884400 10.1074/jbc.M200186200
Nakamura T Yao R Ogawa T Suzuki T Ito C Tsunekawa N Inoue K Ajima R Miyasaka T Yoshida Y Ogura A Toshimori K Noce T Yamamoto T Noda T Oligo-astheno-teratozoospermia in mice lacking Cnot7, a regulator of retinoid X receptor beta Nat Genet 2004 36 528 533 15107851 10.1038/ng1344
Hoh J Ott J Genetic dissection of diseases: design and methods Curr Opin Genet Dev 2004 14 229 232 15172663 10.1016/j.gde.2004.04.006
Lohnes D Kastner P Dierich A Mark M LeMeur M Chambon P Function of retinoic acid receptor gamma in the mouse Cell 1993 21 643 658 8388780 10.1016/0092-8674(93)90246-M
Dufour JM Vo MN Bhattacharya N Okita J Okita R Kim KH Peroxisome proliferators disrupt retinoic acid receptor alpha signaling in the testis Biol Reprod 2003 68 1215 1224 12606456 10.1095/biolreprod.102.010488
Cohen JC Kiss RS Pertsemlidis A Marcel YL McPherson R Hobbs HH Multiple rare alleles contribute to low plasma levels of HDL cholesterol Science 2004 305 869 872 15297675 10.1126/science.1099870
Lohmueller KE Pearce CL Pike M Lander ES Hirschhorn JN Meta-analysis of genetic association studies supports a contribution of common variants to susceptibility to common disease Nat Genet 2003 33 177 182 12524541 10.1038/ng1071
Zilliacus J Holter E Wakui H Tazawa H Treuter E Gustafsson JA Regulation of glucocorticoid receptor activity by 14-3-3-dependent intracellular relocalization of the corepressor RIP140 Mol Endocrinol 2001 15 501 511 11266503 10.1210/me.15.4.501
Ohtake F Takeyama K Matsumoto T Kitagawa H Yamamoto Y Nohara K Tohyama C Krust A Mimura J Chambon P Yanagisawa J Fujii-Kuriyama Y Kato S Modulation of oestrogen receptor signalling by association with the activated dioxin receptor Nature 2003 423 545 550 12774124 10.1038/nature01606
Kumar MB Tarpey RW Perdew GH Differential recruitment of coactivator RIP140 by Ah and estrogen receptors. Absence of a role for LXXLL motifs J Biol Chem 1999 274 22155 22164 10428779 10.1074/jbc.274.32.22155
Peters GA Khan SA Estrogen receptor domains E and F: role in dimerization and interaction with coactivator RIP-140 Mol Endocrinol 1999 13 286 296 9973258 10.1210/me.13.2.286
Galan JJ Buch B Cruz N Segura A Moron FJ Bassas L Martinez-Pineiro L Real LM Ruiz A Multilocus analyses of estrogen-related genes reveal the involvement of ESR1 gene in male infertility and the polygenic nature of the pathology Fertil Steril
|
0
|
PMC1242356
|
CC BY
|
2021-01-04 16:40:17
|
no
|
J Exp Clin Assist Reprod. 2005 Aug 31; 2:12
|
latin-1
|
J Exp Clin Assist Reprod
| 2,005 |
10.1186/1743-1050-2-12
|
oa_comm
|
==== Front
J Negat Results BiomedJournal of Negative Results in Biomedicine1477-5751BioMed Central London 1477-5751-4-61613524510.1186/1477-5751-4-6Brief ReportResting energy expenditure is not influenced by classical music Carlsson Ebba [email protected] Hannah [email protected] Frode [email protected] Dept. of Clinical Nutrition, P. O Box 459, Sahlgrenska Academy at Göteborg University, SE-405 30 Göteborg, Sweden2005 31 8 2005 4 6 6 16 8 2005 31 8 2005 Copyright © 2005 Carlsson et al; licensee BioMed Central Ltd.2005Carlsson et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Obesity shows an increasing prevalence worldwide and a decrease in energy expenditure has been suggested to be one of the risk factors for developing obesity. An increase in resting energy expenditure would have a great impact on total energy expenditure. This study shows that classical music do not influence resting energy expenditure compared to complete silence. Further studies should be performed including other genres of music and other types of stress-inductors than music.
==== Body
Findings
Obesity shows an increasing prevalence worldwide [1] and a decrease in energy expenditure has been suggested to be one of the risk factors for developing obesity [2]. Increasing energy expenditure could be done by increasing physical activity, but resting energy expenditure (REE) is the largest part of an humans' energy expenditure (70–80%), and an increase in REE would have a large impact on total energy expenditure. REE is assessed by indirect calorimetry by measurements of oxygen consumption and carbon dioxide production which, when known, is calculated into energy expenditure [3]. It is known that ingestion of food increases resting energy expenditure – also called diet induced thermogenesis [4] Nicotine and caffeine have also been shown to increase energy expenditure [5]. None has however studied the effect of external sound stimuli, such as music, on REE. The aim of the current study was to assess if classical music has an effect on REE, and if there are differences between different types of classical music.
In this randomized cross-over study, 2 different music CD's were used. Both CD's started with 10 minutes of silence and were followed by 10 minutes of calm classical music and 10 minutes of stressful classical music, presented in Table 1. The order of music differed between the two CD's, which was randomly chosen for each subject. Classical music was chosen for both stressful and calm music to limit confounding effects from the subjects' taste of music. A pre-study power-calculation showed that to be able to detect a statistical significant (p < 0.05) difference at 420 kJ/day (judged as clinical relevant) with a power of 80%, 40 subjects should be included. To allow for drop-out, 43 healthy volunteers (31 women and 12 men) were included, all participants gave written informed consent. Following measurement of height and weight, REE was measured by indirect calorimetry using a ventilated hood system, the Deltatrac™ II Metabolic Monitor (Datex, Helsinki, Finland). Before each measurement, the equipment was calibrated with gas mixtures of known O2 and CO2 contents according to the instructions from the manufacturer. The subjects were instructed to limit their physical activity the evening before measurement. All subjects were measured after an overnight fast and they arrived from their home by car or public transport. After 30 minutes rest in the supine position REE was measured during 35 minutes when the subjects were awake. Due to adaptation to the inside-hood environment, the first five minutes were eliminated from the total result. The music was provided through earphones and measurements were performed in an environmental temperature of 20–24°C. After completion of the measurement, the subjects were asked how they perceived each part of the music, as calm or stressful, or something else. Data are presented as mean and standard deviation. To compare REE during silence to the calm and stressful music, two-sided paired Student's t-tests were performed.
Table 1 Description of the calm and stressful music which each lasted for 10 minutes
Calm music Stressful music
Composer Piece of music Composer Piece of music
Erik Satie Gymnopédie No 1 Béla Bartók String quartet No 4 prestissimo con sordino
Erik Satie Gymnopédie No 3 Igor Stravinsky From The Fire Bird: "Infernal dance of all Kashcers's subjects"
Johann Sebastian Bach Air Hans Werner Henze 2nd movement "Dies irae" from Requiem for piano, trumpet and chamber orchestra
Forty subjects, 29 women and 11 men, completed the study. One subject dropped out because of feeling uncomfortable in the ventilated hood, one subject due to technical issues with the indirect calorimeter, and one subject due to problems with the CD-player. Mean (SD) age of the subjects were 35 (14) y, body height 172 (10) cm, body weight 68 (13) kg, and body mass index 23 (3) kg/m2. Mean (SD) REE during silence was 5720 (1063) kJ/day. No significant differences in REE between silence and the two sets of music were found, 5710 (1054) kJ/day during calm music (p = 0.57) and 5740 (1046) kJ/day during stressful music (p = 0.43). Thirty-eight subjects perceived the calm music as calm and 28 subjects the stressful music as stressful. However, analyzing the results regarding to their own perception of the results, did not yield any statistically significant differences in measured REE between silence and the two music periods.
This study could not detect any statistical significant or clinical relevant influences of music on REE, and then theoretically not on total energy expenditure. We chose to compare classical calm music to classical stressful music. This was to limit the confounding effect of the subjects own music preferences. When the stressful music was selected, not only tempo of the music was taken into consideration. The stressful music was also supposed to be irregular, have large differences between high and low frequencies, include many abrupt sounds, and give a sense of unpredictability. Most of the subjects perceived the calm music as calm and the stressful music as stressful, even if some subjects experienced the stressful music as "other". Maybe the stressful music was not stressful enough. Further studies should be conducted to investigate other types of music, i.e. pop music vs. heavy metal, and preferably also other types of stress-inductors than music combined with measurements of heart rate and other measures of stress. The results from this study do not support that music during rest could be used in obesity prevention or treatment alone, but music could of course be combined with physical activity to achieve an increase in total energy expenditure.
List of abbreviations
REE – resting energy expenditure
CD – compact disc
SD – standard deviation
Authors' contributions
EC and HH participated in the study design, carried out the data collection and analyzed the results. FS conceived the study, and participated in its design and coordination and drafted the manuscript. All authors read and approved the final manuscript.
Acknowledgements
The authors are grateful to Lena Hulthén, professor at Dept of Clinical Nutrition, Sahlgrenska Academy at Göteborg University for valuable input during the study design.
==== Refs
James PT Leach R Kalamara E Shayeghi M The worldwide obesity epidemic Obes Res 2001 9 228S 233S 11707546
Livingstone B Epidemiology of childhood obesity in Europe Eur J Pediatr 2000 159 S14 S34 11011953
Weir JBD New methods for calculating metabolic rate with special reference to protein metabolism J Physiol 1949 109 1 9 15394301
Westerterp KR Diet induced thermogenesis Nutr Metab 2004 1 5 10.1186/1743-7075-1-5
Jessen AB Toubro S Astrup A Effect of chewing gum containing nicotine and caffeine on energy expenditure and substrate utilization in men Am J Clin Nutr 2003 77 1442 7 12791621
|
16135245
|
PMC1242357
|
CC BY
|
2021-01-04 16:37:37
|
no
|
J Negat Results Biomed. 2005 Aug 31; 4:6
|
utf-8
|
J Negat Results Biomed
| 2,005 |
10.1186/1477-5751-4-6
|
oa_comm
|
==== Front
J Transl MedJournal of Translational Medicine1479-5876BioMed Central London 1479-5876-3-341616474910.1186/1479-5876-3-34ReviewImmunologic aspect of ovarian cancer and p53 as tumor antigen Nijman HW [email protected] A [email protected] der Burg SH [email protected] der Zee AGJ [email protected] T [email protected] Dept. of Gynaecologic Oncology, Groningen University Medical Center2 Dept. of Medical Microbiology, Molecular Virology Section, Groningen University Medical Center3 Dept. of Immunohematology and Blood Transfusion, Leiden University Medical Center2005 15 9 2005 3 34 34 21 7 2005 15 9 2005 Copyright © 2005 Nijman et al; licensee BioMed Central Ltd.2005Nijman et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Ovarian cancer represents the fifth leading cause of death from all cancers for women. During the last decades overall survival has improved due to the use of new chemotherapy schedules. Still, the majority of patients die of this disease. Research reveals that ovarian cancer patients exhibit significant immune responses against their tumor. In this review the knowledge obtained thus far on the interaction of ovarian cancer tumor cells and the immune system is discussed. Furthermore the role of p53 as tumor antigen and its potential role as target antigen in ovarian cancer is summarized. Based on the increased knowledge on the role of the immune system in ovarian cancer major improvements are to be expected of immunotherapy based treatment of this disease.
==== Body
Introduction
Ovarian cancer is the most common cause of death from gynecological malignancies. Its nonspecific clinical presentation and the absence of effective screening methods are responsible for the 70% of patients who present with an advanced stage of disease at the time of diagnosis. Primary treatment for advanced stage ovarian cancer is cytoreductive surgery followed by platinum/paclitaxel based chemotherapy. An aggressive surgical approach has been advocated with the intent to remove all macroscopic disease which should yield better survival than leaving residual disease [1-3]. Response rates to primary chemotherapy are 65–80%. When residual or recurrent disease manifests itself, resistance to chemotherapy will prohibit further curative therapy, resulting in an overall survival for patients with advanced stage ovarian disease of only 10–20%[4,5].
Research during the last decades has revealed that ovarian cancer patients exhibit significant immune responses against their tumor (reviewed in this paper). In designing alternative treatments to successfully eradicate ovarian cancer it is important to consider both the positive effects of immune responses to ovarian cancer and the confounding negative effects on the immune system caused by the tumor cells. As the main target for a potential vaccine is the (overexpressed / mutated) p53 protein we will focus on studies aimed at the induction of humoral and cellular responses against this antigen. However, before reviewing these studies we will briefly introduce some general aspects of the cellular immune system including antigen encounter, antigen processing and presentation and factors influencing the outcome of the immune response in ovarian cancer.
General introduction on the cellular immune system
Antigen presenting cells, most likely dendritic cells, can capture tumor antigens that are secreted or shed by tumor cells or by taking up dying tumor cells. The tumor antigens are processed and presented as peptides by major histocompatability complex (MHC) I and II molecules on the cell surface, and recognized by the T-cell receptor on T-cells. This phenomenon is often referred to as the first signal of activation. After cleavage of proteins into peptides by the proteasome complex and loading of peptides into the class I molecules in the endoplasmatic reticulum, these MHC class I – peptide complexes, recognized by cytotoxic T lymphocytes, are transported to the cell surface. MHC class II molecules mainly present exogenous endocytosed proteins. Antigen (peptide) loading of MHC class II molecules occurs within the endocytic pathway (MHC class II compartments). MHC class II – peptide complexes expressed on the cell surface are recognized by the CD4+ T helper cells. Next to this first antigen specific signal there is a need for a second signal. This signal involves the ligation of CD28 or CTLA-4 on lymphocytes by co-stimulatory molecules CD80 (B7.1) or CD86 (B7.2) respectively on antigen presenting cells or target cells. Binding of the CD28 receptor results in proliferation and activation of T cells, in contrast to binding of CTLA-4 which results in T cell anergy. Another important co-activation signal is mediated by interaction of CD40 ligand on T cells and CD40 on the antigen presenting cell. Fully activated CD8+ T cells differentiate into cytotoxic T lymphocytes and can lyse tumor cells. Memory CD4+ and CD8+ T cells play a critical role in maintaining protective immunity. Apart from their role in expanding CD8+ T-cells, CD4+ T-cells are also involved in the activation of CD8+ independent tumoricidal mechanisms which may play a role in the eradication of tumor cells that have lost MHC class I expression [6]. The CD4+ T cells can be divided into at least two subsets of T helper cells (Th), designated Th1 and Th2. Whereas a Th1 type immune response generally stimulates the generation of cellular immunity, a Th2 type response stimulates humoral immunity next to growth and differentiation of mast cells and eosinophils. Th1 cells secrete cytokines like IFN-γ, Il-2 and TNF-α, Th2 type cells mainly produce IL-4 and IL-10. Regulatory or suppressor T cells, represent potentially a major barrier to successful anti-tumor immune responses. These include Natural Killer T cells[7], CD25+CD4+ T cells [8,9] and Th3 cells[10]. The balance of signals processed by regulatory T cells can determine vastly different scenarios in tumor surveillance [11]. In the mouse system, CD25+CD4+ regulatory T cells suppress the activation and proliferation of other CD4+ and CD8+ T cells specific for auto antigens which of course is important to prevent autoimmunity but on the other hand prevents the effective generation of immunity to tumor antigens.
The rules that govern the balance between immunity and tolerance is controlled by the conditions of antigen encounter and activation status of the antigen presenting cell [10,12]. In general, systemic and persistent exposure of T cells to antigen in the absence of costimulation tends to result in T cell tolerization. The type and level of costimulation received during the first encounter with antigen are key determinants in the outcome of an immune response. This depends largely on the activation status of the professional antigen presenting cell that presents the antigenic peptide to naive T cells, in most cases the dendritic cell. The costimulatory state of professional antigen presenting cell is promoted by activated CD4+ T cells, in particular by interaction between CD40L on Th cells and CD40 on the APC [13-16]. This type of T cell help is essential for CTL induction under noninflammatory conditions, whereas lack of CD4+ T cell help can lead to CTL tolerization[17]. Direct demonstration that the activation status of antigen presenting cells influences the outcome of antigen recognition by CD8+ T cells was obtained in studies in which vaccination with mature dendritic cell induced cytotoxic T lymphocyte immunity, whereas infusion of immature dendritic cells failed to do so [15,18]. The conditions involved in setting the balance between tolerance and immunity seem to be different for activated T cells, because circumstances that tolerize naive T cells may not be tolerogenic for memory T cells. More details on the cellular immune system are to be found in recent reviews [19-22])
Ovarian cancer and the immune system
While the interaction between the host immune system and ovarian cancer tumor cells is still not completely understood, several observations suggest that cell-mediated immune responses could be important in controlling ovarian cancer.
As already stated, the presence of antigen presenting cells, most favorable dendritic cells, is crucial in activating the immune system. In cancer patients the number of dendritic cells is decreased and functionally suppressed by the tumor microenvironment, inhibiting immune responses and thereby causing an impaired tumor immunity [23-27]. For several tumor types it was shown that the number of infiltrating dendritic cells correlated with good prognosis. In a retrospective study using immunohistochemistry the same phenomenon was observed in ovarian cancer [28]. The potential role of dendritic cells in ovarian cancer was demonstrated by Schlienger et al[29]. In 50% of ovarian cancer patients dendritic cells derived from peripheral blood mononuclear cells could, in vitro, induce tumor specific T cells upon loading the dendritic cells with tumor antigen derived from autologous tumor. The antigen(s) recognized by these T cells were not defined. Dendritic cells derived from peripheral blood mononuclear cells and tumor associated macrophages obtained from ascites from the same ovarian cancer patients, cultured with IL-4, GM-CSF and TNF-α, comparably stimulated T cell lines[30]. In contrast to the beneficial effects of macrophages and dendritic cells on the tumor specific immune responses, tumor associated macrophages have been shown to secrete the immunosuppressive cytokine IL-10[27,31]. One of the effects of IL-10 is that it induces B7-H1 expression on myeloid derived dendritic cells [32]. B7-H1, belonging to the B7 family of costimulatory molecules, is thought to be involved in the regulation of cellular immune responses through its receptors on activated T and B cells [33,34]. B7-H1 was first described to be expressed by ovarian cancer cells. Later it has been shown to be also present in other human carcinomas [33]. Tumor associated B7-H1 induces apoptosis of activated antigen specific T cells, contributing to the immune evasion of tumor cells [35]. Not only the ovarian cancer tumor cells but also myeloid derived dendritic cells obtained from ovarian tumor tissue and their draining lymph nodes express B7-H1, and are capable to downregulate T cell responses[32]. INF-γ upregulates B7-H1 on the surface of tumor cell lines [35], which might have implications for IFN-γ based cancer immunotherapy. To deal with this issue one could consider blockade of the B7-H1 pathway by e.g. neutralizing mAb. The efficacy of this approach has been shown very nicely in a mouse model for squamous cell carcinoma [36].
In ascites and tumors from patients with ovarian cancer myeloid dendritic cells are outnumbered by plasmacytoid dendritic cells [27,37,38]. The exact role of the plasmacytoid dendritic cells in priming naive T cells needs to be further elucidated. It seems that plasmacytoid dendritic cells produce high levels of the angiogenic cytokines TNFα and IL-8 in contrast to the myeloid dendritic cells which produce cytokine IL-12, an inhibitor of angiogenesis. Thus, the accumulation of plasmacytoid dendritic cells in ascites and ovarian cancer tumors is of benefit for the vascularization of the tumor and thereby promotes tumor growth[39].
In ovarian cancer tumor infiltrating CD4+ and CD8+ T cells have been studied extensively. MHC restricted tumor infiltrating lymphocytes cell lines and clones have been developed from lymphocytes derived from ascites and solid tumors of patients with ovarian cancer [40-44]. A clear association between tumor infiltrating lymphocytes and clinical outcome in ovarian cancer patients has been reported in a landmark paper by Zhang et al[45]. In a large cohort of 186 ovarian cancer patients, the five year survival rate was 38% among patients whose tumors contained T cells and only 4,5% among patients whose tumors contained no T cells. The presence of intratumoral T cells was an independent prognostic factor in a multivariate analysis. One of the other remarkable observations from this study was the correlation between high vascular endothelial growth factor expression and low number of T cells, suggesting that vascular endothelial growth factor reduces the number of T cells. T cells from patients with late-stage ovarian cancer contained increased proportions of regulatory CD25+CD4+ T cells, that secreted the immunosuppressive cytokine TGF-β[9]. In a very elegant study by Curiel et al it was shown that ovarian cancer tumor cells and associated macrophages produce the chemokine CCL22, which mediates trafficking of regulatory T cells in tumors and ascites but not to draining lymph nodes[46]. It was shown that these regulatory T cells suppressed tumor specific T cells and were associated with worse prognosis[46]. The regulatory T cells expressed high levels of CCR4, a receptor for CCL22. By blocking regulatory T cell attracting factors, like CCL22, patients might benefit to a higher extent of immunotherapeutic approaches. Also in the same paper by Curiel it was shown that HER-2/neu specific T cells were blocked by the regulatory T cells in their proliferative function, cytokine production and cytolytic activity. The papers of Zhang et al [45] and Curiel et al [46] seem to have conflicting results with Zhang et al showing a positive correlation between the presence of intratumoral T cells and survival and Curiel et al showing an inverse correlation. However in the first study the total number of T cells was taken into account and in the latter paper only the number of regulatory T cells. One can imagine that ovarian cancer patients with intratumoral T cells have a favorable prognosis as long as regulatory T cells are absent. Nevertheless, it will be important that the data from Zhang et al will be confirmed by others to elucidate the role of intratumoral T cells in ovarian cancer. It has been proposed by Conejo-Garcia et al that the ligand "Letal" (lymphocyte effector cell toxicity-activating ligand), expressed by ovarian cancer tumor cells has a role in survival and expansion of tumor infiltrating lymphocytes [47]. Higher levels of tumor derived "Letal" correlated with stronger lymphocyte infiltration. The same group recently published on a new mechanism of tumor vasculogenesis involving vascular endothelial growth factor in cooperation with antimicrobial inflammatory peptides called β-defensins mediated by a new population of CD11c positive leucocytes (DC precursors) named by these group "vascular leucocytes"' [48,49]. These observations provide a role for the immune system in tumor angiogenesis and need further research to assess what the implications for the clinic could be.
Cytokines and their role in the normal ovary and in ovarian cancer is nicely reviewed by Nash et al[50] and will not be discussed extensively in this review. Ovarian cancer cells probably only partially retain the ability to produce cytokines with important immunostimulatory functions, that are expressed by normal ovarian epithelial cells but lost during neoplastic transformation e.g. the pro-inflammatory cytokine IL-18 [51]. Stat3, a mediator in inflammatory responses and overexpressed in ovarian cancer [52,53], might play an important role in this change in cytokine production by tumor cells suppressing proinflammatory cytokine production[54].
MHC class I down regulation, an often observed immune escape mechanism in different types of cancer, has not been described frequently for ovarian cancer [55-57]. However recently, Vitale et al showed that MHC class I down regulation was associated with higher stage of disease, yet in a multivariate analysis not with survival [58].
The influence of cytoreductive surgery and platinum/paclitaxel based chemotherapy on the immune system in ovarian cancer has not been elucidated up to now. Whether the anti-tumor reactivity in ovarian cancer patients is influenced by surgery and / or chemotherapy remains to be determined. The immunogenicity of dying tumor cells upon chemotherapeutical treatment, does depend on the nature of the cell death (apoptosis or necrosis), but probably as important are local environment and the activation state of the dendritic cells. Platinum based chemotherapy induces apoptosis of ovarian cancer tumor cells. It is therefore encouraging that dendritic cells loaded with autologous apoptotic tumor cells are capable to induce strong tumor specific T cell responses[29]. T cells themselves are susceptible to chemotherapy [59], but high expression of "Letal" by tumor cells protects lymphocytes from cisplatinum induced cell death [47]. For tumor associated antigens like Mov18, OV-TL3 and OC125 only limited differences in expression on the cell surface of ovarian cancer cells were observed before and after chemotherapy[57].
p53 as tumor antigen
General introduction on p53
Specific T cell-mediated immunotherapy requires the identification of tumor-specific antigens carrying T cell epitopes presented in the context of MHC class I and/or MHC class II molecules (reviewed by[19,20,60,61]) An attractive tumor specific antigen in ovarian cancer is the frequently overexpressed and mutated p53 protein. Other possible target antigens like HER-2/neu and MUC-1 are less frequently expressed by ovarian tumor cells. P53 is a tumor suppressor protein. The role of p53 and other cancer genes has been reviewed by Vogelstein and Vousden [62-64]. P53 acts as a transcription factor, playing a key role in coordinating cell cycle arrest, DNA repair and apoptosis following DNA damage to promote genomic stability. P53, as a transcription factor, mediates apoptosis by pathways involving the upregulation of pro-apoptotic genes as well as downregulation of anti-apoptotic genes [65]. P53 also has the capacity to induce apoptosis directly from the cytoplasm via direct activation of Bax to permeabilize mitochondria which will release cytochrome c leading to the induction of apoptosis [66]. In cancer cells loss of wild-type p53 function may lead to more aggressive tumor growth and failure to respond to standard therapy. The most common way of loss of function is through mutation. P53 is one of the most commonly mutated tumor suppressor proteins in human tumors [67], and already more than 4000 different mutations have been described. The majority are point mutations, resulting in single amino-acid substitutions, generally occurring in the central region of the protein (amino acid 100–300). Other tumor suppressor genes often lose their expression after mutation, but the point mutated p53 protein is often more stable and therefore overexpressed in tumor cells. The loss of function of p53 might be due to binding of the mutated protein to the wild type protein (non-functional tetramers) or to loss of the wild type allele (loss of heterozygosity) [67,68]. P53 mutations are associated with poor prognosis. Other ways of inactivation include binding to overexpressed MDM2 or E6 protein of human papillomavirus, both causing rapid p53 protein degradation via the ubiquitin pathway[62,63]. Increased resistance to chemotherapy by mutant p53 has been linked to loss of the presumed triggering role of wild-type p53 in the process of apoptosis.
P53 as tumor antigen (preclinical studies)
P53 protein is overexpressed in 50–60% of ovarian cancers [69-73]. Restoration of the function of p53 in tumor cells is one therapeutic approach. Important progress has been made recently in this field, using viral and non-viral vectors [74], or p53 activating peptides [75]. On the other hand, p53 seems an attractive target for cancer immunotherapy. Due to mutation, nuclear and cytoplasmatic levels of p53 are strongly increased in tumor cells compared to normal cells, thereby providing an immunological window for p53 wild-type specific immune effector cells [76,77]. Still, tolerance against an autoantigen as wild type p53 needs to be overcome, without development of autoreactive T cells. Mutant and wild-type p53 specific CTL have been described in mice [78-85] In mice, eradication of tumors was achieved with vaccines composed of p53 wild type and mutant peptides [81-83], as well as with adoptive transfer of wild type p53 specific T cells [78,85-87]. To immunize with whole p53 protein expressed by e.g. viral vectors or long peptides overlapping a whole protein has the advantage of multiple MHC class I and II restricted epitope expression (dominant as well as cryptic). Mouse dendritic cells transduced with an adenoviral wild type p53 encoding construct generated wild type p53 specific CTL (after i.v. or s.c. immunization) capable of preventing the outgrowth of sarcoma tumors[88,89]. Moreover, the same construct used intratumorally, induced a systemic antitumor response against p53 overexpressing tumors, despite the fact that anti p53 T cell responses could not be measured[90]. Intratumoral injections with recombinant canarypox virus expressing wild type murine p53 (ALVAC-p53) showed antitumor effects in 66% of the mice, however without detectable anti p53 CTL responses [91]. Using different routes of ALVAC-p53 immunizations only intravenous administration was capable of inducing anti-p53 CTL response [92]. More successful than the ALVAC-p53 immunizations in mice was the approach using a recombinant modified vaccinia virus Ankara, expressing wild-type murine p53 (MVAp53). This cell free immunization strategy protected mice for the outgrowth of a syngeneic murine sarcoma by intraperitoneal injection of MVAp53[93]. Mice immunized s.c. with a recombinant vaccinia virus construct expressing wild type p53 were protected against challenge with a p53 overexpressing glioblastome cell line (GL261). Achieving successful p53 based immunization in the presence of well established tumors probably requires active adjuvants. CTLA-4 plays an important role in (negative) regulation of T cell responses [94]. The p53 specific CTL and Th responses can be enhanced by using anti-CTLA-4 at the time of antigenic stimulation, thereby even more effectively breaking tolerance [93,95]. Anti-CTLA-4 blockade in combination with a vaccine adjuvant, CpG ODN (synthetic oligodeoxynucleotide containing unmethylated cytosine-phosphate-guanine motifs) had a synergistic effect on the improvement of MVAp53 induced antitumor immunity[96]. Using MVAp53 based immunization Dafterian et al showed eradication of large, well established tumors in three different tumor models in two different strains of mice[96]. The immune response against p53 can also be enhanced by the activation of CD40 [89,97]. Triggering of the CD40 receptor on dendritic cells is vital for their adequate activation and maturation. Both compounds, anti-CTLA4 and activators of CD40, will become available to test on a wide-based scale in clinical studies within the near future. Another route of enhancement of p53 specific immune response after immunization was obtained by administration of Flt3 Ligand, a strong DC stimulating adjuvant[98]. High steady state levels of p53 are not a pre-requisite for tumor eradication by p53 specific CTL as mentioned in one study[99]. Instead, p53 turnover is an important factor in determining the sensitivity of tumor cells to these CTL [87,100]. CD4+ T helper cells are crucial in the recruitment and regulation of the innate and adaptive immune effector cells[101]. We have demonstrated that CD4+ p53 specific T-helper cells are able to help tumor-specific CTL in controlling p53 overexpressing tumors [102]. Using MHC-transgenic mice has shown to be very efficient in obtaining MHC class I restricted CTL against p53 with high avidity capable of lysing p53 overexpressing tumor cells without lysis of normal cells expressing normal levels of p53 [77]. Very elegantly Kuball et al showed that a CD8-independent p53 specific T cell receptor, generated in HLA A2.1 transgenic mice, could be expressed in human CD8+ and CD4+ T cells with p53 specific tumor recognition[103]. This is at least a very efficient way to obtain p53 specific class I restricted T cells with very high affinity. These model systems might help to answer questions on self tolerance for tumor antigens like p53 and intriguing aspects like cross presentation, cross priming and different aspects of immunotherapy in cancer. So far neither clinical nor immunopathological damage to normal tissue has been observed in different mouse models, despite the fact that wild type p53 is expressed in normal tissue. This indicates that p53 specific T cells are truly tumor-specific. Data available so far support the view that p53 specific immunotherapy may offer a wide therapeutic margin in cancer patients. Proof of the pudding is still in the eating, knowing that their might be important differences in the immune system between preclinical models and men as nicely reviewed by Mestas et al [104].
Cicinnati et al studied the potential of prophylactic vaccination with p53 epitopes using DNA and /or peptide pulsed dendritic cell vaccination in the tumor model giving rise to sarcomas[105]. Compared to control mice a higher incidence of epitope loss tumors were detected in the prophylactic vaccinated group resulting in an increase in tumor growth. Vaccine induced tumor escape therefore could be an important risk in p53 based prophylactic vaccines.
P53 as tumor antigen (clinical studies)
In humans MHC class I restricted p53 specific CTL [106-121], MHC class II restricted p53 specific proliferating Th cells [122-125], and p53 antibody responses (summarized in Table 1) have been observed [123,126-133]. The first phase I/II immunization trials using p53 as an antigen have just finished and new trials are being initiated. In a phase I study, six advanced stage cancer patients were immunized with an adenoviral vector encoding wild type p53[134]. Neither tumor responses nor anti p53 responses were observed, however all patients showed an adenoviral immune response. This strong anti adenoviral specific response may limit a p53 specific response. Based on the results in the mouse system[91,92,135] and rhesus macaques [136], a phase I/II clinical study involving vaccination of end-stage colorectal cancer patients with a recombinant canarypox virus (ALVAC) encoding wild type p53 was performed[137]. Patients were immunized intravenously with an increasing dosage of ALVAC-p53. From this study it appeared that this modality is safe and capable of stimulating p53-specific Th1 (IFNγ) responses in several of these patients. One out of 16 patients showed stable disease for a short period of time after immunization with the highest dose. Fever was the only vaccine related adverse effect. The authors conclude from this trial that repeated immunizations are probably necessary to obtain good clinical responses. Again, anti-vector responses were observed in all patients after vaccination which might have impaired the anti-p53 immune responses. Preclinical data have shown the superiority of prime and boost vaccine strategies using different viral vectors [138,139]. Whether or not the route of administration plays a role is under debate[140]. Clinical studies have shown the safety and effectiveness of prime and boost vaccination protocols using different viral vectors to deliver the antigen of interest[141,142]. An analysis of the p53 specific Th response before and after surgery for colorectal cancer showed that the majority of the Th responses detected were not associated with the immunostimulatory cytokine IFNγ, whereas a number of Th responses even involved secretion of the immunomodulatory cytokine IL-10, pointing at the activity of T-regulatory cells that are known to suppress T cell immunity[143]. These results more or less resemble the cytokine profiles of tumor associated T cells derived from ovarian tumors, which were also associated with a lower zeta chain expression[144]. It is important to further investigate the character of the p53 specific T cell responses, because p53-based vaccination of patients should be aimed at boosting only the desired Th1-type immunity, while stimulation of T-regulatory cells should be avoided. This finding would argue in favor of application of a p53-specific vaccination using a delivery mode specifically stimulating the anti p53 (cytotoxic T cell and) Th1 responses. Autologous dendritic cells expressing the antigen of interest is one of these ways. Svane et al reported on their phase I immunization study in breast cancer patients with p53 peptide pulsed DC[145]. Dendritic cells were pulsed with three wild-type and three modified HLA-A2 restricted p53 peptides combined with a MHC class II binding peptide (PADRE). Patients received ten subcutaneous immunizations with at least 5 × 106 peptide pulsed dendritic cells combined with 6 mIU/m2 IL2. Two out of six patients had a clinical response and three out of six had p53 specific T cell responses (including the two patients with a clinical response), without inducing significant toxicity. Another vaccination strategy would be the use of long peptides encoding the whole protein of interest. The advantage of using long peptides is that, if delivered in the appropriate adjuvant (with dendritic cell stimulatory capacity), all potential MHC class I and class II epitopes within the delivered peptides will be processed and presented to host T cells. Table 2 and 3 summarize the naturally processed wild-type p53 epitopes in MHC class I and II known so far. These vaccines will thus become independent of MHC binding motif prediction or processing algorithms and can be administered to subjects independent of their MHC type. A phase I – II trial using wild- type p53 derived long peptides in ovarian cancer patients will be initiated at the University Medical Center Groningen in 2005.
Table 1 Serum p53 antibodies in patients with epithelial ovarian cancer.
Reference Total no of patients No of patients with p53 serum antibodies (%) Correlation with overall survival
In all patients In patients with stage I/II disease In patients with stage III/IV disease
[146] 86 18 (21) 3 (10) 15 (27) no1
[131] 113 21 (19) 3 (8) 18 (23) yes1,2
[147] 83 38 (46) 5 (26) 33 (52) no2
[148] 193 24 (12) 4 (6) 20 (15) no1,2
[149] 33 12 (36) 3 (21) 9 (47) yes1
[150] 30 10 (33) 2 (22) 8 (38) -
[151] 174 41 (24) 8 (21) 29 (28) no1,2
[133] 113 28 (25) - - no1
[152] 99 25 (25) - - -
[127] 46 4 (9) - - -
[130] 30 8 (27) - - yes1
[153] 30 8 (27) - - -
[129] 40 15 (38) - - -
[126] 46 4 (9) - - -
[154] 38 11(29) - - -
1154 267 (23) 28 (13) 132 (28)
1: tested in an univariate analyses. 2: tested in a multivariate analyses.
Table 2 Naturally processed human wilt-type p53 derived epitopes in MHC class I
Allel amino acid nr. Sequence Reference
HLA-A*0201 65–73 RMPEAAPPV [115,155]
HLA-B*4601 99–107 SQKTYQGSY [117]
HLA-A2 103–111 YQGSYGFRL [120]
HLA-A24 125–134 TYSPALNKMF [156]
HLA-A2 139–147 KTCPVQLWV [120,157]
HLA-A2.1 149–157 STPPPGTRV [84,124]
HLA-A*0201 187–197 GLAPPQHLIRV [115]
HLA-A2 217–225 VPYEPPEVG [118]
HLA-A*0201 264–272 LLFRNSFEV [84,111]
Table 3 Naturally processed human wilt-type p53 derived epitopes in MHC class II
Allel amino acid nr. Sequence Reference
HLA-DR1/HLA-DR4 108–122 GFRLGFLHSGTAKSV [158]
HLA-DRB1*0401 110–124 RLGFLHSGTAKSVTC [124]
HLA-DP5 153–165 PGTRVRAMAIYKQ [125]
HLA-DRB1*1401 193–204 HLIRVEGNLRVE [125]
Conclusion
Progress in the fight against ovarian cancer has been hampered by the lack of highly effective therapy to permanently eradicate disseminated intraperitoneal metastases, which are present in most patients at the time of diagnosis. In order to improve the poor outcome for ovarian cancer patients standard and new treatment modalities, such as targeted or biologic agents and immunotherapy should be combined. In this review we pointed out that ovarian cancer tumor cells may (over)express immunoregulatory molecules such as ligand "Letal", CD40 and Stat-3 which stimulate immune response. On the other hand molecules are expressed which downregulate MHC class I molecules and / or simultaneously produce ligands such as CCL22 attracking regulatory T cells as immune-escape mechanism. Recent data showing the importance of the immune response in the course of ovarian cancer and the availability of new potent immunization strategies urge further exploration of immunotherapy as adjuvant treatment modality in ovarian cancer patients. The immune response against p53 can be enhanced by the activation of CD40, anti CTLA-4 blockade, coadministration of Flt3 Ligand and CpG ODN. Compounds capable of activating or blocking these molecules will become available within the near future to be tested on a wide-based scale in clinical studies. The role of p53 as tumor antigen in ovarian cancer in immunotherapy based trials will be unravled within the near future as well. Next to important issues as safety and immunogenicity of vaccination strategies, clinical effectiveness should be one of the major aims of future trials.
HW Nijman is supported by the Dutch Cancer Society (Grant nr. 2002-2768)
==== Refs
Eisenkop SM Spirtos NM Procedures required to accomplish complete cytoreduction of ovarian cancer: is there a correlation with "biological aggressiveness" and survival? Gynecol Oncol 2001 82 435 441 11520137 10.1006/gyno.2001.6313
Hoskins WJ McGuire WP Brady MF Homesley HD Creasman WT Berman M Ball H Berek JS The effect of diameter of largest residual disease on survival after primary cytoreductive surgery in patients with suboptimal residual epithelial ovarian carcinoma Am J Obstet Gynecol 1994 170 974 979 8166218
Eisenkop SM Friedman RL Wang HJ Complete cytoreductive surgery is feasible and maximizes survival in patients with advanced epithelial ovarian cancer: a prospective study Gynecol Oncol 1998 69 103 108 9600815 10.1006/gyno.1998.4955
McGuire WP Ozols RF Chemotherapy of advanced ovarian cancer Semin Oncol 1998 25 340 348 9633846
Thigpen JT Chemotherapy for advanced ovarian cancer: overview of randomized trials Semin Oncol 2000 27 11 16 10952121
Hung K Hayashi R Lafond-Walker A Lowenstein C Pardoll D Levitsky H The central role of CD4(+) T cells in the antitumor immune response J Exp Med 1998 188 2357 2368 9858522 10.1084/jem.188.12.2357
Godfrey DI Hammond KJ Poulton LD Smyth MJ Baxter AG NKT cells: facts, functions and fallacies Immunol Today 2000 21 573 583 11094262 10.1016/S0167-5699(00)01735-7
Sakaguchi S Regulatory T cells: key controllers of immunologic self-tolerance Cell 2000 101 455 458 10850488 10.1016/S0092-8674(00)80856-9
Woo EY Chu CS Goletz TJ Schlienger K Yeh H Coukos G Rubin SC Kaiser LR June CH Regulatory CD4(+)CD25(+) T cells in tumors from patients with early- stage non-small cell lung cancer and late-stage ovarian cancer Cancer Res 2001 61 4766 4772 11406550
Garza KM Chan SM Suri R Nguyen LT Odermatt B Schoenberger SP Ohashi PS Role of antigen-presenting cells in mediating tolerance and autoimmunity J Exp Med 2000 191 2021 2027 10839816 10.1084/jem.191.11.2021
Smyth MJ Godfrey DI Trapani JA A fresh look at tumor immunosurveillance and immunotherapy Nat Immunol 2001 2 293 299 11276199 10.1038/86297
Zinkernagel RM Ehl S Aichele P Oehen S Kundig T Hengartner H Antigen localisation regulates immune responses in a dose- and time-dependent fashion: a geographical view of immune reactivity Immunol Rev 1997 156 199 209 9176709
Bennett SR Carbone FR Karamalis F Flavell RA Miller JF Heath WR Help for cytotoxic-T-cell responses is mediated by CD40 signalling Nature 1998 393 478 480 9624004 10.1038/30996
Cella M Scheidegger D Palmer-Lehmann K Lane P Lanzavecchia A Alber G Ligation of CD40 on dendritic cells triggers production of high levels of interleukin-12 and enhances T cell stimulatory capacity: T-T help via APC activation J Exp Med 1996 184 747 752 8760829 10.1084/jem.184.2.747
Ridge JP Di Rosa F Matzinger P A conditioned dendritic cell can be a temporal bridge between a CD4+ T- helper and a T-killer cell Nature 1998 393 474 478 9624003 10.1038/30989
Schoenberger SP Toes RE van der Voort EI Offringa R Melief CJ T-cell help for cytotoxic T lymphocytes is mediated by CD40-CD40L interactions Nature 1998 393 480 483 9624005 10.1038/31002
Guerder S Matzinger P A fail-safe mechanism for maintaining self-tolerance J Exp Med 1992 176 553 564 1386876 10.1084/jem.176.2.553
Schuurhuis DH Laban S Toes RE Ricciardi-Castagnoli P Kleijmeer MJ van der Voort EI Rea D Offringa R Geuze HJ Melief CJ Ossendorp F Immature dendritic cells acquire CD8(+) cytotoxic T lymphocyte priming capacity upon activation by T helper cell-independent or -dependent stimuli J Exp Med 2000 192 145 150 10880536 10.1084/jem.192.1.145
Melief CJ Toes RE Medema JP Van der Burg SH Ossendorp F Offringa R Strategies for immunotherapy of cancer Adv Immunol 2000 75 235 282 10879286
Rosenberg SA A new era for cancer immunotherapy based on the genes that encode cancer antigens Immunity 1999 10 281 287 10204484 10.1016/S1074-7613(00)80028-X
Marincola FM Wang E Herlyn M Seliger B Ferrone S Tumors as elusive targets of T-cell-based active immunotherapy Trends Immunol 2003 24 335 342 12810110 10.1016/S1471-4906(03)00116-9
Turtle CJ Hart DN Dendritic cells in tumor immunology and immunotherapy Curr Drug Targets 2004 5 17 39 14738216 10.2174/1389450043490640
Almand B Resser JR Lindman B Nadaf S Clark JI Kwon ED Carbone DP Gabrilovich DI Clinical significance of defective dendritic cell differentiation in cancer Clin Cancer Res 2000 6 1755 1766 10815894
Almand B Clark JI Nikitina E van Beynen J English NR Knight SC Carbone DP Gabrilovich DI Increased production of immature myeloid cells in cancer patients: a mechanism of immunosuppression in cancer J Immunol 2001 166 678 689 11123353
Gabrilovich DI Chen HL Girgis KR Cunningham HT Meny GM Nadaf S Kavanaugh D Carbone DP Production of vascular endothelial growth factor by human tumors inhibits the functional maturation of dendritic cells Nat Med 1996 2 1096 1103 8837607 10.1038/nm1096-1096
Kusmartsev S Gabrilovich DI Immature myeloid cells and cancer-associated immune suppression Cancer Immunol Immunother 2002 51 293 298 12111117 10.1007/s00262-002-0280-8
Zou W Machelon V Coulomb-L'Hermin A Borvak J Nome F Isaeva T Wei S Krzysiek R Durand-Gasselin I Gordon A Pustilnik T Curiel DT Galanaud P Capron F Emilie D Curiel TJ Stromal-derived factor-1 in human tumors recruits and alters the function of plasmacytoid precursor dendritic cells Nat Med 2001 7 1339 1346 11726975 10.1038/nm1201-1339
Eisenthal A Polyvkin N Bramante-Schreiber L Misonznik F Hassner A Lifschitz-Mercer B Expression of dendritic cells in ovarian tumors correlates with clinical outcome in patients with ovarian cancer Hum Pathol 2001 32 803 807 11521223 10.1053/hupa.2001.26455
Schlienger K Chu CS Woo EY Rivers PM Toll AJ Hudson B Maus MV Riley JL Choi Y Coukos G Kaiser LR Rubin SC Levine BL Carroll RG June CH TRANCE- and CD40 Ligand-matured Dendritic Cells reveal MHC class I restricted T cells specific for autologous tumor in late stage ovarian cancer patients. Clin Cancer Res 2003 9 1517 1527 12684428
Chu CS Woo EY Toll AJ Rubin SC June CH Carroll RG Schlienger K Tumor-associated macrophages as a source of functional dendritic cells in ovarian cancer patients Clin Immunol 2002 102 291 301 11890716 10.1006/clim.2001.5179
Loercher AE Nash MA Kavanagh JJ Platsoucas CD Freedman RS Identification of an IL-10-producing HLA-DR-negative monocyte subset in the malignant ascites of patients with ovarian carcinoma that inhibits cytokine protein expression and proliferation of autologous T cells J Immunol 1999 163 6251 6260 10570318
Curiel TJ Wei S Dong H Alvarez X Cheng P Mottram P Krzysiek R Knutson KL Daniel B Zimmermann MC David O Burow M Gordon A Dhurandhar N Myers L Berggren R Hemminki A Alvarez RD Emilie D Curiel DT Chen L Zou W Blockade of B7-H1 improves myeloid dendritic cell-mediated antitumor immunity Nat Med 2003 9 562 567 12704383 10.1038/nm863
Dong H Zhu G Tamada K Chen L B7-H1, a third member of the B7 family, co-stimulates T-cell proliferation and interleukin-10 secretion Nat Med 1999 5 1365 1369 10581077 10.1038/70932
Dong H Chen L B7-H1 pathway and its role in the evasion of tumor immunity J Mol Med 2003 81 281 287 12721664
Dong H Strome SE Salomao DR Tamura H Hirano F Flies DB Roche PC Lu J Zhu G Tamada K Lennon VA Celis E Chen L Tumor-associated B7-H1 promotes T-cell apoptosis: a potential mechanism of immune evasion Nat Med 2002 8 793 800 12091876
Strome SE Dong H Tamura H Voss SG Flies DB Tamada K Salomao D Cheville J Hirano F Lin W Kasperbauer JL Ballman KV Chen L B7-H1 blockade augments adoptive T-cell immunotherapy for squamous cell carcinoma Cancer Res 2003 63 6501 6505 14559843
Colonna M Krug A Cella M Interferon-producing cells: on the front line in immune responses against pathogens Curr Opin Immunol 2002 14 373 379 11973137 10.1016/S0952-7915(02)00349-7
Salio M Cella M Vermi W Facchetti F Palmowski MJ Smith CL Shepherd D Colonna M Cerundolo V Plasmacytoid dendritic cells prime IFN-gamma-secreting melanoma-specific CD8 lymphocytes and are found in primary melanoma lesions Eur J Immunol 2003 33 1052 1062 12672071 10.1002/eji.200323676
Curiel TJ Cheng P Mottram P Alvarez X Moons L Evdemon-Hogan M Wei S Zou L Kryczek I Hoyle G Lackner A Carmeliet P Zou W Dendritic cell subsets differentially regulate angiogenesis in human ovarian cancer Cancer Res 2004 64 5535 5538 15313886
Freedman RS Tomasovic B Templin S Atkinson EN Kudelka A Edwards CL Platsoucas CD Large-scale expansion in interleukin-2 of tumor-infiltrating lymphocytes from patients with ovarian carcinoma for adoptive immunotherapy J Immunol Methods 1994 167 145 160 8308273 10.1016/0022-1759(94)90084-1
Freedman RS Platsoucas CD Immunotherapy for peritoneal ovarian carcinoma metastasis using ex vivo expanded tumor infiltrating lymphocytes Cancer Treat Res 1996 82 115 146 8849947
Ioannides CG Platsoucas CD Rashed S Wharton JT Edwards CL Freedman RS Tumor cytolysis by lymphocytes infiltrating ovarian malignant ascites Cancer Res 1991 51 4257 4265 1868446
Ioannides CG Fisk B Fan D Biddison WE Wharton JT O'Brian CA Cytotoxic T cells isolated from ovarian malignant ascites recognize a peptide derived from the HER-2/neu proto-oncogene Cell Immunol 1993 151 225 234 7691418 10.1006/cimm.1993.1233
Ioannides CG Fisk B Pollack MS Frazier ML Taylor WJ Freedman RS Cytotoxic T-cell clones isolated from ovarian tumour infiltrating lymphocytes recognize common determinants on non-ovarian tumour clones Scand J Immunol 1993 37 413 424 8469924
Zhang L Conejo-Garcia JR Katsaros D Gimotty PA Massobrio M Regnani G Makrigiannakis A Gray H Schlienger K Liebman MN Rubin SC Coukos G Intratumoral T cells, recurrence, and survival in epithelial ovarian cancer N Engl J Med 2003 348 203 213 12529460 10.1056/NEJMoa020177
Curiel TJ Coukos G Zou L Alvarez X Cheng P Mottram P Evdemon-Hogan M Conejo-Garcia JR Zhang L Burow M Zhu Y Wei S Kryczek I Daniel B Gordon A Myers L Lackner A Disis ML Knutson KL Chen L Zou W Specific recruitment of regulatory T cells in ovarian carcinoma fosters immune privilege and predicts reduced survival Nat Med 2004 10 942 949 15322536 10.1038/nm1093
Conejo-Garcia JR Benencia F Courreges MC Gimotty PA Khang E Buckanovich RJ Frauwirth KA Zhang L Katsaros D Thompson CB Levine B Coukos G Ovarian carcinoma expresses the NKG2D ligand Letal and promotes the survival and expansion of Cancer Res 2004 64 2175 2182 15026360
Conejo-Garcia JR Buckanovich RJ Benencia F Courreges MC Rubin SC Carroll RG Coukos G Vascular Leukocytes Contribute to Tumor Vascularization Blood 2004
Conejo-Garcia JR Benencia F Courreges MC Kang E Mohamed-Hadley A Buckanovich RJ Holtz DO Jenkins A Na H Zhang L Wagner DS Katsaros D Caroll R Coukos G Tumor-infiltrating dendritic cell precursors recruited by a beta-defensin contribute to vasculogenesis under the influence of Vegf-A Nat Med 2004 10 950 958 15334073 10.1038/nm1097
Nash MA Ferrandina G Gordinier M Loercher A Freedman RS The role of cytokines in both the normal and malignant ovary Endocr Relat Cancer 1999 6 93 107 10732792 10.1677/erc.0.0060093
Wang ZY Gaggero A Rubartelli A Rosso O Miotti S Mezzanzanica D Canevari S Ferrini S Expression of interleukin-18 in human ovarian carcinoma and normal ovarian epithelium: evidence for defective processing in tumor cells Int J Cancer 2002 98 873 878 11948465 10.1002/ijc.10268
Chen H Ye D Xie X Chen B Lu W VEGF, VEGFRs expressions and activated STATs in ovarian epithelial carcinoma Gynecol Oncol 2004 94 630 635 15350351 10.1016/j.ygyno.2004.05.056
Huang M Page C Reynolds RK Lin J Constitutive activation of stat 3 oncogene product in human ovarian carcinoma cells Gynecol Oncol 2000 79 67 73 11006034 10.1006/gyno.2000.5931
Wang T Niu G Kortylewski M Burdelya L Shain K Zhang S Bhattacharya R Gabrilovich D Heller R Coppola D Dalton W Jove R Pardoll D Yu H Regulation of the innate and adaptive immune responses by Stat-3 signaling in tumor cells Nat Med 2004 10 48 54 14702634 10.1038/nm976
Kooi S Zhang HZ Patenia R Edwards CL Platsoucas CD Freedman RS HLA class I expression on human ovarian carcinoma cells correlates with T-cell infiltration in vivo and T-cell expansion in vitro in low concentrations of recombinant interleukin-2 Cell Immunol 1996 174 116 128 8954611 10.1006/cimm.1996.0301
Nijman HW van Diest PJ Poort-Keesom RJ Mensdorff-Pouilly S Verstraeten RA Kummer A Meijer CJ Melief CJ Hilgers J Kenemans P T cell infiltration and MHC I and II expression in the presence of tumor antigens: An immunohistochemical study in patients with serous epithelial ovarian cancer Eur J Obstet Gynecol Reprod Biol 2001 94 114 120 11134836 10.1016/S0301-2115(00)00294-3
Ravenswaay Claasen HH Fleuren GJ The influence of combination chemotherapy on antigen expression in ovarian cancer Gynecol Oncol 1995 58 16 23 7789884 10.1006/gyno.1995.1177
Vitale M Pelusi G Taroni B Gobbi G Micheloni C Rezzani R Donato F Wang X Ferrone S HLA class I antigen down-regulation in primary ovary carcinoma lesions: association with disease stage Clin Cancer Res 2005 11 67 72 15671529
Zaks TZ Chappell DB Rosenberg SA Restifo NP Fas-mediated suicide of tumor-reactive T cells following activation by specific tumor: selective rescue by caspase inhibition J Immunol 1999 162 3273 3279 10092779
Offringa R Van der Burg SH Ossendorp F Toes RE Melief CJ Design and evaluation of antigen-specific vaccination strategies against cancer Curr Opin Immunol 2000 12 576 582 11007362 10.1016/S0952-7915(00)00145-X
Van Den Eynde BJ van der BP T cell defined tumor antigens Curr Opin Immunol 1997 9 684 693 9368778 10.1016/S0952-7915(97)80050-7
Vogelstein B Lane D Levine AJ Surfing the p53 network Nature 2000 408 307 310 11099028 10.1038/35042675
Vousden KH Lu X Live or let die: the cell's response to p53 Nat Rev Cancer 2002 2 594 604 12154352 10.1038/nrc864
Vogelstein B Kinzler KW Cancer genes and the pathways they control Nat Med 2004 10 789 799 15286780 10.1038/nm1087
Coukos G Rubin SC Chemotherapy resistance in ovarian cancer: new molecular perspectives Obstet Gynecol 1998 91 783 792 9572231 10.1016/S0029-7844(98)00054-4
Chipuk JE Kuwana T Bouchier-Hayes L Droin NM Newmeyer DD Schuler M Green DR Direct activation of Bax by p53 mediates mitochondrial membrane permeabilization and apoptosis Science 2004 303 1010 1014 14963330 10.1126/science.1092734
Greenblatt MS Bennett WP Hollstein M Harris CC Mutations in the p53 tumor suppressor gene: clues to cancer etiology and molecular pathogenesis Cancer Res 1994 54 4855 4878 8069852
de Vries A Flores ER Miranda B Hsieh HM van Oostrom CT Sage J Jacks T Targeted point mutations of p53 lead to dominant-negative inhibition of wild-type p53 function Proc Natl Acad Sci U S A 2002 99 2948 2953 11867759 10.1073/pnas.052713099
Nijman HW Kenemans P Poort-Keesom RJ Verstraeten RA Mensdorff-Pouilly S Verheijen RH Melief CJ Hilgers J Meijer CJ Influence of chemotherapy on the expression of p53, HER-2/neu and proliferation markers in ovarian cancer Eur J Obstet Gynecol Reprod Biol 1999 83 201 206 10391533 10.1016/S0301-2115(98)00317-0
Herod JJ Eliopoulos AG Warwick J Niedobitek G Young LS Kerr DJ The prognostic significance of Bcl-2 and p53 expression in ovarian carcinoma Cancer Res 1996 56 2178 2184 8616869
Klemi PJ Pylkkanen L Kiilholma P Kurvinen K Joensuu H p53 protein detected by immunohistochemistry as a prognostic factor in patients with epithelial ovarian carcinoma Cancer 1995 76 1201 1208 8630898
Marks JR Davidoff AM Kerns BJ Humphrey PA Pence JC Dodge RK Clarke-Pearson DL Iglehart JD Bast RCJ Berchuck A Overexpression and mutation of p53 in epithelial ovarian cancer Cancer Res 1991 51 2979 2984 2032235
van der Zee AG Hollema H Suurmeijer AJ Krans M Sluiter WJ Willemse PH Aalders JG de Vries EG Value of P-glycoprotein, glutathione S-transferase pi, c-erbB-2, and p53 as prognostic factors in ovarian carcinomas J Clin Oncol 1995 13 70 78 7799045
McCormick F Cancer gene therapy: fringe or cutting edge? Nat Rev Cancer 2001 1 130 141 11905804 10.1038/35101008
Snyder EL Meade BR Saenz CC Dowdy SF Treatment of terminal peritoneal carcinomatosis by a transducible p53-activating peptide PLoS Biol 2004 2 E36 14966535 10.1371/journal.pbio.0020036
Offringa R Vierboom MP Van der Burg SH Erdile L Melief CJ p53: a potential target antigen for immunotherapy of cancer Ann N Y Acad Sci 2000 910 223 233 10911916
Voss RH Lotz C Cellary A Theobald M Targeting p53, hdm2, and CD19: vaccination and immunologic strategies Bone Marrow Transplant 2000 25 Suppl 2 S43 S45 10933187 10.1038/sj.bmt.1702353
Hilburger RM Abrams SI Characterization of CD8+ cytotoxic T lymphocyte/tumor cell interactions reflecting recognition of an endogenously expressed murine wild-type p53 determinant Cancer Immunol Immunother 2001 49 603 612 11225991 10.1007/s002620000156
Yanuck M Carbone DP Pendleton CD Tsukui T Winter SF Minna JD Berzofsky JA A mutant p53 tumor suppressor protein is a target for peptide-induced CD8+ cytotoxic T-cells Cancer Res 1993 53 3257 3261 7686815
Mathiassen S Lauemoller SL Ruhwald M Claesson MH Buus S Tumor-associated antigens identified by mRNA expression profiling induce protective anti-tumor immunity Eur J Immunol 2001 31 1239 1246 11298350 10.1002/1521-4141(200104)31:4<1239::AID-IMMU1239>3.0.CO;2-C
Mayordomo JI Loftus DJ Sakamoto H De Cesare CM Appasamy PM Lotze MT Storkus WJ Appella E DeLeo AB Therapy of murine tumors with p53 wild-type and mutant sequence peptide- based vaccines J Exp Med 1996 183 1357 1365 8666894 10.1084/jem.183.4.1357
Noguchi Y Chen YT Old LJ A mouse mutant p53 product recognized by CD4+ and CD8+ T cells Proc Natl Acad Sci U S A 1994 91 3171 3175 7909159
Noguchi Y Richards EC Chen YT Old LJ Influence of interleukin 12 on p53 peptide vaccination against established Meth A sarcoma Proc Natl Acad Sci U S A 1995 92 2219 2223 7892250
Theobald M Biggs J Dittmer D Levine AJ Sherman LA Targeting p53 as a general tumor antigen Proc Natl Acad Sci U S A 1995 92 11993 11997 8618830
Vierboom MP Nijman HW Offringa R van der Voort EI van Hall T van den BL Fleuren GJ Kenemans P Kast WM Melief CJ Tumor eradication by wild-type p53-specific cytotoxic T lymphocytes J Exp Med 1997 186 695 704 9271585 10.1084/jem.186.5.695
Peralta EA Liu X McCarthy TM Wilson TG Diamond DJ Ellenhorn JD Immunotherapy of bladder cancer targeting P53 J Urol 1999 162 1806 1811 10524939 10.1097/00005392-199911000-00074
Vierboom MP Zwaveling S Bos GMJ Ooms M Krietemeijer GM Melief CJ Offringa R High steady-state levels of p53 are not a prerequisite for tumor eradication by wild-type p53-specific cytotoxic T lymphocytes Cancer Res 2000 60 5508 5513 11034095
Ishida T Chada S Stipanov M Nadaf S Ciernik FI Gabrilovich DI Carbone DP Dendritic cells transduced with wild-type p53 gene elicit potent anti-tumour immune responses Clin Exp Immunol 1999 117 244 251 10444254 10.1046/j.1365-2249.1999.00913.x
Nikitina EY Chada S Muro-Cacho C Fang B Zhang R Roth JA Gabrilovich DI An effective immunization and cancer treatment with activated dendritic cells transduced with full-length wild-type p53 Gene Ther 2002 9 345 352 11938454 10.1038/sj.gt.3301670
Murakami T Tokunaga N Waku T Gomi S Kagawa S Tanaka N Fujiwara T Antitumor effect of intratumoral administration of bone marrow-derived dendritic cells transduced with wild-type p53 gene Clin Cancer Res 2004 10 3871 3880 15173096
Odin L Favrot M Poujol D Michot JP Moingeon P Tartaglia J Puisieux I Canarypox virus expressing wild type p53 for gene therapy in murine tumors mutated in p53 Cancer Gene Ther 2001 8 87 98 11263530 10.1038/sj.cgt.7700279
Hurpin C Rotarioa C Bisceglia H Chevalier M Tartaglia J Erdile L The mode of presentation and route of administration are critical for the induction of immune responses to p53 and antitumor immunity Vaccine 1998 16 208 215 9607032 10.1016/S0264-410X(97)00190-4
Espenschied J Lamont J Longmate J Pendas S Wang Z Diamond DJ Ellenhorn JD CTLA-4 blockade enhances the therapeutic effect of an attenuated poxvirus vaccine targeting p53 in an established murine tumor model J Immunol 2003 170 3401 3407 12626601
Krummel MF Allison JP CD28 and CTLA-4 have opposing effects on the response of T cells to stimulation J Exp Med 1995 182 459 465 7543139 10.1084/jem.182.2.459
Hernandez J Ko A Sherman LA CTLA-4 blockade enhances the CTL responses to the p53 self-tumor antigen J Immunol 2001 166 3908 3914 11238635
Daftarian P Song GY Ali S Faynsod M Longmate J Diamond DJ Ellenhorn JD Two distinct pathways of immuno-modulation improve potency of p53 immunization in rejecting established tumors Cancer Res 2004 64 5407 5414 15289349
Erdile LF Smith D CD40 activation enhances the magnitude of cellular immune responses against p53 but not the avidity of the effectors Cancer Immunol Immunother 2000 49 410 416 11043847 10.1007/s002620000135
Parajuli P Pisarev V Sublet J Steffel A Varney M Singh R LaFace D Talmadge JE Immunization with wild-type p53 gene sequences coadministered with Flt3 ligand induces an antigen-specific type 1 T-cell response Cancer Res 2001 61 8227 8234 11719454
Gnjatic S Cai Z Viguier M Chouaib S Guillet JG Choppin J Accumulation of the p53 protein allows recognition by human CTL of a wild-type p53 epitope presented by breast carcinomas and melanomas J Immunol 1998 160 328 333 9551988
Sirianni N Ha PK Oelke M Califano J Gooding W Westra W Whiteside TL Koch WM Schneck JP DeLeo A Ferris RL Effect of human papillomavirus-16 infection on CD8+ T-cell recognition of a wild-type sequence p53264-272 peptide in patients with squamous cell carcinoma of the head and neck Clin Cancer Res 2004 10 6929 6937 15501971
Mitra R Singh S Khar A Antitumour immune responses Expert Rev Mol Med 2003 2003 1 22 14987405 10.1017/S1462399403005623
Zwaveling S Vierboom MP Ferreira Mota SC Hendriks JA Ooms ME Sutmuller RP Franken KL Nijman HW Ossendorp F Van der Burg SH Offringa R Melief CJ Antitumor efficacy of wild-type p53-specific CD4(+) T-helper cells Cancer Res 2002 62 6187 6193 12414646
Kuball J Schmitz FW Voss RH Ferreira EA Engel R Guillaume P Strand S Romero P Huber C Sherman LA Theobald M Cooperation of human tumor-reactive CD4+ and CD8+ T cells after redirection of their specificity by a high-affinity p53A2.1-specific TCR Immunity 2005 22 117 129 15664164 10.1016/j.immuni.2004.12.005
Mestas J Hughes CC Of mice and not men: differences between mouse and human immunology J Immunol 2004 172 2731 2738 14978070
Cicinnati VR Dworacki G Albers A Beckebaum S Tuting T Kaczmarek E DeLeo AB Impact of p53-based immunization on primary chemically-induced tumors Int J Cancer 2005 113 961 970 15514940 10.1002/ijc.20686
Hoffmann TK Nakano K Elder EM Dworacki G Finkelstein SD Appella E Whiteside TL DeLeo AB Generation of T cells specific for the wild-type sequence p53(264-272) peptide in cancer patients: implications for immunoselection of epitope loss variants J Immunol 2000 165 5938 5944 11067956
Houbiers JG Nijman HW Van der Burg SH Drijfhout JW Kenemans P van de Velde CJ Brand A Momburg F Kast WM Melief CJ In vitro induction of human cytotoxic T lymphocyte responses against peptides of mutant and wild-type p53 Eur J Immunol 1993 23 2072 2077 8370389
Nijman HW Houbiers JG Van der Burg SH Vierboom MP Kenemans P Kast WM Melief CJ Characterization of cytotoxic T lymphocyte epitopes of a self-protein, p53, and a non-self-protein, influenza matrix: relationship between major histocompatibility complex peptide binding affinity and immune responsiveness to peptides J Immunother 1993 14 121 126 7506574
Nijman HW Van der Burg SH Vierboom MP Houbiers JG Kast WM Melief CJ p53, a potential target for tumor-directed T cells Immunol Lett 1994 40 171 178 8088874 10.1016/0165-2478(94)90189-9
Nikitina EY Clark JI van Beynen J Chada S Virmani AK Carbone DP Gabrilovich DI Dendritic cells transduced with full-length wild-type p53 generate antitumor cytotoxic T lymphocytes from peripheral blood of cancer patients Clin Cancer Res 2001 7 127 135 11205900
Ropke M Regner M Claesson MH T cell-mediated cytotoxicity against p53-protein derived peptides in bulk and limiting dilution cultures of healthy donors Scand J Immunol 1995 42 98 103 7543215
Ropke M Hald J Guldberg P Zeuthen J Norgaard L Fugger L Svejgaard A Van der BS Nijman HW Melief CJ Claesson MH Spontaneous human squamous cell carcinomas are killed by a human cytotoxic T lymphocyte clone recognizing a wild-type p53-derived peptide Proc Natl Acad Sci U S A 1996 93 14704 14707 8962118 10.1073/pnas.93.25.14704
Theobald M Biggs J Hernandez J Lustgarten J Labadie C Sherman LA Tolerance to p53 by A2.1-restricted cytotoxic T lymphocytes J Exp Med 1997 185 833 841 9120389 10.1084/jem.185.5.833
Umano Y Tsunoda T Tanaka H Matsuda K Yamaue H Tanimura H Generation of cytotoxic T cell responses to an HLA-A24 restricted epitope peptide derived from wild-type p53 Br J Cancer 2001 84 1052 1057 11308253 10.1054/bjoc.2000.1715
Wurtzen PA Pedersen LO Poulsen HS Claesson MH Specific killing of P53 mutated tumor cell lines by a cross-reactive human HLA-A2-restricted P53-specific CTL line Int J Cancer 2001 93 855 861 11519048 10.1002/ijc.1417
Asai T Storkus WJ Mueller-Berghaus J Knapp W DeLeo AB Chikamatsu K Whiteside TL In vitro generated cytolytic T lymphocytes reactive against head and neck cancer recognize multiple epitopes presented by HLA-A2, including peptides derived from the p53 and MDM-2 proteins Cancer Immun 2002 2 3 12747748
Azuma K Shichijo S Maeda Y Nakatsura T Nonaka Y Fujii T Koike K Itoh K Mutated p53 gene encodes a nonmutated epitope recognized by HLA-B*4601-restricted and tumor cell-reactive CTLs at tumor site Cancer Res 2003 63 854 858 12591737
McArdle SE Rees RC Mulcahy KA Saba J McIntyre CA Murray AK Induction of human cytotoxic T lymphocytes that preferentially recognise tumour cells bearing a conformational p53 mutant Cancer Immunol Immunother 2000 49 417 425 11043848 10.1007/s002620000137
Papadopoulos KP Hesdorffer CS Suciu-Foca N Hibshoosh H Harris PE Wild-type p53 epitope naturally processed and presented by an HLA-B haplotype on human breast carcinoma cells Clin Cancer Res 1999 5 2089 2093 10473091
Petersen TR Buus S Brunak S Nissen MH Sherman LA Claesson MH Identification and design of p53-derived HLA-A2-binding peptides with increased CTL immunogenicity Scand J Immunol 2001 53 357 364 11285115 10.1046/j.1365-3083.2001.00887.x
Tokunaga N Murakami T Endo Y Nishizaki M Kagawa S Tanaka N Fujiwara T Human monocyte-derived dendritic cells pulsed with wild-type p53 protein efficiently induce CTLs against p53 overexpressing human cancer cells Clin Cancer Res 2005 11 1312 1318 15709203
Tilkin AF Lubin R Soussi T Lazar V Janin N Mathieu MC Lefrere I Carlu C Roy M Kayibanda M . Primary proliferative T cell response to wild-type p53 protein in patients with breast cancer Eur J Immunol 1995 25 1765 1769 7615005
Van der Burg SH de Cock K Menon AG Franken KL Palmen M Redeker A Drijfhout J Kuppen PJ van V Erdile L Tollenaar RA Melief CJ Offringa R Long lasting p53-specific T cell memory responses in the absence of anti-p53 antibodies in patients with resected primary colorectal cancer Eur J Immunol 2001 31 146 155 11169448 10.1002/1521-4141(200101)31:1<146::AID-IMMU146>3.0.CO;2-T
Chikamatsu K Albers A Stanson J Kwok WW Appella E Whiteside TL DeLeo AB P53(110-124)-specific human CD4+ T-helper cells enhance in vitro generation and antitumor function of tumor-reactive CD8+ T cells Cancer Res 2003 63 3675 3681 12839958
Fujita H Senju S Yokomizo H Saya H Ogawa M Matsushita S Nishimura Y Evidence that HLA class II-restricted human CD4+ T cells specific to p53 self peptides respond to p53 proteins of both wild and mutant forms Eur J Immunol 1998 28 305 316 9485210 10.1002/(SICI)1521-4141(199801)28:01<305::AID-IMMU305>3.0.CO;2-3
Angelopoulou K Rosen B Stratis M Yu H Solomou M Diamandis EP Circulating antibodies against p53 protein in patients with ovarian carcinoma. Correlation with clinicopathologic features and survival Cancer 1996 78 2146 2152 8918407 10.1002/(SICI)1097-0142(19961115)78:10<2146::AID-CNCR15>3.0.CO;2-Z
Angelopoulou K Diamandis EP Detection of the TP53 tumour suppressor gene product and p53 auto- antibodies in the ascites of women with ovarian cancer Eur J Cancer 1997 33 115 121 9071910 10.1016/S0959-8049(96)00295-X
Crawford LV Pim DC Bulbrook RD Detection of antibodies against the cellular protein p53 in sera from patients with breast cancer Int J Cancer 1982 30 403 408 6292117
Gadducci A Ferdeghini M Buttitta F Fanucchi A Annicchiarico C Prontera C Bevilacqua G Genazzani AR Preoperative serum antibodies against the p53 protein in patients with ovarian and endometrial cancer Anticancer Res 1996 16 3519 3523 9042215
Gadducci A Ferdeghini M Buttitta F Cosio S Fanucchi A Annicchiarico C Genazzani AR Serum anti-p53 antibodies in the follow-up of patients with advanced ovarian carcinoma Anticancer Res 1998 18 3763 3765 9854491
Gadducci A Ferdeghini M Buttitta F Cosio S Fanucchi A Annicchiarico C Gagetti O Bevilacqua G Genazzani AR Assessment of the prognostic relevance of serum anti-p53 antibodies in epithelial ovarian cancer Gynecol Oncol 1999 72 76 81 9889034 10.1006/gyno.1998.5101
Labrecque S Naor N Thomson D Matlashewski G Analysis of the anti-p53 antibody response in cancer patients Cancer Res 1993 53 3468 3471 8339249
Vennegoor CJ Nijman HW Drijfhout JW Vernie L Verstraeten RA Mensdorff-Pouilly S Hilgers J Verheijen RH Kast WM Melief CJ Kenemans P Autoantibodies to p53 in ovarian cancer patients and healthy women: a comparison between whole p53 protein and 18-mer peptides for screening purposes Cancer Lett 1997 116 93 101 9177463 10.1016/S0304-3835(97)00168-7
Kuball J Schuler M Antunes FE Herr W Neumann M Obenauer-Kutner L Westreich L Huber C Wolfel T Theobald M Generating p53-specific cytotoxic T lymphocytes by recombinant adenoviral vector-based vaccination in mice, but not man Gene Ther 2002 9 833 843 12080377 10.1038/sj.gt.3301709
Roth J Dittmer D Rea D Tartaglia J Paoletti E Levine AJ p53 as a target for cancer vaccines: recombinant canarypox virus vectors expressing p53 protect mice against lethal tumor cell challenge Proc Natl Acad Sci U S A 1996 93 4781 4786 8643480 10.1073/pnas.93.10.4781
Rosenwirth B Kuhn EM Heeney JL Hurpin C Tartaglia J Bonnet MC Moingeon P Erdile L Safety and immunogenicity of ALVAC wild-type human p53 (vCP207) by the intravenous route in rhesus macaques Vaccine 2001 19 1661 1670 11166889 10.1016/S0264-410X(00)00416-3
Menon AG Kuppen PJ Van der Burg SH Offringa R Bonnet MC Harinck BI Tollenaar RA Redeker A Putter H Moingeon P Morreau H Melief CJ van de Velde CJ Safety of intravenous administration of a canarypox virus encoding the human wild-type p53 gene in colorectal cancer patients Cancer Gene Ther 2003 10 509 517 12833131 10.1038/sj.cgt.7700600
Aarts WM Schlom J Hodge JW Vector-based vaccine/cytokine combination therapy to enhance induction of immune responses to a self-antigen and antitumor activity Cancer Res 2002 62 5770 5777 12384537
Grosenbach DW Barrientos JC Schlom J Hodge JW Synergy of vaccine strategies to amplify antigen-specific immune responses and antitumor effects Cancer Res 2001 61 4497 4505 11389081
Kudo-Saito C Schlom J Hodge JW Induction of an antigen cascade by diversified subcutaneous/intratumoral vaccination is associated with antitumor responses Clin Cancer Res 2005 11 2416 2426 15788693
Marshall JL Hoyer RJ Toomey MA Faraguna K Chang P Richmond E Pedicano JE Gehan E Peck RA Arlen P Tsang KY Schlom J Phase I study in advanced cancer patients of a diversified prime-and-boost vaccination protocol using recombinant vaccinia virus and recombinant nonreplicating avipox virus to elicit anti-carcinoembryonic antigen immune responses J Clin Oncol 2000 18 3964 3973 11099326
Marshall JL Gulley JL Arlen PM Beetham PK Tsang KY Slack R Hodge JW Doren S Grosenbach DW Hwang J Fox E Odogwu L Park S Panicali D Schlom J Phase I study of sequential vaccinations with fowlpox-CEA(6D)-TRICOM alone and sequentially with vaccinia-CEA(6D)-TRICOM, with and without granulocyte-macrophage colony-stimulating factor, in patients with carcinoembryonic antigen-expressing carcinomas J Clin Oncol 2005 23 720 731 15613691 10.1200/JCO.2005.10.206
Van der Burg SH Menon AG Redeker A Franken KL Drijfhout JW Tollenaar RA Hartgrink HH van de Velde CJ Kuppen PJ Melief CJ Offringa R Magnitude and polarization of P53-specific T-helper immunity in connection to leukocyte infiltration of colorectal tumors Int J Cancer 2003 107 425 433 14506743 10.1002/ijc.11419
Rabinowich H Suminami Y Reichert TE Crowley-Nowick P Bell M Edwards R Whiteside TL Expression of cytokine genes or proteins and signaling molecules in lymphocytes associated with human ovarian carcinoma Int J Cancer 1996 68 276 284 8903466 10.1002/(SICI)1097-0215(19961104)68:3<276::AID-IJC2>3.0.CO;2-Z
Svane IM Pedersen AE Johnsen HE Nielsen D Kamby C Gaarsdal E Nikolajsen K Buus S Claesson MH Vaccination with p53-peptide-pulsed dendritic cells, of patients with advanced breast cancer: report from a phase I study Cancer Immunol Immunother 2004 53 633 641 14985857 10.1007/s00262-003-0493-5
Abendstein B Marth C Muller-Holzner E Widschwendter M Daxenbichler G Zeimet AG Clinical significance of serum and ascitic p53 autoantibodies in epithelial ovarian carcinoma Cancer 2000 88 1432 1437 10717627 10.1002/(SICI)1097-0142(20000315)88:6<1432::AID-CNCR22>3.0.CO;2-8
Vogl FD Frey M Kreienberg R Runnebaum IB Autoimmunity against p53 predicts invasive cancer with poor survival in patients with an ovarian mass Br J Cancer 2000 83 1338 1343 11044359 10.1054/bjoc.2000.1446
Vogl FD Stickeler E Weyermann M Kohler T Grill HJ Negri G Kreienberg R Runnebaum IB p53 autoantibodies in patients with primary ovarian cancer are associated with higher age, advanced stage and a higher proportion of p53-positive tumor cells Oncology 1999 57 324 329 10575320 10.1159/000012069
Hogdall EV Hogdall CK Blaakaer J Heegaard NH Glud E Christensen L Bock JE Norgaard-Pedersen B Wiik A Kjaer SK P53 autoantibodies in sera from Danish ovarian cancer patients and their correlation with clinical data and prognosis APMIS 2002 110 545 553 12390412 10.1034/j.1600-0463.2002.11007805.x
Mayerhofer K Tempfer C Kucera E Hefler L Zeisler H Kainz C Zeillinger R Sliutz G Humoral p53 antibody response is a prognostic parameter in ovarian cancer Anticancer Res 1999 19 875 878 10216509
Marx D Frey M Zentgraf H Adelssen G Schauer A Kuhn W Meden H Detection of serum autoantibodies to tumor suppressor gene p53 with a new enzyme-linked immunosorbent assay in patients with ovarian cancer Cancer Detect Prev 2001 25 117 122 11341346
Numa F Umayahara K Suehiro Y Hirakawa H Nawata S Suminami Y Oga A Ito T Sasaki K Kato H Serum anti-p53 antibodies in uterine and ovarian cancer: association with dna sequence copy number abnormalities Tumour Biol 2001 22 162 168 11275794 10.1159/000050611
Montenarh M Harlozinska A Bar JK Kartarius S Gunther J Sedlaczek P p53 autoantibodies in the sera, cyst and ascitic fluids of patients with ovarian cancer Int J Oncol 1998 13 605 610 9683801
Green JA Robertson LJ Campbell IR Jenkins J Expression of the p53 gene and presence of serum autoantibodies in ovarian cancer: correlation with differentiation Cancer Detect Prev 1995 19 151 155 7750102
Barfoed AM Petersen TR Kirkin AF Thor SP Claesson MH Zeuthen J Cytotoxic T-lymphocyte clones, established by stimulation with the HLA- A2 binding p5365-73 wild type peptide loaded on dendritic cells In vitro, specifically recognize and lyse HLA-A2 tumour cells overexpressing the p53 protein Scand J Immunol 2000 51 128 133 10652158 10.1046/j.1365-3083.2000.00668.x
Eura M Chikamatsu K Katsura F Obata A Sobao Y Takiguchi M Song Y Appella E Whiteside TL DeLeo AB A wild-type sequence p53 peptide presented by HLA-A24 induces cytotoxic T lymphocytes that recognize squamous cell carcinomas of the head and neck Clin Cancer Res 2000 6 979 986 10741724
Wurtzen PA Claesson MH A HLA-A2 restricted human CTL line recognizes a novel tumor cell expressed p53 epitope Int J Cancer 2002 99 568 572 11992547 10.1002/ijc.10375
Rojas JM McArdle SE Horton RB Bell M Mian S Li G Ali SA Rees RC Peptide immunisation of HLA-DR-transgenic mice permits the identification of a novel HLA-DRbeta1* Cancer Immunol Immunother 2005 54 243 253 15449037 10.1007/s00262-004-0596-7
|
16164749
|
PMC1243238
|
CC BY
|
2021-01-04 16:39:29
|
no
|
J Transl Med. 2005 Sep 15; 3:34
|
utf-8
|
J Transl Med
| 2,005 |
10.1186/1479-5876-3-34
|
oa_comm
|
==== Front
Mol CancerMolecular Cancer1476-4598BioMed Central London 1476-4598-4-321615329510.1186/1476-4598-4-32ResearchhZIP1 zinc uptake transporter down regulation and zinc depletion in prostate cancer Franklin Renty B [email protected] Pei [email protected] B [email protected] Mohamed M [email protected] Keshav K [email protected] André [email protected] Omar [email protected] Leslie C [email protected] Department of Biomedical Sciences, Dental School. University of Maryland, Baltimore, Md, USA2 Department of Cancer Genetics, Roswell Park Cancer Institute, Buffalo, NY, USA3 Department of Pathology, University of Illinois, Chicago, IL, USA4 Department of Biology; South Carolina Center for Biotechnology; Claflin University, Orangeburg, SC, USA2005 9 9 2005 4 32 32 14 4 2005 9 9 2005 Copyright © 2005 Franklin et al; licensee BioMed Central Ltd.2005Franklin et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The genetic and molecular mechanisms responsible for and associated with the development and progression of prostate malignancy are largely unidentified. The peripheral zone is the major region of the human prostate gland where malignancy develops. The normal peripheral zone glandular epithelium has the unique function of accumulating high levels of zinc. In contrast, the ability to accumulate zinc is lost in the malignant cells. The lost ability of the neoplastic epithelial cells to accumulate zinc is a consistent factor in their development of malignancy. Recent studies identified ZIP1 (SLC39A1) as an important zinc transporter involved in zinc accumulation in prostate cells. Therefore, we investigated the possibility that down-regulation of hZIP1 gene expression might be involved in the inability of malignant prostate cells to accumulate zinc. To address this issue, the expression of hZIP1 and the depletion of zinc in malignant versus non-malignant prostate glands of prostate cancer tissue sections were analyzed. hZIP1 expression was also determined in malignant prostate cell lines.
Results
hZIP1 gene expression, ZIP1 transporter protein, and cellular zinc were prominent in normal peripheral zone glandular epithelium and in benign hyperplastic glands (also zinc accumulating glands). In contrast, hZIP1 gene expression and transporter protein were markedly down-regulated and zinc was depleted in adenocarcinomatous glands and in prostate intra-epithelial neoplastic foci (PIN). These changes occur early in malignancy and are sustained during its progression in the peripheral zone. hZIP1 is also expressed in the malignant cell lines LNCaP, PC-3, DU-145; and in the nonmalignant cell lines HPr-1 and BPH-1.
Conclusion
The studies clearly establish that hZIP1 gene expression is down regulated and zinc is depleted in adenocarcinomatous glands. The fact that all the malignant cell lines express hZIP1 indicates that the down-regulation in adenocarcinomatous glands is likely due to in situ gene silencing. These observations, coupled with the numerous and consistent reports of loss of zinc accumulation in malignant cells in prostate cancer, lead to the plausible proposal that down regulation of hZIP1 is a critical early event in the development prostate cancer.
prostate cancerzincZIP1 zinc transportercitrateZIP1 gene expression
==== Body
Background
Despite the extensive clinical and experimental studies over the recent decades, the pathogenesis of prostate cancer remains unknown. The genetic and molecular mechanisms responsible for and associated with the development of malignant prostate cells and their progression are largely unidentified [for reviews see [1,2]]. The major site for the development of prostate malignancy is the peripheral zone, which comprises about 70% of the prostate gland. It is well established that the normal peripheral zone has the function of accumulating extremely high zinc levels that are 3–10-fold greater than found in other soft tissues [3]. This capability resides in the highly specialized glandular secretory epithelial cells of the peripheral zone, which we characterize as "zinc-accumulating" cells. In contrast, the malignant prostate cells that develop in the peripheral zone do not contain the high zinc levels that characterize the normal secretory epithelial cells. Repeated studies consistently show that the zinc levels of malignant prostate tissue are 62–75% lower than the normal prostate tissue [4-8]. Measurements of pure malignant tissue in the absence of normal glandular epithelium would reveal even lower zinc levels that would approximate the levels found in other soft tissues. This consistency persists in different reports by different investigators employing different populations and tissue samples and involving various stages of malignancy. The studies of Zaichick et al [9] and Vartsky et al [10] further reveal the critically important relationship that, in individual analyses, malignant prostate tissue never exhibits high zinc levels. In addition, Habib [11] reported that the decrease in zinc occurs early in malignancy. These persistent results, and the additional corroborating evidence presented below, firmly establish that the unique zinc-accumulating capability of the normal peripheral zone secretory epithelial cells is lost in the neoplastic transformation to malignant cells; and that zinc-accumulating malignant cells do not exist in situ in prostate cancer. For extensive presentations of the relationships of zinc in normal prostate and prostate cancer, we refer the reader to our recent reviews [12-14].
Established clinical and experimental evidence provides the basis for our concept that zinc accumulation prevents the malignant activities of the neoplastic prostate cell; and that impaired zinc accumulation is an essential requirement for the manifestation of prostate malignancy. If such is the case, one should expect that the zinc-accumulating process that characterizes the normal glandular epithelium is absent or defective in the malignant cells. Until recently, no information had been available regarding the mechanism(s) of zinc accumulation in prostate cells. Recent studies [15-17] have established that the zinc uptake transporter, ZIP1, is important in the uptake and accumulation of zinc by prostate cells. Up-regulation of ZIP1 in prostate cells increases zinc accumulation; and, correspondingly, down-regulation of ZIP1 decreases zinc accumulation in prostate cells. In addition, Rishi et al [18] reported that ZIP1 (and ZIP2) expression in peripheral zone glandular epithelium of black males is down regulated when compared to its expression in white males; which coincides with the race-associated higher incidence of prostate cancer in African-Americans. These relationships suggested that the decrease in zinc in malignant prostate glands might be due to the down regulation of ZIP1 expression. In this report we show, for the first time, the down regulation of hZIP1 gene expression, the loss of ZIP1 transporter protein and the depletion of zinc that is evident in malignant prostate glands. The evidence presented supports the likelihood that down regulation of ZIP1 gene expression in the neoplastic prostate cell is an essential step in the development of prostate malignancy. The studies were conducted independently at three different institutions, which strengthens the validity of these corroborating results.
Results
The studies presented in this report were conducted at the University of Maryland (UMaryland study), the Roswell Park Cancer Institute (Roswell Park Study), and Claflin University (ClaflinU study). Therefore the results will be presented as provided by each separate and independent study, followed by the discussion of the evidence and supporting basis for the genetic/metabolic concept of the role of zinc in prostate malignancy.
The UMaryland Study (RBF, PF, BM, LCC)
Earlier studies [15-17] demonstrated that ZIP1 is expressed in malignant prostate cell lines (PC-3 and LNCaP cells); and that this zinc uptake transporter functions in the uptake and cellular accumulation of zinc. This caused us to initiate preliminary studies to determine if ZIP1 gene expression and/or the level of the transporter protein might be down-regulated in malignant prostate glands in comparison to the expression in normal prostate glandular epithelium. Paraffin mounted serial sections of human prostate tissue were used for ZIP1 immumohistochemistry staining. Hematoxylin and eosin staining was used for pathologic evaluation of normal glands and adencarcinomatous foci. Figure 1A reveals the membrane-associated immunohistochemical identification of ZIP1 in the normal peripheral zone glandular epithelium. In contrast, the malignant glands were essentially devoid of demonstrable membrane-associated ZIP1. It is also apparent that ZIP1 is confined to glandular epithelium and is not demonstrable in the stromal tissue. Figure 1B presents RT-PCR analysis of ZIP1 expression in tissue extracts of malignant tissue versus benign hyperplastic (BPH) glands; which, like normal peripheral zone, are zinc-accumulating glands. The results demonstrate a relatively high level of ZIP1 gene expression in BPH glandular tissue as compared with a barely detectable expression level in malignant tissue. These results provided the initial preliminary evidence that indicated that down regulation of ZIP1 expression is associated with malignant prostate tissue.
Figure 1 (A) Immunohistochemical determination of ZIP 1 transporter levels in normal and malignant prostate glands. The strong positive reaction is evident in the normal gland secretory epithelial cells that border the lumen, and is virtually absent in the malignant glands. Note that ZIP1 is not apparent in the stroma. (B) RT-PCR of RNA extracted from malignant prostate tissue and benign prostatic hyperplasia. Note the marked decrease in ZIP1 mRNA in the malignant tissue. Results are representative of two independent samples. Density of the bands was determined by densitometry scans and GAPDH band intensity used to normalize hZIP1 mRNA. hZIP1/GAPDH for PCa and BPH were 0.71 ± 0.067 and 1.02 ± 0.092 respectively. (C) Immunohistochemical detection of ZIP1 in malignant prostate cell lines. Note the association of ZIP1 with the plasma membrane.
We previously reported the identification by Western blot of the presence of ZIP1 in PC-3 and LNCaP cells under standard culture conditions. These are malignant cell lines that were derived from metastatic prostate tissue. For correlation with the human tissue results, we proceeded to determine the presence of ZIP1 transporter in these cells by immunocytochemistry. Figure 1C shows the localization of ZIP1 in the plasma membrane; which is similar to the localization in normal peripheral zone glandular epithelium. The retention of this gene expression in LNCaP, PC-3, and DU145 (not shown) cells demonstrates that the absence of ZIP1 expression in the malignant glands in situ is not due to the deletion or fatal mutation of the gene. No information exists regarding ZIP1 in metastatic cells in situ in prostate cancer. However, it seems most improbable that the gene would re-appear in metastasis, unless it was reversibly down-regulated in the primary site malignant glands. Therefore the results strongly implicate the epigenetic silencing of hZIP1 gene expression in the primary site malignant cells under the in situ environmental conditions of the malignant prostate gland.
These initial observations dictated the importance of expanding the clinical investigation to establish conclusively that ZIP1 is down regulated in prostate malignancy and is associated with a decrease in zinc accumulation in the malignant cells. To achieve this, independent studies were conducted at Roswell Park Cancer Institute and at Claflin University without prior knowledge of the results of the UMaryland study.
2. The Roswell Park Study (MMD, KKS)
The Roswell Park (RPCI) resources provided the opportunity to conduct ZIP1 immunohistochemical analysis of prostatic adenocarcinoma slides without identification related to patients. Twenty-two cases of prostatic adenocarcinoma were obtained from RPCI that contained both adenocarcinomatous foci and adjacent benign prostatic hyperplasia (table 1). Four of the cases contained normal prostatic glands and five cases contained prostatic intra epithelial neoplastic foci (PIN). The tumors were graded according to the World Health Organization grading system [19]. Grade 1 is defined by well differentiated glands with minimal anaplasia in which the nuclei are almost uniform with minimal variation in size and shape, and few detectable nucleoli. Grade 2 is defined by moderately differentiated glands with moderate nuclear anaplasia with many nucleoli. Grade 3 is defined by poorly differentiated or undifferentiated glands showing marked anaplasia in which the nuclei showed marked variation in size and irregular shapes, vesicular, with marked abnormal mitotic figures.
Table 1 ZIP1 immuno-positivity of glandular components in tissue sections of confirmed cases of prostate cancer.
Case no. Grade ZIP1 IHC scorea
Normal PIN BPH Malignant
1 3 +++ +++ Negative
2 3 + ++ +
3 1 Negative + Negative
4 2 + Negative
5 2 Negative Negative
6 1 Negative +++ Negative
7 2 ++ + +
8 1 +++ +
9 1 ++ +++ Negative
10 2 +++ Negative
11 1 ++ Negative
12 2 Negative Negative
13 1 ++ +
14 1 + Negative
15 1 + Negative
16 1 Negative + Negative
17 2 +++ +
18 1 + ++ +
19 1 + Negative
20 1 + Negative
21 2 + +++ +
22 1 Negative Negative
NEG ZIP1 IHC 3/22 (14%) 15/22 (68%)*
SCORES > + 3/4 0/6 11/22 0/22*
MEAN SCOREb 1.75 0.6 1.68(1.09) 0.32(0.48)*
a Scoring of immunoreactivity was done as follows: negative, no positive cells; score +, <10% positive cells; score ++, 10–50% positive cells; score +++, >50% positive cells
b MEAN SCORE: MEAN(SD) for each group was obtained by the sum of the +'s/number of cases.
* P < 0.01; BPH VS MALIGNANT GROUPS
Figure 2 shows the representative results of the ZIP1 immunohistochemical staining observed in normal peripheral zone glands, BPH glands, adenocarcinomatous glands and in PIN. The glandular epithelium of the normal glands and BPH glands (both being zinc-accumulating glands) exhibit immuno-positive ZIP1 staining that is localized predominantly at the basolateral membrane. In contrast, in the adenocarcinomatous glands and PIN, ZIP1 is negligible in the malignant cells so that the appearance of cell membranes is essentially absent. It is also evident that ZIP1 transporter is not detected in the stromal tissue, which corroborates the results of the U.Maryland preliminary study.
Figure 2 Immunohistochemical detection of ZIP1 transporter protein in malignant and nonmalignant loci of a representative prostate cancer tissue section. (A) BPH, magnification is 1000×, bar = 10 μm. (B) Normal, magnification is 400×, bar = 25 μm. (C) PIN, magnification is 400×, bars = 25 μm. (D) Adenocarcinoma, magnification 400×, bar = 10 μm Note the immuno-positivity of the plasma membrane of BPH and normal glands. The malignant and PIN loci show no detectable ZIP1 so that the plasma membrane of these cells is not visible.
Table 1 is the summary of the immunohistochemical scoring of hZIP1 reactivity of tissue sections from 22 cases of prostate cancer. The analysis involves the comparison of ZIP1 in glands located in the same tissue section. This eliminates, or at least minimizes, any potential technical differences arising from antibody diffusion into the tissue sections and cells for immuno-reactivity. Any comparative differences observed in the immuno-reactivity in the different glands of the same tissue slice would be due to comparative differences in the level of hZIP1. Analysis of the 22 cases (figure 3) for the presence of glands that exhibit ZIP1 immuno-positivity results in a significant difference (P < 0.01) between BPH glands (19 positive/3 negative) and adenocarcinomatous glands (7 positive/15 negative). Analysis for the presence of acini composed of >10% positive cells reveals that BPH glands exhibited this criterion in 50% (11/22) of the cases compared to 0/22 for the adenocarcinmatous glands (figure 3). The average scoring for the twenty-two cases (table 1) was also significantly lower (P < 0.01) for the adenocarcinomatous glands (0.32) as compared to the BPH glands (1.68); i.e. ~5-fold difference. Also, in every case in which the tissue sections showed a positive score for BPH glands, the adenocarcinomatous glands exhibited a lower score. Thus, all the criteria consistently reveal that the immuno-reactive ZIP1 is always reduced and mostly non-detectable in the malignant glandular epithelium. Another important observation is the absence of a correlation between the stage of prostate cancer and the down regulation of ZIP1. This reveals that the down regulation occurs early in the malignant process and persists throughout its progression in the primary site; which is consistent with the early changes in zinc levels.
Figure 3 Comparative results of ZIP1 immuno-positive glands of tissue sections from subjects described in Table 1. A. Summary of glands that exhibited a positive Zip1 reactivity. The number of cases is shown in each bar. B. The number of cases in which the glandular epithelium contained cells that exhibited a ZIP1 score >+ (more than 10% of the cells comprising the acini). The differences in A and B between BPH glands and adenocarcinomatous glands are significant, P < 0.01.
As would be expected, the presence of normal peripheral zone glands in the malignant tissue sections is minimal, and insufficient for statistical analysis. However in three of the 4 cases, the normal glands exhibited the expected higher ZIP1 expression than the adenocarcinomatous glands, and gave results that were similar to BPH; both of which are zinc accumulating glands. In one case the normal gland was negative for ZIP1, which, seemingly, is an anomaly. However an important point needs to be considered. It is consistent with existing evidence (discussed below) that these metabolic changes occur before the appearance of histopathological evidence of malignant cells. Therefore, this "anomaly" might be due to changes that occur in a "premalignant" neoplastic condition that was histologically identified as "normal". Furthermore, in all five cases with PIN, the glands were either negative or + (none was ++ or +++), which mimics the profile of the adenocarcinomatous glands. It is striking that the combined PIN and adenocarcinoma glands showed no instance of ZIP1 positive cells that exceeded 10%. This could be supportive of a malignant relationship between PIN and adenocarcinoma; but further studies with additional PIN and normal peripheral zone glands are needed. Nevertheless, the Roswell Park study clearly establishes a consistent down-regulation of hZIP1 transporter in malignant prostate glands that corroborates and extends the results of the U.Maryland study, and is further corroborated by the following ClaflinU study.
In a parallel study (unpublished information, to be presented in a separate report), the tissue sections were also assayed for the immunohistochemical identification of m-aconitase. m-Aconitase was prevalent and unchanged in BPH, malignant, PIN and normal glands. Thus the down regulation of ZIP1 is specific. Moreover, the differences in citrate levels in malignant versus non malignant glands is not due to altered levels of m-aconitase. This re-emphasizes the role of altered zinc and ZIP1 in the metabolic transformation associated with prostate malignancy.
The ClaflinU Study (AK-D, OB)
In this study, ZIP1 mRNA expression (RT-in situ-PCR) and the relative level of zinc content were determined in the normal peripheral zone glands versus malignant glands from 38 prostate resections. The typical results represented in Figure 4 were consistently observed in all 38 prostate resections. The results show that hZIP1gene expression is evident uniformly in the epithelium of the normal peripheral zone glands; and is absent in the stroma. hZIP1 expression is markedly down regulated to the extent of not being demonstrable in the adenocarcinomatous glands; and presents the same appearance as the surrounding stroma. Of significance is the apparent down regulation of ZIP1 in early-stage as well as in advanced-stage malignant glands; which is consistent with the decrease in ZIP1 transporter protein shown in the Roswell Park study.
Figure 4 In situ detection of ZIP1 mRNA and zinc levels in normal and malignant glands. Panel A. Representative ZIP1 mRNA in Prostate Sections . Sections (inserts 1,2) from two prostate cancer subjects are shown with low magnification. Blue arrows point to acini with normal glandular epithelium that exhibit ZIP1 mRNA. White arrows point to adenocarcinomatous glands in which ZIP1 expression is not demonstrable. Insert 3 is a higher magnification of a section from a cancer patient to show more detail. Blue arrows point to acini with normal glandular epithelium. Red arrows point to malignant glands. Green arrows point to stromal (fibromuscular) tissue. The malignant epithelial cells exhibit a complete absence of detectable ZIP1 mRNA in the glandular epithelium. The normal glandular epithelium exhibits ZIP1 expression; and no ZIP1 expression in the stroma. Normal acini marked 'a' show uniform ZIP1 mRNA expression in the glandular epithelium. Advanced adenocarcinomatous glands marked as 'b" show uniform absence of ZIP1 mRNA. Developing early stage adenocarcinomatous glands marked 'c' show a progression of normal ZIP1 expressing cells and malignant cells that lost the expression of ZIP1. Panel B. Representative Zinc Levels in Prostate Sections. High zinc is represented by Newport Green yellow stain and low zinc is represented by TSQ red stain. The malignant region of the peripheral zone shows a significant depletion of zinc in the malignant glandular epithelium as exhibited by the red staining (white arrows). The depletion of zinc is evident in early differentiated malignant glands as represented by combinations of red and yellow staining in the glandular epithelial cells. As malignancy advances to the undifferentiated stage, the zinc is further depleted as represented by the dominant red stain and no yellow stain in the glandular epithelium of the adenocarcinomatous glands. The depletion of zinc in the malignant glandular region results in the surrounding stroma showing a higher zinc level (green stain) than the glandular epithelium. In contrast, the normal peripheral zone glands exhibit high zinc levels as represented by the uniform yellow stain and absence of red stain in the glandular epithelium. The stroma surrounding the glands exhibits a lower zinc level as shown by the red stain.
Correspondingly, Figure 4 shows the high level of cellular zinc that characterizes the normal glandular epithelial cells (green color). In contrast, the stroma exhibits a low level of zinc. Therefore, the in situ zinc staining provides the expected differential in zinc between normal glandular epithelium and stroma. The marked reduction of cellular zinc in the epithelium of the adenocarcinomatous glands is apparent. Like the expression of ZIP1, the loss of zinc occurs early in malignancy. Due to the depletion of zinc in the malignant glands, the stromal zinc level gives the appearance of relatively higher zinc levels. Many studies have observed that zinc levels are greatly decreased in extracts of resected malignant tissue preparations. However, the ClaflinU study provides the first in situ detection of the depleted cellular zinc levels in adenocarcinomatous glands as compared to the high zinc levels in normal glandular epithelium. An important revelation is that the decrease in zinc level in the malignant glands is due to a decrease in the cellular accumulation of zinc. This establishes that the decrease in intracellular zinc, and not impaired secretion of zinc into the lumen (prostatic fluid), is principally responsible for the decrease in malignant tissue zinc level. Thus, the results of the ClaflinU study are consistent with and corroborate the Roswell Park study and the preliminary results of the UMaryland study.
Discussion
The Zinc-Citrate Connection in Normal and Malignant Prostate
The results of the present study coupled with the numerous and consistent reports of others [[4-8,12-14] for reviews], provide direct overwhelming clinical and experimental evidence that, in prostate cancer, the lost ability of the malignant cells to accumulate zinc is a consistent event in the development of malignancy. However, the significance and further corroboration of this relationship requires the understanding and recognition of the unique role of zinc in normal prostate function and in prostate cancer. The major function of the human prostate gland peripheral zone (as in other animals) is the production and secretion of enormously high levels of citrate; which we refer to as "net citrate production". This capability of the normal secretory epithelial cells is the result of their unique ability to accumulate high levels of zinc; which inhibit m-aconitase activity and citrate oxidation [20,21]. Thus, one must recognize that the production and accumulation of citrate is dependent upon and is preceded by the accumulation of zinc in the glandular epithelial cells. Therefore, changes in the level of citrate in the peripheral zone are the result of and indicative of changes in zinc levels. The recent development of in situ magnetic resonance spectroscopy of prostate citrate levels in normal peripheral zone and malignant loci conclusively establishes that citrate levels are always greatly reduced in malignancy [see reviews [22-24]]. The consistency of this citrate relationship now makes magnetic resonance spectroscopy imaging (MRSI) of the prostate gland the most effective and reliable procedure for the identification, localization and volume estimation of malignant loci in the peripheral zone. Data collected from virtually all the existing MRSI reported studies reveal that there exists no case in which the malignant loci retain the high citrate levels of the normal peripheral zone glands as represented in figure 5[13,22]. These citrate changes revealed by magnetic resonance spectroscopy provide indirect evidence of corresponding changes in the accumulation of zinc, which is the cause of the changes in citrate. This is further verified by the comparative changes in zinc shown in figure 5. The profile of direct measurements of zinc changes associated with malignant prostate tissue strikingly replicates the citrate profile. This is evident despite the fact that these are different studies with different subjects and different stages of cancer. These relationships provide compelling evidence that, in prostate cancer, the malignant cells lose the ability to accumulate high zinc levels; and malignant cells that retain the accumulation of high zinc levels virtually never exist.
Figure 5 Composite of zinc and citrate levels in prostate. The zinc data are taken from Zaichick et al [9] and show the range of zinc levels in resected prostate tissue samples from different subjects. The citrate data are taken from Kurhanewicz et al [41] and show the range of citrate levels as determined by in situ magnetic resonance spectroscopy imaging of the prostate gland of different subjects. The actual zinc and citrate concentrations for normal were set to 100 and the values for BPH and PCa were adjusted accordingly. Note the parallelisms in that zinc and citrate levels are consistently significantly low in malignancy; and that no case exists in which the malignant loci retain the high zinc or high citrate levels that characterize normal or hypertrophic glands. The values above each bar are the number of subjects.
The Concept of the Role of hZIP1 and Zinc in Prostate Cancer
The existence of the zinc and citrate relationships in normal prostate and prostate cancer is irrefutable. How these relationships are involved as factors in the development and progression of prostate malignancy is important to understanding the pathogenesis of prostate cancer. It is well documented that all tumor cells undergo metabolic transformations that are essential for their malignant existence (25, 26 for reviews). It is important to emphasize that these metabolic transformations are not the cause of malignancy. Malignancy requires the genetic transformation of a sane cell to a neoplastic cell that is endowed with the potential capability of malignancy. The metabolic transformation is essential for the neoplastic cells to manifest their malignant capabilities.
The accumulation of zinc in normal prostate glandular epithelial cells results in two important effects; a metabolic effect and a proliferative effect. Its metabolic effect is the inhibition of citrate oxidation that is essential for the prostate function of production and secretion of high levels of citrate [20,21]; and its inhibition of terminal oxidation [27]. This has a bioenergetic cost in that the inhibition of citrate oxidation results in a ~60% loss of ATP production that would arise from complete glucose oxidation. Consequently, zinc-accumulating citrate-producing cells (normal peripheral zone epithelial cells) are energy-inefficient cells. A second effect of zinc is its inhibition of prostate cell proliferation. This effect results from zinc induction of apoptosis in prostate cells [28-31]. These are the consequences imposed upon highly specialized zinc-accumulating citrate-producing cells (i.e. normal peripheral zone secretory epithelial cells) in order to achieve their unique function of net citrate production.
Malignant prostate cells do not exist for the specialized function of citrate production and secretion. They must replace the metabolic pathways associated with net citrate production with metabolic relationships that are suitable for their malignant existence. That the malignant prostate cells in situ never exist as zinc-accumulating, citrate-producing cells is evidence of the incompatibility of the high zinc accumulation and net citrate production for their existence. Their metabolic transformation to energy-efficient citrate-oxidizing cells that have lost the ability to accumulate zinc provides their metabolic/bioenergetic requirements of malignancy. Also, the apoptotic influence of zinc is eliminated, which permits the proliferation of the malignant cells. However, the evidence presented herein clearly establishes hZIP1 down regulation in the primary in situ site and further suggests that this is the explanation for the consistently observed decrease in zinc levels in prostate cancer.
This concept is represented in figure 6. The occurrence of this metabolic transformation is dependent upon the ability of the normal epithelial cells and the inability of the malignant cells to accumulate zinc. The present studies establish that hZIP1 is down-regulated in the adenocarcinomatous glands. This is consistent with the down-regulation of hZIP1 gene expression in the African-American male population, which exhibits a higher incidence of prostate cancer [18]. The functional importance of hZIP1 in the accumulation of zinc in prostate cells has been established [15-17]. Over-expression of hZIP1 results in increased accumulation of zinc which leads to inhibition of cell proliferation; whereas cells with down-regulation of hZIP1 have decreased cellular zinc levels and increased proliferation. Also the accumulation of zinc in the malignant prostate cells in culture and in vivo [31] results in increased citrate levels.
Figure 6 The integrated role of ZIP1, zinc, and citrate metabolism in the pathogenesis of prostate malignancy. The normal glandular epithelial cell expresses ZIP1 that permits zinc accumulation, which inhibits citrate oxidation and terminal respiration. Citrate accumulates and coupled ATP production is reduced. A genetic transformation results in a neoplastic cell with potential malignant capability. ZIP1 expression is silenced by epigenetic factors which eliminate Zip1 transporter and accumulation of zinc in the premalignant cell. The level of cellular zinc decreases which removes the inhibitory effects on citrate oxidation and terminal oxidation. The Krebs cycle is functional and coupled ATP production is increased. The malignant cell is metabolically and bioenergetically capable of manifesting its malignant potential. Additionally, the growth inhibitory effect of zinc is removed, which allows growth and progression of the malignant cell.
Consequently, consistent clinical and experimental evidence strongly implicate the down-regulation of hZIP1 in the lost ability of the malignant cells to accumulate zinc. The existence of hZIP1 insures that prostate cells will accumulate zinc. If ZIP1 is not down regulated in the neoplastic cell, zinc accumulation and its metabolic/energetic and apoptotic effects will prevail; and the neoplastic cell will remain in a pre-malignant dormant state and/or will die. In this concept (figure 6), prostate malignancy requires two essential transformations; the genetic transformation to a neoplastic cell with potential malignant capability; and the metabolic transformation to an energy-efficient citrate-producing cell that has lost the ability to accumulate zinc. These relationships provide a plausible explanation and expectation for the apparent absence of the identification of malignant prostate glands that exhibit high zinc and high citrate levels.
The present studies raise two related important issues that we will be investigating: what is the cause of the down regulation of ZIP1? ; do the ZIP1 and zinc changes persist in the metastatic cells in situ? No information currently exists regarding the latter issue. The fact that hZIP1 is expressed in prostate cancer cell lines (that were established from metastatic lesions) suggests that down regulation of hZIP1 is a reversible phenomenon that occurs in the primary site in situ. This is suggestive of an epigenetic effect imposed by the interaction of the neoplastic cells and their in situ environment. In this case, the in vitro conditions of the cultured cells would eliminate the in situ factor(s) associated with the suppression of hZIP1 expression; thus permitting its re-expression. Moreover, the re-expression in the culture cells results in functional hZIP1 that manifests zinc uptake ; so that a fatal mutation is not involved. It is notable that SLC5A8, a gene that encodes a monocarboxylic acid transporter protein, has been reported to be a tumor suppressor gene in colon cancer [32-35] and other cancers [36,37]. The silencing of that gene occurs by hypermethylation and is a common and early event in human colon cancer. Similarly, it is plausible to propose that hZIP1 is a candidate tumor suppressor gene in prostate cancer. It will be important to determine the in situ conditions and mechanism that initiates the silencing of ZIP1 gene expression; which will then provide an understanding of the etiology of prostate malignancy.
The focus of this report on ZIP1 is not to imply that other zinc transport processes might not be involved in the altered accumulation of zinc. Rishi et al [18] demonstrated that ZIP1 and also ZIP2 are expressed in human prostate glandular epithelium. An increase in export of zinc could also decrease zinc accumulation by "true" malignant cells. However no information currently exists concerning the functional role of zinc exporters in prostate cells. Beck et al [38] reported that ZnT-4 was decreased in peripheral zone malignant tissue when compared to normal peripheral zone tissue samples. ZnT-4 is associated with the sequestering of cytosolic zinc into organelles, and is not involved as a plasma membrane zinc exporter. Moreover, a decrease in ZnT-4 would not be associated with a decrease in cellular zinc level, even as a secretory process. They also reported that ZnT-1 expression was unchanged in malignant versus normal peripheral zone. ZnT-1 does function as a plasma membrane-associated zinc exporter in some cells and possibly in prostate cells. Hasumi et al [39] reported that ZnT-1 expression was significantly lower in malignant prostate tissue samples when compared to BPH samples, which led them to conclude that ZnT-1 was not likely to be associated with the decreased zinc accumulation in the malignant cells. Consequently, a possible role of altered expression of zinc exporters in the genetic/metabolic transformation of the malignant cells in situ is not evident, but more research is required regarding this issue. We are now investigating the possible involvement of other zinc transporters in prostate malignancy.
Conclusion
The present studies, conducted independently in three institutions, collectively establish the presence of hZIP1 gene expression, the presence of membrane-associated hZIP1 transporter protein, and the accumulation of cellular zinc in the normal peripheral zone glandular epithelium and in benign hyperplastic glandular epithelium. The studies reveal that hZIP1 gene expression is down-regulated and hZIP1 transporter protein is depleted in adenocarcinomatous glands in prostate cancer. Correspondingly, the cellular level of zinc is also depleted. These effects occur in early and late stages of malignant development of the peripheral zone. hZIP1 expression is evident in the malignant prostate cell lines in culture. This leads to the likelihood that the lost expression in the adenocarcinomatous glands is due to an epigenetic silencing of hZIP1 that occurs in the in situ environment of the peripheral zone. When coupled with the voluminous clinical and experimental evidence, it becomes irrefutable that the development of malignancy in prostate cancer involves an essential metabolic transformation that results in the lost ability of malignant cells to accumulate zinc. Conversely, as long as the capability of high zinc accumulation exists, the neoplastic cells cannot manifest their malignant potential. Consequently, the expression of hZIP1 that sustains zinc accumulation in prostate cells will prevent the malignant activities and proliferation of the neoplastic cells. This provides a compelling basis for the proposal that hZIP1 down regulation is necessary for tumor progression and could be a tumor suppressor gene in prostate cancer. Consideration of all the clinical and experimental evidence leads to the concept that zinc and citrate-related metabolism play an important role in the pathogenesis and progression of prostate malignancy.
Methods
1. U.Maryland Study
Immunohistochemistry of Human Tissue Sections
Paraffin mounted serial sections of human prostate tissue was used for hZIP1 immunohistochemistry staining. Hematoxylin and eosin staining was used for identification of normal and adenocarcinomatous glands. For immunohistochemistry, slides were dewaxed by incubation in xylene and then rehydrated. Non-specific binding of antibody was blocked by incubation in BlokHen (Aves Labs, Inc.) solution. The slides were washed with PBS, incubated in hZIP1 antibody solution, washed again, and incubated with fluorescein-labeled secondary antibody solution; and then washed and mounted with anti-fade fluorescent medium (Molecular Probes). For control staining, adjacent serial sections were stained as described above except that the antibody-depleted and preimmune preparation were used instead of antihZIP1 antibody
Immunocytochemistry of Prostate Cells
PC-3 and LNCaP cells were plated on cover slips. The cover slips were washed with PBS, and the cells fixed in paraformaldehyde solution. The cells were permeabilized by incubation in 0.2% NP-40 solution, washed in PBS, and stained by the procedure described above for immunohistochemistry.
RT-PCR of Human Tissue mRNA
hZIP and GAPDH cDNA were synthesized from total mRNA isolated from human prostate tissue using 1.0 ug of total RNA, reverse transcriptase and random primers (TaqMan7 reagents, Perkin Elmer). hZIP1 and GAPDH fragments were amplified from the cDNA using 1.0 μM forward and reverse primers and 35 cycles. These conditions were shown to be in the quantitative detection range based on the concentration of template DNA. The cloned cDNA for hZIP1 was used as the template DNA in control reactions to determine the specificity of the PCR reactions. The RT-PCR products were analyzed by agarose gel electrophoresis with ethidium bromide staining and photographed under UV light. No products were detected without reverse transcriptase. The primers for hZIP1 were 5'-TCAGAGCCTCCAGTGCCTGT-3' and 5'-GCAGCAGGTCCAGGAGACAA-3'
2. The Roswell Park Study
Immunohistochemistry of Human Tissue Sections
Embedded prostatic adenocarcinoma slides that contained both benign prostatic hyperplasia (BPH) and adenocarcinomatous foci were obtained from Roswell Park Cancer Institute. Normal glands and intra epithelial neoplastic foci (PIN) were seen in a few cases. Immunohistochemistry with anti-hZIP1 antibody was performed by standard protocol [40]. Briefly, the slides were deparaffinized. Antigen retrieval was done by heating in 10 mM sodium citrate buffer (pH 6.0) at 98°C, incubated in 1% hydrogen peroxide (H2O2), blocked with 5% BlokHen with avidin D, incubated with ZIP1 antibody in 5% BlokHen with biotin (Vector Laboratories) at 4°C over night followed by incubation with Horseradish peroxidase-labeled goat anti chicken IgY secondary antibody in a dilution of 1:200 (AvesLabs, Tigard, Oregon). Color was developed by incubating slides with DAB kit (Vector Laboratories) followed by Hematoxylin counterstaining. Sections were examined with light E600 Nikon microscope. Pictures were taken with Spot advanced soft ware (version. 4.0.1). The appearance of membrane-associated hZIP1 immuno-positivity of the glandular epithelial cells was used for scoring as previously described [40]. The scores employed were; negative, no positive cells; + <10% positive cells; ++ 10–50% positive cells; +++ > 50% positive cells. The mean scores between groups were analyzed by the Student's t-Test.
3. The ClaflinU. Study
RT-in situ-PCR of Human Tissue Sections
Fresh frozen sections from 38 post-prostatectomy of men with clinical histories of prostate cancer were processed for zinc content analyses and RT-in situ-PCR. RT-in situ-PCR of the frozen sections was performed as described in detail by Rishi et al [18]. To preserve the intensity of the hybridized probes, the tissues were not counter-stained. Parallel hematoxylin and eosin-stained slides were used to identify various histologic cell types in the tissue sections. Microscopic examination usually reveals cytoplasmic staining for mRNA versus nuclear staining for DNA. Cell enumeration was performed on coded slides by at least two pathologists.
Determination of Intracellular Zinc Content
The relative intracellular zinc content in situ was determined by utilizing fresh frozen tissues. For this purpose the cells must be biochemically active. The relative concentrations of zinc in various cell types of the prostatic tissues were determined according to the manufacturer's instructions (Molecular Probes, Inc., Eugene, Oregon, USA). Briefly, the frozen tissues were incubated with equal molar concentrations of two zinc-indicator dyes; Newport Green (NPG), and TSQ. The frozen tissues were incubated in 20 ul/section of the zinc indicator cocktail over night and washed in PBS, gently, without disturbing the tissues. The slides were heat-fixed for 10 sec at 104°C to immobilize the signals. These slides were mounted with solution containing 50% glycerol in PBS and observed under a fluorescent microscope. TSQ has a high affinity for zinc (Kd~10 nM) and a detection limit of ~0.1 nM. The ZN-TSQ positive cells stain red. NPG has moderate zinc-binding affinity (Kd ~1 μM). The ZN-NPG positive cells appear yellowish green. Together, TSQ and NPG provide a relative difference in zinc concentrations in various cell types of the prostate. TSQ has about 2–3-log higher affinity for zinc than NPG, but has detection limit of about 3-log lower than NPG. Therefore, the cells that contain very low concentrations of intracellular zinc appear red and the ones with higher concentrations appear green. The cells with no detectable Zn2+ will appear black or dark blue.
Authors' contributions
Umaryland Study: RBF and LCC conceived and directed the study, wrote the Umaryland studies, wrote the final manuscript. BM performed ZIP immunohistochemical study. PF provided malignant cells and performed Western blots. Roswell Park Study: KS directed the study. MD obtained and conducted analyses of prostate cancer slides. KS and MD wrote the Roswell Park studies. ClaflinU Study: AK-D provided human tissue samples, performed histopathology, made the diagnosis and cataloged the tissues. OB performed the in situ RT-PCR on slides, developed the in situ zinc method, wrote the ClaflinU studies
Acknowledgements
The UMaryland study was supported by NIH grants CA 79903 and CA 71207 (RBF and LCC). The Roswell Park study was supported by grants from the National Institutes of Health RO1-097714 and Elsa Pardee Foundation (KKS). The ClaflinU study was supported by DOD grant DAMD 17-02-1-0233 (OB).
==== Refs
Ostrander EA Stanford JL Genetics of prostate cancer: Too many loci; too few genes Am J Hum Genet 2000 67 1367 1375 11067781 10.1086/316916
Porkka KP Visakorpi T Molecular mechanisms of prostate cancer European Urol 2004 45 683 691 10.1016/j.eururo.2004.01.012
Costello LC Franklin RB The novel role of zinc in the intermediary metabolism of prostate epithelial cells and its implications in prostate malignancy Prostate 1998 38 285 296 9609552 10.1002/(SICI)1097-0045(19980601)35:4<285::AID-PROS8>3.0.CO;2-F
Dhar NK Goel TC Dube PC Chowdhury AR Kar AB Distribution and concentration of zinc in the subcellular fractions of benign hyperplastic and malignant neoplastic human prostate Exp Mol Pathol 1973 19 139 142 4127820 10.1016/0014-4800(73)90073-7
Gonic P Oberleas D Knechtges T Prasad AS Atomic absorption determination of zinc in the prostate Invest Urol 1969 6 345 347 5773519
Gyorkey F Min K-W Huff JA Gyorkey P Zinc and magnesium in human prostate gland: Normal, hyperplastic, and neoplastic Cancer Res 1967 27 1349 1353
Ogunlewe JO Osegbe DN Zinc and cadmium concentrations in indigenous blacks with normal, hypertrophic, and malignant prostate Cancer 1989 63 1388 1392 2465818
Feustal A Wennrich R Steiniger D Klauss P Zinc and cadmium concentration in prostatic carcinoma of different histological grading in comparison to normal prostate tissue and adenofibromyomatosis (BPH) Urol Res 1982 10 301 303 6186062
Zaichick VY Sviridova TV Zaichick SV Zinc in the human prostate gland: normal, hyperplastic, cancerous Int Urol Nephr 1997 29 565 574
Vartsky D Shilstein S Bercovich A Huszar M Breskin A Chechik R Korotinsky S Malnick SD Moriel E Prostatic zinc and prostate specific antigen: an experimental evaluation of their combined diagnostic value J Urol 2003 170 2258 62 14634392 10.1097/01.ju.0000095795.86327.b8
Habib FK Mason MK Smith PH Stitch SR Cancer of the prostate: Early diagnosis by zinc and hormone analysis Br J Cancer 1979 39 700 704 87214
Costello LC Franklin RB The intermediary metabolism of the prostate: A key to understanding the pathogenesis and progression of prostate malignancy Oncology 2000 59 269 282 11096338 10.1159/000012183
Costello LC Feng P Milon B Tan M Franklin RB The Role of Zinc in the Pathogenesis and Treatment of Prostate Cancer: Critical Issues to Resolve Prostate Canc Prostate Dis 2004 7 111 117 10.1038/sj.pcan.4500712
Franklin RB Milon B Feng P Costello LC Zinc and zinc transporters in normal prostate function and the pathogenesis of prostate cancer Frontiers in Biosciences 2005 10 2230 2239
Costello LC Liu Y Zou J Franklin RB Evidence for a zinc uptake transporter in human prostate cancer cells which is regulated by prolactin and testosterone J Biol Chem 1999 274 17499 17504 10364181 10.1074/jbc.274.25.17499
Franklin RB Ma J Zou JB Kukoyi B Feng P Costello LC hZIP1 is a major zinc uptake transporter for the accumulation of zinc in prostate cells J Inorg Biochem 2003 96 435 442 12888280 10.1016/S0162-0134(03)00249-6
Gaither AL Eide DJ Functional expression of the human hZip2 zinc transporter J Biol Chem 2000 275 5560 5564 10681536 10.1074/jbc.275.8.5560
Rishi I Baidouri H Abbasi JA Kajdacsy-Balla A Pestaner JP Skacel M Tubbs R Bagasra O Prostate cancer in African-American men is associated with down-regulation of zinc transporters App Immunohistochem Mol Morph 2003 11 253 260 10.1097/00129039-200309000-00009
Mostofi FK (Fathollah Keshvar) Sesterhenn I Davis CJ Sobin LH Histological typing of prostate tumours 2002 2 Berlin; New York : Springer 115
Costello LC Liu Y Franklin RB Kennedy MC Zinc inhibition of mitochondrial aconitase and its importance in citrate metabolism of prostate epithelial cells J Biol Chem 1997 272 28875 28881 9360955 10.1074/jbc.272.46.28875
Costello LC Franklin RB Liu Y Kennedy MC Zinc causes a shift toward citrate at equilibrium of the m-aconitase reaction of prostate mitochondria J Inorg Biochem 2000 78 161 5 10766339 10.1016/S0162-0134(99)00215-9
Costello LC Franklin RB Kurhanewicz J The metabolic diagnosis of prostate by magnetic resonance spectroscopy Encyclopedia of Cancer 2002 3 2 Elsevier Science 167 177
Costello LC Franklin RB Narayan P Citrate in the diagnosis of prostate cancer Prostate 1999 38 237 245 10068348 10.1002/(SICI)1097-0045(19990215)38:3<237::AID-PROS8>3.0.CO;2-O
Kurhanewicz J Swanson MG Nelson SJ Vigneron DB Combined magnetic resonance imaging and spectroscopic imaging approach to molecular imaging of prostate cancer J Magnetic Reson Imag 2002 16 451 463 10.1002/jmri.10172
Costello LC Feng P Franklin RB Mitochondrial Function, Zinc, and Intermediary Metabolism Relationships in Normal Prostate and Prostate Cancer Mitochondrion 2005 5 143 153 16050980 10.1016/j.mito.2005.02.001
Costello LC Franklin RB "Why Do Tumor Cells Glycolyze?": From Glycolysis Through Citrate To Lipogenesis Mol Cell Biochem 2005
Costello LC Guan Z Kukoyi B Feng P Franklin RB Terminal oxidation and the effects of zinc in prostate versus liver mitochondria Mitochond 2004 4 331 338 10.1016/j.mito.2004.07.031
Liang JY Liu YY Zou J Franklin RB Costello LC Feng P Inhibitory effect of zinc on human prostatic carcinoma cell growth Prostate 1999 40 200 207 10398282 10.1002/(SICI)1097-0045(19990801)40:3<200::AID-PROS8>3.0.CO;2-3
Feng P Liang JY Li TL Guan ZX Zou J Franklin RB Costello LC Zinc induces mitochondria apoptogenesis in prostate cells Mol Urol 2000 4 31 35 10851304
Feng P Li TL Guan ZX Franklin RB Costello LC Direct effect of zinc on mitochondrial apoptogenesis in prostate cells Prostate 2002 52 311 318 12210492 10.1002/pros.10128
Feng P Li TL Guan ZX Franklin RB Costello LC Effect of zinc on prostatic tumorigenicity in nude mice Ann N Y Acad Sci 2003 1010 316 320 15033742 10.1196/annals.1299.056
Li H Myeroff L Smiraglia D Romero MF Pretlow TP Kasturi L Lutterbaugh J Rerko RM Casey G Issa JP Willis J Willson JK Plass C Markowitz SD SLC5A8, a sodium transporter, is a tumor suppressor gene silenced by methylation in human colon aberrant crypt foci and cancers Pro of the National Acad of Sci (USA) 2003 100 8412 8417 10.1073/pnas.1430846100
Coady MJ Chang M-H Charron FM Consuelo Plata Bernadette Wallendorff Jerome Sah Frank Sanford MarkowitzD Michael RomeroF Jean-Yves Lapointe The human tumour suppressor gene SLC5A8 expresses a Na+-monocarboxylate cotransporter J Physiol (Lond) 2004 557 719 731 15090606 10.1113/jphysiol.2004.063859
Ganapathy V Gopal E Miyauchi S Prasad PD Biological functions of SLC5A8, a candidate tumour suppressor Biochemical Society Transactions 2005 33 237 240 15667316
Miyauchi S Gopal E Fei YJ Ganapathy V Functional identification of SLC5A8, a tumor suppressor down-regulated in colon cancer, as a Na(+)-coupled transporter for short-chain fatty acids Journal of Biological Chemistry 2004 279 13293 6 14966140 10.1074/jbc.C400059200
Ueno M Toyota M Akino K Suzuki H Kusano M Satoh A Mita H Sasaki Y Nojima M Yanagihara K Hinoda Y Tokino T Imai K Aberrant methylation and histone deacetylation associated with silencing of SLC5A8 in gastric cancer Tumour Biology 2004 253 134 40 15361710 10.1159/000079145
Porra V Ferraro-Peyret C Durand C Selmi-Ruby S Giroud H Berger-Dutrieux N Decaussin M Peix JL Bournaud C Orgiazzi J Borson-Chazot F Dante R Rousset B Silencing of the tumor suppressor gene SLC5A8 is associated with BRAF mutations in classical papillary thyroid carcinomas Journal of Clinical Endocrinology & Metabolism 2005 905 3028 35 15687339 10.1210/jc.2004-1394
Beck FW Prasad AS Butler CE Sakr WA Kucuk O Sarkar FH Differential expression of hZnT-4 in human prostate tissues Prostate 2004 58 374 381 14968438 10.1002/pros.10344
Hasumi M Suzuki K Matsui H Koike H Ito K Yamanaka H Regulation of metallothionein and zinc transporter expression in human prostate cancer cells and tissues Cancer Letters 2003 200 187 95 14568174 10.1016/S0304-3835(03)00441-5
Desouki MM Rowan BG SRC kinase and mitogen-activated protein kinases in the progression from normal to malignant endometrium Clin Cancer Res 2004 10 546 55 14760076
Kurhanewicz J Vigneron DB Hricak H Narayan P Carroll P Nelson SJ Three dimensional hydrogen-1 MR spectroscopic imaging of the in situ human prostate with (0.24-0.7-cm3) high spatial resolution Radiology 1996 198 795 805 8628874
|
16153295
|
PMC1243239
|
CC BY
|
2021-01-04 16:36:35
|
no
|
Mol Cancer. 2005 Sep 9; 4:32
|
utf-8
|
Mol Cancer
| 2,005 |
10.1186/1476-4598-4-32
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-1001614657410.1186/1465-9921-6-100ResearchUp-regulation of Toll-like receptors 2, 3 and 4 in allergic rhinitis Fransson Mattias [email protected] Mikael [email protected]ält Jonas [email protected] Lennart [email protected] Rolf [email protected] Lars-Olaf [email protected] Laboratory of Clinical and Experimental Allergy Research, Department of Oto-Rhino-Laryngology, Malmö University Hospital, Lund University, Malmö, Sweden2 Department of Experimental Medical Science, Lund University Hospital, Lund University, Sweden3 AstraZeneca R&D, Lund, Sweden2005 7 9 2005 6 1 100 100 29 4 2005 7 9 2005 Copyright © 2005 Fransson et al; licensee BioMed Central Ltd.2005Fransson et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Toll-like receptors enable the host to recognize a large number of pathogen-associated molecular patterns such as bacterial lipopolysaccharide, viral RNA, CpG-containing DNA and flagellin. Toll-like receptors have also been shown to play a pivotal role in both innate and adaptive immune responses. The role of Toll-like receptors as a primary part of our microbe defense system has been shown in several studies, but their possible function as mediators in allergy and asthma remains to be established. The present study was designed to examine the expression of Toll-like receptors 2, 3 and 4 in the nasal mucosa of patients with intermittent allergic rhinitis, focusing on changes induced by exposure to pollen.
Methods
27 healthy controls and 42 patients with seasonal allergic rhinitis volunteered for the study. Nasal biopsies were obtained before and during pollen season as well as before and after allergen challenge. The seasonal material was used for mRNA quantification of Toll-like receptors 2, 3 and 4 with real-time polymerase chain reaction, whereas specimens achieved in conjunction with allergen challenge were used for immunohistochemical localization and quantification of corresponding proteins.
Results
mRNA and protein representing Toll-like receptors 2, 3 and 4 could be demonstrated in all specimens. An increase in protein expression for all three receptors could be seen following allergen challenge, whereas a significant increase of mRNA only could be obtained for Toll-like receptor 3 during pollen season.
Conclusion
The up-regulation of Toll-like receptors 2, 3 and 4 in the nasal mucosa of patients with symptomatic allergic rhinitis supports the idea of a role for Toll-like receptors in allergic airway inflammation.
==== Body
Background
Toll-like receptors (TLRs) have recently emerged as key receptors of the innate immune system. They recognize specific pathogen-associated molecular patterns initiating a host defense response [1]. The upper airway encounters potential pathogens like bacteria and viruses in inspired air, and the discovery of TLRs on epithelial cells suggests that the epithelium has a role in the mucosal immune system [2-4]. Activation of TLR-dependent signaling pathways results in the expression of effector molecules, such as cytokines and chemokines, which contributes to the activation of the antigen-specific adaptive immune response [5]. The discovery that microbial components and endogenous ligands are recognized by TLRs may explain how generated signals can affect the initiation, maintenance and progression of inflammatory diseases [6]. Several authors have recently suggested a role for TLRs in the pathophysiology of allergic rhinitis and asthma, thus indicating the importance to study their expression during allergic airway inflammation [7,8].
Ten distinct TLRs have been described in humans, expressed in various combinations in cells of the immune system as well as in other cell types [9,10]. mRNA of all ten TLRs has been described in human nasal airway tissue, but protein verification using Western blot analysis or immunohistochemistry is still lacking for most TLRs in the nose [2,11]. In cultured airway epithelial cells, mRNA and corresponding proteins have been demonstrated for TLR2, TLR3, TLR4 and TLR5 [4,12-14]. TLR2 has also been demonstrated in the lower respiratory tract using immunohistochemistry and both TLR2 and TLR4 have been shown in alveolar epithelial cells with real-time PCR and flow cytometry [15,16]. No difference in the sinonasal mRNA expression of TLR2 and TLR4 could be demonstrated when patients with chronic sinusitis and/or nasal polyps were compared with healthy controls [3]. In another study, TLR4 immunoreactivity was detected in nasal mucosa from both children and adults irrespective of atopy status, revealing a higher expression level among children [17].
The present study was designed to quantify mRNA and protein expressions of TLR2, TLR3 and TLR4 in the nasal mucosa of patients with seasonal allergic rhinitis, before and after pollen exposure, and to compare these findings with data derived from healthy volunteers.
Methods
Skin prick test
Skin prick tests (SPTs) were performed with a standard panel of 10 common airborne allergens (ALK, Copenhagen, Denmark) including pollen (birch, timothy and artemisia), house dust mites (D. Pteronyssimus and D. Farinae), molds (Cladosporium and Alternaria) and animal allergens (cat, dog and horse). SPTs were performed on the volar side of the forearm with saline buffer as negative and histamine chloride (10 mg/ml) as positive control. The wheal reactions were measured after 20 min and designated as 4+, 3+, 2+, 1+ or 0 depending on the size in relation to histamine and saline [18].
Subjects
The study included 42 patients (22 women) with symptomatic birch and/or grass pollen induced intermittent allergic rhinitis and 27 healthy volunteers (13 women), serving as controls. The median (range) age of patients and controls was 36 (18–68) and 27 (16–50) years, respectively. All control patients were healthy and the same goes for the rhinitis patients, with the exception of their allergy. None of the participants were subjected to any other type of surgery than the nasal biopsy described in the research protocol.
The diagnosis of birch and grass pollen induced allergic rhinitis was based on a positive history of intermittent allergic rhinitis for at least 2 years and a positive SPT to birch and/or timothy pollen. All patients were classified as having moderate to severe symptoms (itchy nose and eyes, sneezing, nasal secretion and nasal blockage) during birch and/or grass pollen season and they had all been treated with antihistamines and nasal steroids, often in combination with antihistamine or cromoglycate eye drops, during pollen seasons previous years. Patients had no symptoms of asthma at the time of visit and they did not take any asthma medication (short or long acting β-agonists and inhaled steroids). All patients presented 3+ or 4+ reactions towards birch or timothy in SPT. 13 patients presented positive reactions towards both birch and timothy and 5 patients were in addition positive for mugworth. Patients presenting positive reactions towards animals (4 towards cat, 2 towards dog and 2 towards horse), did not have any regular animal contact. Exclusion criteria included a history of perennial symptoms, upper airway infection during the last 2 weeks before the time of visit and treatment with local or systemic corticosteroids during the last 2 months. Occasional use of antihistamines was accepted.
The controls were all symptom-free, had no history of allergic rhinitis and had negative SPTs to the standard panel of allergens described above They had no history of upper airway infection during the last 2 weeks before the time of visit and they were all free of medication. Before inclusion, all subjects, patients as well as controls, were evaluated by an ear-, nose- and throat-consultant performing nasoscopy. Individuals with signs or symptoms of chronic rhinitis, hypertrophy of turbinates, septum deviation, nasal polyposis or recurrent sinusitis were excluded. None of the participants had been subjected to any form of nasal or sinus surgery before inclusion in the study. The study was approved by the Ethics Committee of the Medical Faculty, Lund University.
Study design
Nasal biopsies for mRNA analysis were obtained from 12 patients with symptomatic seasonal allergic rhinitis, during either birch pollen (5 patients) or grass pollen season (7 patients). They were included when they had experienced substantial symptoms of rhinoconjunctivitis (itchy nose and eyes, sneezing, nasal secretion and nasal blockage) during at least 3 consecutive days. The majority of the patients were seen within 5–10 days after the first appearance of symptoms. A local pollen count confirmed the presence of the relevant pollen in the air during this period. In addition, 19 patients were seen before the pollen season and 18 healthy controls were sampled either before or during the pollen season.
Nasal biopsies for immunohistochemistry were obtained from 11 patients at two separate occasions outside pollen season. The first biopsy was obtained during control conditions without any form of challenge. 2–4 weeks later the same patients were challenged intranasally with relevant pollen. 24 hours after this challenge a second biopsy was obtained. Only one biopsy was taken from each nostril. Patients were challenged with 10,000 SQ/U per nostril of Aquagen (ALK, Denmark) with either birch (3 patients) or grass pollen (8 patients). 9 controls were sampled during the same period.
Nasal biopsy procedure
Biopsies were taken from the inferior turbinate after topical application of local anesthesia containing lidocainhydrochloride/nafazoline (34 mg/mL/0.17 mg/mL) for 20 minutes. Biopsies were in total obtained from 69 different individuals. The nasal biopsies were approximately 2 × 2 × 2 mm large and too small to be divided without losing their morphology. As a consequence, material for mRNA characterization and immunohistochemistry could not be obtained from the same biopsy.
RNA extraction and reverse transcription
Nasal biopsies for mRNA extraction were immediately placed in RNA-later (QIAGEN) for 24 hours and then frozen. RNA was extracted from homogenized nasal biopsies using the RNeasy Mini Kit (QIAGEN GmbH) according to the supplier's protocol. All samples were treated with DNase (QIAGEN). Total RNA quantity and quality were assessed by a spectrophotometer and the wavelength absorption ratio (260/280 nm) was between 1.8 and 2.0 in all preparations. Reverse transcription to cDNA was carried out with Omniscript™ reverse transcriptase kit (QIAGEN GmbH) with oligo-dT primer in a final volume of 20 μl using the Mastercycler personal PCR machine (Eppendorf AG, Germany), at 37°C for 1 hour.
Quantitative real-time PCR
Quantitative real-time PCR assays were performed using the SmartCyclerII system (Cepheid, USA). Two different types of PCR assays were used, each with a different protocol according to the manufacturer's recommendations. For detection of TLR2 and β-actin, intron over-spanning oligonucleotide primers were designed (Table I) using Prime Express® 2.0 software (Applied Biosystems, USA) and synthesized by DNA Technology A/S (Aarhus, Denmark). PCR was performed using QuantiTect™SYBR® Green PCR kit (QIAGEN) in a final volume of 25 μl. Reactions were incubated at 95°C for 15 min, then incubated 46 cycles at 94°C for 30 s followed by 55°C for 60 s (initially 65°C, followed by 2°C decrease the first 6 cycles). Standard curves for TLR2 and β-actin were prepared using half 10log dilutions of PCR products generated from target cDNA. Specific PCR products were analyzed by running melting curves. Melting curve analysis for TLR2 and β-actin revealed a single peak in each sample.
Table 1 Sequences of primers used for PCR amplification of TLR2 and β-actin.
Human TLR2 Forward, 5'-TCACTGCTTTCAACTGGTAGTTGTG-3'
Reverse, 5'-TCCTTGGAGAGGCTGATGATG-3'
Human β-actin Forward, 5'-GCCAACCGCGAGAAGATG-3'
Reverse, 5'-ACGGCCAGAGGCGTACAG-3'
PCR: polymerase chain reaction
TLR: Toll-like receptor
For detection of TLR3, TLR4 and β-actin, primers were purchased from Applied Biosystems. These PCR assays contained a probe and were performed using Taq-Man®Universal PCR Master Mix NoAmpErase®UNG in a final volume of 25 μl. Reactions were incubated at 95°C for 10 min, then incubated 40 cycles at 95°C for 15 s followed by 60°C for 60 s. When using a probe as in these PCR assays, nonspecific amplification is not detected.
Gene expression changes were assessed using the comparative cycle threshold (CT) method . The relative amount of mRNA for TLR2, TLR3 and TLR4 was determined by subtracting the CT values achieved for these genes from the CT value of the housekeeping gene β-actin. The amount of mRNA is expressed in relation to 100,000 mRNA molecules of β-actin (100,000 × 2-ΔCT).
Immunohistochemical analysis of TLRs
Nasal biopsies used for immunohistochemistry were frozen in Tissue Tek® O.C.T mounting media (Histo Lab, Gothenburg, Sweden) immediately after excision. Cryosections, 8 μm thick, were after sectioning post-fixed with 2% buffered formaldehyde for 20 min, rinsed in phosphate buffered saline (PBS; pH 7.6; 3 × 5 min at room temperature (RT)) and placed in 0.1% saponin in PBS for 20 min at RT. Non-specific binding sites were blocked with 5% normal serum (Dako; dilution 1:10 in PBS) for 30 min. Avidin-binding sites were blocked with incubation of Avidin D solution (Vector Laboratories, Burlingame, CA, USA) for 15 min. Thereafter, the sections were rinsed in PBS (3 × 5 min) before blocking of biotin-binding sites with biotin blocking solution (Vector) for 15 min. After additional rinsing (PBS; 3 × 5 min) sections were incubated with the primary antibody overnight at 4°C (in control sections the primary antibody was omitted). The primary antibody was diluted in PBS supplemented with 0.25% Triton X and 0.25% bovine serum albumin. The primary antibodies were: anti-TLR2 (dilution 1:50), anti-TLR3 (dilution 1:100), anti-TLR4 (dilution 1:50). All primary antibodies were purchased from ImmunoKontact, AMS Biotechnology, Oxon, UK. After overnight incubation with primary antibody, the sections were rinsed (3 × 5 min in PBS) and incubated with biotinylated secondary antibody (horse anti-mouse IgG1, Vector, dilution 1:200 or goat anti-rabbit, dilution 1:200) for 45 min at RT. After additional rinsing (3 × 5 min in PBS), the sections were incubated with alkaline phosphatase-labeled Streptavidin (dilution 1:200 for 45 min), rinsed (3 × 5 min in PBS) and alkaline phosphate activity was developed for 6 min at RT using New Fuchsin (Dako) as enzyme substrate. Endogenous alkaline phosphatase activity was inhibited by Levamisol. All sections were counter-stained with Harris's hematoxylin, coated with Aqua Perm mounting medium (484975 Life Sci. International), dried overnight and mounted in DPX. Positive immunoreactivity was identified as bright red precipitate.
For quantification of TLR immunoreactivity, high resolution digital images were obtained from each biopsy, so that the entire mucosal area was captured. Occasional regions where the epithelial lining had been lost due to mechanical damage (such regions were present to an equal extent among the different groups) were excluded from the study. Using an image analysis system (Image-Pro Plus v4.51, Media Cybernetics, Silver Spring, USA) and a preset fixed colour threshold value, the number of positively (i.e. bright red) pixels was automatically calculated. The immunoreactivity of each TLR was expressed in relation to the area of mucosal tissue.
Statistics
Statistical analysis was performed using GraphPad Prism 4. All data are expressed as mean ± SEM, and n equals the number of subjects. For mRNA data, group comparisons were made between controls and patients outside pollen season as well as between patients outside and during pollen season. For immunohistochemical data, group comparisons were made between controls and patients before allergen challenge as well as between patients before and after allergen challenge. Mann-Whitney test was used to determine statistical differences for unpaired comparisons and Wilcoxon signed rank test was used for paired comparisons. P-values less than or equal to 0.05 were considered statistically significant.
Results
Real-time PCR analysis of total RNA extracted from nasal biopsies demonstrated the presence of TLR2, TLR3 and TLR4 as well as β-actin in all samples. The expression of TLR2 mRNA, in relation to 100,000 molecules of β-actin, was 78 ± 13 in controls (n = 17), 112 ± 13 in patients outside pollen season (n = 19) and 122 ± 27 in patients during pollen season (n = 12). The differences seen between the groups were not statistically significant (Figure 1A). The expression of TLR3 mRNA was 318 ± 67 in controls (n = 15), 279 ± 31 in patients outside pollen season (n = 19) and 473 ± 80 in patients during pollen season (n = 11). The increase seen during pollen season was statistically significant (p < 0.05) (Figure 1B). The expression of TLR4 mRNA was 44 ± 7 in controls (n = 14), 55 ± 8 in patients outside pollen season (n = 19) and 103 ± 28 in patients during pollen season (n = 11). The apparent seasonal increase did not reach statistical significance (Figure 1C).
Figure 1 mRNA expression of TLR2, TLR3 and TLR4 in nasal mucosa. Expression levels of TLR2 (A), TLR3 (B) and TLR4 (C) in biopsies of nasal mucosa from healthy volunteers (controls), from patients with intermittent allergic rhinitis outside pollen season and from patients with symptomatic intermittent allergic rhinitis during pollen season. Levels of mRNA were calculated in relation to 100,000 mRNA molecules of β-actin. Bold lines represent mean values. Increase in the mRNA expression of TLR3 in patients during pollen season compared to patients outside pollen season (*p < 0.05).
All patients reported an increase in nasal symptoms both 5 and 15 min after allergen challenge (data not shown). The most intensive immunoreactivity for all three TLRs was seen within the airway epithelium (Figure 2A–F), where the staining was foremost distributed to epithelial cells positioned in the apical region of the epithelium. The same distribution pattern was observed among healthy controls and this pattern was not changed by the allergen challenge. A particularly intense staining was observed in scattered areas covered by a low-height repair epithelium. Immunoreactivity for the different TLRs was also seen in a few scattered intraepithelial leukocytes. For all examined TLRs, immunoreactivity was also present in scattered subepithelial cells. However, although being distinct from controls, the staining was generally variably pale and pallid which precluded a proper quantification. The weak intensity also jeopardized identification of positive cell types by e.g. double labeling techniques. Hence, a more tentative identification was made based on morphological criteria. In this regard, mast cells were identified as large granulated and mononuclear cells, granulocytes by their characteristic polymorph nuclei, lymphocytes as small mononuclear cells with a circular nucleus surrounded by only a thin rim of cytoplasm. Using these morphological criteria, TLR2 immunoreactivity could be observed in mast cells, granulocytes as well as in few scattered large non-granulated mononuclear cells (Figure 3A). Among the subepithelial TLR3-positive cells, only mast cells (Figure 3B) and occasional granulocytes were observed whereas immunoreactivity for TLR4 was present in mast cells, lymphocytes (Figure 3C), granulocytes and large non-granulated mononuclear cells (Figure 3D).
Figure 2 Immunoreactivity of TLR2, TLR3 and TLR4 in nasal mucosa. Immunohistochemical localization of TLR2, TLR3 and TLR4 in biopsies of nasal mucosa. Representative pictures of control (A) and TLR-stained (B-F) sections. No immunoreactivity was observed in control sections when the primary antibody was omitted (A). In an adjacent section, immunoreactivity for TLR3 is seen in the apical part of the epithelial lining and in a few scattered intraepithelial leukocytes (B). Pictures C and D illustrate representative staining patterns of TLR2 and TLR4, respectively. Areas with a thin repairing epithelium display a particularly intense staining, TLR3 (E) and TLR4 (F). Scale bars: A, B and D = 85 μm, C = 30 μm, E and F= 50 μm.
Figure 3 Immunoreactivity of TLR2, TLR3 and TLR4 in subepithelial cells. Bright field micrographs exemplifying TLR-positive cells in the subepithelial tissue. Granulocyte containing TLR2-like immunoreactivity (A), TLR3-positive mast cell (B), TLR4 immunoreactivity in lymphocyte-like cells (C) as well as in a large mononuclear cell (D). Scale bars: A, B and D = 10 μm, C = 8 μm.
The total immunoreactivity for TLR2, in relation to the area of mucosal tissue, was 0.61 ± 0.21 in controls (n = 9), 0.24 ± 0.29 in patients before allergen challenge (n = 11) and 2.16 ± 0.81 after allergen challenge (n = 11). There was an increase in TLR2 immunoreactivity after allergen challenge (p < 0.05; Figure 4A). Immunoreactivity for TLR3 was 1.39 ± 0.44 in controls (n = 9), 0.64 ± 0.25 in patients before allergen challenge (n = 11) and 2.22 ± 0.79 in patients after allergen challenge (n = 11). There was an increase in TLR3 immunoreactivity after allergen challenge (p = 0.05; Figure 4B). Immunoreactivity for TLR4 was 0.93 ± 0.40 in controls (n = 9), 0.47 ± 0.32 in patients before allergen challenge (n = 11) and 2.34 ± 0.80 in patients after allergen challenge (n = 10). There was an increase in TLR4 immunoreactivity after allergen challenge (p < 0.05; Figure 4C). 2 or 3 patients out of the 11 had a heightened response to allergen (Figure 4A–C). These patients were challenged with grass and with the exception of one patient, these data points all emanated from different individuals.
Figure 4 Quantitative immunoreactivity of TLR2, TLR3 and TLR4 in nasal mucosa. Quantitative analysis of TLR2 (A), TLR3 (B) and TLR4 (C) immunostaining in nasal epithelium in biopsies from healthy volunteers (controls), from patients with intermittent allergic rhinitis before nasal allergen challenge and from the same patients after allergen challenge. TLR immunoreactivity is expressed in relation to the area of mucosal tissue. Lines connecting patients before and after allergen challenge represent paired data. Increase in the immunostaining for TLR2 (*p < 0.05), TLR3 (*p = 0.05) and TLR4 (*p < 0.05) after allergen challenge.
Discussion
Real-time PCR demonstrated mRNA expression of TLR2, TLR3 and TLR4 in biopsies from the inferior turbinate of all tested subjects. Immunohistochemistry confirmed the occurrence of the corresponding proteins in the airway epithelium. Differences were not seen in mRNA content or in protein expression, when patients with seasonal allergic rhinitis were examined before pollen season or before allergen challenge and compared with healthy controls. A significant increase of mRNA for TLR3 was obtained during pollen season. The protein expression for all three TLRs increased in the rhinitis patients following a challenge with relevant pollen, supporting the idea of a role for TLRs in allergic airway inflammation.
The expression of TLRs has been established in different cells of the immune system, including dendritic cells, eosinophils, monocytes, macrophages and neutrophils, illustrating the role of TLRs in modulating inflammatory responses [19-21]. However, reports of protein verified TLR expression in human airways including nasal tissue are still lacking for most of the TLRs. In the present study, immunoreactivity for TLR2, TLR3 and TLR4 could be demonstrated in the apical part of the nasal epithelium. This is in line with previous findings for TLR2, TLR3 and TLR4 in cultured epithelial cells [4,14,22,23]. The sites for a possible TLR expression are known to vary between different cell types, i.e. TLR3 is found exclusively in the intracellular compartments of dendritic cells [24], whereas it can be found both on the cell surface and in intracellular parts of human lung fibroblasts [25]. Since TLRs are part of the first line of defense responding to the presence of different types of microbes in the airways, it seems reasonable that these receptors are localized in the apical part of the epithelial cells, facing the lumen. Expression of TLR4 has previously been described in the nasal mucosa from both children and adults irrespective of atopy status, and during control conditions this expression is more marked among children [17]. Only adults participated in the present study and it is unlikely that the slight difference in age between the patients [median age 36 (18–68)] and controls [median age 27 (16–50)] affected the outcome.
The amount or patterns of inflammatory cells with a positive staining for TLR2, TLR3 and TLR4 in the submucosa was limited and did not increase as a result of allergen challenge. Unfortunately the pallid appearance of the immunoreactivity in subepithelial cells precluded a proper identification of the TLR-positive cells (a potential explanation to the different intensity observed among epithelial and subepithelial cells may simply be that it reflects a relative higher expression i.e. more receptor molecules per cell within the epithelium). However, a tentative identification of subepithelial cells was made based on classical morphological criteria. Based on these it is surmised that TLR2 and TLR4 have a rather widespread expression among the immune cells, being present in mast cells, granulocytes, lymphocytes, and large agranular mononucleated cells (likely to be macrophages or dendritic cells). In contrast, subepithelial TLR3-like immunoreactivity was restricted to mast cells and occasional granulocytes.
The mRNA expression of TLR3 was found to be elevated among patients with on-going symptomatic allergic rhinitis. This is well in line with the increase of the corresponding protein seen following allergen challenge. mRNA for TLR4 exhibited a similar tendency, however not reaching statistical significance, whereas no seasonal increase was seen for TLR2 mRNA. The discrepancy between TLR2 and TLR4 mRNA on one hand and their corresponding proteins on the other could be explained in several ways. First of all, we must acknowledge that TLR2, TLR3 and TLR4 appear to be localized mainly in the apical part of the epithelium. mRNA from nasal biopsies is comprised of a heterogeneous mix of different cells not only from the epithelium, but also from the submucosa. An increase in TLR mRNA expression in the epithelial cells might therefore be masked by a lack of increase in the submucosal cells. Another explanation might be related to variations in the mRNA expression between individuals. The mRNA data were derived from two separate groups of patients, one sampled before and the other during pollen season. The protein data were paired, each patient contributed with two biopsies, one before and the other after allergen challenge, thereby omitting variations between individuals. Finally, a single allergen challenge differs from the prolonged allergen exposure seen during pollen season. It might be that the increase in immunoreactivity for TLR2, TLR3 and TLR4 seen after allergen challenge only persists for a limited period of time. If so, the mRNA expression might, after an initial increase, have decreased towards normal a few days into the pollen season.
The increased expression of TLR2, TLR3 and TLR4 as a consequence of allergen exposure is in line with a previous report demonstrating an increase of TLR2 protein in chronic middle ear inflammation [26]. The lack of increase in TLR2 and TLR4 mRNA, discussed above, is in accordance with another study demonstrating no differences in the expression of TLR2 and TLR4 when mRNA from normal turbinates were compared with samples from patients with chronic sinusitis and/or nasal polyps [3]. Proteins were not investigated in the latter study. It is also worth noticing that the inflammation in chronic sinusitis and nasal polyps is different from the acute inflammation seen in the nasal epithelium during intermittent allergic rhinitis.
The role of TLRs as a primary part of our microbe defense system has been shown in several studies, but their possible function as mediators in allergy and asthma remains to be established. Thus, their role as detectors of pathogens has provided molecular mechanisms to underpin the observations leading to the so-called hygiene hypothesis. This theory proposes that a major source of microbial, Th1-like, immune provocations has been lost with the decreased incidence of many infectious diseases due to vaccinations, the use of antibiotics, and a higher hygiene standard [27]. The deficiency in Th1-like provocations leads to Th2-biased immune responses towards environmental allergens and consequently to an increase in allergic airway diseases [28,29]. In line with this hypothesis, it has been shown that children of farmers, known to have a decreased risk of developing allergies, expressed higher levels of TLR2 mRNA in blood compared to children of non-farmers. The expression of TLR4 was not altered in farmers' children, but the level of the co-factor CD14 was markedly increased, indicating a change in receptor mediated signaling activity [29]. However, once the allergic airway inflammation has been established, bacterial and viral infections may be as relevant as allergens in inducing hyperreactive responses [30-33].
Conclusion
An up-regulation of TLR2, TLR3 and TLR4 in the nasal mucosa of patients with symptomatic allergic rhinitis supports the idea of a role for TLRs in allergic inflammation. In several airway model systems, stimulation of TLRs results in changes in the production of effector molecules, such as cytokines and chemokines, thereby affecting and further upgrading the airway inflammation [4,14-17]. It is also well known that viral as well as bacterial infections can worsen the symptoms of allergic rhinitis and trigger exacerbations of asthma [6]. Thus, an increase in the amount of TLRs present in the apical part of the epithelium might induce a hyperreactive response to bacteria and viruses in patients with seasonal allergic rhinitis.
List of abbreviations used
PCR: polymerase chain reaction
RT: room temperature
SPT: skin prick test
TLR: Toll-like receptor
Competing interests
The study was financially supported by the Swedish Medical Research Council, the Swedish Heart Lung Foundation, the Swedish Association for Allergology and AstraZeneca, Sweden.
Lennart Jansson is employed by AstraZeneca R&D.
Authors' contributions
MF acquired and analyzed the mRNA data and drafted the manuscript. MA acquired and analyzed the mRNA data and revised the content of the manuscript. JE and RU acquired and analyzed the immunohistochemistry data and revised the content of the manuscript. LJ contributed to the study-design and revised the data of the manuscript. LOC conceived of the study, participated in its design and coordination, and helped to draft the manuscript
Acknowledgements
The authors would like to thank Ingegerd Larsson and Ann Reutherborg, for skilful technical assistance during the course of this study as well as Anna Karin Bastos and Josefine P Riikonen for logistic support.
==== Refs
Janeway CA JrMedzhitov R Innate immune recognition Annu Rev Immunol 2002 20 197 216 11861602 10.1146/annurev.immunol.20.083001.084359
Vandermeer J Sha Q Lane AP Schleimer RP Innate immunity of the sinonasal cavity: expression of messenger RNA for complement cascade components and toll-like receptors Arch Otolaryngol Head Neck Surg 2004 130 1374 1380 15611395 10.1001/archotol.130.12.1374
Claeys S de Belder T Holtappels G Gevaert P Verhasselt B van Cauwenberge P Bachert C Human beta-defensins and toll-like receptors in the upper airway Allergy 2003 58 748 753 12859553 10.1034/j.1398-9995.2003.00180.x
Sha Q Truong-Tran AQ Plitt JR Beck LA Schleimer RP Activation of airway epithelial cells by toll-like receptor agonists Am J Respir Cell Mol Biol 2004 31 358 364 15191912 10.1165/rcmb.2003-0388OC
Sabroe I Read RC Whyte MK Dockrell DH Vogel SN Dower SK Toll-like receptors in health and disease: complex questions remain J Immunol 2003 171 1630 1635 12902458
Zuany-Amorim C Hastewell J Walker C Toll-like receptors as potential therapeutic targets for multiple diseases Nat Rev Drug Discov 2002 1 797 807 12360257 10.1038/nrd914
O'Neill LA Therapeutic targeting of Toll-like receptors for inflammatory and infectious diseases Curr Opin Pharmacol 2003 3 396 403 12901949 10.1016/S1471-4892(03)00080-8
Heine H Lien E Toll-like receptors and their function in innate and adaptive immunity Int Arch Allergy Immunol 2003 130 180 192 12660422 10.1159/000069517
Sabroe I Parker LC Wilson AG Whyte MK Dower SK Toll-like receptors: their role in allergy and non-allergic inflammatory disease Clin Exp Allergy 2002 32 984 989 12100042 10.1046/j.1365-2745.2002.01451.x
Cario E Podolsky DK Differential alteration in intestinal epithelial cell expression of toll-like receptor 3 (TLR3) and TLR4 in inflammatory bowel disease Infect Immun 2000 68 7010 7017 11083826 10.1128/IAI.68.12.7010-7017.2000
Pitzurra L Bellocchio S Nocentini A Bonifazi P Scardazza R Gallucci L Stracci F Simoncelli C Bistoni F Romani L Antifungal immune reactivity in nasal polyposis Infect Immun 2004 72 7275 7281 15557653 10.1128/IAI.72.12.7275-7281.2004
Muir A Soong G Sokol S Reddy B Gomez M Van Heeckeren A Prince A Toll like receptors in normal and cystic fibrosis airway epithelial cells Am J Respir Cell Mol Biol 2003 4 4 14656745
Guillot L Medjane S Le-Barillec K Balloy V Danel C Chignard M Si-Tahar M Response of human pulmonary epithelial cells to lipopolysaccharide involves Toll-like receptor 4 (TLR4)-dependent signaling pathways: evidence for an intracellular compartmentalization of TLR4 J Biol Chem 2004 279 2712 2718 14600154 10.1074/jbc.M305790200
Monick MM Yarovinsky TO Powers LS Butler NS Carter AB Gudmundsson G Hunninghake GW Respiratory syncytial virus up-regulates TLR4 and sensitizes airway epithelial cells to endotoxin J Biol Chem 2003 278 53035 53044 14565959 10.1074/jbc.M308093200
Hertz CJ Wu Q Porter EM Zhang YJ Weismuller KH Godowski PJ Ganz T Randell SH Modlin RL Activation of toll-like receptor 2 on human tracheobronchial epithelial cells induces the antimicrobial peptide human beta defensin-2 J Immunol 2003 171 6820 6826 14662888
Armstrong L Medford AR Uppington KM Robertson J Witherden IR Tetley TD Millar AB Expression of functional toll-like receptor-2 and -4 on alveolar epithelial cells Am J Respir Cell Mol Biol 2004 31 241 245 15044215 10.1165/rcmb.2004-0078OC
Tulic MK Fiset PO Manoukian JJ Frenkiel S Lavigne F Eidelman DH Hamid Q Role of toll-like receptor 4 in protection by bacterial lipopolysaccharide in the nasal mucosa of atopic children but not adults Lancet 2004 363 1689 1697 15158630 10.1016/S0140-6736(04)16253-3
Aas K Belin L Standardization of diagnostic work in allergy Int Arch Allergy Appl Immunol 1973 45 57 60 4580380
Muzio M Bosisio D Polentarutti N D'Amico G Stoppacciaro A Mancinelli R van't Veer C Penton-Rol G Ruco LP Allavena P Differential expression and regulation of toll-like receptors (TLR) in human leukocytes: selective expression of TLR3 in dendritic cells J Immunol 2000 164 5998 6004 10820283
Kaisho T Akira S Regulation of dendritic cell function through Toll-like receptors Curr Mol Med 2003 3 373 385 12776992 10.2174/1566524033479726
Underhill DM Toll-like receptors: networking for success Eur J Immunol 2003 33 1767 1775 12811836 10.1002/eji.200324037
Soong G Reddy B Sokol S Adamo R Prince A TLR2 is mobilized into an apical lipid raft receptor complex to signal infection in airway epithelial cells J Clin Invest 2004 113 1482 1489 15146246 10.1172/JCI200420773
Greene CM Carroll TP Smith SG Taggart CC Devaney J Griffin S O'Neill SJ McElvaney NG TLR-induced inflammation in cystic fibrosis and non-cystic fibrosis airway epithelial cells J Immunol 2005 174 1638 1646 15661927
Matsumoto M Funami K Tanabe M Oshiumi H Shingai M Seto Y Yamamoto A Seya T Subcellular localization of Toll-like receptor 3 in human dendritic cells J Immunol 2003 171 3154 3162 12960343
Matsumoto M Kikkawa S Kohase M Miyake K Seya T Establishment of a monoclonal antibody against human Toll-like receptor 3 that blocks double-stranded RNA-mediated signaling Biochem Biophys Res Commun 2002 293 1364 1369 12054664 10.1016/S0006-291X(02)00380-7
Shuto T Imasato A Jono H Sakai A Xu H Watanabe T Rixter DD Kai H Andalibi A Linthicum F Glucocorticoids synergistically enhance nontypeable Haemophilus influenzae-induced Toll-like receptor 2 expression via a negative cross-talk with p38 MAP kinase J Biol Chem 2002 277 17263 17270 11867630 10.1074/jbc.M112190200
Strachan DP Hay fever, hygiene, and household size Bmj 1989 299 1259 1260 2513902
Warner JO Worldwide variations in the prevalence of atopic symptoms: what does it all mean? Thorax 1999 54 S46 51 10451693
Lauener RP Birchler T Adamski J Braun-Fahrlander C Bufe A Herz U von Mutius E Nowak D Riedler J Waser M Expression of CD14 and Toll-like receptor 2 in farmers' and non-farmers' children Lancet 2002 360 465 466 12241724 10.1016/S0140-6736(02)09641-1
Use of antibiotics to treat asthma exacerbations J Allergy Clin Immunol 2002 110 S180 183 12518559
Skoner DP Viral infection and allergy: lower airway Allergy Asthma Proc 2002 23 229 232 12221891
Michel O Role of house-dust endotoxin exposure in aetiology of allergy and asthma Mediators Inflamm 2001 10 301 304 11817668 10.1080/09629350152700966
Blease K Kunkel SL Hogaboam CM IL-18 modulates chronic fungal asthma in a murine model; putative involvement of Toll-like receptor-2 Inflamm Res 2001 50 552 560 11766995
|
16146574
|
PMC1243240
|
CC BY
|
2021-01-04 16:23:25
|
no
|
Respir Res. 2005 Sep 7; 6(1):100
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-100
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-1031615939610.1186/1465-9921-6-103ResearchInteraction between human lung fibroblasts and T-lymphocytes prevents activation of CD4+ cells Vancheri Carlo [email protected] Claudio [email protected] Elisa [email protected] Elisa [email protected] Furno Debora [email protected] Maria P [email protected] Massimo [email protected] Rosa Cristina [email protected] Claudia [email protected] Marco [email protected] Nunzio [email protected] Department of Internal and Specialistic Medicine, Section of Respiratory Medicine, University of Catania, Catania, 95125, Italy2005 13 9 2005 6 1 103 103 1 6 2005 13 9 2005 Copyright © 2005 Vancheri et al; licensee BioMed Central Ltd.2005Vancheri et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
T lymphocytes are demonstrated to play an important role in several chronic pulmonary inflammatory diseases. In this study we provide evidence that human lung fibroblasts are capable of mutually interacting with T-lymphocytes leading to functionally significant responses by T-cells and fibroblasts.
Methods
Human lung fibroblast were co-cultured with PMA-ionomycin activated T-CD4 lymphocytes for 36 hours. Surface as well as intracellular proteins expression, relevant to fibroblasts and lymphocytes activation, were evaluated by means of flow cytometry and RT-PCR. Proliferative responses of T lymphocytes to concanavalin A were evaluated by the MTT assay.
Results
In lung fibroblasts, activated lymphocytes promote an increase of expression of cyclooxygenase-2 and ICAM-1, expressed as mean fluorescence intensity (MFI), from 5.4 ± 0.9 and 0.7 ± 0.15 to 9.1 ± 1.5 and 38.6 ± 7.8, respectively. Fibroblasts, in turn, induce a significant reduction of transcription and protein expression of CD69, LFA-1 and CD28 in activated lymphocytes and CD3 in resting lymphocytes. In activated T lymphocytes, LFA-1, CD28 and CD69 expression was 16.6 ± 0.7, 18.9 ± 1.9 and 6.6 ± 1.3, respectively, and was significantly reduced by fibroblasts to 9.4 ± 0.7, 9.4 ± 1.4 and 3.5 ± 1.0. CD3 expression in resting lymphocytes was 11.9 ± 1.4 and was significantly reduced by fibroblasts to 6.4 ± 1.1. Intracellular cytokines, TNF-alpha and IL-10, were evaluated in T lymphocytes. Co-incubation with fibroblasts reduced the number of TNF-alpha positive lymphocytes from 54,4% ± 6.12 to 30.8 ± 2.8, while IL-10 positive cells were unaffected. Finally, co-culture with fibroblasts significantly reduced Con A proliferative response of T lymphocytes, measured as MTT absorbance, from 0.24 ± 0.02 nm to 0.16 ± 0.02 nm. Interestingly, while the activation of fibroblasts is mediated by a soluble factor, a cognate interaction ICAM-1 mediated was demonstrated to be responsible for the modulation of LFA-1, CD28 and CD69.
Conclusion
Findings from this study suggest that fibroblasts play a role in the local regulation of the immune response, being able to modulate effector functions of cells recruited into sites of inflammation.
COX-2ICAM-1CD3CD28LFA
==== Body
Introduction
Interactions between immunocompetent cells, such as lymphocytes and monocytes/macrophages, and other hematopoietic cell lineages is an essential and well known feature of the immune and inflammatory response. Much less attention has been given to the possibility of direct and mutual interactions between immunocompetent cells and resident cells such as fibroblasts. To this regard we have previously shown that normal human lung fibroblasts interact with monocytes suggesting their involvement in the control of the immune and inflammatory response [1,2]. In addition, we have demonstrated that an impairment of fibroblast functions, as observed in fibrotic fibroblasts, may lead to a reduced capability of these cells to modulate monocyte activity [2]. Several data indicate that in pulmonary chronic inflammatory diseases, such as bronchial asthma and interstitial lung diseases, lymphocytes are in an immunologically activated state likely as the result of a persistent and excessive state of immune activation, possibly due to a dysregulation of the fine homeostatic balance governing the immune response [3-5]. In this context, very limited attention has been addressed to potential direct interactions between T lymphocytes and lung fibroblasts [6,7]. Recent studies have in fact provided evidence that the interaction between lymphocytes and fibroblasts might be important to the pathogenesis of chronic inflammatory diseases such as periodontitis and rheumatoid arthritis. In periodontitis, T lymphocytes are often found adjacent to gingival fibroblasts [8] whereas in the inflammed synovium, T lymphocytes and fibroblasts along with monocytes/macrophages, represent the most abundant cell populations. With regard to these disease conditions, it has been demonstrated that T cells induce the activation of both gingival and synovial fibroblasts [9,10]. In addition, it has recently been shown that stromal cells are able to affect T-cell apoptosis, contributing to the accumulation and/or removal of these cells at sites of chronic inflammation [11-13]. However, the inappropriate retention of T-cells within the tissue is unlikely to be the only mechanism leading to the switch from an acute resolving to a chronic persistent inflammatory process and it is reasonable to think that a persistent and excessive condition of immune activation of these cells may be important as well. In view of the above findings, that fibroblasts are capable of interacting with T-lymphocytes, we set out to determine whether the interaction between normal human lung fibroblasts and T-cells could lead to a functionally significant response by T-lymphocytes, influencing their state of immune activation. Our results indicate that lung fibroblasts and T-lymphocytes indeed mutually interact. Activated lymphocytes induce the expression of cyclooxygenase-2 (COX-2) and dramatically increase the expression of intercellular adhesion molecule-1 (ICAM-1) in normal human lung fibroblasts. Fibroblasts, in turn, induce a significant reduction of transcription and protein expression of CD69, considered as a marker of early T cell activation, lymphocyte function associated antigen-1 (LFA-1), CD3 and CD28, all molecules involved in T-lymphocyte activation and costimulation [14-16].
According to this phenotypic down-regulation, lymphocytes co-cultured with fibroblasts, show a significant reduction of the production of tumor necrosis factor-α (TNFα), while the production of interleukin-10 (IL-10) is not affected. This condition of reduced activation is further underlined by a reduced proliferation of lymphocytes co-cultured with fibroblasts in response to a mitogenic stimulus.
It is interesting to note that while the activation of fibroblasts is mediated by a soluble factor, a cognate interaction between ICAM-1 and LFA-1 is responsible for the modulation of LFA-1, CD28 and CD69 on T-cells.
These data confirm and expand the concept that human lung fibroblasts may actively interact with immune cells affecting a large array of functions strictly related to the control and regulation of the local immune response.
Materials and methods
Lung Fibroblasts
Seven primary lines of normal human adult lung fibroblasts were established by using an outgrowth from explant according to the method described by Jordana and coworkers [17]. Fibroblast lines were derived from histologically normal areas of surgical lung specimens from patients undergoing resective surgery for cancer. Their ages ranged from 52 to 61 yr. Five of six patients were men. Lung specimens were chopped into pieces of less than 1 mm3 and washed once with PBS and twice with RPMI-1640 containing 10% FCS, penicillin 100 U/ml, streptomycin 100 mcg/ml, and fungizone 25 mcg/ml (supplemented RPMI) (Gibco, Paisley, UK); eight to ten pieces of washed specimens were then plated in a 100-mm polystyrene dish (Falcon, Becton Dickinson, Lincoln Park, NJ, USA) and overlaid with a coverslip held to the dish with sterile vaseline. Ten milliliters of supplemented RPMI were added and the tissue was incubated at 37°C with 5% CO2. The medium was changed weekly. When a monolayer of fibroblast-like cells covered the bottom of the dish, usually 5 to 6 weeks later, the explant tissue was removed, and the cells were then trypsinized for ten minutes, resuspended in 10 ml of supplemented RPMI, and plated in 100-mm tissue culture dishes. Subsequently, cells were split 1:2 at confluence, usually weekly. Aliquots of cells were frozen and stored in liquid nitrogen. In all experiments we used cell lines at a passage earlier than the tenth.
Lymphocyte isolation procedure
Heparinized venous blood, obtained from healthy donors, was diluted 1:3 with PBS, and 40 ml were then placed on 10 ml of Lymphoprep (Axis-Shield, Oslo, Norway) for centrifugation at 1,600 rpm for 35 minutes at room temperature. Mononuclear cells were collected at the interface, washed three times and resuspended in PBS supplemented with 0.5% bovine serum albumin and 2 mM EDTA. Isolation of human CD4 lymphocytes from mononuclear cells was performed by positive selection of CD4+ cells using a magnetic cell sorting system (MACS, Miltenyi Biotec, Bergisch Gladbach, Germany) according to manufacturer's instructions. Mononuclear cells were magnetically labeled with CD4 microbeads and passed through a separation column placed in the magnetic field of the MACS separator. The magnetically labeled cells were retained in the column while the unlabeled cells run through. After removal of the column from the magnetic field labeled cells, representing the enriched CD4+ cell fraction, passed through the column and were collected as effluent.
Lymphocyte-Fibroblast Co-cultures
Lymphocytes were incubated in 60 mm polystyrene dish (Falcon, Becton-Dickinson) at a concentration of 4 × 106 cells in 4 ml of supplemented RPMI in the absence or presence of 1 μg/ml of ionomycin and 10 ng/ml of PMA, plates were then incubated in a humidified atmosphere of 5% CO2 at 37°C. After 6 hours cells were harvested, washed three times with PBS and counted. 1 × 106 lymphocytes were then seeded on top of 0.5 × 106 fibroblasts in 6-well tissue culture plates in a final volume of 2 ml of supplemented RPMI and incubated for 36 hours. After the 36 hours of co-culture fibroblasts were adherent to the dish and maintained the typical spindle shaped aspect. Lymphocyte viability was assessed by the trypan blue exclusion method that constantly gave a >90% survival. In some experiments cells were separated by a semipermeable membrane (0.4 mcm pores) using a cell culture insert (Falcon, Becton Dickinson). In blocking experiments fibroblasts were pretreated with a blocking anti-ICAM antibody (Calbiochem Corporation, San Diego, CA, USA) for 2 hours before the addition of the lymphocytes and once again when the co-culture started.
RNA Isolation and Reverse Transcriptase-Polymerase Chain Reaction
Total cellular RNA was extracted from cells with the guanidium isothiocyanate/acid-phenol procedure as previously described [18]. The yield and the purity of RNA was measured spectrophotometrically by absorption at 260/280 nm. Total RNA was used for the generation of cDNA. Reverse transcriptase-polymerase chain reaction (RT-PCR) was performed using the SuperScript™ First-Strand Synthesis System for RT-PCR (Invitrogen Inc., Paisley, UK), with some modifications. Briefly, 5 μg of total RNA was reverse transcribed with 50 U of RNase OUT Recombinant (Superscript™ II RT, Invitrogen). The reverse-transcribed product (cDNA) was amplified by PCR (Perkin Elmer Gene Amp PCR System 2400) in the presence of a master mix containing PCR buffer, MgCl2 (under optimal concentrations), 1 U Taq DNA Polymerase Recombinant (Invitrogen), 10 mM dNTPs. The following specific primer pairs were used: ICAM-1 sense 5'-GAGCTGTTTGAGAACACCTC-3' and antisense 5'-TCACACTTCACTGTCACCTC-3' giving a 367 bp PCR product; COX-2 sense 5'-TTCAAATGAGATTGTGGGAAAATTGCT-3' and antisense 5'-AGATCATCTCTGCCTGAGTATCTT-3' (305 bp product); LFA-1 sense 5'-GTCCTCTGCTGAGCTTTACA-3' and antisense 5'-ATCCTTCATCCTTCCAGCAC-3' (337 bp product); CD-28 sense 5'-AAGTTGAGAGCCAAGAGCAG-3' and antisense 5'-CCGACTATTTTTCAGTGACA-3' (304 bp product); CD-69 sense 5' CCTTCCAAGTTCCTGTCC-3' and antisense 5' CATTCCATGCTGCTGACCTC-3' (451 bp product); CD-3 sense 5' GTGTCATTCTCACTGCCTTGTTCC-3' and antisense 5'-TTCAGTGGCTGAGAAGAGTGAACC-3' (496 bp product); beta-actin sense 5'-TGACGGGGTCACCCACACTGTGCCCATCTA-3' and antisense 5'-CTAGAAGCATTGCGGTGGACGATGGAGGG-3' (661 bp product). PCR was performed for 40 cycles, using a cycling program of 94°C for 5 min, 55°C for 59 sec and 72°C for 59 sec in a thermal cycler for the amplification of ICAM-1 and COX-2, for the amplification of LFA-1 and CD-28, PCR was performed for 35 cycles, using a cycling program of 94°C for 5 min, 54°C for 59 sec and 72°C for 59 sec, while for the amplification of CD-69 and CD-3 PCR was performed for 25 and 30 cycles, using a cycling program of 94°C for 5 min, 52°C and 57°C for 59 sec and 72°C for 59 sec, respectively. Final extension was at 72°C for 7 min for all molecules. PCR-amplified products (10 μl) were electrophoresed through a 1,8% agarose gel (Ambion Inc., Austin, Tx, USA) containing 0,5 μg/ml of ethidium bromide and compared with DNA reference markers. Products were visualized by ultraviolet illuminations. Polaroid photographs with ultraviolet exposure were taken with a 665 Polaroid film. Bands were analyzed with the Phoretix 1D version 3.0.
Flow cytometric analysis
Experiments to determine ICAM-1 and COX-2 expression on fibroblasts and LFA-1, CD3, CD28 and CD69 expression on lymphocytes were carried out on cells isolated and co-cultured as described before. After 36 hours of co-colture cells were lightly trypsinized, washed and resuspended in PBS with 0.1% BSA. The cells were incubated with primary antibodies, anti-LFA-1 mAb (Dako Italia, Milan, Italy), anti-CD3 and anti-CD69 mAbs (Beckman Coulter Italia, Milan, Italy), or anti-COX-2 policlonal Ab (Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA) for 60 min at room temperature. Following washing, the secondary antibody, fluorescein (FITC)-conjugated rabbit anti-mouse IgG, was added for 60 min at room temperature. Controls included omission of the primary antibody and incubation only with the secondary antibody. For COX-2 detection cells were in advance permeabilized with Triton 10x for 5 min at 4°C. FITC-labeled anti-ICAM-1 (Dako Italia) and PE-labeled anti-CD28 (BD Pharmingen Italia, Milan, Italy) were also used. Samples were analyzed using a Coulter Epics Elite ESP flow cytometer (Coulter Corporation, Miami, FL, USA). Fibroblasts and lymphocytes were gated on the basis of forward and side scatter profile. Intracellular staining of cytokines was performed using a method originally developed by Laskay and Anderson [19] and recently modified by Assenmacher et al. [20]. Briefly, Brefeldin A at 10 μg/mL (Sigma-Aldrich Co., St Louis, MO, USA), was added to cultures and cocultures CD4+ T cells and fibroblasts described above, for the final 5 hours of our experimental setup. Cells were then harvested and washed once in PBS. Freshly prepared formaldehyde solution (2% in PBS) was added to the cell pellet. Cells were vigorously resuspended and fixed for 20 minutes at room temperature. After washing in saponin buffer (0.5% saponin and 1% BSA in PBS) (Sigma-Aldrich Co., St Louis, MO, USA) the cells were stained, for 1 hour in the dark, in 100 mcl of saponin buffer containing FITC- or PE-conjugated anti-cytokine antibody at the following concentrations: anti-IL10-PE (2.5 μg/ml), anti-IFNγ-PE (2.5 μg/ml), anti-TNFalpha-PE (2.5 μg/ml), anti-IL4-FITC (5 μg/ml) (Caltag Laboratories, Burlinghame, CA, USA). Thereafter, the cells were washed three times with saponin buffer, once with PBS and analyzed by flow cytometry. Gating was always restricted on T cells. Therefore, all depicted data are given in percent of CD4+ T cells. Control stainings with PE- or FITC-coupled isotype-matched antibody were performed in preliminary experiment and never stained >0.3% of CD4+ T cells.
At least 10,000 forward and side scatter gated events were collected per specimen. Cells were excited at 488 nm and the fluorescence was monitored at 525 nm. Fluorescences were collected using logarithmic amplification.
Lymphocytes proliferation assay
After 36 hours co-culture protocol CD4+ lymphocytes were harvested and plated at a density of 2.5 × 105 cells in 24 well plates in supplemented RPMI with 2,5 mcg/ml Concanavalin A (Con A, Sigma-Aldrich Co.) and incubated for 72 hours at 37°C in a 5% CO2 atmosphere, lymphocytes co-cultured with fibroblasts were very gently harvested with warm PBS to detach them from the fibroblasts monolayer, CD4+ T cells harvested this way, had always more than 95% of purity as assessed by differential cell counts and by flow cytometry. Thereafter medium was removed, cells were incubated with fresh medium containing 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) (Sigma-Aldrich Co.) at a final concentration of 0.9 mg/ml for 2 h at 37°C. The solubilization solution, containing acidified isopropanol and 20% SDS, was added and left for 20 min in order to extract the produced formazan which was then evaluated in a plate reader (absorbance = 560 nm).
Statistical analysis
Statistical comparisons of the levels of expression of ICAM-1, COX-2, LFA-1, CD3, CD28, CD69, IL10 and TNF-alpha in all different experimental conditions were performed using a two-way analysis of variance (ANOVA) followed by the Newman-Keuls test for comparisons of specific means, the same tests were used to assess differences among T cells proliferative responses to Con A. A p value of less than 0.05 was considered significant. Results are expressed as mean ± SE.
Results
Cultured lung fibroblasts, under basal conditions, expressed both ICAM-1 and COX-2 as measured by mean fluorescence intensity (MFI) of positive cells at flow cytometric analysis. Exposure of fibroblasts to resting lymphocytes produced an increased expression of both ICAM-1 (from 0.7 ± 0.4 to 2.5 ± 1.4) and COX-2 (from 5.4 ± 0.9 to 6.8 ± 1.1) that however did not yield statistical significance. In contrast, co-incubation of fibroblasts with activated lymphocytes determined a pronounced increase of the expression of both ICAM-1 and COX-2 MFI to 38.6 ± 7.8 (P < 0.001) and 9.1 ± 1.5 (P < 0.001), (Fig. 1a–d). This effect was fully preserved when the two cell types were maintained physically separated by a semi-permeable membrane. The ability of activated lymphocytes to affect ICAM-1 and COX-2 expression was likely exerted at the transcriptional level as suggested by RT-PCR that revealed increased ICAM-1 and COX-2 transcripts in fibroblasts incubated with activated lymphocytes for 36 h (Fig 2a,b). Again, the increased expression was maintained in the presence of a separating semi-permeable membrane.
Figure 1 Representative flow cytometry histograms of ICAM-1 (a) and COX-2 (b) in fibroblast alone (FA), fibroblasts co-cultured with PMA-stimulated lymphocytes (FA+LPMA). ICAM-1 (c) and COX-2 (d) expression in fibroblasts, fibroblasts co-cultured with PMA-stimulated lymphocytes, fibroblasts co-cultured with PMA-stimulated lymphocytes in the presence of a semipermeable membrane (FA+LPMA ins.) Data represent means ± SE of seven independent experiments in which seven different cells lines were used.
Figure 2 Levels of mRNA for ICAM-1 (a) and COX-2 (b) in fibroblasts (FA), fibroblasts co-cultured with PMA-stimulated lymphocytes (FA+LPMA), fibroblasts co-cultured with PMA-stimulated lymphocytes in the presence of a semipermeable membrane (FA+LPMA ins.). In the upper panels, modifications in the appearance of a 367-bp (ICAM-1) and a 305-bp (COX-2) band are compared with that of a β-actin. In the lower panels, the densitometric analysis is shown. Data are from one experiment representative of three.
Expression of LFA-1, CD28 and CD69 in lymphocytes was markedly increased by exposure to 10 ng/ml PMA for 6 h, from 8.4 ± 0.5 for LFA-1, 6.9 ± 0.4 for CD28 and 2.6 ± 0.6 for CD69 to 16.6 ± 0.7 (P < 0.001), 18.9 ± 1.9 (P < 0.001) and 6.6 ± 1.3 (P < 0.01), respectively (Fig. 3a–f). However, co-incubation of activated lymphocytes with fibroblasts, significantly reduced the expression of LFA-1, CD28 and CD69 to 9.4 ± 0.7 for LFA-1 (P < 0.001), 9.4 ± 1.4 for CD28 (P < 0.001) and 3.5 ± 1.0 (P < 0.05) for CD69, an effect that required the physical contact between the two cell types, as suggested by the fact that it was not present any more when a semi-permeable membrane was applied (Fig. 3b,d,f). In this condition, MFI was 16.5 ± 0.6 for LFA-1, 18.4 ± 1.9 for CD28 and 7.1 ± 1.8 for CD69, all these values being statistically increased compared to cells cultured without the membrane (P < 0.01) and not significantly different to those observed in activated lymphocytes cultured in absence of fibroblasts. RT-PCR revealed that the reduced expression of LFA-1, CD28 and CD69 was related to their decreased transcription (Fig. 4a–c) and, similarly to what observed with protein expression, the reducing effect of fibroblasts was prevented by the presence of a membrane between the two cell types (Fig. 4a–c).
Figure 3 Representative flow cytometry histograms of LFA-1 (a), CD28 (c) and CD69 (e) in PMA-stimulated lymphocytes (LPMA), PMA-stimulated lymphocytes co-cultured with fibroblasts (LPMA+FA). LFA-1 (b), CD28 (d) and CD69 (f) expression in resting lymphocytes (LA), PMA-stimulated lymphocytes, PMA-stimulated lymphocytes co-cultured with fibroblasts and PMA-stimulated lymphocytes co-cultured with fibroblasts in the presence of a semipermeable membrane (LPMA+FA ins.). Data represent means ± SE of seven independent experiments in which seven different cells lines were used.
Figure 4 Levels of mRNA for LFA-1 (a), CD28 (b) and CD69 (c) in resting lymphocytes (LA), PMA-stimulated lymphocytes (LPMA), PMA-stimulated lymphocytes co-cultured with fibroblasts (LPMA+FA) and PMA-stimulated lymphocytes co-cultured with fibroblasts in the presence of a semipermeable membrane (LPMA+FA ins.). In the upper panels, modifications in the appearance of a 337-bp (LFA-1), a 304-bp (CD28) and a 451-bp (CD69) band are compared with that of a β-actin. In the lower panels, the densitometric analysis is shown. Data are from one experiment representative of three.
To ascertain whether the reduced protein expression could be ascribed to activation of ICAM-1 on lymphocyte surface, the same experiment was performed in the presence of a blocking anti-ICAM-1 antibody. Under these conditions, lymphocyte expression of LFA-1, CD28 and CD69 was fully restored, being statistically increased compared to cells cultured with fibroblasts and not significantly different to those observed in activated lymphocytes (Fig. 5a–c).
Figure 5 LFA-1 (a), CD28 (b) and CD69 (c) expression in PMA-stimulated lymphocytes (LPMA), PMA-stimulated lymphocytes co-cultured with fibroblasts (LPMA+FA) and PMA-stimulated lymphocytes co-cultured with fibroblasts in the presence of an anti ICAM-1 blocking antibody (LPMA+FA+anti ICAM-1). Data represent means ± SE of four independent experiments.
Expression of CD3 was evaluated in resting lymphocytes as activation by PMA did not significantly modify expression of the protein. Co-incubation of resting lymphocytes with fibroblasts significantly reduced CD3 expression both at translational (Fig. 6a,b) and transcriptional (Fig. 6c) level, an effect that was completely prevented in the presence of a separating membrane. Indeed, CD3 MFI was reduced from 11.9 ± 1.4 to 6.4 ± 1.1 (P < 0.001), while in experiments performed with a separating membrane, CD3 MFI was restored to (11.0 ± 0.9) (P < 0.001).
Figure 6 (a) Representative flow cytometry histogram of CD3 in resting lymphocytes (LA) and in resting lymphocytes co-cultured with fibroblasts (LA+FA). (b) CD3 expression in resting lymphocytes, resting lymphocytes co-cultured with fibroblasts and resting lymphocytes co-cultured with fibroblasts in the presence of a semipermeable membrane (LA+FA ins.). Data represent means ± SE of seven independent experiments in which seven different cells lines were used. (c) Levels of mRNA for CD3 in resting lymphocytes, resting lymphocytes co-cultured with fibroblasts and resting lymphocytes co-cultured with fibroblasts in the presence of a semipermeable membrane. In the upper panel, modifications in the appearance of a 496-bp (CD3) band are compared with that of a β-actin. In the lower panel, the densitometric analysis is shown. Data are from one experiment representative of three.
Finally, in order to evaluate whether the interaction between fibroblasts and lymphocytes could give rise to changes of the function of the latter cell population, two different approaches were used. As a first step, lymphocytes exposed to fibroblasts were tested for their intracellular production of cytokines. Attention has been focused on TNF-alpha and IL-10, two cytokines whose role in lymphocyte function has been thoroughly characterized. As expected, an increase of both TNF-alpha (from 2.5% ± 5.4 to 54.4% ± 6.12, p < 0.01) and IL-10 (from 15.8% ± 5.2 to 53.3% ± 4.6, p < 0.01) positive cells was observed following activation with PMA, (Fig. 7b). Interestingly, co-incubation with fibroblasts significantly reduced the number of TNF-alpha positive lymphocytes (from 54.4% ± 6.12 to 30.8% ± 2.8, p < 0.05), without any significant change of IL-10 positive cells (from 53.3% ± 4.6 to 54.9% ± 5.9, p = NS). Finally we tested the ability of concanavalin A to induce a proliferative response in activated lymphocytes. This proliferation was partially quenched in lymphocytes co-cultured with fibroblasts (Fig. 8). Indeed, proliferation measured as MTT absorbance was reduced from 0.24 nm ± 0.02 of activated lymphocytes to 0.15 nm ± 0.04, p < 0.05, an effect fully maintained when a semi-permeable membrane was applied (0.16 nm ± 0.02, p < 0.01).
Figure 7 (a) Representative flow cytometry dot plots of TNF-α and IL-10 positive PMA-stimulated lymphocytes (LPMA) and PMA-stimulated lymphocytes co-cultured with fibroblasts (LPMA+FA). (b) TNF-α and IL-10 positive cells, expressed as percentage, among resting lymphocytes (LA), PMA-stimulated lymphocytes and PMA-stimulated lymphocytes co-cultured with fibroblasts. Data represent means ± SE of 4 independent experiments.
Figure 8 Proliferative responses of resting lymphocytes (LA), PMA-stimulated lymphocytes (LPMA), PMA-stimulated lymphocytes co-cultured with fibroblasts (LPMA+FA) and PMA-stimulated lymphocytes co-cultured with fibroblasts in the presence of a semipermeable membrane (LPMA+FA ins.), measured by means of MTT, after 72 hours culture in the presence of 2,5 μg/ml concanavalin A. Data represent means ± SE of 6 independent experiments.
Discussion
A great deal of evidence is today available showing that resident cells such as fibroblasts, through the release of soluble signals and/or direct interactions with other cells, may serve as potential regulators of the local inflammatory response [21]. Based on our previous findings, human lung fibroblasts, through the modulation of some monocyte activities, may also participate in the control of the immune response [1]. We have shown that normal human lung fibroblasts are able to interact with monocytes, driving the release of cytokines, whose role is crucial in the regulation of the immune response, i.e. they strongly stimulate interleukin 10 (IL-10) production by LPS-activated monocytes and inhibit interleukin 12 (IL-12). The increase of IL-10 production, induced by fibroblasts, is able, in an autocrine way, to downregulate the expression of human leukocyte-associated antigen-DR (HLA-DR) as well as the expression of CD40 on monocytes, potentially affecting the antigen presenting capacity of these cells as well as their costimulatory function. In addition, we have also shown that an impairment of fibroblast functions, as observed in fibrotic fibroblasts, may lead to a reduced capability of these cells to modulate monocyte activity [1].
In view of the above mentioned experimental evidence we set out to determine whether the interaction between normal human lung fibroblasts and immune cells such as T-lymphocytes could lead to a functionally significant response by these cells.
Several lung diseases, including hypersensitivity pneumonitis, sarcoidosis and bronchial asthma are in fact characterized by an involvement of T-lymphocytes and a subsequent impairment of the immune and inflammatory response [3,22-24]. Another element, common to these diseases, is represented by the possibility that the inflammatory state may ultimately lead to fibroblast activation and tissue fibrosis. Indeed, interstitial lung diseases are marked by fibrosis and also bronchial asthma is characterized by an extensive remodeling of the bronchial wall due to fibroblast activation and collagen deposition [25-27].
Our results indicate that lung fibroblasts and T-lymphocytes mutually interact. Activated lymphocytes induce COX-2 mRNA accumulation and protein expression and dramatically increase both transcription and expression of ICAM-1 in normal human lung fibroblasts. Fibroblasts, in turn, induce a significant reduction of transcription and protein expression of CD69, a marker of early T activation, and LFA-1, CD3 and CD28, all molecules involved in T-lymphocyte activation and costimulation. Moreover, TNF-alpha a typical proinflammatory cytokine [28], was significantly inhibited by fibroblasts, whereas IL-10, commonly considered as a regulatory cytokine [29] was not affected by fibroblasts. Finally, we have demonstrated that lymphocytes co-cultured with fibroblasts show a significantly reduced proliferative response to a mitogenic stimulus.
The enhanced expression of both COX-2 and ICAM-1 on fibroblasts, induced by activated lymphocytes, was not affected by the presence of a semi-permeable membrane separating fibroblasts and lymphocytes, suggesting that the T cell-induced fibroblast activation is likely mediated by soluble factors produced by lymphocytes. The increased expression of COX-2 induced by T cells on fibroblasts is of great interest considering that a large number of studies depicts COX-2 and its products, prostaglandins, as a major pathway occurring in the lung during the control and self-limitation of the inflammatory and reparative process [30]. As regard as the increased expression of ICAM-1 on fibroblasts cocultured with lymphocytes, it has already been shown that various cytokines produced by mononuclear cells enhance adhesiveness of fibroblasts for T cells, through up-regulation of fibroblast ICAM-1 expression, promoting T cell retention, positioning and accumulation in the tissues [31,32]. This observation suggests that T cells, once migrated into the tissue, facilitate their own retention by boosting local fibroblast adhesive properties. This event is commonly considered important for the persistence of inflammation and blockade of the interaction between resident cells and T cells may eventually down-regulate inflammation, representing an ideal target for disrupting immune-non-immune cell interaction. To this regard, in different animal experimental models attenuation of inflammation and reduction in collagen deposition have been described when neutralizing antibodies against adhesion molecules are used [33,34]. However, in an ICAM-1 knockout mice, bleomycin has been reported to induce a more severe pulmonary fibrosis compared to their wild-type counterparts [35]. These results confirm the importance of adhesion molecules for the recruitment and accumulation of inflammatory cells into the tissues and, most importantly, suggest their fundamental role in cell-to-cell communication and inflammation control. Indeed, our findings demonstrate that the adhesion and subsequent interaction between lymphocytes and fibroblasts have an important role in down-regulating T cell activation and likely in dampening the activation state of these cells that characterizes some lung diseases.
With regard to this we have shown that the inhibitory effect exerted by fibroblasts on T lymphocytes activity is evident at multiple levels: surface markers expression, cytokines production and proliferation.
We have shown that the inhibitory effect exerted by fibroblasts on the expression of LFA-1, CD3, CD28 and CD69 on lymphocytes is due to a direct interaction since the inhibitory effect is abolished when a membrane is placed between fibroblasts and lymphocytes. Considering that ICAM-1 expression on fibroblasts makes these cells capable to physically interact with beta 2 integrin positive inflammatory leukocytes and its expression is increased by activated lymphocytes, we hypothesized that the interaction between activated T cells and fibroblasts was ICAM-1 mediated. Indeed, co-culture experiments performed preincubating fibroblasts with a neutralizing anti-ICAM-1 antibody abolished their capacity to reduce LFA-1, CD28 and CD69 expression on activated lymphocytes suggesting that a cognate interaction, LFA-1-ICAM-1 mediated, is requested for the inhibitory effect of fibroblasts on lymphocytes. Similarly, the inhibitory effect induced by fibroblasts on T cell CD3 expression has to be ascribed to a direct interaction, as demonstrated by the disappearance of the inhibitory activity in the presence of a membrane physically separating fibroblasts and lymphocytes. The inhibition of CD3 expression as well as the simultaneous reduction of LFA-1, CD28, CD69 that we have described on lymphocytes cultured with fibroblasts may have important effects on the evolution of the immune response. A full T cell activation depends, in fact, (i) on the interaction of the T cell receptor (TCR)-CD3 complex with the antigen and (ii) on a second signal delivered through the costimulatory molecule CD28, necessary for the complete activation [36,37]. The binding of CD28 and its ligands B7-1 and B7-2 lowers the threshold number of TCR molecules that need to be engaged to initiate T cell activation, protects against the induction of T cell anergy, promotes proliferation, cytokine production and T cell survival [37,38]. In addition, the interaction between T cell-mounted LFA-1 and its ligand ICAM-1 functions not only as an adhesive interaction, but may also deliver a further costimulatory signal able to activate T cells and is considered important for T cell- antigen presenting cell (APC) interaction [39]. Several data suggest that these molecules are crucial in the pathogenesis of a number of lung diseases. Specifically, an influx of activated T-cells characterizes hypersensitivity pneumonitis and the blockade of T-cell costimulation inhibits lung inflammation in a murine model of this disease [23,40]. In pulmonary sarcodosis, a number of reports have emphasized the role of pulmonary lymphocytes and described phenotypic and functional abnormalities of T-cells [22,41] whereas there is evidence showing that T-lymphocytes may also have a role in the fibrotic process, although much remains to be defined. In this regard Okazaky et al. [42] have recently shown that CD28-mediated T-cell costimulation plays a critical role in the development of inflammation and fibrosis in bleomycin-treated mice. On the contrary, the administration of antibodies to LFA-1 leads to a significant reduction in collagen deposition in lung tissue both in experimental hypersensitivity pneumonitis [43] and pulmonary fibrosis [33]. Furthermore, allergen-induced production of T-cell derived cytokines in asthma requires the interaction between costimulatory molecules and points to the CD28-B7 pathway as being important to the inflammation distinctive of the disease [44].
To ascertain whether the phenotypical changes above described were accompanied by a certain degree of functional alteration we studied cytokine production and proliferation of co-cultured T lymphocytes.
Specifically we evaluated the production of TNF-alpha and IL-10 by lymphocytes co-cultured with fibroblasts. TNF-alpha, is a proinflammatory cytokine with many biologic properties thought to be critical in the development of pulmonary fibrosis [28]. Whereas, IL-10 is a regulatory cytokine, that suppresses the production of several proinflammatory cytokines [29], likely through the inhibition of NF-kB activation [45].
According to the immunoregulatory role that we have suggested for fibroblasts, these cells inhibit in T lymphocytes the production of the proinflammatory cytokine TNF-alpha whereas the release of the anti-inflammatory cytokine IL-10 is not affected. Moreover, T-lymphocytes co-cultured with fibroblasts show a significantly reduced proliferative response when exposed to a well known mitogenic stimulus such as concanavalin A [46], underlining the immunoregulatory role exerted by fibroblasts on activated t-lymphocytes.
Thus, several lines of evidence indicate that T-lymphocytes, through the production of soluble signals and direct interactions with other cells orchestrate the immune as well as the inflammatory response and may consequently be involved in the pathogenesis of several lung diseases. In this study, we have shown that CD3, CD28, CD69 and LFA-1, some of the main signals identifying T-cell activation, and taking active part in the immune response, besides TNF-alpha production, are down-regulated by human normal lung fibroblasts, affecting the capacity of these cells to proliferate in response to a non specific stimulus and potentially altering the interaction with antigen presenting cells. This study confirms and expands the concept that tissue resident cells such as fibroblasts do have a role in the regulation of some aspects of the immune response acting as a highly effective local control system of the effector functions of cells recruited into sites of inflammation. An impairment of this modulatory attitude can make the difference between acute self-limited inflammation followed by normal repair as compared to chronic unquenchable inflammation leading to loss of tissue architecture and fibrosis.
Acknowledgements
We are indebted to Dr. Filippo Palermo for his statistical analysis assistance and helpful suggestions.
==== Refs
Vancheri C Mastruzzo C Tomaselli V Sortino MA D'Amico L Bellistri G Pistorio MP Salinaro ET Palermo F Mistretta A Crimi N Normal human lung fibroblasts differently modulate interleukin-10 and interleukin-12 production by monocytes: implications for an altered immune response in pulmonary chronic inflammation Am J Respir Cell Mol Biol 2001 25 592 9 11713101
Vancheri C Sortino MA Tomaselli V Mastruzzo C Condorelli F Bellistri G Pistorio MP Canonico PL Crimi N Different expression of TNF-alpha receptors and prostaglandin E(2). Production in normal and fibrotic lung fibroblasts: potential implications for the evolution of the inflammatory process Am J Respir Cell Mol Biol 2000 22 628 34 10783136
Larche M Robinson DS Kay AB The role of T lymphocytes in the pathogenesis of asthma J Allergy Clin Immunol 2003 111 450 63 12642820 10.1067/mai.2003.169
Agostini C Siviero M Semenzato G Immune effector cells in idiopathic pulmonary fibrosis Curr Opin Pulm Med 1997 3 3 48 55
Sime PJ O'Reilly KM Fibrosis of the lung and other tissues: new concepts in pathogenesis and treatment Clin Immunol 2001 99 308 19 11358425 10.1006/clim.2001.5008
Crowston JG Salmon M Khaw PT Akbar AN T-lymphocyte-fibroblast interactions Biochem Soc Trans 1997 25 529 31 9191149
Hogaboam CM Smith RE Kunkel SL Dynamic interactions between lung fibroblasts and leukocytes: implications for fibrotic lung disease Proc Assoc Am Physicians 1998 110 313 20 9686679
Yucel-Lindberg T Brunius G Wondimu B Anduren I Modeer T Enhanced cyclooxygenase-2 mRNA expression in human gingival fibroblasts induced by cell contact with human lymphocytes Eur J Oral Sci 2001 109 187 92 11456350 10.1034/j.1600-0722.2001.00999.x
Murakami S Hino E Shimabukuro Y Nozaki T Kusumoto Y Saho T Hirano F Hirano H Okada H Direct interaction between gingival fibroblasts and lymphoid cells induces inflammatory cytokine mRNA expression in gingival fibroblasts J Dent Res 1999 78 69 76 10065948
Yamamura Y Gupta R Morita Y He X Pai R Endres J Freiberg A Chung K Fox DA Effector function of resting T cells: activation of synovial fibroblasts J Immunol 2001 166 2270 5 11160281
Hyde H Borthwick NJ Janossy G Salmon M Akbar AN Upregulation of intracellular glutathione by fibroblast-derived factor(s): enhanced survival of activated T cells in the presence of low Bcl-2 Blood 1997 89 2453 60 9116289
Gombert W Borthwick NJ Wallace DL Hyde H Bofill M Pilling D Beverley PC Janossy G Salmon M Akbar AN Fibroblasts prevent apoptosis of IL-2-deprived T cells without inducing proliferation: a selective effect on Bcl-XL expression Immunology 1996 89 397 404 8958053 10.1046/j.1365-2567.1996.d01-759.x
Yarovinsky TO Hunninghake GW Lung fibroblasts inhibit activation-induced death of T cells through PGE(2)-dependent mechanisms Am J Physiol Lung Cell Mol Physiol 2001 281 L1248 56 11597917
Croft M Dubey C Accessory molecule and costimulation requirements for CD4 T cell response Crit Rev Immunol 1997 17 89 118 9034725
Marzio R Manuel J Betz-Corradin S CD69 and regulation of the immune function Immunopharmacol Immunotoxicol 1999 21 565 82 10466080
Wingren AG Parra E Varga M Kalland T Sjogren HO Hedlund G Dohlsten M T cell activation pathways: B7, LFA-3, and ICAM-1 shape unique T cell profiles Crit Rev Immunol 1995 15 235 53 8834450
Jordana M Schulman J McSharry C Irving LB Newhouse MT Jordana G Gauldie J Heterogenous proliferative characteristics of human adult lung fibroblast lines and clonally derived fibroblasts from control and fibrotic tissue Am Rev Respir Dis 1988 137 579 84 3345039
Chomczynski P Sacchi N Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction Anal Biochem 1987 162 156 9 2440339 10.1016/0003-2697(87)90021-2
Laskay T Andersson U Andersson J Kiessling R DeLey M An immunofluorescent method for identifying individual IFN-gamma-producing lymphocytes J Immunol Methods 95 1 1 7 1986 Dec 4 2431063 10.1016/0022-1759(86)90310-8
Assenmacher M Schmitz J Radbruch A Flow cytometric determination of cytokines in activated murine T helper lymphocytes: expression of interleukin-10 in interferon-gamma and in interleukin-4-expressing cells Eur J Immunol 1994 24 5 1097 101 7910138
Boilard E Surette ME Anti-CD3 and concanavalin A-induced human T cell proliferation is associated with an increased rate of arachidonate-phospholipid remodeling J Biol Chem 276 20 17568 75 2001 May 18 11278296 10.1074/jbc.M006152200
Agostini C Meneghin A Semenzato G T-lymphocytes and cytokines in sarcoidosis Curr Opin Pulm Med 2002 8 435 40 12172449 10.1097/00063198-200209000-00016
Israel-Assayag E Fournier M Cormier Y Blockade of T cell costimulation by CTLA4-Ig inhibits lung inflammation in murine hypersensitivity pneumonitis J Immunol 1999 163 6794 9 10586079
Wahlstrom J Berlin M Skold CM Wigzell H Eklund A Grunewald J Phenotypic analysis of lymphocytes and monocytes/macrophages in peripheral blood and bronchoalveolar lavage fluid from patients with pulmonary sarcoidosis Thorax 1999 54 339 46 10092696
Kuhn C McDonald JA The roles of the myofibroblast in idiopathic pulmonary fibrosis. Ultrastructural and immunohistochemical features of sites of active extracellular matrix synthesis Am J Pathol 1991 138 1257 65 2024710
Mornex JF Leroux C Greenland T Ecochard D From granuloma to fibrosis in interstitial lung diseases: molecular and cellular interactions Eur Respir J 1994 7 779 85 8005262 10.1183/09031936.94.07040779
Redington AE Fibrosis and airway remodelling Clin Exp Allergy 2000 30 Suppl 1 42 5 10849474 10.1046/j.1365-2222.2000.00096.x
Piguet PF Ribaux C Karpuz V Grau GE Kapanci Y Expression and localization of tumor necrosis factor-alpha and its mRNA in idiopathic pulmonary fibrosis Am J Pathol 1993 143 651 55 8362967
Pretolani M Interleukin-10: an anti-inflammatory cytokine with therapeutic potential Clin Exp Allergy 1999 29 1164 1171 10469024 10.1046/j.1365-2222.1999.00456.x
Vancheri C Mastruzzo C Sortino MA Crimi N The lung as a priviliged site for the beneficial actions of prostaglandin E2 Trends Immunol 2004 25 40 46 14698283 10.1016/j.it.2003.11.001
Murakami S Shimabukuro Y Saho T Hino E Kasai D Hashikawa T Hirano H Okada H Immunoregulatory roles of adhesive interactions between lymphocytes and gingival fibroblasts J Periodontal Res 1997 32 110 4 9085219
Musso A Condon TP West GA De La Motte C Strong SA Levine AD Bennett CF Fiocchi C Regulation of ICAM-1-mediated fibroblast-T cell reciprocal interaction: implications for modulation of gut inflammation Gastroenterology 1999 117 546 56 10464130
Piguet PF Rosen H Vesin C Grau GE Effective treatment of the pulmonary fibrosis elicited in mice by bleomycin or silica with anti-CD-11 antibodies Am Rev Respir Dis 1993 147 435 41 7679265
Sligh JE JrBallantyne CM Rich SS Hawkins HK Smith CW Bradley A Beaudet AL Inflammatory and immune responses are impaired in mice deficient in intercellular adhesion molecule 1 Proc Natl Acad Sci USA 1993 90 8529 33 8104338
Doherty DE Hilty BA Majewski C ICAM-1 deficiency enhances the development of pulmonary fibrosis in bleomycin-treated mice Am J Resp Crit Care Med 1998 157 A858 (Abstr)
Bugeon L Dallman MJ Costimulation of T cells Am J Respir Crit Care Med 2000 162 S164 8 11029388
Lenschow DJ Walunas T Bluestone JA CD28/B7 system of T cell costimulation Annu Rev Immunol 1996 14 233 58 8717514 10.1146/annurev.immunol.14.1.233
Viola A Lanzavecchia A T cell activation determined by T cell receptor number and tunable thresholds Science 1996 273 104 6 8658175
Chirathaworn C Kohlmeier JE Tibbetts SA Rumsey LM Chan MA Benedict SH Stimulation through intercellular adhesion molecule-1 provides a second signal for T cell activation J Immunol 2002 168 5530 7 12023348
Ando M Suga M Kohrogi H A new look at hypersensitivity pneumonitis Curr Opin Pulm Med 1999 5 299 304 10461534 10.1097/00063198-199909000-00006
Iida K Kadota J Kawakami K Matsubara Y Shirai R Kohno S Analysis of T cell subsets and beta chemokines in patients with pulmonary sarcoidosis Thorax 1997 52 431 7 9176534
Okazaki T Nakao A Nakano H Takahashi F Takahashi K Shimozato O Takeda K Yagita H Okumura K Impairment of bleomycin-induced lung fibrosis in CD28-deficient mice J Immunol 2001 167 1977 81 11489978
Denis M Bisson D Blockade of leukocyte function-associated antigen (LFA-1) in a murine model of lung inflammation Am J Respir Cell Mol Biol 1994 10 481 6 8179910
Lordan JL Davies DE Wilson SJ Dent G Corkhill A Jaffar Z Roberts K Djukanovic R Holgate ST The role of CD28-B7 costimulation in allergen-induced cytokine release by bronchial mucosa from patients with moderately severe asthma J Allergy Clin Immunol 2001 108 976 81 11742276 10.1067/mai.2001.119740
Lentsch AB Shanley TP Sarma V Ward PA In vivo suppression of NF-B and preservation of IB by interleukin-10 and interleukin-13 J Clin Invest 1997 100 2443 2448 9366558
Boilard E Surette ME Anti-CD3 and concanavalin A-induced human T cell proliferation is associated with an increased rate of arachidonate-phospholipid remodeling J Biol Chem 276 20 17568 75 2001 May 18 11278296 10.1074/jbc.M006152200
|
16159396
|
PMC1243241
|
CC BY
|
2021-01-04 16:23:26
|
no
|
Respir Res. 2005 Sep 13; 6(1):103
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-103
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-1051616475510.1186/1465-9921-6-105ResearchInduction of p38- and gC1qR-dependent IL-8 expression in pulmonary fibroblasts by soluble hepatitis C core protein Moorman Jonathan P [email protected] S Matthew [email protected] Deborah C [email protected] Steven A [email protected] David S [email protected] Guha [email protected] Department of Internal Medicine, James H. Quillen College of Medicine, East Tennessee State University, Johnson City, TN, USA2 Medical Service, James H. Quillen VAMC, Johnson City, TN, USA2005 15 9 2005 6 1 105 105 8 4 2005 15 9 2005 Copyright © 2005 Moorman et al; licensee BioMed Central Ltd.2005Moorman et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Recent studies suggest that HCV infection is associated with progressive declines in pulmonary function in patients with underlying pulmonary diseases such as asthma and chronic obstructive pulmonary disease. Few molecular studies have addressed the inflammatory aspects of HCV-associated pulmonary disease. Because IL-8 plays a fundamental role in reactive airway diseases, we examined IL-8 signaling in normal human lung fibroblasts (NHLF) in response to the HCV nucleocapsid core protein, a viral antigen shown to modulate intracellular signaling pathways involved in cell proliferation, apoptosis and inflammation.
Methods
NHLF were treated with HCV core protein and assayed for IL-8 expression, phosphorylation of the p38 MAPK pathway, and for the effect of p38 inhibition.
Results
Our studies demonstrate that soluble HCV core protein induces significant increases in both IL-8 mRNA and protein expression in a dose- and time-dependent manner. Treatment with HCV core led to phosphorylation of p38 MAPK, and expression of IL-8 was dependent upon p38 activation. Using TNFα as a co-stimulant, we observed additive increases in IL-8 expression. HCV core-mediated expression of IL-8 was inhibited by blocking gC1qR, a known receptor for soluble HCV core linked to MAPK signaling.
Conclusion
These studies suggest that HCV core protein can lead to enhanced p38- and gC1qR-dependent IL-8 expression. Such a pro-inflammatory role may contribute to the progressive deterioration in pulmonary function recently recognized in individuals chronically infected with HCV.
==== Body
Background
Hepatitis C virus (HCV), an RNA virus of the Flavivirus family, is the most common blood-borne infection in the United States [1,2]. A striking feature of HCV disease is the high rate of progression to chronicity, with over 80% of acutely infected individuals developing chronic inflammation [3]. This inflammation has been associated with liver failure, hepatocellular carcinoma and autoimmune dysfunction [1]. Treatment for HCV is toxic and of limited efficacy, and the majority of infected individuals do not receive the antiviral therapies available.
Recently, HCV infection has been repeatedly linked to progressive declines in pulmonary function in patients with underlying lung diseases such as asthma and chronic obstructive pulmonary disease (COPD) [4,5]. In patients who already had a diagnosis of COPD, chronic HCV infection led to a more rapid decline in forced expiratory volume (FEV1) and diffusing capacity for carbon monoxide (DLCO), findings that were abrogated in those treated with interferon [4]. In a recent 6-year prospective trial, asthmatic patients with chronic HCV who did not respond to interferon had greater impaired reversibility to bronchodilators when compared to either HCV-negative controls or to HCV-positive individuals who responded to interferon. [5] Some data suggests that HCV infection may alter acetylcholine-mediated airway tone [5]. Other smaller studies also suggest a role for HCV infection in various pulmonary diseases, including idiopathic pulmonary fibrosis [6,7].
While the pathogenesis of the progressive liver disease that occurs with HCV infection involves fibrosis of hepatic tissue in the setting of chronic inflammation, there are few data available that address the inflammatory aspects of HCV infection that lead to declines in lung function. Studies in chronically infected individuals have however demonstrated increased levels of both serum and intrahepatic cytokines, in particular interleukin-8 (IL-8), a chemokine well-known to mediate inflammatory pulmonary processes [8,9]. IL-8 is involved in host inflammatory responses and is synthesized by many different cell types, including fibroblasts and epithelial cells. Expression of IL-8 may inhibit the antiviral activity of interferon γ (IFN) [9] and correlates with the degree of hepatic fibrosis and portal inflammation during HCV infection [10,11]. While IL-8 plays a significant role in pulmonary pathology in general [12], its role in pulmonary disease specifically associated with HCV has not been addressed.
IL-8 signaling is characterized by the integration of at least three different signaling pathways that coordinate induction of mRNA synthesis or that suppress mRNA degradation [13]. Current models suggest that maximal IL-8 can be generated upon de-repression of the gene promoter, activation of NFκB and JNK pathways, and stabilization of the resulting mRNA by p38 MAPK signaling. ERK signaling also contributes to IL-8 induction, although it does not appear to be a potent inducer. TNFα likely activates all of these pathways and has served as a model for robust IL-8 signaling.
Interestingly, we and other investigators have found that the nucleocapsid core protein of HCV may modulate immune signaling pathways, including those mediated by receptors such as gC1qR, TNFR1, and Fas [14-16]. This protein has been found in serum in naked form [17], and soluble core protein can bind and signal extracellularly via the complement receptor, gC1q, on lymphocytes [15]. HCV core appears to be the most potent signal inducer of the IL-8 promoter in hepatocytes transfected with viral protein-reporter expression vectors [18].
We would like to better understand the mechanisms by which chronic HCV infection leads to a more progressive pulmonary decline in individuals with chronic lung disease. Because HCV core antigen can modulate immune signaling pathways that affect IL-8 transcription, we examined the role of soluble HCV core protein in IL-8 signaling in pulmonary fibroblasts. We document an HCV core-induced increase in IL-8 mRNA and protein expression in fibroblasts that is both dose- and time-dependent. We demonstrate that this increase is associated with activation of p38 MAPK that this activation is necessary. We show that co-signaling with TNFα and HCV core leads to augmentation of IL-8 gene and protein expression. Finally, we document that HCV core-mediated IL-8 up-regulation can be inhibited using antibodies that block gC1qR, a known receptor for soluble HCV core antigen.
Materials and methods
Tissue Culture
Normal human lung fibroblasts (NHLF) (Clonetics-BioWhittaker, Walkersville, MD) were grown in fibroblast basal medium (Clonetics-BioWhittaker, Walkersville, MD) at 5% CO2 at 37°C. Media was supplemented with 2% fetal bovine serum, human fibroblast growth factor-B (1.0 μg/mL), insulin (5 mg/mL), gentamicin and amphotericin B. NHLF were cultured in 12 well culture plates at a cell concentration of 5 × 104 cells per well and incubated overnight. β-galactosidase-HCV core antigen (1–191) fusion proteins (core) or β-galactosidase control proteins (β-gal) (endotoxin-negative; Virogen, Watertown, MA) were added at the indicated concentrations based on previous studies of soluble core antigen [15,19] and incubated for 24 hours. Time course experiments were done in the same manner for 12, 24, and 48 hours. Tumor necrosis factor α (TNFα) was added at 1 U/mL with and without HCV core antigen and incubated for 24 hours. Inhibition of p38 MAPK studies were done using the specific inhibitor, SB203580 (Calbiochem, San Diego, CA) as described [20]. NHLF were pretreated with SB203580 (10 μM) for two hours prior to addition of HCV core antigen or control proteins. Murine gC1qR-specific antibodies (Chemicon, Temicula, CA) were used at two different concentrations as described.
Enzyme Linked Immunosorbent Assay (ELISA)
Enzyme linked immunosorbent assay (ELISA) was used to detect IL-8 levels in cell-free supernatants as previously described using commercially available kits (R&D Systems, Minneapolis, MN) [20]. Values were extrapolated or interpolated from a standard curve. Results were analyzed on an ELISA plate reader (Dynatech MR 5000 with supporting software).
IL-8 Gene Expression by RT-PCR
Gene expression for IL-8 was assessed using RT-PCR as previously described [21]. RNA was extracted by a RNAzol technique from cultured cells. Briefly, total cellular RNA was extracted from cultured cells (1 × 106 cells) by the addition of 1.1 mL of RNAzol B (Tel-Test, Inc., Friendswood, Texas). The suspension was shaken for 1 minute and centrifuged at 12,000 × g for 15 minutes at 4°C. The aqueous phase was washed twice with 0.8 ml phenol:chloroform (1:1, v/v), and once with 0.8 mL of chloroform. Each time, the suspension was centrifuged at 12,000 × g for 15 minutes at 4°C. An equal volume of isopropanol was added to the aqueous phase, and the preparation refrigerated at -20°C overnight. After centrifugation at 12,000 × g for 30 minutes at 4°C, the RNA pellet was washed with 75% ethanol. The RNA pellet was air dried and suspended in 20 μl of DEPC-treated water. RNA was quantitated by optical density readings at 260 nm, and the integrity of the 28S and 18S RNA bands determined by electrophoresis in ethidium bromide-stained agarose gels. First strand cDNA was synthesized in the presence of murine leukemia virus reverse transcriptase (2.5 U/μL), 1 mM each of the nucleotides dATP, dCTP, dGTP and dTTP; RNase inhibitor (1 U/μL), 10× PCR buffer (500 mM KCl, 100 mm Tris-HCl, pH 8.3), and MgCl2 (5 mM), using oligo(dT)16 (2.5 mM) as a primer. The preparation was incubated at 42°C for 20 minutes in a DNA thermocycler (Perkin-Elmer Corp., Norwalk, CT) for reverse transcription. PCR amplification was done on aliquots of the cDNA in the presence of MgCl2 (1.8 mM), dNTPs (0.2 mM), and AmpliTaq polymerase (1 U/50 μL), and paired cytokine-specific primers (0.2 nM of each primer) to a total volume of 50 μl. Paired primers for the housekeeping gene hypoxanthine phosphoribosyltransferase (HPRT) were employed as a control for gene expression. PCR consisted of 1 cycle of 95°C for 2 min, 45 cycles of 95°C for 45 sec, 60°C for 45 sec, and 72°C for 1 min 30 sec, and lastly, 1 cycle of 72°C for 10 min. Fourteen microliters of the amplified products were subjected to electrophoresis on a 2% agarose gel stained with ethidium bromide. IL-8 gene products were confirmed by fragment size.
Immunofluorescent Staining
NHLF were cultured on sterile coverslips overnight and subsequently treated with β-gal or core (1 μg/ml) for 30 minutes to three hours. Cells were fixed by immersion in ice-cold methenol:acetone (1:1) for ten minutes at 20°C. Coverslips were air-dried and cells blocked with 1% normal donkey serum (Jackson Laboratories, West Grove, PA) in PBS for thirty minutes. A primary polyclonal rabbit antibody to phosphorylated p38 (Cell Signaling Tech, Beverly, MA) was diluted 1:200 in 1% normal donkey serum/PBS and incubated with cells for 1.5 hr at room temperature. Cells were washed three times using PBS with Tween-20 at five minute intervals. A secondary donkey-anti-rabbit antibody conjugated to Cy tm 3 (Jackson Laboratories, West Grove, PA) was applied at 1:200 and incubated for 45 minutes in the dark. Three washes with PBST were performed at five minute intervals and coverslips were mounted to slides and viewed using an Olympus BX41 fluorescent microscope at 570 nm.
Statistical Analysis
All experiments were done in triplicate. All values are given as the mean ± standard deviation (SD). Statistical analysis was done using the Students t-test and Statistica version 5 computer software (StatSoft, Inc Tulsa, OK). A p-value of < 0.05 was considered significant.
Results
HCV core protein induces IL-8 protein and gene expression
Recent studies have demonstrated that the core protein of HCV can signal extracellularly and that naked core protein is present in the serum of infected individuals [17]. To examine the role of soluble core protein in cytokine signaling in fibroblasts, we employed normal human lung fibroblasts (NHLF), which have been used as a model for cytokine expression. β-galactosidase-HCV core (1–191) fusion proteins or control β-galactosidase (β-gal) proteins were added to NHLF cultures at doses ranging from 1–3 μg/ml and incubated for 24 hours. These commercially available fusion proteins were confirmed to be endotoxin-negative and have been extensively used in studies exploring the role of HCV core in immune modulation [15,22,23]. Culture supernatants were used in an IL-8 ELISA assay (figure 1A) and cells were analyzed for IL-8 mRNA expression at 24 hours using RT-PCR, with HPRT as a control for RNA loading (figure 1B). Cells exposed to HCV core protein demonstrated significant up-regulation of both IL-8 protein and mRNA expression that was not seen in either mock- or β-gal-treated control cells as measured by ELISA. No concentration of β-gal control proteins up to 3 μg/ml elicited any IL-8 gene expression or protein production. In addition, there was no increase in expression of other cytokines, including IL-6, MCP-1, TNFα, and IL-1β, when assayed by ELISA (data not shown).
Figure 1 Dose-dependent increase in IL-8 expression by HCV core protein. A, NHLF were subjected to mock treatment or treatment with β-galactosidase or (β-gal) or β-galactosidase-HCV core protein (Core) at the indicated concentrations, incubated for 24 h, and supernatants assayed for IL-8 by ELISA as described in Materials and Methods. Experiments were done in triplicate. B, NHLF were subjected to mock treatment or treatment with β-galactosidase (β-gal) or β-galactosidase-HCV core protein (Core) at the indicated concentrations and incubated for 24 h. Lysates were harvested, RNA isolated and reverse transcribed, and IL-8 detected using IL-8 specific primers or HPRT as a control for RNA loading as described in Materials and Methods.
To analyze the kinetics of IL-8 induction, IL-8 protein expression was determined at multiple time points following treatment of fibroblasts with HCV core protein (figure 2). Cells treated with HCV core continued to exhibit increasing levels of IL-8 production over 48 hours that were significantly and consistently elevated above mock- or β-gal-treated controls. These data suggested that HCV core significantly induced IL-8 up-regulation leading to enhanced protein expression in a dose- and time-dependent manner.
Figure 2 Time-dependent increase in IL-8 expression by HCV core protein. NHLF were mock-treated or treated with β-galactosidase (β-gal) or β-galactosidase-HCV core protein (Core) at the indicated concentrations. Supernatants were collected at 12, 24, and 48 h and IL-8 expression assayed by ELISA as described in Materials and Methods. All assays were done in triplicate.
Increased IL-8 up-regulation upon co-treatment with HCV core protein and TNFα
Intrahepatic levels of TNFα and IL-8 have been shown to be elevated in individuals with chronic hepatitis C infection [11]. Several investigators have suggested that HCV core protein might modulate or perhaps mimic TNFR1 signaling [24,25] and core protein has been shown to bind the cytoplasmic domain of TNFR1 [16]. Since TNFα is a strong inducer of IL-8 signaling, we wanted to determine how HCV core stimulus interacts with concomitant TNFα signaling. To accomplish this, IL-8 protein and mRNA expression were assayed in NHLF cells co-cultured with HCV core protein and TNFα (figure 3).
Figure 3 Additive increase in TNFα-induced IL-8 expression by HCV core protein. A, NHLF were subjected to mock treatment or treatment with β-galactosidase (β-gal), β-galactosidase-HCV core protein (Core), TNFα, or Core and TNFα at the indicated concentrations and incubated for 24 h. Supernatants were assayed for IL-8 by ELISA as described in Materials and Methods. Experiments were done in triplicate. B, NHLF were treated to the above conditions and incubated for 24 h. Lysates were harvested, RNA isolated and reverse transcribed, and IL-8 detected using IL-8 specific primers or HPRT as a control for RNA loading as described in Materials and Methods.
In these experiments, individual TNFα treatment and HCV core treatments, as expected, led to significant increases in IL-8 induction as measured by both protein (figure 3A) and gene expression (figure 3B). In cells co-treated with both TNFα and HCV core protein, however, there was an additive increase in IL-8 induction beyond what would be expected if core was mimicking TNFα and signaling only through TNFR1. The addition of antibody to TNFα to cells treated with core protein did not inhibit core-mediated IL-8 expression (data not shown).
HCV core-mediated IL-8 secretion is dependent upon p38 phosphorylation
IL-8 signaling is a complex set of events that involves activation of several MAPK members, including most notably NFκB but also variably ERK and JNK. Efficient IL-8 signaling appears to require activation by transcription factors, such as NFκB, as well as stabilization of the resulting mRNA by p38 signaling. Because our data demonstrated a significant induction of IL-8 gene expression and protein production upon HCV core treatment, we wanted to determine if elements of these signaling pathways were being activated.
p38 plays a key role in effective IL-8 responses. [13] To examine the role of p38 signaling in HCV core-mediated responses, NHLF cells were either mock-treated or treated with β-gal or HCV core and incubated for two hours. Cells were harvested, lysed, and analyzed by immunoblotting using antibodies specific for both the native and the phosphorylated forms of p38 (figure 4A). These experiments demonstrated phosphorylation of p38 upon treatment with HCV core protein.
Figure 4 Phosphorylation of p38 in response to HCV core protein. A, Western blot analysis of human fibroblasts. NHLF were subjected to mock treatment (lane 1) or treatment with β-galactosidase (1 μg/ml) (lane 2) or β-galactosidase-HCV core protein (1 μg/ml) (lane 3) and incubated for 2 hours. Whole cell lysates of NHLF were subjected to SDS-PAGE and immunoblotted with antibodies specific for either p38 or phosphorylated p38 (p38-p) as indicated. B, Immunofluorescent staining of NHLF (40×). NHLF cultured on coverslips were subjected to mock treatment or treatment with β-galactosidase (β-gal) or β-galactosidase-HCV core protein (Core) at the indicated concentrations and incubated for 30 min or 3 hr as indicated. Cells were fixed with methanol:acetone (1:1) prior to immunofluorescent staining using an antibody to phosphorylated 38 and a secondary Cy tm3-conjugated antibody. Cells were viewed and photographed using an Olympus BX41 microscope at 570 nm.
To confirm these findings, NHLF cells were again mock-treated or treated with β-gal or HCV core protein and incubated for 30 minutes to three hours. Cells were fixed and immunostained with antibody to phosphorylated forms of p38 and visualized by fluorescent microscopy (figure 4B). We observed nuclear localization of phosphorylated p38 consistent with activation upon treatment with core at 30 minutes. A washout effect was noted, with diminished signaling within the nucleus evident by three hours.
Because HCV core was associated with both increased IL-8 and p38 signaling, we assayed the ability of the p38 inhibitor SB203580 to inhibit overall IL-8 protein expression in these cells (figure 5). NHLF were mock-treated or treated with β-gal or HCV core protein, with or without the addition of SB203580, and incubated for 24 hours. IL-8 protein expression was determined by ELISA. These experiments demonstrated up-regulation of IL-8 protein expression by HCV core that was completely inhibited by the addition of SB203580, and suggested that p38 signaling is necessary for optimal IL-8 expression induced by HCV core protein.
HCV core-induced IL-8 expression is dependent upon gC1qR
Recent investigations into HCV core protein have demonstrated that soluble core interacts with the complement receptor, gC1q and alters intracellular signals including MAPK. Because soluble HCV core affected the p38 MAPK signaling pathway, we wanted to assess whether this was occurring via a gC1qR-dependent process in a manner similar to experiments done in lymphocytes. Expression of gC1qR on pulmonary fibroblasts was confirmed by immunoblotting with antibody to gC1qR (figure 6A). Notably, treatment with β-gal, HCV core protein, or TNFα did not alter receptor expression level.
Figure 5 Inhibition of HCV-core induced IL-8 expression by SB203580. NHLF were either mock-treated or treated with DMSO vehicle, SB203580, β-galactosidase (β-gal), β-galactosidase-HCV core protein (Core), or Core and SB203580 at the indicated concentrations and incubated for 24 hr. IL-8 protein expression was assayed by ELISA as described above.
Figure 6 Inhibition of HCV-core induced IL-8 expression by antibody to gC1qR. A. gC1qR expression in NHLF. NHLF were mock-treated (1) or treated with β-galactosidase (2), β-galactosidase-HCV core protein (3), or β-galactosidase-HCV core protein and TNFα (4) and incubated for 24 h. Whole cell lysates of NHLF were subjected to SDS-PAGE, immunoblotted with antibodies specific for gC1qR, and visualized using ECL. A 33 kD protein was identified. B. Inhibition of HCV core-induced IL-8 expression by gC1qR. NHLF were mock-treated (Mock) or treated with β-galactosidase (β-gal), β-galactosidase-HCV core protein (Core), or β-galactosidase-HCV core protein with either antibody to gC1qR at the indicated concentrations or mouse IgG at 1 μg/ml. Supernatants were collected at 24 h and IL-8 expression assayed by ELISA as described in Materials and Methods. All assays were done in triplicate.
NHLF were subjected to treatment with β-gal, HCV core protein, HCV core protein with anti-gC1qR antibody, or HCV core protein with control isotypic antibody and IL-8 expression was assayed by ELISA (figure 6B). As in our previous experiments, HCV core induced IL-8 expression, and this was not affected by addition of murine IgG antibody. Induction of IL-8 was, however, partially blocked by antibody to gC1qR at two different doses. These data suggest that IL-8 expression induced by HCV core protein is at least partially dependent upon signaling via gC1qR.
Discussion
The association of HCV infection with declines in pulmonary function has only recently been recognized in clinical studies. Our study represents a first attempt to examine the pathogenesis of this process. In this study, we demonstrate that the core nucleocapsid protein of hepatitis C virus induces the up-regulation of a key inflammatory cytokine, IL-8, in pulmonary fibroblasts. Treatment with soluble HCV core antigen led to increases in both IL-8 message and protein that augmented TNFα-induced IL-8 expression. This core-induced up-regulation was associated with p38 activation, which was required for core-induced signaling. IL-8 up-regulation was also dependent upon gC1qR, a known receptor for soluble HCV core.
It is important to note that our studies involved direct extracellular delivery of HCV core protein. The vast majority of studies involving HCV core have employed transfection techniques to deliver HCV proteins intracellularly, with the assumption that this would mimic viral infection of the given cell being studied. Using these techniques, we and other investigators have noted intracellular interactions with various immunomodulatory receptors, including Fas[14], TNFR [16], and LTβR[26], mediated in general through the cytoplasmic domains of those receptors.
Several studies however have now reported that nanomolar amounts of core protein are detected in the circulating blood of HCV-infected patients [17,27,28], and that core protein is secreted from transfected cell lines [29]. This has raised the possibility that HCV core can function extracellularly as a signaling antigen as a means of modulating immune responses. Recent studies focusing on the interaction of soluble core antigen with gC1qR have provided exciting and novel mechanisms by which HCV core might be exerting its immunomodulatory effects [30]. As noted by other investigators, the amounts secreted from transfected cell lines are similar to those employed in our and other investigators' studies of soluble core protein [15,19].
Our findings that HCV core antigen activates MAPK pathways involved in cytokine signaling is supported by multiple previous investigations. In a tetracycline-regulated system used to express HCV core in HepG2 cells, core expression led to activation of ERK, JNK, and p38 pathways and to an increase in cellular proliferation [31]. Similarly, studies have shown that stable transfection of HCV core results in activation of JNK and AP-1 [32]. Upon co-tranfection of HCV proteins and an IL-8 reporter plasmid into mammalian cells, HCV core exhibited the strongest effect on intracellular signaling pathways and activated the IL-8 promoter via NFκB and AP-1 [18]. These studies, however, focused primarily on signaling pathways and promoter activity rather than gene expression per se. Our data demonstrate that activation of at least some of these pathways does ultimately result in detectable gene and protein expression in pulmonary fibroblasts.
It is notable that multiple other HCV gene products have been associated with IL-8 upregulation, including HCV E2 [33], NS4A and 4B [34], and NS5A [9,35]. These studies were performed primarily in hepatocytyes or HeLa cells, but the effect of these gene products on pulmonary fibroblast signaling is yet to be examined. It is certainly feasible that multiple HCV proteins contribute to chemokine upregulation and inflammation in pulmonary fibroblasts, and this possibility should be the focus of future studies.
Our experiments suggest that the ability of HCV core to up-regulate IL-8 expression may be dependent upon p38 phosphorylation. This is perhaps not surprising given the putative role for p38 in stabilizing mRNA following activation of IL-8 at the transcriptional level. In current models of IL-8 signaling, p38 activation is necessary for maximal IL-8 production following stimulation [13]. Our studies would suggest that HCV core provides activation of p38 signaling, which is associated with robust IL-8 up-regulation in multiple cell types [12,36]. We cannot at this point rule out that other MAPKs, such as JNK and perhaps ERK, are not also involved in the HCV core-mediated up-regulation of IL-8. These studies are ongoing in our laboratory.
Soluble core antigen has been shown to inhibit human T cell responses via the complement receptor, gC1qR, which interestingly involved an inhibition of the ERK/MEK MAPK signaling pathway in these T cells [15,19,22]. HCV core can directly and extracellularly bind gC1qR, a phenomenon which is saturable at core concentrations of 3 μg/ml [19]. gC1qR is expressed on pulmonary fibroblasts, and our data suggest that IL-8 up-regulation by soluble HCV core in NHLF is at least partially dependent upon this receptor. It is notable that a similar induction of IL-8 expression via gC1qR was observed in HUVEC endothelial cells and was also mediated through MAPK-dependent processes [37]. Ongoing studies are examining the effect of blocking gC1qR on NFκB signaling and on MAPK signals such as p38, JNK, and ERK.
Notably, IL-8 has been found to be a key mediator of pulmonary inflammation and reactive airway disease [12]. Patients with persistent asthma have an influx of neutrophils and increased local pulmonary IL-8 levels [38]. Because IL-8 has been shown to directly provoke bronchoconstriction [39], it presumably contributes to the establishment of chronic reactive airway disease directly and indirectly by stimulating neutrophil recruitment and activation. Activation of transcription factors or kinase pathways leading to up-regulation of IL-8 expression has been implicated in the pathogenesis of multiple other viral infections, including adenovirus (ERK), CMV (NFκB), and KSHV (p38, JNK) [12].
The pro-inflammatory characteristics of HCV core protein described in our studies may be of key importance to clinical infection. Individuals with chronic HCV infection have evidence of upregulated IL-8 responses, including increased levels of IL-8 mRNA in liver biopsies from infected patients [8,10,11] and increased IL-8 protein detectable in serum compared to controls [8,9](unpublished observations, JPM.) Notably, intrahepatic IL-8 mRNA levels have correlated with both hepatic fibrosis and inflammatory indices and were associated with resistance to interferon therapy [9,11].
It is possible that a similar phenomenon is occurring in pulmonary tissue in response to HCV core and contributes to the declines in pulmonary function associated with active HCV infection [4,5]. In such a model, circulating core antigen would bind to gC1qR displayed on the surface of pulmonary fibroblasts and trigger the phosphorylation/activation of p38, NFκB, and possibly other MAPK mediators. This would lead to enhanced IL-8 gene transcription and protein expression, increased neutrophil recruitment at the local level, and ultimately the deterioration in pulmonary function observed in HCV-infected patients with lung disease [4,5,7]. It is noteworthy that patients with HCV infection even in the absence of pulmonary symptoms have been found to have increased numbers of neutrophils in bronchoalveolar fluid samples [7].
Conclusion
Our studies point to a role for HCV core protein in up-regulating IL-8 mRNA and protein expression in a p38- and gC1qR-dependent manner and support much of the growing body of literature that suggests that HCV core is pro-inflammatory in specific cells. Further dissection of the pathways involved in HCV core-mediated signaling may provide a clearer understanding of the pathogenesis of pulmonary and hepatic disease in HCV-infected individuals and provide targets for modulating its effects.
Abbreviations
HCV: hepatitis C virus
NHLF: normal human lung fibroblasts
IL-8: interleukin 8
MAPK: mitogen-activated protein kinase
BALF: Bronchoalveolar lavage fluid
EBV, Epstein-Barr virus
IFN, interferon
LPS, lipopolysaccharide
CMV: cytomegalovirus
KSHV: Kaposi's sarcoma-associated herpes virus
ELISA, enzyme-linked immunosorbent assay
DLCO: diffusing capacity for carbon monoxide
FEV1: forced expiratory volume at one second
COPD: chronic obstructive pulmonary disease
JNK: jun-kinase
NFκB: nuclear factor kappa B
TNFR: tumor necrosis factor receptor
HPRT: hypoxanthine phosphoribosyltransferase
PMSF: phenylmethylsulfonyl fluoride
PAGE: polyacrylamide gel electrophoresis
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
JPM conceived the study, participated in its design, coordinated all experiments and drafted the manuscript. SMF carried out the IL-8 immunoassays and RT-PCR reactions and helped draft the manuscript. DCP completed all immunofluorescence studies. SAL carried out experiments involving TNFα including immunoassays. DSC participated in study design and data interpretation and provided critical manuscript review. GK participated in the study design, coordinated experiments, and drafted the manuscript in coordination with JPM. All authors read and approved the final manuscript.
Acknowledgements
The authors would like to acknowledge Dr. Donald Hoover for his expertise in fluorescent microscopy. This work was funded by NIH grantsR15 AI-43310 andRO1 HL-63070 (to G.K.), the Chair of Excellence in Medicine (State of Tennessee grant 20233), Cardiovascular Research Institute, and the ResearchDevelopment Committee, East Tennessee State University.
==== Refs
Alter MJ Margolis HS Krawczynski K Judson FN Mares A Alexander WJ Hu PY Miller JK Gerber MA Sampliner RE The natural history of community-acquired hepatitis C in the United States. The Sentinel Counties Chronic non-A, non-B Hepatitis Study Team [see comments] New England Journal of Medicine 1992 327 1899 1905 1280771
Moorman JP Challenges in the HIV patient co-infected with hepatitis C Curr Hep Rep 2002 1 9 15
Moorman JP Joo M Hahn Y Evasion of host immune surveillance by hepatitis C virus: potential roles in viral persistence Arch Immunol Ther Exp (Warsz) 2001 49 189 194 11478392
Kanazawa H Hirata K Yoshikawa J Accelerated decline of lung function in COPD patients with chronic hepatitis C virus infection: a preliminary study based on small numbers of patients Chest 2003 123 596 599 12576384 10.1378/chest.123.2.596
Kanazawa H Yoshikawa J Accelerated decline in lung function and impaired reversibility with salbutamol in asthmatic patients with chronic hepatitis C virus infection: a 6-year follow-up study Am J Med 2004 116 749 752 15144911 10.1016/j.amjmed.2003.12.026
Manganelli P Salaffi F Pesci A [Hepatitis C virus and pulmonary fibrosis] Recenti Prog Med 2002 93 322 326 12050917
Idilman R Cetinkaya H Savas I Aslan N Sak SD Bastemir M Sarioglu M Soykan I Bozdayi M Colantoni A Aydintug O Bahar K Uzunalimoglu O Van Thiel DH Numanoglu N Dokmeci A Bronchoalveolar lavage fluid analysis in individuals with chronic hepatitis C J Med Virol 2002 66 34 39 11748656 10.1002/jmv.2108
Kaplanski G Farnarier C Payan MJ Bongrand P Durand JM Increased levels of soluble adhesion molecules in the serum of patients with hepatitis C. Correlation with cytokine concentrations and liver inflammation and fibrosis Dig Dis Sci 1997 42 2277 2284 9398806 10.1023/A:1018818801824
Polyak SJ Khabar KS Rezeiq M Gretch DR Elevated levels of interleukin-8 in serum are associated with hepatitis C virus infection and resistance to interferon therapy J Virol 2001 75 6209 6211 11390624 10.1128/JVI.75.13.6209-6211.2001
Shimoda K Begum NA Shibuta K Mori M Bonkovsky HL Banner BF Barnard GF Interleukin-8 and hIRH (SDF1-alpha/PBSF) mRNA expression and histological activity index in patients with chronic hepatitis C Hepatology 1998 28 108 115 9657103 10.1002/hep.510280116
Mahmood S Sho M Yasuhara Y Kawanaka M Niiyama G Togawa K Ito T Takahashi N Kinoshita M Yamada G Clinical significance of intrahepatic interleukin-8 in chronic hepatitis C patients Hepatol Res 2002 24 413 419 12479940 10.1016/S1386-6346(02)00136-5
Mukaida N Pathophysiological roles of interleukin-8/CXCL8 in pulmonary diseases Am J Physiol Lung Cell Mol Physiol 2003 284 L566 77 12618418
Hoffmann E Dittrich-Breiholz O Holtmann H Kracht M Multiple control of interleukin-8 gene expression J Leukoc Biol 2002 72 847 855 12429706
Moorman JP Prayther D McVay D Hahn YS Hahn CS The C-terminal region of hepatitis C core protein is required for Fas-ligand independent apoptosis in Jurkat cells by facilitating Fas oligomerization Virology 2003 312 320 329 12919737 10.1016/S0042-6822(03)00208-3
Kittlesen DJ Chianese-Bullock KA Yao ZQ Braciale TJ Hahn YS Interaction between complement receptor gC1qR and hepatitis C virus core protein inhibits T-lymphocyte proliferation J Clin Invest 2000 106 1239 1249 11086025
Zhu N Khoshnan A Schneider R Matsumoto M Dennert G Ware C Lai MM Hepatitis C virus core protein binds to the cytoplasmic domain of tumor necrosis factor (TNF) receptor 1 and enhances TNF-induced apoptosis Journal of Virology 1998 72 3691 3697 9557650
Masalova OV Atanadze SN Samokhvalov EI Petrakova NV Kalinina TI Smirnov VD Khudyakov YE Fields HA Kushch AA Detection of hepatitis C virus core protein circulating within different virus particle populations J Med Virol 1998 55 1 6 9580878 10.1002/(SICI)1096-9071(199805)55:1<1::AID-JMV1>3.0.CO;2-7
Kato N Yoshida H Kioko Ono-Nita S Kato J Goto T Otsuka M Lan K Matsushima K Shiratori Y Omata M Activation of intracellular signaling by hepatitis B and C viruses: C-viral core is the most potent signal inducer Hepatology 2000 32 405 412 10915750 10.1053/jhep.2000.9198
Yao ZQ Eisen-Vandervelde A Waggoner SN Cale EM Hahn YS Direct binding of hepatitis C virus core to gC1qR on CD4+ and CD8+ T cells leads to impaired activation of Lck and Akt J Virol 2004 78 6409 6419 15163734 10.1128/JVI.78.12.6409-6419.2004
Fitzgerald SM Lee SA Hall HK Chi DS Krishnaswamy G Human lung fibroblasts express interleukin-6 in response to signaling after mast cell contact Am J Respir Cell Mol Biol 2004 30 585 593 14565941 10.1165/rcmb.2003-0282OC
Krishnaswamy G Smith JK Mukkamala R Hall K Joyner W Yerra L Chi DS Multifunctional cytokine expression by human coronary endothelium and regulation by monokines and glucocorticoids Microvasc Res 1998 55 189 200 9657919 10.1006/mvre.1998.2079
Yao ZQ Eisen-Vandervelde A Ray S Hahn YS HCV core/gC1qR interaction arrests T cell cycle progression through stabilization of the cell cycle inhibitor p27Kip1 Virology 2003 314 271 282 14517080 10.1016/S0042-6822(03)00419-7
Yao ZQ Nguyen DT Hiotellis AI Hahn YS Hepatitis C virus core protein inhibits human T lymphocyte responses by a complement-dependent regulatory pathway J Immunol 2001 167 5264 5272 11673541
You LR Chen CM Lee YH Hepatitis C virus core protein enhances NF-kappaB signal pathway triggering by lymphotoxin-beta receptor ligand and tumor necrosis factor alpha J Virol 1999 73 1672 1681 9882379
Park KJ Choi SH Koh MS Kim DJ Yie SW Lee SY Hwang SB Hepatitis C virus core protein potentiates c-Jun N-terminal kinase activation through a signaling complex involving TRADD and TRAF2 Virus Res 2001 74 89 98 11226577 10.1016/S0168-1702(00)00251-3
Matsumoto M Hsieh TY Zhu N VanArsdale T Hwang SB Jeng KS Gorbalenya AE Lo SY Ou JH Ware CF Lai MM Hepatitis C virus core protein interacts with the cytoplasmic tail of lymphotoxin-beta receptor Journal of Virology 1997 71 1301 1309 8995654
Maillard P Krawczynski K Nitkiewicz J Bronnert C Sidorkiewicz M Gounon P Dubuisson J Faure G Crainic R Budkowska A Nonenveloped nucleocapsids of hepatitis C virus in the serum of infected patients J Virol 2001 75 8240 8250 11483769 10.1128/JVI.75.17.8240-8250.2001
Widell A Molnegren V Pieksma F Calmann M Peterson J Lee SR Detection of hepatitis C core antigen in serum or plasma as a marker of hepatitis C viraemia in the serological window-phase Transfus Med 2002 12 107 113 11982963 10.1046/j.1365-3148.2002.00359.x
Sabile A Perlemuter G Bono F Kohara K Demaugre F Kohara M Matsuura Y Miyamura T Brechot C Barba G Hepatitis C virus core protein binds to apolipoprotein AII and its secretion is modulated by fibrates Hepatology 1999 30 1064 1076 10498661 10.1002/hep.510300429
Yao ZQ Ray S Eisen-Vandervelde A Waggoner S Hahn YS Hepatitis C virus: immunosuppression by complement regulatory pathway Viral Immunol 2001 14 277 295 11792059 10.1089/08828240152716547
Erhardt A Hassan M Heintges T Haussinger D Hepatitis C virus core protein induces cell proliferation and activates ERK, JNK, and p38 MAP kinases together with the MAP kinase phosphatase MKP-1 in a HepG2 Tet-Off cell line Virology 2002 292 272 284 11878930 10.1006/viro.2001.1227
Shrivastava A Manna SK Ray R Aggarwal BB Ectopic expression of hepatitis C virus core protein differentially regulates nuclear transcription factors Journal of Virology 1998 72 9722 9728 9811706
Balasubramanian A Ganju RK Groopman JE Hepatitis C virus and HIV envelope proteins collaboratively mediate interleukin-8 secretion through activation of p38 MAP kinase and SHP2 in hepatocytes J Biol Chem 2003 278 35755 35766 12824191 10.1074/jbc.M302889200
Kadoya H Nagano-Fujii M Deng L Nakazono N Hotta H Nonstructural proteins 4A and 4B of hepatitis C virus transactivate the interleukin 8 promoter Microbiol Immunol 2005 49 265 273 15782000
Polyak SJ Khabar KS Paschal DM Ezelle HJ Duverlie G Barber GN Levy DE Mukaida N Gretch DR Hepatitis C virus nonstructural 5A protein induces interleukin-8, leading to partial inhibition of the interferon-induced antiviral response J Virol 2001 75 6095 6106 11390611 10.1128/JVI.75.13.6095-6106.2001
Yoshida H Kato N Shiratori Y Otsuka M Maeda S Kato J Omata M Hepatitis C virus core protein activates nuclear factor kappa B-dependent signaling through tumor necrosis factor receptor-associated factor J Biol Chem 2001 276 16399 16405 11278312 10.1074/jbc.M006671200
Xiao S Xu C Jarvis JN C1q-bearing immune complexes induce IL-8 secretion in human umbilical vein endothelial cells (HUVEC) through protein tyrosine kinase- and mitogen-activated protein kinase-dependent mechanisms: evidence that the 126 kD phagocytic C1q receptor mediates immune complex activation of HUVEC Clin Exp Immunol 2001 125 360 367 11531942 10.1046/j.1365-2249.2001.01597.x
Gibson PG Simpson JL Saltos N Heterogeneity of airway inflammation in persistent asthma : evidence of neutrophilic inflammation and increased sputum interleukin-8 Chest 2001 119 1329 1336 11348936 10.1378/chest.119.5.1329
Fujimura M Myou S Nomura M Mizuguchi M Matsuda T Harada A Mukaida N Matsushima K Nonomura A Interleukin-8 inhalation directly provokes bronchoconstriction in guinea pigs Allergy 1999 54 386 391 10371099 10.1034/j.1398-9995.1999.00891.x
|
16164755
|
PMC1243242
|
CC BY
|
2021-01-04 16:23:25
|
no
|
Respir Res. 2005 Sep 15; 6(1):105
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-105
|
oa_comm
|
==== Front
Theor Biol Med ModelTheoretical Biology & Medical Modelling1742-4682BioMed Central London 1742-4682-2-381617429910.1186/1742-4682-2-38ResearchTargeting MDM2 by the small molecule RITA: towards the development of new multi-target drugs against cancer Espinoza-Fonseca L Michel [email protected] Department of Biochemistry, Molecular Biology, and Biophysics, University of Minnesota, Minneapolis 55455, MN, USA2005 20 9 2005 2 38 38 11 7 2005 20 9 2005 Copyright © 2005 Espinoza-Fonseca; licensee BioMed Central Ltd.2005Espinoza-Fonseca; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The use of low-molecular-weight, non-peptidic molecules that disrupt the interaction between the p53 tumor suppressor and its negative regulator MDM2 has provided a promising alternative for the treatment of different types of cancer. Among these compounds, RITA (reactivation of p53 and induction of tumor cell apoptosis) has been shown to be effective in the selective induction of apoptosis, and this effect is due to its binding to the p53 tumor suppressor. Since biological systems are highly dynamic and MDM2 may bind to different regions of p53, new alternatives should be explored. On this basis, the computational "blind docking" approach was employed in this study to see whether RITA would bind to MDM2.
Results
It was observed that RITA binds to the MDM2 p53 transactivation domain-binding cleft. Thus, RITA can be used as a lead compound for designing improved "multi-target" drugs. This novel strategy could provide enormous benefits to enable effective anti-cancer strategies.
Conclusion
This study has demonstrated that a single molecule can target at least two different proteins related to the same disease.
multi-target drugsRITAcancer treatmentblind dockingMDM2p53 tumor suppressor
==== Body
Background
The p53 tumor suppressor is one of the principal mediators of cell-cycle arrest and the activation of apoptosis in response to a broad array of cellular injuries [1-4]. In normal unstressed cells, p53 is regulated by a feedback loop with the negative regulator protein MDM2 (murine double-minute clone 2, referred to as human double-minute clone 2, HDM2, in humans) [1,2,5]. A well-known mechanism for the loss of wild-type p53 activity in cancer cells is the overexpression of MDM2, which leads to constitutive inhibition of p53 and thus allows the tumor cells to escape from p53-induced apoptosis [6].
Recent studies have shown that rescue of p53 function by disruption of the p53-MDM2 interaction may be a promising strategy for developing new anti-cancer drugs [7-9]. To date, different research groups have reported diverse peptidic and non-peptidic molecules that bind at the MDM2-p53 transactivation domain-binding cleft [10-16]. In all cases, these molecules bind to MDM2 and block the p53-MDM2 interaction. In contrast, Issaeva et al. reported the small molecule RITA (reactivation of p53 and induction of tumor cell apoptosis, Figure 1), which binds to p53 and targets it for proteasomal degradation [17]. The most interesting feature of RITA was its ability to increase the p53-dependent antitumor effect in vivo by inducing a conformational change in p53, which prevented MDM2 binding. In principle, targeting MDM2 or p53 should be sufficient to induce apoptosis effectively in cancer cells. However, considering that biological systems are not static, and that proteins present a certain degree of plasticity due to the pre-existence of conformational populations, the traditional single-drug-single-target approach should be replaced by the single-drug-multiple-target approach. By employing the latter, we can obtain benefits from the "promiscuous" behavior of a potential drug by targeting different proteins with a single molecule [18]. Thus, the possibility that RITA binds to both p53 and MDM2 makes it an attractive lead compound for further development of potent and effective anti-cancer drugs.
Figure 1 Chemical structure of RITA [2,5-bis(5-hydroxymethyl-2-thienyl)furan].
In the present study the computational "blind docking" approach [19] is used in order to determine the possibility of RITA binding and its preferential binding sites. It was found not only that RITA can bind efficiently to the MDM2 p53 transactivation domain-binding cleft, but also that is highly specific for its binding site. The results of this study support the effectiveness of the "multi-target" approach in anti-cancer drug design.
Results and discussion
The objective of this study was to demonstrate that RITA, a drug originally found to bind the p53 tumor suppressor, is also able to bind at the MDM2-p53 transactivation domain-binding cleft, which increases its effectiveness and makes it a lead compound for further anti-cancer drug design efforts.
By using the "blind docking" approach, it was found that RITA preferentially binds to the hydrophobic MDM2 p53 transactivation domain-binding cleft. RITA could also bind to other faces of the protein, but this occurred with low frequency. In this case, 81 independent runs out of 100 placed RITA in the MDM2 p53 transactivation domain-binding cleft. The orientation with the most populated cluster is shown in Figure 2. Moreover, "fine docking" focused on the binding cleft showed that 93 out of 100 independent runs accommodated RITA in the same orientation as that observed in the most populated cluster obtained through the "blind docking" procedure. These results imply that RITA is highly specific for the MDM2-p53 transactivation domain-binding cleft. It is also noticeable that RITA covers most of the cleft surface, accommodating horizontally to the cavity and then behaving as a "cap", avoiding p53 to bind to MDM2.
Figure 2 Orientation of the best ranked cluster obtained by using the "blind docking" procedure. RITA is rendered as van der Waals spheres and MDM2 as a surface.
As observed in Figure 3, RITA interacts with the MDM2 as follows: one of the hydroxymethyl-thiophene moieties in the molecule makes contact with residues G58, I61, M62 and Y67, while the other interacts with residues L54, H96, I99, Y100 and I103. Finally, the furan ring makes the only contact with V93. Most of the interactions are favored by van der Waals and hydrophobic interactions. This finding is consistent with the structural composition of both the MDM2-p53 transactivation domain-binding cleft and the thiophene and furan rings, which present large hydrophobic regions.
Figure 3 Perspective of the best raked cluster obtained through the "blind docking" procedure. RITA is shown as van der Waals spheres; residues G58, I61, M62 and Y67 are shown as blue surface; residues L54, H96, I99, Y100 and I103 are shown as lime surface; V93 is shown as red surface.
The calculated binding free energy shows a moderate affinity for MDM2 (Table 1). The stabilizing energy of the complex comes principally from the van der Waals and hydrophobic terms. Subsequently, some hydrogen bonds can be formed between the hydroxyl moieties of the RITA backbone nitrogen, oxygen and hydrogen atoms. However, it is not possible at this stage to determine which residues play an important role in forming such interactions, due to the lack of flexibility of MDM2 during the docking experiments.
Table 1 Calculated free energy of binding of RITA obtained through the blind and fine docking experiments. Kd is the computed dissociation constant, focc is the number of results in the top clusters; Ntot is the number of clusters generated by AutoDock.
Type of docking ΔGb (kcal/mol) Kd (μmol) focc Ntot
Blind -6.36 22 50 9
Fine -6.32 23.3 93 3
As mentioned in the introduction, RITA was previously found to bind to the p53 tumor suppressor, which induces its accumulation in tumor cells and leads to selective apoptosis [17]. It was hypothesized that RITA behaved as an allosteric modulator, inducing conformational changes in p53 and preventing it from binding MDM2 but preserving its functional role. Unfortunately, the full structure of the p53 tumor suppressor is not available, which impedes global screening of the best RITA binding site. Nevertheless, there is enough experimental evidence showing that RITA binds to p53. This work demonstrates that RITA could also bind to the MDM2-p53 transactivation domain-binding cleft.
An important point to consider is that it was shown experimentally that RITA does not bind to MDM2 [17]. Nevertheless it was found in this study that the base structure of the compound (the 2,5-di-thiophen-2-yl-furan moiety) was able to bind to the p53-binding cleft. Preliminary docking studies on different NMR structures of MDM2 showed that RITA can actually bind to the same binding cleft over different MDM2 conformations. This suggests that RITA itself might not be a potent MDM2 inhibitor, but the binding affinity for both MDM2 and p53 might be improved by modifying its structure. The apparent inability of RITA to bind to MDM2 in experiments might be due to the need for high concentrations of this compound to attain effective inhibition. This possibility correlates well with our docking simulations, in which the computed dissociation constant of RITA binding to MDM2 is relatively high.
The experimental Kd obtained for RITA binding to p53 is 1.5 nM, while the computed Kd of RITA binding to MDM2 is 22 μM. This means that RITA binds some ~15000 times more tightly to p53 than to MDM2. Thus, to use RITA as is would be problematic in principle, since much higher concentrations would be necessary to target both p53 and MDM2 efficiently. The physiological consequences are obvious. On this basis, it is proposed in this study that RITA could be markedly improved to increase its effectiveness on MDM2, while retaining its effectiveness on p53. Thus, structural modifications in RITA would alleviate the lack of selectivity for MDM2, making it an effective multi-target drug.
To visualize the approach presented here better, the following factors should be kept in mind: a) biomolecules are not static entities but are constantly involved in dynamic processes; b) MDM2 displays important conformational transitions when binding to different fragments of the N-terminal domain of p53 [20] and can interact with the core domain of the latter [21,22]; c) p53 may mutate in key residues, hindering RITA binding but not altering its physiological role; d) as pointed by Van Regenmortel, the binding site should not be visualized without considering the binding partner [23]; and e) as remarked by Weaver and co-workers, a promiscuous drug candidate is not a collection of different molecules acting in combination on different receptors implicated in the pathogenesis of a disease, but rather a single molecule capable of binding to a range (albeit a limited range) of targets [24]. Thus, improved RITA derivatives could bind not only to p53 but also to MDM2, which would increase or reinforce its therapeutic effect. In other words, the "multi-target" behavior observed by these compounds would compensate their therapeutic deficiencies because of the dynamic nature of the targets involved and the observed promiscuity of MDM2 over p53. Thereby, RITA could serve as a lead compound for designing improved, low-toxicity and highly effective apoptosis inducers via p53 activation and facilitating the effective inhibition of p53 ubiquitination by MDM2. Its effectiveness would then be improved by its action on multiple pathways related to the disease [18].
Currently, additional simulations are being carried out that allow MDM2 to flex, in order to sample the conformational space more thoroughly, and modifications of the RITA structure are in course. These simulations will help to determine the structural keys involved in the molecular recognition mechanism, to modify the structure of RITA and to improve the activity of a new family of potent anti-cancer drugs.
Conclusion
The aim of the present study was to support the viability of the multi-target approach in the design of anti-cancer drugs. For this purpose, a recently described system, the MDM2-p53 complex, was successfully used as a study model.
"Blind" and "fine" docking simulations using the AutoDock program revealed that RITA, a molecule originally investigated for binding to the p53 tumor suppressor, can also bind to the MDM2 p53 transactivation domain-binding cleft. As a multi-target drug acting on several proteins related to the same disease, RITA could be more therapeutically effective in the treatment of some types of cancer. These findings open a new and exciting perspective for effective cancer treatment with low-molecular-weight, non-peptidic molecules.
Structural modifications of RITA may help not only to increase the effectiveness of its binding to MDM2 and p53, but also to elucidate the common structural features of p53 and MDM2 in order to improve the anti-cancer activity of a new family of RITA-derived drugs.
Materials and methods
Protein preparation
The X-ray structure of human MDM2 in complex with the p53 transactivation domain was used in the present study (PDB code: 1YCR). For docking purposes, the p53 fragment was removed from the original PDB file. Hydrogen atoms were added to the protein and the structure was minimized by 500 steps using the conjugate gradient protocol and employing the CHARMM27 force field implemented in NAMD 2.5 software [25]. Subsequently, non-polar hydrogens were merged from the protein and Kollman united atom charges were assigned. Finally, the protein was equipped with fragmental volumes and solvation parameters.
Ligand setup
The structure of RITA [2,5-bis(5-hydroxymethyl-2-thienyl)furan] (Figure 1) was optimized at the AM1 semiempirical level, and the Gasteiger-Marsili formalism [20] was employed to derive the partial charges on the atoms. The AUTOTORS utility, included in the latest version of AutoDock (3.0.5) [26], was used to define the torsions of RITA.
Molecular Docking
Docking simulations were carried out in two stages using version 3.0.5 of the AutoDock program [27]. This program is one of the most reliable docking tools available today because it uses the efficient Lamarckian Genetic algorithm and its scoring function comprises several terms (van der Waals, coulomb potential electrostatics, directional hydrogen bonding, a volume-based solvation term and an estimation of the entropic cost of binding through a weighted sum or torsional degrees of freedom). In addition, the possible binding site need not be specified since the algorithm allows the entire surface of the target to be searched efficiently.
The grid maps representing the protein were calculated with the aid of AutoGrid, a utility of the AutoDock software. Two different grid maps with different dimensions were calculated for the protein. In the first stage, a cubic box of 120 × 120 × 120 points, with a spacing of 0.35 Å between the grid points and centered on the geometric center of the protein, was calculated in order to carry out the "blind docking" experiment. The dimensions of the box were sufficient to cover the entire surface of MDM2. In the second stage, a smaller box (50 × 50 × 50 points, spacing 0.35 Å) was built centered on the most populated binding site, using its geometric center as a reference. This smaller box was employed for performing "fine docking" on the most populated binding site found during the "blind docking" stage. In both cases, docking simulations were carried out using the Lamarckian Genetic Algorithm with an initial population of 300 individuals, a maximum number of 50,000,000 energy evaluations, a maximum number of 50,000 generations, a translation step of 2 Å, a quarternion step of 50° and a torsion step of 50°. For each local search, the pseudo-Solis and Wets algorithm was applied using a maximum number of 300 iterations. Both the blind and refined docking simulations consisted of 100 independent runs. Resulting orientations lying within 2.0 Å in the RMSD were clustered together and represented by the orientation with the most favorable free energy of binding.
Acknowledgements
The author would like to thank David D. Thomas for his hospitality and encouragement, the Minnesota Supercomputing Institute for providing computer resources and Asya Varbanova for her thorough review of the draft and suggested improvements. The author is partially supported by a research fellowship granted by the Department of Biochemisty, Structural Biology and Biophysics-University of Minnesota.
==== Refs
Levine AJ P53, the cellular gatekeeper for growth and division Cell 1997 88 323 331 9039259 10.1016/S0092-8674(00)81871-1
Wu L Levine AJ Differential regulation of the p21/WAF-1 and MDM2 genes after high-dose UV radiation: p53-dependent and p53-independent regulation of the MDM2 gene Mol Med 1997 3 441 451 9260156
Vousden KH Lu X Live or let die: the cell's response to p53 Nat Rev Cancer 2001 2 594 604 12154352 10.1038/nrc864
Vogelstein B Lane D Levine AJ Surfing the p53 network Nature 2000 408 307 310 11099028 10.1038/35042675
Momand J Jung D Wilczynski S Niland J The MDM2 gene amplification database Nucleic Acid Res 1998 26 3453 3459 9671804 10.1093/nar/26.15.3453
Chene P Inhibiting the p53-MDM2 interaction: an important target for cancer therapy Nat Rev Cancer 2003 3 102 109 12563309 10.1038/nrc991
Chene P Inhibition of the p53-MDM2 interaction: Targeting a protein-protein interface Mol Cancer Res 2004 2 20 28 14757842
Chene P Inhibition of the p53-HDM2 interaction with low molecular weight compounds Cell Cycle 2004 3 460 461 14976429
Klein C Vassilev LT Targeting the p53-MDM2 interaction to treat cancer Br J Cancer 2004 91 1415 1319 15452548
Vassilev LT Vu BT Graves B Carvajal D Podlaski F Filipovic Z Kong N Kammlott U Lukacs C Klein C Fotouhi N Liu EA In vivo activation of the p53 pathway by small-molecule antagonists of MDM2 Science 2004 303 844 848 14704432 10.1126/science.1092472
Parks DJ Lafrance LV Calvo RR Milkiewicz KL Gupta V Lattanze J Ramachandren K Carver TE Petrella EC Cummings MD Maguire D Grasberger BL Lu T 1,4-Benzodiazepine-2,5-diones as small molecule antagonists of the HDM2-p53 interaction: discovery and SAR Bioorg Med Chem Lett 2005 15 765 770 15664854 10.1016/j.bmcl.2004.11.009
Grasberger BL Lu T Schubert C Parks DJ Carver TE Koblish HK Cummings MD LaFrance LV Milkiewicz KL Calvo RR Maguire D Lattanze J Franks CF Zhao S Ramachandren K Bylebyl GR Zhang M Manthey CL Petrella EC Pantoliano MW Deckman IC Spurlino JC Maroney AC Tomczuk BE Molloy CJ Bone RF Discovery and co-crystal structure of benzodiazepinedione HDM2 antagonists that activate p53 in cells J Med Chem 2005 48 909 912 15715460 10.1021/jm049137g
Raboisson P Marugan JJ Schubert C Koblish HK Lu T Zhao S Player MR Maroney AC Reed RL Huebert ND Lattanze J Parks DJ Cummings MD Structure based design, synthesis, and biological evaluation of novel 1,4-diazepines as HDM2 antagonists Bioorg Med Chem Lett 2005 15 1857 1861 15780621 10.1016/j.bmcl.2005.02.018
Hardcastle IS Ahmed SU Atkins H Calvert AH Curtin NJ Farnie G Golding BT Griffin RJ Guyenne S Hutton C Kallblad P Kemp SJ Kitching MS Newell DR Norbedo S Northen JS Reid RJ Saravanan K Willems HM Lunec J Isoindolinone based inhibitors of the MDM2-p53 protein-protein interaction Bioorg Med Chem Lett 2005 15 1515 1520 15713419 10.1016/j.bmcl.2004.12.061
Kritzer JA Hodsdon ME Schepartz A Solution structure of a β-peptide ligand for HDM2 J Am Chem Soc 2005 5127 4118 4119 15783163 10.1021/ja042933r
Zhong H Carlson HA Computational studies and peptidomimetic design for the human p53-MDM2 complex Proteins 2005 58 222 234 15505803 10.1002/prot.20275
Issaeva N Bozko P Enge M Protopopova M Verhoef LG Masucci M Pramanik A Selivanova G Small molecule RITA binds to p53, blocks p53-HDM-2 interaction and activates p53 function in tumors Nat Med 2004 10 1321 1328 15558054 10.1038/nm1146
Espinoza-Fonseca LM The benefits of multi-target approach in drug design and discovery Bioorg Med Chem 2005
Hetenyi C van der Spoel D Efficient docking of peptides to proteins without prior knowledge of the binding site Protein Sci 2002 11 1729 1737 12070326 10.1110/ps.0202302
Schon O Friedler A Freund S Fersht AR Binding of p53-derived ligands to MDM2 induces a variety of long-range conformational changes J Mol Biol 2004 336 197 202 14741215 10.1016/j.jmb.2003.11.051
Kussie PH Gorina S Marechal V Elenbaas B Moreau J Levine AJ Pavletich NP Structure of the MDM2 oncoprotein bound to the p53 tumor suppressor transactivation domain Science 1996 274 948 53 8875929 10.1126/science.274.5289.948
Burch LR Midgley CA Currie RA Lane DP Hupp TR Mdm2 binding to a conformationally sensitive domain on p53 can be modulated by RNA FEBS Lett 2000 472 93 98 10781812 10.1016/S0014-5793(00)01427-7
Van Regenmortel MH Molecular recognition in the post-reductionist era J Mol Recognit 1999 12 1 2 10398391 10.1002/(SICI)1099-1352(199901/02)12:1<1::AID-JMR449>3.0.CO;2-P
Stephenson VC Heyding RA Weaver DF The "promiscuous drug concept" with applications to Alzheimer's disease FEBS Lett 2005 579 1338 1342 15733838 10.1016/j.febslet.2005.01.019
Kale L Skeel R Bhandarkar M Brunner R Gursoy A Krawetz N Phillips J Shinozaki A Varadarajan K Schulten K NAMD2: Greater scalability for parallel molecular dynamics J Comput Phys 1999 151 283 10.1006/jcph.1999.6201
Gasteiger J Marsili M Iterative partial equalization of orbital electronegativity – a rapid access to atomic charges Tetrahedron 1980 36 3219 3228 10.1016/0040-4020(80)80168-2
Morris GM Goodsell DS Halliday RS Huey R Hart WE Belew RK Olson AJ Automated docking using Lamarckian genetic algorithm and an empirical binding free energy function J Comput Chem 1998 19 1639 1662 10.1002/(SICI)1096-987X(19981115)19:14<1639::AID-JCC10>3.0.CO;2-B
|
16174299
|
PMC1243243
|
CC BY
|
2021-01-04 16:39:25
|
no
|
Theor Biol Med Model. 2005 Sep 20; 2:38
|
utf-8
|
Theor Biol Med Model
| 2,005 |
10.1186/1742-4682-2-38
|
oa_comm
|
==== Front
Virol JVirology Journal1743-422XBioMed Central London 1743-422X-2-781613892510.1186/1743-422X-2-78ReviewHuman herpesvirus 8 – A novel human pathogen Edelman Daniel C [email protected] University of Maryland Baltimore, School of Medicine, Department of Pathology, 725 West Lombard Street, Rm. S407, Baltimore, Maryland 21201, USA2005 2 9 2005 2 78 78 15 7 2005 2 9 2005 Copyright © 2005 Edelman; licensee BioMed Central Ltd.2005Edelman; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
In 1994, Chang and Moore reported on the latest of the gammaherpesviruses to infect humans, human herpesvirus 8 (HHV-8) [1]. This novel herpesvirus has and continues to present challenges to define its scope of involvement in human disease. In this review, aspects of HHV-8 infection are discussed, such as, the human immune response, viral pathogenesis and transmission, viral disease entities, and the virus's epidemiology with an emphasis on HHV-8 diagnostics.
==== Body
1. The Herpesviruses
1.A. Classification of herpesviruses
More than 100 herpesviruses have been discovered, of which all are double-stranded DNA viruses that can establish latent infections in their respective vertebrate hosts; however, only eight regularly infect humans. The Herpesvirinea family is subdivided into three subfamilies: the Alpha-, Beta-, or Gammaherpesvirinea. This classification was created by the Herpesvirus Study Group of the International Committee on Taxonomy of Viruses using biological properties and it does not rely upon DNA sequence homology. However, researchers have been able to identify and appropriately characterize the viral subfamilies using DNA sequence analysis of the DNA polymerase gene; other investigators have been successful using the glycoprotein B gene [2].
The Alphaherpesvirinea are defined by variable cellular host range, shorter viral reproductive cycle, rapid growth in culture, high cytotoxic effects, and the ability to establish latency in sensory ganglia. In humans, these are termed herpes simplex viruses 1 and 2 (HSV-1 and HSV-2) and varicella zoster virus (VZV), and represent human herpesviruses 1, 2, and 3 [2].
The Betaherpesvirinea have a more restricted host range with a longer reproductive viral cycle and slower growth in culture. Infected cells show cytomegalia (enlargement of the infected cells). Latency is established in secretory glands, lymphoreticular cells, and in tissues such as the kidneys among others. In humans, these are termed human cytomegalovirus (HCMV or herpesvirus 5), human herpesviruses 6A and 6B (HHV-6A and -6B), and human herpesvirus 7 (HHV-7). HHV-7 has also been called the roseolavirus, after the disease roseola infantum it causes in children [2].
The Gammaherpesvirinea have a host range that is found within organisms that are part of the Family or Order of the natural host. In vitro replication of the viruses occurs in lymphoblastoid cells, but some lytic infections occur in epithelial and fibroblasts for some viral species in this subfamily. Gammaherpesviruses are specific for either B or T cells with latent virus found in lymphoid tissues. Only two human Gammaherpesviruses are known, human herpesvirus 4, referred to as Epstein-Barr virus (EBV), and human herpesvirus 8, referred to as HHV-8 or Kaposi's sarcoma-associated herpesvirus (KSHV) [2]. The gammaherpesviruses subfamily contains two genera (a classification of closely related viruses) that includes both the gamma-1 or Lymphocryptovirus (LCV) and the gamma-2 or Rhadinovirus (RDV) virus genera. EBV is the only LCV and HHV-8 is the only RDV discovered in humans. LCV is found only in primates but RDV can be found in both primates and subprimate mammals. RDV DNAs are more diverse across species and are found in a broader range of mammalian species. It is thought that RDVs evolved before LCVs [2].
HHV-8 has sequence homology and genetic structure that is close to another RDV, Herpesvirus saimiri (HVS) [3]. HVS can cause fulminant T-cell lymphoma in its primate host and can immortalize infected T-cells [4]. Rhadinaviruses can infect ungulates, mice, and rabbits and all share a particular genomic organization characterized by large flanking, highly repetitive DNA repeats of high G/C content [5].
1.B. The phenotypic structure of herpesviruses
The phenotypic architecture of the Herpesviridae family viruses characterizes these viruses. Customarily, herpesviruses have a central viral core that contains a linear double stranded DNA. This DNA is in the form of a torus, exemplified by a hole through the middle and the DNA is embedded in a proteinaceous spindle [6]. The capsid is icosadeltahedral (16 surfaces) with 2-fold symmetry and a diameter of 100–120 nm that is partially dependent upon the thickness of the tegument. The capsid has 162 capsomeres. The three dimensional structure of the HHV-8 capsid was determined by cryo-electron microscopy (EM) and was found to be composed of 12 pentons, 150 hexons, and 320 triplexes arranged as expected in the icosadeltahedral lattice with 20 faces; the capsids are 125 nm in diameter [7]. Transmission EM showed a bulls-eye appearance in the virions with electron dense cores and amorphous teguments surrounding the viral core [8]. Interestingly, these structural characteristics were seen in endemic KS lesions as early as 1984, but were not recognized at that time as the possible etiology of the disease [9].
The herpesvirus tegument, an amorphorous proteinaceous material that under EM lacks distinctive features, is found between the capsid and the envelope; it can be asymmetric in distribution. Thickness of the tegument is variable dependent upon its location in the cell and varies between different herpesviruses [10].
The herpesvirus envelope contains viral glycoprotein protrusions on the surface of the virus [2]. As shown by EM there is a trilaminar appearance [11] derived from the cellular membranes [12] and contains some lipid [13]. Glycoproteins protrude from the envelope and are more numerous and shorter than those found on other viruses. The presence of the envelope can influence the size measurement of the virus under EM conditions [2].
1.C. Genomic structure and genes of herpesviruses
There are six defined DNA genomic sequence arrangements for viruses in the Herpesviridae family. Of the human herpesviruses, EBV and HHV-8 are in class C. In this grouping, the number of direct terminal repeats are smaller than for other herpesviruses and there are other repeats found within the genome itself that subdivide the genome into unique stretches [2]. All known herpesviruses have capsid packaging signals at their termini [14].
The majority of herpes genes contain upstream promoter and regulatory sequences, an initiation site followed by a 5' nontranslated leader sequence, the open reading frame (Orf) itself, some 3' nontranslated sequences, and finally, a polyadenylation signal. There are exceptions to this format because initiation from an internal in-frame methionine has been reported [15].
Gene overlaps are common, whereby the promoter sequences of antisense strand (3') genes are located in the coding region of sense strand (5') genes; Orfs can be antisense to one another. Proteins can be embedded within larger coding sequences and yet have different functions. Most genes are not spliced and therefore are without introns and sequences for noncoding RNAs are present [2].
Herpesviruses code for genes that code for proteins involved in establishment of latency, production of DNA, and structural proteins for viral replication, nucleic acid packaging, viral entry, capsid envelopment, for the blocking or modifying host immune defenses, and transitions from latency to lytic growth. Although all herpesviruses establish latency, some (e.g., HSV) do not absolutely require latent protein expression to remain in latency, unlike others (e.g., EBV and HHV-8). Herpesviruses can alter their environment by affecting host cell protein synthesis, host cell DNA replication, immortalizing the host cell, and the host's immune responses (e.g., blocking apoptosis, cell surface MHC I expression, modulation of the interferon pathway) [2].
Gene expression is occurs in two major stages: latency and lytic growth. In the latent phase, there can be replication of circular episomal DNA, and latency typically involves the expression of only a few latently expressed genes. Generally, most host cells infected by herpesviruses exist in a latent phase. When KS tissue or BCBL-1 HHV-8 infected cultured cells are analyzed [8], the vast majority of the infected cells are infected with latent HHV-8 virus. Only a small percent of the cells (≤ 1%) appear to be undergoing lytic replication in a latently infected cell line [16].
The herpesvirus lytic replicative phase can itself be divided into four stages:
1. α or immediate early (IE), which requires no prior viral protein synthesis. In the IE stage, genes involved in trans-activating transcription from other viral genes are expressed.
2. β or early genes (E), whose expression is independent of viral DNA synthesis.
3. Following the E phase, γ1 or partial late genes are expressed in concert with the beginning of viral DNA synthesis.
4. γ2 or late genes, where viral protein expression is totally dependent upon synthesis of viral DNA and where the expression of virion structural genes encoding for capsid proteins and envelope glycoproteins occurs.
1.C.a. Genomic structure and genes of HHV-8
In the viral capsid, HHV-8 DNA is linear and double stranded, but upon infection of the host cell and release from the viral capsid, it circularizes. Reports of the length of the HHV-8 genome have been complicated by its numerous, hard-to-sequence, terminal repeats. Renne et al. [17] reported a length of 170 kilobases (Kb) but Moore et al. [18] suggested a length of 270 Kb after analysis with clamped homogeneous electric field (CHEF) gel electrophoresis. Base pair composition on average across the HHV-8 genome is 59% G/C; however, this content can vary in specific areas across the genome [2]. HHV-8 possesses a long unique region (LUR) at approximately 145 Kb, with at least 87 genes, flanked by terminal repeats (TRs). Varying amounts of TR lengths have been observed in the different virus isolates. These repeats are 801 base pairs in length with 85% G/C content, and have putative packaging and cleavage signals [19]. The LUR is similar to HVS and the HHV-8 genes are named after their HVS counterparts. New genes are still being discovered through transcription experiments with alternative splicing; the initial annotation by Russo et al. [19] was purposely conservative. A "K" prefix denotes no genetic homology to any HVS genes (K1–K15).
HHV-8 possesses approximately 26 core genes, shared and highly conserved across the alpha-, beta-, and gammaherpesviruses. These genes are in seven basic gene blocks, but the order and orientation can differ between subfamilies. These genes include those for gene regulation, nucleotide metabolism, DNA replication, and virion maturation and structure (capsid, tegument, and envelope). HHV-8, being a gammaherpesvirus, encodes more cellular genes than other subfamily viruses. HHV-8 in particular, has a large arrangement of human host gene homologs (at least 12) not shared by other human herpesviruses [19]. These genes seemed to have been acquired from human cellular cDNA as evidenced by the lack of introns. Some retain host function or have been modified to be constitutively active; an example of this is the viral cyclin-D gene [20]. Cellular homologs related to known oncogenes have been identified in HHV-8, including genes encoding viral Bcl-2, cyclin D, interleukin-6, G-protein-coupled receptor, and ribonucleotide reductase [19]. Other genes, such as the chemokine receptor ORF 74, have homologues in other members of the RDV genera [19]. A number of other genes derived from the capsid of HHV-8 have been identified, including Orf 25, Orf 26, and Orf 65 [19]. In addition to virion structural proteins and genes involved in virus replication, HHV-8, typical of a herpesvirus, has genes and regulatory components that interact with the host immune system, presumably as an antidote against cellular host defenses [21].
HHV-8 gene expression has been classified into three stages by current investigators, unlike the four stages of other herpesviruses described above [22]. Class I genes are those that are expressed without the need for chemical induction of the viral lytic phase. Class II genes are induced to increased levels after chemical induction. However, Class III genes, are only expressed after chemical induction.
1.D. The biology of HHV-8
HHV-8 shares four main biological properties with other herpesviruses:
1. A broad array of enzymes involved in nucleic acid metabolism, DNA synthesis, and protein processing.
2. DNA synthesis and capsid formation occur in the nucleus of the host cell and the viral capsid is enveloped at the nuclear membrane.
3. Production of infectious progeny virus in the lytic phase can kill the host cell.
4. The virus can attain a latent state in the host cell with closed circular episomes and a minimal amount of gene expression. Latent genomes, however, can become lytic with the proper stimulation using chemical agents such as sodium butyrate [2].
Several human host cells are permissive for HHV-8 infection. Two prototype cells are the B-cells of the body-cavity-based lymphoma (BCBL) or pleural effusion lymphoma (PEL) [23] and the spindle cells characteristic of Kaposi's sarcoma (KS) [24]. Renne et al. [25] surveyed 38 mammalian cell lines or cell types and was only able to detect by RT-PCR the presence of infectivity from BCBL-1 derived virions in 11 of the 38. However, at least one cell type from lymphoid, endothelial, epithelial, fibroblastoid, and cancer cell types was permissive for infection. The 293 human kidney epithelial cell line was most susceptible in that study [25]. Natural cellular reservoirs for HHV-8 are CD19+ B-cells [26]. Natural infection in other cell types have been reported for endothelium [27], monocytes [28], prostate glandular epithelium [29], dorsal root sensory ganglion cells [30], and spindle cells of KS tumors [27].
Like other rhadinoviruses, HHV-8 might only be pathogenic when other cofactors are involved, such as concurrent infection with HIV or in an immunocompromised host. In the natural healthy host, the virus is relatively benign [5], however, currently, there is no known host other than humans.
1.E. Comparisons of HHV-8 to other herpesviruses
LCV (EBV) and RDV (HHV-8) genomes are more closely related to each other than to the alpha- and betaherpesviruses [18]. HHV-8 does not immortalize B-cells in vitro, as does EBV. HHV-8 has similar large reiterations of the TR as found with EBV but lack EBV's long internal repeats. HHV-8 possesses genes coding for dihydrofolate reductase (DHFR), interferon regulatory factor (IRF), G-protein coupled receptor (GPCR), chemokine analogs, and cyclin-D that are absent from the EBV genome [19]. Fifty-four of 75 HHV-8 genes are collinear with their EBV homologs. Among these 54 genes, the average amino acid identity is 35%. EBV has three forms of viral latency but HHV-8 has only one that has been identified.
1.F. Serodiagnostics of other herpesviruses
I.F.a. Alphaherpesvirinea
HSV infection is optimally detected through direct culture of tissues or secretions with observation of cytopathic effect (CPE) usually occurring in animal embryo cells after 1 – 3 days. Sensitivity of detection of infection is dependent upon the stage of the clinical illness with an average sensitivity of approximately 80%. The shell vial technique, a modified immunofluorescent assay, is also used. VZV grows with more difficulty in culture and it takes 4 to 8 days until CPE is evident, but shell vial techniques can improve the ability to detect VZV infection. Immunofluorescent assay detection (IFA) using monoclonal antibodies (mAb) and using samples taken from the lesions is much quicker than culture methods. However, serology has not been employed conventionally due to the successful culturing techniques. Also, for a successful serological diagnosis, serology requires acute and convalescent samples. Neither culture nor serology has shown optimal sensitivity. Detection of specific glycolsylated proteins can distinguish HSV-1 from HSV-2 infection [2].
I.F.b. Betaherpesvirinea
These viruses (HCMV, HHV-6 & 7) have a more restricted host range than the alpha herpesviruses and exhibit slower growth in culture. They are ubiquitous in the general population but cause serious disease in immunocompromised patients. Diagnosis is difficult due to the absence of clinical disease in healthy persons; virus can be present without pathological effect in humans [2].
Current diagnosis of HCMV is complicated by the intrinsic labiality of the virus and that CPE is not seen in human fibroblast culture cells until after one to three weeks of growth. However, shell vial assays can give results in 24 – 48 hours [2]. The presence of HCMV in peripheral blood is diagnostic for infection even if found in otherwise healthy patients without clinical symptoms. Detection of the HCMV protein, pp65, by an antigen assay is commercially available and can be used for rapid diagnosis of HCMV infection. The pp65 antigen comes from the HCMV lower matrix phosphoprotein customarily found in white blood cells. This antigen test has better sensitivity than culture and can provide positive laboratory results in a few hours. A mAb is used to detect pp65, but the antigen is labile and laboratory tests need to be run within 24 hours of the blood collection [2]. HCMV IgM antibody is diagnostic for HCMV infection in the context of mononucleosis-like disease where the patient is EBV negative. However, acute EBV infection can produce a false positive HCMV IgM test result [31].
For HHV-6 and 7, asymptomatic viral shedding is common in the benign carrier state. Culture of these viruses has been successful with umbilical cord lymphocytes, but there is high background. There are a lack of diagnostic criteria to interpret serologic test results in immunocompromised patients, although the finding of seroconversion in infants is diagnostic [2]. The IFA test using virally infected cells has been commonly used with success [32].
I.F.c. Gammaherpesvirinea and associated antigens
EBV replicates in vivo in lymphoid and epithelial cells and can be cultured in immortalized umbilical cord lymphocytes; EBV antigen is found within the cells. Serology is used for diagnosis of infectious mononucleosis (IM) by detecting IgM heterophile antibodies that agglutinate with red blood cells of horses. Serologic assays can also measure antibodies to the EBV viral capsid antigen (VCA) that is composed of four different proteins, the early antigens (EA) of which there are five proteins, and the nuclear antigens (NA). Testing for IgM against VCA defines acute infection and corresponds to clinical sequelae but lasts only a few months; however, IgG remains for the life of the patient [33]. Anti-EA antibodies arise within a few weeks but are not detectable in all patients with mononucleosis [33]. Anti-NA antibodies arise after the advent of EA antibodies and persist for life [33]. In contrast to acute infection, serology is not useful for post-transplant lymphoproliferative disorder (PTLD) and antigen detection or detection by PCR of viral nucleic acids is required [2]. Antibody production might be compromised due to the host's immunocompromised state or the rapid growth of the polyclonal tumor prior to reactivation of the memory immune response. Antigenic cross reactivity between EBV and other human herpesviruses is rare [2]. This is demonstrated in one study of 42 patients with nasopharyngeal carcinoma, known to be associated with EBV and of all persons positive for EBV VCA, only two showed reactivity to HHV-8 lytic proteins [34].
The humoral antibody response to EBV infection is against four serologically defined antigens [2]:
1. Epstein – Barr virus NA (EBNA) in latently infected cells.
2. EA either in its diffuse (methanol resistant) or restricted (methanol sensitive) compartments, expressed early in the viral lytic cycle.
3. VCA found during the late lytic cycle.
4. Membrane antigen (MA; gp350) as part of the viral envelope and is found on the surface of cells in the lytic phase. Anti-MA antibody levels correlate well with neutralization of the virus.
These EBV antigens are composites of several distinct proteins; e.g. EBNA = EBNA 1, 2, 3A, 3B, 3C. LP and EBNA1 are the most antigenic. The detection of EBV in IM is based upon the use of an enzyme-linked immunosorbant assay (ELISA) to detect IgM specific to BALF2 and BMRF1, the EA antigens, or against VCA components BFRF3 and BLRF2; combinations of these antigens are still recommended [35,36]. Diagnostics of HHV-8 will be discussed at length in Section 8, HHV-8 Diagnostics.
2. HHV-8 Immune Responses and Infectivity
As a prelude to the discussion about HHV-8 immune responses, antibody responses in primary EBV infection are presented as a contrasting system. Upon the appearance of clinical symptoms after EBV infection, most patients have rising IgM antibody titers to VCA and EA; IgA titers are transient [37]. The IgM anti-VCA response disappears over the next few months but the IgG titer falls to a steady state after previously peaking. In comparison, anti-EA IgG titers fall faster and can disappear entirely [2]. Many patients show an EBNA2 IgG response during the acute phase, but an EBNA1 IgG response usually does not appear until convalescence [38]. This delayed EBNA1 response is probably not due to the delay in immune recognition of the latently infected cells or of the released latent antigen because EBNA2 is recognized shortly after infection. Possibly EBNA1 is expressed at a later time point in the virus's life cycle. Latent membrane protein-1 (LMP-1) and LMP-2 antibody responses are rare [39].
Anti-gp350 or membrane antigen (MA) IgM antibodies are neutralizing with the IgG response arising only much later in the infection. These neutralizing antibody (nAb) titers tend to reach a plateau and stay at that level for long periods of time [37]. IgG, IgM and IgA levels are elevated universally in the human host upon EBV infection due to the general activation of B-cells [2]. In addition, heterophile antibodies and autoantibodies, mostly of the IgM class, show a transient increase in titer during acute infection.
In persistent EBV infection, healthy infected individuals are consistently anti-VCA IgG, anti-MA neutralizing antibody positive, and anti-EBNA1 positive. Titers can vary greatly among individuals, but these differences are consistently relative over time [2]. It is unknown why different antibody responses exist for EBV infection.
In general, after herpesvirus infection, some patients present with IgM levels that can be transient or at a low level for varying periods. These can last for up to a year making it difficult to gauge recent infection based upon IgM reactivity alone. In addition, IgM can be detected in viral reactivations [2]. An example of this is found with VZV, which shows an IgM response upon reactivation [40].
2.A. The neutralizing antibody immune response to HHV-8
Neutralizing antibodies are part of the humoral defense system against viral infection. The presence of nAb has been detected by searching for the effect of inhibition by nAb against HHV-8 viral infection in transformed dermal microvascular endothelial cells [41]. By quantifying the level of viral infection by indirect immunofluorescence assay (IFA), inhibition of infection was determined by comparing the level of infection in cells obtained with HHV-8 seropositive sera as compared to the level shown by incubation with seronegative sera. When the seropositive sera was diluted at 1:10 or 1:50 there was significant inhibition compared to the seronegative controls (P = 0.036). However, at a 1:500 dilution, the inhibitory effects of the sera disappeared. The nAb were found in the IgG fraction as shown by depletion of IgG antibody with protein A, which reversed the inhibitory effect.
Similarly, the presence and effect of nAb in the context of HHV-8 infection were investigated by measuring the infectivity in the 293 culture cell line [42]. Kimball et al. also discovered that the nAb were found in the IgG fraction and that compliment was not required for the neutralization. Importantly, their study found that those patients with KS had significantly lower nAb titers than other groups, independent of their HIV status. This suggested a possible role for nAb in the prevention of progression from latent asymptomatic HHV-8 infection to KS disease. They state that the positive effects of nAb were independent of CD4+ counts.
In contrast to these two reports, Inoue et al. observed the effects of nAb action, but concluded that nAb do not affect the progression to KS [43]. These antibodies were found in both KS+ and KS- groups with prevalences of 24% and 31%, respectively, but there was no significance in the difference (P = 0.64). This conflicting finding could perhaps be explained by the specific cohorts used. Other possibilities are the use by Inoue et al. of a colorimetric reporter system and their choice of cutoff at 30% neutralization; where as Kimball et al. used 50% inhibition as the cut off [42]. Additional discussion of HHV-8 antibody responses can be found in Sections 7 and 8.
2.B. Cytologic immune responses to HHV-8
Cell mediated immunology studies of HHV-8 have indicated that there are specific cytotoxic T-lymphocyte (CTL) responses against the virus. In an investigation of five cases of HIV negative subjects that seroconverted to HHV-8, Wang et al. explored the CD8+ T-cell response to five HHV-8 lytic proteins and found that CD8+ T-cells are involved in the control of primary HHV-8 infection [44]. They found that there were no major changes in the numbers of T-cell phenotypes or activation of T-cells, which differed from primary EBV infection that usually produces global increases in the numbers of T-cells. There was also no suppressive effect on other T-cell specificities as seen with EBV infection. They observed distinct CD8+, HLA class I restricted responses and increases in the interferon-gamma (IFN-γ) response to at least three of the five lytic antigens in each of the five subjects. No antigen was dominant in the elicited T-cell response. They observed that HHV-8 antibody titers to lytic IFA proteins paralleled the cytolytic responses. The CD8+ reactivity declined after several years possibly because of the lack of stimulation; the normal biology of HHV-8 is to enter a more latent state after primary infection. More T-cells produced a response of INF-γ production as opposed to CTL precursor production, but neither response was as strong as that observed when the T-cells were challenged with the HCMV pp65 antigenic protein. Osman et al. investigated HLA class I restricted CTL activity directed against the HHV-8 K8.1 lytic antigen [45]. They also investigated an additional lytic protein (K1) and one latent protein (K12) as antigens. Chromium release assays showed that CTL reactivity was detected against all three proteins, but not every patient had reactivity to all three antigens. Specific HLA alleles were able to present more than one of the viral proteins; e.g., HLA B8 could present all three antigens. Most patients with KS and were HIV+ did not have CTL responses indicative of compromised cellular immune systems. In one patient, whose KS had resolved under HAART therapy, CTL activity was restored. In general, these investigators showed that higher titers against HHV-8 LANA1 (Orf 73), i.e., more severe KS, correlated with less CTL response.
In a study of seroconversions in Amsterdam, Goudsmit et al. found that CD4+ T-cell levels did not affect the rate of seroconversions, but once HHV-8 infection had occurred, a decline in CD4+ cells was associated with increasing reactivity against the Orf 65 antigen [46]. Similar findings have been reported by Kimball et al. where persons with KS have higher levels of anti-HHV-8 antibodies and lower CD4+ counts than those without KS, but where both populations have HIV infection [42]. This suggests that viral replication had increased in the context of a more limited CD4 response. Recent investigation [47] has shown that NK cell function is important for the control of latent HHV-8 infection and abrogation of this important immune response can lead to more progressive KS disease.
2.C. Reactivation of HHV-8 infectivity
Using peripheral blood mononuclear cells (PBMCs) culled from KS patients and grown in culture, Monini et al. showed that reactivation of HHV-8 required at least the inflammatory cytokine (IC) INF-γ [48]. They observed that both B-cells and monocytes latently infected with HHV-8 responded to this IC with induction of lytic replication. They proposed that increases in HHV-8 viral load are due to the reactivation of the virus after exposure to INF-γ. They also proposed that a likely scenario of KS pathogenesis is the recruitment of circulating monocytes into peripheral skin tissues, where upon exposure to ICs, their latent HHV-8 genomes enter into the lytic phase. The monocytes then rupture and free virus is available to infect local tissues. The monocytes might also differentiate into macrophages or spindle cells after exposure to the ICs and form the basis of latent HHV-8 infection in the tissues.
Reactivation is possible in the context of autologous peripheral blood stem cell transplantation. Luppi et al. [49] presented a case report that showed HHV-8 viral load in the serum of the transplant patient concomitant with fever, rash, diarrhea, and hepatitis some 17 days after the transplant. The patient had lytic antibodies before and after the transplant indicating a reactivation event.
2.D. Corporeal sites of HHV-8 infection
A number of studies [49-56] have investigated by molecular methods the presence of HHV-8 virions, as evidenced by the presence of viral DNA in body fluids and tissues of several at-risk populations (Table 1). PBMCs were the most commonly studied sample site, but a number of others, including serum or plasma, semen, saliva, and stool have been investigated (Table 1). PCR sensitivities were below 100 copies, although some studies used nested PCR [52] or Southern blotting [50].
Table 1 Compilation of select studies investigating the molecular presence of HHV-8 in different tissues and body fluids. KS, HIV+, and HIV- represent three populations at high, medium, and lower risk of HHV-8 infection, respectively.
KS Lesion Normal Skin PBMC Plasma or Sera Semen Saliva Feces Other
KS+ 63/70 (90%) 17/57 (30%) 94/188 (50%) 33/151 (22%) 7/60 (12%) 26/71 (37%) 0/29
HIV+ 0/10 22/268 (8.2%) 5/164 (3.0%) 4/57 (7%) 9/87 (10%) 10/228 (4.4)
HIV- 0/1 3/381 (0.8%) 0/218 3/168 (1.8%) 7/108 (6.5%) 10/332(3.0)
At least four investigators used the K330 PCR as originally developed by Chang et al. [1]. Five articles described testing KS patients [50-52,54,55] and another five [50-52,55,56] compared HIV+ and HIV- subjects for the presence of HHV-8. Grandadam et al. [53] investigated multicentric Castleman's disease (MCD) in HIV+ patients and Luppi et al. [49] followed the unique case of a viral reactivation. For persons with KS, significant differences were found between sample sites; the HHV-8 prevalence was higher in KS lesions over that found in peripheral blood mononuclear cells (PBMCs), which were about equal in prevalence to saliva (Table 1). These three sites were better for finding the presence of HHV-8 rather than using plasma (P <10-6; P = 0.054; P ≤ 0.02, respectively). For HIV+ persons, saliva and PBMCs were equivalent (P = 0.539) but both had a significant greater frequency of positive samples than were found in plasma (P = 0.016 and P = 0.031, respectively). Analysis of HIV- persons showed that saliva contained significantly more viral sequences than either PBMCs or plasma (P = 0.001 and P = 0.0006, respectively), which were commensurate with each other (P = 0.476).
It is noteworthy to add that several authors have observed the detectable presence of HHV-8 DNA to be intermittent [49,51,57,58]. Perhaps this has contributed to the overall lack of sensitivity of PCR in detecting HHV-8 infection. In keeping with this observation, Simpson et al. [59] stated, "...KSHV genomes were detected in peripheral blood monocyte DNA from KS patients less frequently than antibodies to either KSHV antigen in serum". Smith et al. [60] added that, "Overall, our serologic assay appeared more sensitive than PCR analysis of PBMC for the detection of HHV-8 infection". This last statement was reiterated by other authors (e.g. Angeloni et al. [61], Campbell et al. [62]). HHV-8 viremia is described at more length in Section 8, HHV-8 Diagnostics.
3. Pathogenic Mechanisms of HHV-8
The diversity of the HHV-8 genes allows the virus to assault and modulate its human host with many strategies. These pathogenic effects can promote active changes in the infected human host, such as to increase cytokine production or to suppress MHC Class I (MHC I) presentation of viral proteins to the immune system. The pathogenic activities that are due to HHV-8's unique K-series genes are summarized.
Interleukin-6 (IL-6) is a B-cell growth factor and its altered expression has been linked to several human diseases and malignancies, including MCD with its characteristic plasmacytosis and hypergammaglobulinemia. HHV-8 viral cytokine vIL-6 is encoded by the unique K2 gene, which exhibits 25% amino acid identity with the human homologue [63]. This viral gene is unique to HHV-8 among the other gammaherpesviruses and is the only HHV-8 encoded cytokine. It is a Class II transcript in that it is constitutively expressed in the BCP-1 cell line, but its expression is greatly increased after induction with TPA; it is a Class III transcript in the BC-1 cell line [63]. This feature of the protein implies that its pathogenic effects can be in the context of active viral infection. vIL-6 had activity on human myeloma cells [64], where exogenous application induced DNA synthesis and proliferation in the INA-6 myeloma cell line; this cell line is strictly dependent upon exogenous IL-6 for growth. Expression of vIL-6 mRNA transcripts was detected by in situ hybridization in tissue samples of KS, PEL, and MCD disease patients [65], demonstrating the in vivo expression of this cytokine. Staskus et al. showed that vIL-6 might be important in the pathogenesis of these three HHV-8 associated disorders, but the viral cytokine is variably expressed in the HHV-8 infected cells of these diseases [65]. For example, the number of vIL-6 copies in KS, PEL, and MCD cells was 10–100, 100–1000, and >1000 copies, respectively, per cell. Low levels of vIL-6 have also been observed in KS lesions by immunohistochemistry [63,66].
Several HHV-8 K-genes are active in modulating the adaptive immune response to HHV-8 infection. The K3 and K5 genes allow HHV-8 to evade detection by removing MHC I from the cell surface [21]. The proteins encoded by K3 and K5, MIR-1 and MIR-2, respectively, use a unique mechanism of enhanced endocytosis of the MHC I molecules and their subsequent degradation in lysosomes. MIR-2 protein also down regulates ICAM-1 and B7.2, accessory proteins necessary for proper T-cell stimulation [67].
The lack of MHC I on the cell surface can signal increased natural killer (NK) cell activity, but NK cells are modulated by the K13 gene product, v-FLICE inhibitory protein (vFLIP) [68]. Despite the Fas-dependent signaling (apoptosis triggering) caused by the NK cells, apoptosis is impaired because vFLIP binds to cellular procaspase-8 preventing its proteolytic cleavage into apoptotically active forms.
Another tactic to alter the cell-mediated response to HHV-8 infection is to make sure this response does not occur upon infection. HHV-8 creates a microenvironment where by there is preferential recruitment of T cell type 2 (Th2) lymphocytes with the release of IL-4 and IL-5 cytokines, which polarizes the immune response towards an antibody predominant immune reaction [69]. It is the Th1 response with the characteristic release of Inf-γ that stimulates cell-mediated immunity. Three HHV-8 chemokines, vCCL1, vCCL2, and vCCL3, also referred to as vMIP-1, vMIP-II, and vMIP-III, respectively, are encoded by the K6, K4, and K4.1 Orfs, respectively [70]. These chemokines activate Th2 responses through the CCR8, CCR3, and CCR4 receptors [70], respectively, but are antagonistic for the receptors that result in chemotaxis of Th1 and NK lymphocytes [71]. The vCCL3 is found in KS tumors and is thought to contribute to its pathogenesis [72]. Another HHV-8 gene, K14, encodes a neural cell adhesion-like protein (OX-2) that also promotes Th2 polarization and the production of inflammatory cytokines, such as IL-6 [73]. Other unique K-genes modify the immune system by interacting with the μ-chains of B-cell receptors and blocking transport to the cell surface (K1 or KIS) or by inhibiting interferon signaling (K9 or vIRF-1) [70]. The diverse repertoire of immune suppressive strategies exhibited by HHV-8 could explain the virus's success in establishing a high prevalence in populations where it is being actively transmitted, such as sub-Saharan Africa. However, it then brings into question why HHV-8 is not more successful in establishing infection in developed counties, even with people whose immune systems are compromised or constantly stimulated.
4. Transmission of HHV-8
Patterns of transmission for HHV-8 are being better defined as our understanding of the pathogenesis of this virus increases and testing methods are used strategically. The virus, first thought to be transmitted only sexually, is now also considered transmissible through low risk or more casual behaviors.
4.A. Sexual Transmission
The transmission of HHV-8 through sexual activities has been documented [74]; men with homosexual behaviors showed a 38% prevalence of HHV-8 as compared to 0% of men with no such activity. The increased prevalence correlated with the presence of sexually transmitted diseases (STD) and the number of male sexual partners. The presence of both HIV and HHV-8 produced a 10-year probability of 50% for developing KS [74].
Transmission from male genital secretions, specifically semen, is unlikely due to the low prevalence of detectable HHV-8 in semen samples obtained from both HIV+ or HIV- persons [52,55,56]. In a study of women with KS from Zimbabwe, between 28% and 37% had detectable HHV-8 DNA in their vaginal or cervical samples [75], but HHV-8 DNA was not found in any of the women without KS, even those with HHV-8 seropositivity. A possible explanation why perinatal transmission is infrequent in prevalence studies might be that transmission is limited to immunocompromised mothers where titers might be higher [75].
HHV-8 DNA is found most frequently and with increased viral burdens in saliva or other oral samples [56]. Sexual practices that include oral sex could therefore increase the possibility of transmission. Persons having STDs, such as syphilis and HIV, have an increased risk for greater HHV-8 prevalence [76]. However, in a study of 1,295 women in four USA cities, Cannon et al. did not find an association between the number of sex partners or engagement in commercial sexual practices to be a risk for increased HHV-8 prevalence [76].
4.B. Blood-borne transmission
Identification of HHV-8 in blood donors [58,77] has raised concern about the safety of the blood supply. Other reports [78] have tempered the concern of blood borne transmission after observing no transmission in 18 recipients of HHV-8 seropositive blood components. However, because of the small sample size, additional studies are required for this low prevalence population. In a multicenter study of 1,000 blood donors, approximately 3% of blood donors were considered seropositive, but none of the 138 total seropositive samples had detectable HHV-8 DNA in their PBMCs [79]. Without detectable virus, the possibility of infectious transmission seems remote.
However, blood-borne transmission seems to occur, but rarely. Two epidemiological markers for blood borne viral infection, HCV positivity and daily-injected drug use, were associated with increased HHV-8 infection in four large groups of women in the USA [76]. However, the overall prevalence of HBV and HCV among irregular drug users was higher than found with HHV-8, indicating a lower relative frequency of transmission of this herpesvirus.
Evidence that HHV-8 can be transmitted in populations of intravenous drug users (IVDU) and those HCV+, shows that transmission via blood is possible, albeit with difficulty [80]. Larger studies are required to determine if HHV-8 is a true threat to the blood supply. Such studies will be difficult to conduct due to the difficulty in detecting infectious virus in healthy individuals, the lack of culture methods to tests for cytopathic effect, and the anonymous nature of blood donations, which does not allow for follow up testing.
Important risk factors for transmission of the virus are a spouse's seropositivity and maternal seropositivity [81]. Although spousal seropositivity could include sexual transmission, transmission to children precludes this route, indicating more casual transmission is possible. Horizontal asexual transmission within families has been observed by other investigators [82]. Vertical transmission from mother to child at or before birth is also infrequent with few children from HHV-8 infected mothers showing HHV-8 sequences in their PBMCs at birth [83,84]. In a study of the presence of HHV-8 DNA in matched pairs of breast milk and saliva from the same mother, no HHV-8 sequences were found in the breast milk, but 29% of the saliva samples had HHV-8 DNA; therefore nursing of infants appears unlikely to be a route of infection [85], although, another study seemed to contradict this finding [86].
Of all anatomic sites, HHV-8 DNA is found most frequently in saliva, which also has higher viral concentrations than other secretions [56]. For this reason, it has been hypothesized that saliva could be the route of casual transfer of infectious virus among family members. It has been hypothesized that customarily licking an insect bite, such as from a mosquito, could transfer the virus [87].
4.C. Transplants
4.C.a. Organ
Transmission of other herpesviruses (e.g., HCMV and HHV-6) has been documented [88] and the body of evidence is growing that HHV-8 disease after organ transplantation is a concern for the transplant physician. Most reports in the literature have presented data describing the prevalence and the possible ramifications of HHV-8 infection on donor kidney recipients.
However, the concern of HHV-8 transmission in the context of organ transplantation has two problems. First, there are no large studies of the donor's and the recipient's HHV-8 serostatus and presence of HHV-8 in donor blood and organ. Properly done, both antibody prevalence and a determination of infectious virus by PCR would be necessary. Follow up measuring possible seroreactivity every few months after transplant would be critical. Second, even once the problem is defined, there are no current establish procedures or parameters to monitor the patients both diagnostically and clinically; seemingly, both problems would have to be addressed in tandem.
In areas where endemic KS is not found and in normally healthy people, HHV-8 infection has not been shown to be a life threatening infection. However, in the context of immunosuppression, as with organ transplants, both primary infection and reactivation become a proven concern. Post-transplant immunosuppression can cause iatrogenic KS to appear [89]. The clinical significance of post-transplant KS can be rejection of the graft and death of the patient. In a study of 356 post-transplant patients with KS, 40% had visceral involvement, a manifestation of KS with poor prognosis, and 17% of those with visceral KS died from the tumor [89]. The KS tumor can recede after withdrawal of immunosuppressive therapy, but with immunological recovery, graft loss or organ impairment often emerges as a unwanted condition [89]. In an early study, Parravicini et al. [90] suggest that post-transplant KS is caused by emergence of latent HHV-8 after previously infected but clinically well transplant patients are immunosuppressed. Immunosuppression, such that occurs in transplant recipients, is known to facilitate reactivation of herpesviruses, (e.g., disseminated herpes zoster) and is associated with an increased incidence of herpesvirus associated lymphoproliferative malignancies [91].
Of importance, seroprevalence to HHV-8 increased from 6.4% to 17.7% overall one year after renal transplantation. In addition, seroconversion to HHV-8 occurred within the first year after renal transplantation in 25 of 220 patients and KS developed in two of the 25 within 26 months after transplantation [92]. KS developed within 20 months in two renal transplant recipients from the same cadaveric donor; Orf 73 genotyping confirmed that the virus was transmitted from the donor [93]. Detection of HHV-8 in the allograft kidneys or increases in antibody titer can be prognostic indicators of increased risk for KS [94]. Other studies have found the median time to KS from transplantation to be between 7 months [90] and 24 months [95].
In another study, the increased risk of acquiring HHV-8 infection was shown by 10% of 100 transplant patients who seroconverted to HHV-8, however, there was no pattern associated with the type of organ donated, and none of the donors that could be tested were seropositive [96]. Therefore the investigators concluded that the infection came from sources other than the transplanted organ; however this conclusion is lacking because healthy infected individuals (i.e., healthy organ donors) in the USA are less likely to exhibit antibodies, similar to blood donors, however, the organ might still harbor infectious virus or KS precursor cells [93,94].
In a comparison of kidney and liver transplants, seroconversion was observed in 12% of transplant patients, combined. The incidence of KS in kidney patients was higher than in liver recipients [97]. Importantly, patients already infected with HHV-8 had a greater chance to develop KS from viral reactivation than from primary infections [97]. In a large study of solid organ transplant recipients in Spain (n = 1,328), Munoz et al. [95] reported that the overall KS incidence was 1 in 200 with more males diagnosed with KS than females (6:1 ratio). High HHV-8 antibody titers or seroconversions were prognostic indicators of possible KS development.
Because increased prevalence in transplant patients might be due to reactivation of HHV-8 and the subsequent increase of antibody tiers [98], molecular methods, although normally less sensitive, would be better indicators of transmission. Another possibility would be the use of antibody avidity assays to detect highly avid antibodies that would be indicative of reactivation events [99].
Post-transplant KS can develop in the recipient from transmission of the virus from the donor to the recipient [93,94], and from KS progenitor cells seeded along with the donor organ, which undergo neoplastic change, and progress into KS [100]. HHV-8 DNA can be detected in the KS lesions from patients suffering from post-transplant cutaneous and visceral KS. Other organs without evidence of KS involvement can test positive for HHV-8 sequences [101], as can circulating spindle cells infected with HHV-8 [102]. Disease entities associated with HHV-8 in the context of transplantation continue to be discovered. In at least one report, investigators have suggested that EBV-negative post-transplant lymphoproliferative disorders (PTLD) might be caused by HHV-8 [103].
4.C.b. Bone marrow/Peripheral blood stem cell
Non-neoplastic disease associated with HHV-8 has been documented [49,104]. Bone marrow failure was observed after a kidney transplant and after an autologous peripheral blood stem cell (PBSC) transplant for non-Hodgkin's lymphoma (NHL). HHV-8 produced a syndrome of fever, marrow aplasia and plasmacytosis; these occurred after primary infection and reactivation, respectively [104]. Neither patient presented with KS, but both had detectable HHV-8 sequences by PCR after transplantation and at the presentation of symptoms – both patients died. Another case report [49] showed reactivation of HHV-8 in a seropositive patient and documented nonmalignant disease 17 days after PBSC transplantation in the context of NHL. The patient presented with fever, cutaneous rash, diarrhea, and hepatitis; here too HHV-8 DNA was detected in the serum by PCR with higher viral loads with exacerbation of symptoms. Therefore, transplant patients who are HHV-8 positive could benefit from close clinical follow-up to preempt the occurrence of KS with judicious use of immune suppressive therapy or antiviral drugs, or to begin the early and therefore more effective treatment of the tumor once detected.
5. Diseases of HHV-8
HHV-8 poses challenging questions of diagnosis and pathology related to its role in the etiology of several human malignancies including KS, MCD, PEL, and possibly multiple myeloma (MM) and sarcoidosis, among others.
5.A. Primary infection
Identification of HHV-8 primary infection has been difficult due to the low incidence of infection in most populations studied, and because of the lack of known defining features. By using a diagnosis of exclusion and the temporal occurrence of symptoms and diagnostic criteria, limited studies have suggested several defining clinical sequelae of HHV-8 primary infection. In 15-year longitudinal study of >100 HIV negative men to study the natural history of primary HHV-8 infection, five cases of HHV-8 seroconversion were identified [44]. The effects of HHV-8 primary infection were explored in the absence of HIV coinfection and no debilitating disease was observed in the five seroconverters. Four patients exhibited clinical symptoms, which ranged from mild lymphadenopathy and diarrhea to fatigue and localized rash. These symptoms were significantly associated with HHV-8 seroconversion when compared to the 102 seronegative subjects who remained well.
Organ transplantation is another clinical setting for primary infection. In a patient receiving a renal transplant, bone marrow failure was associated with a syndrome of fever, marrow aplasia, and plasmacytosis [104]. The patient did not present with KS, but HHV-8 sequences were detected by PCR after transplantation and at the presentation of symptoms; the patient did not survive. This limited experience suggests that in the context of immunosuppression, primary infection can be lethal, but in healthy individuals, the infection presents with flu-like symptoms.
5.B. Kaposi's sarcoma
KS was first described by Moritz Kaposi in the 1870s [105] and was described as an aggressive tumor affecting patients younger than those currently observed. For all epidemiological forms of KS, the tumor presents as highly vascularized neoplasm that can be polyclonal, oligoclonal, or monoclonal. It's antigenic profile suggests either endothelial, lympho-endothelial, or macrophage origins [106]. Although the four epidemiological forms of KS have different clinical parameters, such as anatomic involvement and aggressiveness of the clinical course, they have HHV-8 infection in common with indistinguishable histopathology [107]. It is therefore believed that this transforming virus is the causative agent of KS and that HHV-8 fulfills Hill's criteria for causing KS [108,109].
HIV infection substantially increases the risk for development of KS, and therefore, the incidence of KS has increased substantially during the HIV pandemic, particularly in younger HIV-infected patients [110]. Striking differences in risk for acquiring AIDS-KS exist between different HIV transmission groups, varying from a high of 21% for homosexual men to a low of 1% for men with hemophilia. Women who acquired HIV infection by heterosexual contact with bisexual men were also at an increased risk for developing AIDS-KS [110]. Although the incidence of KS has decreased recently with the advent of highly active anti-retroviral (HAART) therapy, the appearance of drug resistant strains of HIV raises concern for a re-emergence of KS cases.
Browning et al., using a cell culture detection method, observed that the characteristic spindle cells of KS are present in the peripheral blood of patients presenting with KS; more importantly, these cells were found in the blood of HIV+ homosexual men, who are at higher risk for developing KS, than HIV+ IVDUs [102].
The first strong evidence that human herpes virus 8 (HHV-8) was the etiological agent of KS came from the use of a novel molecular technique, representational difference analysis (RDA) [1]. This complex molecular method identified viral molecular sequences in KS tumor tissue that were not present in paired normal tissue from the same individual [1]. The presence of nucleic acid sequences of the virus in tissues from all forms of KS [111] throughout the world, and the demonstration of antibodies to HHV-8 in KS patients from a number of serologic studies [112] has supported the association of this virus with KS. Because of its prominent association with KS, the virus is often referred to as Kaposi's sarcoma-associated herpesvirus or KSHV.
Proof of HHV-8's etiology in KS comes from the detection of HHV-8 nucleic acids in KS tissues but not in healthy tissues, from sero-epidemiological and molecular studies showing correlations to the risk of developing KS and progression of KS disease. The detection of antibodies to lytic HHV-8 antigens can be used as a predictor of development of KS [113]. Prospective studies of persons who subsequently developed KS, documented the appearance of infection more than 24 months prior to tumor development [114]. Data have shown that infection of primary endothelial cells with HHV-8 causes long term proliferation and transformation [115]. HHV-8 is detectable in the spindle cells of all forms of KS and in the nearby in situ endothelial cells [27].
5.B.a. Classic KS
The classical or sporadic form of KS (CKS) is an indolent tumor affecting the elderly, preferentially men, in Mediterranean countries such as Italy, Israel, and Turkey [116]. The lesions tend to be found in the lower extremities and the disease, due to its non-aggressive course, usually does not kill those afflicted. HIV infection, unlike HHV-8, is not typically associated with CKS [117].
The older the age of the patient, the greater the risk of CKS disease progression; dissemination of KS lesions is more likely if immunosuppression also exists [118]. Certain behaviors, such as corticosteroid use and infrequent bathing were found to be risk factors for greater incidence of CKS but surprisingly, increased cigarette smoking actually lowered the risk [119]. The increased prevalence in Sardina of HHV-8 and CKS among family members of KS patients indicates that transmission of HHV-8 is probably by asexual routes [61].
5.B.b. AIDS-KS
In the context of the acquired immunodeficiency syndrome (AIDS), KS is the most common malignancy and is an AIDS defining illness [120]. AIDS-KS is a more aggressive tumor than CKS and can disseminate into the viscera with a greater likelihood of death [121]. Unlike CKS, it presents more often multifocally and more frequently on the upper body and head regions [117].
In those with HIV infection, HHV-8 prevalence increases with higher risk of KS, and in patients with HHV-8 seroconversion there is a greater likelihood of KS development [74]. KS was more likely to develop when HHV-8 seroconversion occurred after the patient already had HIV [122,123]. An increased slope of CD4+ cell decline and higher HIV viral loads also suggested increased chances of KS development [122].
However, HIV infection alone might not be enough to increase the risk of KS. In a study of Ethiopians who had immigrated to Israel, only 0.85% of them with AIDS developed KS, as compared with 12.5% of non-Ethiopian AIDS patients (P < 0.001). The low risk of KS exists in the face of high HHV-8 prevalence (above 39%) in HIV+ and HIV- Ethiopian populations [124]. Clearly, other factors are necessary for KS development and ethnic or genetic protective factors might be involved.
5.B.c. Endemic KS
HHV-8 was prevalent in Africa prior to the HIV epidemic, and therefore, was responsible for the large prevalence of KS seen on the continent before HIV changed the scope of KS presentations [125]. Prior to HIV coinfections, endemic KS affected men with an average age of 35 and very young children [126]. In Africa, endemic KS is found more often in women and children than in other areas of the world [125]. It presents in four clinical forms with one form similar to CKS, but found in younger adults; the other three forms are more aggressive, similar to AIDS-KS [117]. They vary in the age of presentation and the sites of involvement.
HIV coinfection has raised the prevalence of KS significantly in Africa. In Uganda, for example, prior to 1970, KS was diagnosed in no more than 7% of the male cancer population and in none of the female cancer population. However, by 1991, KS prevalence had risen to 49% in male cancer patients and to 18% in females [126]. The KS prevalence has increased in Africa, even in HIV negative populations, for unknown reasons [125].
Despite different clinical KS presentations, all forms of KS are associated with HHV-8 infection [111,127]. Paralleling the endemic KS pattern in children, HHV-8 infection in children is also high with seroprevalence reaching adult levels by the age of 20 and in certain locations even earlier [128]. This occurrence of horizontal infection in the young is similar to that seen with EBV in other continents [128]. Despite equal prevalences of HHV-8 in HIV-1 and HIV-2 patients, KS is found almost exclusively in persons infected with HIV-1 [129].
5.B.d. Iatrogenic KS
More extensive information on transplant-associated KS and the involvement of HHV-8 can be found in the Literature Review: Section 4, Transmission of HHV-8. Briefly, iatrogenic KS can present either as a chronic condition or with a more rapid course [117]. Immunosuppression, such that occurs in transplant recipients, is known to facilitate reactivation of herpesviruses [91] and so too with HHV-8, transplant patients under immunosuppressive therapy can present with KS. Withdrawal of the therapy can cause the KS to regress [117].
Iatrogenic KS seems to vary in its geographic prevalences, perhaps reflecting the varying HHV-8 prevalence in the general populations of different countries [125]. KS appears most frequently in renal transplant patients [116] and in conjunction with cyclosporine treatment, used frequently in kidney transplant patients as an immunosuppressive drug; this steroid has been shown to reactivate HHV-8 in vitro [130].
5.C. Primary effusion lymphoma
First identified as a subset of body-cavity-based lymphomas (BCBL), PELs contain HHV-8 DNA sequences [23]. These lymphomas are distinct from malignancies that cause other body cavity effusions. PELs are characterized by several pathological features: 1) They do not exhibit Burkitt lymphoma-like morphology and do not have c-myc gene rearrangements; 2) They have a distinctive morphology comparable to large-cell immunoblastic lymphoma and anaplastic large-cell lymphoma; 3) They occur frequently in men; 4) They present initially as a lymphomatous effusion and remain localized to the body cavity of origin; 5) They express CD45 with frequent absence of B-cell associated antigens; 6) They exhibit clonal immunoglobulin gene rearrangements; 7) They can contain Epstein-Barr virus; 8) They lack oncogene rearrangements in genes such as bcl-2 and p53. Finally, patients with PELs, especially in the context of AIDS, invariably are infected with HHV-8 [23,131]. PEL cell lines have 50–150 copies of HHV-8 episomes per cell [8,132-136].
Divining the association of PELs with HHV-8's etiology has been difficult, because most PELs occur in the context of HIV infection, and the PELs account for only 0.13% of all AIDS malignancies in AIDS patients in the USA [137]. Importantly, PELs occur with an increased frequency in patients with prior KS [125]. In non-AIDS patients, the disease has been termed "classic" PEL by Ascoli et al. [138] where it presents in HIV negative patients, but with similar risk factors as CKS.
5.D. Multicentric Castleman's disease
HHV-8 has been found variably in association with MCD. MCD is a rare polyclonal B-cell angiolymphoproliferative disorder for which vascular proliferation has been found in germinal centers. It presents in heterogeneous forms both clinically and morphologically [139]. However, most of the B-cells in the tumor are not infected with HHV-8, and the HHV-8 infected cells are primarily located in the mantel zone of the follicle [140]. It is thought that uninfected cells are recruited into the tumor through HHV-8 paracrine mechanisms, such as vIL-6 [66], a known growth factor for the tumor. More than 90% of AIDS patients with MCD are HHV-8 positive, whereas MCD in the context of no HIV infection has a HHV-8 prevalence of approximately 40% [141]. Because of it rarity, MCD is difficult to closely associate statistically with HHV-8.
5.E. Other diseases
5.E.a. Sarcoidosis
Sarcoidosis is a multisystemic granulomatous disease of unknown etiology that can involve many different organs such as the lungs, lymph nodes, and skin. Currently, a diagnosis can be established when clinical and radiological findings are confirmed by histological tests showing noncaseous granulomas in more than one tissue [142].
Di Alberti et al. reported that HHV-8 DNA was significantly more prevalent in pulmonary tissues, lymph nodes, skin and oral tissues in 17 Italian patients with sarcoidosis than in tissues from 96 control specimens [143]. However, a study by Belec et al. did not detect HHV-8 sequences in sarcoid tissues from French patients with systemic sarcoidosis [144]. Very little diagnostic HHV-8 serology has been reported on sarcoid patients. In one report, 18% of patients were seropositive, but the investigators concluded that this was not different from the observed prevalences in the patients' respective geographic regions [145].
5.E.b. Multiple myeloma
There is debate concerning the etiology of MM. MM is the most common lymphoid cancer found in Blacks and the second most common in Caucasians [146]. It is a B cell malignancy of clonal origin in which the cancer cells, considered to be plasma cells, secrete monotypic immunoglobin. The pathogenesis of MM has been thought to include an initial antigenic stimulus of B cells followed by further mutagenic events. Studies have shown that autocrine and paracrine loops involving cytokines such as IL-6 [147], TNF, and IL-1β [148] are important as stimuli for growth of the MM cells. It has been believed that T cells and the bone marrow stroma are the sources of these cytokines. Three oncogenes have been implicated in MM; ras, c-myc, and p53 with prevalences of 30%, 25%, and 15–45%, respectively [146].
The possible role of HHV-8 in MM has been debated and a full report of the evidence is beyond the scope of this review. In brief, Rettig et al. [149] who originally reported that there was an association between the virus and the disease, investigated 15 MM patients along with eight patients presenting with monoclonal gammopathy of unknown significance (MGUS). They used PCR to amplify the KS330233 sequence of HHV-8 from bone marrow (BM) mononuclear and stromal cells of the MM patients. Southern blotting of the PCR fragments using an internal fragment confirmed the PCR results. They were able to amplify HHV-8 sequences from cultured BM stromal cells from 15/15 MM patients. However, none of the 23 non-cultured BM mononuclear preparations amplified. Said et al. [150] supported Rettig et al.'s claim that MM and HHV-8 were closely associated by finding 17 out of 20 BM biopsies from MM patients exhibiting HHV-8 positive cells. Gao et al. [151], provided important supportive serological evidence; of 27 MM patients, 81% and 52% possessed lytic and latent antibodies, respectively. All eleven patients with progressive MM were HHV-8 positive. The increased presence of lytic antibodies as opposed to latent antibodies was indicative of past or currently active viral infection in the MM patients.
Contrary to these findings, other groups have found a lack of supporting evidence. Whitby et al. [152] found latent antibodies in only 4/37 MM and in only 2/36 MGUS patients, but these prevalences were not significantly different from patients with Hodgkin's lymphoma, NHL, or normal blood donors. Additionally, whereas Rettig et al. postulated that MGUS might be the precursor of MM through infection with HHV-8 [149], Whitby and colleagues found that 4 persons with MGUS who developed MM were HHV-8 negative, in contrast to two patients with antibodies to HHV-8 who had not exhibited MM symptoms after 36 and 48 months. MacKenzie et al. [153] and Parravicini et al. [154] found only 2/78 and 1/20 MM patients to be seropositive to latent antigen, respectively. The presence of lytic antibodies in MM patients has also been difficult to find by other investigators. Utilizing recombinant ORF 65 antigen in ELISA and Western blot formats, MacKenzie et al. [153] and Parravicini et al. [154] found lytic antibodies in only 2/78 and 1/20 MM patients, respectively. Masood et al. [155] using a lytic IFA and a whole virus lysate ELISA found that only 2/28 MM sera were positive. Perhaps as the pathogenesis of HHV-8 becomes better understood this etiological question will be answered.
5.E.c. Other diseases
Although there are many reports for other diseases and their possible associations with HHV-8, the data are sometimes circumstantial and weak, and many have not been confirmed by extensive investigation in large numbers of patients. Only a few selected diseases or conditions variably associated with HHV-8 are summarized below.
Bone marrow failure is a non-neoplastic disease possibly associated with HHV-8 observed after kidney and autologous peripheral blood stem cell transplants. HHV-8 produced a syndrome of fever, marrow aplasia and plasmacytosis; these occurred after primary infection and reactivation, respectively [104]. Neither patient presented with KS, but both had HHV-8 sequences detected by PCR after transplantation and at the presentation of symptoms.
HHV-8 infection has been associated with congestive heart failure in both KS and PEL patients [138]. Serological evidence has also indicated that Italian patients with cardiovascular disease have a higher prevalence of HHV-8 and HHV-8 DNA has been found in atheromatous plaques [156]. Other studies have suggested possible associations with HHV-8 and pemphigus vulgaris and pemphigus foliaceus [157] and germinotropic lymphoproliferative disorder [158], but not primary central nervous system lymphomas [159].
5.F. Treatment of HHV-8 infection
No single treatment has been found to be completely efficacious for HHV-8 infection. Anti-herpetic drugs such as foscarnet, ganciclovir, cidofovir, and acyclovir inhibit the viral DNA polymerase [107] which, therefore, only allows treatment for replicating viruses in the lytic phase of infection; latent viruses are unaffected. For example, although cidofovir was effective in vitro against BCBL-1 cells [160], intralesional injections were not helpful in reducing the KS tumor burden [161].
Chemotherapy and/radiotherapy are successful treatments for CKS but HHV-8 DNA has been shown to remain at the site of the healed lesion [162]. This might explain the observed reoccurrences of CKS. Treatment for AIDS-KS has centered on HAART. Studies have shown marked decreases in the incidence of AIDS-KS since the use of HAART [163]. However, this reduced risk has been only with triple therapy, and not double or single anti-HIV drug therapy [163]. Additionally, HAART seems to have the best effect on early stage AIDS-KS [164,165]; nonetheless, an 81% reduction in death due to AIDS-KS was observed though HAART [164].
Finally, because HHV-8 can be transferred from organ donor to recipient, the possibility exists that CTLs derived from the donor can be harnessed to provide immunotherapy for the recipient [100]. This has been shown to be an effective treatment for PTLD in the context of EBV reactivation after bone marrow donation [166].
6. HHV-8 Epidemiology
6.A. Serologic prevalence of HHV-8 geographically and in major risk groups
The serologic prevalence of HHV-8 infection has been explored in most continents worldwide and in different populations at different levels of risk of HHV-8 infection. It should be noted that the comparisons of prevalence are limited by whether antibodies to latent or lytic HHV-8 antigens were detected and the test formats used.
6.A.a. North America
Studies from populations from the North American continent have revealed large differences in HHV-8 prevalence between specific populations. Blood donors (BD) have been found to exhibit different levels of infection ranging from no detected infection [167] to as high as 15% [168], with more intermediate levels (~5%) found in most studies [34,59,79]. Individuals infected with HIV infection or having AIDS had more elevated prevalences of 30%–48% [34,74,167], although one study found no evidence of HHV-8 infection in their small HIV cohort [167]. Homosexual men showed prevalences ranging from 20%–38% [74,169,170]. In contrast, the highest prevalences, between 88% and 100%, were found in those patients with KS [34,79,167]. Other miscellaneous populations, such as healthy individuals, the elderly, and those infected with EBV showed a range of 0%–8.6% [74,167,171]. IVDUs had relatively higher prevalences of 10% in both heterosexual men and women; the longer the patient's injected drug use, the higher was the risk of HHV-8 infection, which was not dependent upon sexual behavior or demographic differences [169]. Of note is the exceptionally high level of infection found in children in south Texas, 26% [172]. One report from Quebec, Canada, did not find evidence of HHV-8 infection in 150 renal transplant patients [173].
6.A.b. The Caribbean and Central America
The prevalence of HHV-8 in BDs from Jamaica, Trinidad, and Cuba was 3.6%, 1.2%, and 1.2%, respectively [34,174,175]. Persons with HIV infection from Trinidad, Honduras, and Cuba possessed prevalences of infection at 0%, 24%, and 21%, respectively [34,175,176]. Compared to other studies in KS patients, a relatively low prevalence of HHV-8 infection was found in AIDS KS samples from Cuba (78%) [175]. A very low level of infection was found in attendees of a gynecology clinic in Jamaica (0.7%) [174], but an elevated prevalence was seen in healthy individuals in Honduras (11%) [176]. Commercial sex workers in Honduras showed 19% infection [176].
6.A.c. South America
Evidence of HHV-8 infection has been discovered in South America in at least four countries. In indigenous populations, those without specific risk factors, prevalences of 53% were found among Brazilian Amerindians [177], 16% in northern Brazil [178], and 36% in Amerindians of Ecuador [179]; the prevalences in Ecuador ranged from 20%–100% depending upon the tribe tested [179]. The HHV-8 prevalence was much less in BDs in Brazil (2.8%), Chile (3.0%), and Argentina (4.0%); although in Argentina the prevalence in BDs ranged between 2.4% – 4.3% in three different locales [180]. In contrast, Sosa et al. [181] reported that in Argentinean HIV+ IVDUs, 17.4% showed HHV-8 seropositivity; where as, in HIV negative IVDUs the prevalence was lower at 11.1%. Still lower, HIV negative heterosexuals with no IVDU behavior had a prevalence of 5.7%, similar to that found by Perez et al. [180]. AIDS-KS patients in Brazil had a prevalence of 80% [182].
6.A.d. Europe
In Europe, excluding Italy and its surrounding islands, the prevalence of HHV-8 in BDs was not above 6.5% in six countries: Hungary 0.83%–1.6%, Switzerland 5%, the United Kingdom 1.7%, France 2%, Spain 6.5%, and Germany 3% [59,83,183-186]. In healthy individuals in Switzerland, Greece, and Albania, evidence of HHV-8 infection was 13%, 12%, and 20% [59,184,187]. Persons infected with HIV ranged from a low of 16% in women in Germany to a high of 31% in homosexuals in the United Kingdom [59,184,186]. Homosexuals in Spain however, had an 87% prevalence [185]. IVDUs and persons with STDs in the United Kingdom, Spain, and France showed prevalences of 3.2%–8.4%, 12%–17%, and 13%, respectively [59,185,188]. Similar to North America, the HHV-8 prevalence in patients with classic or endemic KS was 75%, 94%, and 100% in Hungary, Greece, and France, respectively [59,83,183]. The HHV-8 prevalence in AIDS-KS patients in Switzerland (92%), the United Kingdom (81%), France (80%), and Germany (100%) were similar to the prevalence of HHV-8 in classic KS in Europe [59,83,184,186]. IVDUs in the United Kingdom and Spain had prevalences of 0.0%–3.2% and 12%, respectively [59,185].
6.A.e. Italy/Sardinia/Malta/Sicily
Estimations of seroprevalence in Italian BDs were confounded by the variable geographic prevalences and the type of antibodies being detected. Whitby et al. [189] showed that the overall prevalence in 747 BDs in Italy was 14%. However, when these individuals were segregated by North/Central Italy and Southern Italy, the levels of HHV-8 infectivity dispersed to 7.3% and 24.6%, respectively. Even in Rome, centrally located in the country, the prevalence in BDs varied from 2% of people with latent antibodies to 28% with reactivity to lytic antigens [190]. Other reports found prevalences in BDs to be between 3.5% to 18.7% [167,191,192]. In the general population of Sardinia [61], Sicily [193], and for the elderly in Malta [194], antibodies to HHV-8 were found in 11%, 20%, and as high as 54%, respectively. In Italy, those infected with HIV showed a 14% prevalence for latent antibodies, but as high as 61% for lytic antibodies [190]; an intermediate rate (25%) in HIV+ persons was observed by Calabro et al. [192]. In Sicily, 34.6% of HIV+ patients had HHV-8 infection [193]. In regards to other STDs, infections with syphilis were accompanied by HHV-8 infections with 37%–76% showing coinfection, whereas those free from syphilis infection only showed 11%–46% prevalence [190]. No significant differences were seen in persons with or without HCV infections, 10%–50% and 16%–47%, respectively [190]. Perna et al. suggested that the relatively low prevalence of HHV-8 in drug addicts in Sicily (16.6%) was indicative of the poor transmission of HHV-8 parenteraly [193]. Calabro et al. [192] observed 61.5% prevalence in HIV+ homosexuals in Italy, but this rate might have been confounded by the coinfection of HIV because Perna et al. found a lower rate in homosexual men, 32.6% in Sicily [193]. Even healthy adults in Sicily had an elevated prevalence beyond that found in BDs with 36.2% observed with HHV-8 infection [193]. For this central region of the Mediterranean, the prevalence of HHV-8 in CKS normally exceeded that of AIDS-KS. CKS in Italy and Sardinia showed evidence of infection in 95%–100% of patients. However, AIDS-KS were reported to have a much wider range of reactivity in HHV-8 tests: 71%–79% [167], 57.1%–100% [191], 67%–83% [190], and 100% [192] in Italy, and 100% in Sicily [193].
6.A.f. Middle East
Healthy individuals in Israel were found to have a HHV-8 seroprevalence of 4.8% [195], whereas individuals with HBV infection seemed to be at an increased risk of infection (22% prevalence) [81]. Family members from these hepatitis patients also had increased prevalence of HHV-8 at 9.9% [81]. When Ethiopian immigrants to Israel were tested for antibodies against HHV-8, this unique cohort possessed an elevated presence of antibodies against HHV-8 [124]. Fifty seven percent of Ethiopians with HIV infection showed HHV-8 infection, whereas those without HIV had a lower prevalence of 39.1% (P = 0.03). Interestingly, despite the high prevalence of HHV-8 in the HIV+ individuals, in those with AIDS, the occurrence of KS was almost nonexistent (0.85%) compared to non-Ethiopian immigrants with AIDS (12.5%) [124]. Reports on HHV-8 prevalence from Egypt are scarce. Andreoni et al. showed data that in teenagers and young adults, 29% possessed lytic antibodies against HHV-8, but only 5% had latent antibodies [191].
6.A.g. Asia – Southeast and Asia proper
Blood donors and healthy individuals in five Asian countries have shown a 3-fold range in HHV-8 prevalence. In healthy Indian individuals [34], only 3.7% had antibodies, with Thailand, Malaysia, and Sri Lanka exhibiting prevalences no higher than 4.4% [34]. In Taiwan, lytic antibodies were found in 11.7% and 13% of the blood donors tested [196,197]. However, a much higher presence of prior infection was found in the general population of the Uygur people in northwestern China, 47% [198]. The prevalence of infection in HIV positive individuals in Asia varied widely, as well. Prevalences of HHV-8 infection of 0.6% to 11.2%, 2.4%, and 40% where found in Thailand, India, and Taiwan, respectively [34,197,199]. Classical KS still had the highest rate of infection, with 83% of patients in Taiwan [197] and 100% in China [198] showing positivity for HHV-8 antibodies.
6.A.h. The Pacific region
There have been few studies on the seroprevalence of HVV-8 antibodies in the Pacific region. Despite this, the viral infection has been found in both Japan and New Guinea [200,201]. Fujii et al. [200] found a very low prevalence of HHV-8 infection in Japan in BDs where only 0.2% showed reactivity to latent antigen. Comparatively, persons with HIV infection had an elevated prevalence of between 9.8% and 11.6%. In New Guinea, Rezza et al. found a much higher prevalence in the indigenous general population with approximately 25% of the 150 people tested showing prior infection [201].
6.A.i. Sub-Saharan Africa
In sub-Saharan Africa, the seroprevalence of HHV-8 was above 36% in every population reported. In the southern part of the continent, healthy individuals showed a HHV-8 prevalence of 37.5% in Zambia [34], and 54.7%–90% in Botswana, depending upon the test used [179,194]. In Zambia, the HHV-8 prevalence was comparable for HIV+ persons (44%) [34] and 51.1% in HIV+ pregnant women [202]. Cancer patients, in general, in South Africa also had a high prevalence of 36.3% [203]. In comparison, patients with AIDS-KS exhibited a prevalence of 83% in South Africa [203] and 92.3% in Zambia [202].
Central African nations also had HHV-8 prevalences in keeping with those observed in the south. In the Congo, a high prevalence in healthy individuals, 69%–79%, showed prior HHV-8 infection [194]. Somewhat lower percentages were found in healthy individuals in Ghana (41.9%) [34], in Uganda (38.7%) [34], 51%–62% [167]), and 55.5% in Cameroonian pregnant women [83]. Similar HHV-8 prevalences were found for HIV+ persons in Uganda with between 45.7% and 71% HHV-8 prevalence reported depending upon the study and the test used [34,59,167]. The prevalence of HHV-8 infection in AIDS KS patients was relatively higher but did not reach 100%; in Uganda, Gao et al. reported 78% and 89% [167] and Simpson et al. found 82% prevalence [59].
In conclusion, prevalence rates varied depending on the geographic origin of the sera tested and the specific tests used to determine these prevalences; in particular, whether antibodies against latent or lytic antigens were detected could make a difference in the results. Additionally, it is unclear whether these differences were truly due to varying prevalence rates, or perhaps to a lack of sensitivity and specificity of the serologic assays, as has been shown for HIV [204] and HCV [205]. Because most reports indicated high rates of HHV-8 infection in persons with KS, regardless of their origin, it is probable that the assays possess reasonable ability to detect true infection.
6.A.j. Risks of age related HHV-8 infection
Regamey et al. reported that there was a trend of increasing HHV-8 antibody prevalence to Orf 65 antigen with increasing age in HIV negative individuals in Switzerland [184]. Below 30 years of age, the prevalence increased from 15% to 23% and then to 50% in the next three decades. A similar effect was observed in BDs in Hungary [183]. As age increased from 19 until 25 years of age and then for every decade afterwards, the distribution of seropositivity to LANA increased moderately, but significantly (P = 0.048). A similar association was observed with Orf 65 peptide reactivity but the numbers of subjects were too small to calculate statistical significance [183]. In Taiwan, increased progression of antibody response against HHV-8 lytic antigens was observed, starting with a low of 3% in children under five years of age and peaking between age 31 and 40 (19.2%) [196]. Many more examples of this have been reported in Africa [83,128,203], Sardinia [61], and Italy [192]. Perna et al. [193] and others [172,183,185,192] have shown that there most likely exists non-sexual routes of HHV-8 transmission because children worldwide have been infected by HHV-8.
6.B. Molecular prevalence of HHV-8 genotypes and variants
From DNA sequence analysis of distinct loci derived from 60 HHV-8 isolates, the clustering of four major HHV-8 viral subtypes was discovered [206]. These subtypes, A, B, C, and D are based upon DNA sequence derived from the K1 gene, a glycoprotein with transforming properties [207,208], and they exhibit 30% amino acid (aa) variability. These aa substitutions result from an 85% nucleotide substitution rate in this highly variable gene. The four subtypes were further divided into another 13 clades by Hayward [206]. The A1, A4, and C3 variants were predominant in the US AIDS KS samples, but the B variant was predominant in samples from Africa. C variants were observed from samples from Saudi Arabia and Scandinavia. The D subtype was uncommon and was found only in classic KS patients in the Pacific region. Another gene, K15, showed two different alleles (P and M), but these allelic types were not associated with the K1 subtypes [206]. These different genotypes have been investigated to explain the possible pathogenic and epidemiologic variation seen with HHV-8 infection [125]. Studies that are more recent have expanded upon previous work and have shown that the K1 locus can be divided into six subtypes with 24 clades showing strong linkage to the geographic origin of the particular isolate. Data have shown that subtypes A and C are prevalent in Europe, the U.S.A., and northern Asia. Subtypes B and A5 predominate in Africa and the D variant is found in the Pacific. Subtype E has been discovered in Brazilian Amerindians and a unique subtype Z was found in Zambia [125]. In a recent study, Whitby et al. characterized the K1 hypervariability from general populations in South America and Africa: i.e., those without any obvious symptoms of HHV-8 infection [179]. Amerindians from Ecuador carried the E subtype, in keeping with previous studies from South America. In Botswana, subtypes B and A5 were exhibited by subjects from the Bantu and San tribes, similar to the subtypes found there from KS patients. These results show that the same HHV-8 viral strains from similar geographic regions can be found in both diseased and non-diseased individuals, suggesting that there is no association between certain genotypes and disease.
7. HHV-8 Gene Products of Diagnostic Importance
7.A. Orf 73 (LANA1) latency protein
Immunofluorescent observations that PEL cells exhibited a distinct nuclear immunofluorescence after challenge with antisera from KS patients, led to the identification of Orf 73 as the gene responsible for the latency associated nuclear antigen-1 (LANA1) [209-211]. Early gene alignments had suggested that Orf 73 was an immediate early gene with 51% similarity to the Orf 73 of HVS [19]. Studies have since shown that LANA1 is a 222–234 kDa protein that is expressed in the majority of nuclei in KS spindle cells [211,212]; however, the LANA1 protein expression is variable [211] and can depend upon the clinical stage of the KS tumor [213]. The immunodominant epitope has been mapped to the C-terminal domain of the protein [210]. The gene is under latent control as evidenced by reduction in Orf 73 mRNA after chemical induction of the viral lytic phase [210]. The antigenicity of the recombinant LANA1 protein has been shown by Western blot; over 70% of HHV-8 IFA seropositive sera were LANA1 positive in the Western blot [210,211].
7.B. Orf 65 capsid protein
Orf 65 was identified by Russo et al. [19] as a lytic capsid protein with less than 60% similarity to similar capsid proteins from HVS and EBV, but is not cross-reactive with HVS and EBV capsid proteins [59,214-216]. Orf 65 has been shown to be the smallest component of the HHV-8 capsid with a predicted basic isoelectric point of 9.6, similar to other herpesviruses [217]. Because of its embossed structural position on the capsid, Orf 65 might be involved in interactions with the viral tegument and cellular proteins upon infection [218]. First cloned in bacteria by Simpson and colleagues [59] and subsequently by others [215], Orf 65 is a highly antigenic 18–22 kDa protein against which more than 81% of KS patients are seroreactive [59,215]. The dominant eight amino acid epitope has been mapped to the C-terminus, and allowed development of a peptide assay with reactivity in 90% of the KS samples tested [216].
7.C. K8.1 glycoprotein
Originally identified as a single gene locus [19], research has since shown that K8.1 is derived from spliced transcripts [219] for which the transmembrane sequence is appended [220]. This glycoprotein is unique to HHV-8 and is a TPA-inducible lytic protein [221]. On Western blots from induced PEL cells, it measures between 35–40 kDa with the characteristic smear of a glycoprotein [221]. Immunoelectron microscopy suggests that the virion acquires the K8.1 glycoprotein at the cell plasma membrane while budding from the host cell [222]. Two transcripts are produced, K8.1A and K8.1B, of which K8.1B is the shorter by dint of an internal deletion of 61 amino acids. K8.1A, casually referred to as K8.1, is very antigenic, with 97% of HIV+, KS+ patients having antibodies directed against it on Western blot; in HIV+, KS-, persons 61% showed reactivity [219].
7.D. Other antigenic proteins
Orf 25 and Orf 26 code for other major and minor HHV-8 capsid proteins, respectively, and were investigated for their diagnostic utility [223]. Orf 25 possesses 68% identity to the EBV BCLF1 major capsid protein and exhibited considerable cross-reactivity to EBV+ sera and was not used further in their studies. However, Orf 26 has only 49% identity to its EBV gene homologue and showed no cross-reactivity [223]. Only one third of KS patients were reactive to Orf 26, although some exhibited an increase in IgM and IgG reactivity 15 months prior to KS disease.
The Orf 59 protein is another HHV-8 protein that has shown modest diagnostic importance in a few investigations. This gene has about 50% similarity with its HVS and EBV homologues and is presumed to be a DNA replication protein in those viral systems [19]. Orf 59 is a 50 kDa protein with characteristic early-late lytic expression patterns seen for other viral proteins necessary for viral DNA replication [224]. The protein has been localized to the nuclear membrane via IFA and is observed in approximately 30% of induced PEL cells, but in less than 8% of uninduced cells [224]. The Orf 59 gene product, processivity factor-8, has been shown to be present in AIDS-KS tumors (50%) although perhaps not in as many spindle cells as Orf 73 [225]. Approximately 30% of AIDS-KS patients had antibodies against this antigen [226]. Orf 59 might be helpful in identifying aggressive KS disease [225,226].
8. HHV-8 Diagnostics
8.A. HHV-8 serological diagnostics
Presently, the diagnosis of KS requires clinical and histologic evaluation; however, the increasing documentation of its association with HHV-8 has raised the important possibility of being able to predict disease occurrence by demonstrating HHV-8 infection [55]. Additionally, there is a need to develop sensitive and specific serological assays to detect antibodies to HHV-8 for possible blood bank screening, assisting in clinical diagnosis, and in research to facilitate the understanding of the scope of this virus's association with rare, but nonetheless life threatening malignancies. HHV-8 infection can be identified by polymerase chain reaction in tissues and in cells; however, amplification methods are expensive, time consuming, and have been shown to be lacking in sensitivity for easily accessible diagnostic specimens such as plasma and PBMCs [177]. Alternatively, the testing for specific antibodies to HHV-8 offers a simple, inexpensive, and effective means to document infection and a help to define the relationship between infection and disease progression and yield insight into pathogenic mechanisms.
Currently, four methods have been used to demonstrate antibodies to HHV-8: enzyme-linked immunosorbant assay (ELISA), immunofluorescent assay (IFA), Western blot, and immunohistochemistry (IHC). Detection of infection and determination of seroprevalence can be dependent upon which test is selected [227]. ELISA methods vary according to the HHV-8 antigens used and whether they are recombinant antigens, viral lysates, or synthetic peptides. IFA methods incorporate virally-infected cell lines, either latently infected with expression of LANA1, or cells that express lytic antigens following chemical induction (i.e., those representing viral replication). The Western blot technique utilizes electrophoretically separated virally infected cell lysates or whole viral lysates, with transfer to nitrocellulose and then subsequent detection of reactive antigens; it has the advantage of identifying the presence of antibodies to specific antigens. IHC on fixed cells and tissue allows to determination of which cells harbor the virus in vivo and semi-quantitative analysis of infected sell type to help learn more about pathogenesis. IHC is also useful to confirm or rule out the clinical diagnosis of KS. These tests are explored in the following sections.
8.A.a. HHV-8 antigen sources
PEL cell lines have been important sources of antigen mainly for use in IFAs, but also in the form of cell lysates for Western blotting and tools for investigations into HHV-8 pathogenesis [59,210-212,215,224,226]. Over 12 PEL cell lines have been established and they each contain 50–150 episomal copies of HHV-8 per cell [8,132-136]. About half are coinfected with EBV (e.g., BC-1, BC-2, BCBL-2), but others have only latent HHV-8 infection (e.g., BCBL-1, BC-3, KS-1) [135]. Induction of viral replication can be initiated by sodium butyrate (butyrate [228], 12-O-tetradecanoylphorbol-13-acetate (TPA), a phorbal ester [229], or less commonly hydrocortisone [130]. Cell cultures derived from KS spindle cells are not good material for HHV-8 diagnostics because they lose the virus after 2–6 passages [230].
Other sources of antigen have been whole virus lysate, which has been used successfully in the ELISA format [231]. After induction of a PEL cell line, the whole virus is usually purified over a sucrose gradient. The drawback of this method is that it preferentially selects for lytic antigens and does not allow detection of latent antibodies such as LANA1 [112]. In contrast, individual HHV-8 proteins have been incorporated into tests by either expressing them as recombinant proteins or as synthesized peptides. Recombinant proteins such as Orf 65, K8.1, Orf 25, and Orf 26 have been expressed in easy to grow bacterial systems [59,221,223]. Antigenic proteins have also been expressed in more difficult to grow baculovirus systems (insect cells) [232,233], but they have the added benefit of protein glycosylation which bacterial cells can not perform. It has also been reported that LANA1 (Orf73), because of its large size (>200 kDa) is expressed better in insect cells (personal communication, Dr. D. Whitby, NCI-Frederick). Synthesized peptides of immunodominant portions of antigenic proteins (e.g., K8.1, Orf 65) have been developed as a strategy to streamline the production process and to reduce non-specific reactions [183,234].
8.A.b. ELISAs for the detection of HHV-8 infection
ELISA tests are easier to manipulate and technically are the test of choice for large-scale seroprevalence studies. ELISAs based on recombinant antigens of HHV-8 have shown that a specific humoral response is produced against capsid proteins of HHV-8, allowing identification of HHV-8 infection [59,92,223]. Recombinant proteins derived from a truncated Orf 65 minor capsid gene have been used with a relatively high degree of success to differentiate populations of KS patients from BDs [214]. Similarly, recombinant proteins derived from the Orf 25 and Orf 26 genes (major and minor capsid proteins) have been used in ELISA assays to detect IgG and IgM antibodies, but with a lesser degree of success [223]. Seroconversion against capsid proteins has been shown to occur in less than one year after infection using an Orf 65 ELISA [92].
An ELISA based on viral lysate antigens of HHV-8 has also produced encouraging results [231]. Although this assay demonstrated a good sensitivity for detecting infection in patients with classical KS (CKS) and AIDS-KS (80%–90%), normal healthy blood donors had 2–11% prevalence. This ELISA also possessed the ability to differentiate populations based on antibody titer; the mean titer in blood donors was 1:30, while titers ranged from 1:6000 to 1:15000 in AIDS-KS and CKS patients.
Encouraging results have come from a recombinant ELISA based upon the K8.1 gene product [235] and has been considered one of the more sensitive tests with acceptable specificity. Immunodominant peptides from the Orf 65 and K8.1 antigens were incorporated in an ELISA format and used successfully to measure the risk factors in women [76] and to identify HHV-8 infection in allogeneic stem cell transplant patients who are at risk of KS because of their immunocompromised status [236].
Initially, the primary method of detection of latent antibodies was using the LANA IFA, however, subsequent cloning of Orf 73 (LANA) and its application in the ELISA format has begun to replace the LANA IFA. The Orf 73 ELISA has been found to possess the same high specificity, but with a 10% increase in sensitivity [107]. The Orf 73 ELISA has found utility in gauging the progression to KS in HIV+, HHV-8 infected persons [43]. In that study, increasing titers to Orf 73 over time were associated with HIV+ patients acquiring KS.
8.A.c. IFA for the detection of HHV-8 infection
IFAs are a common method to identify antibodies to HHV-8. To detect latent antibodies, an HHV-8 infected PEL cell line (e.g., BCP-1, BCBL-1, BC-3, KS-1) is used to measure antibodies to the primary latent antigen, LANA1 or ORF 73 [107]. This latent antigen corresponds to a ~234 kDa nuclear antigen, which has been shown to be recognized by sera from KS patients [211], and is characterized by its speckled nuclear fluorescent signature in 95% of PEL cells [107]. With this assay, seroprevalences have ranged from 2%–27% in several studies of blood donors where KS is endemic, but lower (0%–15%) for those geographic regions where KS is mainly associated with AIDS and transplant patients [227]. However, the LANA1 assay has been shown to be relatively insensitive and therefore might not be the best choice of assay to screen low titered populations [235].
Lytic antigens can be expressed by these cells following induction with a TPA or with butyrate [237] and have produced encouraging results. The number of induced cells is dependent upon the cell line used, the time of induction, and the chemical used to induce the cells [107,238]. Studies using induced PEL cell lines point to much higher frequencies of infection than have been suggested by serology based on latent proteins in populations not at risk for sexually transmitted diseases [16,239]. However, other studies using lytic IFAs have also indicated that there are higher levels of HHV-8 infection in otherwise healthy individuals [227] and infection would be spread by non-sexual routes in these cases. As with the ELISA, the IFA has been used to determine antibody titers, with sera from HIV-positive persons with KS demonstrating higher titers to lytic and latent antigens as compared to individuals without KS [214,231]. This test method is relatively more sensitive to serum dilutions that are not extensive enough; the correct serum dilution is important to correctly differentiate true positive reactions from those that are non-specific [112,214].
8.A.d. The diagnostic utilities of the Western blot
Western blots using purified viral lysates of HHV-8 have been used to identify immunodominant proteins using sera from pre- and post-KS patients [114,221,240]. This method has shown utility in the diagnosis and prognosis of KS, but it is more cumbersome and expensive than other serologic assays. A 35–37 kDa glycoprotein has been a protein most frequently and intensively detected, and corresponded to the K8.1 Orf of HHV-8 [220].
In a review of articles that used Western blot in their investigations of HHV-8, only four dealt with antigen identification or expression. These reports could influence the development of a confirmatory Western blot as they showed: 1) There are different antigen profiles in diseases associated with HHV-8 [140]. 2) HHV-8 possesses the glycoprotein, gB, found in other herpesviruses and might be a candidate antigen [241]. 3) That different risk groups and different stages of disease could exhibit different antibody profiles [242]. 4) Patients undergoing antiviral therapy might not produce certain antibodies due to a decreased expression of HHV-8 antigens [238]. These findings suggest that it might be necessary to identify specific antigens for use at specific times of infection and even for different disease states.
Nine reports involved the use of Western blots for screening purposes [61,94,210,219,221,226,243-245]. Recombinant Orf 65 was used most often followed by K8.1, Orf 59 and Orf 73, and finally vIL-6 and Orf 47; however, there was no utility in using vIL-6 or Orf 47. In these reports, KS sera were detected by K8.1 with the greatest sensitivity, followed by Orf 65 and then Orf 73. In these studies, there were not sufficient HIV+ sera examined to draw conclusions as to which antigen was best in that specific population. Sera/plasma from healthy controls varied from a low of 0% for Orf 59 and Orf 73 to 6.5% for K8.1 and 8.3% for Orf 65.
Eleven research reports utilized Western blots as tools to confirm the results of previously run serological assays [59,92,183,232,246-252]. Most of these authors used the same antigen found in the ELISA as the confirmatory antigen in the Western blot; however, two reports had the Western blot confirm IFA results. In seven instances, the authors used the Western blot to confirm a single screening assay and in four reports, they used the Western blot it to resolve a disagreement between two screening assays or in duplicate samples.
More recent reports have continued to use the Western blot as both a primary assay and as a confirmatory test [175,253]. The Western blot method has the benefit of allowing identification to one or more antigens. With accessibility to multiple recombinant proteins now possible, several researchers have developed recombinant Western blot utilizing more than one protein [197,254]. In those reports, they accepted reactivity to one of three antigens to be a marker of HHV-8 infection. In this manner, Wang et al. proposed a new antigen, Orf 57, for use in asymptomatic populations [197]. Clearly, despite the technical difficulties in producing Western blots, they are a useful, multitasking serological method in HHV-8 diagnostics.
8.A.e. Comparisons and concordances between assays
Estimates of the prevalence of HHV-8 by different ELISAs have varied. This variance has been shown in reports of multicenter or multitest studies. Spira et al. [255] found a range of concordance between 69% and 94% using seven serologic tests with Kappa values as low as 0.387 (fair agreement) and as high as 0.909 (almost perfect). Rabkin et al. [256] also evaluated seven serologic tests and found a range of concordance between 50% and 94%, with Kappa statistics ranging from -0.08 to 0.86, indicating that the interassay correlation between the assays was less than favorable. The tests frequently disagreed on individual sera, particularly from blood donors. It was concluded that current antibody tests for HHV-8 have uncertain accuracy in asymptomatic HHV-8 infection and that additional tests to define the actual prevalence may be required. Poor correlation for positive results has been observed in other studies [257]. Second generation tests seem to provide better concordances, although the best results came from IFA tests rather than ELISAs in one multicenter study [258]. Even with more optimized assays, sensitivity and specificity can be insufficient for clinical use [235]. As with the detection of infection by many viruses (e.g., HIV), sequential use of screening and confirmatory tests for HHV-8 are likely to be required to address sensitivity and specificity issues; accordingly, an testing algorithm has been reported [235]. These findings supported the need for critical investigation of the parameters that could influence the performance of these tests.
Although there is some variability in prevalence among similar populations with the same test, most data show that there is agreement within a defined range. The lack of concordance in HHV-8 diagnostic assays occurs primarily because not all HHV-8 infected persons exhibit all antibodies against all HHV-8 antigens at the same time [259]. This phenomenon of single antibody reactivity is much more apparent in populations who are at low risk of infection, such as blood donors [259]. Because of this, specificities are more variable than sensitivities among different laboratories [259].
Although refinement of the diagnostics assays is still possible, the greatest chances of success are in developing algorithms that make use of multiple assays for screening and then confirmation or alternatively, the use of assays that incorporate multiple antigens which have been shown to be highly immunogenic, perhaps during different stages of infection [259]. It is possible to use a combination of latent and lytic antigen tests to determine a true positive as has been employed by several laboratories [112]; however, recent data indicate that the humoral response to HHV-8 does not always produce both latent and lytic responses at the same time [260]. In addition, antibodies directed against lytic antigens seem to be more prevalent than those for latent antigens.
8.A.f. IHC for the detection of HHV-8 infection
IHC is a powerful serologic tool, but like Western blots can be tedious to perform. In the field of HHV-8 research and diagnostics, IHC has been used to locate HHV-8 proteins, assess involvement of HHV-8 in malignancies, detect specific HHV-8 gene expression, and to provide diagnosis of KS. The ability to identify which specific cell types or structures within a cell are expressing HHV-8 proteins can assist in the understanding of HHV-8 pathogenesis [261,262] and determine the possible etiology of malignancies [263-267]. Detection of specific HHV-8 gene expression, in particular LANA1 [140,211], has led to possible clinical applications for the diagnosis of KS in tissue samples [268,269]. This allows the exclusion of other neoplasms that can mimic KS [268-270]. The ability of monoclonal and polyclonal antibodies to localize specific HHV-8 antigens should continue to improve HHV-8 diagnosis and our understanding of HHV-8 pathogenesis.
8.B. HHV-8 molecular diagnostics
The diagnostic benefit of the polymerase chain reaction (PCR) for herpesviruses other than HHV-8 has been mixed. Studies have shown a lack of correlation with PCR and positive serological tests results for viral retinitis [271] and no herpesvirus sequences were discovered in the PBMCs of suffers of chronic fatigue syndrome (CFS) [272]. Other studies, however, have found PCR to be useful in diagnosing HHV-6 infection in exanthum subitum during convalescence where IgM is no longer detected [273]. In general, Pearson et al. recommended the use of PCR to better diagnose acute infection or reactivation in herpesvirus infections unless sentinel antigens could be identified [274]. For the detection of HHV-8, the PCR method with optimal performance should fulfill several conditions. The test should be specific for DNA sequences found only in the HHV-8 genome and not other herpesviruses. The K-genes might be good candidates for this, and indeed a real-time PCR test using the K6 region has been used [177]. Sensitivity is an absolute requirement because the virus is found at such low copy numbers due to its latent biology. Most reports have indicated sensitivities from 1–100 copies per reaction [275]. However, the herpes-specific biology makes sampling error a concern. Therefore, strategies are needed to detect the virus in latency, such as induction of the lytic cycle before DNA isolation. There have been a few reports where this has been attempted with success [48,58]. If nested PCR is to be used to gain the needed sensitivity, exceptional care must be taken to avoid false positives. However, nested PCR has the power to provide added specificity and confirmation by amplifying two separate amplicons in the nested PCR reaction. Alternatively, multiplexing in real-time PCR, with the proper optimization and design, could provide this needed level of surety. An easily obtainable diagnostic sample would complete the diagnostic strategy to maximize the effectiveness of PCR for the detection of HHV-8. Reports have shown that saliva contains the highest prevalence of virus in HIV negative persons [56] and in samples from HIV+ patients it is equivalent to PBMCs [51,56]; therefore, it should be considered the sample site of choice. Saliva collection devices are already commercially available (OraSure, Bethlehem, PA) and FDA approved for serologic testing and might be convertible for use for PCR. Finally, the ability to quantify HHV-8 viral loads using quantitative real-time PCR has been employed to measure HHV-8 viral burdens to investigate the association of viral load and progression to KS [276,277] and the pathogenesis and transmission of HHV-8 [86,278].
In a review of the literature, the use of molecular diagnostics, in particular PCR, for the detection of HHV-8 infection has been less than optimal. In most cases, serology is the preferred method to identify HHV-8 infection. Most articles have shown that at least one serological assay had better sensitivity than PCR on the same samples, even better than nested PCR. For example, in a study of AIDS-KS, IFA was able to detect HHV-8 antibodies in 50% (latent) to 100% (lytic) of the patients, whereas, nested PCR detected infection in only 33% [57]. The data from a minority of reports showed that PCR was a more favorable assay in isolated cases [279] or that serology and PCR were comparable [280]. In a composite set of 642 samples from numerous reports, 69% were concordant in their PCR and serology results. However, 179 samples (28%) were positive by serology, but PCR negative; only in 21 samples (3.3%) was there a PCR positive result without a corresponding positive serology.
The utility of PCR in detecting HHV-8 in KS patients is better but not perfect. PCR appears to be very useful when detecting HHV-8 directly in the KS lesions, with sensitivity approaching 100% [50,51,183,279]. However, PBMCs from KS patients were observed to have fewer instances (~50%) of detectable viral sequences [50,51,55]. Detection of HHV-8 DNA by PCR in the PBMCs of HHV-8 infected individuals is not a common event. Only 10–20% of seropositive persons have detectable HHV-8 DNA, but this percentage increases with evidence of KS disease and more severe disease [259]. Even in KS lesions, if the tissue sample is not processed correctly for PCR, there can be false negative results [259]. However, PCR has been found to be useful in detection of early infection or reactivation, especially at times of clinical sequelae of viral primary infection or reactivation [49].
Few reports have used plasma or sera as the analyte for PCR, especially juxtaposed to serological methods [49,56]. However, these investigators seem to indicate it does not perform any better than PBMCs. It is noteworthy to add that several authors observed HHV-8 viremia to be intermittent. In longitudinal samples, several investigators have found that despite enhanced detection schemes and serial samples over periods of time exceeding two years, detection of HHV-8 in PBMCs can be missed 30% of the time or more [54,57,58,281,282]. Even in saliva, which has been shown to carry a relatively higher viral burden, due to intermittent shedding up to 65% of the time, detection of the virus can be missed if only single samples are relied upon for diagnosis [51,56]. Finally, in serum/plasma, detection of can be intermittent with perhaps the best chance of detection at signs of clinical disease [49,54].
Reports on the use of in situ hybridization and reverse-transcriptase PCR (RT-PCR) have been used as mainly research tools to investigate associations of HHV-8 and specific diseases [29,267]. Most RT-PCR reports were concerned with detecting mRNA transcripts to determining infectivity [283] or as a diagnostic method in HHV-8 related disorders, such as PELs [284]. There have been few reports using nucleic acid sequence-based amplification (NASBA) assays to detect and quantitative HHV-8 viral loads in HHV-8 diseases [285,286], although the reports seem to confirm the findings from quantitative PCR studies that increased viral load in to be expected in more advanced KS, both in the lesions and in the PBMCs.
8.C. Commercial sources
Although HHV-8 has been associated with only a few diseases, commercial sources for both testing and kits are available. These include molecular and serologic testing from established laboratories and hospitals, although PCR seems to be the method most used (Table 2). IFA or ELISA serologic kits are also commercially available (Table 3), but no companies seem to be marketing molecular kits except for Celonex, which produces a microarray system for herpesviruses.
Table 2 Companies or institutions that provide molecular testing services or research kits for the detection HHV-8 infection.
Molecular testing & kits
Company Test Utility
Focus Diagnostics, Inc. Herndon, VA, USA PCR, qualitative Method for identifying individuals among HIV+ persons who are at increased risk for developing KS. "The results are for research use only, and should not be used for diagnostic purposes."
ViraCor Laboratories Lee's Summit, MO, USA Real-time PCR, quantitative (100 copies/ml to 1 × 1010 copies/ml) Clinical diagnostics: Determination of HHV-8 primary infection and for determining the risk of developing KS among organ transplant patients and patients taking immune suppressive drugs.
ARUP Laboratories Salt Lake City, UT, USA Real-time PCR, qualitative (limit of detection: 1 in 100,000 cells) Clinical diagnostics: To predict the development of KS, to aid differential diagnosis in other vascular neoplasms and inflammatory conditions that are histologically similar to KS, to diagnose PELs, and to monitor patients with immune compromise or dysregulation.
LabPLUS Auckland City Hospital, New Zealand. PCR, qualitative Clinical diagnostics: Diagnosis in KS, PEL, MCD
Medical Diagnostic Laboratories, L.L.C. Hamilton, NJ, USA PCR, qualitative Clinical diagnostics
UT Southwestern Medical Center Dallas, TX, USA Real-time PCR, qualitative Clinical diagnostics
Celonex Edmonton, Alberta, Canada Single HHV-8 ViruChip™ Gene expression
Table 3 Companies or institutions that provide serologic testing services or research kits for the detection HHV-8 infection.
Serologic testing & kits
Company Test Utility
Fred Hutchinson Cancer Research Center Seattle, WA, USA ELISA Clinical diagnostics
Focus Diagnostics, Inc. Herndon, VA, USA IgG IFA Method for identifying individuals among HIV+ persons who are at increased risk for developing KS. "The results are for research use only, and should not be used for diagnostic purposes."
Quest Diagnostic (Focus Technologies) Baltimore, MD, USA IgG IFA "This test should not be used for diagnosis without confirmation by other medically established means".
Advanced Biotechnologies Inc Columbia, MD, USA 1) IgG Antibody IFA Kit
2) IgG Antibody ELISA Kit (whole virus lysate) For research use only.
Biotrin International The Rise, Mount Merrion Co. Dublin, Ireland 1)IgG IFA assay
2) DIAVIR HHV-8 peptide mix (Orf 65 & K8.1A) ELISA For research use only. To aid in the diagnosis of primary infection or to identify reactivation or reinfection. To determine current or recent infection by testing of paired specimens of plasma or serum taken 7–14 days apart; a ≥ 4-fold rise in titer is indicative of recent infection.
Panbio Inc. Columbia, MD, USA 1) IgG IFA (Lytic)
2) IgG IFA (Latent)
3) DIAVIR HHV-8 peptide mix (Orf 65 & K8.1A) ELISA For research use only
9. Current Diagnostic Issues
Current HHV-8 diagnostic tests are not commonly used in the clinical arena because their procedures are not standardized and the specific patient populations to which they would best be applied are not clearly identified. Investigators have not been able to unambiguously determine if low risk individuals, such as blood donors, who happen to test positive using the current array of assays, are truly infected. Therefore, there is an urgent need for a gold standard, FDA-approved diagnostic test for HHV-8. The difficulty in detecting HHV-8 in patients makes development of a gold standard seroassay difficult at best and the determination of specificity almost impossible. The current, incomplete understanding of how HHV-8 is transmitted, and the risk factors associated with its transmission add to the burden of correlating diagnostic test results to true infection. For example, a patient admitted to an emergency room complaining of myalgia, fever, and headaches could be presenting with symptoms from any number of infectious or non-infectious illnesses. However, if the clinical history indicates a recent walk in the woods with a tick bite, then the diagnostic picture narrows to include the possibility of ehrlichiosis or borreliosis. The translation of research knowledge into the clinical arena will require careful development, evaluation, optimization, and refinement to develop a new standard of care that blends advances in both diagnostic and clinical sciences [287].
There are other deficiencies in HHV-8 diagnostic testing methods. First, there is no effective HHV-8 confirmatory assay similar to the Western blot used with HIV. Because of the large variability of results between current tests and between tested populations, it is difficult to find agreement between two tests, except perhaps, in KS patients. The inconsistent assay results also impede development of effective diagnostic algorithms. Second, the availability of an antigen capture assay (currently unavailable) would benefit HHV-8 diagnostics in several ways. For example, knowledge of the time course and concentrations of virus circulating in patients (temporal antigenemia) could help elucidate the natural history of HHV-8 infection, which in turn could be utilized to detect early HHV-8 infection, to confirm infection, and to monitor therapy. Suitable antigens with high copy number such as capsid proteins would be required. High affinity and high avidity antibodies would need to be identified or developed, and preferential access to the respective recombinant antigen would be required for test development and for use as test controls. Fortunately, commercial and research sources of antibodies exist against both latent and lytic antigens, such as LANA1, K8.1, Orf 65, and Orf 59, which will accelerate development of antigen assays.
There is a deficiency of HHV-8 antigenic proteins for use in diagnostic tests. Current HHV-8 ELISAs target IgG antibodies to one of three viral antigens: K8.1 [235], Orf 65 [59], or Orf 73 [235]. To date, no other useful HHV-8 proteins have been discovered that provide acceptable sensitivity and specificity in all populations tested, despite a viral genome that can express over 47,000 amino acids. Further research into identifying antigenic proteins is needed.
The use of Western blot as a screening tool for HHV-8 is impractical, and Western blot confirmatory tests suffer from nonspecific reactions when whole cell lysates are used. Currently, the choice of HHV-8 antigens is limited for development of recombinant immunoblots making the formulation of confirmatory Western blots difficult.
Although many published reports have confirmed the utility of antibody isotype tests other than IgG for the detection of other herpes viral infections, there is a dearth of reports detecting anti-HHV-8 IgA and IgM antibody isotypes. For example, patients with chronic fatigue syndrome and multiple sclerosis were more apt to have IgM antibodies against HHV-6 [288,289]. IgA against EBV VCA is at a higher seroprevalence and geometric mean titer in patients with EBV-positive gastric carcinomas [290], and is predictive of nasopharyngeal carcinoma [291]. This is in contrast to HHV-8 where there are few reports of HHV-8 antibody isotype assays for IgA and IgM [92,167,223,249,257,292], and none where the investigator compared IgG, IgA, and IgM isotypes concurrently in the same laboratory with the same tests and serum samples. Theoretically, detection of IgA and IgM anti-HHV-8 might improve identification of HHV-8 infection and provide early diagnosis. IgA and IgM isotype detection could also be incorporated into improved diagnostic algorithms to better define the prevalence and disease associations of HHV-8 infection.
If HHV-8 is similar to other herpesviruses, there may be difficulty in identifying specific antigens to which the majority of infected individuals have mounted an antibody response. For example, among the many other viral structural proteins of HCMV, only one, pp150, is recognized by most infected individuals [293] and a p101 protein was found to be most antigenic for HHV-6 [294]. Finding immunodominant antigens may take extended study and application of novel techniques. In addition, determining the sequence of specific antigenic gene products from viral isolates from diverse geographic regions is necessary to ensure that antigens used as a lure for HHV-8 specific antibodies are universally detected [216,295].
In regards to molecular testing, only a few reports have evaluated the utility and efficacy of performing PCR on activated PBMCs isolated from persons potentially infected with HHV-8. Cell culture activation of a blood donor's PBMCs using IL-2, TPA, and hIL-6 increased detection from 1/7 to 5/7 serial samples [58]. Another report showed that the presence of inflammatory cytokines, specifically Inf-γ, increased the HHV-8 viral load to detectable limits in cultured PBMCs derived from both AIDS-KS and non-KS AIDS seropositive patients [48]. Studies to confirm this seemingly useful approach and to define the optimal viral amplification procedures are needed.
The reverse transcriptase PCR (RT-PCR) assay is a popular molecular diagnostic test for retroviruses or RNA viruses, such as HCV or HGV. RT-PCR is usually not necessary for DNA viruses, because the viral genomic DNA itself can be detected without the intermediate step of reverse transcriptase to create cDNA. However, since the unique latent biology of HHV-8 renders DNA PCR of HHV-8 relatively insensitive, RT-PCR should be studied more thoroughly as an alternative diagnostic test for the detection of HHV-8 infection. The rationale is that detection of mRNA provides a built in preamplification step for detection of the viral nucleic acid, because mRNA is at a higher copy number than the corresponding genomic DNA. This method could also allow the detection of both latent and/or lytic transcripts increasing the chances of success. To our knowledge, there are no reports in the literature that RT-PCR has been evaluated seriously as a diagnostic or screening assay for HHV-8 infection.
As an adjunct to the necessity of improved HHV-8 diagnostics, the effective use of HHV-8 viral therapy will depend on the development of sensitive and specific HHV-8 diagnostic tests to gauge the therapy's effectiveness. Accumulating research either has implicated HHV-8 as the etiologic agent of diseases such as KS or has associated the virus indirectly with disease development. The efficacy of clinical therapeutic drug interventions for HHV-8 infection has not been studied thoroughly in clinical settings, rather, mainly through in vitro experiments. Prospective anti-HHV-8 therapeutic trials of anti-herpetic drugs are needed in large and diverse cohorts of AIDS patients presenting with KS. Organ transplant patients, in order to prevent organ rejection, also require intense study to determine the proper anti-HHV-8 intervention in the absence of HAART and in the presence of immunosuppressive therapy. It will be more difficult to study the therapy of patients with PEL and MCD because of the low prevalence of these diseases.
Modern medicine will be able to manage this novel human herpesvirus only through continued research into the dynamics of HHV-8 infection in vivo, and the identification of important and unique antigens and their subsequent development into diagnostics tests. Such advances in turn will result in better understanding of the pathogenesis and associated diseases of HHV-8 and catalyze antiviral therapy and strategies for prevention.
10. Conclusion
Although the prevalence of HHV-8 is not as ubiquitous as other human herpesviruses, there is strong evidence that it is required and quite possibly is the primary etiological agent for the formation of several life threatening neoplasms, including KS. Therefore, the development and optimization of improved diagnostic assays is critical for the identification, diagnosis, and monitoring of HHV-8 infection. Our work at the University of Maryland Baltimore has addressed important issues in the field of HHV-8 investigation; namely, the lack of a gold standard serologic assay to detect the virus or antibodies to the virus, a lack of optimization of current serologic assays, few reliable diagnostic HHV-8 antigens available for serologic tests, the epidemiology of HHV-8, and an incomplete understanding of the host humoral response to HHV-8 infection.
11. Acknowledgements
This review was written to partially fulfill the requirements of my PhD dissertation, and I acknowledge the invaluable assistance of my dissertation committee members: Niel T. Constantine, PhD (advisor), Bill Blattner, MD, Marv Reitz, PhD, Ed Highsmith, PhD, Denise Whitby, PhD, and Judy Johnson, PhD. In addition, editorial support was gratefully provided by Robert Edelman, MD and Janet Barletta, PhD.
==== Refs
Chang Y Cesarman E Pessin MS Lee F Culpepper J Knowles DM Moore PS Identification of herpesvirus-like DNA sequences in AIDS-associated Kaposi's sarcoma Science 1994 266 1865 1869 7997879
Knipe DM Howley PM Fields Virology 2001 1 4 Philadelphia: Lippincott Williams & Wilkins
Moore PS Chang Y Molecular virology of Kaposi's sarcoma-associated herpesvirus Philosophical Transactions of the Royal Society of London – Series B: Biological Sciences 2001 356 499 516 10.1098/rstb.2000.0777
Mesri EA Cesarman E Arvanitakis L Rafii S Moore MA Posnett DN Knowles DM Asch AS Human herpesvirus-8/Kaposi's sarcoma-associated herpesvirus is a new transmissible virus that infects B cells J Exp Med 1996 183 2385 2390 8642350 10.1084/jem.183.5.2385
Neipel F Albrecht JC Fleckenstein B Human herpesvirus 8–the first human Rhadinovirus Journal of the National Cancer Institute Monographs 1998 73 77 9709307
Furlong D Swift H Roizman B Arrangement of herpesvirus deoxyribonucleic acid in the core Journal of Virology 1972 10 1071 1074 4117962
Wu L Lo P Yu X Stoops JK Forghani B Zhou ZH Three-dimensional structure of the human herpesvirus 8 capsid J Virol 2000 74 9646 9654 11000237 10.1128/JVI.74.20.9646-9654.2000
Renne R Zhong W Herndier B McGrath M Abbey N Kedes D Ganem D Lytic growth of Kaposi's sarcoma-associated herpesvirus (human herpesvirus 8) in culture Nat Med 1996 2 342 346 8612236 10.1038/nm0396-342
Walter PR Philippe E Nguemby-Mbina C Chamlian A Kaposi's sarcoma: presence of herpes-type virus particles in a tumor specimen Human Pathology 1984 15 1145 1146 6094332
McCombs RM Brunschwig JP Mirkovic R Benyesh-Melnick M Electron microscopic characterization of a herpeslike virus isolated from tree shrews Virology 1971 45 816 820 4108187 10.1016/0042-6822(71)90203-0
Epstein MA Observations on the mode of release of herpes virus from infected HeLa cells Journal of Cell Biology 1962 12 589 597 13890436 10.1083/jcb.12.3.589
Morgan C Rose HM Mednis B Electron microscopy of herpes simplex virus. I. Entry Journal of Virology 1968 2 507 516 4301316
Asher Y Heller M Becker Y Incorporation of lipids into herpes simplex virus particles Journal of General Virology 1969 4 65 76 4304744
Deiss LP Chou J Frenkel N Functional domains within the a sequence involved in the cleavage-packaging of herpes simplex virus DNA Journal of Virology 1986 59 605 618 3016323
Markovitz NS Filatov F Roizman B The U(L)3 protein of herpes simplex virus 1 is translated predominantly from the second in-frame methionine codon and is subject to at least two posttranslational modifications Journal of Virology 1999 73 8010 8018 10482549
Lennette ET Blackbourn DJ Levy JA Antibodies to human herpesvirus type 8 in the general population and in Kaposi's sarcoma patients Lancet 1996 348 858 861 8826812 10.1016/S0140-6736(96)03240-0
Renne R Lagunoff M Zhong W Ganem D The size and conformation of Kaposi's sarcoma-associated herpesvirus (human herpesvirus 8) DNA in infected cells and virions J Virol 1996 70 8151 8154 8892944
Moore PS Gao SJ Dominguez G Cesarman E Lungu O Knowles DM Garber R Pellett PE McGeoch DJ Chang Y Primary characterization of a herpesvirus agent associated with Kaposi's sarcomae [published erratum appears in J Virol 1996 Dec;70(12):9083] J Virol 1996 70 549 558 8523568
Russo JJ Bohenzky RA Chien MC Chen J Yan M Maddalena D Parry JP Peruzzi D Edelman IS Chang Y Nucleotide sequence of the Kaposi sarcoma-associated herpesvirus (HHV8) Proc Natl Acad Sci U S A 1996 93 14862 14867 8962146 10.1073/pnas.93.25.14862
Arvanitakis L Geras-Raaka E Varma A Gershengorn MC Cesarman E Human herpesvirus KSHV encodes a constitutively active G-protein-coupled receptor linked to cell proliferation Nature 1997 385 347 350 9002520 10.1038/385347a0
Coscoy L Ganem D Kaposi's sarcoma-associated herpesvirus encodes two proteins that block cell surface display of MHC class I chains by enhancing their endocytosis Proc Natl Acad Sci U S A 2000 97 8051 8056 10859362 10.1073/pnas.140129797
Sarid R Flore O Bohenzky RA Chang Y Moore PS Transcription mapping of the Kaposi's sarcoma-associated herpesvirus (human herpesvirus 8) genome in a body cavity-based lymphoma cell line (BC-1) J Virol 1998 72 1005 1012 9444993
Cesarman E Chang Y Moore PS Said JW Knowles DM Kaposi's sarcoma-associated herpesvirus-like DNA sequences in AIDS-related body-cavity-based lymphomas N Engl J Med 1995 332 1186 1191 7700311 10.1056/NEJM199505043321802
Sturzl M Blasig C Schreier A Neipel F Hohenadl C Cornali E Ascherl G Esser S Brockmeyer NH Ekman M Expression of HHV-8 latency-associated T0.7 RNA in spindle cells and endothelial cells of AIDS-associated, classical and African Kaposi's sarcoma Int J Cancer 1997 72 68 71 9212225 10.1002/(SICI)1097-0215(19970703)72:1<68::AID-IJC10>3.0.CO;2-6
Renne R Blackbourn D Whitby D Levy J Ganem D Limited transmission of Kaposi's sarcoma-associated herpesvirus in cultured cells J Virol 1998 72 5182 5188 9573290
Ambroziak JA Blackbourn DJ Herndier BG Glogau RG Gullett JH McDonald AR Lennette ET Levy JA Herpes-like sequences in HIV-infected and uninfected Kaposi's sarcoma patients [letter; comment] Science 1995 268 582 583 7725108
Boshoff C Schulz TF Kennedy MM Graham AK Fisher C Thomas A McGee JO Weiss RA O'Leary JJ Kaposi's sarcoma-associated herpesvirus infects endothelial and spindle cells Nat Med 1995 1 1274 1278 7489408 10.1038/nm1295-1274
Blasig C Zietz C Haar B Neipel F Esser S Brockmeyer NH Tschachler E Colombini S Ensoli B Sturzl M Monocytes in Kaposi's sarcoma lesions are productively infected by human herpesvirus 8 J Virol 1997 71 7963 7968 9311888
Diamond C Brodie SJ Krieger JN Huang ML Koelle DM Diem K Muthui D Corey L Human herpesvirus 8 in the prostate glands of men with Kaposi's sarcoma J Virol 1998 72 6223 6227 9621094
Corbellino M Parravicini C Aubin JT Berti E Kaposi's sarcoma and herpesvirus-like DNA sequences in sensory ganglia [letter] N Engl J Med 1996 334 1341 1342 8609967 10.1056/NEJM199605163342019
Horwitz CA Henle W Henle G Snover D Rudnick H Balfour HH JrMazur MH Watson R Schwartz B Muller N Clinical and laboratory evaluation of cytomegalovirus-induced mononucleosis in previously healthy individuals. Report of 82 cases Medicine 1986 65 124 134 3005799
Manual of Clinical Microbiology
Evans AS Niederman JC Cenabre LC West B Richards VA A prospective evaluation of heterophile and Epstein-Barr virus-specific IgM antibody tests in clinical and subclinical infectious mononucleosis: Specificity and sensitivity of the tests and persistence of antibody Journal of Infectious Diseases 1975 132 546 554 171321
Ablashi D Chatlynne L Cooper H Thomas D Yadav M Norhanom AW Chandana AK Churdboonchart V Kulpradist SA Patnaik M Seroprevalence of human herpesvirus-8 (HHV-8) in countries of Southeast Asia compared to the USA, the Caribbean and Africa Br J Cancer 1999 81 893 897 10555764 10.1038/sj.bjc.6690782
Svahn A Magnusson M Jagdahl L Schloss L Kahlmeter G Linde A Evaluation of three commercial enzyme-linked immunosorbent assays and two latex agglutination assays for diagnosis of primary Epstein-Barr virus infection Journal of Clinical Microbiology 1997 35 2728 2732 9350722
Buisson M Fleurent B Mak M Morand P Chan L Ng A Guan M Chin D Seigneurin JM Novel immunoblot assay using four recombinant antigens for diagnosis of Epstein-Barr virus primary infection and reactivation Journal of Clinical Microbiology 1999 37 2709 2714 10405428
Sixbey JW Yao QY Immunoglobulin A-induced shift of Epstein-Barr virus tissue tropism Science 1992 255 1578 1580 1312750
Henle W Henle G Andersson J Ernberg I Klein G Horwitz CA Marklund G Rymo L Wellinder C Straus SE Antibody responses to Epstein-Barr virus-determined nuclear antigen (EBNA)-1 and EBNA-2 in acute and chronic Epstein-Barr virus infection Proceedings of the National Academy of Sciences of the United States of America 1987 84 570 574 3025881
Meij P Vervoort MB Aarbiou J van Dissel P Brink A Bloemena E Meijer CJ Middeldorp JM Restricted low-level human antibody responses against Epstein-Barr virus (EBV)-encoded latent membrane protein 1 in a subgroup of patients with EBV-associated diseases Journal of Infectious Diseases 1999 179 1108 1115 10191211 10.1086/314704
Arvin AM Koropchak CM Immunoglobulins M and G to varicella-zoster virus measured by solid-phase radioimmunoassay: antibody responses to varicella and herpes zoster infections Journal of Clinical Microbiology 1980 12 367 374 6260833
Dialyna IA Graham D Rezaee R Blue CE Stavrianeas NG Neisters HG Spandidos DA Blackbourn DJ Anti-HHV-8/KSHV antibodies in infected individuals inhibit infection in vitro AIDS 2004 18 1263 1270 15362658 10.1097/00002030-200406180-00004
Kimball LE Casper C Koelle DM Morrow R Corey L Vieira J Reduced levels of neutralizing antibodies to Kaposi sarcoma-associated herpesvirus in persons with a history of Kaposi sarcoma Journal of Infectious Diseases 2004 189 2016 2022 15143468 10.1086/386371
Inoue N Spira T Lam L Corchero JL Luo W Comparison of serologic responses between Kaposi's sarcoma-positive and -negative men who were seropositive for both human herpesvirus 8 and human immunodeficiency virus Journal of Medical Virology 2004 74 202 206 15332267 10.1002/jmv.20167
Wang QJ Jenkins FJ Jacobson LP Kingsley LA Day RD Zhang ZW Meng YX Pellet PE Kousoulas KG Baghian A Primary human herpesvirus 8 infection generates a broadly specific CD8(+) T-cell response to viral lytic cycle proteins Blood 2001 97 2366 2373 11290599 10.1182/blood.V97.8.2366
Osman M Kubo T Gill J Neipel F Becker M Smith G Weiss R Gazzard B Boshoff C Gotch F Identification of human herpesvirus 8-specific cytotoxic T-cell responses J Virol 1999 73 6136 6140 10364372
Goudsmit J Renwick N Dukers NH Coutinho RA Heisterkamp S Bakker M Schulz TF Cornelissen M Weverling GJ Human herpesvirus 8 infections in the Amsterdam Cohort Studies (1984–1997): analysis of seroconversions to ORF65 and ORF73 Proc Natl Acad Sci U S A 2000 97 4838 4843 10781089 10.1073/pnas.97.9.4838
Sirianni MC Vincenzi L Topino S Giovannetti A Mazzetta F Libi F Scaramuzzi D Andreoni M Pinter E Baccarini S NK cell activity controls human herpesvirus 8 latent infection and is restored upon highly active antiretroviral therapy in AIDS patients with regressing Kaposi's sarcoma European Journal of Immunology 2002 32 2711 2720 12355422 10.1002/1521-4141(2002010)32:10<2711::AID-IMMU2711>3.0.CO;2-3
Monini P Colombini S Sturzl M Goletti D Cafaro A Sgadari C Butto S Franco M Leone P Fais S Reactivation and persistence of human herpesvirus-8 infection in B cells and monocytes by Th-1 cytokines increased in Kaposi's sarcoma Blood 1999 93 4044 4058 10361101
Luppi M Barozzi P Schulz TF Trovato R Donelli A Narni F Sheldon J Marasca R Torelli G Nonmalignant disease associated with human herpesvirus 8 reactivation in patients who have undergone autologous peripheral blood stem cell transplantation Blood 2000 96 2355 2357 11001882
LaDuca JR Love JL Abbott LZ Dube S Freidman-Kien AE Poiesz BJ Detection of human herpesvirus 8 DNA sequences in tissues and bodily fluids J Infect Dis 1998 178 1610 1615 9815212 10.1086/314514
Koelle DM Huang ML Chandran B Vieira J Piepkorn M Corey L Frequent detection of Kaposi's sarcoma-associated herpesvirus (human herpesvirus 8) DNA in saliva of human immunodeficiency virus-infected men: clinical and immunologic correlates J Infect Dis 1997 176 94 102 9207354
Howard MR Whitby D Bahadur G Suggett F Boshoff C Tenant-Flowers M Schulz TF Kirk S Matthews S Weller IV Detection of human herpesvirus 8 DNA in semen from HIV-infected individuals but not healthy semen donors AIDS 1997 11 F15 19 9030358 10.1097/00002030-199702000-00001
Grandadam M Dupin N Calvez V Gorin I Blum L Kernbaum S Sicard D Buisson Y Agut H Escande JP Exacerbations of clinical symptoms in human immunodeficiency virus type 1-infected patients with multicentric Castleman's disease are associated with a high increase in Kaposi's sarcoma herpesvirus DNA load in peripheral blood mononuclear cells J Infect Dis 1997 175 1198 1201 9129085
Harrington WJ JrBagasra O Sosa CE Bobroski LE Baum M Wen XL Cabral L Byrne GE Pomerantz RJ Wood C Human herpesvirus type 8 DNA sequences in cell-free plasma and mononuclear cells of Kaposi's sarcoma patients J Infect Dis 1996 174 1101 1105 8896516
Whitby D Howard MR Tenant-Flowers M Brink NS Copas A Boshoff C Hatzioannou T Suggett FE Aldam DM Denton AS Detection of Kaposi sarcoma associated herpesvirus in peripheral blood of HIV-infected individuals and progression to Kaposi's sarcoma Lancet 1995 346 799 802 7674745 10.1016/S0140-6736(95)91619-9
Pauk J Huang ML Brodie SJ Wald A Koelle DM Schacker T Celum C Selke S Corey L Mucosal shedding of human herpesvirus 8 in men N Engl J Med 2000 343 1369 1377 11070101 10.1056/NEJM200011093431904
Camera Pierrotti L Masami Sumita L Santos Freire W Hehl Caiaffa Filho H Akico Ueda Fick de Souza V Detection of human herpesvirus 8 DNA and antibodies to latent nuclear and lytic-phase antigens in serial samples from aids patients with Kaposi's sarcoma J Clin Virol 2000 16 247 251 10738143 10.1016/S1386-6532(99)00078-5
Blackbourn DJ Ambroziak J Lennette E Adams M Ramachandran B Levy JA Infectious human herpesvirus 8 in a healthy North American blood donor Lancet 1997 349 609 611 9057733 10.1016/S0140-6736(96)10004-0
Simpson GR Schulz TF Whitby D Cook PM Boshoff C Rainbow L Howard MR Gao SJ Bohenzky RA Simmonds P Prevalence of Kaposi's sarcoma associated herpesvirus infection measured by antibodies to recombinant capsid protein and latent immunofluorescence antigen Lancet 1996 348 1133 1138 8888167 10.1016/S0140-6736(96)07560-5
Smith MS Bloomer C Horvat R Goldstein E Casparian JM Chandran B Detection of human herpesvirus 8 DNA in Kaposi's sarcoma lesions and peripheral blood of human immunodeficiency virus-positive patients and correlation with serologic measurements J Infect Dis 1997 176 84 93 9207353
Angeloni A Heston L Uccini S Sirianni MC Cottoni F Masala MV Cerimele D Lin SF Sun R Rigsby M High prevalence of antibodies to human herpesvirus 8 in relatives of patients with classic Kaposi's sarcoma from Sardinia J Infect Dis 1998 177 1715 1718 9607855
Campbell TB Borok M Gwanzura L HHV-8 peripheral-blood viral load and the titer of antibodies against HHV-8 [letter; comment] N Engl J Med 1999 341 1241 1242 10523163 10.1056/NEJM199910143411619
Moore PS Boshoff C Weiss RA Chang Y Molecular mimicry of human cytokine and cytokine response pathway genes by KSHV Science 1996 274 1739 1744 8939871 10.1126/science.274.5293.1739
Burger R Neipel F Fleckenstein B Savino R Ciliberto G Kalden JR Gramatzki M Human herpesvirus type 8 interleukin-6 homologue is functionally active on human myeloma cells Blood 1998 91 1858 1863 9490667
Staskus KA Sun R Miller G Racz P Jaslowski A Metroka C Brett-Smith H Haase AT Cellular tropism and viral interleukin-6 expression distinguish human herpesvirus 8 involvement in Kaposi's sarcoma, primary effusion lymphoma, and multicentric Castleman's disease J Virol 1999 73 4181 4187 10196314
Parravinci C Corbellino M Paulli M Magrini U Lazzarino M Moore PS Chang Y Expression of a virus-derived cytokine, KSHV vIL-6, in HIV-seronegative Castleman's disease Am J Pathol 1997 151 1517 1522 9403701
Coscoy L Ganem D A viral protein that selectively downregulates ICAM-1 and B7-2 and modulates T cell costimulation J Clin Invest 2001 107 1599 1606 11413168
Belanger C Gravel A Tomoiu A Janelle ME Gosselin J Tremblay MJ Flamand L Human herpesvirus 8 viral FLICE-inhibitory protein inhibits Fas-mediated apoptosis through binding and prevention of procaspase-8 maturation J Hum Virol 2001 4 62 73 11437316
Murphy PM Viral exploitation and subversion of the immune system through chemokine mimicry Nature Immunology 2001 2 116 122 11175803 10.1038/84214
Moore PS Chang Y Kaposi's sarcoma-associated herpesvirus immunoevasion and tumorigenesis: two sides of the same coin? Annual Review of Microbiology 2003 57 609 639 14527293 10.1146/annurev.micro.57.030502.090824
Boshoff C Endo Y Collins PD Takeuchi Y Reeves JD Schweickart VL Siani MA Sasaki T Williams TJ Gray PW Angiogenic and HIV-inhibitory functions of KSHV-encoded chemokines Science 1997 278 290 294 9323208 10.1126/science.278.5336.290
Stine JT Wood C Hill M Epp A Raport CJ Schweickart VL Endo Y Sasaki T Simmons G Boshoff C KSHV-encoded CC chemokine vMIP-III is a CCR4 agonist, stimulates angiogenesis, and selectively chemoattracts TH2 cells Blood 2000 95 1151 1157 10666184
Chung YH Means RE Choi JK Lee BS Jung JU Kaposi's sarcoma-associated herpesvirus OX2 glycoprotein activates myeloid-lineage cells to induce inflammatory cytokine production J Virol 2002 76 4688 4698 11967286 10.1128/JVI.76.10.4688-4698.2002
Martin JN Ganem DE Osmond DH Page-Shafer KA Macrae D Kedes DH Sexual transmission and the natural history of human herpesvirus 8 infection N Engl J Med 1998 338 948 954 9521982 10.1056/NEJM199804023381403
Lampinen TM Kulasingam S Min J Borok M Gwanzura L Lamb J Mahomed K Woelk GB Strand KB Bosch ML Detection of Kaposi's sarcoma-associated herpesvirus in oral and genital secretions of Zimbabwean women J Infect Dis 2000 181 1785 1790 10823785 10.1086/315426
Cannon MJ Dollard SC Smith DK Klein RS Schuman P Rich JD Vlahov D Pellett PE Group HIVERS Blood-borne and sexual transmission of human herpesvirus 8 in women with or at risk for human immunodeficiency virus infection N Engl J Med 2001 344 637 643 11228278 10.1056/NEJM200103013440904
Lefrere JJ Mariotti M Girot R Loiseau P Herve P Transfusional risk of HHV-8 infection [letter; comment] Lancet 1997 350 217 9250205 10.1016/S0140-6736(05)62382-3
Engels EA Eastman H Ablashi DV Wilks RJ Braham J Manns A Risk of transfusion-associated transmission of human herpesvirus 8 J Natl Cancer Inst 1999 91 1773 1775 10528029 10.1093/jnci/91.20.1773
Pellett PE Wright DJ Engels EA Ablashi DV Dollard SC Forghani B Glynn SA Goedert JJ Jenkins FJ Lee TH Multicenter comparison of serologic assays and estimation of human herpesvirus 8 seroprevalence among US blood donors Transfusion 2003 43 1260 1268 12919429 10.1046/j.1537-2995.2003.00490.x
Goedert JJ Charurat M Blattner WA Hershow RC Pitt J Diaz C Mofenson LM Green K Minkoff H Paul ME Risk factors for Kaposi's sarcoma-associated herpesvirus infection among HIV-1-infected pregnant women in the USA Aids 2003 17 425 433 12556697 10.1097/00002030-200302140-00017
Davidovici B Karakis I Bourboulia D Ariad S Zong J Benharroch D Dupin N Weiss R Hayward G Sarov B Seroepidemiology and molecular epidemiology of Kaposi's sarcoma-associated herpesvirus among Jewish population groups in Israel J Natl Cancer Inst 2001 93 194 202 11158187 10.1093/jnci/93.3.194
Plancoulaine S Abel L van Beveren M Tregouet DA Joubert M Tortevoye P de The G Gessain A Human herpesvirus 8 transmission from mother to child and between siblings in an endemic population Lancet 2000 356 1062 1065 11009141 10.1016/S0140-6736(00)02729-X
Gessain A Mauclere P van Beveren M Plancoulaine S Ayouba A Essame-Oyono JL Martin PM de The G Human herpesvirus 8 primary infection occurs during childhood in Cameroon, Central Africa Int J Cancer 1999 81 189 192 10188717 10.1002/(SICI)1097-0215(19990412)81:2<189::AID-IJC4>3.0.CO;2-E
Brayfield BP Phiri S Kankasa C Muyanga J Mantina H Kwenda G West JT Bhat G Marx DB Klaskala W Postnatal human herpesvirus 8 and human immunodeficiency virus type 1 infection in mothers and infants from Zambia Journal of Infectious Diseases 2003 187 559 568 12599072 10.1086/367985
Brayfield BP Kankasa C West JT Muyanga J Bhat G Klaskala W Mitchell CD Wood C Distribution of Kaposi sarcoma-associated herpesvirus/human herpesvirus 8 in maternal saliva and breast milk in Zambia: implications for transmission Journal of Infectious Diseases 2004 189 2260 2270 15181574 10.1086/421119
Dedicoat M Newton R Alkharsah KR Sheldon J Szabados I Ndlovu B Page T Casabonne D Gilks CF Cassol SA Mother-to-child transmission of human herpesvirus-8 in South Africa Journal of Infectious Diseases 2004 190 1068 1075 15319855 10.1086/423326
Coluzzi M Calabro ML Manno D Chieco-Bianchi L Schulz TF Ascoli V Reduced seroprevalence of Kaposi's sarcoma-associated herpesvirus (KSHV), human herpesvirus 8 (HHV8), related to suppression of Anopheles density in Italy Medical & Veterinary Entomology 2003 17 461 464 14651663 10.1111/j.1365-2915.2003.00465.x
Ho M Suwansirikul S Dowling JN Youngblood LA Armstrong JA The transplanted kidney as a source of cytomegalovirus infection New England Journal of Medicine 1975 293 1109 1112 171567
Penn I Kaposi's sarcoma in transplant recipients Transplantation 1997 64 669 673 9311700 10.1097/00007890-199709150-00001
Parravicini C Olsen SJ Capra M Poli F Sirchia G Gao SJ Berti E Nocera A Rossi E Bestetti G Risk of Kaposi's sarcoma-associated herpes virus transmission from donor allografts among Italian posttransplant Kaposi's sarcoma patients Blood 1997 90 2826 2829 9326251
Rady PL Yen A Rollefson JL Orengo I Bruce S Hughes TK Tyring SK Herpesvirus-like DNA sequences in non-Kaposi's sarcoma skin lesions of transplant patients Lancet 1995 345 1339 1340 7752756 10.1016/S0140-6736(95)92538-4
Regamey N Tamm M Wernli M Witschi A Thiel G Cathomas G Erb P Transmission of human herpesvirus 8 infection from renal-transplant donors to recipients N Engl J Med 1998 339 1358 1363 9801396 10.1056/NEJM199811053391903
Luppi M Barozzi P Santagostino G Trovato R Schulz TF Marasca R Bottalico D Bignardi L Torelli G Molecular evidence of organ-related transmission of Kaposi sarcoma-associated herpesvirus or human herpesvirus-8 in transplant patients Blood 2000 96 3279 3281 11050015
Sarid R Pizov G Rubinger D Backenroth R Friedlaender MM Schwartz F Wolf DG Detection of human herpesvirus-8 DNA in kidney allografts prior to the development of Kaposi's sarcoma Clin Infect Dis 2001 32 1502 1505 11317254 10.1086/320153
Munoz P Alvarez P de Ory F Pozo F Rivera M Bouza E Incidence and clinical characteristics of Kaposi sarcoma after solid organ transplantation in Spain: importance of seroconversion against HHV-8 Medicine 2002 81 293 304 12169884 10.1097/00005792-200207000-00005
Jenkins FJ Hoffman LJ Liegey-Dougall A Reactivation of and primary infection with human herpesvirus 8 among solid-organ transplant recipients Journal of Infectious Diseases 2002 185 1238 1243 12001040 10.1086/340237
Andreoni M Goletti D Pezzotti P Pozzetto A Monini P Sarmati L Farchi F Tisone G Piazza A Pisani F Prevalence, incidence and correlates of HHV-8/KSHV infection and Kaposi's sarcoma in renal and liver transplant recipients J Infect 2001 43 195 199 11798259 10.1053/jinf.2001.0899
Hudnall SD Rady PL Tyring SK Fish JC Serologic and molecular evidence of human herpesvirus 8 activation in renal transplant recipients J Infect Dis 1998 178 1791 1794 9815236 10.1086/314482
Gray JJ Avidity of EBV VCA-specific IgG antibodies: distinction between recent primary infection, past infection and reactivation Journal of Virological Methods 1995 52 95 104 7769043 10.1016/0166-0934(94)00147-9
Barozzi P Luppi M Facchetti F Mecucci C Alu M Sarid R Rasini V Ravazzini L Rossi E Festa S Post-transplant Kaposi sarcoma originates from the seeding of donor-derived progenitors comment erratum appears in Nat Med 2003 Jul;9(7):975 Nature Medicine 2003 9 554 561 12692543 10.1038/nm862
Moosa MR Treurnicht FK van Rensburg EJ Schneider JW Jordaan HF Engelbrecht S Detection and subtyping of human herpesvirus-8 in renal transplant patients before and after remission of Kaposi's sarcoma Transplantation 1998 66 214 218 9701267 10.1097/00007890-199807270-00013
Browning PJ Sechler JM Kaplan M Washington RH Gendelman R Yarchoan R Ensoli B Gallo RC Identification and culture of Kaposi's sarcoma-like spindle cells from the peripheral blood of human immunodeficiency virus-1-infected individuals and normal controls Blood 1994 84 2711 2720 7522639
Kapelushnik J Ariad S Benharroch D Landau D Moser A Delsol G Brousset P Post renal transplantation human herpesvirus 8-associated lymphoproliferative disorder and Kaposi's sarcoma Br J Haematol 2001 113 425 428 11380409 10.1046/j.1365-2141.2001.02740.x
Luppi M Barozzi P Schulz TF Setti G Staskus K Trovato R Narni F Donelli A Maiorana A Marasca R Bone marrow failure associated with human herpesvirus 8 infection after transplantation N Engl J Med 2000 343 1378 1385 11070102 10.1056/NEJM200011093431905
Kaposi M Idiopathic multiple pigmented sarcoma of the skin Archiv fur Dermatologie und Syphilis 1872 4 265 273 10.1007/BF01830024
Antman K Chang Y Kaposi's sarcoma N Engl J Med 2000 342 1027 1038 10749966 10.1056/NEJM200004063421407
Ablashi DV Chatlynne LG Whitman JE Cesarman E Spectrum of Kaposi's sarcoma-associated herpesvirus, or human herpesvirus 8, diseases Clinical Microbiology Reviews 2002 15 439 464 12097251 10.1128/CMR.15.3.439-464.2002
Fredericks DN Relman DA Sequence-based identification of microbial pathogens: a reconsideration of Koch's postulates Clin Microbiol Rev 1996 9 18 33 8665474
Hill AB The environment and disease: Association or causation? Proceedings of the Royal Society of Medicine 1965 58 295 300 14283879
Tappero JW Conant MA Wolfe SF Berger TG Kaposi's sarcoma. Epidemiology, pathogenesis, histology, clinical spectrum, staging criteria and therapy J Am Acad Dermatol 1993 28 371 395 8445054
Buonaguro FM Tornesello ML Beth-Giraldo E Hatzakis A Mueller N Downing R Biryamwaho B Sempala SD Giraldo G Herpesvirus-like DNA sequences detected in endemic, classic, iatrogenic and epidemic Kaposi's sarcoma (KS) biopsies Int J Cancer 1996 65 25 28 8543391 10.1002/(SICI)1097-0215(19960103)65:1<25::AID-IJC5>3.0.CO;2-3
Chatlynne LG Ablashi DV Seroepidemiology of Kaposi's sarcoma-associated herpesvirus (KSHV) Semin Cancer Biol 1999 9 175 185 10343069 10.1006/scbi.1998.0089
Davis DA Humphrey RW Newcomb FM O'Brien TR Goedert JJ Straus SE Yarchoan R Detection of serum antibodies to a Kaposi's sarcoma-associated herpesvirus-specific peptide J Infect Dis 1997 175 1071 1079 9129068
Gao SJ Kingsley L Hoover DR Spira TJ Rinaldo CR Saah A Phair J Detels R Parry P Chang Y Seroconversion to antibodies against Kaposi's sarcoma-associated herpesvirus-related latent nuclear antigens before the development of Kaposi's sarcoma N Engl J Med 1996 335 233 241 8657239 10.1056/NEJM199607253350403
Flore O Rafii S Ely S O'Leary JJ Hyjek EM Cesarman E Transformation of primary human endothelial cells by Kaposi's sarcoma-associated herpesvirus Nature 1998 394 588 592 9707121 10.1038/29093
Iscovich J Boffetta P Franceschi S Azizi E Sarid R Classic kaposi sarcoma: epidemiology and risk factors Cancer 2000 88 500 517 10649240 10.1002/(SICI)1097-0142(20000201)88:3<500::AID-CNCR3>3.0.CO;2-9
Hengge UR Ruzicka T Tyring SK Stuschke M Roggendorf M Schwartz RA Seeber S Update on Kaposi's sarcoma and other HHV8 associated diseases Part 1: epidemiology, environmental predispositions, clinical manifestations, and therapy The Lancet Infectious Diseases 2002 2 281 292 12062994 10.1016/S1473-3099(02)00263-3
Brenner B Weissmann-Brenner A Rakowsky E Weltfriend S Fenig E Friedman-Birnbaum R Sulkes A Linn S Classical Kaposi sarcoma: prognostic factor analysis of 248 patients Cancer 2002 95 1982 1987 12404293 10.1002/cncr.10907
Goedert JJ Vitale F Lauria C Serraino D Tamburini M Montella M Messina A Brown EE Rezza G Gafa L Risk factors for classical Kaposi's sarcoma Journal of the National Cancer Institute 2002 94 1712 1718 12441327
Goedert JJ The epidemiology of acquired immunodeficiency syndrome malignancies Semin Oncol 2000 27 390 401 10950365
Schwartz RA Kaposi's sarcoma: advances and perspectives J Am Acad Dermatol 1996 34 804 814 8632078 10.1016/S0190-9622(96)90018-3
Jacobson LP Jenkins FJ Springer G Munoz A Shah KV Phair J Zhang Z Armenian H Interaction of human immunodeficiency virus type 1 and human herpesvirus type 8 infections on the incidence of Kaposi's sarcoma J Infect Dis 2000 181 1940 1949 10837173 10.1086/315503
Renwick N Halaby T Weverling GJ Dukers NH Simpson GR Coutinho RA Lange JM Schulz TF Goudsmit J Seroconversion for human herpesvirus 8 during HIV infection is highly predictive of Kaposi's sarcoma AIDS 1998 12 2481 2488 9875587 10.1097/00002030-199818000-00018
Grossman Z Iscovich J Schwartz F Azizi E Klepfish A Schattner A Sarid R Absence of Kaposi sarcoma among Ethiopian immigrants to Israel despite high seroprevalence of human herpesvirus 8 Mayo Clinic Proceedings 2002 77 905 909 12233922
Dourmishev LA Dourmishev AL Palmeri D Schwartz RA Lukac DM Molecular genetics of Kaposi's sarcoma-associated herpesvirus (human herpesvirus-8) epidemiology and pathogenesis Microbiology & Molecular Biology Reviews 2003 67 175 212 table of contents. 12794189 10.1128/MMBR.67.2.175-212.2003
Wabinga HR Parkin DM Wabwire-Mangen F Mugerwa JW Cancer in Kampala, Uganda, in 1989–91: changes in incidence in the era of AIDS International Journal of Cancer 1993 54 26 36
Chang Y Ziegler J Wabinga H Katangole-Mbidde E Boshoff C Schulz T Whitby D Maddalena D Jaffe HW Weiss RA Kaposi's sarcoma-associated herpesvirus and Kaposi's sarcoma in Africa. Uganda Kaposi's Sarcoma Study Group Arch Intern Med 1996 156 202 204 8546554 10.1001/archinte.156.2.202
Mayama S Cuevas LE Sheldon J Omar OH Smith DH Okong P Silvel B Hart CA Schulz TF Prevalence and transmission of Kaposi's sarcoma-associated herpesvirus (human herpesvirus 8) in Ugandan children and adolescents Int J Cancer 1998 77 817 820 9714046 10.1002/(SICI)1097-0215(19980911)77:6<817::AID-IJC2>3.0.CO;2-X
Ariyoshi K Schim van der Loeff M Cook P Whitby D Corrah T Jaffar S Cham F Sabally S O'Donovan D Weiss RA Kaposi's sarcoma in the Gambia, West Africa is less frequent in human immunodeficiency virus type 2 than in human immunodeficiency virus type 1 infection despite a high prevalence of human herpesvirus 8 J Hum Virol 1998 1 193 199 10195242
Hudnall SD Rady PL Tyring SK Fish JC Hydrocortisone activation of human herpesvirus 8 viral DNA replication and gene expression in vitro Transplantation 1999 67 648 652 10096517 10.1097/00007890-199903150-00002
Nador RG Cesarman E Chadburn A Dawson DB Ansari MQ Sald J Knowles DM Primary effusion lymphoma: a distinct clinicopathologic entity associated with the Kaposi's sarcoma-associated herpes virus Blood 1996 88 645 656 8695812
Arvanitakis L Mesri EA Nador RG Said JW Asch AS Knowles DM Cesarman E Establishment and characterization of a primary effusion (body cavity-based) lymphoma cell line (BC-3) harboring kaposi's sarcoma-associated herpesvirus (KSHV/HHV-8) in the absence of Epstein-Barr virus Blood 1996 88 2648 2654 8839859
Boshoff C Gao SJ Healy LE Matthews S Thomas AJ Coignet L Warnke RA Strauchen JA Matutes E Kamel OW Establishing a KSHV+ cell line (BCP-1) from peripheral blood and characterizing its growth in Nod/SCID mice Blood 1998 91 1671 1679 9473233
Cesarman E Moore PS Rao PH Inghirami G Knowles DM Chang Y In vitro establishment and characterization of two acquired immunodeficiency syndrome-related lymphoma cell lines (BC-1 and BC-2) containing Kaposi's sarcoma-associated herpesvirus-like (KSHV) DNA sequences Blood 1995 86 2708 2714 7670109
Drexler HG Uphoff CC Gaidano G Carbone A Lymphoma cell lines: in vitro models for the study of HHV-8+ primary effusion lymphomas (body cavity-based lymphomas) Leukemia 1998 12 1507 1517 9766492 10.1038/sj.leu.2401160
Gaidano G Cechova K Chang Y Moore PS Knowles DM Dalla-Favera R Establishment of AIDS-related lymphoma cell lines from lymphomatous effusions Leukemia 1996 10 1237 1240 8684008
Mbulaiteye SM Biggar RJ Goedert JJ Engels EA Pleural and peritoneal lymphoma among people with AIDS in the United States Journal of Acquired Immune Deficiency Syndromes: JAIDS 2002 29 418 421
Ascoli V Lo Coco F Torelli G Vallisa D Cavanna L Bergonzi C Luppi M Human herpesvirus 8-associated primary effusion lymphoma in HIV–patients: a clinicopidemiologic variant resembling classic Kaposi's sarcoma Haematologica 2002 87 339 343 11940475
Soulier J Grollet L Oksenhendler E Miclea JM Cacoub P Baruchel A Brice P Clauvel JP d'Agay MF Raphael M Molecular analysis of clonality in Castleman's disease Blood 1995 86 1131 1138 7620166
Katano H Sato Y Kurata T Mori S Sata T Expression and localization of human herpesvirus 8-encoded proteins in primary effusion lymphoma, Kaposi's sarcoma, and multicentric Castleman's disease Virology 2000 269 335 344 10753712 10.1006/viro.2000.0196
Soulier J Grollet L Oksenhendler E Cacoub P Cazals-Hatem D Babinet P d'Agay MF Clauvel JP Raphael M Degos L Kaposi's sarcoma-associated herpesvirus-like DNA sequences in multicentric Castleman's disease Blood 1995 86 1276 1280 7632932
Marcoval J Mana J Moreno A Gallego I Fortuno Y Peyri J Foreign bodies in granulomatous cutaneous lesions of patients with systemic sarcoidosis Archives of Dermatology 2001 137 427 430 11295921
Di Alberti L Piattelli A Artese L Favia G Patel S Saunders N Porter SR Scully CM Ngui SL Teo CG Human herpesvirus 8 variants in sarcoid tissues Lancet 1997 350 1655 1661 9400509 10.1016/S0140-6736(97)10102-7
Belec L Mohamed AS Lechapt-Zalcman E Authier FJ Lange F Gherardi RK Lack of HHV-8 DNA sequences in sarcoid tissues of French patients [letter] Chest 1998 114 948 949 9743199
Lebbe C Agbalika F Flageul B Pellet C Rybojad M Cordoliani F Farge D Vignon-Pennamen MD Sheldon J Morel P No evidence for a role of human herpesvirus type 8 in sarcoidosis: molecular and serological analysis Br J Dermatol 1999 141 492 496 10583053 10.1046/j.1365-2133.1999.03043.x
Mendelson J Howley P Israel M Liotta L The Molecular Basis of Cancer 1995 Philadelphia, PA: WB Saunders Company
Kawano M Hirano T Matsuda T Taga T Horii Y Iwato K Asaoku H Tang B Tanabe O Tanaka H Autocrine generation and requirement of BSF-2/IL-6 for human multiple myelomas Nature 1988 332 83 85 3258060 10.1038/332083a0
Lichtenstein A Berenson J Norman D Chang MP Carlile A Production of cytokines by bone marrow cells obtained from patients with multiple myeloma Blood 1989 74 1266 1273 2475182
Rettig MB Ma HJ Vescio RA Pold M Schiller G Belson D Savage A Nishikubo C Wu C Fraser J Kaposi's sarcoma-associated herpesvirus infection of bone marrow dendritic cells from multiple myeloma patients Science 1997 276 1851 1854 9188529 10.1126/science.276.5320.1851
Said JW Rettig MR Heppner K Vescio RA Schiller G Ma HJ Belson D Savage A Shintaku IP Koeffler HP Localization of Kaposi's sarcoma-associated herpesvirus in bone marrow biopsy samples from patients with multiple myeloma Blood 1997 90 4278 4282 9373238
Gao SJ Alsina M Deng JH Harrison CR Montalvo EA Leach CT Roodman GD Jenson HB Antibodies to Kaposi's sarcoma-associated herpesvirus (human herpesvirus 8) in patients with multiple myeloma J Infect Dis 1998 178 846 849 9728556
Whitby D Boshoff C Luppi M Torelli G Kaposi's sarcoma-associated herpesvirus infection and multiple myeloma [letter; comment] Science 1997 278 1971 1972 discussion 1972–1973. 9417644
MacKenzie J Sheldon J Morgan G Cook G Schulz TF Jarrett RF HHV-8 and multiple myeloma in the UK [letter] Lancet 1997 350 1144 1145 9343509 10.1016/S0140-6736(05)63792-0
Parravicini C Lauri E Baldini L Neri A Poli F Sirchia G Moroni M Galli M Corbellino M Kaposi's sarcoma-associated herpesvirus infection and multiple myeloma [letter; comment] Science 1997 278 1969 1970 9417642 10.1126/science.278.5345.1969
Masood R Zheng T Tupule A Arora N Chatlynne L Handy M Whitman J Jr Kaposi's sarcoma-associated herpesvirus infection and multiple myeloma [letter; comment] Science 1997 278 1970 1971 9417643
Carletti F Mandolini C Rossi A Capobianchi MR Borgia MC Prevalence of human herpesvirus (HHV)-8 infection among carriers of cardiovascular disease Journal of Biological Regulators & Homeostatic Agents 2002 16 110 113 12144122
Memar OM Rady PL Goldblum RM Yen A Tyring SK Human herpesvirus 8 DNA sequences in blistering skin from patients with pemphigus Arch Dermatol 1997 133 1247 1251 9382563 10.1001/archderm.133.10.1247
Du MQ Diss TC Liu H Ye H Hamoudi RA Cabecadas J Dong HY Harris NL Chan JK Rees JW KSHV- and EBV-associated germinotropic lymphoproliferative disorder Blood 2002 100 3415 3418 12384445 10.1182/blood-2002-02-0487
Gomez-Brouchet A Delisle MB Vital A Brousset P Immunohistochemical assessment of human herpesvirus 8 infection in primary central nervous system large B cell lymphomas J Clin Pathol 2001 54 617 618 11477117 10.1136/jcp.54.8.617
Kedes DH Ganem D Sensitivity of Kaposi's sarcoma-associated herpesvirus replication to antiviral drugs. Implications for potential therapy J Clin Invest 1997 99 2082 2086 9151779
Simonart T Noel JC De Dobbeleer G Parent D Van Vooren JP De Clercq E Snoeck R Treatment of classical Kaposi's sarcoma with intralesional injections of cidofovir: report of a case J Med Virol 1998 55 215 218 9624609 10.1002/(SICI)1096-9071(199807)55:3<215::AID-JMV6>3.0.CO;2-4
Guillot B Carles MJ Foulongne V Dereure O Ramuz M Lack of modification of virological status after chemotherapy or radiotherapy for classic Kaposi's sarcoma Br J Dermatol 2002 146 337 338 11903258 10.1046/j.1365-2133.2002.4653_7.x
Jones JL Hanson DL Dworkin MS Jaffe HW Incidence and trends in Kaposi's sarcoma in the era of effective antiretroviral therapy J Acquir Immune Defic Syndr 2000 24 270 274 10969352
Tam HK Zhang ZF Jacobson LP Margolick JB Chmiel JS Rinaldo C Detels R Effect of highly active antiretroviral therapy on survival among HIV-infected men with Kaposi sarcoma or non-Hodgkin lymphoma Int J Cancer 2002 98 916 922 11948473 10.1002/ijc.10274
Paparizos VA Kyriakis KP Papastamopoulos V Hadjivassiliou M Stavrianeas NG Response of AIDS-associated Kaposi sarcoma to highly active antiretroviral therapy alone JAIDS Journal of Acquired Immune Deficiency Syndromes 2002 30 257 258
Rooney CM Smith CA Ng CY Loftin S Li C Krance RA Brenner MK Heslop HE Use of gene-modified virus-specific T lymphocytes to control Epstein-Barr-virus-related lymphoproliferation Lancet 1995 345 9 13 7799740 10.1016/S0140-6736(95)91150-2
Gao SJ Kingsley L Li M Zheng W Parravicini C Ziegler J Newton R Rinaldo CR Saah A Phair J KSHV antibodies among Americans, Italians and Ugandans with and without Kaposi's sarcoma Nat Med 1996 2 925 928 8705864 10.1038/nm0896-925
Baillargeon J Deng J Hettler E Harrison C Grady JJ Korte LG Alexander J Montalvo E Jenson HB Gao S Seroprevalence of Kaposi's Sarcoma-Associated Herpesvirus Infection among Blood Donors from Texas Ann Epidemiol 2001 11 512 518 11557184 10.1016/S1047-2797(01)00242-3
Atkinson J Edlin BR Engels EA Kral AH Seal K Gamache CJ Whitby D O'Brien TR Seroprevalence of human herpesvirus 8 among injection drug users in San Francisco Journal of Infectious Diseases 2003 187 974 981 12660944 10.1086/368332
O'Brien TR Kedes D Ganem D Macrae DR Rosenberg PS Molden J Goedert JJ Evidence for concurrent epidemics of human herpesvirus 8 and human immunodeficiency virus type 1 in US homosexual men: rates, risk factors, and relationship to Kaposi's sarcoma J Infect Dis 1999 180 1010 1017 10479125 10.1086/315039
Engels EA Clark E Aledort LM Goedert JJ Whitby D Kaposi's sarcoma-associated herpesvirus infection in elderly Jews and non-Jews from New York City International Journal of Epidemiology 2002 31 946 950 12435765 10.1093/ije/31.5.946
Baillargeon J Leach CT Deng JH Gao SJ Jenson HB High prevalence of human herpesvirus 8 (HHV-8) infection in south Texas children Journal of Medical Virology 2002 67 542 548 12116002 10.1002/jmv.10136
Delorme S Houde I Deschenes L Seroprevalence of antibodies against human herpesvirus 8 in a population of renal transplant recipients at Hotel-Dieu de Quebec Hospital Journal of Clinical Microbiology 2003 41 5207 5208 14605164 10.1128/JCM.41.11.5207-5208.2003
Manns A Strickler HD Hanchard B Manassaram DM Waters D Ablashi DV Age- and sex-specific seroprevalence of human herpesvirus 8 in Jamaica [letter; comment] J Natl Cancer Inst 1998 90 1102 1104 9672261 10.1093/jnci/90.14.1102
Kouri V Eng SM Rodriguez ME Resik S Orraca O Moore PS Chang Y Seroprevalence of Kaposi's sarcoma-associated herpesvirus in various populations in Cuba Pan American Journal of Public Health 2004 15 320 325 15231079
Sosa C Klaskala W Chandran B Soto R Sieczkowski L Wu MH Baum M Wood C Human herpesvirus 8 as a potential sexually transmitted agent in Honduras J Infect Dis 1998 178 547 551 9697741
Biggar RJ Whitby D Marshall V Linhares AC Black F Human herpesvirus 8 in Brazilian Amerindians: a hyperendemic population with a new subtype J Infect Dis 2000 181 1562 1568 10823754 10.1086/315456
Freitas RB Freitas MR Linhares AC Prevalence of human herpesvirus 8 antibodies in the population of Belem, Para, Brazil Revista do Instituto de Medicina Tropical de Sao Paulo 2002 44 309 313 12532213
Whitby D Marshall VA Bagni RK Wang CD Gamache CJ Guzman JR Kron M Ebbesen P Biggar RJ Genotypic characterization of Kaposi's sarcoma-associated herpesvirus in asymptomatic infected subjects from isolated populations Journal of General Virology 2004 85 155 163 14718630 10.1099/vir.0.19465-0
Perez C Tous M Gallego S Zala N Rabinovich O Garbiero S Martinez MJ Cunha AM Camino S Camara A Seroprevalence of human herpesvirus-8 in blood donors from different geographical regions of Argentina, Brazil, and Chile Journal of Medical Virology 2004 72 661 667 14981770 10.1002/jmv.20029
Sosa C Benetucci J Hanna C Sieczkowski L Deluchi G Canizal AM Mantina H Klaskala W Baum M Wood C Human herpesvirus 8 can be transmitted through blood in drug addicts Medicina (Mex) 2001 61 291 294
Keller R Zago A Viana MC Bourboulia D Desgranges C Casseb J Moura WV Dietze R Collandre H HHV-8 infection in patients with AIDS-related Kaposi's sarcoma in Brazil Brazilian Journal of Medical & Biological Research 2001 34 879 886 11449306
Juhasz A Remenyik E Konya J Veress G Begany A Andirko I Medgyessy I Hunyadi J Gergely L Prevalence and age distribution of human herpesvirus-8 specific antibodies in Hungarian blood donors J Med Virol 2001 64 526 530 11468739 10.1002/jmv.1081
Regamey N Cathomas G Schwager M Wernli M Harr T Erb P High human herpesvirus 8 seroprevalence in the homosexual population in Switzerland J Clin Microbiol 1998 36 1784 1786 9620422
Gambus G Bourboulia D Esteve A Lahoz R Rodriguez C Bolao F Sirera G Muga R del Romero J Boshoff C Prevalence and distribution of HHV-8 in different subpopulations, with and without HIV infection, in Spain AIDS 2001 15 1167 1174 11416719 10.1097/00002030-200106150-00012
Preiser W Szep NI Lang D Doerr HW Rabenau HF Kaposi's sarcoma-associated herpesvirus seroprevalence in selected german patients: evaluation by different test systems Medical Microbiology & Immunology 2001 190 121 127 11827200
Graffeo R Ranno S Marchetti S Capodicasa N Schito AM Fuga L Amico R Cattani P Fadda G HHV 8 seroprevalence and transmission within Albanian family groups New Microbiologica 2003 26 1 6 12578305
Janier M Agbalika F de La Salmoniere P Lassau F Lagrange P Morel P Human herpesvirus 8 seroprevalence in an STD clinic in Paris: a study of 512 patients Sexually Transmitted Diseases 2002 29 698 702 12438907 10.1097/00007435-200211000-00013
Whitby D Luppi M Barozzi P Boshoff C Weiss RA Torelli G Human herpesvirus 8 seroprevalence in blood donors and lymphoma patients from different regions of Italy J Natl Cancer Inst 1998 90 395 397 9498490 10.1093/jnci/90.5.395
Rezza G Lennette ET Giuliani M Pezzotti P Caprilli F Monini P Butto S Lodi G Di Carlo A Levy JA Prevalence and determinants of anti-lytic and anti-latent antibodies to human herpesvirus-8 among Italian individuals at risk of sexually and parenterally transmitted infections Int J Cancer 1998 77 361 365 9663596 10.1002/(SICI)1097-0215(19980729)77:3<361::AID-IJC9>3.0.CO;2-M
Andreoni M El-Sawaf G Rezza G Ensoli B Nicastri E Ventura L Ercoli L Sarmati L Rocchi G High seroprevalence of antibodies to human herpesvirus-8 in Egyptian children: evidence of nonsexual transmission J Natl Cancer Inst 1999 91 465 469 10070947 10.1093/jnci/91.5.465
Calabro ML Sheldon J Favero A Simpson GR Fiore JR Gomes E Angarano G Chieco-Bianchi L Schulz TF Seroprevalence of Kaposi's sarcoma-associated herpesvirus/human herpesvirus 8 in several regions of Italy J Hum Virol 1998 1 207 213 10195244
Perna AM Bonura F Vitale F Viviano E Di Benedetto MA Ajello F Villafrate MR Prestileo T Mancuso S Goedert JJ Antibodies to human herpes virus type 8 (HHV8) in general population and in individuals at risk for sexually transmitted diseases in Western Sicily Int J Epidemiol 2000 29 175 179 10750620 10.1093/ije/29.1.175
Engels EA Sinclair MD Biggar RJ Whitby D Ebbesen P Goedert JJ Gastwirth JL Latent class analysis of human herpesvirus 8 assay performance and infection prevalence in sub-saharan Africa and Malta Int J Cancer 2000 88 1003 1008 11093828 10.1002/1097-0215(20001215)88:6<1003::AID-IJC26>3.0.CO;2-9
Iscovich J Fischbein A Fisher-Fischbein J Freedman LS Eng SM Boffetta P Vudovich A Glasman C Goldschmidt R Livingston M Seroprevalence of Kaposi's sarcoma-associated herpesvirus in healthy adults in Israel Anticancer Res 2000 20 2119 2122 10928163
Huang LM Huang SY Chen M Chao M Lu CY Tien HF Lee CY Jeang KT Geographical differences in human herpesvirus 8 seroepidemiology: a survey of 1,201 individuals in Asia J Med Virol 2000 60 290 293 10630961 10.1002/(SICI)1096-9071(200003)60:3<290::AID-JMV7>3.0.CO;2-G
Wang YF Lee SB Cheng LC Tai MH Su IJ Detection of serum antibodies to three different recombinant antigens of human herpesvirus 8 by immunoblotting: seroprevalence studies in Taiwan Clinica Chimica Acta 2002 320 37 42 10.1016/S0009-8981(02)00023-2
Dilnur P Katano H Wang ZH Osakabe Y Kudo M Sata T Ebihara Y Classic type of Kaposi's sarcoma and human herpesvirus 8 infection in Xinjiang, China Pathol Int 2001 51 845 852 11844050 10.1046/j.1440-1827.2001.01293.x
Ayuthaya PI Katano H Inagi R Auwanit W Sata T Kurata T Yamanishi K The seroprevalence of human herpesvirus 8 infection in the Thai population Southeast Asian Journal of Tropical Medicine & Public Health 2002 33 297 305 12236428
Fujii T Taguchi H Katano H Mori S Nakamura T Nojiri N Nakajima K Tadokoro K Juji T Iwamoto A Seroprevalence of human herpesvirus 8 in human immunodeficiency virus 1-positive and human immunodeficiency virus 1-negative populations in Japan J Med Virol 1999 57 159 162 9892401 10.1002/(SICI)1096-9071(199902)57:2<159::AID-JMV12>3.0.CO;2-A
Rezza G Danaya RT Wagner TM Sarmati L Owen IL Monini P Andreoni M Suligoi B Ensoli B Pozio E Human herpesvirus-8 and other viral infections, Papua New Guinea Emerging Infectious Diseases 2001 7 893 895 11747707
He J Bhat G Kankasa C Chintu C Mitchell C Duan W Wood C Seroprevalence of human herpesvirus 8 among Zambian women of childbearing age without Kaposi's sarcoma (KS) and mother-child pairs with KS J Infect Dis 1998 178 1787 1790 9815235 10.1086/314512
Sitas F Carrara H Beral V Newton R Reeves G Bull D Jentsch U Pacella-Norman R Bourboulia D Whitby D Antibodies against human herpesvirus 8 in black South African patients with cancer N Engl J Med 1999 340 1863 1871 10369849 10.1056/NEJM199906173402403
Constantine NT Fox E Abbatte EA Woody JN Diagnostic usefulness of five screening assays for HIV in an east African city where prevalence of infection is low AIDS 1989 3 313 317 2504236
Callahan JD Constantine NT Kataaha P Zhang X Hyams KC Bansal J Second generation hepatitis C virus assays: performance when testing African sera J Med Virol 1993 41 35 38 7693861
Hayward GS KSHV strains: the origins and global spread of the virus Semin Cancer Biol 1999 9 187 199 10343070 10.1006/scbi.1998.0116
Lee H Guo J Li M Choi JK DeMaria M Rosenzweig M Jung JU Identification of an immunoreceptor tyrosine-based activation motif of K1 transforming protein of Kaposi's sarcoma-associated herpesvirus Molecular & Cellular Biology 1998 18 5219 5228 9710606
Kasolo FC Monze M Obel N Anderson RA French C Gompels UA Sequence analyses of human herpesvirus-8 strains from both African human immunodeficiency virus-negative and -positive childhood endemic Kaposi's sarcoma show a close relationship with strains identified in febrile children and high variation in the K1 glycoprotein J Gen Virol 1998 79 3055 3065 9880022
Kedes DH Lagunoff M Renne R Ganem D Identification of the gene encoding the major latency-associated nuclear antigen of the Kaposi's sarcoma-associated herpesvirus J Clin Invest 1997 100 2606 2610 9366576
Kellam P Boshoff C Whitby D Matthews S Weiss RA Talbot SJ Identification of a major latent nuclear antigen, LNA-1, in the human herpesvirus 8 genome J Hum Virol 1997 1 19 29 10195227
Rainbow L Platt GM Simpson GR Sarid R Gao SJ Stoiber H Herrington CS Moore PS Schulz TF The 222- to 234-kilodalton latent nuclear protein (LNA) of Kaposi's sarcoma-associated herpesvirus (human herpesvirus 8) is encoded by orf73 and is a component of the latency-associated nuclear antigen J Virol 1997 71 5915 5921 9223481
Katano H Sato Y Kurata T Mori S Sata T High expression of HHV-8-encoded ORF73 protein in spindle-shaped cells of Kaposi's sarcoma Am J Pathol 1999 155 47 52 10393835
Dupin N Fisher C Kellam P Ariad S Tulliez M Franck N van Marck E Salmon D Gorin I Escande JP Distribution of human herpesvirus-8 latently infected cells in Kaposi's sarcoma, multicentric Castleman's disease, and primary effusion lymphoma Proc Natl Acad Sci U S A 1999 96 4546 4551 10200299 10.1073/pnas.96.8.4546
Edelman DC Ketema F Saville RD Herman J Sill AM Gill PS Blattner WA Constantine NT Specifics on the refinement and application of two serological assays for the detection of antibodies to HHV-8 J Clin Virol 2000 16 225 237 10738141 10.1016/S1386-6532(99)00085-2
Lin SF Sun R Heston L Gradoville L Shedd D Haglund K Rigsby M Miller G Identification, expression, and immunogenicity of Kaposi's sarcoma-associated herpesvirus-encoded small viral capsid antigen J Virol 1997 71 3069 3076 9060668
Pau CP Lam LL Spira TJ Black JB Stewart JA Pellett PE Respess RA Mapping and serodiagnostic application of a dominant epitope within the human herpesvirus 8 ORF 65-encoded protein J Clin Microbiol 1998 36 1574 1577 9620379
Nealon K Newcomb WW Pray TR Craik CS Brown JC Kedes DH Lytic replication of Kaposi's sarcoma-associated herpesvirus results in the formation of multiple capsid species: isolation and molecular characterization of A, B, and C capsids from a gammaherpesvirus J Virol 2001 75 2866 2878 11222712 10.1128/JVI.75.6.2866-2878.2001
Lo P Yu X Atanasov I Chandran B Zhou ZH Three-dimensional localization of pORF65 in Kaposi's sarcoma-associated herpesvirus capsid Journal of Virology 2003 77 4291 4297 12634386 10.1128/JVI.77.7.4291-4297.2003
Chandran B Bloomer C Chan SR Zhu L Goldstein E Horvat R Human herpesvirus-8 ORF K8.1 gene encodes immunogenic glycoproteins generated by spliced transcripts Virology 1998 249 140 149 9740785 10.1006/viro.1998.9316
Chandran B Smith MS Koelle DM Corey L Horvat R Goldstein E Reactivities of human sera with human herpesvirus-8-infected BCBL-1 cells and identification of HHV-8-specific proteins and glycoproteins and the encoding cDNAs Virology 1998 243 208 217 9527930 10.1006/viro.1998.9055
Raab MS Albrecht JC Birkmann A Yaguboglu S Lang D Fleckenstein B Neipel F The immunogenic glycoprotein gp35-37 of human herpesvirus 8 is encoded by open reading frame K8.1 J Virol 1998 72 6725 6731 9658120
Li M MacKey J Czajak SC Desrosiers RC Lackner AA Jung JU Identification and characterization of Kaposi's sarcoma-associated herpesvirus K8.1 virion glycoprotein J Virol 1999 73 1341 1349 9882339
Andre S Schatz O Bogner JR Zeichhardt H Stoffler-Meilicke M Jahn HU Ullrich R Sonntag AK Kehm R Haas J Detection of antibodies against viral capsid proteins of human herpesvirus 8 in AIDS-associated Kaposi's sarcoma J Mol Med 1997 75 145 152 9083932 10.1007/s001090050099
Chan SR Bloomer C Chandran B Identification and characterization of human herpesvirus-8 lytic cycle-associated ORF 59 protein and the encoding cDNA by monoclonal antibody Virology 1998 240 118 126 9448696 10.1006/viro.1997.8911
Sato-Matsumura KC Matsumura T Nabeshima M Katano H Sata T Koizumi H Serological and immunohistochemical detection of human herpesvirus 8 in Kaposi's sarcoma after immunosuppressive therapy for bullous pemphigoid Br J Dermatol 2001 145 633 637 11703292 10.1046/j.1365-2133.2001.04413.x
Katano H Sata T Suda T Nakamura T Tachikawa N Nishizumi H Sakurada S Hayashi Y Koike M Iwamoto A Expression and antigenicity of human herpesvirus 8 encoded ORF59 protein in AIDS-associated Kaposi's sarcoma J Med Virol 1999 59 346 355 10502268 10.1002/(SICI)1096-9071(199911)59:3<346::AID-JMV15>3.0.CO;2-4
Hudnall SD Crazy 8: unraveling human herpesvirus 8 seroprevalence Clinical Infectious Diseases 2004 39 1059 1061 15472861 10.1086/424020
Luka J Kallin B Klein G Induction of the Epstein-Barr virus (EBV) cycle in latently infected cells by n-butyrate Virology 1979 94 228 231 220786 10.1016/0042-6822(79)90455-0
zur Hausen H O'Neill FJ Freese UK Hecker E Persisting oncogenic herpesvirus induced by the tumour promotor TPA Nature 1978 272 373 375 204874 10.1038/272373a0
Flamand L Zeman RA Bryant JL Lunardi-Iskandar Y Gallo RC Absence of human herpesvirus 8 DNA sequences in neoplastic Kaposi's sarcoma cell lines JAIDS: Journal of Acquired Immune Deficiency Syndromes 1996 13 194 197
Chatlynne LG Lapps W Handy M Huang YQ Masood R Hamilton AS Said JW Koeffler HP Kaplan MH Friedman-Kien A Detection and titration of human herpesvirus-8-specific antibodies in sera from blood donors, acquired immunodeficiency syndrome patients, and Kaposi's sarcoma patients using a whole virus enzyme-linked immunosorbent assay Blood 1998 92 53 58 9639499
Zhu L Wang R Sweat A Goldstein E Horvat R Chandran B Comparison of human sera reactivities in immunoblots with recombinant human herpesvirus (HHV)-8 proteins associated with the latent (ORF73) and lytic (ORFs 65, K8.1A, and K8.1B) replicative cycles and in immunofluorescence assays with HHV-8-infected BCBL-1 cells Virology 1999 256 381 392 10191203 10.1006/viro.1999.9674
Wang FZ Akula SM Pramod NP Zeng L Chandran B Human herpesvirus 8 envelope glycoprotein K8.1A interaction with the target cells involves heparan sulfate J Virol 2001 75 7517 7527 11462024 10.1128/JVI.75.16.7517-7527.2001
Lam LL Pau CP Dollard SC Pellett PE Spira TJ Highly sensitive assay for human herpesvirus 8 antibodies that uses a multiple antigenic peptide derived from open reading frame K8 1 J Clin Microbiol 2002 40 325 329 11825937 10.1128/JCM.40.02.325-329.2002
Engels EA Whitby D Goebel PB Stossel A Waters D Pintus A Contu L Biggar RJ Goedert JJ Identifying human herpesvirus 8 infection: performance characteristics of serologic assays J Acquir Immune Defic Syndr 2000 23 346 354 10836758
Sergerie Y Abed Y Roy J Boivin G Comparative evaluation of three serological methods for detection of human herpesvirus 8-specific antibodies in Canadian allogeneic stem cell transplant recipients Journal of Clinical Microbiology 2004 42 2663 2667 15184449 10.1128/JCM.42.6.2663-2667.2004
Yu Y Black JB Goldsmith CS Browning PJ Bhalla K Offermann MK Induction of human herpesvirus-8 DNA replication and transcription by butyrate and TPA in BCBL-1 cells J Gen Virol 1999 80 83 90 9934688
Zoeteweij JP Eyes ST Orenstein JM Kawamura T Wu L Chandran B Forghani B Blauvelt A Identification and rapid quantification of early- and late-lytic human herpesvirus 8 infection in single cells by flow cytometric analysis: characterization of antiherpesvirus agents J Virol 1999 73 5894 5902 10364341
Kedes DH Operskalski E Busch M Kohn R Flood J Ganem D The seroepidemiology of human herpesvirus 8 (Kaposi's sarcoma-associated herpesvirus): distribution of infection in KS risk groups and evidence for sexual transmission [published erratum appears in Nat Med 1996 Sep;2(9):1041] Nat Med 1996 2 918 924 8705863 10.1038/nm0896-918
Miller G Rigsby MO Heston L Grogan E Sun R Metroka C Levy JA Gao SJ Chang Y Moore P Antibodies to butyrate-inducible antigens of Kaposi's sarcoma-associated herpesvirus in patients with HIV-1 infection N Engl J Med 1996 334 1292 1297 8609946 10.1056/NEJM199605163342003
Baghian A Luftig M Black JB Meng YX Pau CP Voss T Pellett PE Kousoulas KG Glycoprotein B of human herpesvirus 8 is a component of the virion in a cleaved form composed of amino- and carboxyl-terminal fragments Virology 2000 269 18 25 10725194 10.1006/viro.2000.0198
Lang D Hinderer W Rothe M Sonneborn HH Neipel F Raab M Rabenau H Masquelier B Fleury H Comparison of the immunoglobulin-G-specific seroreactivity of different recombinant antigens of the human herpesvirus 8 Virology 1999 260 47 54 10405355 10.1006/viro.1999.9804
Olsen SJ Tarte K Sherman W Hale EE Weisse MT Orazi A Klein B Chang Y Evidence against KSHV infection in the pathogenesis of multiple myeloma Virus Res 1998 57 197 202 9870587 10.1016/S0168-1702(98)00074-4
Raje N Kica G Chauhan D Zhang Y Teoh G Treon SP Hideshima T Deng JH Gao SJ Alsina M Kaposi's sarcoma-associated herpesvirus gene sequences are detectable at low copy number in primary amyloidosis Amyloid 2000 7 126 132 10842716
Rivas C Thlick AE Parravicini C Moore PS Chang Y Kaposi's sarcoma-associated herpesvirus LANA2 is a B-cell-specific latent viral protein that inhibits p53 J Virol 2001 75 429 438 11119611 10.1128/JVI.75.1.429-438.2001
Calabro ML Fiore JR Favero A Lepera A Saracino A Angarano G Schulz TF Chieco-Bianchi L Detection of human herpesvirus 8 in cervicovaginal secretions and seroprevalence in human immunodeficiency virus type 1-seropositive and -seronegative women J Infect Dis 1999 179 1534 1537 10228077 10.1086/314765
Lyall EG Patton GS Sheldon J Stainsby C Mullen J O'Shea S Smith NA De Ruiter A McClure MO Schulz TF Evidence for horizontal and not vertical transmission of human herpesvirus 8 in children born to human immunodeficiency virus-infected mothers Pediatr Infect Dis J 1999 18 795 799 10493340 10.1097/00006454-199909000-00010
Wilkinson D Sheldon J Gilks CF Schulz TF Prevalence of infection with human herpesvirus 8/Kaposi's sarcoma herpesvirus in rural South Africa S Afr Med J 1999 89 554 557 10416461
Katano H Iwasaki T Baba N Terai M Mori S Iwamoto A Kurata T Sata T Identification of antigenic proteins encoded by human herpesvirus 8 and seroprevalence in the general population and among patients with and without Kaposi's sarcoma J Virol 2000 74 3478 3485 10729121 10.1128/JVI.74.8.3478-3485.2000
Sheldon J Henry S Mourad M Bodeus M Squifflet JP Schulz TF Goubau P Human herpes virus 8 infection in kidney transplant patients in Belgium Nephrol Dial Transplant 2000 15 1443 1445 10978405 10.1093/ndt/15.9.1443
Almuneef M Nimjee S Khoshnood K Miller G Rigsby MO Prevalence of antibodies to human herpesvirus 8 (HHV-8) in Saudi Arabian patients with and without renal failure Transplantation 2001 71 1120 1124 11374413 10.1097/00007890-200104270-00019
Juhasz A Konya J Beck Z Remenyik E Veress G Begany A Medgyessy I Hunyadi J Gergely L HHV-8 ELISA based on a one-step affinity capture of biotinylated K8.1 antigen J Virol Methods 2001 94 163 172 11337051 10.1016/S0166-0934(01)00289-0
Caterino-De-Araujo A de los Santos-Fortuna E Carbone PH Cibella SE Moreira AA Human herpesvirus-8 (HHV-8) antibodies in women from Sao Paulo, Brazil Association with behavioral factors and Kaposi's sarcoma Brazilian Journal of Infectious Diseases 2003 7 395 401 14636479 10.1590/S1413-86702003000600007
Tedeschi R Kvarnung M Knekt P Schulz TF Szekely L De Paoli PD Aromaa A Teppo L Dillner J A prospective seroepidemiological study of human herpesvirus-8 infection and the risk of multiple myeloma British Journal of Cancer 2001 84 122 125 11139326 10.1054/bjoc.2000.1527
Spira TJ Lam L Dollard SC Meng YX Pau CP Black JB Burns D Cooper B Hamid M Huong J Comparison of serologic assays and PCR for diagnosis of human herpesvirus 8 infection J Clin Microbiol 2000 38 2174 2180 10834972
Rabkin CS Schulz TF Whitby D Lennette ET Magpantay LI Chatlynne L Biggar RJ Interassay correlation of human herpesvirus 8 serologic tests. HHV-8 Interlaboratory Collaborative Group J Infect Dis 1998 178 304 309 9697708
Enbom M Sheldon J Lennette E Schulz T Ablashi DV Neipel F Biberfeld P Carlberg H Ljungman P Nilsson A Antibodies to human herpesvirus 8 latent and lytic antigens in blood donors and potential high-risk groups in Sweden: variable frequencies found in a multicenter serological study J Med Virol 2000 62 498 504 11074479 10.1002/1096-9071(200012)62:4<498::AID-JMV15>3.0.CO;2-B
Schatz O Monini P Bugarini R Neipel F Schulz TF Andreoni M Erb P Eggers M Haas J Butto S Kaposi's sarcoma-associated herpesvirus serology in Europe and Uganda: multicentre study with multiple and novel assays J Med Virol 2001 65 123 132 11505454 10.1002/jmv.2011
Schulz TF Kaposi's sarcoma-associated herpesvirus (human herpesvirus 8): epidemiology and pathogenesis J Antimicrob Chemother 2000 45 15 27 10855768 10.1093/jac/45.suppl_4.15
Biggar RJ Engels EA Whitby D Kedes DH Goedert JJ Antibody reactivity to latent and lytic antigens to human herpesvirus-8 in longitudinally followed homosexual men Journal of Infectious Diseases 2003 187 12 18 12508141 10.1086/345866
Katano H Sato Y Sata T Expression of p53 and human herpesvirus-8 (HHV-8)-encoded latency-associated nuclear antigen with inhibition of apoptosis in HHV-8-associated malignancies Cancer 2001 92 3076 3084 11753987 10.1002/1097-0142(20011215)92:12<3076::AID-CNCR10117>3.0.CO;2-D
Bryant-Greenwood P Sorbara L Filie AC Little R Yarchoan R Wilson W Raffeld M Abati A Infection of mesothelial cells with human herpes virus 8 in human immunodeficiency virus-infected patients with Kaposi's sarcoma, Castleman's disease, and recurrent pleural effusions Modern Pathology 2003 16 145 153 12591967 10.1097/01.MP.0000052374.61768.79
Klussmann JP Muller A Wagner M Guntinas-Lichius O Jungehuelsing M Sloots T Ablashi DV Krueger GR Human herpesvirus type 8 in salivary gland tumors J Clin Virol 2000 16 239 246 10738142 10.1016/S1386-6532(99)00077-3
Meggetto F Cesarman E Mourey L Massip P Delsol G Brousset P Detection and characterization of human herpesvirus-8-infected cells in bone marrow biopsies of human immunodeficiency virus-positive patients Hum Pathol 2001 32 288 291 11274637 10.1053/hupa.2001.22749
Engels EA Pittaluga S Whitby D Rabkin C Aoki Y Jaffe ES Goedert JJ Immunoblastic lymphoma in persons with AIDS-associated Kaposi's sarcoma: a role for Kaposi's sarcoma-associated herpesvirus Modern Pathology 2003 16 424 429 12748248 10.1097/01.MP.0000056629.62148.55
Klussmann JP Wagner M Guntinas-Lichius O Muller A Detection of HHV-8 sequences and antigens in a MALT lymphoma associated with Sjogren's syndrome Journal of Oral Pathology & Medicine 2003 32 243 245 12653865
Cioc AM Allen C Kalmar JR Suster S Baiocchi R Nuovo GJ Oral plasmablastic lymphomas in AIDS patients are associated with human herpesvirus 8.[see comment] American Journal of Surgical Pathology 2004 28 41 46 14707862
Robin YM Guillou L Michels JJ Coindre JM Human herpesvirus 8 immunostaining: a sensitive and specific method for diagnosing Kaposi sarcoma in paraffin-embedded sections American Journal of Clinical Pathology 2004 121 330 334 15023036 10.1309/96U1-6LRR-AN5H-WWVE
Patel RM Goldblum JR Hsi ED Immunohistochemical detection of human herpes virus-8 latent nuclear antigen-1 is useful in the diagnosis of Kaposi sarcoma Modern Pathology 2004 17 456 460 14990970 10.1038/modpathol.3800061
Cheuk W Wong KO Wong CS Dinkel JE Ben-Dor D Chan JK Immunostaining for human herpesvirus 8 latent nuclear antigen-1 helps distinguish Kaposi sarcoma from its mimickers American Journal of Clinical Pathology 2004 121 335 342 15023037 10.1309/B8TC-0LBV-H8XY-5MFV
Pendergast SD Werner J Drevon A Wiedbrauk DL Absence of herpesvirus DNA by polymerase chain reaction in ocular fluids obtained from immunocompetent patients Retina 2000 20 389 393 10950418 10.1097/00006982-200007000-00012
Wallace HL 2ndNatelson B Gause W Hay J Human herpesviruses in chronic fatigue syndrome Clinical & Diagnostic Laboratory Immunology 1999 6 216 223 10066657
Kondo K Hayakawa Y Mori H Sato S Kondo T Takahashi K Minamishima Y Takahashi M Yamanishi K Detection by polymerase chain reaction amplification of human herpesvirus 6 DNA in peripheral blood of patients with exanthem subitum Journal of Clinical Microbiology 1990 28 970 974 2351740
Pearson GR Recommended procedures for the diagnosis of persistent active herpesvirus infections Journal of Virological Methods 1988 21 291 300 2846618 10.1016/0166-0934(88)90074-2
Taylor MM Chohan B Lavreys L Hassan W Huang ML Corey L Ashley Morrow R Richardson BA Mandaliya K Ndinya-Achola J Shedding of human herpesvirus 8 in oral and genital secretions from HIV-1-seropositive and -seronegative Kenyan women Journal of Infectious Diseases 2004 190 484 488 15243920 10.1086/421466
Engels EA Biggar RJ Marshall VA Walters MA Gamache CJ Whitby D Goedert JJ Detection and quantification of Kaposi's sarcoma-associated herpesvirus to predict AIDS-associated Kaposi's sarcoma Aids 2003 17 1847 1851 12891072 10.1097/00002030-200308150-00015
Polstra AM Cornelissen M Goudsmit J van der Kuyl AC Retrospective, longitudinal analysis of serum human herpesvirus-8 viral DNA load in AIDS-related Kaposi's sarcoma patients before and after diagnosis Journal of Medical Virology 2004 74 390 396 15368522 10.1002/jmv.20192
Martro E Cannon MJ Dollard SC Spira TJ Laney AS Ou CY Pellett PE Evidence for both lytic replication and tightly regulated human herpesvirus 8 latency in circulating mononuclear cells, with virus loads frequently below common thresholds of detection Journal of Virology 2004 78 11707 11714 15479812 10.1128/JVI.78.21.11707-11714.2004
Fouchard N Lacoste V Couppie P Develoux M Mauclere P Michel P Herve V Pradinaud R Bestetti G Huerre M Detection and genetic polymorphism of human herpes virus type 8 in endemic or epidemic Kaposi's sarcoma from West and Central Africa, and South America Int J Cancer 2000 85 166 170 10629072
Belec L Mohamed AS Authier FJ Hallouin MC Soe AM Cotigny S Gaulard P Gherardi RK Human herpesvirus 8 infection in patients with POEMS syndrome-associated multicentric Castleman's disease Blood 1999 93 3643 3653 10339470
Dupon M Masquelier B Cazorla C Chene G Dumon B Ragnaud JM de Barbeyrac B Bebear C Lacut JY Fleury HJ Acquired immunodeficiency syndrome-associated Kaposi's sarcoma and human herpesvirus 8 DNA detection in serial peripheral blood mononuclear cell samples Res Virol 1997 148 417 425 9498013 10.1016/S0923-2516(97)83631-X
Corbellino M Bestetti G Scalamogna C Calattini S Galazzi M Meroni L Manganaro D Fasan M Moroni M Galli M Long-term remission of Kaposi sarcoma-associated herpesvirus-related multicentric Castleman disease with anti-CD20 monoclonal antibody therapy Blood 2001 98 3473 3475 11719390 10.1182/blood.V98.12.3473
Vieira J Huang ML Koelle DM Corey L Transmissible Kaposi's sarcoma-associated herpesvirus (human herpesvirus 8) in saliva of men with a history of Kaposi's sarcoma J Virol 1997 71 7083 7087 9261440
Wakely PE JrMenezes G Nuovo GJ Primary effusion lymphoma: cytopathologic diagnosis using in situ molecular genetic analysis for human herpesvirus 8 Modern Pathology 2002 15 944 950 12218212 10.1097/01.MP.0000024287.19070.96
Polstra AM Goudsmit J Cornelissen M Development of real-time NASBA assays with molecular beacon detection to quantify mRNA coding for HHV-8 lytic and latent genes BMC Infectious Diseases 2002 2 18 12207829 10.1186/1471-2334-2-18
Polstra AM Goudsmit J Cornelissen M Latent and lytic HHV-8 mRNA expression in PBMCs and Kaposi's sarcoma skin biopsies of AIDS Kaposi's sarcoma patients Journal of Medical Virology 2003 70 624 627 12794727 10.1002/jmv.10440
Cannon MJ Laney AS Pellett PE Human herpesvirus 8: current issues Clinical Infectious Diseases 2003 37 82 87 12830412 10.1086/375230
Patnaik M Komaroff AL Conley E Ojo-Amaize EA Peter JB Prevalence of IgM antibodies to human herpesvirus 6 early antigen (p41/38) in patients with chronic fatigue syndrome.[erratum appears in J Infect Dis 1995 Dec;172(6):1643] Journal of Infectious Diseases 1995 172 1364 1367 7594679
Ablashi DV Eastman HB Owen CB Roman MM Friedman J Zabriskie JB Peterson DL Pearson GR Whitman JE Frequent HHV-6 reactivation in multiple sclerosis (MS) and chronic fatigue syndrome (CFS) patients J Clin Virol 2000 16 179 191 10738137 10.1016/S1386-6532(99)00079-7
Shinkura R Yamamoto N Koriyama C Shinmura Y Eizuru Y Tokunaga M Epstein-Barr virus-specific antibodies in Epstein-Barr virus-positive and -negative gastric carcinoma cases in Japan J Med Virol 2000 60 411 416 10686024 10.1002/(SICI)1096-9071(200004)60:4<411::AID-JMV8>3.0.CO;2-8
Chien YC Chen JY Liu MY Yang HI Hsu MM Chen CJ Yang CS Serologic markers of Epstein-Barr virus infection and nasopharyngeal carcinoma in Taiwanese men New England Journal of Medicine 2001 345 1877 1882 11756578 10.1056/NEJMoa011610
Mbopi-Keou FX Legoff J Piketty C Hocini H Malkin JE Inoue N Scully CM Porter SR Teo CG Belec L Salivary production of IgA and IgG to human herpes virus 8 latent and lytic antigens by patients in whom Kaposi's sarcoma has regressed Aids 2004 18 338 340 15075558 10.1097/00002030-200401230-00030
Vornhagen R Plachter B Hinderer W The TH Van Zanten J Matter L Schmidt CA Sonneborn HH Jahn G Early serodiagnosis of acute human cytomegalovirus infection by enzyme-linked immunosorbent assay using recombinant antigens Journal of Clinical Microbiology 1994 32 981 986 8027354
Yamamoto M Black JB Stewart JA Lopez C Pellett PE Identification of a nucleocapsid protein as a specific serological marker of human herpesvirus 6 infection Journal of Clinical Microbiology 1990 28 1957 1962 2172295
Ma HJ Sjak-Shie NN Vescio RA Kaminsky M Mikail A Pold M Parker K Beksac M Belson D Moss TJ Human herpesvirus 8 open reading frame 26 and open reading frame 65 sequences from multiple myeloma patients: a shared pattern not found in Kaposi's sarcoma or primary effusion lymphoma Clin Cancer Res 2000 6 4226 4233 11106236
|
16138925
|
PMC1243244
|
CC BY
|
2021-01-04 16:38:59
|
no
|
Virol J. 2005 Sep 2; 2:78
|
utf-8
|
Virol J
| 2,005 |
10.1186/1743-422X-2-78
|
oa_comm
|
==== Front
PLoS GenetPLoS GenetpgenplgeplosgenPLoS Genetics1553-73901553-7404Public Library of Science San Francisco, USA 1621754710.1371/journal.pgen.001004305-PLGE-RA-0110R1plge-01-04-01Research ArticleEvolutionInfectious DiseasesMicrobiologyGenetics/GenomicsGenetics/Population GeneticsGenetics/Comparative GenomicsEubacteriaGain and Loss of Multiple Genes During the Evolution of Helicobacter pylori
Gene Gain and Loss in
H. pyloriGressmann Helga 1Linz Bodo 1Ghai Rohit 2Pleissner Klaus-Peter 3Schlapbach Ralph 4Yamaoka Yoshio 5Kraft Christian 6Suerbaum Sebastian 6Meyer Thomas F 1Achtman Mark 1*1 Department of Molecular Biology, Max-Planck-Institut für Infektionsbiologie, Berlin, Germany
2 Institut für Medizinische Mikrobiologie, Justus-Liebig-Universität, Giessen, Germany
3 Core Facility Bioinformatics, Max-Planck-Institut für Infektionsbiologie, Berlin, Germany
4 Functional Genomics Center Zurich, ETH Zurich/University of Zurich, Zurich, Switzerland
5 Department of Medicine, M.E. DeBakey Veterans Affairs Medical Center and Baylor College of Medicine, Houston, Texas, United States of America
6 Medizinische Hochschule Hannover, Institut für Medizinische Mikrobiologie und Krankenhaushygiene, Hannover, Germany
Gojobori Takashi EditorNational Institute of Genetics, Japan* To whom correspondence should be addressed. E-mail: [email protected] 2005 7 10 2005 1 4 e4318 5 2005 26 8 2005 Copyright: © 2005 Gressmann et al.2005This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.Sequence diversity and gene content distinguish most isolates of Helicobacter pylori. Even greater sequence differences differentiate distinct populations of H. pylori from different continents, but it was not clear whether these populations also differ in gene content. To address this question, we tested 56 globally representative strains of H. pylori and four strains of Helicobacter acinonychis with whole genome microarrays. Of the weighted average of 1,531 genes present in the two sequenced genomes, 25% are absent in at least one strain of H. pylori and 21% were absent or variable in H. acinonychis. We extrapolate that the core genome present in all isolates of H. pylori contains 1,111 genes. Variable genes tend to be small and possess unusual GC content; many of them have probably been imported by horizontal gene transfer. Phylogenetic trees based on the microarray data differ from those based on sequences of seven genes from the core genome. These discrepancies are due to homoplasies resulting from independent gene loss by deletion or recombination in multiple strains, which distort phylogenetic patterns. The patterns of these discrepancies versus population structure allow a reconstruction of the timing of the acquisition of variable genes within this species. Variable genes that are located within the cag pathogenicity island were apparently first acquired en bloc after speciation. In contrast, most other variable genes are of unknown function or encode restriction/modification enzymes, transposases, or outer membrane proteins. These seem to have been acquired prior to speciation of H. pylori and were subsequently lost by convergent evolution within individual strains. Thus, the use of microarrays can reveal patterns of gene gain or loss when examined within a phylogenetic context that is based on sequences of core genes.
Synopsis
The Gram-negative pathogenic bacterium Helicobacter pylori colonizes the stomach of 50% of mankind and has probably infected humans since their origins. Due to geographic isolation and frequent local recombination, phylogeographic differences within H. pylori have arisen, resulting in multiple populations and subpopulations that mirror ancient human migrations and genetic diversity. We have examined the gene content of representatives of these populations by whole genome microarrays. Only 1,111 genes are predicted to exist in all H. pylori of the 1,531 that are present on average in two sequenced genomes. Missing genes fall into two classes: one class contains genes within the cag pathogenicity island that was acquired en bloc after speciation and is present only in particular populations. The second class contains a variety of genes whose function may be unimportant for the cell and that were acquired prior to speciation. Their absence in individual isolates reflects convergent evolution through gene loss. Thus, patterns of gene gain or loss can be identified by whole genome microarrays within a phylogenetic context that can be supplied by sequences of genes from the core genome.
Citation:Gressmann H, Linz B, Ghai R, Pleissner KP, Schlapbach R, et al. (2005) Gain and loss of multiple genes during the evolution of Helicobacter pylori. PLoS Genet 1(4): e43.
==== Body
Introduction
The Gram-negative pathogenic bacterium Helicobacter pylori colonizes the stomach of 50% of mankind [1] and has probably infected humans since their origins [2]. H. pylori is acquired during childhood by intrafamilial transmission [3,4], and infection continues lifelong in the absence of antibiotic therapy. It has been postulated [5] that this lifelong association is accompanied by the selection of well-adapted, host-specific variants with particular patterns of expression of adhesins [6,7] and other surface molecules [8,9]. Recombination is frequent during transient colonization with multiple strains due to DNA transformation, resulting in variants within individual hosts that differ in sequence content [10] and genomic composition [11]. As a result of frequent recombination, almost all H. pylori isolates from unrelated hosts possess unique sequences [12], unlike other bacteria, where identical sequences of core housekeeping genes are found in multiple isolates [13]. H. pylori differ between individual hosts [12,14–18], but even greater differences are found when sequences are compared with isolates from different continents [18–20], possibly reflecting genetic drift during geographic isolation [12], as well as adaptation to genetic differences between different ethnic groups of humans [7,19].
H. pylori have been grouped into multiple populations and subpopulations on the basis of sequence differences in seven core genes [12]. These were designated hpEurope, which is common in Europe and countries colonized by Europeans; hpAfrica1, with subpopulations hspWAfrica (West Africa, South Africa, and the Americas) and hspSAfrica (South Africa); hpEastAsia, with subpopulations hspMaori (Polynesians), hspAmerind (Native Americans), and hspEAsia (East Asia); and hpAfrica2, which is very distinct and has been isolated only in South Africa [12]. Recent work (B.Linz, unpublished data) has defined an additional H. pylori population, hpAsia2, which is found in Central and South Asia.
Two genome sequences are available, namely from strains 26695 [21] and J99 [22]. These strains belong to the hpEurope population and the hspWAfrica subpopulation of hpAfrica1, respectively [12]. Of their coding sequences (CDSs), 6% are genome-specific, due to multiple insertions and deletions. Half of these genome-specific CDSs are located in two regions called plasticity zones 1 and 2 [22] that are located 600 kilobases (kb) apart in strain 26695 but are joined in strain J99. These plasticity zone(s) cover a total of 68 kb in 26695 and 45 kb in J99. In addition to plasticity zones 1 and 2, some strains of H. pylori possess a 37 kb pathogenicity island, called the cag pathogenicity island (cag PAI), while others do not [23]. In microarray studies, 22% of the CDSs in the two genomic sequences were absent in at least one of 15 H. pylori strains [24]. It seemed possible that the differential presence or absence of multiple CDSs might correlate with H. pylori population structure, but this question has not yet been addressed.
Bacterial population structures differ with the taxa under investigation [25,26]. Some species, such as Mycobacterium tuberculosis and Yersinia pestis, are of such recent origin that very little sequence diversity has yet accumulated [27,28]. In such largely clonal species, only a few CDSs are variably present among different isolates [29–32], and the absence of individual CDSs correlates with sequence differences between strains of Y. pestis [28]. Similarly, microarray analysis of Listeria monocytogenes identified groups of CDSs whose presence correlated with subdivisions according to serotyping [33]. Microarray analyses of common serovars of Salmonella enterica grouped isolates together that were known to be closely related according to multilocus enzyme electrophoresis [34], with exceptions that may reflect horizontal genetic exchange and subsequent selection. Based on these results, it might be anticipated that whole genome comparisons based on microarrays would not only provide inferences about phenotypic differences within a species but also reveal the general population structure of the bacteria under investigation. However, population structure according to microarrays has not yet been investigated in any species in which a global population structure has already been determined by established methods. And the accuracy of population structures according to microarrays has not been examined for bacteria with high sequence diversity and frequent homologous recombination, such as Neisseria meningitidis [35] or H. pylori. We addressed these issues by performing microarray analyses with isolates that are representative of the global diversity of H. pylori. In order to provide a close out-group for resolving the timing of import of foreign genes by horizontal genetic exchange, we also tested strains of Helicobacter acinonychis, the only close relative of H. pylori according to 16S RNA sequences [36,37]. The results indicate that microarrays provide useful information on variation of genome content. However, the phylogenetic history inferred by microarray data is distorted due to homoplasies, homologous recombination, and horizontal gene transfer. Given an independent phylogenetic context based on sequence diversity in core genes, these distortions can be used to elucidate the history of horizontal gene transfer into a bacterial species.
Results
An H. pylori Whole Genome Microarray
We designed a microarray from the genomes of 26695 and J99 that contains 1,649 PCR products, corresponding to 98% of the CDSs present within both genomes (Table S1). Most of the PCR products correspond to the entire CDSs, but for longer genes, only an N-terminal segment of less than 2.5 kb was amplified. Hybridizations with microarrays were measured by fluorescence with test DNAs mixed with 26695/J99 DNAs that had been differentially labeled with Cy3 and Cy5. The ratios between these fluorescence levels were categorized as reflecting presence or absence of CDSs on the basis of cut-off values that were optimized for control hybridizations between strains 26695 and J99 (accuracy: > 98%; sensitivity: 98%–100%; specificity: 82%–86% [see Materials and Methods]). Subsequent, retrospective analyses of genome content in two recently completed Helicobacter genomes yielded similar estimates of accuracy, sensitivity, and specificity. The estimates of specificity are probably too low: most of the apparent false-negatives seem to reflect the high efficiency of FASTA in detecting partial homologies rather than inefficient hybridization with orthologous CDSs (see Materials and Methods). Bioinformatic analyses also indicated that 98% of the PCR products on the microarray should have hybridized only with the homologous genomic region while the remainder (33/1,649) probably hybridized with two distinct regions (Table S2). Hereafter, we ignore these potential sources of error and treat positive hybridizations as indicating the presence of a CDS and negative hybridizations as indicating its absence.
Gene Content in Representative Strains
The population structure based on the sequence diversity of seven core genes has been determined for 370 isolates from a global collection of H. pylori [12]; in unpublished work, this approach has been extended to more than 800 isolates (B. Linz, et al., unpublished data). We chose 56 strains of H. pylori to reflect the diversity within this collection and to serve as a reference strain collection (Table S3). We also tested four strains of H. acinonychis, a close relative of H. pylori that infects large felines including tigers, lions, and cheetahs [38–40]. These 60 isolates were screened against the whole gene microarrays.
Microarray experiments with the 56 H. pylori strains showed that 499 CDSs were absent from at least one strain (Figure 1 and Table S4). As expected, genes within the cag PAI (28 CDSs tested) and plasticity zones 1 and 2 (108 CDSs) were lacking in numerous isolates (Figure S1), but these three regions accounted for only 27% of the 499 variably present CDSs; 73% (363 CDSs) were located in multiple regions that contained only one to eight CDSs and were scattered around the virtual genome (Figures 1 and S2). Thus, hundreds of CDSs are variably present within H. pylori without any obvious genomic clustering.
Figure 1 Genes Present and Absent in 56 Strains of H. pylori and Four Strains of H. acinonychis
CDSs used in microarrays are shown to scale along a virtual genome consisting of CDSs from both 26695 and J99 in the gene order found within 26695. Circle contents from outside to inside: (1) virtual chromosome (1.76 Mb) with ticks every 220 kb (2), GC content indicated in colors (orange, < 39%; purple, > 39%; green, rRNA genes) (3–9), numbers of missing CDSs from individual strains according to population, color-coded according to presence in both 26695 and J99 (gray) or specific to either 26695 (red) or J99 (blue). Circle, population; 3, hpAfrica2; 4, hpAfrica1; 5, hpEurope; 6, hpAsia2; 7, hpEastAsia; 8, AmerindB; 9, H. acinonychis.
How many genes belong to the core genome present in all H. pylori strains? The two genomes contain 1,590 [21] and 1,495 [22] CDSs, respectively, for a weighted average of 1,531 CDSs (see Materials and Methods). The number of universally present CDSs decreased with the number of strains examined (Figure 2), and only 1,150 CDSs were present in all 56 strains tested (Table S5). Extrapolation to infinity indicates that the core genome consists of 1,111 CDSs (Figure 2), or 73% of the weighted average, which would be universally present in all strains even if a much larger set of isolates were tested.
Figure 2 Extrapolated Number of Universally Present CDSs in H. pylori
The fraction of CDSs present in a sample of strains (“common CDSs”) was calculated on random samples of one to 56 strains taken without replacement. Mean fractions of common CDSs were calculated from 100 iterations of this sampling procedure. The graph shows the results of fitting an exponential decay model to these calculations, in which y0 approaches the minimum number of universally common CDSs at infinity (0.674 × 1,649 CDSs = 1,111 universally present CDSs).
Properties of Variably Present Genes
The GC content of many CDSs in the plasticity zones and within the cag PAI is lower than the average GC content of the entire genome [22]. An unusually low GC content is also typical of many of the variable genes outside these regions (Figure 1), but the frequency distribution of GC content among variable CDSs is unusually broad, skewing in the direction of both low and high GC content (Figure 3, bottom). In fact, most H. pylori CDSs with a GC content of less than 36% or more than 50% are variably present within H. pylori (Figure 3, top). This observation provides support for the inference that many variable genes may have been imported by horizontal gene transfer from other species [22]. The 499 variable CDSs (Table S4) have depressed the average GC content of the genome to 39%, whereas the average GC content of the 1,150 universally present genes is 40.2%.
Figure 3 GC Content of CDSs That Are Universally Present or Variable within H. pylori
CDSs were binned according to GC content in steps of 2% (24–26, 26–28, etc.). Top: Fraction of all CDSs within a bin that are variable. Bottom: Fraction of universally present (n = 1,150) or variable (n = 499) CDSs by GC content. One universally present CDS with a GC content of 62% (HP0359) has been excluded from the figure.
Many of the variable genes in the plasticity region and the cag PAI were classified as “genes of unknown function” [22]. The same association was observed with the current dataset (Table 1): 22% of CDSs encoding outer membrane proteins, 44% of CDSs of unknown function, 54% of CDSs associated with DNA metabolism (often encoding restriction and modification enzymes), and 100% of CDSs that were assigned to “other categories,” including transposons, were variably present. Categories associated with housekeeping functions contained only a few variable CDSs (Tables 1 and S6). These observations provide additional support to the inference that many of these CDSs may have been acquired by horizontal gene transfer. Many of the same CDSs that were variable within H. pylori were also either lacking or variable among the four strains of H. acinonychis that were tested (see below).
Table 1 Variable CDSs in H. pylori and H. acinonychis by Functional Category
As reported by Alm et al. [22], the plasticity zone in J99 contains one of the two 5S/23S rDNA copies, and plasticity zone 2 in 26695 is flanked by an orphan 5S rDNA. Our data now show that a second 5S/23S rDNA in the genome is also associated with numerous variable CDSs, unlike the two 16S rDNA loci, where no association with variable CDSs was noted. In addition to these plasticity zones, we also noticed that the virtual genome contained 24 small regions, designated rA through rX (Figure 1), consisting of two to eight CDSs that were present in at least six isolates but largely absent in at least one population including H. acinonychis and AmerindB (see below).
Differences in Genome Content by Population
Of the 499 variable CDSs, 145 (29%) were uniformly absent within at least one H. pylori population. These included the 28 CDSs within the cag PAI, 53 CDSs within plasticity zones 1 and 2 (Figure S1), 27 CDSs in regions rC, rD, rF, rM, rP, rQ, rV, rW, and rX (Figure S3), as well as 37 singleton CDSs. (However, no CDS was specific to and uniformly present in any single population, except for JHP0914, which is exclusively present in hpAfrica1.) A further indication of population structure within the microarray data was obtained by comparing the mean number of CDSs present within each population (Table 2). hpAfrica1 and hpEurope hybridized with significantly more PCR products on the microarray than did the other populations of H. pylori, as might be expected because the microarrays were designed using genomes from those populations (26695: hpEurope; J99: hpAfrica1). However, significant differences between the numbers of CDSs were also found in comparisons between the hpAfrica2, hpEastAsia, and hpAsia2 populations (Table 2). And H. acinonychis hybridized with fewer spots than did any of the H. pylori populations.
Table 2 Significance of Differences in Mean Numbers of CDSs between Different Populations
In order to test the strength of these correlations, we constructed pair-wise difference matrices for all 60 strains based on the nucleotide content for all seven core genes and on hybridization with the microarray. These matrices were used to calculate Neighbor-joining phylogenetic trees whose branch order was compared by Pearson correlation coefficients (r). The phylogenetic trees differed considerably in branch order (Figure 4A versus 4C; r = 0.49). Many of the discrepancies seem to reflect the 28 CDSs in the cag PAI, because cag
+ isolates clustered separately from cag− isolates (Figure 4C), unlike the relationships according to the seven core genes. In microarray trees within which the CDSs within the cag PAI were excluded (Figure 4B), cag
+ and cag− isolates of hpEurope were intermingled, and isolates from hpAfrica1, hpAfrica2, hspEAsia, hspMaori, and H. acinonychis formed distinct clusters, similar to the sequence tree (Figure 4A). However, this tree still differed considerably from the sequence-based tree (r = 0.51). In the microarray tree, hpEurope and hpAsia2 were intermingled, the hspAmerind population was split into two distinct groupings, and hpAfrica2 was more closely related to hpEurope than according to sequence data, where it formed a highly distinct branch. Finally, the sequence tree is tripartite, consisting of H. acinonychis, hpAfrica2, and a continuum of strains from the four other populations. In contrast, according to the microarray data, H. acinonychis is most similar to three strains (strains 41–43) of the hspAmerind subpopulation that were isolated from Athabaskans in Canada and which we shall subsequently refer to as AmerindB. We note that hpAfrica2 and hspAmerind were identified only due to considerable efforts to cover the global phylogeographical diversity of H. pylori [12]. We therefore tested whether a less globally representative population sample would have yielded stronger correlations between microarrays and sequences of core genes. After excluding the hpAfrica2 population and hspAmerind isolates, the matrices were indeed much more similar (r = 0.85, regardless of whether or not the cag PAI was excluded).
Figure 4 Phylogenetic Structure (Neighbor-Joining Trees) According to (A) Sequences of Seven Core Genes, (B) Microarray Data Excluding cag PAI, and (C) Microarray Data Including the cag PAI for 56 Strains of H. pylori and Four Strains of H. acinonychis
Filled triangles indicate strains possessing the cag PAI, open circles indicate strains lacking it, and filled circles indicate hspAmerind strains that lack HP0536–0548 from the cag PAI. Colors indicate population assignments by Structure based on the sequence data (B. Linz, unpublished data). Numbers at the tips of the twigs are strain numbers (Table S3), while blue numbers next to nodes are bootstrap values over 75% after 250 iterations.
cag PAI Association with Population
The entire cag PAI was lacking in the four isolates of H. acinonychis and eight isolates of hpAfrica2 that were tested by microarrays (Figure S1), which sequence experiments showed was due to the presence of an empty site. Other experiments using PCR amplification also demonstrated an empty site at the location of the cag PAI in five additional isolates of H. acinonychis and ten additional isolates of hpAfrica2 (unpublished data). All isolates of the populations hpAfrica1 and hpAsia2, as well as the subpopulations hspEAsia and hspMaori, contained the entire cag PAI. In contrast, some isolates of hpEurope contained the cag PAI while others lacked it. Similarly, most hspAmerind isolates lacked the cag PAI while others possessed only part of the cag PAI and were lacking CDSs between HP0536 and HP0548. The AmerindB group all lacked the entire cag PAI.
Similarities between H. acinonychis and AmerindB
H. acinonychis and AmerindB are not particularly closely related on the basis of nucleotide sequences. The observation that they cluster near each other according to the microarray data (Figures 4B, 4C, and S3) was intriguing as it indicates that multiple CDSs are absent in both groups, possibly due to convergent evolution. From the microarray, 243 CDSs were universally absent in the four isolates of H. acinonychis and 170 others were variably present. Within the three isolates of AmerindB, 223 CDSs were absent and 110 were variable. These sample sizes are very small, and it may be more relevant to compare with CDSs that were either absent or variably present. Of the 413 such CDSs in H. acinonychis and 333 in AmerindB, the great majority (288), including 83 in the plasticity zones and 54 in regions rA to rX, were variable or absent in both groups. These included the cag PAI (Figure S1) as well as 181 CDSs of unknown function. Interestingly, of those CDSs for which functional assignments had been made, eight were associated with molybdenum, namely HP0172 and HP0798–0801 (region rK), which are involved in molybdopterin biosynthesis, as well as HP0473–0475 (region rH), which encode a molybdenum ABC transporter. These eight CDSs were generally present in other strains, including other hspAmerind isolates (Table S4). Similarly, ten outer membrane protein genes (HP0009, 78, 79, 252, 492, 722, 725, 922, 1243, and 1453) were absent in these two groups, as were four restriction or modification genes. One possible explanation for these observations is that some of the CDSs lacking in both groups reflect adaptation to a similar ecological niche of currently unknown nature.
Other CDSs were specific for only one of the two groups: H. acinonychis lacks CDSs encoding six outer membrane proteins and 17 hypothetical proteins that are present in AmerindB, whereas AmerindB lacks 12 CDSs encoding proteins with various functions that are present in H. acinonychis (Table S4). AmerindB strains also lack amiE (HP0294) and amiF (HP1238), encoding aliphatic amidases that are implicated in ammonia production. All other strains that were tested harbor both CDSs, except for two H. acinonychis isolates that possess only amiE. The absence of these CDSs was unexpected because ammonia production was thought to be very important for colonization of the stomach by Helicobacter [41].
Discussion
Microarrays containing 1,649 CDSs that are present in the genomes of 26695 and J99 were tested against a representative sample of the population genetic diversity within a global collection of H. pylori and H. acinonychis. After weighting for CDSs that are specific to each of the two genomes, 25% of the CDSs within these two genomes were absent in one or more of 56 isolates of H. pylori, and 21% of the H. pylori CDSs were absent or variable within four isolates of H. acinonychis. After extrapolation, we infer that the core genome that is universally present within all strains of H. pylori consists of only 1,111 CDSs (73% of the weighted average) (Figure 2).
Variable Genes: Loss or Gain?
Does the variable presence of CDSs reflect acquisition by some isolates, or loss by others, or a combination of both processes? The sequence-based tree in Figure 4A provides a framework for addressing this question because it reflects the evolutionary history of H. pylori relative to its closest relative, H. acinonychis [36]. This tree is tripartite and consists of three lobes: H. acinonychis, hpAfrica2, and a near continuum of isolates from hpEastAsia, hpAsia2, hpEurope, and hpAfrica1. Sequence comparisons with other Helicobacter species place the root of this tree near the branch between hpAfrica2 and H. acinonychis (unpublished data). Therefore, CDSs that are present within the other populations, but absent in hpAfrica2 and H. acinonychis (and other Helicobacter species), are likely to have been acquired after speciation by horizontal gene transfer from unrelated species. Once imported, such CDSs can spread between isolates by DNA transformation and homologous recombination at the flanking sequences. Alternatively, spread of an empty site by transformation [9] can also lead to loss of such sequences.
Based on this reasoning, the cag PAI was probably imported by horizontal gene transfer from a different species after the tripartite split within the tree because neither H. acinonychis nor the hpAfrica2 population (nor the distant species Helicobacter hepaticus [42] or Wolinella succinogenes [43]) possess any of the CDSs within the cag PAI. The cag PAI is present in almost all hpAfrica1, hpEastAsia, and hpAsia2 isolates and many hpEurope isolates. However, it is lacking completely or in part in most hspAmerind isolates, which also belong to hpEastAsia, and half of the hpEurope isolates. One possibility to explain these observations is that selection for the type-four secretion system encoded by the cag PAI [44–46] may have resulted in the descent of these four populations from an ancestor that had already imported the cag PAI. In that case, strains lacking the cag PAI, or parts of it, have lost the island through transformation with an empty site or through deletion mutations. Alternatively, the cag PAI was imported after subdivision into the hpAfrica1, hpAsia2, hpEastAsia, and hpEurope populations. Its current presence in all of these populations would then reflect spread by transformation from the cells that had first acquired it, coupled with selection for its expression. Extant isolates lacking the cag PAI would then represent the ancestral state prior to its acquisition, or in some cases secondary loss due to transformation of an empty site. Both alternative scenarios infer that H. pylori containing the cag PAI are fitter than those lacking it and that the high number of strains carrying this island reflects positive selection for cag
+ strains. However, the first alternative infers that the selection pressures may not be very high. Although H. pylori possessing the cag PAI are more virulent than strains lacking it [2], both cag
+ and cag
− strains are of comparable incidence within Spain [47], and it remains unclear whether cag
+ bacteria are any fitter in terms of transmission than are cag
− bacteria. Weak support for the first alternative is also provided by the consideration that transformation of the cag PAI may be very rare due to its size (38 kb); the median size of DNA fragments that are exchanged by recombination is only 450 bp [10]. Further sequence analyses of the regions flanking the cag PAI would be needed to distinguish between these alternative scenarios
Other Variably Present CDSs
The situation for most of the other variable CDSs differs from that of the cag PAI. These CDSs tend to be short (Figure S4) and are located within multiple regions, many of which contain fewer than eight CDSs (Figure 1). Thus, they should be readily transmissible between strains by homologous recombination at flanking sites after DNA transformation. Like the cag PAI, these CDSs might have represented genes that were imported recently and have spread due to selection. However, most of the variable CDSs were found in many or all of the H. pylori populations, including hpAfrica2, indicating that they were probably inherited from the last common ancestor of this species. If they were imported from an unrelated species, this must then have happened prior to the existence of that last common ancestor. Their low GC content (Figure 3) is not incompatible with this interpretation, because amelioration of GC content is a process that can take millions of years [48]. Furthermore, CDSs that are absent within certain groups of isolates, such as the AmerindB group plus H. acinonychis (Figure S2), probably reflect convergent, independent gene loss, because the populations lacking these genes are not particularly closely related. Similar conclusions have been drawn for variably present CDSs within the Enterobacteriaceae that were imported once and have subsequently been lost on multiple occasions during subsequent evolution [49].
Most of the variable CDSs encode proteins of unknown function, or selfish DNA such as restriction or modification enzymes [50], and may not possess functional attributes that are targets for positive selection. Thus, it seems likely that repeated loss rather than recent acquisition accounts for the variability of so much of the H. pylori genome. Note that if variable CDSs were imported after speciation, their presence in multiple populations would reflect spread by transformation and would result in less geographical diversity than is the case for neutral, housekeeping genes, whose genetic diversity reflects genetic drift associated with geographical separation. Sequence comparisons of such genes could test this interpretation.
Evolutionary Analyses with Microarray Data
A major presumption within our analyses is that the population structure revealed by sequencing housekeeping genes is a more accurate representation of the genetic descent than are the relationships revealed by microarray analyses. We also infer that this may be true for bacteria in general. For Y. pestis, a species with only limited diversity, microarray analysis involving 4,000 CDSs [29,32] was concordant with the branch order revealed by 44 synonymous single nucleotide polymorphisms [28], but less informative. Within S. enterica, microarrays based on 4,300 CDSs yielded trees that were generally concordant with those based on multilocus enzyme electrophoresis using 24 enzymes [34], but multiple exceptions were found. Our observations with H. pylori parallel those with other species, namely that there is general concordance between phylogenetic relationships based on a microarray with 1,649 CDSs and sequences from seven gene fragments. However, multiple discrepancies were found and the two methods yielded distinct patterns of relationship. The Pearson correlation coefficient between dissimilarity matrices from microarrays and sequence data for the same 60 isolates was only 0.5. Microarrays grouped organisms together (AmerindB and H. acinonychis) that belong to different species and are not known to have any particularly close evolutionary relationship. As a result, the microarray data did not detect the tripartite population structure within these isolates that is revealed by sequence analysis. It might be argued that microarrays based on all the CDSs within a genome provide a better overview of relationships than do the sequences of a few core genes. We note that at least for H. pylori, this is probably not the case because the information content in the sequences of seven gene fragments (1,480 polymorphic sites can yield 41,480 distinct combinations), is 730 orders of magnitude greater than the presence or absence of 535 CDSs (2535 distinct combinations). Secondly, hybridization data from microarrays have a certain methodologically inherent inaccuracy, unlike sequence data. Furthermore, established phylogenetic methods and theory are available for evaluating sequence differences, including the ability to calculate population structures and the time since the existence of a last common ancestor. Until now, microarray data have been evaluated predominantly by clustering techniques, and it is unclear whether changes occur according to a molecular clock or not. In addition, high-throughput methods allow sequencing of multiple gene fragments from thousands of isolates (http://www.mlst.net), which can be necessary for population studies, whereas microarray analysis of more than 100 isolates remains a major effort. Finally, microarrays are based on the gene complement of the genomes that have been chosen for sequencing, which can provide a biased view of the global diversity of that particular species [51]. It therefore seems most appropriate to continue to use sequences of multiple core genes for determining population structure.
The data presented here also show that the comprehensive overview of the genomic content of multiple isolates from microarray data can be used in the context of a known population structure in order to identify discrepancies that reflect evolution by gene acquisition and loss rather than by descent. Such discrepancies can then be used to infer when genes were acquired (or lost) and to infer the selective advantages of particular genes on a genome-wide scale. We therefore conclude that the power of genome-wide analyses of microarrays is first released when analyzed in the context of a population structure that has been defined by sequence based methods.
Conclusions
The data presented here provide a rich source of information on variability within H. pylori and H. acinonychis. Unlike previous conclusions [22], the genome of these organisms is plastic and a weighted average of 27% of the genome is variably present in different isolates. Our data provide a phylogenetic hypothesis for when the cag PAI and other variable regions were imported into these species and indicate that convergent evolution has occurred within the AmerindB group and H. acinonychis. In addition, they also provide a list of 1,150 core genes (Table S5), most of which are universally present within H. pylori, and which includes genes that are essential for the unique physiology of this organism.
Materials and Methods
Bacterial isolates.
H. pylori strains, numbered 1 to 56, that are representative of the hspWAfrica (n = 3) and hspSAfrica (n = 5) subpopulations of hpAfrica1, hpAfrica2 (n = 8), hpEurope (n = 15), hpAsia2 (n = 6), and the hspEAsia (n = 6), hspMaori (n = 3), and hspAmerind (n = 10) subpopulations of hpEastAsia were investigated. Four H. acinonychis strains, numbered 57–60, were from a cheetah, a lion, and two tigers from the United States and Russia. Details about these bacterial strains are summarized in Table S3. Bacterial strains were cultivated as previously described [45]. Genomic DNA was isolated using Qiagen Genomic DNA isolation kits according to the manufacturer's instructions (Qiagen, Valencia, California, United States).
Housekeeping genes.
Fragments of the housekeeping genes atpA, efp, mutY, ppa, trpC, ureI, and yphC were amplified, and both strands were sequenced from independent PCR products as described [18].
Microarray experiments.
The H. pylori array consisted of 1,649 PCR fragments corresponding to 98% of the CDSs within the genomes of 26695 and J99 [21,22]. Primers were designed manually that amplified each entire CDS, or, for long CDSs, the N-terminal 2 kb, with flanking universal linkers to allow reamplification. PCR products were amplified from genomic DNA of 26695 (1,558 CDSs), except for 91 CDSs that were amplified from J99, to which they are specific. After dilution, these PCR products were then used as templates for three secondary rounds of amplification with primers specific for the universal linkers. For each PCR product, agarose gel electrophoresis was used to confirm that single bands of the correct size had been amplified. For PCR products that did not meet this criterion, novel primers flanked by universal linkers were designed using PrimeArray [52]. The final list of primers and sizes of amplified genomic DNA within the PCR products (size range: 62–2,568 bp) is presented in Table S1. Upon bioinformatic analysis of the information in Table S1, we noticed that the PCR product for JHP1032 should not correspond to that CDS, because the primers are oriented in the wrong direction. All other primers were confirmed by FASTA analyses to correspond to the sequence positions listed in Table S1. The same microarray has been used previously for transcriptional analyses [53].
PCR products from the secondary round of amplification were purified using Millipore MultiScreen-PCR plates( Millipore, Billerica, Massachusetts, United States), resuspended in printing buffer (150 mM sodium phosphate [pH 8.5], 0.01% N-lauroyl sarcosine [Sigma, St. Louis, Missouri, United States]) and spotted in duplicate with a Microgrid II spotter (Biorobotics; Genomic Solutions, Ann Arbor, Michigan, United States) on glass slides coated with poly-L-lysine (http://cmgm.stanford.edu/pbrown/protocols/1_slides.html). Additional negative controls were also spotted onto the slides, consisting of 20 PCR fragments from human IMAGE clones from the I.M.A.G.E. Consortium (http://image.llnl.gov/), printing buffer or water. Bacterial genomic DNA was fluorescently labeled with either Cy3 or Cy5 during three rounds of extension with Klenow enzyme and purified as described (http://cmgm.stanford.edu/pbrown/protocols/4_genomic.html). 2 μg of purified labeled DNA from each test strain was mixed with 1 μg each from 26695 and J99 that had been labeled with the alternate dye and hybridized in DIG Easy Hyb buffer (Roche, Basel, Switzerland) to a microarray slide for 15–18 h at 37 °C. DIG Easy Hyb buffer contains urea in a concentration that results in hybridization conditions that are comparable to 50% formamide content. The arrays were washed in washing buffer 1 (1×SSC, 0.03% SDS) until the cover slip fell off, followed by 10-min washes with washing buffers 2 (0.2×SSC) and 3 (0.05×SSC) and dried under a stream of gaseous nitrogen. Fluorescent signal intensities were measured with a G2565AA scanner (Agilent Technologies, Palo Alto, California, United States), and feature extraction was performed with ImaGene (Biodiscovery, El Segundo, California, United States). Empty spots and spots with impurities or high backgrounds were flagged and excluded from the analysis (“missing data”). Spots with signal intensities below the average cross-hybridization signal of the IMAGE clone spots were also excluded. Local background values were subtracted from each spot, the intensities for each fluorescent signal were normalized to the mean intensities over the entire microarray, and the signal intensity ratios between the two dyes were calculated for each PCR product.
The data presented here are based on mean ratios of signal intensities from two experiments each with dye-swapping. For half of the experiments, multiple spots were excluded due to high backgrounds and one to two additional slides were therefore tested. The mean ratios were based on the data from all 2–4 slides that were tested, except when more than two slides were tested, in which case unusually high or low single values were excluded from the mean ratios. The mean ratios were log2 transformed and were assigned a 0 or 1 on the basis of an optimized cut-off calculated by using Gack [54] (settings: data smoothing by moving window average 7; peak modeling based on the normal curve; binary output with 25% EPP cut-off). These settings were determined by trial and error to yield the greatest accuracy (maximal percentage of correct assignments) in separate, control hybridizations between 26695 and J99 (J99: 97%; 26695: 99%) on the basis of the original assignments of genome specificity [22]. We estimated the number of false-positives (J99, 24/135 (18%); 26695, 12/91 (13%) and false-negatives (J99, 25/1514 (1.7%); 26695: 0/1558) and calculated sensitivity (J99: 98%; 26695: 100%) and specificity (J99: 82%; 26695: 86%) values as described in Table S7. We note that 67 of these original assignments to genome-specific genes were predicted to yield false-positive results by Salama et al. [24] on the basis of bioinformatic analyses whereas still other genes are predicted to yield apparently false assignments by Kim et al. [54].
Two sets of retrospective bioinformatic analyses were performed to determine the specificities of the assignments. Firstly, we identified the best FASTA [55] hits for each PCR product included on the array within two unpublished genome sequences (H. acinonychis strain Sheeba [from S. C. Schuster] and hpAfrica2 strain 162.0 [from SS and MA] of isolates that had been tested by microarrays). For each best hit, we calculated a measure of similarity, zfp, consisting of the normalized Z-score [56] multiplied by the homology, multiplied by the fractional length of the hit compared with the size of the PCR product. Most PCR products that hybridized with the microarrays possessed high zfp values for both genomes (Figure S5, bottom), but the distribution of zfp values for PCR products that did not hybridize was very broad (Figure S5, top), and overlapped in part with the distribution of the positive results. The validity of using zfp as a predictor for hybridization was investigated by visual examination of genomic comparisons in ACT [57], resulting in the assignments summarized as + and − for predicted hybridization in Figure S5. According to these manual analyses, most PCR products with zfp > 91 had high homologies over extensive stretches to a genomic region and might be expected to result in positive hybridizations while most PCR products with lower zfp values had only poor homologies to short stretches. Based on a cut-off value of 91, we calculated the accuracy as 93%–96%, sensitivity as 96%–98%, and specificity as 80%–83% (Table S7). These are reflected by zfp values below 91 in 2.1% (Sheba: 28/1314) to 2.5% (162.0: 36/1453) of positive hybridizations and zfp values > 91 in 13.7% (Sheba; 43/315) to 14.6% (162.0; 26/178) of negative hybridizations. However, zfp may not be fully adequate to predict hybridization because 79% (Sheba) to 81% (162.0) of the hits for apparent false-negatives (zfp > 91; hybridization-positive) possessed either < 83% homology or were homologous to a stretch of < 250 bp. As a result, the calculated sensitivity and specificity values are probably much too conservative.
In the second approach, we used FASTA analysis to identify PCR products that would hybridize in microarrays with other sequences within the genome rather than that of the CDS for which the primers had been designed. To this end, we identified all FASTA hits with zfp values > 91 that were not within the location targeted by the PCR primers. We predict that 2% (33/1649) of the PCR products should each hybridize detectably with a second location (Table S2).
Data analysis.
A virtual genome was calculated in which J99 CDSs plus specific flanking DNA were inserted at the appropriate positions based on flanking CDSs within the 26695 genome, and printed using a modified version (http://www.uniklinikum-giessen.de/genome/) of GenomeViz [58] (Figures 1 and S2). The number of CDSs lacking from this virtual genome was calculated for individual strains after weighting CDSs that were specific for 26695 (134 CDSs, weighting factor of 0.4) or J99 (91 CDSs, 0.6).
Functional annotations of genes are according to http://genolist.pasteur.fr/PyloriGene/. 1,627 genes were assigned to a single functional category, and 22 were assigned to two categories.
Statistical analyses were performed using functions that are implemented in R [59]. The statistical significance of differences between mean numbers of CDSs per population in Table 2 was calculated using the “t-test” function for all the data and for random samples of comparable size (“sample” function) over 100 iterations.
Neighbor-joining trees were calculated from difference matrices based on Hamming distances with Mega V2.0 [60]. The correlations between these trees were calculated by Mantel tests with GENETIX (http://www.univ-montp2.fr/~genetix/genetix/intro.htm).
Supporting Information
Figure S1 CDSs Present or Absent within the cag PAI and the Plasticity Zones among 56 Strains of H. pylori and Four Strains of H. acinonychis
Red, absent; yellow, present; black, missing data.
(1 MB PDF)
Click here for additional data file.
Figure S2 A Higher-Resolution Version of Figure 1
(313 KB PDF)
Click here for additional data file.
Figure S3 CDSs Present or Absent within 24 Small Regions of 56 H. pylori and Four H. acinonychis Strains
These regions are scattered along the chromosome, contain two to nine CDSs and contain CDSs that are uniformly lacking in at least one population. Red, absent; yellow, present; black, missing data.
(1 MB PDF)
Click here for additional data file.
Figure S4 Size Distribution of Variably or Universally Present CDSs within H. pylori and H. acinonychis
(10 KB PDF)
Click here for additional data file.
Figure S5 Bioinformatic Predictions of Hybridization based on FASTA Analyses of the Genomic Sequences of hpAfrica2 Strain 162.0 and H. acinonychis strain Sheeba
PCR products on the microarray that hybridized with each of these strains are shown in the lower quadrants, while those that did not hybridize are shown in the upper quadrants. Each quadrant shows a histogram of the number of PCR products versus zfp from FASTA searches (bottom) and whether individual PCR products were predicted to hybridize on the basis of manual ACT comparisons (top). zfp was calculated as the normalized Z-score × homology × fractional length of hit. The arrow in the top left quadrant shows the position of zfp = 91, which was used as a cut-off value to distinguish between FASTA hits that would be expected to hybridize based on the ACT comparisons. These data form the basis for the calculations in Table S7.
(50 KB PDF)
Click here for additional data file.
Table S1 List of PCR Products
(152 KB TXT)
Click here for additional data file.
Table S2 List of PCR Products with Multiple Hits within the Genome
(3 KB TXT)
Click here for additional data file.
Table S3 List of Strains
(27 KB XLS)
Click here for additional data file.
Table S4 List of 499 Variable CDSs in H. pylori
(287 KB XLS)
Click here for additional data file.
Table S5 List of 1,150 PCR Products that Hybridized with All H. pylori Strains
(14 KB TXT)
Click here for additional data file.
Table S6 Variable CDSs in H. pylori (sub)populations
(7 KB PDF)
Click here for additional data file.
Table S7 Accuracy, Sensitivity, Specificity and Predictive Value of Hybridizations with the Genomic Sequences of Sheba and 162.0
(47 KB PDF)
Click here for additional data file.
We are grateful to Jörg Angermann and Christiana Stamer for technical support, Martina Böhme for assistance with bioinformatics, and Michaela Dehio for designing primers. We gratefully acknowledge the permission of Stephan C. Schuster, Penn State University, to analyze the unpublished genome of strain Sheeba and the helpful comments of two anonymous reviewers. The work was supported by grants to SS, TFM, and MA from the German Federal Ministry for Education and Research (BMBF) in the framework of the competence center of the PathoGenoMik Network (Grant 03U213), NGFN-2 grant 01GS0401 to T. Chakraborty, and the Fonds der Chemischen Industrie to TFM
Competing interests. The authors have declared that no competing interests exist.
Author contributions. SS, TFM, and MA conceived and designed the experiments. HG performed the experiments. HG, BL, KPP, and RS analyzed the data. RG, YY, and CK contributed reagents/materials/analysis tools. MA wrote the paper.
Abbreviations
cag PAI
cag pathogenicity island
CDScoding sequence
kbkilobase
==== Refs
References
Suerbaum S Michetti P 2002
Helicobacter pylori infection New Engl J Med 347 1175 1186 12374879
Covacci A Telford JL Del Giudice G Parsonnet J Rappuoli R 1999
Helicobacter pylori virulence and genetic geography Science 284 1328 1333 10334982
Suerbaum S Maynard Smith J Bapumia K Morelli G Smith NH 1998 Free recombination within Helicobacter pylori
Proc Natl Acad Sci U S A 95 12619 12624 9770535
Li L Genta RM Go MF Gutierrez O Kim JG 2002
Helicobacter pylori strain and the pattern of gastritis among first-degree relatives of patients with gastric carcinoma Helicobacter 7 349 355 12485121
Kuipers EJ Israel DA Kusters JG Gerrits MM Weel J 2000 Quasispecies development of Helicobacter pylori observed in paired isolates obtained years apart from the same host J Infect Dis 181 273 282 10608776
Backstrom A Lundberg C Kersulyte D Berg DE Boren T 2004 Metastability of Helicobacter pylori bab adhesin genes and dynamics in Lewis b antigen binding Proc Natl Acad Sci U SA 101 16923 16928
Aspholm-Hurtig M, Dailide G, Lahmann M, Kalia A, Ilver D et al. 2004 Functional adaptation of BabA, the H. pylori ABO blood group antigen binding adhesin Science 305 519 522 15273394
Salaun L Linz B Suerbaum S Saunders NJ 2004 The diversity within an expanded and redefined repertoire of phase-variable genes in Helicobacter pylori
Microbiology 150 817 830 15073292
Kersulyte D Chalkauskas H Berg DE 1999 Emergence of recombinant strains of Helicobacter pylori during human infection Mol Microbiol 31 31 43 9987107
Falush D Kraft C Taylor NS Correa P Fox JG 2001 Recombination and mutation during long-term gastric colonization by Helicobacter pylori : Estimates of clock rates, recombination size, and minimal age Proc Natl Acad Sci U S A 98 15056 15061 11742075
Israel DA Salama N Krishna U Rieger UM Atherton JC 2001
Helicobacter pylori genetic diversity within the gastric niche of a single human host Proc Natl Acad Sci U S A 98 14625 14630 11724955
Falush D Wirth T Linz B Pritchard JK Stephens M 2003 Traces of human migrations in Helicobacter pylori populations Science 299 1582 1585 12624269
Urwin R Maiden MC 2003 Multi-locus sequence typing: A tool for global epidemiology Trends Microbiol 11 479 487 14557031
Akopyanz N Bukanov NO Westblom TU Berg DE 1992 PCR-based RFLP analysis of DNA sequence diversity in the gastric pathogen Helicobacter pylori
Nucleic Acids Res 20 6221 6225 1361982
Akopyanz N Bukanov NO Westblom TU Kresovich S Berg DE 1992 DNA diversity among clinical isolates of Helicobacter pylori detected by PCR-based RAPD fingerprinting Nucleic Acids Res 20 5137 5142 1408828
Go MF Kapur V Graham DY Musser JM 1996 Population genetic analysis of Helicobacter pylori by multilocus enzyme electrophoresis: Extensive allelic diversity and recombinational population structure J Bacteriol 178 3934 3938 8682800
Jiang Q Hiratsuka K Taylor DE 1996 Variability of gene order in different Helicobacter pylori strains contributes to genome diversity Mol Microbiol 20 833 842 8793879
Achtman M Azuma T Berg DE Ito Y Morelli G 1999 Recombination and clonal groupings within Helicobacter pylori from different geographical regions Mol Microbiol 32 459 470 10320570
van der Ende A Pan ZJ Bart A van der Hulst RW Feller M 1998
cagA -positive Helicobacter pylori populations in China and The Netherlands are distinct Infect Immun 66 1822 1826 9573056
van Doorn LJ Figueiredo C Megraud F Pena S Midolo P 1999 Geographic distribution of vacA allelic types of Helicobacter pylori
Gastroenterology 116 823 830 10092304
Tomb JF White O Kerlavage AR Clayton RA Sutton GG 1997 The complete genome sequence of the gastric pathogen Helicobacter pylori
Nature 388 539 547 9252185
Alm RA Ling LS Moir DT King BL Brown ED 1999 Genomic-sequence comparison of two unrelated isolates of the human gastric pathogen Helicobacter pylori
Nature 397 176 180 9923682
Censini S Lange C Xiang Z Crabtree JE Ghiara P 1996
cag, a pathogenicity island of Helicobacter pylori, encodes type-I specific and disease-associated virulence factors Proc Natl Acad Sci U S A 93 14648 14653 8962108
Salama N Guillemin K McDaniel TK Sherlock G Tompkins L 2001 A whole-genome microarray reveals genetic diversity among Helicobacter pylori strains Proc Natl Acad Sci U S A 97 14668 14673
Maynard Smith J Smith NH O''Rourke M Spratt BG 1993 How clonal are bacteria? Proc Natl Acad Sci U S A 90 4384 4388 8506277
Achtman M 2002 A phylogenetic perspective on molecular epidemiology Sussman M Molecular medical microbiology London Academic Press 485 509
Sreevatsan S Pan X Stockbauer K Connell ND Kreiswirth BN 1997 Restricted structural gene polymorphism in the Mycobacterium tuberculosis complex indicates evolutionarily recent global dissemination Proc Natl Acad Sci U S A 94 9869 9874 9275218
Achtman M Morelli G Zhu P Wirth T Diehl I 2004 Microevolution and history of the plague bacillus, Yersinia pestis
Proc Natl Acad Sci U S A 101 17837 17842 15598742
Hinchliffe SJ Isherwood KE Stabler RA Prentice MB Rakin A 2003 Application of DNA microarrays to study the evolutionary genomics of Yersinia pestis and Yersinia pseudotuberculosis
Genome Res 13 2018 2029 12952873
Tsolaki AG Hirsh AE DeRiemer K Enciso JA Wong MZ 2004 Functional and evolutionary genomics of Mycobacterium tuberculosis : Insights from genomic deletions in 100 strains Proc Natl Acad Sci U S A 101 4865 4870 15024109
Hirsh AE Tsolaki AG DeRiemer K Feldman MW Small PM 2004 Stable association between strains of Mycobacterium tuberculosis and their human host populations Proc Natl Acad Sci U S A 101 4871 4876 15041743
Zhou D Han Y Song Y Tong Z Wang J 2004 DNA microarray analysis of genome dynamics in Yersinia pestis : Insights into bacterial genome microevolution and niche adaptation J Bacteriol 186 5138 5146 15262950
Doumith M Cazalet C Simoes N Frangeul L Jacquet C 2004 New aspects regarding evolution and virulence of Listeria monocytogenes revealed by comparative genomics and DNA arrays Infect Immun 72 1072 1083 14742555
Porwollik S Boyd EF Choy C Cheng P Florea L 2004 Characterization of Salmonella enterica subspecies I genovars by use of microarrays J Bacteriol 186 5883 5898 15317794
Feil EJ Holmes EC Bessen DE Chan MS Day NP 2001 Recombination within natural populations of pathogenic bacteria: Short-term empirical estimates and long-term phylogenetic comparisons Proc Natl Acad Sci U S A 98 182 187 11136255
Solnick JV Schauer DB 2001 Emergence of diverse Helicobacter species in the pathogenesis of gastric and enterohepatic diseases Clin Microbiol Rev 14 59 97 11148003
Suerbaum S Kraft C Dewhirst FE Fox JG 2002
Helicobacter nemestrinae ATCC 49396 is a strain of Helicobacter pylori (Marshall et al. 1985) Goodwin et al. 1989 and Helicobacter nemestrinae Bronsdon et al.1991 is therefore a junior heterotypic synonym of Helicobacter pylori
Int J Syst Evol Microbiol 52 437 439 11931154
Eaton KA Dewhirst FE Radin MJ Fox JG Paster BJ 1993
Helicobacter acinonyx sp. nov., isolated from cheetahs with gastritis Int J Syst Bacteriol 43 99 106 8379970
Schroder HD Ludwig C Jakob W Reischl U Stolte M 1998 Chronic gastritis in tigers associated with Helicobacter acinonyx
J Comp Pathol 119 67 73 9717128
Dailidiene D Dailide G Ogura K Zhang M Mukhopadhyay AK 2004
Helicobacter acinonychis : Genetic and rodent infection studies of a Helicobacter pylori -like gastric pathogen of cheetahs and other big cats J Bacteriol 186 356 365 14702304
Bury-Moné S Skouloubris S Dauga C Thiberge JM Dailidiene D 2003 Presence of active aliphatic amidases in Helicobacter species able to colonize the stomach Infect Immun 71 5613 5622 14500481
Suerbaum S Josenhans C Sterzenbach T Drescher B Brandt P 2003 The complete genome sequence of the carcinogenic bacterium Helicobacter hepaticus
Proc Natl Acad Sci U S A 100 7901 7906 12810954
Baar C Eppinger M Raddatz G Simon J Lanz C 2003 Complete genome sequence and analysis of Wolinella succinogenes
Proc Natl Acad Sci U S A 100 11690 11695 14500908
Stein M Rappuoli R Covacci A 2000 Tyrosine phosphorylation of the Helicobacter pylori CagA antigen after cag-driven host cell translocation Proc Natl Acad Sci U S A 97 1263 1268 10655519
Backert S Ziska E Brinkmann V Zimny-Arndt U Fauconnier A 2000 Translocation of the Helicobacter pylori CagA protein in gastric epithelial cells by a type IV secretion apparatus Cellular Microbiol 2 155 164
Odenbreit S Puls J Sedlmaier B Gerland E Fischer W 2000 Translocation of Helicobacter pylori CagA into gastric epithelial cells by type IV secretion Science 287 1497 1500 10688800
Alarcon T Domingo D Martinez MJ Lopez-Brea M 1999
cag A gene and vac A alleles in Spanish Helicobacter pylori clinical isolates from patients of different ages FEMS Immunol Med Microbiol 24 215 219 10378423
Lawrence JG Ochman H 1997 Amelioration of bacterial genomes: Rates of change and exchange J Mol Evol 44 383 397 9089078
Lerat E Daubin V Ochman H Moran NA 2005 Evolutionary origins of genomic repertoires in bacteria PLoS Biol 3 e130. DOI: 10.1371/journal.pbio.0030130 15799709
Kobayashi I Nakayama Y 1998 Restriction-modification gene complexes as selfish gene entities: Roles of a regulatory system in their establishment, maintenance, and apoptotic mutual exclusion Proc Natl Acad Sci U S A 95 6442 6447 9600985
Pearson T Busch JD Ravel J Read TD Roton SD 2004 Phylogenetic discovery bias in Bacillus anthracis using single-nucleotide polymorphisms from whole-genome sequencing Proc Natl Acad Sci U S A 101 13536 13541 15347815
Raddatz G Dehio M Meyer TF Dehio C 2001 PrimeArray: Genome-scale primer design for DNA-microarray construction Bioinformatics 17 98 99 11222267
Niehus E Gressmann H Ye F Schlapbach R Dehio M 2004 Genome-wide analysis of transcriptional hierarchy and feedback regulation in the flagellar system of Helicobacter pylori
Mol Microbiol 52 947 961 15130117
Kim CC Joyce EA Chan K Falkow S 2002 Improved analytical methods for microarray-based genome-composition analysis Genome Biol 3 RESEARCH0065 Epub 2002 Oct 29.
Pearson WR 2000 Flexible sequence similarity searching with the FASTA3 program package Methods Mol Biol 132 185 219 10547837
Pearson WR 1998 Empirical statistical estimates for sequence similarity searches J Mol Biol 276 71 84 9514730
Carver TJ Rutherford KM Berriman M Rajandream MA Barrell BG 2005 ACT: The artemis comparison tool Bioinformatics 21 3422 3423 15976072
Ghai R Hain T Chakraborty T 2004 GenomeViz: Visualizing microbial genomes BMC Bioinformatics 5 198 15601465
Ihaka R Gentleman R 1996 R: A language for data analysis and graphics J Comput Graph Stat 5 299 314
Kumar S Tamura K Jakobsen IB Nei M 2001 MEGA2: Molecular evolutionary genetics analysis software Bioinformatics 17 1244 1245 11751241
Boneca IG de Reuse H Epinat JC Pupin M Labigne A 2003 A revised annotation and comparative analysis of Helicobacter pylori genomes Nucleic Acids Res 31 1704 1714 12626712
|
16217547
|
PMC1245399
|
CC BY
|
2021-01-05 07:59:37
|
no
|
PLoS Genet. 2005 Oct 7; 1(4):e43
|
utf-8
|
PLoS Genet
| 2,005 |
10.1371/journal.pgen.0010043
|
oa_comm
|
==== Front
PLoS Comput BiolPLoS Comput. BiolpcbiplcbploscompPLoS Computational Biology1553-734X1553-7358Public Library of Science San Francisco, USA 1621754810.1371/journal.pcbi.001004505-PLCB-RA-0091R3plcb-01-05-01Research ArticleBioinformatics - Computational BiologyEvolutionStatisticsNoneProtein Molecular Function Prediction by Bayesian Phylogenomics Bayesian PhylogenomicsEngelhardt Barbara E 1*Jordan Michael I 12Muratore Kathryn E 3Brenner Steven E 341 Department of Electrical Engineering and Computer Sciences, University of California, Berkeley, California, United States of America
2 Department of Statistics, University of California, Berkeley, California, United States of America
3 Department of Molecular and Cell Biology, University of California, Berkeley, California, United States of America
4 Department of Plant and Microbial Biology, University of California, Berkeley, California, United States of America
Eisen Jonathan EditorThe Institute for Genomic Research, United States of America* To whom correspondence should be addressed. E-mail: [email protected] 2005 7 10 2005 1 5 e454 5 2005 29 8 2005 Copyright: © 2005 Engelhardt et al.2005This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.We present a statistical graphical model to infer specific molecular function for unannotated protein sequences using homology. Based on phylogenomic principles, SIFTER (Statistical Inference of Function Through Evolutionary Relationships) accurately predicts molecular function for members of a protein family given a reconciled phylogeny and available function annotations, even when the data are sparse or noisy. Our method produced specific and consistent molecular function predictions across 100 Pfam families in comparison to the Gene Ontology annotation database, BLAST, GOtcha, and Orthostrapper. We performed a more detailed exploration of functional predictions on the adenosine-5′-monophosphate/adenosine deaminase family and the lactate/malate dehydrogenase family, in the former case comparing the predictions against a gold standard set of published functional characterizations. Given function annotations for 3% of the proteins in the deaminase family, SIFTER achieves 96% accuracy in predicting molecular function for experimentally characterized proteins as reported in the literature. The accuracy of SIFTER on this dataset is a significant improvement over other currently available methods such as BLAST (75%), GeneQuiz (64%), GOtcha (89%), and Orthostrapper (11%). We also experimentally characterized the adenosine deaminase from Plasmodium falciparum, confirming SIFTER's prediction. The results illustrate the predictive power of exploiting a statistical model of function evolution in phylogenomic problems. A software implementation of SIFTER is available from the authors.
Synopsis
New genome sequences continue to be published at a prodigious rate. However, unannotated sequences are of limited use to biologists. To computationally annotate a hypothetical protein for molecular function, researchers generally attempt to carry out some form of information transfer from evolutionarily related proteins. Such transfer is most successfully achieved within the context of phylogenetic relationships, exploiting the comprehensive knowledge that is available regarding molecular evolution within a given protein family. This general approach to molecular function annotation is known as phylogenomics, and it is the best method currently available for providing high-quality annotations. A drawback of phylogenomics, however, is that it is a time-consuming manual process requiring expert knowledge. In the current paper, the authors have developed a statistical approach—referred to as SIFTER (Statistical Inference of Function Through Evolutionary Relationships)—that allows phylogenomic analyses to be carried out automatically.
The authors present the results of running SIFTER on a collection of 100 protein families. They also validate their method on a specific family for which a gold standard set of experimental annotations is available. They show that SIFTER annotates 96% of the gold standard proteins correctly, outperforming popular annotation methods including BLAST-based annotation (75%), GOtcha (89%), GeneQuiz (64%), and Orthostrapper (11%). The results support the feasibility of carrying out high-quality phylogenomic analyses of entire genomes.
Citation:Engelhardt BE, Jordan MI, Muratore KE, Brenner SE (2005) Protein molecular function prediction by Bayesian phylogenomics. PLoS Comput Biol 1(5): e45.
==== Body
Introduction
The post-genomic era has revealed the nucleic and amino acid sequences for large numbers of genes and proteins, but the rate of sequence acquisition far surpasses the rate of accurate protein function determination. Sequences that lack molecular function annotation are of limited use to researchers, so automated methods for molecular function annotation attempt to make up for this deficiency. But the large number of errors in protein function annotation propagated by automated methods reduces their reliability and utility [1–3].
Most of the well-known methods or resources for molecular function annotation, such as BLAST [4], GOFigure [5], GOtcha [6], GOblet [7], OntoBlast [8], GeneMine [9], PFUNCTIONER [10], PEDANT [11], MAGPIE [12], GeneQuiz [13], the COGs database [14], and HOVERGEN/HOBACGEN [15], rely on sequence similarity, such as a BLAST E-value, as an indicator of homology. A functional annotation is heuristically transferred to the query sequence based on reported functions of similar sequences.
SIFTER (Statistical Inference of Function Through Evolutionary Relationships) takes a different approach to function annotation. Phylogenetic information, if leveraged correctly, addresses many of the weaknesses of sequence-similarity-based annotation transfer [16], such as ignoring variable mutation rates [17,18]. Orthostrapper [19] and RIO [20] provide examples of methods that exploit phylogenetic information, but these methods simplify the problem by extracting pairwise comparisons from the phylogeny, and by using heuristics to convert these comparisons into annotations. SIFTER is a more thoroughgoing approach to automating phylogenomics that makes use of a statistical model of molecular function evolution to propagate all observed molecular function annotations throughout the phylogeny. Thus, SIFTER is able to leverage high-quality, specific annotations and to combine them according to the overall pattern of phylogenetic relationships among homologous proteins.
Other approaches, referred to as context methods, predict protein function using evolutionary information and protein expression and interaction data [21–26]. These methods provide predictions for functional interactions and relationships. They complement detailed predictions from SIFTER and the sequence-based approaches mentioned above, which predict features that evolve in parallel with molecular phylogenetic relationships, such as molecular function.
Phylogenomics
Phylogenomics is a methodology for annotating the specific molecular function of a protein using the evolutionary history of that protein as captured by a phylogenetic tree [17]. Phylogenomics has been used to assign precise functional annotations to proteins encoded in a number of recently sequenced genomes [27,28] and specific protein families [29], despite being a time-consuming manual process. Phylogenomic ideas have also proven helpful for addressing general evolutionary questions, such as showing that horizontal gene transfer is much less common between bacteria and human genes than was suggested in the original publication of the human genome [30,31].
Phylogenomics applies knowledge about how molecular function evolves to improve function prediction. Specifically, phylogenomics is based on the assertion that protein function evolves in parallel with sequence [32], implying that a phylogeny based on protein sequences accurately represents how molecular function evolved for that particular set of proteins. Additionally, molecular function tends to evolve more rapidly after duplication than after speciation because there are fewer mutational constraints; thus, mutations that alter function may more easily fixate in one of the copies [33–35]. These observations give rise to the phylogenomics method, which involves building a phylogenetic tree from homologous protein sequences, marking the location of duplication events, and propagating known functions within each clade descendant from a duplication event. This produces a set of function predictions supported by the evolutionary principles outlined above.
It is broadly recognized that this method produces high-quality results for annotating proteins with specific molecular functions [16]. Three problems limit its feasibility for universal application. First, phylogenomic analysis is a labor-intensive manual process that requires significant effort from dedicated scientists. Second, the quality of the predictions depends on the expertise of the scientist performing the annotation and the quality and availability of functions for the homologous proteins. Third, phylogenomics does not provide a consistent methodology for reporting when a function has insufficient support because of sparse, conflicting, or evolutionarily distant evidence. These three problems motivate the development of a statistical methodology for phylogenomics.
Bayesian Statistics in Biology
Bayesian methodologies have influenced computational biology for many years [36]. Bayesian methods give robust, consistent means of incorporating evidence, even when it is sparse, and enable different types of evidence to be integrated in a meaningful way. The specific inference method we developed for phylogenomics (see Materials and Methods) is based on the general formalism of probabilistic graphical models [37]. It has roots in the peeling methods for pedigree analysis [38,39], and later in maximum likelihood methods for reconstructing phylogenies [40]. We have chosen to take a Bayesian approach to calculating the posterior probabilities of each molecular function for each protein, addressing the uncertainty in the unobserved variables in the phylogeny using Bayesian inference but assuming the phylogeny is known. This is in contrast to the bootstrap approach as taken in RIO and Orthostrapper, which calculate bootstrapped confidence values representing the percentage of trees in which two proteins are orthologous. These methods address the uncertainty of the phylogeny structure (using a frequentist approach), but assume the values of the unobserved variables are known given the phylogeny.
Three properties of the Bayesian approach make it uniquely suited to molecular function prediction. First, Bayesian inference exploits all of the available observations, a feature that proves to be essential in this inherently observation-sparse problem. Second, the constraints of phylogenomics—that function mutation tends to occur after a duplication event or that function evolution proceeds parsimoniously—are imposed as prior biases, not as hard constraints. This provides a degree of robustness to assumptions that is important in a biological context. Third, Bayesian methods also tend to be robust to errors in the data. This is critical in our setting, not only because of existing errors in functional annotations, but also because phylogeny reconstruction and reconciliation often imperfectly reflect evolutionary history.
The current instantiation of SIFTER uses Bayesian inference to combine all molecular function evidence within a single phylogenetic tree, using an evolutionary model of molecular function. A fully Bayesian approach to phylogenomics would integrate over all sources of uncertainty in the function annotation problem, including uncertainty in the phylogeny and its reconciliation, and uncertainty in the evolutionary model for molecular function. It is important to be clear at the outset that the current instantiation of SIFTER stops well short of full Bayesian integration. Rather, we have focused on a key inferential problem that is readily treated with Bayesian methods and is not accommodated by current tools in the literature—that of combining all of the evidence within a single inferred tree using probabilistic methods. Technically, this limited use of the Bayesian formalism is referred to as “empirical Bayes” [41].
Extensions to a more fully Bayesian methodology are readily contemplated; for example, we could use techniques such as those used by MrBayes [42] to integrate across phylogenies. In preliminary investigation of the robustness of SIFTER, however, we have performed bootstrap resampling of the reconciled trees and found little variation in our results across bootstrap samples (results will be detailed elsewhere). This suggests that much of the gain in using Bayesian methods may accrue at the level of inference within a single tree, a suggestion supported by the results that we present here comparing SIFTER to Orthostrapper, which is similar to SIFTER in its use of phylogenomic concepts but differs critically in that it does not integrate evidence within trees.
Molecular Function Annotations
All automated function annotation methods require a vocabulary of molecular function names, whether the names are from the set of Enzyme Commission (EC) numbers, Gene Ontology (GO) molecular function names [43], or words derived from existing manual annotations (e.g., Swiss-Prot functional descriptions). In our method, we currently use the well-curated molecular function ontology from GO, which provides annotations for many proteins in Swiss-Prot and TrEMBL. Each annotation in the GO annotation (GOA) database [44] includes an evidence code, which describes how the annotation was determined. These codes include IDA (inferred by direct assay), IMP (inferred by mutant phenotype), and IEA (inferred by electronic annotation), and they can be used to crudely estimate the reliability of the reported function for a protein.
SIFTER Approach
SIFTER builds upon phylogenomics by employing statistical inference algorithms to propagate available function annotations within a phylogeny, instead of relying on manual inference, as fully described in Materials and Methods. Statistical inference requires a probabilistic model of how the character states (in this case, molecular function) evolve; to this end, we constructed a model of molecular function evolution to infer function in a reconciled phylogeny. Our model takes into account evidence of varying quality and computes a posterior probability for every possible molecular function (from the set of GO molecular function terms) for each protein in the phylogeny, including ancestor proteins. In our model, each molecular function may evolve from any other function, and a protein's function may evolve more rapidly after duplication events than after speciation events.
A “duplication event” captures a single instance of a gene duplicating into divergent copies of that gene within a single genome; a “speciation event” captures a single instance of a gene in an ancestral species evolving into divergent copies of a gene in distinct genomes of different species. Each of the internal nodes of a phylogeny represents one of these two events, although a standard phylogeny does not distinguish between the two. The reconciled phylogeny for a protein family, which discriminates duplication events and speciation events [45,46], specifies the tree-structured graphical model used in inference. In this work, we do not estimate the locations of gene deletion, as it can be difficult to differentiate gene deletion from partial sampling of genes within a particular family.
The available, or observed, function annotations, associated with individual proteins at the leaves of the phylogeny, are propagated towards the root of the phylogeny and then propagated back out to the leaves of the phylogeny, based on a set of update equations defined by the model of function evolution. The result of the inference procedure is a posterior probability of each molecular function for every node in the tree (including the leaves), conditioned on the set of observed functions. The posterior probabilities at each node do not actually select a unique functional annotation for that node, so functional predictions may be selected using a decision rule based on the posterior probabilities of all of the molecular functions. This procedure gives statistical meaning to the phylogenomic notion of propagating functional annotations throughout each clade descendant from a molecular function mutation event. We do not require that the mutation event coincide with a duplication event.
The inference algorithm used in SIFTER has linear complexity in the size of the tree and thus is viable for large families. The complexity of SIFTER is exponential in the number of possible molecular functions in a family, owing to the fact that we compute posterior probabilities for all possible subsets of functions. In the families that we studied, the number of functions was small and this computation was not rate-limiting; in general, however, it may be necessary to restrict the computation to smaller collections of subsets. The rate-limiting step of applying SIFTER is phylogeny reconstruction; a full-genome analysis, given limited computational resources, might use lower-quality or precomputed phylogenies along with bootstrapping, or a subset of closely related species for the larger protein families. We found that lower-quality trees do not significantly diminish the quality of the results (results will be detailed elsewhere).
In this report, we use only GO IDA- and IMP-derived annotations as observations for SIFTER, because of the high error rate and contradictions in the non-experimental annotations (i.e., all annotation types besides IDA and IMP). However, SIFTER can also incorporate other types of annotations, weighted according to their reliability.
Results
We first present results for SIFTER's performance on a large set of proteins to show general trends in prediction and to evaluate the scalability of SIFTER. We then present results for a single protein family with a gold standard set of function characterizations to evaluate prediction quality in detail. We also describe results for the lactate/malate dehydrogenase family, although it does not have a gold standard dataset. The decisive benefit of a statistical approach to phylogenomics is evidenced on each of these different datasets.
Results for 100 Pfam Families
To evaluate the scalability, applicability, and relative performance of SIFTER, we predicted molecular function for proteins from 100 protein families available in Pfam [47], using experimental annotations from the GOA database as evidence. On this broad set of proteins, there are no gold standard functional annotations to which we can meaningfully compare SIFTER's predictions. Instead, we compared SIFTER's predictions to the non-experimental annotations from the GOA database, GOtcha [6], Orthostrapper [19], and BLAST-based predictions [4] in order to measure trends of agreement and compatibility.
For each family in our 100-family dataset, we ran SIFTER on the associated reconciled tree with the experimental annotations (IDA and IMP) from the GOA database. SIFTER produced a total of 23,514 function predictions; we selected the subset of 18,736 that had non-experimental annotations from the GOA database and applied BLASTC, GOtcha, and Orthostrapper to this set (Table 1). We did not compare annotations for experimentally characterized proteins, as those observations were used for inference in SIFTER and Orthostrapper. We compared SIFTER's predictions against non-experimental annotations from the GOA database and function predictions from the other methods. In addition to considering identical GO terms, we also considered terms on the same path to the root of the GO directed acyclic graph (DAG); we call the latter “compatible” annotations because although they are not identical they do not disagree, even though one may be much more specific than the other (and possibly incorrect).
Table 1 Comparison of Predicted Annotations on 18,736 Proteins from 100 Pfam Families
We chose these 100 families to meet one of the following two criteria: (1) greater than 10% proteins with experimental annotations (and more than 25 proteins), or (2) more than nine experimental annotations. Families with fewer than two incompatible experimental GO functions were excluded. The families had an average of 235 proteins, ranging from 25 to 1,116 proteins. On average, 3.3% of the proteins in a family had IDA annotations, and 0.4% had IMP annotations. Both SIFTER and Orthostrapper relied on this particularly sparse dataset for inference; evaluative techniques involving the removal of any of these annotations from inference tended to trivialize the results (e.g., removing a lone experimental annotation for a particular function did not enable that function prediction for homologous proteins). Selecting well-annotated families via these criteria assists SIFTER, but it should also enhance the performance of all of the function transfer methods evaluated here. Note also that SIFTER does not require this level of annotation accuracy to be effective, as discussed below. Finally, it is important to note that many of the IEA annotations from the GOA database may come from one of the assessed methods, so we can expect consistency to be quite high.
Of the 8,501 SIFTER predictions that were either identical or incompatible to the GO non-experimental annotations, 83.1% were identical. The average percentage of identical function predictions by family was 82.9%, signifying that the size of the family does not appear to impact this percentage. The median identity by family was 90.7%, and the mode was 100% (representing 25 families). The minimum identity was 14.4% (Pfam family PF00536). We estimate that 38 of the families contained non-enzyme proteins, and we found no difference in the identity percentage of SIFTER on enzyme families versus non-enzyme families. Similarly, the total number of functional annotations used as observations in SIFTER does not appear to impact the identity percentage (although percentage of proteins with annotations does appear to impact identity percentage). These data suggest that a large percentage of incompatibility is concentrated within a few families. It is not entirely clear what property of those families contributes to the greater incompatibility; it may reflect how well studied the families are relative to the number of proteins in the family.
Annotation rates.
Not all of the annotation methods predict functions for 100% of the proteins. Indeed, as shown in Table 2, Orthostrapper was able to annotate only 7% of the proteins. In an effort to improve the annotation rate, we implemented a variant of Orthostrapper (referred to as “Ortho-ns” in the tables) in which functional annotations were transferred within non-significant orthologous clusters. The nominal mode of operation of Orthostrapper is to transfer function within “statistically significant clusters,” defined as those in which proteins are transitively orthologous with one another in at least 75% of the phylogenies built from a bootstrapped alignment. For Ortho-ns we lowered the criterion from 75% to 0.1%. This yielded an annotation rate of 77%, significantly higher than Orthostrapper, but still well short of the rate of the other methods.
Table 2 Prediction Coverage on 18,736 Proteins from 100 Pfam Families
The percentage of Orthostrapper predictions that were identical or compatible with the non-experimental GOA database function annotations in the 100-family dataset was 88%, but only 7% of proteins received Orthostrapper predictions (Table 2). When function is transferred within non-statistically significant clusters, agreement or compatibility goes to 92% for the 77% of proteins that now receive predictions.
The difficulties encountered by Orthostrapper arise from the small number of proteins that are placed in statistically supported clusters, and the lack of annotations in these clusters. The latter limits the usefulness of the method to protein families with a high percentage of known protein functions, or to observed annotations with a low error rate. These results highlight the impact of the modeling choices in SIFTER and Orthostrapper. SIFTER uses Bayesian inference in a single phylogeny, addressing uncertainty in the ancestral variables in the phylogeny but presently not addressing uncertainty in the phylogeny itself. In contrast, Orthostrapper's approach of bootstrapped orthology addresses uncertainty in the phylogeny, but neglects uncertainty in the ancestral variables. Our results indicate that the gains to be realized by treating uncertainty within a tree may outweigh those to be realized by incorporating uncertainty among trees, but it would certainly be of interest to implement a more fully Bayesian version of SIFTER that accounts for both sources of uncertainty.
Prediction comparisons.
We compared SIFTER's prediction (the function with the single highest posterior probability) to the top-ranked prediction from BLAST-based methods, the top-ranked prediction from GOtcha, the unranked set of non-experimental terms from the GOA database, and unranked Orthostrapper predictions. On this broad set of proteins, SIFTER's predictions were compatible or identical with the non-experimental annotations from the GOA database for 80% of the predictions, while 67% of BLAST-based predictions, 80% of GOtcha predictions, and 78% of GOtcha-ni predictions were compatible or identical to the non-experimental GOA database annotations. It is not entirely clear what these numbers represent, in particular because some unknowable fraction of the IEA annotations in the GOA database were derived using these or related methods. Orthostrapper predictions achieved 88% (Ortho) and 92% (Ortho-ns) compatibility or identity with the GOA database, but because of the small percentage of proteins receiving predictions using Orthostrapper, the absolute number of compatible or identical annotations is much lower. All pairwise comparison data are in Table 1.
The number of incompatible annotations is noteworthy: exact term agreement ranges from 16% to 91%, and the percentage of compatible or identical terms ranges from 45% to 95%. Collectively the methods must be producing a large number of incorrect annotations as evidenced by the high percentage of disagreement in predictions. It appears that there is no gold standard for comparison in the case of electronic annotation methods other than experimental characterization.
Adenosine-5′-Monophosphate/Adenosine Deaminase: A Gold Standard Family
We selected a well-characterized protein family, the adenosine-5′-monophosphate (AMP)/adenosine deaminase family, for evaluation of SIFTER's predictions against a gold standard set of function annotations. We assessed these using experimental annotations that we manually identified in the literature, accepting only first-hand experimental results that were successful in unambiguously characterizing the specific chemical reaction in question. References are provided in Dataset S1 for each protein characterized in this way. The “accuracy” percentages presented here reflect the product of the percentage of proteins that received a prediction and, of those, the percentage that were “correct,” i.e., had the same GO terms as the gold standard test set.
The AMP/adenosine deaminase Pfam family contains 128 proteins. Based on five proteins with experimental annotations from the GOA database, we ran SIFTER to make predictions for the remaining 123 proteins. Of these remaining proteins, 28 had experimental characterizations found by the manual literature search. SIFTER achieved 96% accuracy (27 of 28) for predicting a correct function against this gold standard dataset. SIFTER performed better than BLAST, GeneQuiz, GOtcha, GOtcha-exp (GOtcha transferring only experimental GO annotations), and Orthostrapper (75%, 64%, 89%, 79%, and 11% accuracy, respectively). The comparative results are summarized in Figure 1A. The complete data for these analyses are available in Dataset S1.
Figure 1 Percentage of Proteins with Incorrect or Omitted Molecular Function Prediction of the AMP/Adenosine Deaminase Family, Assessed on a Gold Standard Test Set
Results for SIFTER, BLASTA (the most significant non-identity annotated sequence), BLASTB (the most significant non-identity sequence), GeneQuiz, GOtcha, GOtcha-exp (only experimental GO annotations used), Orthostrapper (significant clusters), and Orthostrapper-ns (non-significant clusters). The gold standard test set was manually compiled based on a literature search. All percentages are of true positives relative to the test set.
(A) Results for discrimination between just the three deaminase substrates, as a percentage of the 28 possible correct functions.
(B) Results for discrimination between the three deaminase substrates plus the additional growth factor domain, as a percentage of the 36 possible correct functions; for BLAST, GeneQuiz, Orthostrapper, and Orthostrapper-ns, we required the transferred annotation to contain both functions; for SIFTER, GOtcha, and GOtcha-exp we required that the two correct functions have the two highest ranking posterior probabilities or scores.
The general role of the AMP/adenosine deaminase proteins is to remove an amine group from the purine base of the substrate. The AMP/adenosine deaminase family has four GO functions associated with member proteins (Figure 2). Adenine deaminase (GO:0000034; EC:3.5.4.2) catalyzes the hydrolytic deamination of adenine to ammonia and hypoxanthine, which is a metabolic nitrogen source [48]. Adenosine deaminase (GO:0004000; EC:3.5.4.4) modifies post-transcriptional RNA, converting adenosine to inosine, resulting in a protein with a sequence different from that coded in the genome by the standard codon table [49]. A mutation in the adenosine deaminase protein in Homo sapiens results in one form of severe combined immune deficiency syndrome [50]. AMP deaminase (GO:0003876; EC:3.5.4.6) converts AMP into inosine-5′-monophosphate and ammonia, and is critical in carbohydrate metabolism [51]. A subset of the adenosine deaminase proteins include multi-domain proteins, in which the additional domain is associated with growth factor activity (GO:0008083, not an enzyme function) (e.g., [52]), and we discuss this additional domain later in Results. Given the functionally important and distinct roles of these related proteins, being able to differentiate at the level of substrate specificity is a critical aspect of function prediction.
Figure 2 Gene Ontology Hierarchy Section Representing the Functions Associated with the Three Substrate Specificities Found in the AMP/Adenosine Deaminase Pfam Family, and the Growth Factor Activity Associated with a Few Members of the Family
Double ovals represent the four functions, none of which are compatible, corresponding to the random variables associated with the random vector used for inference in SIFTER.
The prediction results, a subset of which are shown in Figure 3, illustrate how statistical inference captures the phylogenomic principle of propagating function throughout clades descendant from duplication or speciation events where a function mutation may have occurred. The posterior probability for each annotation provides a measure of confidence in each possible function annotation, based on the model of function evolution and the reported functions for the five proteins with GOA database experimental annotations. The confidence for a particular function annotation tends to drop as the tree-based distance from the closest observation of that function increases.
Figure 3 Results for Pruned Version of the AMP/Adenosine Deaminase Family
The reconciled phylogeny used in inference is shown, along with inferential results (both the posterior probabilities for the deaminase substrates and the function prediction based on the maximum posterior probability). Eight of the proteins in this tree were annotated with growth factor activity, with the second highest probability being adenosine deaminase. The function observations used for inference are denoted by filled boxes to the left of the column with the posterior probabilities. For each substrate specificity that arises, a single edge in the phylogeny identifies a possible location for that mutation. The highlighted sequences are discussed in the text. The blue vertices represent speciation events and the red vertices represent duplication events. The tree was rendered using ATV software, version 1.92 [68].
An alternate method to evaluate prediction accuracy is the receiver operating characteristic (ROC) plot. Figure 4 shows the ROC plot for discriminating the three deaminase substrates (AMP, adenine, and adenosine) using the posterior probabilities from SIFTER, with 64% coverage (i.e., percentage of proteins annotated correctly) at 1% false positives. We logarithmically scaled the false positive axis to focus on true positive percentages when the percentage of false positives is low. The purpose of the ROC plot here is to show that a user-specified cutoff value (based on percentage of false positives at that cutoff) may be used to identify when a functional prediction should not be made for a particular protein. With such a cutoff, we can identify proteins for which the posterior probability of every molecular function is too low to support a prediction.
Figure 4 ROC Plots for the AMP/Adenosine Deaminase Family Functional Predictions from BLASTC, SIFTER, and SIFTER-N (Normalized)
These ROC curves were computed over the 28 proteins in the test set for the deaminase family. This figure presents the ROC plot for both the posterior probabilities produced by SIFTER (and normalized for SIFTER-N) and the E-value significance scores from BLASTC, where they are used to annotate proteins, selecting between deaminase substrates AMP, adenine, and adenosine. The false positive axis is scaled logarithmically to focus on true positive percentages when the percentage of false positives is low. FN, false negative; FP, false positive; TN, true negative; TP, true positive.
Unmodified posterior probabilities allow us to assess the quality of a functional prediction across a family. When considering subsets of functional predictions, however, the maximum posterior for a protein may be small compared to the maximum posterior for other proteins, but we still would like to select this functional assignment because the other posteriors for this protein are smaller still. This can be achieved by normalizing the posteriors for a given protein across the subset of functional assignments of interest. For discriminating the three deaminase substrates, Figure 4 shows the results using renormalized posteriors as the curve labeled SIFTER-N. SIFTER-N achieved 79% coverage at 1% false positives, showing that the correct function has the dominant posterior probability for nearly all proteins. This result implies that choosing a single cutoff value as a decision rule for the unmodified posteriors may not be appropriate for certain biological questions.
Comparison with existing methods.
We compared SIFTER's predictions in this family to four available protein function annotation methods: BLAST (in two approaches called BLASTA and BLASTB, as described in Materials and Methods), GeneQuiz, GOtcha, and Orthostrapper. The complete summary results are shown in Figure 1A.
On this gold standard annotation dataset, SIFTER predictions were more accurate than the alternative methods. Caveats must be mentioned for two of the methods. We ran GOtcha in two different ways on this dataset, detailed in Materials and Methods. GOtcha-exp, which includes only experimental GO annotations for each GOtcha prediction, allows GOtcha access to the same annotation data that SIFTER and Orthostrapper use for inference, creating a more comparable set of predictions. For GOtcha-exp, of the 22 correct annotations, nine were ties between the correct substrate and an incorrect substrate that we resolved in favor of the correct substrate. Orthostrapper was inhibited by failing to annotate some proportion of the proteins with functional characterizations, as in the 100-family dataset. Orthostrapper provided correct annotations for 11% of the proteins; this is because it annotated three correctly, and failed to annotate 25. Orthostrapper using clusters without statistical support (Ortho-ns) provided correct function prediction for 39% of the proteins, correctly annotating 11 of the 28 characterized proteins and omitting the remainder. All of the other methods annotated all of the proteins in the gold standard test set.
If we accept compatible annotations, the accuracy for BLASTA improves by a single protein (Q9VHH7) to 79% (22 of 28), and BLASTB results remain the same (see Materials and Methods for these protocol definitions). To evaluate BLAST in a more sophisticated way, we applied the BLASTC protocol (see Materials and Methods) to find the E-value of the top non-identity hit with an annotation in our selected subset of functions (i.e., adenosine deaminase, adenine deaminase, and AMP deaminase). Although BLAST is not generally applied to molecular function prediction in this way, this comparison enables a more critical assessment of SIFTER. These E-values were used to plot BLAST on the ROC plot (Figure 4, BLAST label). BLASTC achieved 21% coverage at 1% false positives, and is visibly inferior to the coverage provided by both SIFTER and SIFTER-N.
Multi-domain proteins.
The deaminase results focus on a single homologous protein domain that deaminates three possible substrates (AMP, adenosine, and adenine). A few of the proteins in this Pfam family have an additional N-terminal domain. This extra domain (PB003508) has growth factor activity (GO:0008083, not an enzyme function), while the AMP/adenosine binding domain (PF00962) has adenosine deaminase function. We built the phylogeny for this family using only the common AMP/adenosine binding domain; the functional annotations, however, are affiliated with the entire protein sequence. Because the phylogenomic model currently does not explicitly address domain fusion events, we did not consider the molecular function associated with the additional domain in the analyses described thus far.
We reevaluated the results, requiring, where appropriate, growth factor activity to be in the transferred functional annotation (for BLAST, Orthostrapper, and GeneQuiz), or requiring “adenosine deaminase” and “growth factor activity” to have the two highest rankings (ranked by posterior probabilities for SIFTER, or scores for GOtcha and GOtcha-exp). This provided a total of 36 molecular functions to be annotated on the 28 proteins. When evaluating the ability of methods to also correctly annotate this additional role, the accuracy for every method decreases or remains consistent. SIFTER achieved 78% accuracy (28 of 36), while BLAST achieved 75% accuracy (27 of 36), and GeneQuiz achieved 58% accuracy (21 of 36). GOtcha predictions achieved 89% accuracy in the multi-domain setting (32 of 36), and GOtcha-exp achieved 75% (27 out of 36). Of the 11% of proteins that Orthostrapper annotated, none were in the set of proteins with growth factor (so overall accuracy is 8% of the 36 functions to annotate); considering non-statistically significant annotations, Orthostrapper (Ortho-ns) achieved 35% (58% accuracy for 61% of proteins annotated). These results are summarized in Figure 1B. This degradation trend in prediction quality highlights a problem with function annotation methods and their application to multifunction or multi-domain proteins [16,53]. SIFTER in particular appears prone to this degradation, which may be addressed in part by a more problem-specific decision rule that selects function predictions from posterior probabilities, although ultimately the statistical model for SIFTER could explicitly take protein domain architecture into account.
SIFTER prediction experimentally confirmed.
We experimentally characterized the substrate specificity of a deaminase (Q8IJA9) from the human malarial parasite, Plasmodium falciparum. SIFTER predicted that the preferred substrate for this enzyme is adenosine. SIFTER also predicted that the enzyme would not catalyze reactions in which AMP or adenine is the substrate. Saturation kinetics were evaluated by fitting the data to the Michaelis–Menten equation:
where E is the concentration of enzyme and S is the concentration of substrate. Kinetic analysis proves that this deaminase does, in fact, exhibit activity towards adenosine with a kcat of 9.3 ± 0.5 s−1 and a Km of 11 ± 2 μM (Figure 5). No activity was detected with either AMP or adenine at enzyme concentrations up to 860 nM (Figure 5, inset). Since the kcat/Km values for AMP and adenine are less than 10 M−1 s−1, this enzyme shows a preference for adenosine by at least five orders of magnitude.
Figure 5 The Dependence of the Rate of Deamination of Adenosine upon Substrate Concentration with 17 nM Q8IJA9_PLAFA
The open circles are individual data points, while the solid line is the fit of the data to Equation 1. The inset shows raw data for the deamination of three substrates by Q8IJA9_PLAFA as detected by loss of absorbance at 265 nm. The bold, thin, and dashed lines are data for 100 μM adenine, AMP, and adenosine, respectively. The reactions with adenine and AMP contained 860 nM enzyme, while the assay containing adenosine had only 17 nM enzyme. Reaction conditions for all assays were 25 °C in 50 mM potassium phosphate (pH 7.4).
Lactate/Malate Dehydrogenase Family
A second family we chose for detailed analysis and validation is the lactate/malate dehydrogenase family. We used the Pfam family PF00056, representing the NAD+/NADP+ binding domain of this family of proteins. This Pfam family contains 605 proteins, 34 of which have function annotations supported by experimental evidence in the GOA database or in literature references; these 34 were used as evidence for SIFTER.
There are three GO functions associated with proteins in this family. L-lactate dehydrogenase (L-LDH) (GO:0004459; EC:1.1.1.27) catalyzes the final step in anaerobic glycolysis, converting L-lactate to pyruvate and oxidizing NADH [54]. L-malate dehydrogenase (L-MDH) NAD+ (GO:0030060; EC:1.1.1.37) and MDH NADP+ (GO:0046554; EC:1.1.1.83) catalyze the reversible reaction of malate to oxaloacetate using either NADH or NADPH as a reductant [55]. Although the detailed analysis will be described elsewhere, two aspects of the analysis illuminate the power of SIFTER and are discussed briefly here.
Rapid function mutation.
An interesting property of the SIFTER analysis is that it reports three instances of convergent evolution in the dehydrogenase family, all of which are supported in the literature, but only one of which is explicitly documented as convergent evolution [56]. One type of convergent evolution, homoplasy, occurs when a substrate specificity arises from mutations at multiple locations independently in a single phylogenetic tree. Convergent evolution demonstrates that small changes in sequence space do not necessarily correspond to small changes in function space. In particular, when substrate specificity is correlated with a small number of amino acids, molecular function may evolve rapidly. Standard phylogenomics and sequence-based annotation transfer methods are less effective at reporting convergent evolution due to rapid function mutation because of the built-in assumption that sequence and molecular function evolve parsimoniously in parallel. The impact of a particular function annotation associated with a large number of proteins within a significantly short evolutionary distance of the query protein does not allow a small clade with a different function prediction to emerge, since lack of evidence and small evolutionary distances are often insufficient to support a function mutation. By making this assumption probabilistic, SIFTER was able to report three instances of convergent evolution within this family, illustrating another benefit of approaching the problem using Bayesian methods (details provided elsewhere).
SIFTER predictions are specific.
Although there is no gold standard dataset for this family of proteins, based on a manual literature search we gathered 421 proteins in this family that scientists have non-experimentally annotated with a specific function, including substrate and cofactor. Our comparison metric is consistency, or the percentage of protein predictions that are identical to the set of 421 available (although non-experimental) annotations. It appears that the task of discriminating the substrates of this enzyme, i.e., predicting LDH or MDH, is not a difficult one, as all of the methods achieve a high consistency: SIFTER achieves 97% consistency, BLASTA achieves 93% consistency, GeneQuiz achieves 98% consistency, and GOtcha achieves 95% consistency with a set of non-experimental annotations. The methods mentioned here made predictions for all of the 421 proteins.
When we changed the task to include discriminating function at the cofactor level (i.e., predicting one of L-LDH, MDH NAD+, and MDH NADP+, so predicting LDH or MDH is inconsistent), the prediction task became more difficult. On this task, BLASTA consistency drops to 32%, GeneQuiz consistency drops to 68%, and GOtcha consistency drops to 73%. SIFTER's consistency, however, drops only slightly, to 95%, on this set of 421 proteins. One of the primary advantages of SIFTER for scientists is the ability to produce specific function annotations, which was originally a motivation for performing a manual phylogenomic analysis. Often substrate specificity or cofactor changes the biological role of a protein significantly, as in this case and in the deaminase family; being able to differentiate between protein functions with greater precision facilitates characterization and allows subtle but significant functional distinctions to be made.
This point can also be illustrated on a larger scale using the 100-family dataset. For each compatible (but not identical) pair of predictions, we checked which of the two predictions was more specific in the GO DAG. The results are shown in Table 3. On this set of 18,736 proteins, SIFTER predictions were more specific than BLAST, GOtcha, or Orthostrapper predictions at rates of 95%–100%. GOtcha made particularly general predictions, never having more than 35% of its predictions more specific than other methods. This reflects the tendency of the scoring metric in GOtcha to give higher weights to less specific function terms. Although we can make no claim regarding the correctness of these predictions because this dataset is not a gold standard, it is clear from these data that for a diverse set of protein families, the predictions produced by SIFTER are more specific than those produced by these competitive methods.
Table 3 Comparison of Compatible (but Not Identical) Predicted Annotations on 18,736 Proteins from 100 Pfam Families
Discussion
Annotation of protein function through computational techniques relies on many error-prone steps and incomplete function descriptions; SIFTER is no different than other methods in this regard. But a significant component of SIFTER is a statistical model of how protein function evolves. Devos and Valencia postulate that “the construction of a complete description of function requires extensive knowledge of the evolution of protein function that is not yet available” [57]. Although the naive model proposed here for molecular function evolution is too simple to represent how function evolves in detail, the quality of the predictions implies that it is a critical first step to building a complete statistical model that accurately captures much of protein function evolution, and has broad predictive power.
Molecular Function Evolution
The accuracy of SIFTER's results lets us revisit the assumptions of phylogenomics with an eye towards lessons about molecular function evolution. The improvement obtained by using a tree-structured evolutionary history and evolutionary distance, as in SIFTER, versus a measure of evolutionary distance alone, as in BLAST, GOtcha, or GeneQuiz, implies that the information in evolutionary tree structure corrects the systematic errors inherent in pairwise distance methods [17] and goes much further in exploiting the parallel sequence-based tree structure to incorporate sparse data robustly. While the quality of the phylogenetic tree impacts the function predictions, bootstrap resampling of the reconciled trees illustrates that this impact is limited (results will be detailed elsewhere). Nonetheless, it would be useful to extend this analysis to a more fully Bayesian approach that integrates over reconciled phylogenies so that the method is more robust to choices of phylogeny reconstruction and reconciliation methods.
Specific Function Annotation
Comparing SIFTER to BLAST, GeneQuiz, and GOtcha at the cofactor level for the dehydrogenase family (95% consistency versus 32%, 68%, and 73%, respectively) exemplifies the power of SIFTER over other methods for specific function prediction, and the results from the 100-family dataset lend further strength to this comparison. The difference in consistency of BLAST predictions for general substrate discrimination versus specific cofactor discrimination (61% difference) reflects the disparity between the availability of general function descriptions and specific function descriptions evolutionarily proximate to each query protein. By employing phylogenomic principles, SIFTER leveraged evolutionarily distant function observations, incorporating more specific but sparser annotations and enabling SIFTER to make specific function predictions across an entire family. BLAST, GeneQuiz, and GOtcha were limited in their ability to detail molecular function at the cofactor level because of the relative sparsity of functions reported at the cofactor level.
Multi-Domain Proteins
A single protein sequence may contain multiple domains with several functions. There are many cases of individual domains in multi-domain proteins assuming a diverse set of functions, depending on the adjoined domains (e.g., [58]). As illustrated here, phylogenomics and models of molecular function evolution tend to lose predictive power when an additional distantly related function appears (e.g., a large path distance in the GO DAG) [16]. Because this is a relatively rare event, few models based on protein sequence exist to describe these distant functional changes (a notable exception is [59]). A more complex model including domain fusion events would improve the accuracy of SIFTER for many protein families.
Availability of High-Quality Function Data
The sparsity of reliable data is inherent to the task of predicting protein function. In the case of the 100-family dataset, 3.7% of proteins (on average) had experimental function annotations; in the AMP/adenosine deaminase family, 2.6% of proteins had experimental function annotations. Despite this sparseness, SIFTER achieves 96% accuracy in predicting function for homologous proteins for the latter gold standard dataset. Relying exclusively on evidence derived from experimental assays ensured that the quality of the annotations was high. For the AMP/adenosine family in Pfam, there were 348 non-experimental GO annotations (for 127 of the proteins) versus three proteins with IDA (inferred from direct assay) annotations and two proteins with IMP (inferred from mutant phenotype) annotations.
There are methods of extracting annotations from literature (e.g., [60,61]) and other sources of function annotations, such as EC numbers. SIFTER can be readily modified to incorporate these alternative sources of annotations. Based on our results, it appears that SIFTER makes a prediction for a query protein at least as often as BLAST searches do. Our ongoing work focuses on quantifying this transfer rate on a genomic scale. If the posterior probabilities for the small number of specific functional terms produced by SIFTER are propagated toward the root term of the GO DAG, we have posterior probabilities for each molecular function term between the most specific and most general. We can then annotate each protein with either the most specific function prediction available at a certain confidence level or all functions with posterior probabilities above a certain cutoff.
SIFTER's primary role may be to reliably predict protein function for many of the Pfam families or more generic sets of homologous proteins. The argument can be made that no automated function annotation method should be used in some of these cases because the data within a family are too sparse to support annotation transfer. Thus, a second role for SIFTER may be to quantify the reliability of function transfer in under-annotated sets of homologous proteins, by using the posterior probabilities as a measure of confidence in annotation transfer. A third role may be to select targets for functional assays so as to provide maximum coverage based on function transfer for automated annotation techniques. Because of its Bayesian foundations, SIFTER is uniquely qualified to address these alternate questions in a quantifiable and robust way.
Molecular function predictions cannot replace direct experimental evidence for producing flawless function annotations [62]. However, computational methods for functional annotation are being called upon to fill the gap between sequence availability and functional characterization. Unfortunately, large-scale automated methods for function annotation have resulted in widespread annotation errors that reside in current databases [2,18,57,63–65]. These errors impede the progress of experimental studies by providing imprecise or incorrect molecular functions, with little indication of confidence, and minimal recourse to trace the history and origin of that function prediction. The methodology presented here aims to produce high-quality, precise, and traceable sets of possible functions for a protein with a meaningful measure of the reliability of the annotation, thereby facilitating experimental assays of molecular function and inhibiting the propagation of incorrect annotations. SIFTER is unique among function prediction methods in that it exploits phylogenomic information to infer function using formal Bayesian methods. SIFTER's prediction results, as presented here and compared with results from popular methods of function annotation, illustrate the potency and potential of exploiting evolutionary information through a statistical model of molecular function evolution.
Materials and Methods
In this section we first present the modeling, algorithmic, and implementational choices that were made in SIFTER. We then turn to a discussion of the methods that we chose for empirical comparisons. Finally, we present the protocol followed for the deaminase activity assays.
SIFTER model.
In classical phylogenetic analysis, probabilistic methods are used to model the evolution of characters (e.g., nucleotides and amino acids) along the branches of a phylogenetic tree [40] and to make inferences about the ancestral states. For example, if the characters of interest are the nucleotides at an aligned site in DNA sequences, the Jukes–Cantor model [66] defines a transition probability for the four nucleotides at a given node in the phylogeny, conditional on the nucleotide at the ancestor of that node. The Jukes–Cantor model provides a simple example of a parametric model of nucleotide evolution—the transition probability is a parametric function of the branch length, with longer branch lengths yielding a distribution that is more nearly uniform. Given a model such as Jukes–Cantor for each branch in a phylogenetic tree, the overall joint probability of an assignment of nucleotides to all of the nodes in the tree is obtained by taking the product of the branch-wise conditional probabilities (together with a marginal probability distribution for the root). Conditioning on observed values of some of the nodes (e.g., the leaves of the tree, corresponding to extant species), classical dynamic programming algorithms (e.g., the “pruning” algorithm) can be used to infer posterior probability distributions on the states of the unobserved nodes [40].
SIFTER borrows much of the probabilistic machinery of phylogenetic analysis in the service of an inference procedure for molecular function evolution. The major new issues include the following: (1) given our choice of GO as a source of functional labels, functions are not a simple list of mutually exclusive characters, but are vertices in a DAG; (2) we require a model akin to Jukes–Cantor but appropriate for molecular function; (3) generally only a small subset of the proteins in a family are annotated, and the annotations have different degrees of reliability. We describe our approach to these issues below.
The first step of SIFTER is conventional sequence-based phylogenetic reconstruction and reconciliation.
Phylogenetic reconstruction is the computational bottleneck in the application of SIFTER. Thus, in the current implementation of SIFTER we have made use of parsimony methods instead of more computationally intense likelihood-based or Bayesian methods in phylogenetic reconstruction. This “empirical Bayes” simplification makes it possible to apply SIFTER to genome-scale problems.
In detail, the steps of phylogenetic reconstruction implemented in SIFTER are as follows. Given a query protein, we (1) find a Pfam family of a homologous domain [47], and extract the multiple sequence alignment from the Pfam database (release 12.0); (2) build a rooted phylogenetic tree with PAUP* version 4.0b10 [67], using parsimony with the BLOSUM50 matrix; (3) apply Forester version 1.92 [68] to estimate the location of the duplication events at the internal nodes of the phylogeny by reconciling the topological differences between a reference species tree (taken from the Pfam database) and the protein tree.
The result of this procedure is a “reconciled phylogeny,” a rooted phylogenetic tree with branch lengths and duplication events annotated at the internal nodes [45,46].
Subsequent stages of SIFTER retain these structural elements of the phylogeny, but replace the amino acid characters with vectors of molecular function annotations and place a model of molecular function evolution on the branches of the phylogeny.
We use the following process to define a vector of candidate molecular function annotations for a given query protein and for the other proteins in the phylogeny.
Given a Pfam family of a homologous domain for a query protein, we index into the GOA database [44] (we used the version of January 6, 2004) and form an initial raw list of candidate molecular functions by taking the union of the experimental annotations associated with all of the proteins in the Pfam family. We then prune this list by making use of the structure of GO in the following way. Recall that GO is organized into a DAG of functions, with the more specific function names at the leaves. Given our initial list of functions, we choose those functions that are closest to the leaves of the DAG, under the constraint that the corresponding nodes form a “nad” subset—a subset of nodes that are “not ancestors and not descendants” of each other in GO [69].
We treat the elements of this nad subset as the indices of a vector of candidate functional annotations, to be referred to as the “annotation vector” in the remainder of this section. Subsequent inferential stages of SIFTER treat this vector as a Boolean random vector. That is, we assume that each function can be asserted as either present or absent for a given protein, and we allow more than one function to be asserted as being present.
The goal is to compute posterior probabilities for all unannotated proteins in the family of interest, conditioning on experimentally derived annotations (i.e., IDA or IMP annotations) associated with some of the proteins in the family. To accommodate the fact that IDA annotations may be more reliable than IMP annotations according to the experiments by which they are generated, and to allow users to make use of other, possibly less reliable, annotations, SIFTER distinguishes between a notion of “true function” and “annotated function,” and defines a likelihood function linking these variables. In particular, the current implementation of SIFTER defines expert-elicited probabilities that an experimentally derived annotation is correct given the method of annotation: IDA annotations are treated as having a likelihood of 0.9 of being correct, and IMP as having a likelihood of 0.8.
GOA database annotations are not restricted to the leaves of the ontology but can be found throughout the DAG. To incorporate all such annotations in SIFTER, we need to propagate annotations to the nad subset. In particular, annotations at nodes that are ancestors to nad nodes need to be propagated downward to the nad nodes. (By definition, there can be no annotated descendants of nad nodes.) We do this by treating evidence at an ancestor node as evidence for all possible combinations of its descendants, according to the distribution Q(S) = 1/η|S|, where S is an arbitrary subset of the nad nodes, |S| is the cardinality of S, and the value of η is fixed by the requirement that
. Finally, to combine annotations at a given node we take one minus the product of their errors (where error is one minus their likelihood).
We turn to a description of the model of molecular function evolution that SIFTER associates with the branches of the phylogeny. For each node in the phylogeny, corresponding to a single protein, this model defines the conditional probability for the vector of function annotations at the node, conditioning on the value of the vector of function annotations at the ancestor of the node. Figure 6 provides an overview of the model and its role in the inference procedure.
Figure 6 A Depiction of a Fragment of a Phylogeny and the Noisy-OR Model
(A) Two proteins, Q9VFS0 and Q9VFS1, both from Drosophila melanogaster, related by a common ancestor protein.
(B) Protein Q9VFS1 has a functional observation for adenosine deaminase (the center rectangle). Also shown are the posterior probabilities for each molecular function as grayscale (white indicating zero and black indicating one) of the annotation vector after inference. Each component of the vector corresponds to a particular deaminase substrate.
(C) The noisy-OR model that underlies the inference procedure. We focus on the adenosine deaminase random variable in protein Q9VFS0. The transition probability for this random variable depends on all of the ancestor random variables and the transition parameters qm,n.
We chose a statistical model known as a loglinear model for the model of function evolution. We make no claims for any theoretical justification of this model. It is simply a phenomenological model that captures in broad outlines some of the desiderata of an evolutionary model for function and has worked well in practice in our phylogenomic setting.
Let Xi denote the Boolean vector of candidate molecular function annotations for node i and let
denote the mth component of this vector. Let M denote the number of components of this vector. Let πi denote the immediate ancestor of node i in the phylogeny, so that
denotes the annotation vector at the ancestor. We define the transition probability associated with the branch from πi to i as follows:
where di and qm,n are parametric functions of branch lengths in the phylogeny and path lengths in GO, respectively. This functional form is known as a “noisy-OR” function [70], and it has the following interpretation. Suppose that
is equal to one for only a single value of m and is equal to zero for all other values of m. (Thus, a single function is asserted as present for the parent.) Suppose that di is equal to one. Then the probability that node i has the nth function (i.e., that
) is equal to qm,n. Thus, qm,n has an interpretation as a local transition probability between the mth function and the nth function. The multiplication in Equation 2 corresponds to an assumption of independence (specifically, independence of the events that an ancestor function m fails to trigger a function n in the descendant).
To capture the notion that a transition should be less probable the less “similar” two functions are, we defined qm,n to be a decreasing function of the path length lm,n in GO. Specifically, we let
, where s is a free parameter. This parameter is taken to be different for speciation and duplication events; in particular, it is larger in the latter case, corresponding to the phylogenomic assumption that evolutionary transitions are more rapid following a duplication event. To set the parameters s
speciation and s
duplication, we can in principle make use of resampling methods such as cross-validation or the bootstrap. In the case of the deaminase family, however, the number of observed data points (five) is too small for these methods to yield reasonable results, and in our analyses of this family we simply fixed the parameters to the values s
speciation = 3 and s
duplication = 4 and did not consider other values. For the 100-family dataset, we ran each family with a few different parameter settings, because the number of annotations available for the families was in general prohibitively small, and fixed them at the set of values that produced predictions most closely aligned with the non-experimental annotations from the GOA database. We define qm,m = (1/r)s
/2 for self-transitions; this normalizes the self-transition probability with respect to the number of components of the annotation vector.
We also need to parameterize the transition rate as a function of the branch length in the phylogeny. This is achieved by defining di to be a decreasing nonlinear function of the branch length. (Thus, for greater branch lengths, transitions become more probable.) Specifically, we set
, where bi is the most parsimonious number of amino acid mutations along the branch from πi to i.
Having defined a probabilistic transition model for the branches of the phylogeny, and having defined a mechanism whereby evidence is incorporated into the tree, it remains to solve the problem of computing the posterior probability of the unobserved functions in the tree conditional on the evidence.
This problem is readily solved using standard probabilistic propagation algorithms. Specifically, all posterior probabilities can be obtained in linear time via the classical pruning algorithm [40], also known as (a special case of) the junction tree algorithm [37]. This algorithm propagates probabilistic “messages” from the leaves of the tree to the root, and from the root back to the leaves, performing a constant number of operations at each node. The computational complexity of the algorithm is thus linear in the number of leaves in the tree.
Methods for comparison.
The BLAST version 2.2.4 [4] assessment was performed on the non-redundant set of proteins from Swiss-Prot downloaded from the NCBI Web site on March 7, 2004. We ran BLASTP with an E-value cutoff of 0.01. We transferred annotation from the highest scoring non-identity protein (BLASTB), which was determined by checking the alignment for 100% identity and identical species name. We also transferred annotation from the highest scoring annotated non-identity protein (BLASTA), which was the highest scoring non-identity protein that had a functional description (i.e., not “hypothetical protein” or “unknown function”). Phrases modifying a functional annotation such as “putative” and “-related” were ignored. An annotation including an EC number was considered unambiguous.
To build the ROC plots for the BLASTC comparison, for each protein in the selected families we searched the BLAST output for the highest scoring sequence (with the most significant E-value) that had a function description from the appropriate set: for the deaminase family we searched for “adenosine deaminase,” “adenine deaminase,” “AMP deaminase,” and, for the results on multiple functions, “growth factor activity.” A reference could also be in the form of an EC number or unambiguous phrase (e.g., “growth and transcription activator” was interpreted as “growth factor activity”). We plotted the false positives (one minus specificity) versus true positives (sensitivity) as the acceptance cutoff for E-values ranges from 0.01 to zero, where proteins were annotated with a function if the most significant E-value for a protein with that particular function was less than the acceptance cutoff.
To build the BLASTC set of annotations for the set of 18,736 proteins from the 100-family dataset, we built a keyword search with 260 GO terms, including all of the terms from the SIFTER analysis and other terms common to the BLAST search results. From this keyword search we extracted a set of terms ranked by E-values, facilitated by BioPerl [71]. The first of the ranked set of terms was then compared against predictions from alternate methods, using GO molecular function term comparisons. The full set of data including the keyword search code is available in Dataset S1.
For GeneQuiz, we ran each member of the AMP/adenosine deaminase and lactate/malate dehydrogenase families on the GeneQuiz server (publicly available at EBI) from August 22, 2004, to September 1, 2004 [13]. The function predictions from GeneQuiz are not based on an ontology, and we manually converted them to equivalent GO numbers. If there was an EC number, the annotation correlated exactly with a GO term. We ignored phrases such as “putative,” “fragment,” or “weakly similar to,” and only interpreted the functional words.
For GOtcha, we ran the first publicly available version of the GOtcha software [6], kindly provided by D. Martin, on the set of 18,736 proteins from the 100-family dataset. We searched the protein sequences against all seven available genome databases, gathering results using all annotations (GOtcha), excluding IEA annotations (GOtcha-ni), and finally including only IMP and IDA annotations (GOtcha-exp). The output is a ranked list of GO terms. It is a property of GOtcha that the top-ranked terms are the more general ones (e.g., “molecular function,” the most general term in the molecular function ontology, is always ranked first). We parsed out the molecular function annotations, retaining the relative rank of each term, and discarding terms that were too general but had compatible terms ranked below them. We broke ties in rank in favor of the correct term (for the deaminase family). We compared the top member of this ranked list against predictions from the other methods.
For Orthostrapper [19], we ran the version from February 6, 2002, on each of the 100 Pfam families in our dataset. Species 1 and species 2 were all of the proteins with any type of GO annotation containing the proteins from eukaryotes and non-eukaryotes, respectively, when both sets were not empty (and mammals and non-mammals otherwise). We clustered the bootstrapped analysis according to the cluster program in Orthostrapper, using a bootstrap cutoff of 750 and then using a cutoff of one, resulting in the statistically significant clusters and non-statistically significant clusters, respectively.
In each cluster, we transferred all experimentally derived GO annotations from member proteins onto the remaining proteins without experimentally derived GO annotations. If a cluster did not contain a protein with an experimentally derived GO annotation, no functions were transferred; if a protein was present in multiple clusters, it would receive annotations transferred within each of those clusters. This method yields an unranked set of predictions for each protein.
Deaminase activity assays.
Purified Q8IJA9_PLAFA was the kind gift of Erica Boni, Chris Mehlin, and Wim Hol of the Stuctural Genomics of Pathogenic Protozoa project at the University of Washington. Adenosine and adenine were from Sigma-Aldrich (St. Louis, Missouri, United States), AMP was from Schwarz Laboratories (Mt. Vernon, New York, United States), and monobasic and dibasic potassium phosphate were from EMD Chemicals (Gibbstown, New Jersey, United States).
The loss of absorbance at 265 nm was monitored with an Agilent Technologies (Palo Alto, California, United States) 8453 spectrophotometer. The Δɛ between substrate adenosine and product inosine is 7,740 AU M−1 cm−1 [72].
Supporting Information
Dataset S1 SIFTER Supplemental Data
(7.5 MB TGZ)
Click here for additional data file.
Accession Numbers
The Swiss-Prot (http://www.ebi.ac.uk/swissprot/) accession number for H. sapiens adenosine deaminase is P00813 and for P. falciparum adenosine deaminase is Q8IJA9. The Pfam (http://www.sanger.ac.uk/Software/Pfam/) accession number for the AMP/adenosine deaminase family is PF00962.
This material is based upon work supported under a National Science Foundation Graduate Research Fellowship, National Institutes of Health (NIH) grant K22 HG00056, the Searle Scholars Program (1-L-110), a grant from Microsoft Research, a grant from the Intel Corporation, NIH grant R33 HG003070, an IBM SUR grant, and NIH grant GM35393 to J. F. Kirsch, whose generous help enabled the rapid characterization of Q8IJA9_PLAFA. Purified Q8IJA9_PLAFA was the kind gift of Erica Boni, Chris Mehlin, and Wim Hol of the Stuctural Genomics of Pathogenic Protozoa project at the University of Washington.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. BEE, MIJ, and SEB conceived and designed the experiments. BEE performed the experiments. KEM conceived, designed, and performed the experimental characterization of the adenosine deaminase protein. BEE, MIJ, and SEB analyzed the data. BEE, MIJ, KEM, and SEB contributed reagents/materials/analysis tools. BEE, MIJ, KEM, and SEB wrote the paper.
Abbreviations
AMPadenosine-5′-monophosphate
DAGdirected acyclic graph
ECEnzyme Commission
GOGene Ontology
GOAGene Ontology annotation
LDHlactate dehydrogenase
MDHmalate dehydrogenase
ROCreceiver operating characteristic
==== Refs
References
Galperin MY Koonin EV 1998 Sources of systematic error in functional annotation of genomes: Domain rearrangement, non-orthologous gene displacement, and operon disruption In Silico Biol 1 7
Brenner SE 1999 Errors in genome annotation Trends Genet 15 132 133 10203816
Koonin EV 2000 Bridging the gap between sequence and function Trends Genet 16 16 10647294
Altschul SF Gish W Miller W Myers EW Lipman DJ 1990 Basic local alignment search tool J Mol Biol 215 403 410 2231712
Khan S Situ G Decker K Schmidt CJ 2003 GoFigure: Automated gene ontology annotation Bioinformatics 18 2484 2485
Martin DMA Berriman M Barton GJ 2004 GOtcha: A new method for prediction of protein function assessed by the annotation of seven genomes BMC Bioinformatics 5 178 195 15550167
Groth D Lehrach H Henning S 2004 GOblet: A platform for Gene Ontology annotation of anonymous sequence data Nucleic Acids Res 32 313 317
Zehetner G 2003 Ontoblast function: From sequences similarities directly to potential function annotations by ontology terms Nucleic Acids Res 31 3799 3803 12824422
Lee CI Irizarry K 2001 The Genemine system for genome/proteome annotation and collaborative data mining IBM Syst J 40 592 603
Pazos F Sternberg MJE 2004 Automated prediction of protein function and detection of functional sites from structure Proc Natl Acad Sci U S A 101 14754 14759 15456910
Frishman D Mokrejs M Kosykh D Kastenmuller G Kolesov G 2003 The PENDANT genome database Nucleic Acids Res 31 207 211 12519983
Gaasterland T Sensen CW 1996 Magpie: Automated genome interpretation Trends Genet 12 76 78 8851977
Andrade MA Brown NP Leroy C Hoersch S deDaruvar A 1999 Automated genome sequence analysis and annotation Bioinformatics 15 391 412 10366660
Tatusov RL Galperin MY Natale DA Koonin EV 2000 The COG database: A tool for genome-scale analysis of protein functions and evolution Nucleic Acids Res 28 33 36 10592175
Perriere G Duret L Gouy M 2000 HOBACGEN: Database system for comparative genomics in bacteria Genome Res 10 379 385 10720578
Sjölander K 2004 Phylogenomics inference of protein molecular function: Advances and challenges Bioinformatics 20 170 179 14734307
Eisen JA 1998 Phylogenomics: Improving functional predictions for uncharacterized genes by evolutionary analysis Genome Res 8 163 167 9521918
Rost B 2002 Enzyme function less conserved than anticipated J Mol Biol 318 595 608 12051862
Storm CE Sonnhammer EL 2002 Automated ortholog inference from phylogenetic trees and calculation of ortholog reliability Bioinformatics 18 92 99 11836216
Zmasek CM Eddy SR 2002 Rio: Analyzing proteomes by automated phylogenomics using resampled inference of orthologs BMC Bioinformatics 3 14 12028595
Marcotte EM Pellegrini M Thompson MJ Yeates TO Eisenberg D 1999 A combined algorithm for genome-wise prediction of protein function Nature 402 83 86 10573421
Huynen M Snel B Lathe W Bork P 2000 Exploitation of gene context Curr Opin Struct Biol 10 366 370 10851194
Mateos A Dopazo J Jansen R Tu Y Gerstein M 2002 Systematic learning of gene functional classes from DNA array expression data by using multilayer perceptrons Genome Res 12 1703 1715 12421757
Pavlidis P Weston J Cai J Noble WS 2002 Learning gene functional classifications from multiple data types J Comput Biol 9 401 411 12015889
Letovsky S Kasif S 2003 Predicting protein function from protein/protein interaction data: A probabilistic approach Bioinformatics 19 197 204
Troyanskaya OG Dolinski K Owen AB Altman RB Botstein D 2003 A Bayesian framework for combining heterogeneous data sources for gene function prediction (in S. cerevisiae ) Proc Natl Acad Sci U S A 100 8348 8353 12826619
Nierman WC Feldblyum TV Laub MT Paulsen IT Nelson KE 2001 Complete genome sequence of Caulobacter crescentus
Proc Natl Acad Sci U S A 98 4136 4141 11259647
Heidelberg JF Eisen JA Nelson WC Clayton RA Gwinn ML 2000 The genome sequence of Vibrio cholerae and the etiologic agent of cholera Nature 406 477 483 10952301
Eisen JA Hanawalt PC 1999 A phylogenomics study of DNA repair genes, proteins, and processes Mutat Res 3 171 213
Stanhope MJ Lupas A Italia MJ Koretke KK Volker C 2001 Phylogenetic analyses do not support horizontal gene transfers from bacteria to vertebrates Nature 411 940 944 11418856
Lander ES Linton LM Birren B Nusbaum C Zody MC 2001 Initial sequencing and analysis of the human genome Nature 409 860 921 11237011
Atchley W Fitch W 1997 A natural classification of the basic helix-loop-helix class of transcription factors Proc Natl Acad Sci U S A 94 5172 5176 9144210
Ohno S 1970 Evolution by gene duplication New York Springer-Verlag 160 p.
Fitch WM 1970 Distinguishing homologous from analogous proteins Syst Zool 19 99 113 5449325
Lynch M Conery JC 2000 The evolutionary fate and consequences of duplicate genes Science 290 1151 1154 11073452
Beaumont M Rannala B 2004 The Bayesian revolution in genetics Nat Rev Genet 5 251 261 15131649
Cowell RG Dawid AP Lauritzen SL Spiegelhalter DJ 2003 Probabilistic networks and expert systems New York Springer-Verlag 321 p.
Hilden J 1970 Genex—An algebraic approach to pedigree probability calculus Clin Genet 1 319 348
Elston RC Stewart J 1971 A general model for the analysis of pedigree data Hum Hered 21 523 542 5149961
Felsenstein J 1981 Evolutionary trees from DNA sequences: A maximum likelihood approach J Mol Evol 17 368 376 7288891
Carlin BP Louis TA 2000 Bayes and empirical Bayes methods for data analysis, 2nd ed Boca Raton Chapman and Hall/CRC 419 p.
Huelsenbeck JP Ronquist F 2001 MrBayes: Bayesian inference of phylogenetic trees Bioinformatics 17 754 755 11524383
Ashburner M Ball CA Blake JA Botstein D Butler H 2002 Gene Ontology: Tool for the unification of biology. The Gene Ontology Consortium Nat Genet 25 25 29
Camon E Magrane M Barrell D Lee V Dimmer E 2004 The Gene Ontology annotation (GOA) database: Sharing knowledge in UNIPROT with Gene Ontology Nucleic Acids Res 32 262 266
Goodman M Cselusniak J Moore G Romero-Herrera A Matsuda G 1979 Fitting the gene lineage into its species lineage: A parsimony strategy illustrated by cladograms constructed from globin sequences Syst Zool 28 132 168
Page RDM 1998 Genetree: Comparing gene and species phylogenies using reconciled trees Bioinformatics 14 819 820 9918954
Bateman A Birney E Cerruti L Durbin R Etwiller L 2002 The Pfam protein families database Nucleic Acids Res 30 276 280 11752314
Ribard C Rochet M Labedan B Daignan-Fornier B Alzari P 2003 Sub-families of alpha/beta barrel enzymes: A new adenine deaminase family J Mol Biol 334 1117 1131 14643670
Keegan LP Leroy A Sproul D O'Connell MA 2004 Adenosine deaminases acting on RNA (ADARS): RNA-editing enzymes Genome Biol 5 209 14759252
Hirschhorn R Ellenbogen A 1986 Genetic heterogeneity in adenosine deaminase (ADA) deficiency: Five different mutations in five new patients with partial ADA deficiency Am J Hum Genet 38 13 25 3946419
Lushchak VI 1996 Functional role and properties of AMP-deaminase Biokhimiia 61 195 211 8717492
Maier SA Podemski L Graham SW McDermid HE Locke J 2001 Characterization of the adenosine deaminase-related growth factor (ADGF) gene family in Drosophila
Gene 280 27 36 11738815
Doolittle RF 1995 The multiplicity of domains in proteins Annu Rev Biochem 64 287 314 7574483
Auerbach G Ostendorp R Prade L Korndorfer I Dams T 1998 Lactate dehydrogenase from the hyperthermophilic bacterium Thermotoga maritima: The crystal structure at 21 A resolution reveals strategies for intrinsic protein stabilization Structure 6 769 781 9655830
Miller SS Driscoll BT Gregerson RG Gantt JS Vance CP 1998 Alfalfa malate dehydrogenase (MDH): Molecular cloning and characterization of five different forms reveals a unique nodule-enhanced MDH Plant J 15 173 184 9721676
Wu G Fiser A ter Kuile B Sali A Muller M 1999 Convergent evolution of Trichomonas vaginalis lactate dehydrogenase from malate dehydrogenase Proc Natl Acad Sci U S A 96 6285 6290 10339579
Devos D Valencia A 2000 Practical limits of function prediction Proteins 41 98 107 10944397
Neer EJ Schmidt CJ Nambudripad R Smith TF 1994 The ancient regulatory-protein family of WD-repeat proteins Nature 371 297 300 8090199
Gerlt JA Babbitt PC 2001 Diverent evolution of enzymatic function: Mechanistically diverse superfamilies and functionally distinct suprafamilies Annu Rev Biochem 70 209 246 11395407
Raychaudhuri S Chang JT Sutphin PD Altman RB 2002 Associating genes with Gene Ontology codes using a maximum entropy analysis of biomedical literature Genome Res 12 203 214 11779846
Andrade MA Valencia A 1998 Automatic extraction of keywords from scientific text: Application to the knowledge domain of protein families Bioinformatics 14 600 607 9730925
Roberts RJ 2004 Identifying protein function: A call for community action PLoS Biology 2 e42. DOI: 10.1371/journal.pbio.0020042 15024411
Bork P Koonin EV 1998 Predicting functions from protein sequences—Where are the bottlenecks? Nat Genet 18 313 318 9537411
Hadley C 2003 Righting the wrongs EMBO Rep 4 829 831 12949580
Karp P 2004 Call for an enzyme genomics initiative Genome Res 5 401
Jukes TH Cantor CR 1969 Evolution of protein molecules Munro HN Allison JB Mammalian protein metabolism New York Academic Press 21 132
Swofford D 2001 PAUP*: Phylogenetic analysis using parsimony (*and other methods) Sunderland (Massachusetts) Sinauer Associates
Zmasek CM Eddy SR 2001 A simple algorithm to infer gene duplication and speciation events on a gene tree Bioinformatics 17 821 828 11590098
King OD Foulger RE Dwight SS White JV Roth FP 2003 Predicting gene function from patterns of annotation Genome Res 13 896 904 12695322
Pearl J 1988 Probabilistic reasoning in intelligent systems: Networks of plausible inference San Mateo (California) Morgan Kaufman Publishers 552 p.
Stajich JE Block D Boulez K Brenner SE Chervitz SA 2002 The BioPerl toolkit: Perl modules for the life sciences Genome Res 12 1611 1618 12368254
White MF Vasquez J Yang SF Kirsh JF 1994 Expression of the apple 1-aminocyclopropane-1-carboxylate synthase in E. coli: Kinetic characterization of wild-type and active-site mutant forms Proc Natl Acad Sci U S A 91 12428 12432 7809054
|
16217548
|
PMC1246806
|
CC BY
|
2021-01-05 09:18:22
|
no
|
PLoS Comput Biol. 2005 Oct 7; 1(5):e45
|
utf-8
|
PLoS Comput Biol
| 2,005 |
10.1371/journal.pcbi.0010045
|
oa_comm
|
==== Front
PLoS Comput BiolPLoS Comput. BiolpcbiplcbploscompPLoS Computational Biology1553-734X1553-7358Public Library of Science San Francisco, USA 1626119110.1371/journal.pcbi.001004605-PLCB-RA-0122R4plcb-01-05-02Research ArticleBioinformatics - Computational BiologySystems BiologyDrug TargetsFbaGene EssentialityPathway ModellingFlux Balance Analysis of Mycolic Acid Pathway: Targets for Anti-Tubercular Drugs FBA of Mycolic Acid PathwayRaman Karthik Rajagopalan Preethi Chandra Nagasuma *Bioinformatics Centre, Supercomputer Education and Research Centre, Indian Institute of Science, Bangalore, IndiaSegre Daniel EditorBoston University, United States of America* To whom correspondence should be addressed. E-mail: [email protected] 2005 14 10 2005 4 8 2005 1 5 e462 6 2005 31 8 2005 Copyright: © 2005 Raman et al.2005This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.
Mycobacterium tuberculosis is the focus of several investigations for design of newer drugs, as tuberculosis remains a major epidemic despite the availability of several drugs and a vaccine. Mycobacteria owe many of their unique qualities to mycolic acids, which are known to be important for their growth, survival, and pathogenicity. Mycolic acid biosynthesis has therefore been the focus of a number of biochemical and genetic studies. It also turns out to be the pathway inhibited by front-line anti-tubercular drugs such as isoniazid and ethionamide. Recent years have seen the emergence of systems-based methodologies that can be used to study microbial metabolism. Here, we seek to apply insights from flux balance analyses of the mycolic acid pathway (MAP) for the identification of anti-tubercular drug targets. We present a comprehensive model of mycolic acid synthesis in the pathogen M. tuberculosis involving 197 metabolites participating in 219 reactions catalysed by 28 proteins. Flux balance analysis (FBA) has been performed on the MAP model, which has provided insights into the metabolic capabilities of the pathway. In silico systematic gene deletions and inhibition of InhA by isoniazid, studied here, provide clues about proteins essential for the pathway and hence lead to a rational identification of possible drug targets. Feasibility studies using sequence analysis of the M. tuberculosis H37Rv and human proteomes indicate that, apart from the known InhA, potential targets for anti-tubercular drug design are AccD3, Fas, FabH, Pks13, DesA1/2, and DesA3. Proteins identified as essential by FBA correlate well with those previously identified experimentally through transposon site hybridisation mutagenesis. This study demonstrates the application of FBA for rational identification of potential anti-tubercular drug targets, which can indeed be a general strategy in drug design. The targets, chosen based on the critical points in the pathway, form a ready shortlist for experimental testing.
Synopsis
M. tuberculosis, a deadly human pathogen, owes many of its unique qualities to its thick, waxy coat, containing fatty acids called mycolic acids. Several front-line drugs used for treating tuberculosis indeed inhibit mycolic acid synthesis. Understanding the biochemical pathway that makes these compounds is therefore of great interest. Availability of the genome sequence and various computational methods enable us to study pathways as whole functional units, rather than having to infer from the study of individual proteins. Here, we present a comprehensive identification of the components of the mycolic acid pathway and represent it mathematically based on reaction stoichiometry. Such models are amenable to perturbations and simulations using flux balance analysis, allowing the study of pathways from a metabolic capacity perspective, and yielding information about reaction fluxes. The perturbations studied here are in silico gene knock-outs and drug effects, which led us to identify genes essential to the pathway and hence for survival of the pathogen. The results are in good agreement with essentiality determined through experimental genetics. Such essential genes can be good targets for drug design, especially when they do not have homologues in the human proteome. FBA followed by sequence analyses have resulted in identification of potential anti-tubercular drug targets.
Citation:Raman K, Rajagopalan P, Chandra N (2005) Flux balance analysis of mycolic acid pathway: Targets for anti-tubercular drugs. PLoS Comput Biol 1(5): e46.
==== Body
Introduction
Genomics is rapidly changing the very foundation of several aspects of drug discovery research, one of them being a systems biology and bioinformatics approach for rational identification of drug targets. It is now possible to carry out a metabolic analysis of a system to gain insights into fundamental molecular mechanisms of several processes such as those that are critical for the survival of the pathogen. Here, we seek to apply such an approach to identify targets for designing anti-tubercular drugs. Despite the availability of several drugs and the Bacillus Calmette-Guérin vaccine, tuberculosis remains a major health concern worldwide, warranting identification of new drug targets for the design of more efficacious drugs.
The mycobacterial cell wall is distinctive and is associated with the pathogenicity of Mycobacterium tuberculosis [1–4]. The three polymers in the cell wall, arabinogalactan-mycolate [5] covalently linked with peptidoglycan and trehalose dimycolate, provide a thick layer that protects the tubercle bacillus from general antibiotics and the host's immune system [6]. The synthesis of mycolic acids—which are long-chain α-alkyl-β-hydroxy fatty acids, the major constituents of this protective layer—has been shown to be critical for the survival of M. tuberculosis [7]. InhA (EC 1.3.1.9, enoyl-[acyl-carrier-protein] reductase), involved in mycolic acid synthesis, also turns out to be the target for front-line anti-tubercular drugs [8], such as isoniazid [9] and ethionamide [10].
The mycolic acid pathway (MAP) has been of great interest, and a large amount of biochemical and genetic information is available in the literature, in addition to the entire genome sequence of M. tuberculosis. It is possible to exploit these large volumes of data to construct an in silico model of the pathway, which can then be simulated and analysed [11]. Constructing such models forms an important step in understanding the underlying molecular mechanisms of disease, and facilitates rational approaches to drug design. Several computational methods have emerged in recent years to simulate biochemical models, which aid in the systems approach to understanding pathways, processes, and whole-cell metabolism [12–18]. Flux balance analysis (FBA), a stoichiometric analysis technique, has been applied to study the metabolic capabilities of several systems [19,20], which has provided useful insights into cellular behaviour, including response to perturbations such as gene deletions [21,22].
Given the biological importance of mycolic acids, it would be useful to understand the behaviour of the pathway as a whole and of its individual components, both in a normal mycobacterial cell as well as upon perturbations, such as when a drug is acting upon it. We have therefore built a model of the MAP, represented it as a stoichiometric matrix, and performed FBA of this model, to gain insights into the critical steps of the pathway. Further, we have used the knowledge gained from these analyses for rational identification of putative drug targets and estimated their appropriateness by sequence analysis.
Results/Discussion
Model Description
The model of the MAP built here (Figure 1), containing 219 reactions and 197 metabolites, mediated through 28 proteins (Table 1), is according to our knowledge more complete and accurate with annotations from the latest literature than that from any other publicly available resource. The model in Systems Biology Markup Language (SBML) format is available as Dataset S1. The list of reactions has also been given in a flat file as Dataset S2. The biosynthesis of mycolic acids can be considered as made up of four sub-pathways: (A) production of malonyl CoA, (B) fatty acid synthase-I (FAS-I) pathway, (C) fatty acid synthase-II (FAS-II) pathway, and (D) condensation of FAS-II and FAS-I products into α- (D1), methoxy- (D2), and keto-mycolic acids (D3). FAS-I and FAS-II (B and C sub-pathways) are dependent upon the production of malonyl CoA (produced in A). The products of B and C are then converted into different mycolic acids in D.
Figure 1 Schematic Diagram of the MAP in M. tuberculosis
(A–D) refer to the four sub-pathways (see text). The key metabolites are indicated in larger type. Proteins catalysing each reaction are indicated to the right of the reaction arrows, while the reaction numbers are indicated to the left. Reaction cycles have been indicated as, for instance, 65:6:185 (for fabG1), which is to be interpreted as 65, 71, 77, ..., 185. ⊗ indicates inhibition of InhA by isoniazid and ethionamide in the pathway. The reaction numbers in parentheses indicate reactions for the trans forms in (D2) and (D3).
Table 1 Table of Proteins and Their Corresponding Genes in MAP and Sources for Inference of Their Reactions
The FAS-I system, present predominantly in eukaryotes, is capable of de novo fatty acid synthesis, whereas the FAS-II system in mycobacteria, although similar to that in other bacteria, elongates the products of FAS-I, resulting in the production of meromycolates, key precursors of mycolic acids [2]. The basic reactions in FAS-I and FAS-II are a repetition of a cycle of four reactions, each cycle culminating in the extension of the alkyl chain by a two-carbon unit (Figure 2). The FAS-I enzyme is a single polypeptide with multiple domains catalysing a cycle of reactions to generate short-chain acyl CoA esters [23]. FAS-I exhibits a bimodal product distribution: C16 to C18 and C24 to C26 acyl CoAs [24]. These form the substrates for the FAS-II reaction cycle and the polyketide synthase enzyme, respectively. β-ketoacyl-synthase III forms a pivotal link between FAS-I and FAS-II [25,26]. The FAS-II system is composed of four enzyme reactions iteratively converting C16-acyl CoA to C58-acyl-ACP (meromycolate) [27–31].
Figure 2 Flux Distributions Obtained from FBA Using the MAP Model and Objective Function c
1
(A) in an unperturbed state, (B) upon deletion of inhA, (C) upon deletion of pcaA, and (D) upon inhibition of InhA. Insets in (A) and (C) refer to enlarged versions of the indicated portions. Note that the scale for (D) is different. It may be noted that the lines joining the various flux points have been drawn to aid in discerning the flux peaks clearly; the lines as such have no physical significance.
Except for reactions 190–192, experimental data clearly indicating the involvement of the appropriate proteins are available in literature. However, clear identification of the proteins referred to as UNK1 and UNK2 is not yet available. Although no explicit experimental evidence is available for reactions 190–192, the involvement of desA1, desA2, and desA3 has been suggested based on sequence annotations [27] and indirect experimental evidence [32], thus justifying their inclusion in this model. The cis-unsaturated meromycolate chain further undergoes cyclo-propanation, processing for keto- and methoxy-mycolic acids [3,4,33–39] and Claisen condensation with the FAS-I product C24-acyl CoA [6,40], to yield α-, methoxy-, or keto-mycolic acids, as shown in Figure 1D.
Biochemical characterisation of mycolic acids in M. tuberculosis H37Rv cell cultures clearly indicate that α-mycolate is the predominant mycolic acid and it comprises as much as 49% of the mycolates in the cell wall, whereas methoxy- and keto-mycolates are present in smaller quantities of 27% and 24%, respectively [41]. These data have been considered during FBA of the MAP model.
FBA
The MAP system outlined here, though involving only 28 proteins, has 197 metabolites participating in 219 reactions with high interconnectivity, and thus benefits from a systematic FBA. The system considered here, though relatively small in comparison to genome-scale metabolic models previously studied by FBA [19–22], is still complete in its own right (analogous to a separate module) and gives profound insights into mycolic acid metabolism. More importantly, with a single pathway in question, the objective functions for optimisation can be better defined and have specific biological relevance that can be related to experimental data quite readily.
The stoichiometric matrix for this system is of size 197 × 247. The vector v (for details, see Materials and Methods) has 247 fluxes, including 28 exchange fluxes:
Objective function: Mycolic acids are known to play a key role in the structural integrity of the mycobacterial cell wall and have been shown to be critical both for growth [42–44] of the bacillus and its pathogenicity [1–4] by several experiments. These data imply that the mycobacterial cell would be geared toward optimal production of mycolic acids, in terms of maximal molar yield of the appropriate mycolic acids for a given genotype.
Given that the cell wall contains different types of mycolates in varying proportions, the optimal production of mycolates can be captured in two different ways: (a) only the most important mycolate is produced, and (b) the known ratios of different mycolates are fixed. We encode these two scenarios as two objective functions, c
1 and c
2, respectively:
and
where v
mycolates represents the flux of a hypothetical reaction:
0.4926 α-mycolate + 0.2334 cis-methoxy-mycolate
+ 0.0327 trans-methoxy-mycolate + 0.2117 cis-keto-mycolate
+ 0.0297 trans-keto-mycolate → mycolate-biomass.
The coefficients of the fluxes of α-mycolates, cis-methoxy and trans-methoxy mycolates, as well as cis-keto and trans-keto mycolates indicated in the above equations are based on the biochemical data about α-, methoxy-, and keto-mycolates present in the cell wall in the ratio of 1.0:0.54:0.49, with the cis forms dominating the trans forms of the methoxy- and keto-mycolates, in a ratio of 1:0.14 [41]. Production of different mycolates follows much of the same pathway but differs only in the end stages (Figure 1D), suggesting that they may not all be made simultaneously.
The objective functions used in previous FBA studies (such as in [21]) have been capturing the optimal production of the biomass, which implicitly fixes the proportions of the different components of the biomass. The objective function c
2 captures an analogous scenario, where the total mycolate content can be considered as the biomass. We feel that c
1 reflects the biological situation more closely than c
2 for this analysis for the following reasons: (a) it has been reported [4] that the mycobacterial cell can survive in the absence of one of the mycolate components, which cannot be accounted by c
2; (b) the objective function c
2 directly precludes the possibility of the cell surviving in the absence of even a single mycolic acid; for any genotype, c
2 necessitates that the mycolates be produced in an all-or-none fashion in the corresponding phenotype. On the other hand, c
1 favours the production of the most important mycolate that can be produced under the given conditions (i.e., following some gene deletions); and (c) reported biological data [44] suggest that the three major mycolates are in fact made at different phases of cell growth.
However, to make the analysis comprehensive, both objective functions have been used independently and the results presented. In the first case, the objective function c
1 (Equation 2) accounts for the relative importance of the mycolates. The coefficients are indicative of the precise ranking order of the various mycolates, based on cell wall composition. On the other hand, the objective function in the second case, c
2 (Equation 3), fixes the relative ratios of the mycolates based on the absolute values of the coefficients.
Once an objective function is fixed, the system translates to solving a linear programming (LP) problem as in Equation 7 (listed in Materials and Methods). The solution of the LP problem, using objective functions c
1 and c
2 yields flux distributions specifying the fluxes of all the internal reactions and the exchanges, as shown in Figures 2A and 3A, respectively. Both figures indicate two major peaks at reactions 3–4 and 63, apart from intense peaks at reactions 220–247. The peaks at reactions 3–4 and 63 correspond to the production of metabolites BCCP-biotin, malonyl CoA, and malonyl acyl carrier protein. The positive and negative peaks at reactions 220–247 correspond to the exchange fluxes originating from external metabolites such as ATP, NADP, NADPH, and CO2, indicating their exit from or entry into the MAP system, respectively. A repetitive pattern is observed for reactions 65–172, which is comprehensible in view of the cyclic nature of the reactions involved in extension of the carbon chain in FAS-II. Malonyl CoA is an important metabolite, since it is required not only for the formation of C4-acyl-ACP, but also for each of the ten steps of chain elongation by the FAS-I system, where the chain length of the fatty acid component of mycolic acid grows from four carbons to 24 carbons (see Figure 1). Besides, it is also required for the synthesis of malonyl acyl carrier protein, which in turn is required for chain elongation in each of the 20 chain elongation steps catalysed by the FAS-II system, where the chain length grows from C16 to C52–C58. The large fluxes seen for reaction 4, which produces malonyl CoA, and reaction 3, which produces BCCP-biotin, an immediate precursor of malonyl CoA, are explained by their high requirement. The exchange fluxes for external metabolites such as ATP, ADP, NADP, and NADPH are also understandably very high, since they are either utilised or produced in large quantities in the pathway.
Figure 3 Flux Distributions Obtained from FBA Using the MAP Model and Objective Function c
2
(A) in an unperturbed state, (B) upon deletion of pcaA. Inset in A refers to an enlarged version of the indicated portion. Note that the scale for (B) is different. It may be noted that the lines joining the various flux points have been drawn to aid in discerning the flux peaks clearly; the lines as such have no physical significance.
While the flux distributions using either objective function are largely similar for the reactions belonging to sub-pathways A, B, and C (see Figure 1), significant differences are observed for the fluxes of the reactions producing mycolates and those immediately related to them. With objective function c
1 (Figure 2A), a small peak at reaction 197 corresponding to α-mycolate is seen, as shown in more detail in the enlarged inset. It should be noted that, in this unperturbed state, the amounts of methoxy- and keto-mycolates produced are negligible, compared with that of α-mycolate. On the other hand, with objective function c
2, all the mycolates are produced in the ratio mentioned earlier (Figure 3A, enlarged image in inset), as imposed by the objective function.
Perturbations
Effects of in silico gene deletions, using c
1.
The perturbations carried out on the MAP model using FBA were (a) in silico gene deletions and (b) inhibition by known drugs. Each of the 28 genes and hence its gene product was systematically deleted from the MAP model, one at a time, and its effect on the flux distribution was analysed (Table 2). Figures 2B and 2C are examples of flux distributions upon deletion of inhA and pcaA, respectively. Upon deletion of inhA, which catalyses every sixth reaction from 69 to 189 (see Figure 1C), the fluxes of almost all reactions were seen to be zero, except for reactions 1–2 and their corresponding external metabolites. On the other hand, deletion of pcaA, which is involved only in the production of α-mycolate (Figure 1D1), the flux pattern remained largely unaltered, except for an increase in the flux corresponding to cis-methoxy-mycolate (Figure 1D2). A flat flux distribution profile (of near zero) was observed upon deletion of 16 of the genes (and hence their gene products) in the MAP model. In D1, D2, and D3 sub-pathways, genes common to all three, such as desA1, fall into this category. On the other hand, genes responsible for the production of one or the other mycolate, such as pcaA, involved in D1, when deleted, did not significantly alter the overall flux distribution and are hence classified as non-essential. This is comprehensible, since in these cases, the system is still capable of producing the other two mycolates. Although it has been reported that all three mycolates in appropriate proportions are required for mycobacterial persistence and virulence, growth and survival to varying extents have been observed in the presence of any one component [4,45]. Experimental data available for pcaA deletion indeed show that the mutant bacilli can grow and survive for limited periods by producing a significant excess of keto-mycolate, to compensate for the absence of α-mycolate. Our results too show a similar compensatory effect upon deletion of pcaA, consistent with the experimental results about its non-essentiality [4,46]. In our results, however, methoxy- rather than keto-mycolate was produced in excess, because of the incorporation of their relative abundances available in literature [41] into the objective function. If the proportion of keto-mycolate was higher than that of methoxy-mycolate, then we would have observed a higher proportion of keto- rather than methoxy-mycolate, in our results. However, given the lack of more experimental details on the composition under different conditions, or the exact functional role of each of the mycolates, this difference does not seem too significant, at this stage.
Table 2 Results of In Silico Gene Deletion Studies for the 28 Genes in the MAP Model, Using Objective Functions c
1 and c
2
Effects of in silico gene deletions, using c
2.
Systematic in silico gene deletion studies were carried out using the MAP model and with c
2 as the objective function. Here too, each of the 28 genes and hence its product was systematically deleted from the MAP model, one at a time, and its effect on the flux distribution was analysed, as presented in Table 2. An example where a significant difference was found with respect to the perturbations using c
1 is illustrated by pcaA deletion. Upon deletion of pcaA, using c
2, almost all of the reaction fluxes were seen to have dropped to zero, except for reactions 1–2 and their corresponding external metabolites (Figure 3B). This is in contrast to the corresponding gene deletion, using c
1, observed in Figure 2C, where the flux distribution is very similar to that in the unperturbed case. These results suggest pcaA to be essential if c
2 is used, in contrast to the results obtained using c
1 as well as reported biological data discussed above, indicating c
1 to be a better objective function. Since the objective function c
2 demands the production of all mycolates in the appropriate ratio, apart from pcaA, five other genes that were non-essential in the analysis using c
1 were classified as essential.
Minimisation of metabolic adjustment (MOMA), using the methodology described by Segrè and co-workers [47], was also carried out for all the gene deletions (using c
1 and c
2 as objective functions in separate studies). There were no significant changes in the flux profiles, and hence in the interpretations, as compared with those described above.
Experimental data from systematic gene deletion studies using the transposon site hybridisation mutagenesis technique are available in the literature [46]. Comparison of such data for these 28 genes with the results obtained from our FBA study (using both objective functions) is shown in Table 2. With c
1, a good correlation was observed for 19 genes, no experimental data was available for four genes, and disagreement was seen for only five genes. High correlation with experimentally observed data about the essentiality of individual genes indicates the usefulness of our MAP model and its study using FBA. With c
2, a good correlation was observed for 13 genes and disagreement was seen for 11 genes.
On analysing the results of the gene deletion studies obtained using c
1 and c
2, it is apparent that the objective function c
1 is able to reflect the biological situation better. c
2 requires the production of all mycolates in a definite ratio under all conditions, which is not very appropriate, considering the fact that cells can survive even in the absence of one or more mycolates [4] and that in any state, only a single mycolate is produced in the cell. Hence, for the rest of the analyses, we have restricted the discussion to the gene deletion results obtained using c
1.
The possible reasons for the disagreements (using c
1) for accD3, fabG2, fabH, inhA, and desA3 are discussed below. We had identified malonyl CoA as an internal metabolite, in the absence of concrete experimental evidence of its being produced by other reactions in the cell. However, if indeed malonyl CoA is produced by the cell through some other means, then AccD3 would no longer be essential, agreeing with the experimental results. fabG2 has been reported as essential, while our analysis identified it as non-essential. Clearly, in our model, FabG2 can be substituted for by either FabG1 or FabG4. However, it may be possible that FabG2 could be responsible for catalysing some other critical reaction outside the MAP, which could contribute to its essentiality. fabH is a critical gene, whose product catalyses the important step that links FAS-I to FAS-II. Our sequence analysis studies also show that FabH has no homologues in the mycobacterial proteome. It is unclear as to why this is a non-essential gene in experimental studies. Similarly, inhA, identified as essential in our analysis but reported as non-essential in the experimental studies using transposon site hybridisation mutagenesis, is a well-known target for drugs such as isoniazid and ethionamide. While that may be the case for the conditions under which the transposon site hybridisation mutagenesis experiment was carried out, it is well-known that inhibition of InhA leads to a significant reduction in the growth of mycobacteria, making its inhibitors as front-line drugs. InhA is also known to be essential for mycolic acid synthesis [48], which in turn is known to be essential for survival of the pathogen [7]. In fact, InhA has been shown to be one of the few highly over-expressed proteins inside an infected macrophage [48]. The topology of the curated reaction network clearly makes InhA an essential gene for the system, which is in agreement with its known role for mycobacterial survival [48]. Another possibility could be that structural but not sequence homologues of InhA and FabH, which are not yet well-characterised (and hence not a part of our model), may substitute in their absence. desA3 has been reported as non-essential but was identified as essential from our analysis. It is possible that our model may not have accounted for its exact physiological role, due to lack of information in literature.
Inhibition studies were carried out for isoniazid, since its inhibition of InhA has been well-characterised in the literature [9,10]. Inhibition in the context of FBA is in fact similar to that of deletion studies of the corresponding gene, except that the latter will lead to total inactivation of that gene, whereas inhibition by a drug need not necessarily lead to total inactivation. Just to represent the relative effect upon partial inactivation, we have considered a scenario where isoniazid would inhibit InhA to an extent of 90%. The flux profile shown in Figure 2D indicates much lower fluxes for each of the reactions. Similar results will be obtained for any inhibitor of InhA. Thus, the model and the method, besides being consistent with a requirement of a high percentage of inhibition for an ideal drug, also show their utility in analysing drug action when any quantitative data become available.
This method also has the potential to consider inhibition at multiple points. For example, isoniazid is thought to act at two points in the pathway (InhA and KasA), although conclusive experimental proof is still awaited [49]. The FBA study here presents a ready framework to analyse the effects of such drug inhibitions, which would be extremely difficult to judge by an inspection of the reaction map alone.
While the usefulness of FBA for large systems with high network connectivities and redundancy is well established, its application for specific pathways, which can be considered as simpler systems, has not yet been well explored in the literature. The study reported here illustrates the usefulness of FBA even for individual pathways. The effects of the perturbations to a system even of this size, either at single points or at multiple points, are beyond unambiguous comprehension and thus benefit from systematic studies such as FBA to get meaningful results. Moreover, FBA provides a handle to systematically identify essential genes in the pathway, irrespective of the size of the system, in a systematic, efficient, and much simplified manner. An advantage of considering specific pathways individually is that the objective functions for optimisation can be better defined, with specific biological relevance, to generate hypotheses useful for designing molecular biology studies quite readily. It must be noted that at the present time our understanding of systems, and hence their reconstructions in general, is not sufficient to generate knowledge that can replace biochemical or genetic experiments. However, an in silico framework for predicting gene essentiality, while complementing experimental data where available, has the additional advantage of enabling studies under various environmental conditions such as during low nutrition or upon oxidative stress, or even in the presence of drugs, which are difficult to perform experimentally.
Identification of Drug Targets
Those genes that were classified as essential in the above analysis automatically form a first list of putative targets for anti-tubercular drugs, since their total inactivation results in loss of production of mycolic acids and hence the viability or the pathogenicity of the bacillus. However, it was reasoned that an ideal target should be essential not only in terms of the reaction it can catalyse, but also as the only protein coded by the genome that can perform the same task. Moreover, an ideal target should also have no recognisable homologue in the host system, which can in principle compete with the same drug, leading to unintended/adverse effects in the host system. Sequence analysis with the M. tuberculosis H37Rv and human proteomes was therefore carried out for each of the identified targets and the results are summarised in Table 3.
Table 3 Homologues Present in M. tuberculosis and Human Proteomes for Genes Identified As Essential, Based on In Silico Gene Deletion Studies
Of the 16 proteins classified as essential in Table 2, no close homologues were observed within the M. tuberculosis H37Rv proteome for seven proteins: FabH, AccD3, InhA, FabD, Fas, Pks13, and DesA3. Similarities greater than 50% using the BLOSUM62 substitution matrix with an e-value of less than 0.1 for a length greater than 70% of the query sequence were considered as close homologues (in both proteomes). For identifying more distant/fast-evolving homologues in the human proteome, homologues were identified with a second set of criteria, considering similarities greater than 30% using the BLOSUM62 substitution matrix with an e-value of less than 10−5 for greater than 70% of the query sequence length. None of the seven proteins have close homologues in the human proteome. A distant homologue was identified with the second set of criteria for FabD. Those proteins, which have no other homologues either in the mycobacterial proteome or in the human proteome, are therefore obvious potential drug targets. Proteins lacking homologues with multiple cut-offs can be considered targets with greater confidence. Identification of their presence only in the bacterial cell helps in the process of validation as useful drug targets. Front-line anti-tubercular drugs in current clinical practice, isoniazid and ethionamide, in fact turn out to be inhibitors of InhA [9,10], preventing mycolic acid synthesis. Thus, the inhibition of the identified targets, which would all lead to impairment of mycolic acid synthesis, appears to be a promising strategy for designing anti-tubercular agents.
Besides the seven targets mentioned above, DesA1, DesA2, and FadD32 do not have any close homologues in the human proteome but have homologues in the M. tuberculosis H37Rv proteome. Of these, DesA1 and DesA2 are homologues of each other, but it is not clear as yet whether they are required together or if they can substitute for one another. KasA and KasB too are homologues of each other and do not share similarities with any other mycobacterial protein. However, they share considerable sequence similarities with a hypothetical human protein FLJ20604. FadD32 too exhibits distant homology with six other proteins in the human proteome. Such mycobacterial proteins would also be interesting drug targets, provided such homologies are considered during drug design.
In conclusion, the work presented here provides a framework to rationally identify targets for use in tuberculosis drug design and provides a ready shortlist that can be experimentally tested. This also outlines a general strategy for analysis of microbial metabolism, providing insights into targets for drug design. Systems approaches are being increasingly applied for understanding the metabolic capabilities of organisms, which can be exploited for drug design. A major bottleneck in this process is the accuracy of the model, which requires expert curation of available literature. The MAP model presented here should be of value not only in drug design but also for understanding mycolic acid synthesis in general. The model can also be adapted to perform quantitative simulations when kinetic data become available, and it can be used as a framework for incorporating newer or alternate components when such information becomes available.
Materials and Methods
Model building.
Initially, the Kyoto Encyclopedia of Genes and Genomes database was explored to obtain information about various proteins that comprise the MAP in M. tuberculosis H37Rv. Information about the FAS-I and FAS-II pathways in M. tuberculosis was available in the Kyoto Encyclopedia of Genes and Genomes database [50], but appeared to be based on pre-genomic annotations and was also incomplete and inconsistent in parts. The BioCyc repository of pathway models [51], on the other hand, had more recent annotations and provided a basic framework to build the MAP model. It contained information about 11 proteins. However, no explicit data were available about the specific reactions during the fatty acid elongation steps, catalysed by the FAS-I and FAS-II enzyme systems. Further, no information was available for the conversion of the FAS-II products to mycolates. Available literature (detailed in Table 1), as well as annotations in the TubercuList (http://genolist.pasteur.fr/TubercuList/) database, were therefore carefully analysed to fill the missing links and obtain a comprehensive model of the MAP, which contained information about 28 proteins, catalysing 219 reactions involving 197 metabolites, thus yielding a detailed pathway landscape. The model was encoded using SBML Level 2 Version 1 [52].
The set of reactions in the landscape were then mathematically represented as a stoichiometric matrix, S
m × n, with every metabolite being represented by a row (m metabolites) and every reaction by a column (n reactions). The entries in each column correspond to the stoichiometric coefficients of the metabolites (negative for reactants and positive for products) for each reaction. The ith row of the matrix defines the participation of a particular metabolite across all metabolic reactions, and the jth column provides the stoichiometry of all metabolites in that reaction. Exchange fluxes were considered for metabolites that are typically exchanged with the environment (e.g., ATP, ADP, NAD[P], and NAD[P]H) and those that are not produced by the system, or those that are produced in the system for use in other metabolic pathways. Such metabolites, referred to as external metabolites, will have corresponding exchange fluxes in the pathway. The other metabolites were regarded as internal, which are produced by the system and consumed within the system itself. The external metabolites in the MAP model were identified manually. A list of external metabolites has also been supplied as part of Dataset S2.
FBA.
FBA [12,53,54] has been shown to be a very useful technique for analysis of metabolic capabilities of cellular systems [20,21,55,56]. FBA involves carrying out a steady-state analysis, using the stoichiometric matrix for the system in question. The system is assumed to be optimised with respect to functions such as maximisation of biomass production or minimisation of nutrient utilisation, following which it is solved to obtain a steady-state flux distribution. This flux distribution is then used to interpret the metabolic capabilities of the system. The dynamic mass balance of the metabolic system is described using the stoichiometric matrix, relating the flux rates of enzymatic reactions, v
n
× 1 to time derivatives of metabolite concentrations, x
m
× 1 as
where vi signifies the internal fluxes, bi represents the exchange fluxes in the system, and n
ext is the number of external metabolites in the system. At steady-state,
Therefore, the required flux distribution belongs to the null space of S. Since m < n, the system is under-determined and may be solved for v fixing an optimisation criterion, following which, the system translates into an LP problem:
where c represents the objective function composition, in terms of the fluxes. Further, we can constrain:
which necessitates all internal irreversible reactions to have a flux in the positive direction and allows exchange fluxes to be in either direction. Practically, a finite upper bound can be imposed, so that the problem does not become unbounded. This upper bound may also be decided based on the knowledge of cellular physiology. In our analysis, the upper bound was set as unity, which then effectively gives the relative ratios of the reaction fluxes. MATLAB (http://www.mathworks.com/) was used for solving the LP problem. The “linprog” routine from the Optimization Toolbox was used, which uses a large-scale interior point algorithm.
Perturbations.
FBA also has the capabilities to address the effect of gene deletions and other types of perturbations on the system. Gene deletion studies were performed by constraining the reaction flux(es) corresponding to the gene(s) (and therefore, of their corresponding proteins[s]), to zero. Effects of inhibitors of particular proteins were also studied in a similar way, by constraining the upper bounds of their fluxes to any defined fraction of the normal flux, corresponding to the extents of inhibition.
MOMA.
MOMA [18,47] is a technique similar to FBA, particular to the analysis of perturbed systems and has been reported to outperform FBA in certain cases. MOMA circumvents the use of an objective function for optimisation under perturbed conditions and rather solves for a flux distribution closest to the unperturbed system, subject to the new constraints imposed, minimising the metabolic adjustment of the system.
Analysis of the effects of deletion of individual genes on the flux profiles of the five mycolates provided us a handle to define essential and non-essential genes. Those deletions that resulted in zero or near-zero fluxes of all the mycolates were considered as essential, and the rest were considered as non-essential.
Feasibility analysis of putative targets.
Sequence analysis was carried out using BLAST [57] to adjudge the feasibility of the putative targets identified through FBA. Homologues were searched for within the M. tuberculosis H37Rv and human proteomes, using BLOSUM62 substitution matrix. The BLAST outputs were parsed with home-grown scripts using BioPython modules, to identify homologues that satisfied various length and similarity criteria.
Supporting Information
Dataset S1 SBML Model of the MAP
An SBML model for the MAP in M. tuberculosis.
(124 KB XML)
Click here for additional data file.
Dataset S2 Flat File Containing Reactions in the MAP
A flat file containing all the reactions in the MAP, along with the genes corresponding to proteins that catalyse the reactions. The external metabolites have also been indicated in the file.
(19 KB TXT)
Click here for additional data file.
Accession Number
The SwissPROT/TrEMBL (http://www.ebi.ac.uk/trembl/) accession number for hypothetical human protein FLJ20604 is Q9NWU1. Gene accession numbers are listed in Table 1.
Financial support from the computational genomics initiative of the Department of Biotechnology is gratefully acknowledged. Use of facilities at the Interactive Graphics Based Molecular Modeling Facility and Distributed Information Centre (both supported by the Department of Biotechnology) and the facilities at the Supercomputer Education and Research Centre are also gratefully acknowledged.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. KR, PR, and NC conceived and designed the experiments. KR and PR performed the experiments. KR, PR, and NC analysed the data, and wrote the paper.
A previous version of this article appeared as an Early Online Release on August 4, 2005 (DOI: 10.1371/journal.pcbi.0010046.eor).
Abbreviations
FAS-Ifatty acid synthase-I
FAS-IIfatty acid synthase-II
FBAflux balance analysis
LPlinear programming
MAPmycolic acid pathway
MOMAminimisation of metabolic adjustment
SBMLSystems Biology Markup Language
==== Refs
References
Smith I 2003
Mycobacterium tuberculosis : Pathogenesis and molecular determinants of virulence Clin Microbiol Rev 16 463 496 12857778
Barry CE III, Lee RE, Mdluli K, Simpson AE, Schroeder BG, et al. 1998 Mycolic acids: Structure, biosynthesis and physiological functions Prog Lipid Res 37 143 179 9829124
Dubnau E, Chan J, Raynaud C, Mohan V, Lanéelle M, et al. 2000 Oxygenated mycolic acids are necessary for virulence of Mycobacterium tuberculosis in mice Mol Microbiol 36 630 637 10844652
Glickman M Cox J Jacobs W 2000 A novel mycolic acid cyclopropane synthetase is required for cording, persistence, and virulence of Mycobacterium tuberculosis
Mol Cell 5 717 727 10882107
Crick DC Mahapatra S Brennan PJ 2001 Biosynthesis of the arabinogalactan-peptidoglycan complex of Mycobacterium tuberculosis
Glycobiology 11 107R 118R
Takayama K Wang C Besra GS 2005 Pathway to synthesis and processing of mycolic acids in Mycobacterium tuberculosis
Clin Microbiol Rev 18 81 101 15653820
Draper P Daffé M 2005 The cell envelope of Mycobacterium tuberculosis with special reference to the capsule and outer permeability barrier Cole ST Tuberculosis and the tubercle bacillus Washington, D.C. American Society of Microbiology Press 261 273
Pasqualoto KFM Ferreira EI Santos-Filho OA Hopfinger AJ 2004 Rational design of new antituberculosis agents: Receptor-independent four-dimensional quantitative structure-activity relationship analysis of a set of isoniazid derivatives J Med Chem 47 3755 3764 15239654
Lei B Wei C Tu S 2000 Action mechanism of antitubercular isoniazid. Activation by Mycobacterium tuberculosis KatG, isolation, and characterization of inhA inhibitor J Biol Chem 275 2520 2526 10644708
Banerjee A, Dubnau E, Quemard A, Balasubramanian V, Um K, et al. 1994
inhA , a gene encoding a target for isoniazid and ethionamide in Mycobacterium tuberculosis
Science 263 227 230 8284673
Ma H Zeng AP 2003 Reconstruction of metabolic networks from genome data and analysis of their global structure for various organisms Bioinformatics 19 270 277 12538249
Bonarius HPJ Schmid G Tramper J 1997 Flux analysis of underdetermined metabolic networks: The quest for the missing constraints Trends Biotech 15 308 314
Schuster S Dandekar T Fell DA 1999 Detection of elementary flux modes in biochemical networks: A promising tool for pathway analysis and metabolic engineering Trends Biotech 17 53 60
Voit EO 2000 Computational analysis of biochemical systems: A practical guide for biochemists and molecular biologists Cambridge Cambridge University Press 531 p.
Stelling J Klamt S Bettenbrock K Schuster S Gilles ED 2002 Metabolic network structure determines key aspects of functionality and regulation Nature 420 190 193 12432396
Cornish-Bowden A Hofmeyr JHS 2002 The role of stoichiometric analysis in studies of metabolism: An example J Theor Biol 216 179 191 12079370
Price ND Papin JA Schilling CH Palsson BO 2003 Genome-scale microbial in silico models: The constraints-based approach Trends Biotech 21 162 169
Segrè D, Zucker J, Katz J, Lin X, D'haeseleer P, et al. 2003 From annotated genomes to metabolic flux models and kinetic parameter fitting OMICS 7 301 316 14583118
Edwards JS Palsson BO 1999 Systems properties of the Haemophilus influenzae Rd metabolic genotype J Biol Chem 274 17410 17416 10364169
Forster J Famili I Fu P Palsson BO Nielsen J 2003 Genome-scale reconstruction of the Saccharomyces cerevisiae metabolic network Genome Res 13 244 253 12566402
Edwards JS Palsson BO 2000 The Escherichia coli MG1655 in silico metabolic genotype: Its definition, characteristics, and capabilities Proc Natl Acad Sci U S A 97 5528 5533 10805808
Becker SA Palsson BO 2005 Genome-scale reconstruction of the metabolic network in Staphylococcus aureus N315: An initial draft to the two-dimensional annotation BMC Microbiology 5 8 15752426
Schweizer E Hofmann J 2004 Microbial type I fatty acid synthases (FAS): Major players in a network of cellular FAS systems Microbiol Mol Biol Rev 68 501 517 15353567
Kikuchi S Rainwater DL Kolattukudy P 1992 Purification and characterization of an unusually large fatty acid synthase from Mycobacterium tuberculosis var. bovis BCG
Arch Biochem and Biophys 295 318 326 1586161
Choi KH Kremer L Besra GS Rock CO 2000 Identification and substrate specificity of β-ketoacyl (acyl carrier protein) synthase III (mtFabH) from Mycobacterium tuberculosis
J Biol Chem 275 28201 28207 10840036
Scarsdale JN Kazanina G He X Reynolds KA Wright HT 2001 Crystal structure of the Mycobacterium tuberculosis β-ketoacyl-acyl carrier protein synthase III J Biol Chem 276 20516 20522 11278743
Cole ST, Brosch R, Parkhill J, Garnier T, Churcher C, et al. 1998 Deciphering the biology of Mycobacterium tuberculosis from the complete genome sequence Nature 393 537 544 9634230
Kremer L, Dover L, Carrère S, Nampoothiri K, Lesjean S, et al. 2002 Mycolic acid biosynthesis and enzymic characterization of the beta-ketoacyl-ACP synthase A-condensing enzyme from Mycobacterium tuberculosis
Biochem J 364 423 430 12023885
Marrakchi H, Ducasse S, Labesse G, Montrozier H, Margeat E, et al. 2002 MabA (FabG1), a Mycobacterium tuberculosis protein involved in the long-chain fatty acid elongation system FAS-II Microbiology 148 951 960 11932442
Quémard A, Sacchettini JC, Dessen A, Vilcheze C, Bittman R, et al. 1995 Enzymatic characterization of the target for isoniazid in Mycobacterium tuberculosis
Biochemistry 34 8235 8241 7599116
Schaeffer ML, Agnihotri G, Volker C, Kallender H, Brennan PJ, et al. 2001 Purification and biochemical characterization of the Mycobacterium tuberculosis β-ketoacyl-acyl carrier protein synthases KasA and KasB J Biol Chem 276 47029 47037 11600501
Phetsuksiri B, Jackson M, Scherman H, McNeil M, Besra GS, et al. 2003 Unique mechanism of action of the thiourea drug isoxyl on Mycobacterium tuberculosis
J Biol Chem 278 53123 53130 14559907
Glickman M Cahill S Jacobs W 2001 The Mycobacterium tuberculosis cmaA2 gene encodes a mycolic acid trans-cyclopropane synthetase J Biol Chem 276 2228 2233 11092877
Dinadayala P Laval F Raynaud C Lemassu A Laneelle M 2003 Tracking the putative biosynthetic precursors of oxygenated mycolates of Mycobacterium tuberculosis
J Biol Chem 278 7310 7319 12473649
George K Yuan Y Sherman D Barry C 1995 The biosynthesis of cyclopropanated mycolic acids in Mycobacterium tuberculosis . Identification and functional analysis of CMAS-2 J Biol Chem 270 27292 27298 7592990
Glickman MS 2003 The mmaA2 gene of Mycobacterium tuberculosis encodes the distal cyclopropane synthase of the α-mycolic acid J Biol Chem 278 7844 7849 12502719
Yuan Y Barry CE III 1996 A common mechanism for the biosynthesis of methoxy and cyclopropyl mycolic acids in Mycobacterium tuberculosis
Proc Natl Acad Sci U S A 93 12828 12833 8917504
Yuan Y Lee R Besra G Belisle J Barry C 1995 Identification of a gene involved in the biosynthesis of cyclopropanated mycolic acids in Mycobacterium tuberculosis
Proc Natl Acad Sci U S A 92 6630 6634 7604045
Yuan Y Crane D Musser J Sreevatsan S Barry C 1997 MMAS-1, the branch point between cis - and trans -cyclopropane-containing oxygenated mycolates in Mycobacterium tuberculosis
J Biol Chem 272 10041 10049 9092547
Portevin D de Sousa-D'Auria C Montrozier H Houssin C Stella A 2005 The acyl-AMP ligase FadD32 and AccD4-containing acyl-CoA carboxylase are required for the synthesis of mycolic acids and essential for mycobacterial growth J Biol Chem 280 8862 8874 15632194
Watanabe M Aoyagi Y Ridell M Minnikin D 2001 Separation and characterization of individual mycolic acids in representative mycobacteria Microbiology 147 1825 1837 11429460
Takayama K Wang L David HL 1972 Effect of isoniazid on the in vivo mycolic acid synthesis, cell growth, and viability of Mycobacterium tuberculosis
Antimicrob Agents Chemother 2 29 35 4208567
Vilchèze C Morbidoni HR Weisbrod TR Iwamoto H Kuo M 2000 Inactivation of the inhA -encoded fatty acid synthase II (FASII) enoyl-acyl carrier protein reductase induces accumulation of the FASI end products and cell lysis of Mycobacterium smegmatis
J Bacteriol 182 4059 4067 10869086
Yuan Y Zhu Y Crane DD Barry CE III 1998 The effect of oxygenated mycolic acid composition on cell wall function and macrophage growth in Mycobacterium tuberculosis
Mol Microbiol 29 1449 1458 9781881
Watanabe M Aoyagi Y Mitome H Fujita T Naoki H 2002 Location of functional groups in mycobacterial meromycolate chains; the recognition of new structural principles in mycolic acids Microbiology 148 1881 1902 12055308
Sassetti CM Boyd DH Rubin EJ 2003 Genes required for mycobacterial growth defined by high density mutagenesis Mol Microbiol 48 77 84 12657046
Segrè D Vitkup D Church GM 2002 Analysis of optimality in natural and perturbed metabolic networks Proc Natl Acad Sci U S A 99 15112 15117 12415116
Monahan IM Betts J Banerjee DK Butcher PD 2001 Differential expression of mycobacterial proteins following phagocytosis by macrophages Microbiology 147 459 471 11158363
Mdluli K Slayden RA Zhu Y Ramaswamy S Pan X 1998 Inhibition of a Mycobacterium tuberculosis β-ketoacyl ACP synthase by isoniazid Science 280 1607 1610 9616124
Kanehisa M Goto S Kawashima S Okuno Y Hattori M 2003 The KEGG resource for deciphering the genome Nucleic Acids Res 32 D277 D280
Karp PD Paley S Romero P 2002 The pathway tools software Bioinformatics 18 S1 S8
Hucka M Finney A Bornstein B Keating S Shapiro B 2004 Evolving a lingua franca and associated software infrastructure for computational systems biology: The Systems Biology Markup Language (SBML) project Syst Biol 1 41 53
Edwards JS Covert M Palsson BO 2002 Metabolic modelling of microbes: The flux-balance approach Environ Microbiol 4 133 133 12000313
Kauffman KJ Prakash P Edwards JS 2003 Advances in flux balance analysis Curr Opin Biotech 14 491 496 14580578
Alvarez-Vasquez F Sims K Cowart L Okamoto Y Voit E 2005 Simulation and validation of modelled sphingolipid metabolism in Saccharomyces cerevisiae
Nature 433 425 430 15674294
Edwards JS Ibarra RU Palsson BO 2001
In silico predictions of Escherichia coli metabolic capabilities are consistent with experimental data Nat Biotechnol 19 125 130 11175725
Altschul SF Madden TL Schäffer AA Zhang J Zhang Z 1997 Gapped BLAST and PSI-BLAST: A new generation of protein database search programs Nucleic Acids Res 25 3389 3402 9254694
Kremer L Nampoothiri KM Lesjean S Dover LG Grahami S 2001 Biochemical characterization of acyl carrier protein (AcpM) and malonyl-CoA:AcpM transacylase (mtFabD), two major components of Mycobacterium tuberculosis fatty acid synthase II J Biol Chem 276 27967 27974 11373295
Portevin D De Sousa-D'Auria C Houssin C Grimaldi C Chami M 2004 A polyketide synthase catalyzes the last condensation step of mycolic acid biosynthesis in mycobacteria and related organisms Proc Natl Acad Sci U S A 101 314 319 14695899
|
16261191
|
PMC1246807
|
CC BY
|
2021-01-05 09:18:23
|
no
|
PLoS Comput Biol. 2005 Oct 14; 1(5):e46
|
utf-8
|
PLoS Comput Biol
| 2,005 |
10.1371/journal.pcbi.0010046
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0112-000943Research ArticleIntroductionReviews of Environmental Health, 2004 Burkhart James National Institute of Environmental Health Sciences, National Institutes of Health, Department of Health and Human Services, Research Triangle Park, North Carolina, USA6 2004 112 9 943 943 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
Our broadening realization of the interconnectivity of well-being among species and ecosystems must bring new perspective to environmental health if we are to deal successfully with the dynamics of global change and human activity. Individuals might have interest and expertise in particular areas, but understanding the actions we must pro-actively take to enhance human and environmental health requires us to integrate many fields into articulated direction. EHP is committed to facilitating interdisciplinary communication from the most basic biological mechanism to public policy issues in the interest of health.
Realistically an individual is challenged by the prospect of seeing the broad patterns that might come from integrating interdisciplinary information. Language, hypotheses, methods, and data interpretation vary greatly across fields. The rapid evolution of new technologies in all disciplines produces even more daunting challenges to environmental scientists, educators, and policymakers. Yet this integration is profoundly important to the individual disciplines because larger views can produce more relevant questions and direction within each field of study.
The scientific complexity of environmental health in this time can produce confusion, or as we would hope, excitement and enthusiasm about discovery and our capacity to affect change in a positive direction. There are many important decisions and debates that must be met in the near future by the environmental health community. These involve global and local environments, exposure of humans and wildlife to new and old pollutants, children’s health, environmental medicine, conservation biology, and emerging technologies. The objective of the EHP Annual Review issue is to provide the latest information on particular topics in a way that is useful both to a very broad readership and to the specialists. As the Annual Review issue continues to evolve, EHP would like to engage the environmental health communities to consider, suggest, and submit for consideration reviews that summarize new developments in environmentally relevant areas and provide balanced background and perspective.
The review by Theo Colborn (2004) in this issue explores the role of environmental contaminants in disrupting thyroid signaling and its impact on the developing brain. The health issues associated with thyroid-mediated developmental disruption transcend humans and wildlife. Developmental pathways are conserved across diverse phylogenies. It is possible that classes of environmental contaminants capable of altering development and behavior in species such as amphibians may also contribute to attention deficit hyperactivity disorder and behavioral problems for humans in developed countries.
Kamel and Hoppin (2004) continue to explore the role of pesticide exposure and neurotoxicity. Read about the unresolved issues surrounding acute versus chronic exposure and the potential for changes in neurobehavioral performance reflecting cognitive and psychomotor dysfunction. The issues are pivotal if we are to plot a course toward better environmental health quality.
The increasing human population produces enormous waste disposal problems and responsibilities. All those interested should find the review by Rideout et al. (2004) of value, as it explores the possibility of accumulating human exposure to polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans through sewage sludge recycling as a nutrient source in agriculture.
Learn from the review by Armstrong et al. (2004) the issues associated with occupational exposure to polycylcic aromatic hydrocarbons and lung cancer, together with how the results might influence risk assessment.
A large percentage of the world human population lives close to coastlines, and human activity has placed many of those ecosystems at serious risk. The ecologic stress has increased toxic substances and pathogens. Niemi et al. (2004) provide the recent developments and suggest we need to have new coastal ecologic indicators if we are to successfully protect environmental health.
We all are aware of the potentially devastating effects of lead exposure on children and the recent successes in reducing lead in the environment. Now learn from Koller et al. (2004) about the possibilities for intellectual impairment at low levels and the relationships of other environmental factors in the health outcome.
Mini-monographs containing up to six manuscripts dealing with specific topics within a broadly important topic have increased in popularity over the last year in EHP. We seek to weave together different aspects of a larger topic so that the readers can come away with perspectives not easily attainable with a single review. Here we have included the mini-monograph “Health and Environment Information Systems for Exposure and Disease Mapping, and Risk Assessment” (EHP 2004) because it will give readers the opportunity to understand contributions of new geographic information systems (GIS) technology to environmental health and exposure tracking. The importance of this technology lies in the fact that exposures to large numbers of potentially toxic agents are uneven geographically and temporally. The GIS might be used to provide more rapid exposure and risk assessments in order to better protect humans and the environment.
The reviews and monograph articles included here should be of interest to the broad environmental health community. Your comments, insights, and suggestions for future directions would be appreciated.
==== Refs
References
Armstrong B Hutchinson E Unwin J Fletcher T 2004 Lung cancer risk after exposure to polycyclic aromatic hydrocarbons: a review and meta-analysis Environ Health Perspect 112 970 978 15198916
Colborn T 2004 Neurodevelopment and endocrine disruption Environ Health Perspect 112 944 949 15198913
EHP 2004 Health and Environment Information Systems for Exposure and Disease Mapping, and Risk Assessment Environ Health Perspect 112 995 1044 15198919
Kamel R Hoppin JA 2004 Neurologic dysfunction and disease Environ Health Perspect 112 940 958
Koller K Brown T Spurgeon A Levy L 2004 Recent developments in low-level lead exposure and intellectual impairment in children Environ Health Perspect 112 987 994 15198918
Niemi G Wardrop D Brooks R Anderson S Brady V Paerl H 2004 Rationale for a new generation of indicators for coastal waters Environ Health Perspect 112 979 986 15198917
Rideout K Teschke K 2004 Potential for increased human foodborne exposure to PCDD/F when recycling sewage sludge on agricultural land Environ Health Perspect 112 959 969 15198915
|
0
|
PMC1247185
|
CC0
|
2021-01-04 23:39:44
|
no
|
Environ Health Perspect. 2004 Jun; 112(9):943
|
utf-8
|
Environ Health Perspect
| 2,004 |
nan
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0112-00094410.1289/ehp.660115198913Research ArticleReviewsNeurodevelopment and Endocrine Disruption Colborn Theo Department of Zoology, University of Florida, Gainesville, Florida, USA, and The Endocrine Disruption Exchange, Paonia, Colorado, USAAddress correspondence to T. Colborn, P.O. Box 1253, Paonia, CO 81428 USA. Telephone/Fax: (970) 527-6548. E-mail:
[email protected] thank the Joyce Foundation, the Women’s Donor Network, and the Linda Zidell Foundation for supporting this article. In addition, I thank K. Howdeshell and M. Smolen and two unidentified individuals for reviewing this article.
The author declares no competing financial interests.
6 2004 17 11 2003 112 9 944 949 23 7 2003 17 11 2003 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. In this article I explore the possibility that contaminants contribute to the increasing prevalence of attention deficit hyperactivity disorder, autism, and associated neurodevelopmental and behavioral problems in developed countries. I discuss the exquisite sensitivity of the embryo and fetus to thyroid disturbance and provide evidence of human in utero exposure to contaminants that can interfere with the thyroid. Because it may never be possible to link prenatal exposure to a specific chemical with neurodevelopmental damage in humans, I also present alternate models where associations have been made between exposure to specific chemicals or chemical classes and developmental difficulties in laboratory animals, wildlife, and humans.
ADHDautismbehaviorendocrine disruptorenvironmental contaminantsneurologic effectsprenatal exposurethyroid
==== Body
Approximately 12 years ago the scientific community acknowledged that certain synthetic chemicals are capable of crossing the placental and brain barriers and interfering with development and function (Colborn and Clement 1992). The chemicals mimic or interfere with endogenous hormones and other signaling chemicals of the endocrine system. These chemicals, distinguished as endocrine disruptors, bridge many chemical classes and are an integral part of the world economy and commerce. To date no validated or standardized screens or assays have been developed to test chemicals for their possible endocrine-disrupting effects. Consequently, none of the thousands of chemicals used today have been tested systematically for these effects for regulatory purposes. Despite this, the list is growing of known endocrine disruptors having a wide range of mechanisms of action that can interfere with brain development (Brucker-Davis 1998; Howdeshell 2002).
The production and use of industrial and agricultural chemicals have increased at an almost exponential rate for the past 50 years, with roughly 10 new chemicals currently being introduced each day. The U.S. Environmental Protection Agency (U.S. EPA) estimates that 87,000 chemicals are in use today. In the United States the plastics industry has grown at the rate of 6–12% per year since the mid-1940s, with annual production in the United States reaching 85 billion pounds (> 338 pounds per person per year) in 1996. In developing countries, plastics production is expanding at the rate of 40% per year (Society of the Plastics Industry 1997). Plastics are used in toys, cosmetics, perfumes, cleaning compounds, clothing, telecommunication equipment, computers, almost all household products, high-impact sporting equipment, and construction material from buildings to automobiles, airplanes, and aerospace vehicles.
Currently approximately 875 active ingredients registered as pesticides by the U.S. EPA have been formulated into 21,000 pesticide products, with many more new products entering the market each month (Short and Colborn 1999). In 1995 the United States produced 1.3 billion pounds of pesticide active ingredients, of which herbicides (weed killers) are the most widely used. It is estimated that herbicides cover > 14% of the land surface of the United States. This does not include nonfarm use for lawns, gardens, golf courses, parks, roadsides, railways, airports, forests, federal applications on government lands, and vast rights-of-way by states, counties, and municipalities. More than 60% of herbicides are documented endocrine disruptors (Short and Colborn 1999). Among the most widely used herbicides that interfere with the thyroid system are 2,4-dichlorophenoxyacetic acid (2,4-D), acetochlor, aminotriazole, amitrole, bromacil, bromoxynil, pendamethalin, and the thioureas (Brucker-Davis 1998; Howdeshell 2002; Short and Colborn 1999).
Historical Perspective of Exposure and Human Disorders
When data on the growth in synthetic chemical production are compared with the data on increasing prevalence of neurodevelopmental and other developmental disorders in humans, the data begin to merge around 1970. At approximately the same time, the first generation of humans exposed in the womb to synthetic chemicals on a large scale began to have children of their own (Table 1). For example, a plastic monomer, bisphenol A (BPA), was introduced in the early 1920s. Polychlorinated biphenyls (PCBs) were introduced in 1929. DDT became available for retail sale in 1938, and the large-scale, widespread commercial use of a vast number of synthetic chemicals commenced near the end of World War II (WWII) in the 1940s. Companies previously producing chemicals for warfare converted to making pesticides and plastics as the petroleum industry began to find more uses for its by-products from gasoline production. Although indivduals were being exposed to these chemicals since the early 1920s, it was not until the end of WWII that exposure increased to such an extent that vast numbers of adults exposed daily were accumulating significant amounts of these chemicals in their bodies. In terms of generation time, these individuals in the 1950s produced the first generation of offspring exposed to numerous synthetic chemicals in the womb and at increased levels. By 1970 these post-WWII babies were having children of their own. It was during the 1970s that what appeared to be increases in unusual, previously rare neurodevelopmental disorders began to catch the attention of health professionals.
Terms such as learning disabilities, autism, attention deficit hyperactivity disorder (ADHD), childhood cancers, juvenile diabetes, and juvenile delinquency became household words by the mid-1990s. Parental support groups emerged across the nation for each anomaly, and in response, health authorities began to acknowledge these increases. In 1995 the U.S. EPA established the Children’s Environmental Health Program to develop preventive measures to protect children from exposure to environmental contaminants, and in 2000, a presidential initiative led to the establishment of children’s health centers nationwide to develop treatments and cures for these problems.
Gershon and Rieder (1992) provided one of the earliest alerts that neurodevelopmental problems were increasing in the United States. They reported that suicides were 10-fold higher in teenagers 15–19 years of age born in the 1950s than those born in the 1930s (Gershon and Rieder 1992). In addition the rates of depression and mania were continuing to rise with each new birth cohort they examined. They called it a “mystery.” They wrote that mood disorders could arise from the interactions between genes and “some aspects of the environment.” They implied that some individuals were more susceptible to environmental stresses than others because of their genetic makeup. Although it was suggested that the increases reflected better diagnoses, there was still much discussion in the media and among parents. Then, in 1987, Berkow and Fletcher (1987) estimated that as many as 10% of the children < 13 years of age in the United States suffered ADHD. A study by Weiss et al. (1993) on children born between 1979 and 1982 found thyroid abnormalities 5 times more frequent in children with ADHD. A study by Rowland et al. (2002) on children born in the late 1980s and early 1990s found that 15% of the boys and 5% of the girls from grades 1–5 in a North Carolina countywide school district had been diagnosed with ADHD (p < 0.001). The study also found that 11% of the boys and 3% of the girls were on medication. The authors suggested this is an underreported anomaly because sampling depended solely on parental responses.
In the 1980s, occasional reports concerning the increased prevalence of autism began to appear in the peer-reviewed literature from the United States and other countries. More accurate clinical diagnoses and reporting of autism occurred after the American Psychiatric Association defined this syndrome in 1994 (American Psychiatric Association 1994). Although estimates on the prevalence of the disorder vary widely depending on the scope of the definition of the term, the most recent studies consistently reveal higher prevalence or incidence rates (depending how the study was designed) of both the narrow and broad definitions. Autism disorder includes a limited number of classic symptoms, using a narrow set of criteria, whereas autistic spectrum disorder includes a broader scope of symptoms that has been made possible in recent years with better diagnostic tests and technologies. A 2001 study in the United Kingdom reported 16.8 autistic children per 10,000 using the narrow definition and 62.6 per 10,000 using the broader definition (Chakrabarti and Fombonne 2001). A study in 1998 of Brick Township, New Jersey, found 40 cases per 10,000 in the narrow definition group and 67 per 10,000 in the broader definition group (Bertrand et al. 2001). This study was initiated because of community concern about exposure to industrial emissions. A 2003 study by the U.S. Centers for Disease Control and Prevention (CDC) (Yeargin-Allsopp et al. 2003) found 19–47 cases per 10,000 from a random sampling of children in 1996 between 3 and 10 years of age from five counties in metropolitan Atlanta, Georgia; the male–female ratio was 4:1. As with ADHD, boys are significantly more likely to develop autism than girls. “Autism is the fastest-growing developmental disability, increasing at a rate of 10 to 17 percent annually,” according to the Autism Society of America (Grossman 2002).
By 1996, on the basis of extensive magnetic resonance imaging (MRI) and histological examination in both humans and laboratory animals, it became apparent that the original autism lesion occurs just before or shortly after neural tube closure. This is around week 6 or 7 in the human fetus (Bayer et al. 1993; Howdeshell 2002; Rodier et al. 1996). During this stage of development the large cerebellar neurons begin to develop. The increased brain weight of autistic children, the packing of cells in the limbic system, the deficiency in Purkinje cells and granule cells of the cerebellum, and a number of other differences in the autistic brain suggest that numerous tissues and stages of brain development can be affected as the autistic syndrome evolves. In addition one retrospective study showed that when thalidomide exposure occurred between gestation days 20 and 24 (week 3), approximately 30% of the phocomelia cases were also autistic (Miller and Strömland 1993).
Sensitivity of Neurodevelopment to Thyroid Hormones
Although it has been known for a century that hypothyroidism leads to retardation and other serious developmental effects, the role of thyroid hormones in brain development is still not completely understood (Rice and Barone 2000). It is also accepted that thyroid hormones transferred from the mother to the embryo and fetus are critical for normal brain development (Lazarus 1999), even though the thyroid gland of a fetus starts producing thyroid hormones at about 10 weeks (Shepard 1967) (Figure 1).
We now recognize that only a slight difference in the concentration of thyroid hormones during pregnancy can lead to significant changes in intelligence in children. In pregnant women, normal thyroid hormones circulate bound to protein at parts per billion (ppb) and as free hormone at parts per trillion (ppt). In a long-term study Haddow et al. (1999) collected and stored blood from women in their second trimester of pregnancy. Years later, their children were tested between 7 and 9 years of age for intelligence, attention, language, reading ability, school performance, and visual–motor performance. The data of Haddow et al. revealed a possible association between a relatively slight reduction in the amount of circulating free thyroid hormone levels in the mothers (2.6 ppt) and the intellectual development of their children (Haddow et al. 1999). Children of mothers with a geometric mean of 9.1 ppt free thyroxine (fT4) during gestation scored 4 points higher in IQ than children of mothers with a geometric mean of 7.5 ppt (p = 0.002). In this study the thyroxine (T4) level in the lower IQ cohort was at the low end of what is considered the normal T4 range (Haddow et al. 1999). Haddow et al. (1999) were not attempting to link exposure to synthetic chemicals with loss of intelligence in the children. However, their results demonstrate that synthetic chemicals, which can interfere with the thyroid system, would not have to be present in very high concentrations to affect the intellectual and behavioral development of embryos and fetuses. Their study unexpectedly demonstrates the fragile relationship between a mother and her developing offspring.
Mechanisms of Action of Thyroid-Disrupting Chemicals
For investigators to understand the role of endocrine disruption in brain development, it was necessary to first understand how thyroid hormones regulate brain development. The complexity of the development of both the neurologic and thyroid systems offers numerous opportunities for chemicals to interfere as the systems develop, mature, and function. Howdeshell (2002) provides a current road map about the simultaneous development of the neurologic and thyroid systems (Figure 1). She also provides a list of those synthetic environmental chemicals (aside from pharmaceuticals and designer chemicals) known to interfere with these systems for which mechanism of action has been determined [see Howdeshell (2002) for further list and discussion]. Below is a list of demonstrated thyroid–pituitary disruptions that result from environmental exposure.
Inhibition of active transport of inorganic iodide into the follicular cell
Interference with the sodium/iodide transporter system
Inhibition of thyroid peroxidases to convert inorganic iodide into organic iodide to couple iodinated tyrosyl moieties into thyroid hormone
Damage to follicular cells
Inhibition or enhancement of thyroid hormone release into the blood
Inhibition or activation of the conversion of T4 to T3 by 5′-monodeiodinase at various sites in the body, for example, the fetal brain
Enhancement or interference of the metabolism and excretion of thyroid hormone by liver uridine diphosphate
Interference with transport of thyroid hormones
Vitamin A (retinol) disturbances
Blocking of or interfering with thyroid receptors
Briefly, there are chemicals that interfere with iodine uptake (the herbicides 2,4-D and man-cozeb, several PCB congeners, and thiocyanates) and peroxidation at the molecular level (the herbicides aminotriazole and thioureas, the insecticides endosulfan and malathion, and PCBs). They also interfere with the protein transporter that provides a pathway for iodine to enter the cell (military and aerospace chemicals, perchlorates). Certain antagonists (PCBs, the herbicides aminotriazole and dimethoate, and the insecticide fenvalerate) prevent the release of thyroid hormone from the cell and inhibit conversion of T4 to triiodothyronine (T3). Various chemicals enhance excessive excretion of thyroid hormones, some through activation of the cytochrome P450 system (dioxin, hexa-chlorobenzene, and fenvalerate). Some PCBs, phthalates, and other widely used chemicals compete for sites on the thyroid transport proteins that deliver thyroid hormones throughout the body. New research focuses on the role of chemicals as they interfere with vitamin A (retinols), the retinol receptors, and the essential dimerization of thyroid hormone with retinols, a process essential for thyroid hormone expression. There is still no evidence that environmental chemicals directly block the thyroid receptor.
For years it was thought that in humans transthyretin (TTR) played a special role among the thyroid transport proteins, albumin and thyroglobulin, to transport fT4 into the fetal brain where it is converted by the enzyme deiodinase to free T3 (fT3) (Porterfield 1994). However, recent research using TTR knockout mice reveals that TTR is not necessary for transport of fT4 to the fetal murine brain (Palha et al. 2002). Nonetheless, Brouwer et al. (1998) pointed out that during normal enzyme detoxification of PCBs in the maternal liver, certain PCB congeners are hydroxylated. This metabolic step enhances the binding affinity of the hydroxylated PCBs to TTR (Brouwer et al. 1998). Traveling on TTR in the blood, the hydroxylated PCBs cross the placenta, enter the fetus, and ultimately the fetal brain. Through their high-affinity binding the hydroxylated congeners displace essential fT4 that must get to the fetal brain to be converted to fT3. Schuur et al. (1998) demonstrated that the hydroxylated PCBs also interfere with the normal excretion of thyroid hormones by inhibiting their sulfation. PCB hydroxylates also have estrogenic and antithyroid activity (Iwasaki et al. 2002). For example, developing brain cells exposed to the PCB hydroxylate 4(OH)-2′,3,3′,4′5′-pentachlorobiphenyl (10−10 M) displayed the strongest suppression of thyroid hormone–activated transcription compared with any other developing cells tested in an in vitro assay (Iwasaki et al. 2002). This is another example of the ultrasensitivity of brain development to PCBs.
At the organism level, U.S. EPA scientists were able to demonstrate the critical role of thyroid hormones in the development of the ear, normal hearing, and motor control. As shown in Figure 1, the cochlea in the human ear begins to form around 6 weeks in utero. The cochlea is connected directly with the brainstem, allowing for immediate transmission to and interpretation of sound in the brain. A properly constructed cochlea is critical for hearing. Pregnant rats fed the antithyroid pharmaceutical propyl thiouracil had both difficulty hearing low and intermediate frequency clicking sounds and loss of motor coordination (Goldey et al. 1995b). Another set of pregnant rats was fed Aroclor 1254, a commercial mixture of PCBs, which also induced hypothyroidism and the accompanying hearing and motor problems (Goldey et al. 1995a). To confirm that the PCBs interfered with the thyroid system, pregnant rats were fed Aroclor 1254 supplemented with T4, and motor problems in the pups were attenuated (Goldey and Crofton 1998). In line with the U.S. EPA findings, Lavado-Autric et al. (2003) report abnormal cell migration and cytoarchitecture in the hippocampus and primary somatosensory cortex in rat pups whose dams were fed a low-iodine diet. They point out that the pups exhibited detectable functional audio deficits related to this abnormal development (Lavado-Autric et al. 2003).
Overcoming the Difficulty of Making Causal Links
It is almost impossible to make causal links between prenatal contaminant exposure and developmental damage in humans. Because of this, scientists have used laboratory and wild-animal models to better understand the effects of synthetic chemicals on development. Making a strong association between a particular chemical or class of chemicals and an adverse condition in the field is sometimes difficult, yet in some instances an association can be made by supplementing field research with well-designed laboratory research. For example, reports of serious developmental and reproductive problems among birds in the North American Great Lakes and other regions in developed countries date back to the 1960s and early 1970s (Gilbertson 1975; Leatherland 1992, 1999; Moccia et al. 1986). Thyroid gland and hormone abnormalities in particular were repeatedly reported in Great Lakes herring gulls used in a Canadian monitoring program to track PCBs and other organochlorine chemicals in the lakes. In addition, Leatherland (1992) reported that every top predator fish they examined in the Great Lakes had enlarged thyroid glands. In the late 1990s the problem persisted as the thyroid glands in Lake Erie fish began to rupture because the glands were becoming so large (Leatherland 1999), although the concentrations of a number of the organochlorine chemicals, including the PCBs, had declined considerably in the late 1970s in the Great Lakes. To date, no clear link has been established with a specific chemical in the Great Lakes that causes these thyroid problems. Lack of iodine has been ruled out as a possible link.
Conversely, field observations of damaged brains and spinal cords in bald eagles and great blue herons (Henshel et al. 1993, 1996) led to laboratory experimentation that provided a strong association between the anomalies and 2,3,7,8-tetrachlorodibenzo-p-dioxin, which was discovered in elevated levels in the birds. The laboratory results not only provided a causal link with a contaminant but also demonstrated the extreme sensitivity of the developing brain to chemical interference. Myelination is critical for proper nervous system development, and in humans it commences at approximately 12 weeks in utero (Figure 1). Myelination was reduced in the spinal cords of chicks exposed to dioxin (10, 100, and 1,000 ppt) in a dose–response manner when injected during their egg stage. Effects were visible and similar to the damage seen in the wild birds at an exposure of 10 ppt, which is within the range of human exposure (Henshel 1998; U.S. EPA 2000).
The Human Connection
Concern in the 1970s over the widespread health problems among Great Lakes wildlife led to a human epidemiologic study that examined the health effects in infants whose mothers ate two to three meals a month of Lake Michigan fish for at least 6 or more years before their pregnancies (Jacobson and Jacobson 1996). Only healthy mothers and infants were selected for this study. Within 24 hr of birth, significant delays in neuromuscular and neurologic development were detected in the children whose mothers ate the most fish contaminated with PCBs. At 4 years of age some children showed an association between short-term memory problems and the amount of PCBs in the mothers’ blood at delivery. The same children at 11 years of age displayed significantly reduced academic skills accompanied by a mean 6.2-point IQ reduction. This again was associated with their prenatal exposure to PCBs, not the concentration of PCBs in their blood at the time of testing. Although there is no way to prove that PCBs interfered with the development of the cochlea in these children, the affected children had difficulty with audiovisual discrimination and information processing. Some children were as much as 2 years behind their peers in school, were hyperactive, and had attention problems (Jacobson and Jacobson 1996).
Another healthy mother–infant study commenced 12 years later in Oswego, New York, on Lake Ontario to replicate and expand the design of the Lake Michigan study (Darvill et al. 2000). As in the Lake Michigan study, the high fish eaters consumed about the same amount of Lake Ontario fish before their pregnancies (Stewart et al. 1999). Again, prenatal exposure to PCBs was associated with neurodevelopmental changes in their children at approximately 4 years of age. Using another battery of tests, Stewart et al. (2000b) found that the temperaments of the affected children were altered compared with those of lesser-exposed children. They smiled less, were more fearful, and had difficulty adapting to changes in their environment.
Stewart et al. (2000b) also compared the content of the highly chlorinated PCB homologs or isomers with 7, 8, or 9 chlorines (septa-, octa-, and nonyl-chlorinated biphenyls) in the mothers’ blood at the end of the first trimester with their fish diet. The mothers who never ate fish from the lakes had the lowest concentrations of the highly chlorinated PCBs in their blood. In 1984 when the first fish advisories were issued warning pregnant women not to eat the fish from Lake Ontario, the mothers who stopped eating fish had less of the isomers than the mothers who only stopped eating Lake Ontario fish when they found they were pregnant. The mothers who did not stop eating fish throughout their pregnancies had the highest PCB isomer concentrations (Stewart et al. 1999). Stewart et al. (2000a) found a dose–response relationship between the prenatal exposure of the children to the highly chlorinated PCBs and increases in their reflexive and autonomic deviations from the norm and their reduced ability to habituate under various conditions. MRI examination of the most highly exposed children in this study revealed an inverse dose–response association between their PCB cord blood and the volume/size of the splenium of the corpus callosum at 7.8 years of age, and their response inhibition at 4.5 years of age, a behavioral characteristic seen in ADHD children where they do not adapt well to their environment and have trouble settling down (Stewart et al. 2003). The splenium is the bridge between the right and left sides of the corpus callosum.
Another set of studies with healthy mothers and infants performed in the Netherlands examined a cross-section of the population, not just fish eaters (Koopman-Esseboom et al. 1996). This team found neuromuscular delays in the children at 3 months of age in association with in utero exposure to PCBs and dioxins measured as dioxin toxicity equivalents (TEQs) (Brouwer et al. 1995). Additionally, an inverse dose–response association was observed between increased TEQs with thyroid levels in the children and a positive association with unusual changes in their immune system (Weisglas-Kuperus et al. 1995). Further comparisons are difficult because the same battery of tests as those used in the United States was not employed in this series of studies.
The Netherlands research team divided the mothers into two groups, low TEQ (< 30.74 pg/g fat) and high TEQ (> 30.75 pg/g fat), on the basis of the equivalents in the plasma of the mother during the last month of pregnancy (Koopman-Esseboom et al. 1994). The differences between the two cohorts in total T4 (177.5–159.9 nmol/L) and thyroid-stimulating hormone (TSH) (1.9–2.6 μIU/mL) were significant at 2 weeks of age. However, the TSH levels of the mothers were within the normal range (3.0 IU/mL is the cutoff).
All the effects reported in the children in the studies described above were linked to the children’s prenatal experience. In each study mentioned, even though the decrements among the children were statistically significant at the population level, the parents or doctors of the infants would not have known they were compromised. It took skilled psychologists and technicians to quantify the changes in the children.
In the Lake Michigan study trained psychologists were able to measure developmental delays in infants shortly after birth if the blood fat of the mother held 1.00 parts per million (ppm) PCBs (Jacobson and Jacobson 1996). At 1.25 ppm PCBs, the change was statistically significant (p < 0.001) because there was so little variance. The intelligence and behavioral impairments reported in this study are populationwide. They are not rare events such as cancer. In this healthy mother–infant study, at 11 years of age, 11% of the children were affected (Jacobson and Jacobson 1996). At 4 years of age, 17 children were removed from the study because they were too hyperactive and would not take the tests (Jacobson et al. 1990). If the outliers had remained in the study, 20% of the children would have been affected. It was later determined that the children who were removed from the study were the children of the mothers with the highest PCB concentrations in the study. Another child was removed from the study at the end because he or she had an IQ below 70 (Jacobson and Jacobson 1996). These researchers noted that consuming fish is not the only source of PCBs, but these compounds are found in many other foods such as meats, fatty foods, fast foods, cheeses, ice cream, and even in the most rigid vegan diet (Schecter et al. 2001).
A Japanese group measured TEQs in breast milk at approximately 3 months postpartum and compared those with T3 and T4 in blood of the infants at 1 year of age (Nagayama et al. 1998). Only healthy mothers and full-term infants from southern Japan were selected for this study. There was a significant inverse relationship between total TEQs and T3 and T4 of the infants (n = 40). In this study, TSH had no association with the contaminants, suggesting that TSH may not be as sensitive an end point as previously considered without accompanying T3 and T4 measurements. No behavioral results accompanied these data. This study confirms, however, the transfer of contaminants, measured as TEQs, from the mother to the child and a change in the circulating thyroid hormones of the child distinguishable at the population level.
Opening the Black Box of Exposure
The most difficult task for epidemiologists is to demonstrate exposure during development in human populations. Unlike the scenario where dioxin exposure was correlated with brain damage in great blue herons, in most epidemiologic studies attempting to determine etiology, the timing, scope of chemicals involved or range of exposure, and the actual body burdens of the chemicals are not knowable. Fortunately, in the past 5 years, technology for measuring human exposure to synthetic chemicals has advanced considerably. For example, the CDC can now monitor human blood and urine for > 116 chlorinated and nonchlorinated chemicals and their metabolites. These include contemporary-use pesticides and industrial chemicals used in cosmetics, perfumes, detergents, toys, plastics, and fire retardants—many of which are high-production-volume chemicals widely used in commerce (Burse et al. 1996; CDC 2003).
CDC chemists have begun to open the black box of exposure not only for a better picture of human organochlorine chemical exposure but for a number of other widely used chemicals that have not been studied as intensely. For example, they discovered that some metabolites of a class of chemicals called phthalates were 9 times higher in the urine of women between 20 and 40 years of age—women of childbearing age—than in any other segment of the population (Blount et al. 2000). Phthalates make plastics flexible and soft; they are used to improve delivery systems in perfumes, nail polish, shampoos, cosmetics, and dermal and intravenous applications of medications. They have been widely used as inert ingredients in pesticide formulations. Three of the phthalates, diethylhexyl phthalate, di(n-octyl) phthalate, and di(n-hexyl) phthalate, are antiandrogens in laboratory animals, producing hypospadias, cryptorchidism, and other male developmental disorders. They also interfere with the thyroid system (Hinton et al. 1977; Price et al. 1988).
Discussion
Increases in the prevalence of neurodevelopmental disorders over the past 30 years make it imperative to reverse this trend. Because it appears that this trend could be partly the result of exposure to environmental contaminants, it is also imperative to prevent further exposure to synthetic chemicals that are suspect. Fortunately, technological improvements in the past 10 years have broadened the scope and sensitivity of detection for not only synthetic chemicals in human tissues but also for natural endogenous hormones. The evidence that certain hormones operate at parts per trillion and parts per billion and equivalent exposure to endocrine-active chemicals is equivalent or higher reveals the extreme vulnerability of development to chemical perturbation. This detection technology, combined with large-scale epidemiologic studies, is beginning to reveal a panorama of subtle biological differences within populations that would never be recognized at the individual level. The work of Haddow et al. (1999) is an example of what we can learn by stepping back and observing what is happening in the population. Their work suggests that a) it is time to reassess what is considered euthyroid; b) quantification of T4,fT4, T3, and fT3 is needed to determine the thyroid hormone status of a pregnant woman; and c) routine maternal thyroid hormone monitoring throughout gestation should become standard practice.
It is important to note that increasing numbers of children are exhibiting attention deficit disorder and ADHD-like symptoms in the classroom, in turn placing more and more responsibility on teachers, families, social services, and taxpayers. A new approach for exploring the etiology of these disorders is needed that could include the use of more comprehensive case histories for the entire family. For example, the road map of brain/thyroid development presented by Howdeshell (2002) provides a starting place for specifically seeking the etiology of neurodevelopmental disorders. A vertical dissection of Figure 1 at various times throughout gestation cuts through many brain development events occurring simultaneously that are thyroid dependent. If a chemical were to interfere with the commencement of development of the cochlea at 6 weeks through its antithyroid effect, it might also affect other thyroid-sensitive tissues emerging at the same time. This might partly explain the list of irregularities or sequelae of anomalies that fit within the definition of the ADHD and autism syndromes. Overlay maps must be constructed of other developing systems to determine where the presence of a foreign chemical in the womb could interfere simultaneously with the developing brain and thyroid system. Comorbidities should begin to appear on these overlays, for example, the relationship between autism and hypospadias at 7–8 weeks when sexual differentiation commences at the same time as the development of the urogenital system and the hippocampus. Could one or more of the chemicals that are known antiandrogens and antithyroidals be involved in the latter? These junctures could be the tip of the iceberg in terms of determining the etiology of a disorder. To take advantage of this approach, more extensive case histories of autistic children are needed to determine other anomalies they are experiencing. In addition, exposure histories are needed of the parents of children with obvious developmental disorders, with emphasis on their lifestyle and occupational exposure before the birth of their child, not just focusing on the early postnatal and immediate exposure of the child.
Because it is now the rule to demand only statistically significant results in studies, a great deal of insight can be lost, as was the case with the Lake Michigan studies. It is time to reassess how statistics are applied in health-related studies and the conclusions used in risk analyses. For example, because an average IQ deficit of 6 points is within two standard deviations of normal, it could be dismissed as an adverse effect. If this statistical criterion were applied, it would exclude the most sensitive but still responding individuals. However, when viewed from the population level, this can have a tremendous impact on the economy and integrity of a society. Unfortunately, the true robustness of the Lake Michigan PCB studies mentioned previously is lost in the final statistical analysis for several reasons. First, because only the healthiest pairs of infants and mothers were used in any of the these studies, the most vulnerable segment of the population was eliminated before the study started. Second, by eliminating the outliers (n = 18), another vulnerable segment of the population was removed. In so doing, the overall population effect was artificially reduced, thus minimizing the importance and significance of the study results. It is time for a thorough reassessment of statistical applications and study designs in long-term, large-scale epidemiologic studies to fully assess damage in the entire population, not only in representative subpopulations. This would assist those concerned about etiology and prevention.
During the organizational stages of gestation, responses to endocrine disruption are unlike the typical responses in adulthood. Consequently, testing with mature animals misses the organizational damage from prenatal exposure. In addition, most traditional toxicological tests use doses 1,000–1,000,000 times that of the equivalent physiological range at which the endocrine systems operate and well above real-world exposure concentrations to synthetic chemicals. The high doses used in toxicological testing far exceed the normal threshold or peak concentrations at which homeostatic negative-feedback control from the brain shuts down cellular responses. Consequently, other nonendocrine toxic effects might be expressed in exposed adult animals but not the same effects if exposure had taken place during their construction and programming. Thus, in endocrine disruption, extrapolating down from several high doses to determine the lowest safe dose or no-effect dose of a chemical will not protect the fetus. Fortunately, many innovative and entirely new protocols for detecting endocrine disruption are in the early stages of being validated and standardized in dozens of countries around the world, but unfortunately, it will take years before many will be ready for use.
Figure 1 Role of thyroid hormones in fetal neurologic development in relation to timing of several landmark stages of development. Figure adapted from Howdeshell (2002).
Table 1 Chronology of human exposure.
Years Exposure scenario
1920s–1930s BPA, PCBs, and DDT commercially introduced. Chlorine industry expanding. Discrete postnatal and prenatal exposure.
1940s–WWII First wide-scale production and exposure to the above and other chemicals including plastics and chlorinated compounds as technology advanced.
1940s–1950s First generation widely exposed postnatally and some who may have been exposed prenatally.
1950s–1970s First generation born that was widely exposed prenatally.
1970s–1990s First generation that was widely exposed prenatally reached reproductive age.
1980s–present Second generation born that was exposed in the womb and beginning to produce the third generation. Production volume and exposure still increasing.
==== Refs
References
American Psychiatric Association 1994. Diagnostic and Statistical Manual of Mental Disorders. 4th ed. Washington, DC:American Psychiatric Association.
Bayer SA Altman J Russo RJ Zhang X 1993 Timetables of neurogenesis in the human brain based on experimentally determined patterns in the rat Neurotoxicology 14 83 144 8361683
Berkow R Fletcher A 1987. The Merck Manual of Diagnosis and Therapy. 15th ed. Rahway, NJ:Merck & Co., Inc.
Bertrand J Mars A Boyle C Bove F Yeargin-Allsopp M Decoufle P 2001 Prevalence of autism in a United States population: the Brick Township, New Jersey, investigation Pediatrics 108 1155 1161 11694696
Blount BC Silva MJ Caudill SP Needham LL Pirkle JL Sampson EJ 2000 Levels of seven urinary phthalate metabolites in a human reference population Environ Health Perspect 108 972 982
Brouwer A Ahlborg UG van den Berg M Birnbaum LS Boersma ER Bosveld B 1995 Functional aspects of developmental toxicity of polyhalogenated aromatic hydrocarbons in experimental animals and human infants Eur J Pharmacol 293 1 40 7545581
Brouwer A Morse DC Lans MC Schuur AG Murk AJ Klasson-Wehler E 1998 Interactions of persistent environmental organohalogens with the thyroid hormone system: mechanisms and possible consequences for animal and human health Toxicol Ind Health 14 59 84 9460170
Brucker-Davis F 1998 Effects of environmental synthetic chemicals on thyroid function Thyroid 8 827 856 9777756
Burse VW Patterson DG Brock JW Needham LL 1996 Selected analytical methods used at the CDC for measuring environmental pollutants in serum Toxicol Ind Health 12 481 498 8843565
CDC 2003. Second National Report on Human Exposure to Environmental Chemicals. NCEH Publ. no 02-0716. Atlanta, GA:Centers for Disease Control and Prevention, National Center for Environmental Health.
Chakrabarti S Fombonne E 2001 Pervasive developmental disorders in preschool children JAMA 285 3093 3099 11427137
Colborn TE Clement C 1992. Chemically Induced Alterations in Sexual and Functional Development: the Wildlife/Human Connection. Princeton, NJ:Princeton Scientific.
Darvill T Lonky E Reihman J Stewart P Pagano J 2000 Prenatal exposure to PCBs and infant performance on the Fagan test of infant intelligence Neurotoxicology 21 1029 1038 11233749
Gershon ES Rieder RO 1992 Major disorders of mind and brain Sci Am 267 126 133 1502514
Gilbertson MA 1975 Great Lakes tragedy Nat Can 4 22 25
Goldey ES Crofton KM 1998 Thyroxine replacement attenuates hypothyroxinemia, hearing loss, and motor deficits following developmental exposure to Aroclor 1254 in rats Toxicol Sci 45 94 105 9848116
Goldey ES Kehn LS Lau C Rehnberg GL Crofton KM 1995a Developmental exposure to polychlorinated biphenyls (Aroclor 1254) reduced circulating thyroid hormone concentrations and causes hearing deficits in rats Toxicol Appl Pharmacol 135 77 88 7482542
Goldey ES Kehn LS Rehnberg GL Crofton KM 1995b Effects of developmental hypothyroidism on auditory and motor function in the rat Toxicol Appl Pharmacol 135 67 76 7482541
Grossman L 2002 President’s message Advocate 35 3 3
Haddow JE Palomaki GE Allan WC Williams JR Knight GJ Gagnon J 1999 Maternal thyroid deficiency during pregnancy and subsequent neuropsychological development of the child N Engl J Med 341 549 555 10451459
Henshel DS 1998 Developmental neurotoxic effects of dioxin and dioxin-like compounds on domestic and wild avian species Environ Toxicol Chem 17 88 98
Henshel DS Cheng KM Norstrom R Whitehead P Steeves JD 1993. Morphometric and histologic changes in brains of great blue heron hatchlings exposed to PCDDs: preliminary analyses. In: Environmental Toxicology and Risk Assessment (Landis WG, Hughes JS, Lewis MA, eds). Environmental Toxicology and Risk Assessment, ASTM STP 1179. Philadelphia:American Society for Testing and Materials, 262–277.
Henshel DS Martin JW Best D Cheng KM Elliot J Rosenstein D Sikarskie J 1996. Evaluating gross brain asymmetry: a potential biomarker for TCDD-related neurotoxicity. In: Environmental Toxicology and Risk Assessment: Biomarkers and Risk Assessment Vol 5 (DA Bengston, DS Henshel, eds). ATSM STP 1306. Philadelphia:American Society for Testing and Materials, 230–238.
Hinton RH Mitchell FE Mann A Chescoe D Price SC Nunn A 1977 Effects of phthalate esters on the liver and thyroid Environ Health Perspect 70 195 210 3830106
Howdeshell KL 2002 A model of the development of the brain as a construct of the thyroid system Environ Health Perspect 110 suppl 3 337 348 12060827
Iwasaki T Miyazaki W Takeshita A Kuroda Y Koibuchi N 2002 Polychlorinated biphenyls suppress thyroid hormone-induced transactivation Biochem Biophys Res Commun 299 384 388 12445811
Jacobson JL Jacobson SW 1996 Intellectual impairment in children exposed to polychlorinated biphenyls in utero N Engl J Med 335 783 789 8703183
Jacobson JL Jacobson SW Humphrey HEB 1990 Effects of exposure to PCBs and related compounds on growth and activity in children Neurotoxicol Teratol 12 319 326 2118230
Koopman-Esseboom C Morse DC Weisglas-Kuperus N Lutke-Schipholt IJ Van der Paauw CG Tuinstra LGMT 1994 Effects of dioxins and polychlorinated biphenyls on thyroid hormone status of pregnant women and their infants Pediatr Res 36 468 473 7816522
Koopman-Esseboom C Weisglas-Kuperus N de Ridder MAJ Van der Paauw C Tuinstra LGMT Sauer PJJ 1996 Effects of polychlorinated biphenyl/dioxin exposure and feeding type on infants’ mental and psychomotor development Pediatrics 97 700 706 8628610
Lavado-Autric R Auso E Garcia-Velasco JV Arufe MdC Escobar del Ray F Berbel P 2003 Early maternal hypothyroxemia alters histogenesis and cerebral cortex cytoarchitecture of the progeny J Clin Invest 111 1073 1082 12671057
Lazarus JH 1999 Thyroid hormone and intellectual development: a clinician’s view Thyroid 9 659 660 10447010
Leatherland JF 1992. Endocrine and reproductive function in Great Lakes salmon. In: Chemically Induced Alterations in Sexual And Functional Development: The Wildlife/Human Connection (Colborn T, Clement C, eds). Princeton, NJ:Princeton Scientific, 129–145.
Leatherland JF 1999. Contaminant-altered thyroid function in wildlife. In: Environmental Endocrine Disrupters: An Evolutionary Perspective (Guillette LJ Jr, Crain DA, eds). New York:Taylor and Francis, 155–181.
Miller MT Strömland K 1993 Thalidomide embryopathy: an insight into autism? Teratology 47 387 388
Moccia RD Fox GA Britton A 1986 A quantitative assessment of thyroid histopathology of herring gulls (Larus argentatus ) from the Great Lakes and a hypothesis on the causal role of environmental contaminants J Wildl Dis 22 60 70 3951063
Nagayama J Tsuji H Iida T Hirakawa H Matsueda T Okamura K 1998 Postnatal exposure to chlorinated dioxins and related chemicals on thyroid hormone status in Japanese breast-fed infants Chemosphere 37 1789 1793 9828307
Palha JA Nissanov J Fernandes R Sousa JC Bertrand L Dratman MB 2002 Thyroid hormone distribution in the mouse brain: the role of transthyretin Neuroscience 113 4 837 847 12182890
Porterfield SP 1994 Vulnerability of the developing brain to thyroid abnormalities: environmental insults to the thyroid system Environ Health Perspect 102 125 130 7925183
Price SC Chescoe D Grasso P Wright M Hinton RH 1988 Alterations in the thyroids of rats treated with di(2-ethyl-hexyl) phthalate or with hypolipidaemic agents Toxicol Lett 40 37 46 3341048
Rice D Barone S Jr 2000 Critical periods of vulnerability for the developing nervous system: evidence from humans and animal models Environ Health Perspect 108 suppl 3 511 533 10852851
Rodier PM Ingram JL Tisdale B Nelson S Romano J 1996 Embryological origin for autism: developmental anomalies of the cranial nerve motor nuclei J Comp Neurol 370 247 261 8808733
Rowland AS Umbach DM Stallone L Naftel AJ Bohlig EM Sandler DP 2002 Prevalence of medication treatment for attention deficit-hyperactivity disorder among elementary school children in Johnston County, North Carolina Am J Public Health 92 231 234 11818297
Schecter A Cramer P Boggess K Stanley J Papke O Olson J 2001 Intake of dioxins and related compounds from food in the U.S. population J Toxicol Environ Health 63 101 118
Schuur AG Brouwer A Bergman A Coughtrie MWH Visser TJ 1998 Inhibition of thyroid hormone sulfation by hydroxylated metabolites of polychlorinated biphenyls Chem Biol Interact 109 293 297 9566753
Shepard TH 1967 Onset of function in the human fetal thyroid: biochemical and radioautographic studies from organ culture J Clin Endocrinol 27 945 958
Short P Colborn T 1999 Pesticide use in the U.S. and policy implications: a focus on herbicides Toxicol Ind Health 15 240 275 10188206
Society of the Plastics Industry 1997. Facts and Figures of the U.S. Plastics Industry. 1996 ed. Washington, DC:Society of the Plastics Industry.
Stewart P Darvill T Lonky E Reihman J Pagano J Bush B 1999 Assessment of prenatal exposure to PCBs from maternal consumption of Great Lakes fish: an analysis of PCB pattern and concentration Environ Res 80 S87 S96 10092422
Stewart P Fitzgerald S Reihman J Gump B Lonky E Darvill T 2003 Prenatal PCB exposure, the corpus callosum, and response inhibition Environ Health Perspect 111 13 1670 1682 14527849
Stewart P Pagano J Sargent D Darvill T Lonky E Reihman J 2000a Effects of Great Lakes fish consumption on brain PCB pattern, concentration, and progressive-ratio performance Environ Res 82 18 32 10677143
Stewart P Reihman J Lonky E Darvill T Pagano J 2000b Prenatal PCB exposure and Neonatal Behavioral Assessment Scale (NBAS) performance Neurotoxicol Teratol 22 21 29 10642111
U.S. EPA 2000. Exposure and Human Health Reassessment of 2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) and Related Compounds. Draft. Washington, DC:U.S. Environmental Protection Agency.
Weisglas-Kuperus N Sas TCJ Koopman-Esseboom C Van der Zwan CW de Ridder MAJ Beishuizen A 1995 Immunologic effects of background prenatal and postnatal exposure to dioxins and polychlorinated biphenyls in Dutch infants Pediatr Res 38 404 410 7494667
Weiss RE Stein MA Trommer B Refetoff S 1993 Attention-deficit hyperactivity disorder and thyroid function J Pediatr 123 539 545 8410504
Yeargin-Allsopp M Rice C Karapurkar T Doernberg N Boyle C Murphy C 2003 Prevalence of autism in a US metropolitan area JAMA 289 49 55 12503976
|
15198913
|
PMC1247186
|
CC0
|
2021-01-04 23:39:51
|
no
|
Environ Health Perspect. 2004 Jun 17; 112(9):944-949
|
utf-8
|
Environ Health Perspect
| 2,003 |
10.1289/ehp.6601
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0112-00095010.1289/ehp.713515198914Research ArticleReviewsAssociation of Pesticide Exposure with Neurologic Dysfunction and Disease Kamel Freya Hoppin Jane A. National Institute of Environmental Health Sciences, National Institutes of Health, Department of Health and Human Services, Research Triangle Park, North Carolina, USAAddress correspondence to F. Kamel, Epidemiology Branch, MD A3-05, NIEHS Box 12233, Research Triangle Park, NC 27709. Telephone: (919) 541-1581. Fax: (919) 541-2511. E-mail:
[email protected] appreciate the thoughtful comments of D. Baird and M. Longnecker on an earlier version of this paper.
This work was supported by internal funding to the Epidemiology Branch, NIEHS.
The authors declare they have no competing financial interests.
6 2004 20 5 2004 112 9 950 958 30 3 2004 19 5 2004 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Poisoning by acute high-level exposure to certain pesticides has well-known neurotoxic effects, but whether chronic exposure to moderate levels of pesticides is also neurotoxic is more controversial. Most studies of moderate pesticide exposure have found increased prevalence of neurologic symptoms and changes in neurobehavioral performance, reflecting cognitive and psychomotor dysfunction. There is less evidence that moderate exposure is related to deficits in sensory or motor function or peripheral nerve conduction, but fewer studies have considered these outcomes. It is possible that the most sensitive manifestation of pesticide neurotoxicity is a general malaise lacking in specificity and related to mild cognitive dysfunction, similar to that described for Gulf War syndrome. Most studies have focused on organophosphate insecticides, but some found neuro-toxic effects from other pesticides, including fungicides, fumigants, and organochlorine and carbamate insecticides. Pesticide exposure may also be associated with increased risk of Parkinson disease; several classes of pesticides, including insecticides, herbicides, and fungicides, have been implicated. Studies of other neurodegenerative diseases are limited and inconclusive. Future studies will need to improve assessment of pesticide exposure in individuals and consider the role of genetic susceptibility. More studies of pesticides other than organophosphates are needed. Major unresolved issues include the relative importance of acute and chronic exposure, the effect of moderate exposure in the absence of poisoning, and the relationship of pesticide-related neurotoxicity to neurodegenerative disease.
fumigantfungicideinsecticideneurobehavioral performanceneurodegenerative diseaseneurologic symptomsorganophosphateParkinson diseasepesticide
==== Body
Pesticides are used extensively throughout the world. In the United States, more than 18,000 products are licensed for use, and each year > 2 billion pounds of pesticides are applied to crops, homes, schools, parks, and forests [U.S. Environmental Protection Agency (EPA) Office of Pesticide Programs 2002]. Such widespread use results in pervasive human exposure.
Evidence continues to accumulate that pesticide exposure is associated with impaired health. Occupational exposure is known to result in an annual incidence of 18 cases of pesticide-related illness for every 100,000 workers in the United States (Calvert et al. 2004). The best-documented health effects involve the nervous system. The neurotoxic consequences of acute high-level pesticide exposure are well established: Exposure is associated with a range of symptoms as well as deficits in neurobehavioral performance and abnormalities in nerve function (Keifer and Mahurin 1997). Whether exposure to more moderate levels of pesticides is also neurotoxic is more controversial. Pesticide exposure may also be associated with increased risk of neurodegenerative disease, particularly Parkinson disease (Le Couteur et al. 1999).
In this review, we summarize briefly what is known about the neurotoxic effects of high-level exposure, describe in more detail the existing data on neurotoxic effects of chronic exposure at lower levels, and then discuss the relationship of pesticide exposure to neurologic disease. Although pesticide exposure may have significant effects on neurodevelopment (Eskenazi et al. 1999), this review focuses on effects in adults ≥ 18 years of age. Since differences in approach to evaluating pesticide exposure may play a crucial role in creating inconsistencies among studies, we first consider pesticide exposure assessment.
Pesticide Exposure
Pesticides are a broad range of substances most commonly used to control insects, weeds, and fungi (plant diseases). They are frequently classified by target organism or mode of use as insecticides, herbicides, fungicides, or fumigants. Insecticides are often subclassified by chemical type as organophosphates (OPs), organochlorines, carbamates, and pyrethroids. Individuals are frequently exposed to many different pesticides or mixtures of pesticides, either simultaneously or serially. These exposures are often highly correlated, particularly within functional or chemical groups, making it difficult to identify effects of particular agents.
Studies of pesticide neurotoxicity have typically evaluated either the long-term sequelae of pesticide poisoning or the effects of occupational exposure (Table 1). Pesticide poisoning may go undiagnosed, especially among farm-workers with poor access to medical care (Moses et al. 1993) and particularly among women (London et al. 2002). Thus, workers who have never been diagnosed with pesticide poisoning may still have sustained high exposures or experienced pesticide-related illness; therefore using diagnosed poisoning as a criterion for inclusion in an exposed group or exclusion from a comparison group may incorrectly classify individuals.
Some studies of occupational pesticide exposure have classified as exposed all members of an occupational group—typically farmers or farmworkers—sometimes also considering job duration. The potential for misclassification with this approach is high. Farm owners who employ others to apply pesticides may have limited personal exposure to pesticides. Even among pesticide applicators, exposure can vary widely. For example, farmworkers with little access to information about safety practices or protective equipment (Gomes et al. 1999) may sustain far more exposure than well-trained and equipped commercial applicators (Maizlish et al. 1987). Further, farm-workers who do not apply pesticides as part of their job may still be exposed, and even family members with no direct occupational exposure may be exposed at home or elsewhere (Fenske 1997; Gladen et al. 1998), so neither of these may be an appropriate comparison group.
Factors such as application method, use of personal protective equipment, work practices related to hygiene, spills, and attitudes toward risk may all influence the degree of pesticide exposure and can be incorporated into exposure estimates (Alavanja et al. 2004; Buchanan et al. 2001; Dosemeci et al. 2002; Gomes et al. 1999; Hernandez-Valero et al. 2001; London and Myers 1998; Ohayo-Mitoko et al. 1999; Stewart et al. 2001). The relationship of these factors to exposure can be complex. For example, wearing gloves can increase exposure under some circumstances (Hines et al. 2001), perhaps because fabric (as opposed to chemically impervious) gloves can become impregnated with pesticide and serve as a reservoir of exposure. The same may be true of other types of protective clothing (Ohayo-Mitoko et al. 1999). In developing countries, use of closed pesticide mixing and loading systems may increase exposure when the equipment is used to speed up work and increase productivity rather than to protect workers (McConnell et al. 1992). Additional factors may be crucial for evaluating exposure in farmworkers, such as availability of washing and drinking water, interval between application of pesticides to a field and re-entry of workers, and housing conditions (Arcury and Quandt 1998; Gomes et al. 1999; Hernandez-Valero et al. 2001; Tielemans et al. 1999). Studies of neurotoxicity have used all these kinds of information to evaluate pesticide exposure (Gomes et al. 1999; Ohayo-Mitoko et al. 1999). The most sophisticated approaches were employed by London and Myers (1998), who used a crop-and job-specific job exposure matrix to evaluate exposure in a study of the neurotoxicity of chronic OP exposure among South African farmworkers, and by Buchanan et al. (2001), who developed an exposure algorithm to predict diazinon exposure for a study of chronic neurologic effects among sheep dippers in the United Kingdom.
Both historic and current exposures may be relevant to neurotoxicity and need to be characterized. Even among people who remain in the same occupation, current exposure may not reflect past exposure patterns because both available products and methods of use change over time. The need to evaluate past as well as current exposure has limited the utility of bio-markers; most modern pesticides are not persistent, so studies of chronic exposure rely primarily on questionnaire-based methods. Biomarkers are, however, useful in some situations. For example, organochlorines have a long half-life, so serum levels can be used as a marker of exposure to these pesticides. OP inhibition of erythrocyte acetylcholinesterase (AChE) can also be used as an exposure marker. The effect lasts 3–4 months, so AChE activity in whole blood or erythrocytes can be used to evaluate subchronic exposure, although interpretation can be complicated by acute exposure. Although the clinical utility of this biomarker in individuals may be limited by variability in baseline levels, in populations chronic OP exposure is associated with small but reliable decreases in erythrocyte AChE activity (Karr et al. 1992; Ohayo-Mitoko et al. 1997). OPs also inhibit plasma butylcholinesterase, but the effect lasts at most a few weeks and is therefore not useful for evaluating chronic exposure. Cholinesterase inhibition by carbamates lasts only minutes, so it is not a useful marker of chronic exposure to these pesticides.
Estimating lifetime pesticide exposure quantitatively is difficult because it is affected by many factors, including the multiple chemicals involved, uncertainty regarding the degree of exposure related to specific job tasks or other events, and contributions from multiple sources of exposure, including sources unrelated to occupation. Further, the biologically relevant exposure measure is not known: Peak or average exposure intensity might be more important than cumulative exposure. Thus, attempts to assess quantitative dose–response relationships may be problematic. The goal of exposure assessment in epidemiologic studies is not, however, to assign quantitative dose estimates but rather to rank individuals by relative exposure level. Assignment of either exposed or unexposed individuals to the wrong category can be a significant problem, as can combining individuals with low and high levels of exposure into one group. Random misclassification of exposure, unrelated to health outcome, will typically weaken studies by making associations more difficult to detect, although it will not undermine the validity of any association that is observed. As discussed above, assuming that all farmers or even all pesticide applicators are equally exposed is likely to entail significant misclassification, as is assuming that all farmworkers who are not applicators are not exposed. Further, studies that identify only a single highly exposed group for study cannot evaluate the neurotoxicity of moderate exposure, which may have great significance to public health. Methods described above can correctly categorize study participants with respect to their relative exposure levels, and using such methods to increase precision of exposure assessment may help minimize inconsistencies among studies.
Neurotoxicity of High-Level Exposure
Most types of pesticides, including OP, carbamate, and organochlorine insecticides as well as fungicides and fumigants, can be neurotoxic, but only OPs have been studied in detail (Keifer and Mahurin 1997). The response to OPs can occur within minutes. Less severe cases of OP poisoning display symptoms including headache, dizziness, nausea, vomiting, pupillary constriction, and excessive sweating, tearing, and salivation. More severe cases develop muscle weakness and twitches, bronchospasm, and changes in heart rate and can progress to convulsions and coma. The mechanism of OP neurotoxicity in most cases involves overstimulation of postsynaptic cholinergic receptors after inhibition of AChE (Keifer and Mahurin, 1997), although other macromolecular targets may also be involved (Pope 1999). An intermediate syndrome, occurring 1–4 days after exposure, is characterized by muscle weakness and can be fatal if respiratory muscles are affected. Two to five weeks after exposure, some patients develop OP-induced delayed polyneuropathy, a well-characterized syndrome involving sensory abnormalities, muscle cramps, weakness, and even paralysis, primarily in the legs. These symptoms are a consequence of axonal death following OP inhibition of a neural enzyme called neuropathy target esterase and may be irreversible (Keifer and Mahurin 1997).
Several studies have shown that OP poisoning has additional long-term sequelae. Studies of individuals with a history of pesticide poisoning—farmworkers (London et al. 1998; McConnell et al. 1994; Rosenstock et al. 1991; Wesseling et al. 2002), farmers (Stallones and Beseler 2002), rescue workers (Nishiwaki et al. 2001), or individuals identified from hospitals or pesticide registries (Miranda et al. 2002; Savage et al. 1988; Steenland et al. 1994)—have found that increased symptom prevalence, deficits in cognitive and psychomotor function, decreased vibration sensitivity, and motor dysfunction can occur long after the immediate episode is resolved. In some cases, effects were observed ≥ 10 years after poisoning (Savage et al. 1988), suggesting that the residual damage is permanent. Even less severe poisoning can have long-term consequences: Banana farm workers who had been treated for intoxication with OPs or carbamates but did not require hospitalization performed worse on tests of cognitive and psychomotor function than did nonpoisoned workers when tested > 2 years later (Wesseling et al. 2002).
Neurotoxicity of Low-Level Exposure
Findings from studies of acute exposure to moderate levels of pesticides are inconsistent. Some studies of well-trained and -equipped pesticide applicators in the United States reported that exposure to OPs sustained during a single work shift (Maizlish et al. 1987) or assessed using a short-lived urinary bio-marker (Dick et al. 2001) was associated with little neurotoxicity. However, several studies in developing countries, where exposures may have been higher, found that acute exposure to OPs was associated with increased symptom prevalence in commercial applicators (Misra et al. 1985) and farmworkers (London et al. 1998; Ohayo-Mitoko et al. 2000). Acute and chronic exposures are often correlated, sometimes making it difficult to separate their effects. The following discussion focuses on the effects of chronic exposure to moderate levels of pesticides, although in many studies acute exposure may also have occurred. Several types of neurologic end points are considered, including symptom prevalence, neurobehavioral performance, sensory and motor dysfunction, and direct measures of nerve function. Studies are summarized in Table 2.
Symptom Prevalence
Studies of symptom prevalence are often based on variations of an established checklist (Lundberg et al. 1997) and evaluate a broad range of symptoms, including headache, dizziness, fatigue, insomnia, nausea, chest tightness, and difficulty breathing as well as symptoms suggesting cognitive (confusion, difficulty concentrating), motor (weakness, tremor), and sensory (numbness, tingling, visual disturbance) dysfunction. Pesticide exposure is associated with increases in prevalence of many symptoms, with little evidence for specificity. Most studies have focused on OPs; most of these found an association of exposure with increased symptom prevalence. Farmworkers (Gomes et al. 1998), greenhouse workers (Bazylewicz-Walczak et al. 1999), and factory workers (Bellin and Chow 1974) exposed to OPs reported increased symptom prevalence compared to unexposed workers. In particular, farmers and farmworkers who applied OPs had higher symptom prevalence than nonapplicators (London et al. 1998; Ohayo-Mitoko et al. 2000; Smit et al. 2003), as did commercial applicators (Misra et al. 1985; Steenland et al. 2000) and sheep dippers (Pilkington et al. 2001). Pesticides other than OPs also affect symptom prevalence: one study found that exposure to dichlorodiphenyl-trichloroethane (DDT) was associated with increased symptom prevalence (van Wendel de Joode et al. 2001), as did one study of fumigants (Anger et al. 1986) although not another (Calvert et al. 1998). Additional studies have evaluated changes in mood and affect, using either self-report or validated scales. Workers exposed to OPs (Bazylewicz-Walczak et al. 1999; Steenland et al. 2000; Stokes et al. 1995) or DDT (van Wendel de Joode et al. 2001) reported higher levels of tension, anger, or depression on standard symptom questionnaires, and OP applicators showed elevated levels of anxiety on personality tests (Levin et al. 1976). Three studies found no association of OPs with symptom prevalence or affect (Ames et al. 1995; Fiedler et al. 1997; Korsak and Sato 1977).
Increased symptom prevalence was correlated with inhibition of erythrocyte AChE in four studies of OP exposure (Bellin and Chow 1974; Gomes et al. 1998; Leng and Lewalter 1999; Ohayo-Mitoko et al. 2000) and with inhibition of both erythrocyte AChE and plasma cholinesterase in two of these (Bellin and Chow 1974; Leng and Lewalter 1999). Another study found no relationship of symptom prevalence to inhibition of either erythrocyte or plasma cholinesterase (Lee et al. 2003). One study found that increased symptom prevalence was associated with self-reported pesticide exposure but not with depressed erythrocyte AChE activity (Ciesielski et al. 1994). Effects of OP exposure may not necessarily be caused by AChE inhibition (Pope 1999). Further, farmworkers have complex work histories and are likely to be exposed to pesticides other than OPs that may affect symptom prevalence without affecting AChE.
Neurobehavioral Performance
Neurobehavioral test batteries, including the World Health Organization Neurobehavioral Core Test Battery (Anger et al. 2000), the Neurobehavioral Evaluation System (Letz et al. 1996), and portions of other batteries, have been used to evaluate pesticide effects on cognitive and psychomotor function. Tests included in these batteries assess memory, attention, visuospatial processing, and other aspects of cognitive function; commonly used tests include symbol digit, digit span, visual retention, pattern memory, trail making, and others. Most studies indicate that pesticide exposure is associated with deficits in cognitive function. Sheep dippers (Stephens et al. 1995), nursery workers (Bazylewicz-Walczak et al. 1999), and other workers (Korsak and Sato 1977) exposed to OPs, malaria-control workers who sprayed DDT (van Wendel de Joode et al. 2001), vineyard workers exposed to fungicides (Baldi et al. 2001), fumigators exposed to sulfuryl fluoride but not those exposed to methyl bromide (Anger et al. 1986; Calvert et al. 1998), and farmers (Cole et al. 1997), farm-workers (Gomes et al. 1998; Kamel et al. 2003), and pesticide applicators (Farahat et al. 2003) exposed to multiple pesticides all performed worse on tests of cognitive function. There are some inconsistencies among these studies. Although most studies found deficits on one or more tests of cognitive function, different tests were affected in different studies, and a few studies found no relationship of OP exposure to any test (Ames et al. 1995; Daniell et al. 1992; Fiedler et al. 1997; Rodnitzky et al. 1975; Steenland et al. 2000).
Deficits in psychomotor function could be caused by impairment of sensory input, motor output, or associative delays; tests used include reaction time, tapping, pursuit aiming, Santa Ana and other pegboard tests, and others. Most studies indicate that pesticide exposure is associated with deficits in psychomotor function. Farmworkers (Daniell et al. 1992; London et al. 1997), farmers (Fiedler et al. 1997) and termiticide applicators (Steenland et al. 2000) exposed to OPs, malaria-control workers who sprayed DDT (van Wendel de Joode et al. 2001), vineyard workers exposed to fungicides (Baldi et al. 2001), fumigators exposed to methyl bromide or sulfuryl fluoride (Anger et al. 1986; Calvert et al. 1998), and farmworkers with multiple exposures (Gomes et al. 1998; Kamel et al. 2003) all showed worse performance on tests of psychomotor function. Again, results for individual tests were not fully consistent within or among studies, and no change in psychomotor function was evident in two studies of OP exposure (Ames et al. 1995; Cole et al. 1997).
Sensory and Motor Dysfunction
Neurobehavioral test batteries are often supplemented with tests of sensory or motor function. One frequently used test is vibration sensitivity, which evaluates peripheral somatosensory function. Most available evidence suggests this is not affected by moderate pesticide exposure. One study of farmers exposed to OPs found decreased sensitivity (Stokes et al. 1995), and another of farmers exposed to multiple pesticides found both decreased sensitivity and other signs of peripheral neuropathy (Cole et al. 1998). However, other studies of individuals exposed to OPs (Ames et al. 1995; London et al. 1998; Pilkington et al. 2001; Steenland et al. 2000), DDT (van Wendel de Joode et al. 2001), fumigants (Anger et al. 1986; Calvert et al. 1998), or multiple pesticides (Kamel et al. 2003) found no relationship of exposure to vibration sensitivity or other measures of somatosensory function.
Few studies have evaluated other aspects of sensory function. One study suggested that the sense of smell was not affected by OPs (Steenland et al. 2000); another study suggested a relationship with fumigants (Calvert et al. 1998). Visual contrast sensitivity was not affected by exposure to OPs (Steenland et al. 2000; van Wendel de Joode et al. 2001) or multiple pesticides (Kamel et al. 2003), but color vision was (Steenland et al. 2000). Retinal degeneration was associated with fungicide exposure in a case–control study of licensed pesticide applicators (Kamel et al. 2000). In general, these data are too limited to draw conclusions about the relationship to pesticide exposure to sensory function.
Similarly, few studies have considered motor function, and few inferences can be made about its relationship to pesticide exposure. Tremor was related to exposure to multiple pesticides in one study (Davignon et al. 1965) but not to OPs in two others (London et al. 1998; Steenland et al. 2000). Grip strength was not related to exposure to fumigants (Anger et al. 1986), DDT (van Wendel de Joode et al. 2001), or multiple pesticides (Kamel et al. 2003).
Balance is an integrated sensorimotor function. An early study found deficits in balance in apple farmers exposed to multiple pesticides (Davignon et al. 1965). In modern studies, balance is commonly evaluated by a test of postural sway; varying the conditions of the test may indicate whether impaired balance is related to deficits in visual, proprioceptive, or vestibular input. Three studies of individuals exposed to OPs (Steenland et al. 2000) or to multiple pesticides (Kamel et al. 2003; Sack et al. 1993) found that impaired postural sway was associated with exposure, but effects were small and another study found no relationship of OP exposure to postural sway (Ames et al. 1995). Effects were most evident when both visual and proprioceptive inputs were removed, suggesting that vestibular function may be affected (Kamel et al. 2003; Sack et al. 1993).
Nerve Function
Studies that have evaluated peripheral nerve conduction have produced largely negative results. Several studies of OPs found little evidence of impaired nerve conduction (Ames et al. 1995; Engel et al. 1998; Steenland et al. 2000). One study of fumigators found deficits in nerve conduction (Calvert et al. 1998), but another did not (Anger et al. 1986). In contrast, fungicide exposure was related to impaired nerve conduction in a study of bulb farmers, which also found deficits in autonomic nerve function (Ruijten et al. 1994). One study found changes in electroencephalogram (EEG) associated with OP exposure (Korsak and Sato 1977).
Three studies have performed clinical neurologic examinations in a subset of individuals identified by field studies as having deficits related to OP exposure. Beach et al. (1996) studied sheep dippers with increased symptom prevalence (Stephens et al. 1995); Horowitz et al. (1999) studied apple farmers with decreased vibration sensitivity (Stokes et al. 1995); and Jamal et al. (2002) studied sheep dippers with peripheral neuropathy (Pilkington et al. 2001). In general, clinical examination confirmed the results of the field studies, although clinically recognizable neurologic abnormalities were minor and not present in all individuals identified by the field studies.
Genetic Susceptibility to Pesticide Neurotoxicity
Individual response to pesticide exposure may be affected by polymorphisms in genes affecting pesticide metabolism. The best-known example is paraoxonase, an enzyme that hydrolyzes active metabolites of OPs (Costa et al. 2003). Animal studies suggest that changes in serum paraoxonase activity alter susceptibility to OP toxicity (Costa et al. 2003). In humans, paraoxonase polymorphisms affect the relationship of OP exposure to both erythrocyte AChE inhibition and symptom prevalence (Lee et al. 2003; Leng and Lewalter 1999; Mackness et al. 2003; Sozmen et al. 2002). Although Costa et al. (2003) have suggested that adequate evaluation of susceptibility requires measuring serum paraoxonase activity as well as genotype, recent population-based studies have suggested that the discrepancy between genotype and phenotype is relatively small and that nongenetic factors contribute relatively little to variation in serum activity (Ferre et al. 2003; Vincent-Viry et al. 2003).
Neurodegenerative Disease
Parkinson Disease
An extensive literature suggests that pesticide exposure may increase risk of Parkinson disease (Le Couteur et al. 1999). Many studies have found an association of Parkinson disease risk with living in rural areas, drinking well water, and farming as an occupation (Priyadarshi et al. 2001). More specifically, case–control studies have observed that pesticide exposure is associated with increased Parkinson disease risk, although results are not fully consistent. Studies published before 1999 were reviewed by Le Couteur et al. (1999), who noted that 12 of 20 studies found a positive association, with 1.6- to 7-fold increases in risk. Some of these studies evaluated risks associated with ever exposure to any pesticide. This broad definition of exposure permits significant misclassification, which could minimize the magnitude of any association observed.
Recent studies with more detailed exposure assessment have generally found an association of pesticide exposure with Parkinson disease, with 1.5- to 7-fold increases in risk. Case–control studies found increased risk associated with possession of a pesticide use license (Baldereschi et al. 2003), cumulative pesticide exposure based on complete occupational histories (Baldi et al. 2003a; Fall et al. 1999), or occupational or other pesticide use (Herishanu et al. 2001). A cross-sectional study found an association of parkinsonism with exposure to any pesticide, although not with specific pesticides or pesticide classes (Engel et al. 2001), and an ecologic study found that Parkinson disease mortality was higher in California counties where pesticides were used than in counties where they were not (Ritz and Yu 2000). Two cohort studies with detailed exposure information confirmed these findings: Risk was related to years of plantation work and to self-reported pesticide exposure in men enrolled in the Honolulu Heart Program cohort (Petrovitch et al. 2002), and occupational exposure to pesticides assessed with a job-exposure matrix was strongly associated with Parkinson disease risk (5.6-fold increase in risk) in an older cohort living in a vineyard-growing region of France (Baldi et al. 2003b). Three case–control studies found no association of pesticide exposure with Parkinson disease (Behari et al. 2001; Kuopio et al. 1999; Taylor et al. 1999).
Most studies of pesticide exposure and Parkinson disease risk have been unable to implicate specific pesticides. Several studies found increased risk associated with exposure to either insecticides or herbicides (Butterfield et al. 1993; Gorell et al. 1998; Semchuk et al. 1992), and one study indicated that risk was elevated by exposure to organochlorines, OPs, or carbamates (Seidler et al. 1996). Several studies have implicated the herbicide paraquat (Hertzman et al. 1990; Liou et al. 1997), which produces selective degeneration of neurons involved in Parkinson disease (McCormack et al. 2002). Case reports have described Parkinson disease in individuals exposed to OPs (Bhatt et al. 1999; Davis et al. 1978); to herbicides including glyphosate (Barbosa et al. 2001), paraquat (Sanchez-Ramon et al. 1987), and diquat (Sechi et al. 1992); and to fungicides including maneb (Meco et al. 1994) and other dithiocarbamates (Hoogenraad 1988). Higher concentrations of organochlorines, particularly dieldrin, have been found in postmortem brains of Parkinson disease patients compared to patients with other neurologic diseases (Corrigan et al. 2000; Fleming et al. 1994).
Animal models have also implicated pesticide exposure in the etiology of Parkinson disease. In rats, systemic administration of rotenone has been shown to produce highly selective neural degeneration similar to that found in Parkinson disease as well as a parkinsonian behavioral disorder (Betarbet et al. 2000). Treatment of mice with both paraquat and maneb reduced motor activity and striatal tyrosine hydroxylase activity, at doses at which neither compound was effective alone (Thiruchelvam et al. 2000).
Other Neurodegenerative Diseases
Information on pesticide exposure and other neurologic diseases is more limited. Several studies have suggested that risk of amyotrophic lateral sclerosis (ALS) is related to farming as an occupation, although not necessarily to living in rural areas (Nelson 1995–1996). Pesticide exposure has been considered in six case–control studies; three found some evidence for an association (Deapen and Henderson 1986; McGuire et al. 1997; Savettieri et al. 1991), whereas three others found none (Chancellor et al. 1993; Granieri et al. 1988; Gunnarsson et al. 1992). Only one study presented detailed exposure information (McGuire et al. 1997): Based on an industrial hygiene assessment of a complete occupational history, pesticide exposure was associated with > 2-fold increase in ALS risk, with greater risk at higher levels of exposure. This study did not implicate specific pesticides in ALS etiology. However, a cohort study found increased risk of ALS among workers exposed to the herbicide 2,4-dichlorophenoxyacetic acid (2,4-D) compared to other company employees, although this result was based on only three deaths (Burns et al. 2001). Case reports have described ALS after exposure to OPs (Bidstrup et al. 1953) and organochlorines (Fonseca et al. 1993).
Dementia has also been related to pesticide exposure. Occupational exposure to unspecified pesticides and fertilizers was associated with risk of Alzheimer disease in a large case–control study (McDowell et al. 1994), although another smaller study of environmental exposure in the general population found no relationship to herbicides, insecticides, or pesticides (Gauthier et al. 2001). Occupational exposure to any pesticide assessed with a job–exposure matrix was associated with 2-fold increase in risk of Alzheimer disease in a cohort of older individuals living in a vineyard-growing region of France and exposed primarily to dithiocarbamate fungicides (Baldi et al. 2003b). Occupational pesticide exposure was also associated with mild cognitive dysfunction in a population-based prospective study (Bosma et al. 2000), with vascular dementia (Lindsay et al. 1997), and with risk of dementia among Parkinson disease patients (Hubble et al. 1998). Understanding the relationship of pesticide exposure to Alzheimer disease may be complicated by the fact that the basic neurochemical defect in Alzheimer disease is loss of cholinergic neurons, and that to increase cholinergic tone Alzheimer disease is sometimes treated with OP cholinesterase inhibitors (Ringman and Cummings 1999).
Conclusion
Most studies of neurotoxicity have documented an increase in symptom prevalence and changes in neurobehavioral performance reflecting cognitive and psychomotor dysfunction, but many found little effect of pesticide exposure on sensory or motor function or direct measures of nerve function. There are several potential explanations for these findings. Except for vibrotactile sensitivity, information on sensory and motor function is limited, and further study may reveal associations with pesticide exposure. Another possibility is that the increase in symptom prevalence is due to bias: Most studies were cross-sectional in design, and individuals with greater exposure or a history of poisoning may have been more motivated to recall or report symptoms. Confounding by head injury or neurologic disease, either of which might be related to both pesticide exposure and increased symptom prevalence, could also create the appearance of an association. Consistency of findings across many studies argues against these explanations, as do the positive findings of some studies that used more quantitative exposure measures. Further, bias and confounding are less likely to account for changes in neurobehavioral performance, which is assessed using objective test batteries. Thus, moderate pesticide exposure may in fact have greater effects on symptom prevalence and neurobehavioral performance than on sensory or motor function. The lack of specificity of the symptomatic response is also interesting. It is possible that the earliest or most general response to pesticide neurotoxicity is a general malaise lacking in specificity and related to mild cognitive dysfunction, similar to that described for Gulf War syndrome (White et al. 2001).
Although the weight of the evidence suggests that pesticide use is associated with increased symptom prevalence and deficits in neurobehavioral performance, there were some inconsistencies that future studies should attempt to resolve. It may be that certain functional domains are more sensitive to pesticides than others, but the current literature is too limited to resolve this question. Some of the inconsistencies among studies are likely due to methodologic differences. A critical concern is exposure assessment. Qualitative and quantitative aspects of the exposure under consideration differed among studies, as did the ability of the studies to assess exposure. Exposure measures ranged from job title to detailed assessment of cumulative exposure based on work history. There was, however, no clear-cut relationship between the quality of exposure assessment and the results of the studies.
The choice of comparison group may also influence results. Responses to symptom questionnaires and neurobehavioral performance are influenced by age, education, and cultural background (Anger et al. 1997), so it is important for comparison groups to be demographically similar to exposed populations. However, using a comparison group from the same community or workplace as the exposed participants can create problems. Although the former may have no documented exposure, they may nevertheless not be truly unexposed, limiting the power of the study to detect effects. There may be no one best solution to this problem.
Other aspects of study design, such as size, neurologic end points considered, and data analytic strategies including control for confounding, are likely to influence results. More than half of the studies considered were small, with < 100 exposed participants, and therefore had limited power to detect associations. Poor response rates in some studies may have biased results. Symptom questionnaires, neurobehavioral test batteries, and other methods for evaluating neurologic outcomes also varied among studies. In particular, different neurobehavioral batteries employ different tests of cognitive and psychomotor function. However, results were variable even for tests used in many studies. Implementation of a given test may vary between batteries; for example, a computerized version may differ from a paper-and-pencil model, but even this consideration may not explain all differences. A study of styrene found that grouping results of neurobehavioral tests provided increased power to detect effects of exposure, compared to evaluating individual tests (Heyer et al 1996). Use of similar analytic strategies might reduce inconsistencies among studies of pesticides.
Pesticide exposure may be associated with increased risk of Parkinson disease. Inconsistencies among studies are again likely to be caused by variations in study methodology, particularly lack of detailed exposure assessment in some earlier studies. The positive results from recent studies with more comprehensive exposure assessment, together with support from animal models, reinforces the hypothesis of an association. Results for ALS and Alzheimer disease are suggestive but too sparse to support firm conclusions. Whether the subtle signs of neurotoxicity found in studies of poisoning and occupational exposure are related to the later development of neurodegenerative disease is a question not adequately addressed by the literature, although one study showed that short- and long-term responses to moderate exposure are not necessarily related (Stephens et al. 1996).
Historically, most studies have focused on OPs, first to document sequelae of acute poisoning and then to explore the effects of chronic moderate exposure. There is also evidence suggesting that other types of pesticides, including organochlorines, carbamates, fungicides, and fumigants, are neurotoxic. No study has evaluated the association of herbicides with symptom prevalence or neurobehavioral performance, but these chemicals have been implicated as risk factors for Parkinson disease. Although it is important to identify classes of pesticides and even specific chemicals associated with neurotoxicity, it is also important to recognize that most workers are exposed to complex mixtures of pesticides, which may contribute synergistically to neurotoxicity.
Other aspects of the relationship of pesticide exposure to neurotoxicity remain to be clarified. Participants in most studies have sustained both chronic and acute exposures; because these are often correlated, the studies have not been able to disentangle their effects. It is also possible that studies of chronic moderate exposure have been influenced by inclusion of individuals with a history of pesticide poisoning in the exposed population. Several studies in which such individuals were excluded found no relationship of chronic exposure to neurobehavioral performance or nerve function (Ames et al. 1995; Engel et al. 1998; Fiedler et al. 1997), but other studies of nonpoisoned individuals have found associations (Kamel et al. 2003; Stephens et al. 1995; van Wendel de Joode et al. 2001), suggesting that moderate as well as high-level pesticide exposure is neurotoxic. An issue receiving increasing attention is genetic susceptibility to pesticide neurotoxicity. In particular, genetic variation in paraoxonase has been related to OP neurotoxicity.
In conclusion, there is mounting evidence that chronic moderate pesticide exposure is neurotoxic and increases risk of Parkinson disease. To substantiate these findings, future studies must employ more detailed assessment of exposure in individuals and consider the role of genetic susceptibility. More studies of pesticides other than OPs and greater attention to disentangling the effects of different types of pesticides are also needed. Better information is required to clarify the relative importance of acute and chronic exposure and the role of moderate exposure in the absence of poisoning. Finally, it will be important to clarify the relationship of pesticide-related neurotoxicity to neurodegenerative disease.
Table 1 Studies of chronic pesticide exposure and neurotoxicity: exposure measurement.a
Reference Exposed population Chemicalb Exposure measurec No. Comparison group No.
Ames et al. 1995 Pesticide registry OP Mild poisoning 45 Friends 90
Anger et al. 1986 Fumigators Fumigants High pesticide use 74 Fumigators, low exposure 29
Baldi et al. 2001 Vineyard workers FNG Apply pesticide 528 Farmworkers, not exposed 216
Work in vineyards 173
Bazylewicz-Walczak et al. 1999 Greenhouse workers OP Work with plants 26 Greenhouse workers, not exposed 25
Bellin and Chow 1974 Factory workers OP, CAR AChE inhibition 83 Faculty, students, staff 56
Calvert et al. 1998 Fumigators Fumigants High pesticide use 123 Friends, neighbors 120
Ciesielski et al. 1994 Farmworkers Multiple Self-report; AChE inhibition 202 Local population 42
Cole et al. 1997 Farmers, some applicators OP, CAR, FNG Apply pesticide 144 Local population 72
Cole et al. 1998 Farmers, some applicators OP, CAR, FNG Apply pesticide 144 Local population 72
Daniell et al. 1992 Farmworker applicators OP Apply pesticide 49 Slaughterhouse workers 40
Davignon et al. 1965 Apple farmers Multiple Apply pesticide 441 Local population 162
Engel et al. 1998 Farmworkers OP Current farmwork 67 Local population 68
Farahat et al. 2003 Farmworker applicators OP, CAR, PYR Apply pesticide 52 Clerks, administrators 50
Fiedler et al. 1997 Fruit tree farmers OP Cumulative exposure 57 Berry farmers; storeowners 42
Gomes et al. 1998 Farmworkers Multiple Past and current farmwork 226 Domestic workers 226
Current farmwork 92
Kamel et al. 2003 Farmworkers Multiple Years of work 288 Local population 51
Korsak and Sato 1977 Occupational exposure OP High cumulative exposure 16 Low cumulative exposure 16
Levin et al. 1976 Pesticide applicators OP Current pesticide use 24 Farmers 24
London et al. 1997 Fruit farm applicators OP Cumulative exposure 163 Farmworkers, not applicators 84
London et al. 1998 Fruit farm applicators OP Cumulative exposure 164 Farmworkers, not applicators 83
McConnell et al. 1994 Farmworkers OP Poisoning 36 Friends, siblings 36
Miranda et al. 2002 Hospital patients OP Poisoning 62 Cattle ranchers, fishermen 39
Misra et al. 1985 Commercial applicators OP Apply pesticide 22 Hospital workers 20
Nishiwaki et al. 2001 Rescue workers OP Poisoning 56 Rescue workers, not exposed 52
Ohayo-Mitoko et al. 2000 Farmworker applicators OP, CAR AChE inhibition 256 Farmworkers 152
Pilkington et al. 2001 Sheep dippers OP Cumulative exposure 612 Farmers, ceramic workers 160
Rodnitzky et al. 1975 Pesticide applicators OP Current pesticide use 23 Farmers 24
Rosenstock et al. 1991 Farmworkers OP Poisoning 36 Friends, siblings 36
Ruijten et al. 1994 Flower bulb farmers FNG Apply pesticide 131 Population 67
Sack et al. 1993 Commercial applicators Multiple Apply pesticide 37 Students, staff 35
Savage et al. 1988 Registry OP Poisoning 100 Multiple sources 100
Smit et al. 2003 Farmers Multiple Apply pesticide 216 Fishermen 44
Stallones and Beseler 2002 Farmers, spouses Multiple Poisoning 69 Other farm residents 692
Steenland et al. 1994 Pesticide registry OP Poisoning 128 Friends 90
Steenland et al. 2000 Commercial applicators OP Apply pesticide 191 Friends, blue-collar workers 189
Stephens et al. 1995 Sheep dippers OP Apply pesticide 146 Quarry workers 143
Stokes et al. 1995 Apple orchard applicators OP Apply pesticide 68 Population 68
van Wendel de Joode et al. 2001 Pesticide applicators DDT Years apply pesticide 27 Guards, drivers 27
Wesseling et al. 2002 Farmworkers OP, CAR Poisoning 81 Farmworkers 130
Abbreviations: CAR, carbamates; FNG, fungicides; PYR, pyrethroids.
aOnly studies of chronic exposure in adults ≥ 18 years of age with comparison groups are included. Two studies (Baldi et al. 2001 and Gomes et al. 1998) evaluated two exposed groups. In four cases, two references report studies of different neurologic outcomes in the same population: Cole et al. (1997, 1998); Levin et al. (1976) and Rodnitzky et al. (1975); London et al. (1997, 1998); and McConnell et al. (1994) and Rosenstock et al. (1991). Studies are listed alphabetically.
bIdentifies the chemical emphasized by the study; participants may have been exposed to others.
cExposure measure that was used for evaluation of relationship of chronic exposure to neurotoxicity; if more than one measure was used for analysis, then the one providing the most specific information on individual exposure is listed.
Table 2 Studies of chronic pesticide exposure and neurotoxicity: neurologic outcomes.
Reference Symptoms, affect Cognitive function Psychomotor function Vibration sensitivity Balance Tremor Nerve functiona
Ames et al. 1995 0 0 0 0 0 0
Anger et al. 1986 1 1 1 0 0
Baldi et al. 2001 1 1
Bazylewicz-Walczak et al. 1999 1 1
Bellin and Chow 1974 1
Calvert et al. 1998 0 1 1 0 1
Ciesielski et al. 1994 1
Cole et al. 1997 1 0
Cole et al. 1998 1
Daniell et al. 1992 0 1
Davignon et al. 1965 1 1
Engel et al. 1998 0
Farahat et al. 2003 1
Fiedler et al. 1997 0 0 1
Gomes et al. 1998 1 1 1
Kamel et al. 2003 1 1 0 1
Korsak and Sato 1977 0 1 1
Levin et al. 1976 1
London et al. 1997 1 0
London et al. 1998 1 0 0
McConnell et al. 1994 1
Misra et al. 1985 1
Nishiwaki et al. 2001 1 0 0 0
Ohayo-Mitoko et al. 2000 1
Pilkington et al. 2001 1 0
Rodnitzky et al. 1975 0
Rosenstock et al. 1991 1 1 1
Ruijten et al. 1994 1
Sack et al. 1993 1
Savage et al. 1988 1 1 1 0
Smit et al. 2003 1
Stallones and Beseler 2002 1
Steenland et al. 1994 1 1 1 1 0 0
Steenland et al. 2000 1 0 1 0 1 0 0
Stephens et al. 1995 1
Stokes et al. 1995 1 1
van Wendel de Joode et al. 2001 1 1 1 0
Wesseling et al. 2002 1 1 1
1 indicates the study found some relationship of pesticide exposure to the general category of outcome, although not necessarily for all tests; 0 indicates no relationship was observed for any test.
aPeripheral nerve conduction, EEG.
==== Refs
References
Alavanja M Hoppin J Kamel F 2004 Health effects of chronic pesticide exposure: cancer and neurotoxicity Annu Rev Public Health 25 155 197 15015917
Ames R Steenland K Jenkins B Chrislip D Russo J 1995 Chronic neurologic sequelae to cholinesterase inhibition among agricultural pesticide applicators Arch Environ Health 50 440 444 8572722
Anger W Moody L Burg J Brightwell WS Taylor BJ Russo JM 1986 Neurobehavioral evaluation of soil and structural fumigators using methyl bromide and sulfuryl fluoride Neurotoxicology 7 137 156 3822255
Anger WK Liang YX Nell V Kang SK Cole D Bazylewicz-Walczak B 2000 Lessons learned—15 years of the WHO-NCTB: a review Neurotoxicology 21 837 846 11130289
Anger WK Sizemore OJ Grossmann SJ Glasser JA Letz R Bowler R 1997 Human neurobehavioral research methods: impact of subject variables Environ Res 73 18 41 9311528
Arcury T Quandt S 1998 Chronic agricultural chemical exposure among migrant and seasonal farmworkers Soc Nat Resour 11 829 843
Baldereschi M Di CA Vanni P Ghetti A Carbonin P Amaducci L 2003 Lifestyle-related risk factors for Parkinson’s disease: a population-based study Acta Neurol Scand 108 239 244 12956856
Baldi I Cantagrel A Lebailly P Tison F Dubroca B Chrysostome V 2003a Association between Parkinson’s disease and exposure to pesticides in southwestern France Neuroepidemiology 22 305 310 12902626
Baldi I Filleul L Mohammed-Brahim B Fabrigoule C Dartigues J Schwall S 2001 Neuropsychologic effects of long-term exposure to pesticides: results from the French Phytoner study Environ Health Perspect 109 839 844 11564621
Baldi I Lebailly P Mohammed-Brahim B Letenneur L Dartigues JF Brochard P 2003b Neurodegenerative diseases and exposure to pesticides in the elderly Am J Epidemiol 157 409 414 12615605
Barbosa ER Leiros da Costa MD Bacheschi LA Scaff M Leite CC 2001 Parkinsonism after glycine-derivate exposure Mov Disord 16 565 568 11391760
Bazylewicz-Walczak B Majczakowa W Szymczak M 1999 Behavioral effects of occupational exposure to organophosphorous pesticides in female greenhouse planting workers Neurotoxicology 20 819 826 10591517
Beach JR Spurgeon A Stephens R Heafield T Calvert IA Levy LS 1996 Abnormalities on neurological examination among sheep farmers exposed to organophosphorous pesticides Occup Environ Med 53 520 525 8983462
Behari M Srivastava AK Das RR Pandey RM 2001 Risk factors of Parkinson’s disease in Indian patients J Neurol Sci 190 49 55 11574106
Bellin JS Chow I 1974 Biochemical effects of chronic moderate exposure to pesticides Res Comm Chem Pathol Pharmacol 9 325 337
Betarbet R Sherer TB MacKenzie G Garcia-Osuna M Panov AV Greenamyre JT 2000 Chronic systemic pesticide exposure reproduces features of Parkinson’s disease Nat Neurosci 3 1301 1306 11100151
Bhatt MH Elias MA Mankodi AK 1999 Acute and reversible parkinsonism due to organophosphate pesticide intoxication: five cases Neurology 52 1467 1471 10227636
Bidstrup P Bonnel JA Beckett AG 1953 Paralysis following poisoning by a new organic phosphorus insecticide (Mipafox): report on two cases Br Med J 1 1068 1072 13042137
Bosma H van Boxtel MP Ponds RW Houx PJ Jolles J 2000 Pesticide exposure and risk of mild cognitive dysfunction Lancet 356 912 913 11036900
Buchanan D Pilkington A Sewell C Tannahill SN Kidd MW Cherrie B 2001 Estimation of cumulative exposure to organophosphate sheep dips in a study of chronic neurological health effects among United Kingdom sheep dippers Occup Environ Med 58 694 701 11600724
Burns CJ Beard KK Cartmill JB 2001 Mortality in chemical workers potentially exposed to 2,4-dichlorophenoxyacetic acid (2,4-D) 1945-94: an update Occup Environ Med 58 24 30 11119631
Butterfield PG Valanis BG Spencer PS Lindeman CA Nutt JG 1993 Environmental antecedents of young-onset Parkinson’s disease Neurology 43 1150 1158 8170560
Calvert GM Mueller CA Fajen JM Chrislip DW Russo J Briggle T 1998 Health effects associated with sulfuryl fluoride and methyl bromide exposure among structural fumigation workers Am J Public Health 88 1774 1780 9842373
Calvert GM Plate DK Das R Rosales R Shafey O Thomsen C 2004 Acute occupational pesticide-related illness in the US, 1998–1999: surveillance findings from the SENSOR-pesticides program Am J Ind Med 45 14 23 14691965
Chancellor AM Slattery JM Fraser H Warlow CP 1993 Risk factors for motor neuron disease—a case-control study based on patients from the Scottish motor neuron disease register J Neurol Neurosurg Psychiatry 56 1200 1206 8229031
Ciesielski S Loomis DP Mims SR Auer A 1994 Pesticide exposures, cholinesterase depression, and symptoms among North Carolina migrant farmworkers Am J Publ Health 84 446 51
Cole DC Carpio F Julian J Leon N 1998 Assessment of peripheral nerve function in an Ecuadorian rural population exposed to pesticides J Toxicol Environ Health A 55 77 91 9761129
Cole DC Carpio F Julian J Leon N Carbotte R De Almeida H 1997 Neurobehavioral outcomes among farm and nonfarm rural Ecuadorians Neurotoxicol Teratol 19 277 286 9253006
Corrigan FM Wienburg CL Shore RF Daniel SE Mann D 2000 Organochlorine insecticides in substantia nigra in Parkinson’s disease J Toxicol Environ Health 59 229 234
Costa LG Cole TB Furlong CE 2003 Polymorphisms of paraoxonase (PON1) and their significance in clinical toxicology of organophosphates J Toxicol Clin Toxicol 41 37 45 12645966
Daniell W Barnhart S Demers P Costa LG Eaton DL Miller M 1992 Neuropsychological performance among agricultural pesticide applicators Environ Res 59 217 228 1425511
Davignon LF St Pierre J Charest G Tourangeau FJ 1965 A study of the chronic effects of insecticides in man Can Med Assoc J 92 597 602 14264969
Davis K Yesavage J Berger P 1978 Single case study. Possible organophosphate induced parkinsonism J Nerv Ment Dis 166 222 225 641541
Deapen D Henderson BE 1986 A case-control study of amyotrophic lateral sclerosis Am J Epidemiol 123 790 799 3962963
Dick R Steenland K Krieg E Hines C 2001 Evaluation of acute sensory-motor effects and test sensitivity using termiticide workers exposed to chlorpyrifos Neurotoxicol Teratol 23 381 393 11485841
Dosemeci M Alavanja MC Rowland AS Mage D Zahm SH Rothman N 2002 A quantitative approach for estimating exposure to pesticides in the Agricultural Health Study Ann Occup Hyg 46 245 260 12074034
Engel LS Checkoway H Keifer MC Seixas NS Longstreth WT Jr Scott KC 2001 Parkinsonism and occupational exposure to pesticides Occup Environ Med 58 582 589 11511745
Engel LS Keifer MC Checkoway H Robinson LR Vaughan TL 1998 Neurophysiological function in farm workers exposed to organophosphate pesticides Arch Environ Health 53 7 14 9570304
Eskenazi B Bradman A Castorina R 1999 Exposures of children to organophosphate pesticides and their potential adverse health effects Environ Health Perspect 107 suppl 3 409 419 10346990
Fall PA Fredrikson M Axelson O Granerus AK 1999 Nutritional and occupational factors influencing the risk of Parkinson’s disease: a case-control study in southeastern Sweden Mov Disord 14 28 37 9918341
Farahat T Abdelrasoul GM Amr MM Shebl MM Farahat FM Anger WK 2003 Neurobehavioral effects among workers occupationally exposed to organophosphorous pesticides Occup Environ Med 60 279 286 12660376
Fenske RA 1997 Pesticide exposure assessment of workers and their families Occup Med 12 221 237 9220483
Ferre N Camps J Fernandez-Ballart J Arija V Murphy MM Ceruelo S 2003 Regulation of serum paraoxonase activity by genetic, nutritional, and lifestyle factors in the general population Clin Chem 49 1491 1497 12928230
Fiedler N Kipen H Kelly-McNeil K Fenske R 1997 Long-term use of organophosphates and neuropsychological performance Am J Ind Med 32 487 496 9327072
Fleming L Mann JB Bean J Briggle T Sanchez Ramos JR 1994 Parkinson’s disease and brain levels of organochlorine pesticides Ann Neurol 36 100 103 7517654
Fonseca RG Resende LAL Silva MD Camargo A 1993 Chronic motor neuron disease possibly related to intoxication with organochlorine insecticides Acta Neurol Scand 88 56 58 7690513
Gauthier E Fortier I Courchesne F Pepin P Mortimer J Gauvreau D 2001 Environmental pesticide exposure as a risk factor for Alzheimer’s disease: a case-control study Environ Res 86 37 45 11386739
Gladen BC Sandler DP Zahm SH Kamel F Rowland AS Alavanja MC 1998 Exposure opportunities of families of farmer pesticide applicators Am J Ind Med 34 581 587 9816416
Gomes J Lloyd O Revitt MD Basha M 1998 Morbidity among farm workers in a desert country in relation to long-term exposure to pesticides Scand J Work Environ Health 24 213 219 9710374
Gomes J Lloyd OL Revitt DM 1999 The influence of personal protection, environmental hygiene and exposure to pesticides on the health of immigrant farm workers in a desert country Int Arch Occup Environ Health 72 40 45 10029229
Gorell JM Johnson CC Rybicki BA Peterson EL Richardson RJ 1998 The risk of Parkinson’s disease with exposure to pesticides, farming, well water, and rural living Neurology 50 1346 1350 9595985
Granieri E Carreras M Tola R 1988 Motor neuron disease in the province of Ferrara, Italy, in 1964–1982 Neurology 38 1604 1608 3419606
Gunnarsson LG Bodin L Soderfeldt B Axelson O 1992 A case-control study of motor neurone disease - Its relation to heritability, and occupational exposures, particularly to solvents Br J Ind Med 49 791 798 1463680
Herishanu YO Medvedovski M Goldsmith JR Kordysh E 2001 A case-control study of Parkinson’s disease in urban population of southern Israel Can J Neurol Sci 28 144 147 11383940
Hernandez-Valero MA Bondy ML Spitz MR Zahm SH 2001 Evaluation of Mexican American migrant farmworker work practices and organochlorine pesticide metabolites Am J Ind Med 40 554 560 11675624
Hertzman C Wiens M Bowering D Snow B Calne D 1990 Parkinson’s disease: a case-control study of occupational and environmental risk factors Am J Ind Med 17 349 355 2305814
Heyer NJ Bittner AC Jr Echeverria D 1996 Analyzing multivariate neurobehavioral outcomes in occupational studies: a comparison of approaches Neurotoxicol Teratol 18 401 406 8866530
Hines C Deddens J Tucker S Hornung R 2001 Distributions and determinants of pre-emergent herbicide exposures among custom applicators Ann Occup Hyg 45 227 239 11295146
Hoogenraad T 1988 Dithiocarbamates and Parkinson’s disease [Letter] Lancet 1 767 2895292
Horowitz SH Stark A Marshall E Mauer MP 1999 A multimodality assessment of peripheral nerve function in organophosphate-pesticide applicators J Occup Environ Med 41 405 408 10337611
Hubble JP Kurth JH Glatt SL Kurth MC Schellenberg GD Hassanein RE 1998 Genetoxin interaction as a putative risk factor for Parkinson’s disease with dementia Neuroepidemiology 17 96 104 9592786
Jamal GA Hansen S Pilkington A Buchanan D Gillham RA Abdel-Azis M 2002 A clinical neurological, neuro-physiological, and neuropsychological study of sheep farmers and dippers exposed to organophosphate pesticides Occup Environ Med 59 434 441 12107290
Kamel F Boyes WK Gladen BC Rowland AS Alavanja MC Blair A 2000 Retinal degeneration in licensed pesticide applicators Am J Ind Med 37 618 628 10797505
Kamel F Rowland A Park L Anger W Baird D Gladen B 2003 Neurobehavioral performance and work experience in Florida farmworkers Environ Health Perspect 111 1765 1772 14594629
Karr C Demers P Costa L Daniell W Barnhart S Miller M 1992 Organophosphate pesticide exposure in a group of Washington State orchard applicators Environ Res 59 229 237 1425512
Keifer M Mahurin R 1997 Chronic neurologic effects of pesticide overexposure Occup Med 12 291 304 9220487
Korsak RJ Sato MM 1977 Effects of chronic organophosphate pesticide exposure on the central nervous system Clin Toxicol 11 83 95 872544
Kuopio AM Marttila RJ Helenius H Rinne UK 1999 Environmental risk factors in Parkinson’s disease Mov Disord 14 928 939 10584666
Le Couteur DG McLean AJ Taylor MC Woodham BL Board PG 1999 Pesticides and Parkinson’s disease Biomed Pharmacother 53 122 130 10349500
Lee BW London L Paulauskis J Myers J Christiani DC 2003 Association between human paraoxonase gene polymorphism and chronic symptoms in pesticide-exposed workers J Occup Environ Med 45 118 122 12625227
Leng G Lewalter J 1999 Role of individual susceptibility in risk assessment of pesticides Occup Environ Med 56 449 453 10472315
Letz R Green RC Woodard JL 1996 Development of a computer-based battery designed to screen adults for neuropsychological impairment Neurotoxicol Teratol 18 365 370 8866526
Levin HS Rodnitzky RL Mick DL 1976 Anxiety associated with exposure to organophosphate compounds Arch Gen Psychiatry 33 225 228 1252099
Lindsay J Hebert R Rockwood K 1997 The Canadian Study of Health and Aging: risk factors for vascular dementia Stroke 28 526 530 9056606
Liou HH Tsai MC Chen CJ Jeng JS Chang YC Chen SY 1997 Environmental risk factors and Parkinson’s disease: a case-control study in Taiwan Neurology 48 1583 1588 9191770
London L De Grosbois S Wesseling C Kisting S Rother H Mergler D 2002 Pesticide usage and health consequences for women in developing countries: out of sight, out of mind? Int J Occup Environ Health 8 46 59 11843440
London L Myers JE 1998 Use of a crop and job specific exposure matrix for retrospective assessment of long-term exposure in studies of chronic neurotoxic effects of agri-chemicals Occup Environ Med 55 194 201 9624271
London L Myers JE Nell V Taylor T Thompson ML 1997 An investigation into neurologic and neurobehavioral effects of long-term agrichemical use among deciduous fruit farm workers in the western cape, South Africa Environ Res 73 132 145 9311539
London L Nell V Thompson M Myers J 1998 Effects of long-term organophosphate exposures on neurological symptoms, vibration sense and tremor among South African farm workers Scand J Work Environ Health 24 18 29 9562397
Lundberg I Hogberg M Michelsen H Nise G Hogstedt C 1997 Evaluation of the Q16 questionnaire on neurotoxic symptoms and a review of its use Occup Environ Med 54 343 350 9196457
Mackness B Durrington P Povey A Thomson S Dippnall M Mackness M 2003 Paraoxonase and susceptibility to organophosphorus poisoning in farmers dipping sheep Pharmacogenetics 13 81 88 12563177
Maizlish N Schenker M Weisskoph C Seiber J Samuels S 1987 A behavioral evaluation of pest control workers with short-term, low-level exposure to the organophosphate diazinon Am J Ind Med 12 153 172 3661569
McConnell R Cordon M Murray DL Magnotti R 1992 Hazards of closed pesticide mixing and loading systems—the paradox of protective pechnology in the phird world Br J Ind Med 49 615 619 1390266
McConnell R Keifer M Rosenstock L 1994 Elevated quantitative vibrotactile threshold among workers previously poisoned with methamidophos and other organophosphate pesticides Am J Ind Med 25 325 334 8160653
McCormack AL Thiruchelvam M Manning-Bog AB Thiffault C Langston JW Cory-Slechta DA 2002 Environmental risk factors and Parkinson’s disease: selective degeneration of nigral dopaminergic neurons caused by the herbicide paraquat Neurobiol Dis 10 119 127 12127150
McDowell I Hill G Lindsay J Helliwell B Costa L Beattie L 1994 The Canadian study of health and aging: risk factors for Alzheimer’s disease in Canada Neurology 44 2073 2080 7969962
McGuire V Longstreth W Jr Nelson L Koepsell T Checkoway H Morgan M 1997 Occupational exposures and amyotrophic lateral sclerosis: a population-based case-control study Am J Epidemiol 145 1076 1088 9199537
Meco G Bonifati V Vanacore N Fabrizio E 1994 Parkinsonism after chronic exposure to the fungicide maneb (manganese ethylene-bis-dithiocarbamate) Scand J Work Environ Health 20 301 305 7801076
Miranda J Lundberg I McConnell R Delgado E Cuadra R Torres E 2002 Onset of grip- and pinch-strength impairment after acute poisonings with organophosphate insecticides Int J Occup Environ Health 8 19 26 11843436
Misra U Nag D Bushan V Ray PK 1985 Clinical and biochemical changes in chronically exposed organophosphate workers Toxicol Lett 24 187 193 3983971
Moses M Johnson ES Anger WK Burse VW Horstman SW Jackson RJ 1993 Environmental equity and pesticide exposure Toxicol Ind Health 9 913 959 8184449
Nelson LM 1995–1996 Epidemiology of ALS Clin Neurosci 3 327 331 9021253
Nishiwaki Y Maekawa K Ogawa Y Asukai N Minami M Omai K 2001 Effects of sarin on the nervous system in rescue team staff members and police officers 3 years after the Tokyo subway sarin attack Environ Health Perspect 109 1169 1173 11713003
Ohayo-Mitoko GJ Kromhout H Karumba PN Boleij JS 1999 Identification of determinants of pesticide exposure among Kenyan agricultural workers using empirical modelling Ann Occup Hyg 43 519 525 10616325
Ohayo-Mitoko GJ Kromhout H Simwa JM Boleij JS Heederik D 2000 Self reported symptoms and inhibition of acetyl-cholinesterase activity among Kenyan agricultural workers Occup Environ Med 57 195 200 10810102
Ohayo-Mitoko GJA Heederik DJJ Kromhout H Omondi BEO Boleij JSM 1997 Acetylcholinesterase inhibition as an indicator of organophosphate and carbamate poisoning in Kenyan agricultural workers Int J Occup Environ Health 3 210 220 9891121
Petrovitch H Ross GW Abbott RD Sanderson WT Sharp DS Tanner CM 2002 Plantation work and risk of Parkinson disease in a population-based longitudinal study Arch Neurol 59 1787 1792 12433267
Pilkington A Buchanan D Jamal GA Gillham R Hansen S Kidd M 2001 An epidemiological study of the relations between exposure to organophosphate pesticides and indices of chronic peripheral neuropathy and neuropsychological abnormalities in sheep farmers and dippers Occup Environ Med 58 702 710 11600725
Pope CN 1999 Organophosphorus pesticides: do they all have the same mechanism of toxicity? J Toxicol Environ Health B Crit Rev 2 161 181 10230392
Priyadarshi A Khuder SA Schaub EA Priyadarshi SS 2001 Environmental risk factors and Parkinson’s disease: a metaanalysis Environ Res 86 122 127 11437458
Ringman JM Cummings JL 1999 Metrifonate: update on a new antidementia agent J Clin Psychiatry 60 776 782 10584768
Ritz B Yu F 2000 Parkinson’s disease mortality and pesticide exposure in California 1984–1994 Int J Epidemiol 292 323 329 10817132
Rodnitzky RL Levin HS Mick DL 1975 Occupational exposure to organophosphate pesticides: a neurobehavioral study Arch Environ Health 30 98 103 1115534
Rosenstock L Keifer M Daniell WE McConnell R Claypoole K the Pesticide Health Effects Study Group 1991 Chronic central nervous system effects of acute organophosphate pesticide intoxication Lancet 338 223 227 1676786
Ruijten M Salle HJA Verberk MM Smink M 1994 Effect of chronic mixed pesticide exposure on peripheral and autonomic nerve function Arch Environ Health 49 188 195 8185390
Sack D Linz D Shukla R Rice C Bhattacharya A Suskind R 1993 Health status of pesticide applicators: postural stability assessments J Occup Med 35 1196 1202 8113922
Sanchez-Ramon J Hefti F Weiner WI 1987 Paraquat and Parkinson disease [Letter] Neurology 37 728
Savage E Keefe TJ Mounce LM Heaton RK Lewis JA Burcar PJ 1988 Chronic neurological sequelae of acute organophosphate pesticide poisoning Arch Environ Health 43 38 45 3355242
Savettieri G Salemi G Arcara A Cassata M Castiglione MG Fierro B 1991 A case-control study of amyotrophic lateral sclerosis Neuroepidemiology 10 242 245 1798425
Sechi G Agnetti V Piredda M Canu M Deserra F Omar HA 1992 Acute and persistent parkinsonism after use of diquat Neurology 42 261 263 1734316
Seidler A Hellenbrand W Robra BP Vieregge P Nischan P Joerg J 1996 Possible environmental, occupational, and other etiologic factors for Parkinson’s disease: a case-control study in Germany Neurology 46 1275 1284 8628466
Semchuk K Love EJ Lee RG 1992 Parkinson’s disease and exposure to agricultural work and pesticide chemicals Neurology 42 1328 1335 1620342
Smit LAM van Wendel de Joode BN Heederik D Peiris-John RJ van der Hoek W 2003 Neurological symptoms among Sri Lankan farmers occupationally exposed to acetyl-cholinesterase-inhibiting insecticides Am J Ind Med 44 254 264 12929145
Sozmen EY Mackness B Sozmen B Durrington P Girgin FK Aslan L 2002 Effect of organophosphate intoxication on human serum paraoxonase Hum Exp Toxicol 21 247 252 12141395
Stallones L Beseler C 2002 Pesticide illness, farm practices, and neurological symptoms among farm residents in Colorado Environ Res 90 89 97 12483798
Steenland K Dick RB Howell RJ Chrislip DW Hines CJ Reid TM 2000 Neurologic function among termiticide applicators exposed to chlorpyrifos Environ Health Perspect 108 293 300 10753086
Steenland K Jenkins B Ames RG O’Malley M Chrislip D Russo J 1994 Chronic neurological sequelae to organophosphate pesticide poisoning Am J Public Health 84 731 736 8179040
Stephens R Spurgeon A Berry H 1996 Organophosphates: the relationship between chronic and acute exposure effects Neurotoxicol Teratol 18 449 453 8866537
Stephens R Spurgeon A Calvert IA Beach J Levy LS Berry H 1995 Neuropsychological effects of long-term exposure to organophosphates in sheep dip Lancet 345 1135 1139 7723544
Stewart PA Prince JK Colt JS Ward MH 2001 A method for assessing occupational pesticide exposures of farmworkers Am J Ind Med 40 561 570 11675625
Stokes L Stark A Marshall E Narang A 1995 Neurotoxicity among pesticide applicators exposed to organophosphates Occup Environ Med 52 648 653 7489054
Taylor CA Saint-Hilaire MH Cupples LA Thomas CA Burchard AE Feldman RG 1999 Environmental, medical, and family history risk factors for Parkinson’s disease: a New England-based case control study Am J Med Genet 88 742 749 10581500
Thiruchelvam M Richfield EK Baggs RB Tank AW Cory-Slechta DA 2000 The nigrostriatal dopaminergic system as a preferential target of repeated exposures to combined paraquat and maneb: implications for Parkinson’s disease J Neurosci 20 9207 9214 11124998
Tielemans E Louwerse E de Cock J Brouwer D Zielhuis G Heederik D 1999 Exposure to fungicides in fruit growing: re-entry time as a predictor for dermal exposure Am Ind Hyg Assoc J 60 789 793 10635545
U. S. EPA Office of Pesticide Programs 2002. FY 2002 Annual Report. Washington, DC:U.S. Environmental Protection Agency. Available: http://www.epa.gov/oppfead1/annual/2002/2002annualreport.pdf [accessed 25 March 2004].
van Wendel de Joode B Wesseling C Kromhout H Monge P Garcia M Mergler D 2001 Chronic nervous-system effects of long-term occupational exposure to DDT Lancet 357 1014 1016 11293598
Vincent-Viry M Sass C Bastien S Aguillon D Siest G Visvikis S 2003 PON1-192 phenotype and genotype assessments in 918 subjects of the Stanislas cohort study Clin Chem Lab Med 41 535 540 12747599
Wesseling C Keifer M Ahlbom A McConnell R Moon J Rosenstock L 2002 Long-term neurobehavioral effects of mild poisonings with organophosphate and n-methyl carbamate pesticides among banana workers Int J Occup Environ Health 8 27 34 11843437
White RF Proctor SP Heeren T Wolfe J Krengel M Vasterling J 2001 Neuropsychological function in Gulf War Veterans: relationships to self-reported toxicant exposures Am J Ind Med 40 42 54 11439396
|
15198914
|
PMC1247187
|
CC0
|
2021-01-04 23:39:44
|
no
|
Environ Health Perspect. 2004 Jun 20; 112(9):950-958
|
utf-8
|
Environ Health Perspect
| 2,004 |
10.1289/ehp.7135
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0112-00095910.1289/ehp.680215198915Research ArticleReviewsPotential for Increased Human Foodborne Exposure to PCDD/F When Recycling Sewage Sludge on Agricultural Land Rideout Karen 12Teschke Kay 131School of Occupational and Environmental Hygiene2Institute for Resources, Environment and Sustainability3Department of Health Care and Epidemiology, University of British Columbia, Vancouver, British Columbia, CanadaAddress correspondence to K. Rideout, Institute for Resources, Environment and Sustainability, University of British Columbia, 2206 East Mall, Room 491, Vancouver, BC V6T 1Z3 Canada. Telephone: (604) 732-3571. Fax: (604) 822-9250. E-mail:
[email protected] authors are grateful for the thoughtful review of drafts of this article by the BC Ministry of Water, Land and Air Protection, the BC Ministries of Health, and the Greater Vancouver Regional District.
Funding was provided by the BC Ministry of Water, Land and Air Protection and Environment Canada.
The authors declare they have no competing financial interests.
6 2004 26 4 2004 112 9 959 969 15 10 2003 26 4 2004 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Sewage sludge from municipal wastewater treatment is used in agriculture as a nutrient source and to aid in moisture retention. To examine the potential impact of sludge-amended soil on exposures to polychlorinated dibenzo-p-dioxins and dibenzofurans (PCDD/Fs) from plant and animal foods, we conducted a review of published empirical data from international sources. Levels of PCDD/F in municipal sewage sludge ranged from 0.0005 to 8,300 pg toxic equivalents (TEQ)/g. Background levels in soil ranged from 0.003 to 186 pg TEQ/g. In sludge-amended soils, levels of PCDD/F ranged from 1.4 to 15 pg TEQ/g. Studies that measured levels before and after sludge treatment showed an increase in soil concentration after treatment. Relationships between PCDD/F levels in soil and resulting concentrations in plants were very weakly positive for unpeeled root crops, leafy vegetables, tree fruits, hay, and herbs. Somewhat stronger relationships were observed for plants of the cucumber family. In all cases, large increases in soil concentration were required to achieve a measurable increase in plant contamination. A considerably stronger positive relationship was observed between PCDD/F in feed and resulting levels in cattle tissue, suggesting bioaccumulation. Although PCDD/Fs are excreted in milk, no association was found between feed contamination and levels of PCDD/Fs measured in milk. There is a paucity of realistic data describing the potential for entry of PCDD/Fs into the food supply via sewage sludge. Currently available data suggest that sewage sludge application to land used for most crops would not increase human exposure. However, the use of sludge on land used to graze animals appears likely to result in increased human exposure to PCDD/F.
agriculturebioaccumulationbiosolidsdioxinsexposure assessmentfood chainfuransland recyclingPCDD/Fplant uptakesewage sludge
==== Body
In populations not industrially exposed to polychlorinated dibenzo-p-dioxins and dibenzofurans (PCDD/Fs), diet is responsible for virtually all (~ 98%) human exposure to these compounds (Pohl et al. 1995; Travis and Hattemer-Frey 1987). PCDD/Fs are common contaminants in municipal sewage sludge; thus, it is important to consider the risk of increased exposure to these contaminants if sewage sludge is to be applied to agricultural lands. There is currently much interest in agricultural use of sewage sludge to reap its benefits as fertilizer, as an aid in moisture retention, and to provide an alternative to incineration or landfills for disposal. The term “sewage sludge” is used here to refer to the solid by-product of municipal sewage or wastewater treatment processes. It includes but is not limited to “biosolids,” a term that usually refers to a stabilized product that has been treated to reduce pathogen content and vector attraction potential. The more inclusive term is used here because the data used in this review included all forms of municipal sewage sludge and because PCDD/F content is not affected by the additional treatment processes.
Some authors who have examined food-borne exposure to PCDD/F via sewage sludge have conducted deterministic modeling, using a number of assumptions including sludge application rates, exposure duration, PCDD/F concentration in sewage sludge, application methods, timing of application with respect to harvesting or sampling, and impact of atmospheric deposition. Those interested in such reports are referred to Duarte-Davidson and Jones (1996), Jackson and Eduljee (1994), Jones and Sewart (1997), Rappe and colleagues (1999), Wild and Jones (1992), and Wild and colleagues (1994). The U.S. Environmental Protection Agency (U.S. EPA) has recently modeled disease risks (cancer) from land-applied sewage sludge (U.S. EPA 2004).
In this article, we review the international empirical evidence of the impact of contaminated soil on the concentrations of PCDD/F in plant and animal tissue. We undertook this review to provide guidance regarding agricultural use of sewage sludge to federal, provincial, and municipal governments in Canada. Our purpose was to examine only the empirical literature and to use that literature to describe the potential transfer of PCDD/F from soil to foodstuffs, to derive empirical models of the transfer, and to identify data gaps in the science. We also wanted to determine whether some agricultural uses present greater likelihood than others of increased PCDD/F consumption by humans.
To organize the literature review process, we considered the pathways by which PCDD/F might be transferred from sewage products to humans via the food supply. Contaminants may adhere directly to plant surfaces or they may move from the sludge into the soil. From the soil, they may be transferred to crops, which are then consumed by humans or animals. These animals may in turn be consumed by humans. Animals also consume soil while grazing, which potentially increases their contaminant load.
Methods
Literature Search
A systematic search of the published literature was conducted using the following databases: MEDLINE (http://gateway2.ovid.com/), TOXLINE (http://toxnet.nlm.nih.gov/), Agricola (http://agricola.nal.usda.gov/), National Technical Information Service (http://www.ntis.gov/search/index.asp?loc=3-0-0), EMBASE (http://www.embase.com/), CAB International Abstracts (http://www.cabi.org/), Environmental Sciences and Pollution Management (http://ca1.csa.com), Food Science and Technology Abstracts (http://www.foodsciencecentral.com), Web of Science (http://isiknowledge.com), Compendex (http://www.engineeringvillage2.org), Dissertation Abstracts (http://wwwlib.umi.com/dissertations/gateway), Public Affairs Information Service, and Canadian Institute for Scientific and Technical Information (http://cat.cisti.nrc.ca/screens/opacmenu.html).
Combinations of the following key words were used in the searches: agricultural, agriculture, animals, application to land, application to soil, biosolids, crops, cropland, dibenzofuran, dioxin(s), fluid waste disposal, food contamination, forage, furan(s), land application, PCDD/F, PCDD, PCDF, plants, polychlorinated dibenzo-p-dioxin, polychlorinated dibenzofuran, sewage, sewage sludge, sewage as fertilizer, soil, soil ingestion, and soil pollutant.
In addition, literature previously gathered by the British Columbia Ministry of Water, Land and Air Protection was provided to us. Reference lists of all relevant articles including review articles were used as a source of additional citations.
Literature was sought in relation to the following issues: a) levels of PCDD/F in municipal sewage sludge; b) background levels of PCDD/F in soil; c) levels of PCDD/F in soil after sewage sludge application; d) transfer of PCDD/F from soil to plant tissue; e) transfer of PCDD/F from soil or feed to tissue of grazing animals.
Inclusion and Exclusion Criteria
All articles identified by the search were reviewed for relevance using the title and/or abstract. Articles were considered relevant if they reported PCDD/F concentrations in the following sample types: sludge from sewage or wastewater treatment plants handling municipal wastes; agricultural soil with historical or experimental treatment with sewage sludge; agricultural soil with no previous application of sewage sludge or experimental contamination with PCDD/Fs; food or forage plants grown in sludge-amended soil or soil treated experimentally with PCDD/Fs; tissue or milk of animals fed food grown in sludge-amended soil or food otherwise contaminated with PCDD/F; tissue of animals grazing on sludge-amended soil; or plant food, forage crops, animal tissue, or milk not believed to be contaminated from a specific PCDD/F source, that is, background concentrations in these types of food.
The following types of publications were excluded from further review: those that were not peer reviewed; those that reported about sites of industrial accidents (e.g., Seveso, Italy), nonmunicipal sources of sludge (e.g., industrial waste, pulp mill sludge), or plants grown by soil-free methods (e.g., hydroponics); studies conducted before 1980 when the limits of analytical chemical methods were insufficient to detect low PCDD/F concentrations; or studies that used nonstandard analytical methods (e.g., bioassays to determine dioxin-like activity).
Sixty-five papers met the above criteria.
Data Treatment and Analysis
All PCDD/F concentrations were converted to equivalent units using the international toxicity equivalency system (U.S. EPA 1999).
To examine the relative uptake of PCDD/Fs from soil to different plant and animal tissues, simple linear regressions were conducted to estimate the relationships between soil or feed PCDD/F toxic equivalents (TEQ) concentration (independent variable) and plant or animal tissue concentration (dependent variable) for each tissue type with a minimum of five data points. The resulting regression coefficients and standard errors were used to predict potential tissue PCDD/F concentrations (in TEQ) over the range of soil PCDD/F concentrations observed in agricultural settings where sewage sludge had been applied to the land. All analyses were performed using JMP statistical analysis software, version 3.2 (SAS Institute, Cary, NC).
Results
Sewage Sludge Contamination by PCDD/F
In municipal sewage sludge, levels of PCDD/F ranged from 0.0005 to 8,300 pg TEQ/g (Table 1).
Soil Contamination by PCDD/F
Background levels of PCDD/F in untreated soils ranged from 0.003 to 186 pg TEQ/g (Table 2). In studies of soil after sludge application, concentrations of PCDD/F ranged from 1.4 to 15 pg TEQ/g (Table 2). Although this range is very similar to the range of background values in untreated soils, all studies that measured soil PCDD/F concentrations before and after sludge application found increased contamination after sludge amendment (Figure 1). PCDD/F concentrations increased by factors of 1.4 to 17.0 (mean 7.1) after sludge application, indicating that application of sewage sludge increases PCDD/F contamination in soil.
Crop Contamination by PCDD/F
Table 3 is a list of the levels of PCDD/F in root crops, including carrots, potatoes, and beets. Mean levels in crops grown in uncontaminated soil or soil with low levels of PCDD/F ranged from below detection limits (<0.01) to 0.6 pg TEQ/g dry weight (dw).
Root vegetables grown either in naturally contaminated soil or soil to which PCDD/F had been added for experimental purposes had concentrations ranging from below detection limits (detection limit not stated) (Prinz et al. 1991) to 6,488 pg TEQ/g (dw) (Table 3). All experimental studies that examined root uptake of PCDD/F used soils that were much more highly contaminated than sludge-amended agricultural land. PCDD/F concentrations in experimentally contaminated soil ranged from 56 to 112,800 pg TEQ/g soil, whereas the highest level found in treated agricultural soil was 49 pg TEQ/g soil.
Table 4 indicates the levels of PCDD/F in crops with edible parts grown above the ground, including lettuce, silver beet, peas, and zucchini. The concentrations of PCDD/F in the aboveground parts of crops grown in soil with low levels of PCDD/F contamination ranged from < 0.01 to 10.2 pg TEQ/g (dw) (Table 4).
When grown in more highly contaminated soil, aboveground plants, including lettuce, silver beet, peas, zucchini, pumpkin, kale, chives, endive, leeks, beans, kohlrabi, and savoy, had PCDD/F concentrations ranging from 0.04 to 55.2 pg TEQ/g (dw) (Table 4). Tree fruits such as plums, strawberries, and apples had PCDD/F concentrations ranging from 0.8 to 1.4 pg TEQ/g (dw) when grown in soil containing 670 pg TEQ/g PCDD/F. Apples and pears grown in soil containing from 48 to 1,950 pg TEQ/g (dw) PCDD/F contained from 8 to 142 pg TEQ/g fresh weight (fw) PCDD/F (Table 5).
Measured concentrations of grasses and hay grown in soil with low levels of dioxin and furan contamination were all ≤ 1 pg TEQ/g (Table 6).
The contamination levels found in grass and hay grown in contaminated soil were generally higher (0.1–39 pg TEQ/g) (Table 6). Of the two studies that examined PCDD/F contamination of forage grown in contaminated soil, one did not state whether the plants were washed before analysis (Prinz et al. 1991), and the other used sand or clay pebbles on the soil surface to prevent soil–leaf contact (Hulster and Marschner 1993).
Relationships between PCDD/F in Soil and Crops
Tables 3–7 and Figure 2 show the relationship between PCDD/F concentrations in soil and resulting concentrations in crop tissues. The contaminant levels in whole carrot and potato showed weak positive relationships with the contaminant level of the soil. The concentration in peeled potatoes, however, did not change over a wide range of soil concentrations. This suggests that most of the PCDD/F contamination in potatoes accumulates in the peel.
A positive relationship was found between some members of the cucumber (Cucurbitaceae) family (namely zucchini, pumpkin, and cucumber) and soil contamination levels. Concentration of PCDD/F in green leafy vegetables also showed a positive (though weaker) relationship with soil concentration. Among aboveground crops, the weakest positive relationship was present between soil PCDD/F concentrations and contamination of tree fruits such as apples and pears. The data were insufficient to estimate the relationship between soil and plant concentrations of PCDD/F in peas and beans. Weak positive relationships were observed between soil and plant concentrations of hay and herbs. No positive relationship was observed between concentrations of PCDD/F in soil and grass (Figure 2; Table 7).
Animal Food Contamination by PCDD/F
Background contamination of beef ranged from less than the detection limit to 30.8 pg TEQ/g fat (Table 8); all mean values were < 5 pg/g. Dairy products were contaminated in the range of 0.3–1.4 pg TEQ/g fat (Table 8). Unfortunately, the contamination level of the feed eaten by the animals tested in these studies is not known.
Tissue concentrations from cattle consuming feed contaminated with PCDD/Fs ranged from 0.6 to 130 pg TEQ/g, in such tissues as fat, liver, kidney, muscle, and plasma (Table 8). Cattle were fed food with an extremely wide range of PCDD/F concentrations, ranging from those typically expected from forage crops (e.g., 2–3 pg/g) to extremely high levels (equivalent to thousands of picograms per gram) higher than the levels observed in sludge. For example, Jones et al. (1989) fed cattle 0.05 μg 2,3,7,8-tetra-chlorodibenzo-p-dioxin (TCDD)/kg body weight, which corresponds to a dose of 24.4 × 106 to 32.5 × 106 pg. Based on an estimated daily dry feed intake of 8 kg for beef cattle (Jones and Sewart 1997), this dose represents a feed contamination level of approximately 3,050–4,063 pg (mean 3,557) TEQ/g (dw). In those studies that used feed grown on sludge-amended land (Jilg et al. 1992; McLachlan et al. 1990, 1994; McLachlan and Richter 1998; Richter and McLachlan 2001), it was not stated whether the plants were washed or otherwise treated to remove soil or sludge particles before analysis and feeding. In practice, it is highly unlikely that grass, hay, or other forage would be washed before feeding to animals.
One of the great difficulties facing those studying animal uptake and contamination is the long duration required for animals to reach steady-state body burdens. The elimination half-life of PCDD in lactating cows is estimated to be in the range of 50–76 days (Firestone et al. 1979; Tuinstra et al. 1992), although one study based on a large single dose of 2,3,7,8-TCDD found that most was excreted in the milk within 14 days (Jones et al. 1989). The biological half-life of PCDD/F in cattle has been estimated to be somewhat longer, on the order of 93–148 days (Jensen et al. 1981; Thorpe et al. 2001), based on two experiments in which the animals were fed for 28 days and 18 weeks. Furthermore, McLachlan et al. (1994) found higher PCDD/F concentrations in cows that had calved several times than in those that had calved only once, suggesting that steady state had not been achieved in the younger cows. The exposure time in most of the feeding studies found in the literature search ranged from a single dose to 19 weeks. Given that it takes about five biological half-lives to reach steady state, the estimated minimum time to reach steady state would be 250 days in lactating animals and 465 days for nonlactating animals. None of the feeding studies were of sufficient duration.
Concentrations of milk from cows consuming PCDD/F-contaminated feed ranged from 0.031 to 3.0 pg TEQ/g (Table 8). Cows were fed food with PCDD/F concentrations typically expected from forage crops (e.g., 0.3–3 pg/g). As in the animal tissue studies, none of the studies was of sufficient duration for the body burden to reach steady state, although because of the shorter PCDD/F half-life in lactating animals and a minimum feeding duration of 17 days, the milk studies were generally more realistic. It should be noted that in most of these studies, milk was sampled while contaminated feed was still being consumed (Fries et al. 1999; Jilg et al. 1992; McLachlan et al. 1990, 1994) or within a week after the contaminated feeding ceased (Jilg et al. 1992; McLachlan and Richter 1998).
Among those who studied PCDD/F levels in milk with differing levels of soil or feed contamination, two reported little or no effect (Furst et al. 1993; McLachlan and Richter 1998), although the latter study did observe a slight increase in whole milk PCDD/F concentrations from 0.015 pg TEQ/g before the intervention to 0.049 pg/g after 23 days of consuming feed contaminated with 3.2 pg TEQ/g. Fries et al. (1999) found a 17-fold increase in dioxin and furan contamination of milk fat after pentachlorophenol-treated wood (contaminated with PCDD/F) was added to the cow’s diet for 58 days. McLachlan et al. (1994) found that the application of sewage sludge as fertilizer for harvested feed can increase the PCDD/F concentration in milk under certain circumstances, that is, in cows with a low level of milk production or in cows lactating after their first calving.
Relationships between PCDD/F in Feed and Animal Tissues
Table 8 and Figure 3 show the relationship between PCDD/F concentrations in feed and resulting concentrations in animal tissues. Because all results were reported per gram of lipid and there was no consistent pattern by tissue type, (i.e., muscle, fat, plasma, kidney, liver), all values were included in a single regression curve. The contaminant levels in beef tissue showed a strong positive relationship with the contaminant level in the feed.
No clear pattern was observed in the data from five studies examining the relationships between contamination of feed or grazing land and milk contamination from cows (Fries et al. 1999; Jilg et al. 1992; McLachlan et al. 1990, 1994; McLachlan and Richter 1998).
Discussion
Sewage Sludge and Soil
Soils treated with sewage sludge had relatively low levels of contamination when compared with those of the sludge itself. It is important to note, however, that in every case, the concentration of PCDD/F in the soil increased measurably after sludge application (Eljarrat et al. 1997; McLachlan and Reissinger 1990; McLachlan et al. 1996b; Molina et al. 2000; Wilson et al. 1997) (Figure 1). The elevated concentration of PCDD/F in sludge-amended soil also persisted over time. Most of the studies (Eljarrat et al. 1997; Molina et al. 2000; Wilson et al. 1997) measured PCDD/F concentrations up to 1 year after application of sewage sludge. One study that measured contamination on reclaimed quarry soil found elevated PCDD/F concentrations 4 years after a single treatment with sludge (Molina et al. 2000). Another study using archived soil samples from land that received a single sludge application in 1968 found that 59% of the PCDD/F contamination detected in 1972 was still present 18 years later (McLachlan et al. 1996b). McLachlan and Reissinger (1990) compared fields with 10–30 years of regular sludge treatments (application rate not known) with an untreated field on the same farm and found higher PCDD/F concentrations in the treated fields. Only one other study examined the effect of multiple sludge treatments (Eljarrat et al. 1997); after four annual treatments, the authors reported soil contamination levels no higher than those reported in other studies of single sludge treatments. In another study that compared the effects of plowing sewage sludge into the soil with surface application on meadowland, the authors found that elevated PCDD/F concentrations persisted for at least 260 days after application of sewage sludge and appeared to be slightly more persistent when plowed into the soil (Wilson et al. 1997). The half-life of PCDD/F in soil is estimated to be at least 10 years (Jackson and Eduljee 1994; Rappe et al. 1999).
Plant Foods
Studies that examined the uptake of PCDD/F by plants growing in contaminated soils used either field soils that were highly contaminated because of proximity to heavy industry or experimentally contaminated soils with extremely high levels of PCDD/F. The PCDD/F concentrations in the soils used as controls in these studies are closer to if slightly lower than the concentrations found in sludge-amended agricultural soils. Furthermore, differences in soil properties, such as organic matter content, between contaminated and sludge-amended soils may affect plant uptake.
In our estimates of the relationships between soil and plant concentrations, the slopes of the regression lines were very shallow, suggesting that large increases in soil contamination would be required for small increases in plant contamination (Table 7, Figure 2). The regression coefficients and standard errors were used to estimate mean PCDD/F contamination levels in crops grown in soil with contamination levels in the range found for sludge-amended soils. These estimates indicate that very little change in plant contamination is expected over the probable soil contamination range of 1–30 pg TEQ/g soil. Even at an extremely high estimate for soil concentration, one that assumes a concentration equivalent to that of the highest sludge concentration reported, the predicted increases in plant concentrations were only moderately elevated. It is important to note, however, that the predicted plant values at the lower soil contamination levels have been back-extrapolated, as no empirical data are available at these lower soil concentrations. This adds uncertainty to the estimates.
Interpretation of the coefficients listed in Table 7 must take into account that they are based on relatively few data points, from only one or a few studies. Taken together, they suggest that for most plants, large increases in soil contamination (200–10,000 pg TEQ/g; namely, much higher than the increases expected from sewage sludge treatment) are required to produce small increases (1 pg TEQ/g) in plant contamination. They also suggest that plants in the family Cucurbitaceae (pumpkin, zucchini, cucumber) show a sufficiently strong association between soil PCDD/F levels and plant contamination that application of sewage sludge may increase the contamination levels of the plants.
The data suggest that different plants have different potentials for uptake of PCDD/Fs based on the different coefficients for the relationships between soil contamination levels and plant concentrations. All studies that examined the uptake of PCDD/Fs from soil by carrots and by certain members of the cucumber family found that these plants take up more PCDD/Fs from the soil than do other plants. In a study comparing different members of the family Cucurbitaceae (Hulster et al. 1994) grown in contaminated soil (148 pg TEQ/g soil), zucchini fruits and the outer layer of pumpkin (genus Cucurbita) had much higher levels of PCDD/F contamination [20.0 and 11.8 pg TEQ/g (dw), respectively] than did cucumber (genus Cucumis) [2.35 pg TEQ/g (dw)]. In a study that compared the ability of root exudates to absorb PCDD/F from soil (Hulster and Marschner 1994), zucchini root exudates absorbed 4 times more PCDD/F than tomato root exudates.
In a study that measured PCDD/F uptake by carrots grown in contaminated soil (Muller et al. 1994), more than 75% of the contamination was concentrated in the peel [mean concentration, 3 pg TEQ/g (dw)]. The inner parts of the carrot had PCDD/F concentrations more comparable to other plants [mean cortex concentration, 0.29 pg TEQ/g (dw); mean stele concentration, 0.40 pg TEQ/g (dw)]. When the congener profiles were compared, although the control (uncontaminated) soil had primarily octachlorodibenzo-p-dioxin (OCDD) and the contaminated soil had mostly higher chlorinated furans, the carrots from either soil contained mostly lower chlorinated furans. The lower-chlorinated PCDD/F congeners tend to be more bioavailable in lipid environments (Muller et al. 1993), which declines from the outer to inner parts of the carrot root.
Although the published empirical data for any one crop are very limited, the collective body of work indicates that high levels of PCDD/F in soil are associated with increased contamination of plant crops. However, at the soil contamination levels expected from treatment with sewage sludge, it appears that there would be minimal or no increase in the dioxin and furan content of most food crops.
To date, there is no evidence related to the potential for increased dioxin and furan contamination of other root vegetables (e.g., beets, parsnips, turnips, sweet potatoes, ginger, garlic, onions) or aboveground plant foods (e.g., cruciferous vegetables, berries, tomatoes, corn, peppers, grains).
Forage Crops
Studies that have examined the uptake of PCDD/Fs by forage crops, such as the studies on other plant foods, used soils with extremely high levels of PCDD/F. Within this wide range of soil contamination levels, weak positive relationships were seen between soil and hay or herb concentrations of PCDD/F, but not between soil and grass concentrations. Potential contamination levels of hay and herbs grown on sludge-amended land were estimated using the regression coefficients (Table 7). Over the soil contamination range of 1–1,250 pg TEQ/g soil, there is virtually no change in predicted crop contamination levels.
Although the evidence for forage crops appears consistent with that of other plants with edible parts grown aboveground, there are outstanding issues relating to adherence of soil particles to the plants. In one study that measured the soil content of freshly cut forage from a pasture, the soil content ranged from approximately 1 to 46% of the dry weight of the plant, depending on the time of year. In winter the soil content was consistently greater than 23% of plant dry weight (Beresford and Howard 1991). Two other studies that measured the soil content of harvested cattle feed found that soil contributed less than 1% of the dry weight of the feed (Fries et al. 1981; Zach and Mayoh 1984). It is reasonable to assume that forage is not washed before feeding animals under normal conditions. However, many of the plant crop studies and one of the two studies of forage crops used experimental methods that either protected the leaves from contact with soil or washed it away after harvesting. Thus, the contribution of contaminated soil to harvested forage crop PCDD/F contamination may not have been adequately assessed by the studies to date. More evidence is needed to evaluate this potentially important contributor to animal uptake of PCDD/Fs.
Animal Foods
The results of this review indicate that consumption of contaminated feed or grazing of cattle on treated land is likely to increase the PCDD/F levels in meat products. Unlike the plant studies, most of the studies examining the impact of PCDD/F contamination on animal tissue used feed contaminated at levels low enough that they might be encountered in practice.
The relationship between feed contamination levels and concentrations in the fatty tissue of cattle (Figure 2, Table 7) is considerably stronger than that for plant tissues, with a coefficient two to three orders of magnitude higher than for most plants and one order higher than for the family Cucurbitaceae. The coefficient of the relationship is greater than 1, suggesting bioaccumulation. As an example, the PCDD/F concentration in beef tissue may increase by up to 10 pg TEQ/g fat at the relatively low contamination level of 5 pg TEQ/g in feed (Table 7). This suggests that the use of dioxin/furan-contaminated sewage sludge on grazing land or on land used to grow cattle feed may result in increased human exposure to PCDD/Fs through the diet, especially if the sludge is highly contaminated.
There were insufficient data to conclude whether consumption of feed grown on land treated with sewage sludge or grazing of animals on sludge-amended land is likely to increase the PCDD/F levels in milk products. Few studies examined the relationships between contamination of feed or grazing land and milk contamination from cows (Fries et al. 1999; Jilg et al. 1992; McLachlan et al. 1990, 1994; McLachlan and Richter 1998), and no clear relationship could be seen in the data. Overall, the studies that examined the relationship between feed or soil PCDD/F concentration and milk concentration show that PCDD/Fs are excreted in milk. The amount excreted appears to be dependent on the timing of PCDD/F contamination in the diet (Jilg et al. 1992; Jones et al. 1989). There may be only a minimal impact of sewage sludge use on milk, especially if a sufficient time lag is provided between sludge application and milking for human consumption. However, the data are still very limited.
The application of sewage sludge to grazing or forage land presents additional exposure risk to animals beyond that resulting from direct uptake of PCDD/Fs by the crops. Animals consume soil along with fodder, either by eating the soil directly while grazing or by consuming plants (e.g., grass, hay, or beetroot) to which soil has adhered (McLachlan et al. 1996a; Zach and Mayoh 1984). As a result, they may directly ingest sludge that has been applied to pastureland. Although estimates vary, cattle, sheep, and swine may consume an average of 6–7% (up to 18% during seasons of sparse forage) of their ingested dry matter as soil (Fries 1996; Pohl et al. 1995). Studies from the Netherlands and the United States, where grazing is seasonal and cattle are given plenty of supplemental feed, suggest that cows may ingest an average of 150–300 g of soil per day (1–2% of their dry matter intake) (McLachlan et al. 1996a). At a worst-case estimate of 30 pg TEQ/g soil, this would correspond to an additional intake of up to 9 ng PCDD/F per cow per day. Based on an analysis of studies from New Zealand, the United Kingdom, and the United States, Fries (1996) estimated that a 500-kg dairy cow would ingest 900 g of soil per day. With a PCDD/F concentration of 30 pg TEQ/g soil, this would contribute 27 ng PCDD/F per cow per day.
Limitations
One of the primary limitations of this review is the small number of studies relevant to the subject at hand. All the data related to plant foods were taken from only six articles, and the variety of plant species represented is quite small relative to the number of food crops that could potentially be exposed to recycled sewage sludge. No studies were identified that measured PCDD/Fs in animals other than cattle fed from sludge-amended land. Although there were eight articles reporting background concentrations of PCDD/F in animal tissue, the level of PCDD/F contamination in the feed or grazing land of these animals was not reported.
There were no field-based plant studies and few animal uptake studies that examined the effects of real sludge application practices. This is especially important with respect to harvested forage crops, for which the contribution of soil adherence is not known.
Many studies did not describe the details of the analytical methods used (including limits of detection) or state whether crop samples were washed before analysis. Field practices such as sludge application rate, application method, PCDD/F concentration, and fertilization/harvesting time may influence the uptake of PCDD/F. Unfortunately, such factors could not be considered in this review because the information was not usually reported in the published studies. Furthermore, although the TEQ system is useful when comparing samples with differing congener profiles, it is somewhat limited in that any differences in uptake or behavior of individual congeners is not taken into account.
Gaps in the Published Research
Although there is some empirical evidence to suggest that there is an impact of sewage sludge application on PCDD/F uptake by grazing animals but minimal uptake from sludge to plants, there are a number of significant gaps in the data. Controlled field studies are needed that include variables such as application rate, timing, and method and that assess crops and animals exposed under realistic conditions. Repeat studies must be conducted to determine the reliability of the data, and more species need to be assessed. It is essential that the complex issue of additional animal exposure to sewage sludge through soil consumption or adherence to forage crops be examined. Information is also needed on the effects on animals other than cows, for example, swine and poultry.
Conclusions
The results reported here, based on published empirical data, were compared with the results of studies that used pathway modeling to predict the effect of land application of sewage sludge on PCDD/F contamination in food and were similar. Investigators using models have concluded that a) sewage sludge application may lead to slight increases in PCDD/F concentration in the peel of root crops (Duarte-Davidson and Jones 1996; Jackson and Eduljee 1994; Wild and Jones 1992) or in members of the Cucurbitaceae family (Jones and Sewart 1997), but would have a negligible impact on other aboveground plants (Duarte-Davidson and Jones 1996; Jones and Sewart 1997; Rappe et al. 1999; Wild and Jones 1992; and that b) sewage sludge application on grazing or forage land could significantly increase human dietary exposure to PCDD/F (Duarte-Davidson and Jones 1996; Jackson and Eduljee 1994; Jones and Sewart 1997; McLachlan et al. 1996a; Rappe et al. 1999; Wild and Jones 1992; Wild et al. 1994). A recent human health risk assessment (U.S. EPA 2004) found that land application of sewage sludge would lead to a negligible increase in cancer cases even among the most highly exposed groups. Noncancer health risks were not assessed. Our review examined the potential for increased human foodborne exposure rather than potential health outcomes.
In conclusion, the available empirical evidence indicates that application of sewage sludge to agricultural land may have a small impact on the levels of PCDD/F found in root vegetables, aboveground plant foods, and forage crops. The impact in animal tissues is likely to be considerably greater. Therefore, before sludge application, careful consideration should be given to the types of agricultural products grown. Minimizing the PCDD/F content would also reduce human exposure potential in land application of sewage sludge.
Figure 1 Change in concentration of PCDD/F in soil after sludge application. The numbers above the bars indicate the factor by which the soil PCDD/F concentration increased after application of sewage sludge.
Figure 2 Relationship between PCDD/F concentrations in plant foods and soil contamination levels. The plant data include data from Tables 3, 4, and 6 that relate to those plants for which relationships could be found between plant and soil PCDD/F concentrations. The following data were omitted: a) measurements in which the soil PCDD/F concentration was much higher (8- and 20-fold) (Hulster and Marschner 1993) than in the other samples and not remotely relevant to the soil concentrations likely to result from sewage sludge application; and b) the result of a study that did not use natural growing conditions (plants growing in pots of uncontaminated soil placed in or on top of contaminated soil (Hulster et al. 1994). Data were taken from the following sources: potato: Prinz et al. (1991), Hulster and Marschner (1993); carrot: Prinz et al. (1991), Schroll and Scheunert (1993), Muller et al. (1994); leafy vegetable: Prinz et al. (1991), Hulster and Marschner (1993), Muller et al. (1994); Cucurbitaceae: Prinz et al. (1991), Hulster et al. (1994); hay: Hulster and Marschner (1993).
Figure 3 Projected increases in PCDD/F concentrations in plant foods and beef per unit increase in soil or feed contamination levels. CL, confidence limit.The data are derived from the regression curves for plant and animal foods shown in Table 7. This figure illustrates the increases in PCDD/F concentrations in beef fed feed or forage contaminated with PCDD/F and demonstrates how much more pronounced this effect is in beef than in the plant foods grown in sludge-treated soils. The regression curve for beef includes all values from Table 8 relating to concentration of PCDD/F in beef tissue (not milk) that provided the feed PCDD/F level (Jensen et al. 1981; Jilg et al. 1992; Richter and McLachlan 2001; Thorpe et al. 2001) except one study that used an experimental dose 87 times higher than in the other studies (Jones et al. 1989).
Table 1 Concentrations of PCDD/F in sewage sludge, sorted by country and year.
Reference Country Year Source of material n Mean concentration (pg TEQ/g) Range (pg TEQ/g)
Ho and Clement 1990 Canada 1986 Treated municipal sludge 50 NA 0.0005–0.0015
Raw municipal sludge 50 NA 0.0026–0.0051
van Oostdam and Ward 1995 Canada 1990–1993 Primary sludge 4 16.6 (dw) 2.3–49.6
Healey and Bright 2000 Canada 1998–1999 Municipal wastewater treatment plants 26 40 (dw) 5.6–250
Lamparski et al. 1984 USA 1933 Treated municipal sludge 1 87.7 (dw)
1981 Treated municipal sludge 1 88.9 (dw)
1982 Treated municipal sludge 1 80.8 (dw)
Telliard et al. 1990 USA 1988–1989 Public-owned sewage treatment works 211 38.38 (ww) 0.039–1252.9
Malloy et al. 1993 USA 1990–1992 Municipal yard waste compost 11 29.6 5–91
Municipal solid waste compost 6 46.5 19–96
Municipal solid waste + dewatered sewage sludge compost 4 56 37–87
Wilson et al. 1997 U.K. NA Anaerobically digested sewage sludge 1 19 (dw)
McLachlan et al. 1996b U.K. 1968 Rural uncontaminated sewage sludge 2 230 (dw) 200–280
Sewart et al. 1995 U.K. 1992 Digested sludges from sewage treatment plants 8 72 (dw) 19–206
1942–1960 Archived samples from 1942 to 1960 7 148 (dw) 18–402
Rappe et al. 1989 Sweden NA Urban sludge 1 23.9
Rural sludge 1 23.1
Naf and Broman 1990 Sweden May–Aug 1989 Anaerobically digested sludge from urban wastewater treatment plant 1 31 (dw)
Broman et al. 1990 Sweden May–Aug. 1989 Digested and dewatered sludge 4 79 (ow) 41–130
Grossi et al. 1998 Brazil 1990–? Municipal solid waste compost from the following:
Urban 11 57 (dw) 11–150
Small cities 5 27 (dw) 3–163
Coastal sandy 3 8 (dw) 5–11
New, some industrial waste 2 54 (dw) 10–99
Disse et al. 1995 Germany NA Undigested sludge from rural area 1 9 (dw)
Undigested sludge from municipal area with no heavy industry 1 20 (dw)
Undigested sludge from municipal area with metal industry 1 200 (dw)
McLachlan and Reissinger 1990 Germany NA Local wastewater treatment plant 1 42 (dw)
Horstmann et al. 1992 Germany 1991 Anaerobically digested sewage sludge 1 48 (dw)
Primary sludge (dry conditions) 9 31.4 (dw) 15–64
Primary sludge (rainy conditions) 2 28.5 (dw) 21–36
Eljarrat et al. 1999 Spain 1994–1998 Sludges from rural, urban, and industrial wastewater treatment plants 19 55 (dw) 7–160
1979–1987 Archived samples from 1979 to 1987 24 620 (11.3-fold increase) 29–8,300
Molina et al. 2000 Spain NA Aerobic sewage treatment plant 1 68.1 (dw)
Eljarrat et al. 1997 Spain 1986, 1987 Sludge from urban wastewater treatment plants (aerobic digestion) 7 144 (dw) 74–260
Abbreviations: dw, dry weight; NA, no data available; ow, organic weight; ww , wet weight.
Table 2 Concentrations of PCDD/F in soil (background and sludge amended), sorted by year of publication.
Reference Country Year Source of material Sludge concentration (pg TEQ/g) n Mean concentration (pg TEQ/g) Range (pg TEQ/g)
Creaser et al. 1989 U.K. Soil at intersection points of a 50-km grid NA 77 23.4 (dw) 1.2–161.9
Broman et al. 1990 Sweden 1989 Agricultural land near major roads NA 4 29 (ow) 13–49
Agricultural land not near major roads NA 4 17 (ow) 9–32
McLachlan and Reissinger 1990 Germany Farmland NA 1 0.84 (dw)
Farmland 42 (dw) 2 6.55 (dw) 3.7–9.4
Meadow 42 (dw) 1 15 (dw)
Kjeller et al. 1991 U.K. 1986 Semirural experimental plots NA 3 1.4 (dw)
Sund et al. 1993 Australia 1990 Soil from urban and industrial areas NA 7 2.3 0.09–8.2
van Oostdam and Ward 1995 Canada 1990–1993 Background soil NA 53 5.0 (dw) ND–57
McLachlan et al. 1996b U.K. 1968, 1972, 1976, 1981, 1985, 1990 Experimental agricultural land NA 6 1.3 (dw) 0.88–2.0
Sludge applied experimentally in 1968 230 (dw) 5 8.8 (dw) 6.5–13
Eljarrat et al. 1997 Spain 1986–1987 Acidic and basic agricultural soil NA 2 1.7 (dw) 0.3–3.1
Urban wastewater treatment plants (aerobic digestion) 144 (dw) 4 4.6 (dw) 2.4–8.6
Wilson et al. 1997 U.K. Plowed plot NA 4 2.0 (dw) 1.8–2.2
Pasture plot NA 4 1.9 (dw) 1.7–2.0
Plowed plot (15–20 cm) 19 4 2.7 (dw) 2.4–3.0
Pasture plot (surface application) 19 4 2.8 (dw) 1.6–4.3
Molina et al. 2000 Spain Alkaline soil NA 2 0.37 (dw) 0.34–0.39
7.5% sludge (time 0) 68.1 (dw) 1 2.43 (dw)
7.5% sludge (1 year) 68.1 (dw) 1 2.37 (dw)
15% sludge (time 0) 68.1 (dw) 1 5.28 (dw)
15% sludge (1 year) 68.1 (dw) 1 4.61 (dw)
Quarry NA 2 0.84 dw) 0.76–0.92
Direct application of 7.5% sludge (time 0) 68.1 (dw) 1 1.4 (dw)
Direct application of 7.5% sludge (4 years) 68.1 (dw) 1 12.1 (dw)
Soil–sludge mixture 7.5% (time 0) 68.1 (dw) 1 3.14 (dw)
Soil–sludge mixture 7.5% (4 years) 68.1 (dw) 1 4.24 (dw)
Direct application of 15% sludge (time 0) 68.1 (dw) 1 5.26 (dw)
Direct application of 15% sludge (4 years) 68.1 (dw) 1 8.50 (dw)
Soil–sludge mixture 15% (time 0) 68.1 (dw) 1 2.56 (dw)
Soil–sludge mixture 15% (4 years) 68.1 (dw) 1 4.24 (dw)
Abbreviations: dw, dry weight; NA, no data available; ND, not detected; ow, organic weight.
Table 3 PCDD/F concentrations in root vegetables, sorted by year of publication.
Reference Growing environment Source of PCDD/F Soil concentration (pg TEQ/g) Plant type (part) n Mean plant concentration (pg TEQ/g) (dw) Range of plant concentration (pg TEQ/g) (dw)
Prinz et al. 1991 Field conditions None 68 (dw) Potato (tuber) 2 ~ 0.5
Incinerator 274 (dw) Potato (tuber) 2 < LOD
Incinerator 670 (dw) Potato (tuber) 2 ~ 0.6
Incinerator 788 (dw) Potato (tuber) 2 ~ 0.3
None 68 (dw) Carrot (root) 2 ~ 0.6
Incinerator 274 (dw) Carrot (root) 2 ~ 0.6
Incinerator 670 (dw) Carrot (root) 2 ~ 2.8
Incinerator 788 (dw) Carrot (root) 2 ~ 2.0
Incinerator 670 (dw) Celery 2 ~ 0.4
Incinerator 788 (dw) Red beet (tuber) 2 ~ 0.4
Hulster and Marschner 1993 Field conditions None 4.8 Potato (unpeeled) NA ~ 0.2
Incinerator 328 Potato (unpeeled) NA ~ 0.6
845 Potato (unpeeled) NA ~ 1.2
2,390 Potato (unpeeled) NA ~ 1.6
None 4.8 Potato (peeled) NA ~ 0.1
Incinerator 328 Potato (peeled) NA ~ 0.1
845 Potato (peeled) NA ~ 0.1
2,390 Potato (peeled) NA ~ 0.1
Schroll and Scheunert 1993 Closed system None 0 Carrots (roots) 2 < LOD
Growing chamber OCDD added to soil 6,400 (dw) Carrots (roots) 2 4,811.1 397.8 (fw) 3134.3–6488.5 259.1–536.4 (fw)
Muller et al. 1994 Field conditions None 5 (dw) Carrots (peel) 1 0.55
Incinerator 56 (dw) Carrots (peel) 2 3.08 2.86–3.3
None 5 (dw) Carrots (cortex) 1 0.27
Incinerator 56 (dw) Carrots (cortex) 2 0.29 0.28–0.3
None 5 (dw) Carrots (stele) 1 0.32
Incinerator 56 (dw) Carrots (stele) 2 0.395 0.29–0.5
None 5 (dw) Carrots (whole) 1 0.35
Incinerator 56 (dw) Carrots (whole) 2 0.96 0.87–1.05
Abbreviations: dw, dry weight; fw, fresh weight; LOD, limit of detection; NA, no data available.
Table 4 PCDD/F concentrations in crops with edible parts grown aboveground, sorted by year of publication.
Reference Growing environment Source of PCDD/F Soil concentration (pg TEQ/g) Plant type (part) n Mean plant concentration (pg TEQ/g dw) Range of plant concentration (pg TEQ/g dw)
Prinz et al. 1991 Field conditions None 68 (dw) Salad 2 ~ 0.4
Incinerator 200 (dw) Salad 2 ~ 3.2
Incinerator 274 (dw) Salad 2 ~ 4.3
Incinerator 670 (dw) Salad 2 ~ 9.2
Incinerator 788 (dw) Salad 2 ~ 6.6
None 68 (dw) Silver beet 2 ~ 0.3
Incinerator 25 (dw) Silver beet 2 ~ 3.5
Incinerator 670 (dw) Silver beet 2 ~ 9.8
Incinerator 788 (dw) Silver beet 2 ~ 7.0
Incinerator 199 (dw) Kale 2 ~ 7.3
Incinerator 200 (dw) Kale 2 ~ 6.6
Incinerator 274 (dw) Kale 2 ~ 6.3
Incinerator 788 (dw) Kale 2 ~ 2.0
Incinerator 274 (dw) Endive 2 ~ 2.5
Incinerator 788 (dw) Endive 2 ~ 17.8
Incinerator 670 (dw) Leek 2 ~ 1.6
Incinerator 670 (dw) Cucumber 2 ~ 0.8
Incinerator 670 (dw) Bean 2 ~ 0.6
Incinerator 788 (dw) Kohlrabi 2 ~ 0.3
Incinerator 788 (dw) Savoy 2 ~ 0.5
Hulster and Marschner 1993 Field conditions None 4.8 Lettuce leaves NA ~ 0.2
Incinerator 845 Lettuce leaves NA ~ 0.3
Incinerator 328 Lettuce leaves NA ~ 1.3
None 4.8 Lettuce (whole) NA ~ 0.2
Incinerator 845 Lettuce (whole) NA ~ 0.4
Incinerator 328 Lettuce (whole) NA ~ 1.4
Schroll and Scheunert 1993 Closed system Treated soil 6,400 (dw) Carrots (stem) 2 2306.2 2029.4–2582.9
Hulster et al. 1994 Field conditions None 0.4 (dw) Zucchini (fruit) 2 1.0 0.9–1.1
0.4 (dw) Zucchini (fruit) 2 0.6 0.5–0.7
Chlorine–alkaline– 148 (dw) Zucchini (fruit) 2 20.0 19.1–21.0
electrolysis residues 148 (dw) Zucchini (no soil–fruit contact) 2 20.5 19.4–21.6
328 (dw) Zucchini (fruit) 2 17.2 17.0–17.4
2,390 (dw) Zucchini (fruit) 2 54.9 54.6–55.2
Chlorine–alkaline– 148 (dw) Pumpkin (outer fruit) 2 11.8 11.6–12.0
electrolysis residues 148 (dw) Pumpkin (inner fruit) 2 3.25 3.1–3.4
148 (dw) Cucumber (outer fruit) 2 2.35 2.3–2.4
148 (dw) Cucumber (inner fruit) 2 0.2 0.2–0.2
Muller et al. 1994 Field conditions None 5 (dw) Peas (pods) 1 0.13
Incinerator 56 (dw) Peas (pods) 1 0.12
None 5 (dw) Peas (seeds) 1 < 0.01
Incinerator 56 (dw) Peas (seeds) 1 0.04
None 5 (dw) Peas (whole) 1 0.08
Incinerator 56 (dw) Peas (whole) 1 0.09
None 5 (dw) Lettuce (outer leaves) 1 0.13
Incinerator 56 (dw) Lettuce (whole) 2 0.21 0.21–0.21
Abbreviations: dw, dry weight; NA, no data available.
Table 5 PCDD/F concentrations in tree fruits, sorted by year of publication.
Reference Growing environment Source of PCDD/F Soil concentration (pg TEQ/g) (dw) Plant type (part) n Mean plant concentration (pg TEQ/g) Range of plant concentration (pg TEQ/g)
Prinz et al. 1991 Field conditions Incinerator 670 Plum 2 ~ 1.1 (dw)
Strawberry 2 ~ 0.8 (dw)
Apple 2 ~ 1.4 (dw)
Muller et al. 1993 Field conditions Chlorine–alkaline–electrolysis residues 48 (subsoil) Pear 2 (washed, whole) 1 25 (fw)
14,530 (subsoil) Pear 1 (unprocessed, whole) 2 33 (fw) 20–46
Pear 1 (washed, peel) 2 123.5 (fw) 105–142
Pear 1 (washed, pulp) 2 15 (fw) 8–22
Pear 1 (washed, whole) 2 36 (fw) 27–45
Pear 1 (wrapped, whole) 2 14 (fw) 11–17
1,950 (subsoil) Apple (washed, pulp) 1 8 (fw)
Apple (washed, peel) 1 46 (fw)
Apple (washed, whole) 1 14 (fw)
Abbreviations: dw, dry weight; fw, fresh weight.
Table 6 PCDD/F concentrations in forage crops.
Reference Growing environment Source of PCDD/F Soil concentration (pg TEQ/g) Plant type (part) n Mean plant concentration (pg TEQ/g dw)
Hulster and Marschner 1993 Field conditions None 4.8 Hay NA ~ 1
Incinerator 328 Hay NA ~ 4
Incinerator 845 Hay NA ~ 3
Incinerator 2,390 Hay NA ~ 10
Incinerator 5,752 Hay NA ~ 6
None 4.8 Herbs (hay) NA < LOD
Incinerator 328 Herbs (hay) NA ~ 0.5
Incinerator 845 Herbs (hay) NA ~ 0.7
Incinerator 2,390 Herbs (hay) NA ~ 0.8
Incinerator 5,752 Herbs (hay) NA ~ 0.9
None 4.8 Grass (hay) NA < LOD
Incinerator 328 Grass (hay) NA ~ 0.1
Incinerator 845 Grass (hay) NA ~ 0.2
Incinerator 2,390 Grass (hay) NA ~ 0.1
Incinerator 5,752 Grass (hay) NA ~ 0.2
Abbreviations: dw, dry weight; LOD, limit of detection; NA, no data available.
Table 7 Mean projected increase in concentration of PCDD/F in food with a given increase in soil or feed concentration.
Food type Increase in soil or feed PCDD/F concentration (pg TEQ/g dw)a
1b 5 10 15 30
n Projected increase in food concentration (pg TEQ/g dw)
Herbs 5 0.0001 (0.00006)c 0.00 (0.00)d 0.00 (0.00)d 0.00 (0.00)d 0.00 (0.00)d
Potato tuber 9 0.0004* (0.000063) 0.00 (0.00) 0.00 (0.01) 0.01 (0.01) 0.01 (0.02)
Hay 5 0.0008 (0.000703) 0.00 (0.01) 0.00 (0.02) 0.00 (0.03) 0.00 (0.06)
Tree fruits (fw) 9 0.0016 (0.00185) 0.01 (0.02) 0.02 (0.05) 0.02 (0.07) 0.05 (0.15)
Carrot root 13 0.0027* (0.000608) 0.01 (0.01) 0.03 (0.02) 0.04 (0.05) 0.08 (0.11)
Leafy vegetables 26 0.0042 (0.00255) 0.01 (0.04) 0.03 (0.09) 0.06 (0.13) 0.12 (0.21)
Cucurbitaceae 9 0.019* (0.00503) 0.07 (0.12) 0.17 (0.26) 0.27 (0.41) 0.55 (0.84)
Animal tissue 18 1.458* (0.278)5.80 (8.00) 13.1 (18.0) 21.9 (28.0) 47.4 (58.0)
aSoil/feed concentration values are intended to represent the following potential scenarios: 0–1 pg TEQ/g represents the likely concentrations found in forage crops grown in soil with minimal background PCDD/F contamination (Hulster and Marschner 1993); 0.1–4 pg TEQ/g represents the likely concentrations found in forage grown in sludge-amended soil; 1–10 pg TEQ/g is the typical range in sludge-amended agricultural soil; and the concentrations found in forage grown in highly contaminated soil (> 670 pg TEQ/g) (Hulster and Marschner 1993; Prinz et al. 1991); 15 pg TEQ/g represents the maximum concentration reported in sludge-amended soil (McLachlan and Reissinger 1990); 30 pg TEQ/g represents the maximum mean concentration reported in soil (not sludge amended) (Broman et al. 1990).
bCoefficient of relationship between food concentration and soil or feed concentration.
cValues in parentheses are standard error of the coefficient.
dValues in parentheses are upper 95% confidence limits of the increase in food concentration.
*Regression coefficient significant at p < 0.05.
Table 8 Concentrations of PCDD/F in food from cattle, sorted by year of publication.
Reference Source of PCDD/F Feeding time Mean food concentration (pg TEQ/g) Tissue No. of animals Mean tissue concentration (pg TEQ/g fat) Range of tissue concentration (pg TEQ/g)
Jensen et al. 1981 Experimental 28 days 24 ± 5 Fat 7 84 66–95
Liver 7 8.2 7–10
Kidney 7 7 6–8
Muscle 7 2 2
Jones et al. 1989 Single oral dose in grain 1 dose ~ 3557 Fat 2 105 80–130
Single oral dose in soil 1 dose ~ 3557 Fat 2 155 130–180
McLachlan et al. 1990 None NA 6.9 Milk 1 1.39
Jilg et al. 1992 Hay grown in contaminated soil (1,944 pg TEQ/g dw) 19 weeks 2 (range 0.5–8.7) Plasma 4 1.95 0.8–4.1
Fat 4 1.1 0.6–2.8
Muscle 4 1.75 1.3–2.8
Milk (weeks 1–19) 4 1.88 0.8–3.0
Milk (weeks 20–28) 3 1.13 0.6–2.1
McLachlan et al. 1994 None 6 months 0.19 (dw) Milk 12 0.9
None 6 months 0.22 (dw) Milk 12 1.3
Silage from sludge- treated land 6 months 0.35 (dw) Milk 12 1.2
Silage from sludge- treated land 6 months 1.2 (dw) Milk 12 2.3
Schecter et al. 1994 None NA Beef 4 0.578 (ww) 0.04–1.5 (ww)
NA Dairy 5 0.348 (ww) 0.04–0.7 (ww)
Winters et al. 1996 None NA Back fat 63 0.35 (SE 0.08) < LOD–3.8
Fiedler et al. 1997 None NA Fat 3 0.67 ± 0.17 0.528–1.1
NA Dairy fat 9 0.77 ± 0.10 0.416–0.970
Feil and Ellis 1998 None NA Perirenal fat 20 4.1275 (ww) 0.3341–30.8373
McLachlan and Richter 1998 None 12 weeks 0.2 (dw) Milk (whole) 4 0.015 (whole milk) 0.010–0.02 (whole milk)
Silage from sludge- treated land 17 days 3.2 (dw) Milk (whole) 4 0.049 (day 23) 0.031–0.069 (day 23)
Fries et al. 1999 None NA Milk 4 0.315
PCP-treated wood 58 days 0.289 (dw) Milk 4 5.518
Richter and McLachlan 2001 None 10 weeks 0.2 (dw) Muscle 2 0.41 0.30–0.51
Fat 2 0.47 0.34–0.61
Liver 2 6.5 5.1–7.9
Kidney 2 0.50 0.41–0.58
Silage from sludge- treated land 17 days 3.2 (dw) Muscle 2 0.70 0.54–0.91
Fat 2 0.64 0.49–0.79
Liver 2 20.5 17.0–24.0
Kidney 2 0.74 0.61–0.86
Thorpe et al. 2001 None (testing at 31 weeks) 28 days NA Liver 4 3.9
Muscle 4 5.9
Fat 4 3.7
Prepared pellets (testing at 31 weeks) 28 days ~ 41.3 (330,000 pg TEQ/day) Liver 4 118.5
Muscle 4 57.3
Fat 4 27.2
Abbreviations: dw, dry weight; LOD, limit of detection; NA, no data available; PCP, pentachlorophenol.
==== Refs
References
Beresford NA Howard BJ 1991 The importance of soil adhered to vegetation as a source of radionuclides ingested by grazing animals Sci Total Environ 107 237 254 1785051
Broman D Naf C Rolff C Zebuhr Y 1990 Analysis of polychlorinated dibenzo-p -dioxins (PCDD) and polychlorinated dibenzofurans (PCDF) in soil and digested sludge from Stockholm, Sweden Chemosphere 21 1213 1220
Creaser CS Fernandes AR Al-Haddad A Harrad SJ Homer RB Skett PW 1989 Survey of background levels of PCDDs in UK soils Chemosphere 18 767 776
Disse G Weber H Haman R Haupt HJ 1995 Comparison of PCDD and PCDF concentrations after aerobic and anaerobic digestion of sewage sludge Chemosphere 3 3617 3625 8528650
Duarte-Davidson R Jones KC 1996 Screening of the environmental fate of organic contaminants in sewage sludge applied to agricultural soils. II. The potential for transfers to plants and grazing animals Sci Total Environ 185 59 70 8643960
Eljarrat E Caixach J Rivera J 1997 Effects of sewage sludges contaminated with polychlorinated dibenzo-p -dioxins, dibenzofurans, and biphenyls on agricultural soils Environ Sci Technol 31 2765 2771
Eljarrat E Caixach J Rivera J 1999 Decline in PCDD and PCDF levels in sewage sludges from Catalonia (Spain) Environ Sci Technol 33 2493 2498
Feil VJ Ellis RL 1998 The USDA perspective on dioxin concentrations in dairy and beef J Anim Sci 76 152 159 9464896
Fiedler H Cooper KR Bergek S Hjelt M Rappe C 1997 Polychlorinated dibenzo-p -dioxins and polychlorinated dibenzofurans (PCDD/PCDF) in food samples collected in southern Mississippi Chemosphere 34 1411 1419 9134675
Firestone D Clower M Borsetti AR Teske RH Long PE 1979 Polychlorodibenzo-p -dioxin and pentachlorophenol residues in milk and blood of cows fed technical pentachlorophenol J Agric Food Chem 27 1171 1177 544646
Fries GF 1996 Ingestion of sludge applied organic chemicals by animals Sci Total Environ 185 93 108 8643962
Fries GF Marrow GS Snow PA 1981 Soil ingestion by dairy cattle J Dairy Sci 65 611 618 7096726
Fries G Paustenbach D Mather D Luksemburg W 1999 A congener specific evaluation of transfer of chlorinated dibenzo-p -dioxins and dibenzofurans to milk of cows following ingestion of pentachlorophenoltreated wood Environ Sci Technol 33 1165 1170
Furst P Krause GHM Hein D Delschen T Wilmers K 1993 PCDD/PCDF in cow’s milk in relation to their levels in grass and soil Chemosphere 27 1349 1357
Grossi G Lichtig J Kraub P 1998 PCDD/F, PCB, and PAH content of Brazilian compost Chemosphere 37 2153 2160 9828334
Healey N Bright D 2000. Validation of Risk-Based Contaminant Management Strategies for the Beneficial Recycling of Wastewater Treatment Plant Biosolids: Great Vancouver Regional District, Vol 1. Victoria, British Columbia, Canada:Royal Roads University.
Ho AKW Clement RE 1990 Chlorinated dioxins/furans in sewage sludge of municipal water pollution control plants Chemosphere 20 1549 1552
Horstmann M McLachlan MS Reissinger M Hutzinger O 1992 Temporal variability of PCDD/F concentrations in sewage sludge Chemosphere 25 1463 1468
Hulster A Marschner H 1993 Transfer of PCDD/PCDF from contaminated soils to food and fodder crop plants Chemosphere 27 439 446
Hulster A Marschner H 1994 The influence of root exudates on the uptake of PCDD/PCDF by plants Organohal Comp 20 31 34
Hulster A Muller JF Marschner H 1994 Soil–plant transfer of polychlorinated dibenzo-p -dioxins and dibenzofurans to vegetables of the cucumber family (Cucurbitaceae) Environ Sci Technol 28 1110 1115 22176237
Jackson AP Eduljee GH 1994 An assessment of the risks associated with PCDDs and PCDFs following the application of sewage sludge to agricultural land in the UK Chemosphere 29 2523 2543
Jensen DJ Hummel RA Hahle NH Kocher CW Higgins HS 1981 Residue study on beef cattle consuming 2,3,7,8-tetra-chlorodibenzo-p -dioxin J Agric Food Chem 29 265 268 7229210
Jilg T Muller W Hagenmaier H Papke O 1992 Carry-over of polychlorinated dibenzodioxins (PCDD) and dibenzofurans (PCDF) from feed into the milk of the dairy cow Agribiol Res 45 303 310
Jones D Safe S Morcom E Holcomb M Coppock C Ivie W 1989 Bioavailability of grain and soil-borne tritiated 2,3,7,8-tetrachlorodibenzo-p -dioxin (TCDD) administered to lactating Holstein cows Chemosphere 18 1257 1263
Jones KC Sewart AP 1997 Dioxins and furans in sewage sludges: a review of their occurrence and sources in sludge and of their environmental fate, behaviour and significance in sludge-amended agricultural systems Crit Rev Environ Sci Technol 27 1 85
Kjeller L-O Rappe C Jones KC Johnston AE 1991 Increases in the polychlorinated dibenzo-p -dioxins and -furan (PCDDs/PCDFs) content of soils and vegetation since the 1840s Environ Sci Technol 25 1619 1627
Lamparski LL Nestrick TJ Strenger VA 1984 Presence of chlorinated dioxins in a sealed 1933 sample of dried municipal sewage sludge Chemosphere 13 361 365
Malloy T Goldfarb T Surico M 1993 PCDDs, PCDFs, PCBs, chlorophenols (CPs) and chlorobenzenes (CBzs) in samples from various types of composting facilities in the United States Chemosphere 27 325 334
McLachlan MS Hinkel M Reissinger M Hippelein M Kaupp H 1994 A study of the influence of sewage sludge fertilization on the concentrations of PCDD/F and PCB in soil and milk Environ Pollut 85 337 343 15091665
McLachlan MS Horstmann M Binkel M 1996a Poly-chlorinated dibenzo-p -dioxins and dibenzofurans in sewage sludge: sources and fate following sludge application to land Sci Total Environ 185 109 123
McLachlan M Richter W 1998 Uptake and transfer of PCDD/Fs by cattle fed naturally contaminated feedstuffs and feed contaminated as a result of sewage sludge application. 1. Lactating cows J Agric Food Chem 46 1166 1172
McLachlan MS Reissinger M 1990 The influence of sewage sludge fertilization on the PCDD/F concentration in soil. An example from northeastern Bavaria Organohal Comp 1 577 582
McLachlan MS Sewart AP Bacon JR Jones KC 1996b Persistence of PCDD/F in a sludge-amended soil Environ Sci Technol 30 2567 2571
McLachlan MS Thoma H Reissinger M Hutzinger O 1990 PCDD/F in an agricultural food chain. Part 1: PCDD/F mass balance of a lactating cow Chemosphere 20 1013 1020
Molina L Diaz-Ferrero J Coll M Marti R Broto-Puig F Comellas L 2000 Study of evolution of PCDD/F in sewage sludge-amended soils for land restoration purposes Chemosphere 40 1173 1178 10739059
Muller JF Hulster A Papke O Ball M Marschner H 1993 Transfer pathways of PCDD/PCDF to fruits Chemosphere 27 195 201
Muller JF Hulster A Papke O Ball M Marschner H 1994 Transfer of PCDD/PCDF from contaminated soils into carrots, lettuce and peas Chemosphere 29 2175 2181 7850372
Naf C Broman D 1990 PCDDs and PCDFs in water sludge and air samples from various levels in a waste water treatment plant with respect to composition changes and total flux Chemosphere 20 1503 1510
Pohl H DeRosa C Holler J 1995 Public health assessment for dioxins exposure from soil Chemosphere 31 2437 2454 7670858
Prinz B Krause GHM Radermacher L 1991 Criteria for the evaluation of dioxins in the environment Chemosphere 23 1743 1761
Rappe C Kjeller L-O Andersson R 1989 Analyses of PCDDs and PCDFs in sludge and water samples Chemosphere 19 13 20
Rappe C Oberg L Nilsson C Hakansson H 1999. Biosolids in Land Application—A Study on the Dioxin Situation. Washington, DC:U.S. Conference of Mayors’ Urban Water Council.
Richter W McLachlan MS 2001 Uptake and transfer of PCDD/Fs by cattle fed naturally contaminated feedstuffs and feed contaminated as a result of sewage sludge application. 2. Nonlactating cows J Agric Food Chem 49 5857 5865 11743775
Schecter A Startin J Wright C 1994 Dioxins in US food and estimated daily intake Chemosphere 29 2261 2265 7850373
Schroll R Scheunert I 1993 Uptake pathways of octachlorodibenzo-p -dioxin from soil by carrots Chemosphere 26 1631 1640
Sewart AP Harrad SJ McLachlan MS McGrath SP Jones KC 1995 PCDD/Fs and non-o -PCBs in digested UK sewage sludges Chemosphere 30 51 67 7874468
Sund KG Carlo GL Crouch RL Senefelder BC 1993 Background soil concentrations of phenolic compounds, chlorinated herbicides, PCDDs and PCDFs in the Melbourne metropolitan area Aust J Public Health 17 157 161 8399710
Telliard WA McCarty HB King JR Hoffman JB 1990 USEPA National Sewage Sludge Survey results for polychlorinated dibenzo-p -dioxins and polychlorinated dibenzofurans Organohal Comp 2 307 310
Thorpe S Kelly M Startin J Harrison N Rose M 2001 Concentration changes for 5 PCDD/F congeners after administration in beef cattle Chemosphere 43 869 879 11372879
Travis CC Hattemer-Frey HA 1987 Human exposure to 2,3,7,8-TCDD Chemosphere 16 2331 2342
Tuinstra LGMT Roos AH Berende PLM Van Rhijn JA Traag WA Mengelers MJB 1992 Excretion of polychlorinated dibenzo-p -dioxins and -furans in milk of cows fed on dioxins in the dry period J Agric Food Chem 40 1772 1776
U.S. EPA 1999 Standards for the Use or Disposal of Sewage Sludge; 40 CFR Part 503. Proposed rule Fed Reg 64 72045 72062
U.S. EPA 2004. Use and Disposal of Biosolids (Sewage Sludge). Baltimore, MD:U.S. Environmental Protection Agency. Available: http://www.epa.gov/waterscience/biosolids/index.html [accessed 8 April 2004].
van Oostdam JC Ward JEH 1995. Dioxins and Furans in the British Columbia Environment. Victoria, British Columbia, Canada:BC Environment Environmental Protection Division.
Wild SR Harrad SJ Jones KC 1994 The influence of sewage sludge applications to agricultural land on human exposure to polychlorinated dibenzo-p -dioxins (PCDDs) and furans (PCDFs) Environ Pollut 83 357 369 15091742
Wild SR Jones KC 1992 Organic chemicals entering agricultural soils in sewage sludges: screening for their potential to transfer to crop plants and livestock Sci Total Environ 119 85 119 1631535
Wilson SC Alcock RE Sewart AP Jones KC 1997 Persistence of organic contaminants in sewage sludge-amended soil: a field experiment J Environ Qual 26 1467 1477
Winters D Cleverly D Meier K Dupuy A 1996 A statistical survey of dioxin-like compounds in United States beef: a progress report Chemosphere 32 469 478 8907224
Zach R Mayoh KR 1984 Soil ingestion by cattle: a neglected pathway Health Physics 46 426 431 6693273
|
15198915
|
PMC1247188
|
CC0
|
2021-01-04 23:36:23
|
no
|
Environ Health Perspect. 2004 Jun 26; 112(9):959-969
|
utf-8
|
Environ Health Perspect
| 2,004 |
10.1289/ehp.6802
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0112-00097010.1289/ehp.689515198916Research ArticleReviewsLung Cancer Risk after Exposure to Polycyclic Aromatic Hydrocarbons: A Review and Meta-Analysis Armstrong Ben 1Hutchinson Emma 1Unwin John 2Fletcher Tony 11London School of Hygiene and Tropical Medicine, London, United Kingdom2Health and Safety Laboratory, Sheffield, United KingdomAddress correspondence to B. Armstrong, London School of Hygiene and Tropical Medicine, Keppel St., London WC1E 7HT, U.K. Telephone: 44 0 207 927 2232. Fax: 44 0 207 580 4524. E-mail:
[email protected] authors thank the Health and Safety Executive for financial support, the investigators of the included studies for providing additional information, and M. Stear for guidance on the exposure assessment.
The authors declare they have no competing financial interests.
6 2004 7 4 2004 112 9 970 978 5 12 2003 7 4 2004 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Typical polycyclic aromatic hydrocarbon mixtures are established lung carcinogens, but the quantitative exposure–response relationship is less clear. To clarify this relationship we conducted a review and meta-analysis of published reports of occupational epidemiologic studies. Thirty-nine cohorts were included. The average estimated unit relative risk (URR) at 100 μg/m3 years benzo[a]pyrene was 1.20 [95% confidence interval (CI), 1.11–1.29] and was not sensitive to particular studies or analytic methods. However, the URR varied by industry. The estimated means in coke ovens, gasworks, and aluminum production works were similar (1.15–1.17). Average URRs in other industries were higher but imprecisely estimated, with those for asphalt (17.5; CI, 4.21–72.78) and chimney sweeps (16.2; CI, 1.64–160.7) significantly higher than the three above. There was no statistically significant variation of URRs within industry or in relation to study design (including whether adjusted for smoking), or source of exposure information. Limited information on total dust exposure did not suggest that dust exposure was an important confounder or modified the effect. These results provide a more secure basis for risk assessment than was previously available.
cancerlungmeta-analysisPAHpolycyclicsreview
==== Body
Airborne polycyclic aromatic hydrocarbons (PAHs), which are emitted when organic matter is burned, are ubiquitous in the occupational and general environment. It has long been known that several PAHs can produce cancers in experimental animals, and epidemiologic studies of exposed workers, especially in coke ovens and aluminum smelters, have shown clear excesses of lung cancer and highly suggestive excesses of bladder cancer [Boffetta et al. 1997; International Agency for Research on Cancer (IARC) 1984, 1985, 1987; Mastrangelo et al. 1996; Negri and La Vecchia 2001]. The animal experiments have included some using airborne exposure and have been mixtures and individual compounds, including particularly benzo[a]pyrene (BaP). Although the existence of a cancer risk is beyond reasonable doubt, considerable uncertainty exists as to the exposure–response relationship, and hence as to the risks posed at today’s levels in the workplace and general environment. Information on this relationship is clearly important for setting of occupational and environmental standards.
Estimating exposure–response relationships by extrapolation from animal studies is possible [Collins et al. 1991; U.S. Environmental Protection Agency (U.S. EPA) 1984], but the limitation of this approach, particularly species differences, makes sole reliance on it problematic. Data from a large cohort of coke oven workers in the United States, which has been followed since the 1960s (Costantino et al. 1995; Lloyd 1971), have been used to estimate risk per unit residential exposure [Nisbet and LaGoy 1992; World Health Organization (WHO) 1987]. However, many other studies provide information that has not yet been systematically used to quantitatively assess risk.
The fact that PAHs comprise a mixture, several components of which are animal carcinogens, adds to the complexity of the task. One issue is whether a single index of exposure, such as BaP or total benzene soluble matter (BSM) or cyclohexane soluble matter (CSM) is adequate to determine risk. If such an index is used, risk per unit exposure may differ between studies (and unstudied exposures) because of differences in the ratio of this index to the total carcinogenic potential of the mixture. It is possible that such variation, if present, can be adequately described by classifying exposures in broad categories (e.g., by source). However, this approach remains untested.
We conducted a review and meta-analysis that aimed to use all relevant published evidence from epidemiologic studies to obtain an estimate or estimates of the relationship of PAH exposure with lung and bladder cancer and to identify sources of variation in this relationship. Here we report the results for lung cancer.
Methods
The methods summarized here are described at greater length in a technical report on this work (Armstrong et al. 2002).
Literature Search
We sought all potentially informative peer-reviewed publications reporting epidemiologic studies on the occupational PAH–lung cancer exposure–response relationship. Specifically, we searched the following online electronic databases: MEDLINE (http://www.nlm.nih.gov/databases/databases_medline.html); EMBASE (http://www.embase.com/); OLDMEDLINE (http://www.nlm.nih.gov/databases/databases_oldmedline.html); NIOSHTIC-2 (http://www2.cdc.gov/nioshtic2/niosh2.htm); and CancerLit (http://www.cancer.gov/search/cancer_literature/). We searched publication dates 1958–February 2001) by text phrases and supplemented these publications with articles cited in the studies we obtained.
We excluded the following:
Studies of workplaces where PAHs were considered unlikely to be the predominant lung or bladder carcinogen, to reduce potential for confounded results. Such workplaces included those in the rubber industry; those where primary exposure was from diesel exhaust; foundries; and steel works (because of co-exposure to silica), unless there were separate analyses specifically of coke oven workers.
Studies for which it was not possible to quantify exposure to PAHs. Hospital- and population-based case–control and registry studies were excluded for this reason.
Studies of occupational exposure other than by inhalation.
Superceded publications. Where repeated follow-ups of the same workforce were reported in several articles, only the most recent was included.
Biomarker studies because it was difficult to deduce relationships of exposure concentration to cancer incidence.
Proportional cancer analyses.
Articles not written in English.
After these exclusions, 34 articles remained, of which 5 reported two distinct cohorts for which results were presented separately. Thus, there were 39 cohorts. For each included cohort we systematically extracted general descriptive information, information on potential modifiers of risk associated with PAHs, and information from which we estimated unit relative risk (URR) increments (see next two subsections).
Exposure Estimation
We distinguished studies according to author reports of the following:
exposures to PAHs measured as BaP (10 cohorts)
exposures to PAHs measured by a proxy that we could convert to BaP: benzene soluble matter (BSM), total PAHs, carbon black (6 cohorts)
no measures of exposure (n = 23 cohorts)
For those studies with no exposure measures, we estimated exposure to PAHs for each workgroup for which cancer risk estimates were presented (e.g., for top-oven workers, side-oven workers). These exposure estimates were based on published exposure estimates in the same industries (IARC 1984; Lindstedt and Sollenberg 1982) and other published epidemiologic studies. Principle estimates thus derived are presented in Table 1.
Cumulative exposure.
We sought to relate cancer risk to mean cumulative exposures to BaP (duration × time-weighted mean concentration). Where risk by cumulative exposure was not published, it was derived as the product of mean estimated concentration of exposure in each group for which risk was reported and the mean duration of exposure in that group. In the absence of information on duration of exposure, 20 years was assumed, representing the average found in studies for which duration was reported.
Dust exposure.
At the inception of this study, we had not planned to seek information on potentially confounding exposures beyond those noted by the authors. However, interest has sharply increased recently in the hypothesis that inhaled dust carries a risk of lung cancer regardless of composition. We therefore sought to add information that we could find on dust exposure. Because few publications reported such estimates, we relied entirely on supplementary data and the judgment of the hygienists on the research team. We were aware that this would be a very rough assessment and chose a simple scale [low (< 1 mg/m3 total dust); moderate (1–5 mg/m3); high (5–10 mg/m3); very high (10–25 mg/m3)] and broad job groups or, in some cases, entire industries. Assessments are listed in the final column of Table 1. A list of references on which assessments were based is included in Appendix B4 of the full research report (Armstrong et al. 2002).
Estimation of Unit Risks from Studies
We estimated relative risks (RRs) per 100 μg/m3 years cumulative BaP for each study, using a log-linear model, RR = exp(bx), where RR is relative risk, x is cumulative exposure in micrograms per cubic meter years and b is the slope of the exposure–response relationship. (For this model, RR = 1 if x = 0.) Thus, relative risk represents the risk of lung cancer at a specified exposure (x) relative to that at zero exposure. For example, RR = 1.30 at x = 100 μg/m3 years BaP exposure implies that at this exposure, lung cancer risk is 1.3 times that of an unexposed person—a 30% excess.
Slopes b were estimated by Poisson regression, using data from each study from published tables of risk [usually standard mortality ratios (SMRs) or internal RRs] by cumulative exposure, duration of exposure, or job group. For 13 cohorts with only one published SMR, these estimates depended on assuming that at zero exposure those cohorts would have experienced the same rates as those of the general population (allowing for age and calendar time). For the remaining 23 cohorts, rates at zero exposure were inferred from the cohort itself (i.e., exposure–response curves were not constrained to SMR = 1 at zero exposure). Standard errors were from a scale-overdispersion model, reflecting variation in observed deaths above the Poisson expected, and estimated jointly across studies.
Unit relative risks were defined as those predicted by the models at 100 μg/m3 years BaP [i.e., exp(100b)]. The exposure of 100 μg/m3 years BaP is close to the mean of the maximum exposures in included studies and corresponds to a concentration of 2.5 μg/m3 BaP over 40 years.
Many studies reported more than one contrast of risk in differently exposed subgroups, so several URRs could be estimated. For example, there may be tables of risk by duration of service (sometimes subdivided by job group) and by job group (e.g., coke oven top-worker, side-worker, and distillation products), as well as on overall SMR. In these instances we selected the following, in order of importance:
internal comparisons (risk in groups of different exposure in the same study) over external (a single SMR)
large contrasts of exposure across exposure groups
mortality outcomes over morbidity
confounder-controlled contrasts over uncontrolled (e.g., smoking-adjusted vs. unadjusted)
estimates without latency or lag restrictions over those with such restrictions (to maximize comparability in primary analyses).
Meta-Analysis and Meta-Regression
We sought to describe the distribution of URRs across studies and to identify determinants, allowing for sampling uncertainty of each estimate and additional random variation between studies (i.e., random effects) if present (Sutton 2000). We proceeded on the assumption that sampling variation of the logged URRs and any additional random variation were reasonably approximated by a normal distribution. Cochran’s test was used to determine significance of variation in URRs between studies. Meta-regression, using a log-linear random effects model with restricted maximum likelihood, was used to clarify patterns in URRs (e.g., a tendency for different URRs for each industry) and to identify whether such patterns could have occurred by chance (Sutton 2000).
Results
Study Characteristics
Characteristics of the 39 cohorts are shown in Table 2, and the frequencies of selected study characteristics are given in Table 3.
All were essentially cohort studies, but three used nested case–control samples, and one (Armstrong et al. 1994) used case-cohort sampling from within the cohort. For 13 of the cohorts, only single SMRs were reported; for the remainder there were risk comparisons (contrasts) across 2 or more exposure groups (maximum 7). Of these, the contrasts selected according to our criteria were by cumulative exposure (8), duration of exposure (12), and job group (6).
A remarkable feature was the large range of exposures. Table 2 lists the cumulative exposure in the highest exposure group in each study, which ranged across three orders of magnitude from 0.75 to 805 μg/m3 years BaP. This corresponds approximately to con-centrations in air of 0.04–40 μg/m3. This large range was the predominant reason for the large range in the precision with which the URR was estimated.
Unit Relative Risks
Relative risks predicted at 100 μg/m3 years BaP from the log-linear model are shown to the right of the cohort characteristics in Table 2. They ranged from 0 to > 1,000. The precision with which these relative risks were estimated also varied substantially, with standard errors (log scale) ranging from 0.02 to >1,000. Most of the variation in precision was due to variation in the degree of exposure contrast in the studies. Many of the studies at the bottom of the table (power and carbon black industries) have low exposures. This limits the range of exposures being compared in the studies, which causes imprecision in estimated URRs, shown as wide confidence limits. (URRs are essentially regression coefficients—a narrow range in the x variable leaves uncertainty in the slope of the line.) Some variation in precision was also due to variation in size of cohort populations and duration of follow-up, which is reflected in the number of cases.
For cohorts without any exposure groups with mean higher than 100 μg/m3 BaP, the estimate of relative risk at this value (the URR) is an extrapolation. The extreme values of URRs found in such cohorts are thus theoretical. To give an indication of the actual relative risks found in the cohorts, we also show for each cohort (Table 2, last column) the relative risks for the group with the highest exposure in that cohort, as predicted by the (log-linear) model.
Twenty-eight (72%) of the URRs were > 1, with the lower confidence limit > 1 (p < 0.05) in 14 of these URRs. The mean (estimated by random effects meta-analysis), overall and in subgroups, is shown in Table 3. A graph of all results loses definition catastrophically in the more precise studies. Limiting the graph to studies with standard errors < 10 (Figure 1A) and 1 (Figure 1B) allows focus on the more precise and consequently influential cohorts.
The overall mean URR was 1.20 and significantly > 1 (p < 0.001). There was no one cohort dominating this estimate, and it was little changed on removal of the less precise cohorts. However, there was significant heterogeneity of URRs across cohorts (p < 0.001).
Meta-regression revealed that much of the heterogeneity was explained by variation in URRs across industries (p = 0.002), although coke ovens, gasworks, and aluminum smelters exposed to coal tar volatiles at similar levels had similar mean URRs. There was no significant heterogeneity of URRs within industry groups. We therefore examined variation in URRs according to other factors after allowing for the differences across industries by including industry in the meta-regression. After doing so, there was no difference more than could easily be explained by chance (p > 0.20) when studies were grouped according to source of exposure information, continent, whether the outcome of studies was mortality or morbidity, or exposure contrast (cumulative exposure, duration, etc). Neither did maximum exposure explain variation. The higher mean URR in the three nested case-control studies (p = 0.10) and that in the four smoking-adjusted studies (p = 0.05) are not independent. Both reflect high URRs in two case–control studies also adjusted for smoking.
Publication Bias
There was little evidence that the URR was related to its standard error or to number of cases (p > 0.20), factors that might relate to publication. It is evident in Table 2 that although the very high URRs derive from the smaller studies with lower exposures, some of the extremely low estimated URRs do also. Further, neither Egger’s test nor Begg’s test (p > 0.20) gave evidence for publication bias (Sutton 2000). Applying a trim-and-fill analysis (designed to correct for publication bias, if any) made negligible difference to the mean.
Dust
Because our information on dust exposure was for each cohort or sometimes for broad job group within studies (Table 1), we could not use conventional methods for controlling for confounding (stratification or inclusion of dust in multiple regression analyses). We adopted an ad hoc approach to use the data we had in order to shed what light we could on this issue:
We compared relative risks estimated at 100 μg/m3 years BaP in cohorts in which we had identified substantial dust exposure with those in which there was less. If generic dust were an important cause of lung cancer in these cohorts, one would expect greater apparent risks per unit PAH (BaP) where it was accompanied by dust. Results are shown at the bottom of Table 3. There was no significant association between estimated relative risk per unit PAH (BaP) exposure and dust exposure in the industry. This gives some reassurance that dust is not the predominant cause of the association seen in this cohort between PAH and lung cancer.
Sensitivity Analysis
By investigating dependence of URRs on study characteristics (Table 3), we have already implicitly examined sensitivity of results to these characteristics (study design, smoking adjustment, exposure information, etc) and found little such sensitivity. Here we report investigations of sensitivity of our results to three statistical modeling assumptions.
First, we repeated analyses using the linear model (RR = 1 + bx). We found very similar rankings of URRs (Spearman’s correlation = 0.99). Fitted relative risks at the maximum exposure found in each plant were also similar. However, there was some variation in URRs of individual cohorts; those with lower exposures typically had lower URRs with the linear model, and those with higher exposures higher URRs. For example, the URR for Swaen’s 1997 study of asphalt workers was 15.23 with the exponential model but 3.13 with the linear model; the relative risk predicted at the actual mean exposure in this cohort of 10 μg/m3 years, however, was 1.31 for both models. Because methods are not available to rigorously allow for the highly non-regular sampling error in the linear estimates in meta-analyses, we view means and the assessment of heterogeneity of URRs estimated under this model cautiously. Nevertheless, it is reassuring that the mean estimated relative risk at 100 μg/m3 years BaP was similar (1.19 compared with 1.20, both highly significant). The patterns of variation of risk across industries were broadly similar, although with some important differences (e.g., means for coke, gas, aluminum, and other were 1.22, 2.25, 1.04, and 4.41, respectively, in linear model vs 1.17, 1.15,1.16, and 10.9, respectively, in log-linear model).
Second, we repeated analyses using alternative criteria for choice of contrast:
Minimum standard error
Minimum standard error but using internal comparisons instead of single SMRs whenever available.
In either case, the mean URR and the basic pattern of URRs between industries changed little, although estimates for individual studies changed, sometimes substantially.
Finally, we investigated dependence of our results on extrapolation of risks from very high exposures, by repeating analyses three times, excluding exposure groups with means more than 80 μg/m3 years BaP (40 years at 2 μg/m3), 40 μg/m3 years BaP (1 μg/m3), and 20 μg/m3 years BaP (0.5 μg/m3). For example, the large U.S. coke ovens study (Costantino et al. 1995) had seven groups with means 0.0, 14.8, 73.7, 162.4, 251.2, 339.9, and 805.4 but contributed only the first three groups to the first reestimated URR (means ≤ 80) and only the first two to the second and third reestimated URRs (means ≤ 40 and ≤ 20). Overall mean URRs and mean URRs for coke ovens, gasworks, and aluminum smelters are given in Table 4. The mean URR increases substantially on removal of higher exposure groups. This is partly explained by the greater weight given by URRs from industries with lower exposures, most of which have higher URRs. However, looking at the results for coke ovens, gasworks, and aluminum smelters only (right side of Table 4), we see that even within these industries restricting analyses to groups with lower cumulative exposures led to higher mean URRs, suggestive of an exposure–response curve steeper at lower exposures than at higher exposures. However, for all these analyses except those excluding all exposures above 20 μg/m3 years BaP, which was imprecise, there was significant heterogeneity between studies. These results should therefore be interpreted with caution.
Discussion
That our meta-analysis supports the conclusions of previous reviews that lung cancer is associated with PAH exposure is reassuring but not surprising. Our attention to quantification of this relationship in a comprehensive review is novel. Although other reviews have cited unit risk estimates from single studies, and one (Gibbs 1997) calculated such estimates from eight studies, no meta-analyses of unit risk estimates have been published.
Our results for coke ovens, gasworks, and aluminum production are relatively well supported by evidence from multiple studies, although biases should be considered. Our findings of higher URRs for other industries are more tentative. In the following sections, we discuss biases and possible explanations for patterns of variation in URRs.
Possible Biases
Each study included in this meta-analysis is subject to the usual range of potential biases in epidemiologic studies, in particular, confounding and information bias (exposure error).
Our first concern is potential confounding by smoking, which was uncontrolled in most studies. However, for two reasons, this seems unlikely to have caused major bias: a) Although only four studies controlled for smoking, two were large studies with substantial exposure allowing precise estimates of URRs. The mean URR in smoking-adjusted studies was statistically compatible with but somewhat higher than that for the studies uncontrolled for smoking and was statistically significant. b) Several methodological articles (Axelson and Steenland 1988; Blair et al. 1988; Siemiatycki et al. 1988) have explored mathematically the potential for confounding by smoking. One common conclusion was that because comparisons are generally between groups with only moderately differing smoking habits (particularly different groups of manual workers, as in most studies in this analysis), substantial confounding is unlikely.
Confounding by other occupational exposure is also possible, but we limited that potential by excluding cohorts in which PAHs were judged unlikely to be the predominant carcinogen. We did not exclude subjects exposed to high levels of total dust, however, because the hypothesis that dust may cause lung cancer regardless of composition has gained credence only recently (Pope et al. 2002) and because dust is a universal co-exposure of PAHs. The analysis that we conducted addressing the possibility of confounding by dust gave no support to the hypothesis that dust plays a major confounding role. Other ad hoc investigations of confounding potential, in particular noting the absence of lung cancer excess in prebake aluminum workers (exposed to dust but little PAH), came to similar conclusions (Armstrong et al. 2002). However, none of our analyses could rule out confounding completely because of the limited information on total dust exposure available to us and the lack of control for this exposure in the published studies. Further evaluation will be possible when assessments of the dust hypothesis are carried out, which was not possible in this study.
Exposure is likely to have been inaccurately estimated in many studies, in particular those for which no exposure data were published in the report of the epidemiologic study itself, so we made estimates. Random exposure error tends to bias exposure–response slopes toward the null value (Armstrong 1998). However, if our estimates were systematically too high or too low, exposure response slopes would be underestimated or overestimated, respectively. We included estimates if we believed them to be within 4 times the true exposure, so considerable margin for uncertainty remains. It is somewhat reassuring that the mean URR in those studies for which we estimated exposure was not much different from the mean URR in those studies with author-provided exposure information (Table 3). However, errors in exposure estimation might explain particularly high or low URRs in specific studies or industries. In those industries (tar distillation, chimney sweeping, power) with no studies reporting investigators’ own exposure estimates, interpretation should be particularly cautious.
Finally, could bias be introduced by selection of cohorts or contrasts for inclusion? Our sensitivity analyses suggest no strong sensitivity, and standard tests for publication bias were negative. However, the overall mean URR is strongly influenced by the predominance of coke ovens and aluminum smelters in the sample.
Explanations for Variation in Unit Relative Risks
Unit relative risks may vary between industries and cohorts for three reasons: a) chance, b) biases, or c) because risk per unit BaP really varies. We established that variation in URRs between industries cannot be explained by chance (particularly coke ovens and aluminum production vs. asphalt and chimney sweeping), but variation within industry can be. Therefore, it seems sensible to focus attention on explaining variation between industries. We discussed biases and confounding in the preceding section. Biases, in particular from inevitably inaccurate exposure estimation, could explain some variation. Confounding by other occupational exposures, perhaps dust, could also play a part, although we found no evidence for this.
Two reasons might account for true variation in URRs. a) A factor that modifies the effect is present to varying degrees in different industries. An example is smoking. Even if different PAH exposure groups in each cohort smoke to the same extent (so there is no confounding), a heavily smoking cohort might exhibit greater or lesser effect on relative risk per unit occupational PAH than a lightly smoking cohort. Unfortunately, we did not have the information to address this aspect. Generally, we can assume that our cohorts were mixed smokers and nonsmokers, so the exposure–response relationships are most likely to predict risk well in similarly mixed groups. Other occupational exposures might also modify the effect per unit PAH by promoting or inhibiting the action of PAHs. Such a hypothesis is too general to evaluate without making it more specific. Finally, cumulative exposure may not be the right metric. If another metric (e.g., early adult exposure, lagged exposure, or another time-weighted exposure) were the relevant one, modification would arise if the time pattern of exposure differed across cohorts. Information on timing of exposure was insufficient for us to evaluate such hypotheses, but in any case we expect that timing of exposure would be too similar across cohorts for informative results to emerge. A few studies reported risk by lagged cumulative exposure, but these generally differed little from tables of risk by overall cumulative exposure. b) The carcinogenic potency of the PAH mixture varies across industries. As we noted earlier, many PAHs aside from BaP are carcinogenic in animals (IARC 1985). BaP is used as an indicator of the total risk, not because it is the sole causal agent but because at least in some industries it correlates well with other agents (Expert Panel on Air Quality Standards 1999). To the extent that PAH mixtures in different industries have different relative concentrations of the various carcinogenic PAHs (their profiles), this could thus explain differences in risk per unit BaP. Krewski et al. (1989) have proposed an approach that derives a risk metric by combining information on PAH profiles with information on relative carcinogenic potency from animal studies. To apply that approach to this meta-analysis, however, would require estimates of PAH profiles for each study or at least each industry. We did not have such information, which is not readily available, for this study. However, PAH profiles are slowly being ascertained and some are published (Appendix B3, Armstrong et al. 2002), so this approach could probably be applied in the future.
Some specific studies with URRs quite different from the mean for their industry deserve specific mention:
The two cohorts of gasworks worker studies (Doll et al. 1972) have high URRs. The estimate of exposure for retort workers in these plants was 3 μg/m3 BaP and was heavily influenced by measurements reported in 1965 from mask samples. These measurements may have been underestimates.
The very high URR estimated from one study of carbon anode plant workers (Liu et al. 1997) was based on exposure estimates reported by Liu for just one of seven plants, which may not have been representative.
The low and precisely estimated URR from the study of several Norwegian aluminum production plants (Romundstad et al. 1998) has no obvious explanation. Exposure estimation was based on substantial hygiene data for most plants. The nonsignificance of the test for heterogeneity in URRs among studies of aluminum production workers indicates that the absence of excess risk in this study could have been due to chance, but the result remains noteworthy.
Comparisons with Other Unit Risk Assessments
The most directly comparable study estimated lifetime risks of lung cancer per 100,000 men from 50 years of continuous exposure to 1 ng/m3 BaP [unit lifetime risk (ULR)] from nine studies, using a linear no-threshold model (Gibbs 1997). Eight of the nine studies were occupational (four from coke ovens, two gasworks, one aluminum production, and one asphalt), and they or their updates were included in our analysis. The ninth was a study of domestic exposure to smoky coal in China. To translate Gibbs’ risk lifetime estimates from continuous exposure to relative risk estimates from occupational exposure (per 100 μg/m3 BaP), we used the conversion factors used by Gibbs to do the reverse:
where ULR is the lifetime risk per nanogram per cubic meter continuous (23 m3/day vs. 10 occupational) exposure (365 days vs. 230 occupational—our assumption) over 50 years, assuming a 9% baseline lifetime risk. Gibbs’ finding of ULR 0.3, 4.2, 4.4, 5.8, 6.6, 7.2, 7.8, and 9.5 translates to URRs 1.02, 1.26, 1.27, 1.35, 1.40, 1.44, 1.48, and 1.58, which are somewhat higher on average than the estimates for the same studies in this analysis but not grossly different.
It is also possible to compare our meta-analytic estimates with those published from the U.S. coke oven cohort (latest update reported by Costantino 1995):
The U.S. EPA (1984), cited by WHO (1987) estimated a lifetime risk from continuous exposure per nanogram per cubic meter BaP of 8.7/100,000 using the linearized multistage model. Following the translation we used for the Gibbs study, this corresponds to a URR (relative risk from 100 μg/m3 years BaP) of about 1.53.
Moolgavkar et al. (1998) estimated a unit absolute risk from continuous 1 μg/m3 BSM of 15/100,000 using the two-stage clonal expansion model. Roughly re-expressing this using the Gibbs study translation gives a URR = 1.13.
The second of these alternative estimates of URR is very similar to our estimate from Costantino (1995) using the exponential model (Table 2; URR = 1.15).
A review of 10 studies with risk estimates published for two or more exposure groups (Mastrangelo et al. 1996) emphasized the unit risk estimate published in one aluminum smelter study (Armstrong et al. 1994). The unit risk estimate cited by Mastrangelo for the Armstrong study used the linear model and was somewhat higher (1.39 translated into the units we have used) than the log-linear URR for this study (1.22) that we used in our meta-analysis.
Our findings of a larger URR in the asphalt industry than in coke ovens or aluminum smelters was tentative. Recent publication of a very large European study of mortality in the asphalt industry (Boffetta and Burstyn 2003) will add important information on this question. The study was not published in time for formal inclusion in this meta-analysis. It found an association of lung cancer with exposure to bitumen fumes in some but not other analyses. Estimates of exposure to PAHs as BaP were made, allowing as far as possible for knowledge of the extent to which coal tar was used as an additive, time trends in exposure levels, and type of asphalt paving. In the asphalt industry PAH exposure originates from bitumen, coal tar (now banned in Western Europe), and diesel exhaust. Contribution of diesel exhaust to PAH exposure was not incorporated into quantitative PAH exposure metric because available data did not permit the investigators to identify groups of asphalt pavers within the cohort with different diesel exhaust exposure. The technical report of this study (Boffetta et al. 2001) includes a table (8.9.4) of lung cancer rate ratios in relation to cumulative exposure to PAHs (as BaP). From this table, it was possible to estimate a URR in the method that was standard for our meta-analysis. The estimate [44.9; 95% confidence interval (CI), 25.0–64.8] is similar to the that of other asphalt worker studies included in this review, adding support to the hypothesis that risk per unit BaP is higher in this industry than in coke ovens or aluminum production. However, analysis of risk by quantitative estimate of PAH exposure was possible only for workers employed in paving (including mastic paving). It may be that other groups in the study (e.g., roofers) showed different patterns.
Interpretation for Risk Assessment
We have used a benchmark of 100 μg/m3 years exposure to provide a scale for presenting the URR, but risk predictions at other exposures (x) can be made using the formula
For example, relative risk consequent on exposure to 1 μg/m3 for 40 years (40 μg/m3 years) according to the mean estimate for coke ovens is 1.17(40/100) = 1.06. (At these moderate to low relative risks, log-linear interpolation is close to linear interpolation.) Risk estimates calculated this way for a range of URRs and exposure concentrations are given in Table 5.
Overall or industry-specific means?
The URRs overall had significant and substantial heterogeneity. There was evidence that risk per unit BaP varied across cohorts. The mean in the presence of this heterogeneity is a rather artificial one, reflecting those industries and cohorts that happen to have been studied. Within industries there was no significant heterogeneity, so that the industry-specific means could be interpreted as representative of each industry. These considerations favor use of industry-specific means. Means for coke ovens, gas works, and aluminum production are consistent and relatively precisely estimated. The combined mean URR for these industries was 1.17 (95% CI, 1.12–1.22) and might reasonably be used for all these industries. However, means for other industries are imprecise. Risk assessment for these industries will inevitably be uncertain, whether the imprecise industry-specific mean or the overall mean was used.
Model choice.
Risk assessment depends on the form of the model, in particular for extrapolation of risk to exposure ranges far from those observed. We adopted the log-linear model because the linear model is not amenable to rigorous statistical evaluation; estimates and confidence intervals for means, and p-values for heterogeneity are unreliable. However, evidence suggests (Appendix C, Armstrong et al. 2002) that the linear model fits the data and arguments on mechanism better than the log-linear model. That the overall mean and broad pattern of URRs under the linear and log-linear models were similar is reassuring, but having model choice forced by statistical tractability is not ideal. The development of methods to allow better meta-analysis of linear relative risk models would be useful.
Apart from the log-linear and linear models, models with very different assumptions about increments at low exposures, such as threshold models, could predict very different risks at these levels. However, information was insufficient to fit these or other more elaborate models (e.g., two-stage, multistage) with the information published. In particular, lack of information precluded our investigating dependence of risk on timing or exposure beyond the cumulative exposure model, for example, risk eventually declining after exposure. The sensitivity analysis (Table 4) investigating dependence of results on high exposures was suggestive of an exposure–response curve steeper at lower exposure than at higher exposure.
Attributable burden of disease.
The number of cancers caused by occupational exposure to PAHs depends on three factors beyond the exposure–response relationship: a) the number of persons exposed; b) the levels at which they are exposed, and c) the background rate of lung cancer on which relative risks will act. As an example, we have made an estimate of cases that would be caused in U.K. coke oven workers by PAH exposures continuing at current levels, ignoring probably higher past exposures. There are currently about a thousand coke oven workers in the United Kingdom, with mean exposure about 1.5 μg/m3 BaP (Unwin J, personal communication). General population lifetime risk of lung cancer in U.K. males, using 1997 rates, is 8% (Office for National Statistics 2000). Using the mean URR of 1.17 for coke ovens, 1 year of exposure will therefore lead eventually to a lifetime excess risk of 0.08 × (1.17(1.5/100)–1) = 1.9 × 10–4, which among 1,000 workers will lead to 0.2 cases. Forty years of such exposure would lead to 40 × 0.2 = 8 cases.
Assessing risk in the general environment.
Included cohorts were all occupationally exposed, and our study was aimed primarily at informing risk assessment in an occupational setting. However, given the limited number and informativeness of direct studies of risks from PAHs in the general ambient exposure, these data also provide a possible basis for estimating these risks. A full discussion is beyond the scope of this article, but we note that for this purpose our (occupational) ULRs would have to be converted to apply to continuous (24-hr, 365-day) exposure, such as with the assumptions of Gibbs discussed above.
Uncertainty.
We have acknowledged many sources of uncertainty in risk estimates made from a summary URR. Many such sources, notably model choice and exposure uncertainty, are not incorporated in the confidence intervals, which should be regarded as lower bounds of uncertainty.
Methodological Lessons
Compared with their widespread use in clinical trials, meta-analyses are relatively new to occupational epidemiology, and even more rare in investigations of exposure–response relationships. In entering this poorly charted territory, this study presented several methodological challenges for which we found reasonable but ad hoc solutions. It might be useful to future similar meta-analyses for us to draw attention to the principle issues:
We needed to choose one contrast from each study from which to estimate an exposure–response relationship. To be objective, we selected a simple choice algorithm and explored sensitivity of results to it, but it may be that this procedure could be improved.
We needed to estimate mean exposure in upper-exposure groups for which only a lower limit was published.
A priori considerations and data in the meta-analysis studies suggested use of linear rather than log-linear models, but estimates of URRs from linear models proved intractable in meta-analysis, so we worked with log-linear models. It would be preferable not to have to compromise. One possibility would be to apply a random effects linear relative risk model to semiaggregated data (see below), but to our knowledge, such models have not been discussed in the statistical literature, nor can they be fitted with standard software.
We proceeded in this meta-analysis to estimate first a single effect measure (URR) from each study, then analyze these measures using standard meta-analytic methods. However, it appears to us that once semiaggregated data have been assembled for cases, exposures, and relative risks in each exposure group in each study (Appendix E, Armstrong et al. 2002), it would be possible to use methods developed more generally for hierarchical data (multilevel models).
Conclusion
Considerable independent data are now available that allow us to conclude that occupational exposure to PAHs by inhalation is associated with a risk of lung cancer. For exposures in the coke ovens, gasworks, and aluminum industries, the risk can be estimated and is equivalent to a relative risk of 1.06 for a working lifetime at 1 μg/m3 exposure to BaP. Exposures in other industries with PAH exposure, in particular carbon anode plants, asphalt use, and tar distilleries, suggest higher risks at equivalent BaP exposure, but the risk estimates are much less precise.
Figure 1 Estimated URRs (squares) and 95% (lines) for each cohort. Confidence intervals are truncated at the edges of the graph. Diamonds at the bottom of each show the overall mean and its confidence interval. The solid vertical line is at RR = 1, and the broken line is at the mean URR. (A) All studies with standard errors of log(URR) < 10. (B) Further restricted to studies with SEs < 1. Imprecise estimated URRs (SE > 10) are not graphed.
Table 1 Main supplementary exposure estimates (BaP; total dust).
Industry Job group BaP (μg/m3) Dusta
Coke ovens Top 20 H
Side 10 M
Other 0.5 M
Typical plant mean 10 H/M
Coal gas production Retorts 3 M
By-products 0.5 L
Typical plant M/L
Aluminum smelting Soderberg potroom 15 VH
Prebake potroom 0.05 H
Carbon plant 2 H
Typical plant mean: Soderberg 3 H
Typical plant mean: Prebake 0.5 H/M
Carbon anode plants — 1 H
Asphalt — 0.5 M
Tar distillation — 0.5 M
Chimney sweep — 1 VH
Thermoelectric power — 0.05 L
Carbon black — 0.05 H
aDust classification: L, low (< 1 mg/m3); M, moderate (1–5 mg/m3); H, high (5–10 mg/m3); VH, very high (10–25 mg/m3).
Table 2 Cohort characteristics and URR estimates.
Cohort characteristics
URR
RR
First author and year Industrya Country Designb Author exposurec Contrastd Outcomee Smoking adjustment Cases (n) Exposure (n) Maximum exposuref Estimate (95% CI) SEg Maximum exposure
Bye 1998 Coke Norway Cohort Proxy cum.exp. Morb No 7 4 10.0 > 1,000 (0.01 to > 1,000) 11.43 6.10
Chau 1993 Coke France Cohort None jobgroup Mort Yes 24 5 294.4 1.00 (0.68–1.46) 0.20 0.99
Costantino 1995 Coke USA Cohort Proxy cum.exp. Mort No 458 7 805.4 1.15 (1.10–1.21) 0.02 3.18
Franco 1993 Coke Italy Cohort None singleSMR Mort No 19 1 186.0 1.41 (1.01–1.97) 0.17 1.90
Hurley 1983 Coke U.K. Cohort Proxy cum.exp. Mort No 182 4 252.9 1.36 (1.04–1.79) 0.14 2.19
Hurley 1983 Coke U.K. Cohort Proxy cum.exp. Mort No 59 4 262.9 1.19 (0.77–1.85) 0.22 1.60
Reid 1956 Coke U.K. Cohort None jobgroup Mort No 21 3 400.0 0.94 (0.64–1.39) 0.20 0.79
Sakabe 1975 Coke Japan Cohort None singleSMR Mort No 15 1 200.0 1.13 (0.80–1.60) 0.18 1.28
Swaen 1991 Coke Holland Cohort None jobgroup Mort No 273 3 200.0 1.19 (0.97–1.45) 0.10 1.41
Xu 1996 Coke China Nested BaP duration Morb Yes 194 3 453.8 1.33 (1.14–1.56) 0.08 3.65
Berger 1992 Gas Germany Cohort BaP singleSMR Mort No 78 1 747.6 1.15 (1.11–1.20) 0.02 2.88
Doll 1972 Gas U.K. Cohort None jobgroup Mort No 79 3 60.0 4.01 (1.16–13.87) 0.63 2.30
Doll 1972 Gas U.K. Cohort None jobgroup Mort No 110 2 60.0 5.82 (1.06–32.00) 0.87 2.88
Gustavsson 1990 Gas Sweden Cohort BaP singleSMR Mort No 0 1 28.7 0.00 (0.00–66.56) 1,450 0.00
Armstrong 1994 Alum Canada Ca-coh BaP cum.exp. Mort Yes 338 5 413.1 1.22 (1.09–1.37) 0.06 2.30
Milham 1979 Alum USA Cohort None duration Mort No 35 6 99.2 0.19 (0.00 to > 1,000) 6.15 0.19
Moulin 2000 Alum France Cohort None duration Mort No 19 5 200.0 1.11 (0.46–2.66) 0.45 1.23
Mur 1987 Alum France Cohort None duration Mort No 17 3 248.2 0.69 (0.31–1.54) 0.41 0.40
Rockette 1983 Alum USA Cohort None duration Mort No 64 5 116.1 1.85 (0.53–6.53) 0.64 2.05
Rockette 1983 Alum USA Cohort None duration Mort No 133 5 15.4 0.06 (0.00–9.58) 2.59 0.65
Romundstad 2000 Alum Norway Cohort BaP cum.exp. Morb No 189 4 222.4 0.99 (0.79–1.22) 0.11 0.97
Spinelli 1991 Alum Canada Cohort Proxy cum.exp. Morb No 37 5 251.1 1.31 (0.72–2.39) 0.30 1.99
Donato 2000 Carbon Italy Cohort None duration Mort No 34 3 36.4 0.18 (0.01–5.61) 1.75 0.54
Liu 1997 Carbon China Cohort BaP jobgroup Mort No 50 4 17.3 53.07 (3.44–819) 1.40 1.99
Moulin 1989 Carbon France Nested BaP duration Morb Yes 7 4 94.9 2.82 (0.20–40.59) 1.36 2.67
Moulin 1989 Carbon France Nested BaP duration Mort No 13 4 5.8 0.00 (0.00 to > 1,000) 24.21 0.41
Hammond 1976 Asphalt USA Cohort BaP duration Mort No 121 4 66.8 5.63 (0.89–35.53) 0.94 3.17
Hansen 1991 Asphalt Denmark Cohort BaP singleSMR Mort No 25 1 20.3 189.59 (13.5 to > 1,000) 1.35 2.90
Swaen 1997 Asphalt Holland Cohort None singleSMR Mort No 39 1 10.0 15.23 (0.21 to > 1,000) 2.19 1.31
Hansen 1989 Tar Denmark Cohort None singleSMR Mort No 16 1 10.0 35.76 (0.04 to > 1,000) 3.42 1.43
Maclaren 1987 Tar U.K. Cohort None singleSMR Mort No 12 1 6.0 > 1,000 (0.01 to > 1,000) 6.58 1.60
Swaen 1997 Tar Holland Cohort None singleSMR Mort No 48 1 10.0 5.32 (0.11–89.4) 1.97 1.18
Evanhoff 1993 Chimney Sweden Cohort None duration Mort No 53 4 40.0 9.88 (0.60–162) 1.43 2.50
Hansen 1983 Chimney Denmark Cohort None singleSMR Mort No 5 1 30.0 44.63 (0.82 to > 1,000) 2.04 3.13
Cammarano 1986 Power Italy Cohort None singleSMR Mort No 5 1 1.0 > 1,000 (0.00 to > 1,000) 61.16 1.77
Forastiere 1989 Power Italy Cohort None duration Mort No 8 3 1.5 0.02 (0.00 to > 1,000) 110.37 0.94
Petrelli 1989 Power Italy Cohort None singleSMR Mort No 6 1 1.0 > 1,000 (0.00 to > 1,000) 55.83 1.36
Robertson 1996 C_black USA Cohort None singleSMR Mort No 34 1 1.0 0.00 (0.00 to > 1,000) 23.45 0.84
Sorahan 2001 C_black U.K. Cohort Proxy cum.exp. Mort No 64 4 0.8 > 1,000 (0.00 to > 1,000) 58.15 1.48
aIndustry: Alum, aluminum smelter; Carbon, carbon anode plant; Tar, tar distillery; Chimney, chimney sweep; Power, thermoelectric power plant; C_black, carbon black.
bDesign: Nested, nested case–control; Ca-coh, case–cohort.
cAuthor exposure: information provided by the authors on exposure to BaP.
dContrast: Basis of risk comparison from which URR was estimated. cum.exp., cumulative exposure.
eOutcome: Morb, morbidity; Mort, mortality.
fExposure in micrograms per cubic meter years BaP in the highest exposure group.
gStandard error of URR (log scale).
Table 3 Distribution and determinants of URRs.
Group Studies (n) Mean URRa (95% CI) Significance testsb
All cohorts 39 1.20 (1.11–1.29) p(het) = 0.007
Excluding less precise URR estimates
Restricted to URRs with SE < 10 31 1.20 (1.11–1.30) p(het) = 0.002
Restricted to URRs with SE < 1 19 1.18 (1.12–1.23) p(het) = 0.19
By industry p = 0.002
Coke ovens 10 1.17 (1.12–1.22)
Gasworks 4 1.15 (1.11–1.20)
Aluminum 8 1.16 (1.05–1.28)
(above three combined) 22 1.17 (1.12–1.22) p(het) > 0.20]
Carbon 4 4.30 (0.81–22.79)
Asphalt 3 17.50 (4.21–72.78)
Tar distillery 3 12.28 (0.48–314.4)
Chimney sweep 2 16.24 (1.64–160.7)
Power 3 > 1,000 (0 to > 1,000)
Carbon black 2 0 (0 to > 1,000)
By exposure information from authors p > 0.20
BaP 10 1.29 (1.11–1.49)
Proxy 6 1.16 (1.11–1.21)
None 23 1.17 (1.03–1.33)
By contrast p > 0.20
Cumulative exposure 8 1.16 (1.11–1.22)
Duration 12 1.27 (1.10–1.48)
Job group 6 1.16 (0.99–1.36)
Single SMR 13 1.20 (0.95–1.51)
By study design p = 0.10
Cohort 36 1.16 (1.11–1.21)
Nested case–control 3 1.33 (1.14–1.55)
By smoking adjustment p = 0.05
No 35 1.16 (1.11–1.21)
Yes 4 1.31 (1.16–1.48)
By continent p > 0.20
Asia 3 1.30 (1.13–1.50)
Europe 28 1.13 (1.02–1.26)
North America 8 1.16 (1.11–1.22)
By outcome p > 0.20
Mortality 34 1.17 (1.12–1.22)
Morbidity 5 1.21 (1.06–1.38)
By dust exposure for industry p = 0.12
Low 3 > 1,000 (0 to > 1,000)
Moderate 10 1.16 (1.11–1.21)
High 24 1.17 (1.13–1.22)
Very high 2 16.24 (1.64–14.8)
aRR at 100 μg/m3 BaP years. Adjusted for differences across industries by including industry indicator in a meta-regression. Means are scaled to show fitted values for coke ovens, although ratios would apply to any industry.
bGenerally, the Wald test for significance of variation in mean URRs across the groups indicated was used; “p(het)” indicates the test for heterogeneity across all studies.
Table 4 Investigating the dependence of mean URR on high exposures.
All studies
Coke, gas, aluminum
Exclusions na URR (95% CI) p(het)b na URR (95% CI) p(het)b
No exclusions 39 1.20 (1.11–1.29) < 0.001 22 1.17 (1.12–1.22) 0.20
> 80 μg/m3 34 3.46 (2.03–5.90) < 0.001 17 1.88 (1.22–2.91) 0.02
> 40 μg/m3 30 6.49 (1.99–21.12) < 0.001 14 2.42 (0.56–10.40) 0.01
> 20 μg/m3 21 4.54 (1.26–16.30) > 0.20 7 1.87 (0.24–14.22) > 0.20
aRemaining number of studies from which URRs could be estimated.
bTest for heterogeneity between URRs.
Table 5 Relative risks for contracting cancer estimated to follow from 40 years of occupational exposure.
Exposure to BaP in μg/m3 for a working life of 40 years
URR used for circulation 0.1 0.2 0.5 1
Overall mean URR (1.20) 1.007 1.015 1.035 1.076
Mean URR for coke ovens, aluminum smelters, and gasworks (1.16) 1.006 1.012 1.030 1.061
Mean URR for asphalt (17.5) 1.12 1.26 1.78 3.14
==== Refs
References
Armstrong BG 1998 Effect of measurement error on epidemio-logical studies of environmental and occupational exposures Occup Environ Med 55 651 656 9930084
Armstrong BG Hutchinson E Fletcher T 2002. Cancer Risk following Exposure to Polycyclic Aromatic Hydrocarbons (PAHs): A Meta-Analysis. Rep No 068. Sudbury, UK:This Health and Safety Executive. Available: http://www.hse.gov.uk/research/rrhtm/rr068.htm [accessed 19 May 2004].
Armstrong B Tremblay C Baris D Theriault G 1994 Lung cancer mortality and polynuclear aromatic hydrocarbons: a case-cohort study of aluminum production workers in Arvida, Quebec, Canada Am J Epidemiol 139 250 262 8116600
Axelson O Steenland K 1988 Indirect methods of assessing the effects of tobacco use in occupational studies Am J Ind Med 13 105 118 3344750
Berger J Manz A 1992 Cancer of the stomach and the colon-rectum among workers in a coke gas plant Am J Ind Med 22 825 834 1463028
Blair A Steenland K Shy C O’Berg M Halperin W Thomas T 1988 Control of smoking in occupational epidemiologic studies: methods and needs Am J Ind Med 13 3 4 3344754
Boffetta P Burstyn I 2003 Studies of carcinogenicity of bitumen fumes in humans Am J Ind Med 43 1 2 12494416
Boffetta P Burstyn I Partanen T Kromhout H Svane O Langard S 2001. IARC Epidemiological Study of Cancer Mortality among European Asphalt Workers. Internal Report. Lyon:International Agency for Research on Cancer.
Boffetta P Jourenkova N Gustavsson P 1997 Cancer risk from occupational and environmental exposure to polycyclic aromatic hydrocarbons Cancer Causes Control 8 444 472 9498904
Bye T Romundstad PR Ronneberg A Hilt B 1998 Health survey of former workers in a Norwegian coke plant. Part 2. Cancer incidence and cause specific mortality Occup Environ Med 55 622 626 9861185
Cammarano G Crosignani P Berrino F Berra G 1986 Additional follow-up of cancer mortality among workers in a thermoelectric power plant Scand J Work Environ Health 12 631 632 3823815
Chau N Bertrand JP Mur JM Figueredo A Patris A Moulin JJ 1993 Mortality in retired coke oven plant workers Br J Ind Med 50 127 135 8435345
Collins JF Brown JP Dawson SV Marty MA 1991 Risk assessment for benzo[a]pyrene Regul Toxicol Pharmacol 13 170 184 1852928
Costantino JP Redmond CK Bearden A 1995 Occupationally related cancer risk among coke oven workers: 30 years of follow-up J Occup Environ Med 37 597 604 7640988
Doll R Vessey MP Beasley RWR Buckley AR Fear EC Fisher REW 1972 Mortality of gasworkers - final report of a prospective study Br J Ind Med 29 394 406 15625750
Donato F Monarca S Machionna G Rossi A Cicioni C Chiesa R 2000 Mortality from cancer and chronic respiratory diseases among workers who manufacture carbon electrodes Occup Environ Med 57 484 487 10854502
Evanoff BA Gustavsson P Hogstedt C 1993 Mortality and incidence of cancer in a cohort of Swedish chimney sweeps: an extended follow up study Br J Ind Med 50 450 459 8507598
Expert Panel on Air Quality Standards 1999. A recommendation for a United Kingdom air quality standard for polycyclic aromatic hydrocarbons. Internal Report. London:Department for Environment, Transport, and the Regions.
Forastiere F Pupp N Magliola E Valesini S Tidei F Perucci CA 1989 Respiratory cancer mortality among workers employed in thermoelectric power plants Scand J Work Environ Health 15 383 386 2617254
Franco F Chellini E Seniori Costantini A Gioia A Carra G Paolinelli F 1993 Mortality in the coke oven plant of Carrara, Italy Med Lav 84 443 447 8177128
Gibbs GW 1997 Estimating residential polycyclic aromatic hydrocarbon (PAH) related lung cancer risks using occupational data Ann Occup Hyg 41 49 53
Gustavsson P Reuterwall C 1990 Mortality and incidence of cancer among Swedish gas workers Br J Ind Med 47 169 174 2328224
Hammond EC Selikoff IJ Lawther PL Seidman H 1976 Inhalation of benzpyrene and cancer in man Ann NY Acad Sci 271 116 124 1069493
Hansen ES 1983 Mortality from cancer and ischemic heart disease in Danish chimney sweeps: a five-year follow-up Am J Epidemiol 117 160 164 6829545
Hansen ES 1989a Cancer incidence in an occupational cohort exposed to bitumen fumes Scand J Work Environ Health 15 101 105 2772572
Hansen ES 1989b Cancer mortality in the asphalt industry: a ten year follow up of an occupational cohort Br J Ind Med 46 582 585 2775679
Hansen ES 1991 Mortality of mastic asphalt workers Scand J Work Environ Health 17 20 24 2047802
Hurley JF Archibald RM Collings PL Fanning DM Jacobsen M Steele RC 1983 The mortality of coke workers in Britain Am J Ind Med 4 691 704 6650509
IARC 1984 Polynuclear Aromatic Compounds. Part 3: Industrial Exposures to Aluminum Production, Coal Gasification, Coke Production, and Iron and Steel Founding IARC Monogr Eval Carcinog Risks Hum 34 1 219
IARC 1985 Polynuclear Aromatic Compounds. Part 4: Bitumens, Coal-Tars and Derived Products, Shale Oils and Soots IARC Monogr Eval Carcinog Risks Hum 35 1 271
IARC 1987 Overall evaluations of carcinogenicity: an updating of IARC monographs 1–42 IARC Monogr Eval Carcinog Risks Hum Suppl 7 1 440 3482203
Krewski D Thorslund T Withey J 1989 Carcinogenic risk assessment of complex mixtures Toxicol Ind Health 5 851 867 2683227
Lindstedt G Sollenberg J 1982 Polycyclic aromatic hydrocarbons in the occupational environment: with special reference to benzo[a]pyrene measurements in Swedish industry Scand J Work Environ Health 8 1 19 6291144
Liu N Wang Z Dong D Chen K Qin L 1997 Cancer mortality among carbon workers in China: retrospective cohort study J Occup Health 39 325 330
Lloyd JW 1971 Long-term mortality study of steelworkers. V. Respiratory cancer in coke plant workers J Occup Med 13 53 68 5546197
Maclaren WM Hurley JF 1987 Mortality of tar distillation workers Scand J Work Environ Health 13 404 411 3433042
Mastrangelo G Fadda E Marzia V 1996 Polycyclic aromatic hydrocarbons and cancer in man Environ Health Perspect 104 1166 1170 8959405
Milham S Jr 1979 Mortality in aluminum reduction plant workers J Occup Med 21 475 480 469612
Moolgavkar SH Luebeck EG Anderson EL 1998 Estimation of unit risk for coke oven emissions Risk Anal 18 813 825 9972584
Moulin JJ Clavel T Buclez B Laffitte-Rigaud G 2000 A mortality study among workers in a French aluminium reduction plant Int Arch Occup Environ Health 73 323 330 10963416
Moulin JJ Wild P Mur JM Lafontaine M Lefer M Mercier-Gallay M 1989 Risk of lung, larynx, pharynx and buccal cavity cancers among carbon electrode manufacturing workers Scand J Work Environ Health 15 30 37 2922586
Mur JM Moulin JJ Meyer-Bisch C Massin N Coulon JP Loulergue J 1987 Mortality of aluminium reduction plant workers in France Int J Epidemiol 16 257 264 3610453
Negri E La Vecchia C 2001 Epidemiology and prevention of bladder cancer Eur J Cancer Prev 10 7 14 11263594
Nisbet IC LaGoy PK 1992 Toxic equivalency factors (TEFs) for polycyclic aromatic hydrocarbons (PAHs) Regul Toxicol Pharmacol 16 290 300 1293646
Office for National Statistics 2000 Registrations of Cancer Diagnosed 1994–1997 Health Stat Q 7 71 82
Petrelli G Menniti-Ippolito F Taroni F Raschetti R Magarotto G 1989 A retrospective cohort mortality study on workers of two thermoelectric power plants: fourteen-year follow-up results Eur J Epidemiol 5 87 89 2707397
Pope CA III Burnett RT Thun MJ Calle EE Krewski D Ito K 2002 Lung cancer, cardiopulmonary mortality, and long-term exposure to fine particulate air pollution JAMA 287 1132 1141 11879110
Reid DD Buck C 1956 Cancer in coking plant workers Br J Ind Med 13 265 269 13364156
Robertson JM Inman KJ 1996 Mortality in carbon black workers in the United States J Occup Environ Med 38 569 570 8794954
Rockette HE Arena VC 1983 Mortality studies of aluminum reduction plant workers: potroom and carbon department J Occup Med 25 549 557 6886861
Romundstad PR Anderson A Haldorsen T 2000 Cancer incidence among workers in six Norwegian aluminum plants Scand J Work Environ Health 26 461 469 11201392
Romundstad PR Ronneberg A Leira HL Bye T 1998 Health survey of former workers in a Norwegian coke plant. Part 1. Estimation of historical exposures Occup Environ Med 55 616 621 9861184
Sakabe H Tsuchiya K Takekura N Nomura S Koshi S Takemoto K Matsushita H 1975 Lung cancer among coke oven workers Ind Health 13 57 68
Siemiatycki J Wacholder S Dewar R Cardis E Greenwood C Richardson L 1988 Degree of confounding bias related to smoking, ethnic group, and socioeconomic status in estimates of the associations between occupation and cancer J Occup Med 30 617 625 3171718
Sorahan T Hamilton L Van Tongeren M Gardiner K Harrington JM 2001 A cohort mortality study of U.K. carbon black workers, 1951–1996 Am J Ind Med 39 158 170 11170158
Spinelli JJ Band PR Svirchev LM Gallagher RP 1991 Mortality and cancer incidence in aluminum reduction plant workers J Occup Med 33 1150 1155 1765856
Sutton AJ 2000. Meta-analysis in Medical Research. London:Wiley.
Swaen GM Slangen JJ Volovics A Hayes RB Scheffers T Sturmans F 1991 Mortality of coke plant workers in the Netherlands Br J Ind Med 48 130 135 1998607
Swaen GM Slangen JM 1997 Mortality in a group of tar distillery workers and roofers Int Arch Occup Environ Health 70 133 137 9253642
U.S. EPA 1984. Health Effects Assessment for Benzo(a)pyrene (BaP). Internal Report. Washington, DC:U.S. Environmental Protection Agency.
WHO 1987. Air Quality Guidelines for Europe. WHO Regional Publications, European Series no 23. Copenhagen:WHO Regional Office for Europe.
Xu Z Brown LM Pan GW Liu TF Gao GS Stone BJ 1996 Cancer risks among iron and steel workers in Anshan, China, Part II: case-control studies of lung and stomach cancer Am J Ind Med 30 7 15 8837676
|
15198916
|
PMC1247189
|
CC0
|
2021-01-04 23:39:20
|
no
|
Environ Health Perspect. 2004 Jun 7; 112(9):970-978
|
utf-8
|
Environ Health Perspect
| 2,004 |
10.1289/ehp.6895
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0112-00097910.1289/ehp.690315198917Research ArticleReviewsRationale for a New Generation of Indicators for Coastal Waters Niemi Gerald 1Wardrop Denice 2Brooks Robert 2Anderson Susan 3Brady Valerie 1Paerl Hans 4Rakocinski Chet 5Brouwer Marius 5Levinson Barbara 6McDonald Michael 71Natural Resources Research Institute, University of Minnesota, Duluth, Minnesota, USA2Cooperative Wetlands Center, Pennsylvania State University, University Park, Pennsylvania, USA3Bodega Marine Laboratory, University of California Davis, Bodega Bay, California, USA4Institute of Marine Sciences, University of North Carolina-Chapel Hill, Morehead City, North Carolina, USA5Department of Coastal Sciences, University of Southern Mississippi, Ocean Springs, Mississippi, USA6U.S. Environmental Protection Agency, National Center for Environmental Research, Washington DC, USA7U.S. Environmental Protection Agency, Environmental Monitoring and Assessment Program, Research Triangle Park, North Carolina, USAAddress correspondence to G. Niemi, Natural Resources Research Institute, University of Minnesota, 5013 Miller Trunk Hwy., Duluth, MN 55811-1442. Telephone: (218) 720-4270. Fax: (218) 720-4328. E-mail:
[email protected] thank T. Barnwell for comments on an earlier version of this manuscript. This is contribution number 364 from the Center for Water and the Environment of the Natural Resources Research Institute.
Although the research described in this article has been funded wholly or in part by the U.S. Environmental Protection Agency’s Science to Achieve Results (STAR) program through cooperative agreements (R-82867501, R-82867701, R-82867601, R-82945801) to the University of Minnesota, University of North Carolina, Pennsylvania State University, University of California, and the University of Southern Mississippi, respectively, it has not been subjected to the Agency’s required peer and policy review and therefore does not necessarily reflect the views of the Agency and no official endorsement should be inferred.
The authors declare they have no competing financial interests.
6 2004 11 5 2004 112 9 979 986 9 12 2003 10 5 2004 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. More than half the world’s human population lives within 100 km of the coast, and that number is expected to increase by 25% over the next two decades. Consequently, coastal ecosystems are at serious risk. Larger coastal populations and increasing development have led to increased loading of toxic substances, nutrients and pathogens with subsequent algal blooms, hypoxia, beach closures, and damage to coastal fisheries. Recent climate change has led to the rise in sea level with loss of coastal wetlands and saltwater intrusion into coastal aquifers. Coastal resources have traditionally been monitored on a stressor-by-stressor basis such as for nutrient loading or dissolved oxygen. To fully measure the complexities of coastal systems, we must develop a new set of ecologic indicators that span the realm of biological organization from genetic markers to entire ecosystems and are broadly applicable across geographic regions while integrating stressor types. We briefly review recent developments in ecologic indicators and emphasize the need for improvements in understanding of stress–response relationships, contributions of multiple stressors, assessments over different spatial and temporal scales, and reference conditions. We provide two examples of ecologic indicators that can improve our understanding of these inherent problems: a) the use of photopigments as indicators of the interactive effects of nutrients and hydrology, and b) biological community approaches that use multiple taxa to detect effects on ecosystem structure and function. These indicators are essential to measure the condition of coastal resources, to diagnose stressors, to communicate change to the public, and ultimately to protect human health and the quality of the coastal environment.
coastalecologicestuarinehealthindicatorsmarinenutrientsresponsesstressors
==== Body
More than half the world’s human population resides within 100 km of the coastline [National Oceanic and Atmospheric Administration (NOAA) 1998; Vitousek et al. 1997], with increases likely over the next two decades (Stegeman and Solow 2002). The coastal zone represents at least half the value of global ecologic services (Costanza et al. 1997), and in economic terms, is the single most important source of recreational and residential income worldwide as well as fisheries (Jackson JBC et al. 2001; Ray and McCormick-Ray 2004).
Human development of coastal watersheds has greatly accelerated environmental pressure on downstream estuarine and coastal ecosystems; yet, unfortunately, assessing detrimental changes in these systems is complex. The symptoms of degradation include deterioration of water quality, loss of habitat and biodiversity, beach closings, fishery declines, fish consumption advisories, and an overall decline in the livability in the coastal zone [Boesch et al. 2001; Elofson et al. 2003; Hobbie 2000; Nixon 1995; National Research Council (NRC) 2000; Rabalais et al. 1996; Richardson 1997]. Coastal systems are hydrologically complex and are among the most susceptible to direct disturbance through global climate change. For instance, sea levels have been rising over the past century and even greater rises are predicted over the next 50 years (Jackson RB et al. 2001; Pilkey and Cooper 2004). These changes will continue to affect coastlines and will dramatically increase salt-water intrusion into freshwater coastal aquifers as well as the displacement of coastal agriculture (Jackson RB et al. 2001, McCarthy et al. 2001). Climate change is also believed to have increased events of heavy precipitation and flooding, which recently have become more common. These events increase the flushing of nutrients and toxic chemicals into coastal regions (McCarthy et al. 2001). In addition, many estuaries are highly urbanized and reference conditions are difficult or impossible to specify because of large changes in habitat, water diversion, and the introduction of exotic species (Nichols et al. 1986).
Not only are estuarine systems complex, but many stressors are also difficult to control. For example, non-point source nutrient pollution from coastal watersheds is a major problem, that has affected more than 60% of coastal rivers and bays (Howarth et al. 2000). Increases in nitrogen and phosphorus coming into coastal ecosystems have led to disruptions of basic ecologic functions [e.g., rising frequency and proliferation of harmful algal blooms (HABs) and increasing oxygen depletion (hypoxia)] with major damage to coastal fisheries and biodiversity (Jackson JBC et al. 2001; NRC 2000; Sundareshwar et al. 2003). In addition, nonpoint sources of toxic substances (e.g., agricultural chemicals and urban runoff) have impaired habitat quality for aquatic life and the human use of numerous coastal watersheds (Detenbeck et al. 1999; Kuivila and Foe 1995).
It is increasingly evident that ecosystem and human health are intricately linked (Stegeman and Solow 2002). For example, HABs can cause diseases in humans that result from consumption of contaminated seafood or inhalation of toxins entrapped in sea spray. Moreover, the distribution and frequency of HAB events have increased along U.S. coastlines over the last 30 years (Van Dohla 2000). People are also exposed to waterborne diseases through recreational contact, and the incidence of these diseases is increasing worldwide, as is the cost to the economy of frequent beach closures (Harvell et al. 1999). The principal agents of these diseases are bacteria, viruses, and protists. In the case of bacteria, this includes both native marine organisms (e.g., Vibrio species) and human or animal-derived pathogens from sewage and runoff (Rose et al. 2003). The importance of these connections between coastal conditions and human health was realized by the National Institute of Environmental Health Sciences (Research Triangle Park, NC) and National Science Foundation (Washington, DC), and they subsequently established their Centers for Oceans and Human Health.
Given the importance of coastal systems and the increasing pressure on them, quantitative measures of coastal ecologic conditions are absolutely essential for detection of change as well as for the design of control measures and restoration activities. Many approaches to comprehensive assessment of condition are cost prohibitive; thus, there has been a tendency to use broadly applicable indicators such as water clarity, nutrient or contaminant loads and levels, and various biodiversity measures such as species richness as metrics (NRC 2000; Whittier et al. 2002). Yet measurements of environmental condition are becoming more sophisticated and more applicable across space, time, and biological organization (Cottingham 2002). These new indicators include such metrics as diagnostic photopigments of algal functional groups to assess eutrophication (Paerl et al. 2002, 2003); biochemical and genetic indicators of toxicant exposure and stress (Anderson et al. 1994; Huggett et al. 1992; McCarthy and Shugart 1990); molecular techniques to assess human fecal bacterial distribution (Field et al. 2003); isotopic techniques to evaluate nutrient enrichment (Page 1995); indices of biological integrity or other biological community responses (Karr 1981; Simon 2003); ecosystem and population modeling approaches (Gentile et al. 2001); landscape metrics (DeAngelis et al. 1998; Whittier et al. 2002); and remote sensing techniques to detect large-scale land use impacts and change (Guerschman et al. 2003; Wolter and White 2002). While these indicators represent impressive advancements in both science and technology, there are limitations on their widespread and integrated use (NRC 2000).
Our primary goal in this review is to identify four critical areas in which scientific advancements are needed before improvements can be made in indicator development of coastal regions. In addition, we provide examples of two promising approaches to improvements in indicators. Defining the limitations of previous approaches and developing new approaches was a major goal of the U.S. Environmental Protection Agency (U.S. EPA) Science to Achieve Results Program (STAR) in establishing the Estuarine and Great Lakes (EaGLe) Research Program. The EaGLe investigators (Atlantic, Pacific, Gulf of Mexico, and the Great Lakes coastal areas) realize there is increasing urgency to develop indicators capable of detecting and diagnosing environmental conditions over space and time at cellular, organism, habitat, ecosystem, and regional levels. As the explosion of technical and conceptual advances in various disciplines ranging from molecular biology to ecosystems ecology and from remote sensing to bioinformatics continues to provide new and better tools, the goals of the EaGLe program and the science surrounding them are works in progress.
Limitations of Current Coastal Indicators
Most indicators were designed to provide specific information on local conditions such as water clarity or eutrophication or to provide broad-based snapshots of regional-scale water quality and habitat condition as exemplified by the total maximum daily load or Environmental Monitoring and Assessment Program (EMAP) (NRC 1994; U.S. EPA 1999). From these studies we know that large areas of the U.S. coastal zones are impaired (Bricker et al. 1999; Environment Canada and U.S. EPA 2003; U.S. EPA 2001), and many of the specific stressors within coastal watersheds that contribute to impairments have been identified (U.S. EPA 1999). Traditionally, the focus has been on indicators associated with water quality and toxic substances; however, there are major concerns on many other stressors, including habitat and landscape change, exotic species, global climate change, and pathogens. We focus on some of the limitations of current indicators that must be overcome for future advancements to be made on broad aspects of this problem in the coastal region (Figure 1):
Stress with response.
Most current indicators of coastal condition are not linked with specific stressors; hence, it is unclear what causes the change in the indicator and, therefore, what management solutions should be implemented to affect an improvement.
Multiple stressors.
Stressors in coastal ecosystems are diverse and originate from both anthropogenic and natural perturbations. Most current indicators are incapable of providing diagnostic information and are unable to discern the relative contribution of the various stressors to the observed changes in the indicators.
Space and time.
Sources of stress to coastal environments operate over a range of spatial scales (e.g., square meters to entire landscapes) and time (seconds to decades). Current indicators are typically not explicit in how they relate condition with stressors over these different scales.
Reference conditions.
The interpretation of the condition or change of an indicator is based on a comparison to reference conditions or benchmarks. Frequently, these reference conditions are not quantitatively defined; thus condition or meaning of a change in indicators is subject to considerable interpretation and debate.
Linking Stress with Response
From a management perspective, among the greatest limitations of many indicators of coastal condition is the lack of a link with the cause for change (Suter et al. 2002). In the development of indicators, distinguishing between measurements of disturbance or stress and measurements of ecosystem response is imperative. The terminology in this area of the literature varies considerably, but recent reviews (NRC 2000; U.S. EPA 2003) provide clarification.
In developing stress–response relationships, natural experiments or surveys across a disturbance gradient from relatively pristine to highly disturbed areas are used most often (Karr and Chu 1999). In these situations, it is often difficult to control the stress because multiple factors often vary across any environmental gradient. Thus, inherent limitations in the scope of inference result and must be explicitly recognized. Understanding the response variable is also dependent on the strength of conceptual models used to describe the factors structuring the ecosystem, and on the extent to which anthropogenic disturbances influence that ecosystem. Obviously, detection of a response to a stressor is best accomplished by experiments in which the stressor can be manipulated in frequency, intensity, and duration. Unfortunately, these experiments often lack realism, as they are typically limited to laboratory situations or small-scale field-based mesocosms. However, a combination of a gradient design with field and laboratory experiments can be a powerful approach for the initial phases of indicator development, but as the indicator is applied at larger spatial scales or in uncharacterized sites, additional approaches are needed. Multivariate statistical approaches, novel modeling approaches, and techniques to aggregate indicators according to their use in management may all be valuable approaches. For example, these may permit more realistic diagnosis of stressors as complex as nutrient inputs, pathogen effects, and toxicant bioaccumulation.
A number of existing approaches are available to couple toxicant stress with response. For example, evaluation of the effects of toxic substances on ecosystems can involve multiple approaches: a) comparison of a toxic response or specific dose level of a contaminant to a water quality standard that has been linked to biological effects; b) assessment of environmental effects of multiple toxic substances by using well-characterized individual contaminant exposures; or c) the coupling of physiologic and genetic indicators with environmental chemistry and ecologic responses at multiple spatial and temporal scales. However, these too have their limitations. The first approach is limited when multiple contaminants and multiple stressor types are present. The second approach is primarily limited because often no direct integration of the toxic response and exposure occurs. The result is data correlation, which is also limited to the spatial and temporal scopes of the immediate investigation. The third approach considers multiple stressors and permits direct integration and scaling, but significant challenges remain to fully develop indicators for a range of habitats and model organisms. There are also combinations of these approaches. A simple index of sediment contamination, the Sediment Quality Triad, combines the first two. It provides a framework for analyzing benthic community data, analytical chemistry, and toxicity test data to assess whether a site is affected by toxicants, and is widely used throughout the nation (Long 2000; MacDonald and Ingersoll 2002). Yet, the action levels derived for specific contaminants are often unknown and benthic data are highly variable. Moreover, there is often a lack of specific reference conditions that precludes clear interpretation. The triad approach is also based on acute lethality, whereas sublethal effects can also be very important.
In summary, there is a need for controlled experiments in laboratory and field settings for those stressors (e.g., toxicants, nutrients, turbidity) amenable to manipulation. For larger-scale stressors such as introduction of exotic species, climate change, habitat loss, or landscape change such as fragmentation that are not easily amenable to manipulation, field experimental designs that test responses over gradients of stressor levels are among the options for linking stress with response (Danz et al., in press). Coupling the approaches of laboratory and field methods are essential for future development of appropriate response indicators.
Multiple Stressors of Environmental Condition
Stress on coastal ecosystems is usually a combined effect of natural and anthropogenic disturbances. Natural disturbances in the U.S. coastal zones include water-level fluctuations from droughts and floods, wind events such as hurricanes, natural soil/sediment deposition, insect infestations, and forest fires. These natural disturbances vary in intensity both spatially and temporally. Major anthropogenic disturbances to the watershed of coastal ecosystems include permanent land cover conversions of native vegetation to agricultural, residential, and industrial areas; and temporary conversions of land due to forestry. These conversions result in concomitant disturbance to coastal ecosystems, including a) landscape effects of fragmentation, b) increased surface water runoff, c) increased nutrient and sediment input, d) increased pesticide and other chemical inputs, e) increased water temperature, and f ) greater human disturbance from recreational use, increased fish and shellfish extraction, and noise. Climate change and the resulting change in weather patterns is a combination of both natural, stochastic events, and human-induced warming which affects vegetation, water levels, and virtually all types of disturbance. In arid regions, water diversion and water use patterns also result in landscape and ecosystem-level alterations (Bennett and Moyle 1996). Detecting the effects of both individual disturbances and the simultaneous influences of natural and anthropogenic disturbances in coastal ecosystems is a challenging and complex task.
The relative effects of anthropogenic disturbance must be distinguished from the ranges of variation in natural disturbance regimes, but because of the large size and variability of coastal ecosystems, manipulative experiments to untangle the complexities of the varying disturbance regimes are difficult except on a relatively small scale. Combining specific indicators with modeling efforts can clarify and distinguish anthropogenic from natural stress in individual ecosystems and regions (DeAngelis et al. 1998; Gentile et al. 2001). However, the general trend has moved away from using direct diagnostic measures of stressors to using integrated indicators of ecosystem structure and function (NRC 2000). Yet, we know this approach may be inadequate for many applications because stressors vary among regions, implying that indicators are needed with diagnostic and prognostic capabilities. Indicators, therefore, could be grouped to fit regional needs related not only to assessing condition but also to developing appropriate management responses.
Characterizing the effects of multiple stressors on any ecosystem is among the most challenging tasks facing scientists today because multiple stressors can have synergistic, additive, or antagonistic effects on biological responses. Disentangling the various effects of multiple stressors will likely require a combination of controlled laboratory experiments, large-scale studies over multidimensional gradients of stress, and insightful modeling of ecosystem responses and change.
Spatial and Temporal Explicitness
Ecologic indicators are constructed or selected to assess the condition of ecosystems and to detect environmental change related to human disturbance. Condition is often assessed by documenting the state or rate of ecologic processes such as productivity, respiration, or the structuring of biological communities. Indicators may do this by either measuring those processes directly (such as primary productivity) or inferring process from pattern (such as indices of biotic integrity (IBIs) as descriptors of community structure). Ecologic processes operate over a range of spatial and temporal scales, and the resulting patterns are expressed over varied scales. Hence, the relevant scale of each indicator must be specified to relate pattern to process in the appropriate conceptual model. Levin (1992) stated “the concepts of scale and pattern are ineluctably intertwined. The description of pattern is the description of variation, and the quantification of variation requires the determination of scales.”
Many studies have sought to quantify spatio–temporal patterns across a range of scales. Unfortunately, few have determined whether patterns are consistent across scales or whether related phenomena cross scales (Caldow and Racey 2000). In addition, because of technical, logistical, and financial reasons, most ecologic studies have focused on small systems (e.g., site or plot level) and short periods of time, which in turn has limited the development over large spatial scales (Innes 1998). Alternatives such as top-down approaches do exist. Here, large-scale processes form the basis of inferring process from pattern and are being applied in regional classification schemes (Hawkins et al. 2000).
The need for scale explicitness is complicated by multiple stressors arising from human and natural disturbances. Most ecologic indicators are related to multiple stressors and scales. For example, in a study of littoral macroinvertebrate communities (Johnson and Goedkoop 2002), 23% of the variance in taxonomic composition was associated with habitat factors, but greater spatial scales (riparian, catchment, ecoregion classification) accounted for 24% of the variance. If indicators are to be used effectively in management, it is necessary that we know the relevant scale(s), so that the scale of management actions matches the scale of the phenomena being measured (Hobbs 1998). Experimental approaches that allow for the partitioning of the variance among different stress components and over a hierarchy of spatial scales will be critical in the development of new ecologic indicators.
Reference Conditions
To interpret any set of indicators, one must compare results of monitoring to standards or benchmarks. One of the preferred benchmarks is a reference site or condition. The use of reference sites has become increasingly common as ecologists and managers search for reasonable and scientifically based methods to measure and describe the inherent variability in natural aquatic systems (Kentula et al. 1992; Rheinhardt et al. 1999). As there are likely few places on earth unaffected by anthropogenic disturbances, true reference areas remain elusive. For example, even coastal regions of Greenland or Antarctica have been affected by atmospheric chemical inputs and climate change. Alternatively, even in coastal regions with long histories of human occupation and, hence, anthropogenic disturbances, reference sites can be established using specific estuaries, watersheds, or lotic systems entering the coastal zone. These may represent the best attainable environmental conditions for a specific geographic setting, a historic representation using paleolimnologic data, a simulated reference condition, or a situation where conditions fall within the range of natural variability for the system.
Determining reference condition of a system is highly dependent on the indicators used and the locations where samples were gathered. Benthic indicators will provide different results than fish indicators. Similarly, indicators will be different in large, ephemerally stratified systems (e.g., Chesapeake Bay, Maryland–Virginia; Pamlico Sound, North Carolina; Mobile Bay, Alabama; San Francisco Bay, California; or Green Bay, Wisconsin) compared with smaller, well-flushed systems. For example, phytoplankton growth responses to nutrient enrichment will not be as profound as those for benthic microalgae in well-flushed systems. Here, benthic microalgae may be more sensitive and meaningful indicators of ecosystem response to nutrient enrichment. Indicators of community structure (i.e., diversity indices, keystone species) may gauge ecosystem conditions quite distinct from indicators of function (e.g., primary and secondary production, respiration, and nutrient cycling). IBI, habitat suitability indices, and chemical monitoring are specific examples of indicators that in combination can assess structure, physical–chemical quality, and biological measures of reference condition.
Historical information from a site, such as survey data, paleolimnologic studies, and habitat reconstruction, can be extremely helpful for determining reference conditions. Finding such information, however, can also be very difficult and most historical information is not quantitative such as for urban estuaries (Nichols et al. 1986). Hughes (1995) summarized the basic characteristics necessary for a suitable reference condition including reasonableness and political acceptability, sufficient number of reference sites within the area of interest, and suitable data on natural conditions of the site.
Examples of New Indicators
Environmental indicators can have an enormous number of possible end points, reflecting the breadth and diversity of the scientific underpinnings in biology, chemistry, and physics (McKenzie et al. 1992; Noss 1990; NRC 2000; O’Neill et al. 1988). For example, biological indicators span the realm of biological organization from genetic markers to entire ecosystems. Chemical indicators reflect a variety of spatial or temporal scales ranging from oxygen demand for a specific point source to global carbon dioxide distributions in the atmosphere. Physical indicators can include elevational, morphologic, transport, circulation, exchange and stratification processes with all their attendant ramifications for ecosystem structure and function. Because of the massive amounts of information that can be gathered across levels of physical, chemical, and biological organization and across spatial or temporal scales (Dixit et al. 1992; Karr and Chu 1999), the challenge to integrate data across levels of organization in space and time is daunting.
Current programmatic and academic funding scenarios exacerbate our lack of integration. Most funding is limited by amount and duration. These monetary and time deficiencies have been recognized by academic funding sources, such as NSF (e.g., Long Term Ecological Research, Biocomplexity, Global Ocean Flux, Biotechnology, and other centers) and U.S. EPA (EaGLe). As such programs mature, advances in integrating across a variety of trophic levels as well as organizational or spatial and temporal scales will likely occur.
We provide two brief examples of new types of indicators; one that links productivity (function) and hydrology and another linking community (structural) patterns. We believe these new types of indicators will substantially improve our ability to measure and understand the complexity, response, and condition of coastal systems. For example, analyses of photopigments provide a means to explicitly link nutrient and hydrologic stressors with specific phytoplankton groups and over explicit spatial scales when combined with remote sensing information. If data are gathered systematically over time, then temporal changes can also be linked with specific stress events (e.g., hurricanes). In the second example, multitaxa types of approaches provide a wide range of possibilities for improving our knowledge of stress–response relationships, the identification of multiple stressor effects, spatial and temporal explicitness, and the identification of suitable reference conditions. For example, Luoma et al. (2001) provide eight case studies that specify cause and effect linkages using community analyses of macroinvertebrates to decipher the effects of multiple stressors. Paleolimnologic data derived from sediment and water column sampling of diatom communities are among the most powerful techniques for identifying suitable reference conditions within aquatic systems (Dixit et al. 1992).
Photopigments as Integrators of Estuarine Nutrients and Hydrology
Nitrogen availability most frequently controls microalgal and higher plant primary production in estuarine and coastal waters (Nixon 1995; Ryther and Dunstan 1971). Loading rates of this nutrient directly reflects human population density and activity in coastal water- and airsheds (Peierls et al. 1991). Excessive nitrogen loading is a key causative agent for accelerating primary production or eutrophication (Nixon 1995; Paerl 1997). Symptoms include phytoplankton blooms, which may accumulate as ungrazed organic matter in the sediments, providing the “fuel” for oxygen consumption and depletion in bottom waters and sediments. This chain of events is particularly problematic in salinity- or temperature-stratified waters, where oxygen may not be easily replenished from the atmosphere. Hypoxic conditions alter nutrient cycling and promote fish disease and mortality (Paerl et al. 1998).
Suspended microalgae or phytoplankton account for the bulk of estuarine and coastal primary production in many estuarine and coastal ecosystems. Their composition and activity are key in determining fertility, eutrophication, and water quality. Water discharge controls transport of phytoplankton through these systems and plays an interactive role with nutrient supply to control phytoplankton growth, competition, succession, and community composition. For example, high rates of freshwater discharge reduce the salinity and residence time. These conditions favor fast-growing oligohaline phytoplankton, such as chlorophytes (green algae). In contrast, low-discharge conditions promote long water residence, high salinity conditions, which favor slower growing halophylic taxa such as dinoflagellates and certain cyanobacteria. Phytoplankton community composition affects the structure and function of estuarine food webs, nutrient cycling, habitat condition, fishery resources, and overall ecosystem condition (Paerl et al. 2002, 2003).
Chlorophyll a has been used for many years as a sensitive indicator of phytoplankton biomass. However, because virtually all phytoplankton contain this pigment, it alone cannot be used to determine community composition. Using additional diagnostic chlorophyll and carotenoid photopigments as indicators of major phytoplankton functional groups (i.e., diatoms, dinoflagellates, chlorophytes, cyanobacteria, cryptomonads), we can examine the interactive effects of nutrient and hydrologically driven changes of phytoplankton community composition and activity. HPLC, coupled to photodiode array spectrophotometry is used to determine phytoplankton group composition based on the diagnostic photopigments. Photopigment markers include chlorophyll b and lutein (chlorophytes), zeaxanthin, myxoxanthophyll, and echinenone (cyanobacteria), fucoxanthin (diatoms), peridinin (dinoflagellates), and alloxanthin (cryptomonads). A statistical procedure, ChemTax (Mackey et al. 1996), partitions chlorophyll a (i.e., total microalgal biomass) into the major algal groups to determine the relative and absolute contributions of each group.
Examples from ongoing studies in the Neuse River estuary in North Carolina and Pamlico Sound (1994–present) show that these systems have experienced the combined stresses of anthropogenic nutrient enrichment, droughts (reduced flushing combined with minimal nutrient inputs), and since 1996, elevated hurricane activity (high flushing accompanied by elevated nutrient inputs) (Figure 2). Seasonal and hurricane-induced variations in river discharge, and the resulting changes in flushing rates, and hence, estuarine residence times, have differentially affected phytoplankton taxonomic groups as a function of their contrasting growth characteristics. For example, the relative contribution of chlorophytes, cryptophytes, and diatoms to the total chlorophyll a pool was strongly controlled by periods of elevated river flow (Figure 2). These effects are due to the efficient growth rates and enhanced nutrient uptake rates of these groups. Cyanobacteria, on the other hand, showed greater relative biomass when flushing was minimal (i.e., longer residence times) during the summer.
Further evidence that hydrologic changes have altered phytoplankton community structure is provided by the observed historical trends in dinoflagellate and chlorophyte abundance. Both decreases in the occurrence of winter-spring dinoflagellate blooms and increases in the abundance of chlorophytes coincided with the increased frequency and magnitude of tropical storms and hurricanes since 1996. The relatively slow growth rates of dinoflagellates may have led to their reduced abundance during these high river discharge events. These changes in the phytoplankton community have been linked to altered trophodynamics and nutrient cycling, which subsequently affects fishery habitats and yields.
Diagnostic photopigment analyses are able to detect significant changes in phytoplankton community composition over a broad range of time scales (< 24 hr to decades) and as such are well suited for monitoring programs designed to assess short- and long-term trends in water quality in response to: hydrographic features (circulation, upwelling); nutrient enrichment; climatic; and hydrologic perturbations (floods, droughts). In addition “top down” effects of grazing have been examined using an HPLC-based technique. Finally, these analyses have proven useful as a means of ground-truthing and calibrating remotely sensed estimates of phytoplankton bloom events (Harding et al. 1999; Millie et al. 1997). This coupling of indicator technology with remote sensing enables “scaling up,” namely, mapping the spatial distributions of phytoplankton groups over large geographic areas not amenable to routine field sampling, evaluating the effectiveness of nutrient management strategies, use as an early warning system for blooms of nuisance or toxic species (Millie et al. 1997), and as a sensitive bioindicator of overall water quality conditions (Pinckney et al. 2001; Paerl et al. 2003).
Biological Community Responses as Condition and Change Indicators
Plant and animal community structure and function have been extensively measured to describe the condition of both aquatic and terrestrial systems. The major challenge is how to scale and aggregate the responses of species populations at the site level to reflect conditions of the biological community level for specific taxa or to provide assessments of large scale patterns, such as IBI (Karr 1981), biological species profiles (Simon 2003), multitaxa indices (O’Connor et al. 2000), or indices of environmental integrity (Paul 2003). These approaches hold tremendous potential for assessments of environmental conditions over large landscape or regional areas, as well as for detection of temporal change. However, these indicators will also require considerable development in the areas of a) providing linkages with single and multiple stressors; b) exploration of analytical techniques to integrate and synthesize multiple biological signals from species or functional groups within the biological community; c) parsing these multivariate responses among stressors and over varying spatial scales; and d) providing explicit spatial or temporal scales for the indicators which are consistent with the scales of management actions (Niemi and McDonald, in press).
The strength of the community approach lies in the differential sensitivity of individual species, functional groups (e.g., guilds), or trophic levels to different stressors. Each of these levels can respond differently to stress. For example, O’Connor et al. (2000) found a correlation among many taxa (diatoms, benthos, zooplankton, fish, birds) to the gross condition of lakes, but fish provided the best measure of condition in the near-shore environment. In an experiment on the pesticide effects of mosquito control agents in wetlands, of the zooplankton, aquatic insect, and bird communities studied, only the aquatic insects exhibited a response to treatment (Hershey et al. 1998; Niemi et al. 1999). In this case, aquatic insects were the best indicator of pesticide effects. Niemi and McDonald (in press) provide many examples of responses by different taxa to diverse stressors.
The unique aspect of the community approach is the ability to sample a wide variety of taxa; each of which has a unique life history capable of being disrupted by stress at various scales. All coastal regions are represented by thousands of species including taxa such as bacteria, plankton, macroinvertebrates, fish, vascular and nonvascular plants, amphibians, and birds. Many associations between these taxa and stress exist. For example, diatoms are particularly sensitive to nutrients (Dixit et al. 1992), whereas benthic invertebrates are responsive to sediment contamination in both lakes and estuaries (Bailey et al. 1995, Rakocinski et al. 1997). Fish communities are sensitive to human development (Brazner 1997) and exotic species (Rahel 2000). Wetland vegetation is directly affected by hydrologic modifications such as dikes and road building (Herdendorf 1992). Amphibians are sensitive to water quality in wetlands (Kutka and Bachman 1990). Bird populations are affected by landscape-level habitat change and fragmentation (Robinson et al. 1995). Moreover, many of these taxa have well-established sampling methods, and some have long-term nationwide monitoring programs that are currently still in use (Likens 1989; Robbins et al. 1989). The combination of species- or taxa-specific responses by plant and animal communities to stressors and the availability of extant monitoring programs allows for the partitioning of multiple stress–response relationships.
Probability-based and standardized sampling of communities within specific sites but over large landscapes have proven useful for regional-scale assessments of environmental conditions (Olsen et al. 1999). For example, these approaches have been used for the identification of imperiled systems (Stein et al. 2000); development of biological indicators for the Mid-Atlantic Highlands (Herlihy et al. 1998); and establishing the condition of streams and estuaries through U.S. EPA’s EMAP (U.S. EPA 2002). One important aspect of these large-scale approaches is the development of indicators that can identify areas that have the most severe problems and greatest need of management attention, action, and potentially restoration. Integration of these types of data will be challenging and will require multivariate, integrative approaches of multitaxa biological communities over large-scale landscapes and regions.
Promising new techniques to achieve integration of community measurements over multiple spatial, temporal, and biological scales are evolving such as development of multimetric indices (e.g., Karr 1981; Paul 2003), and statistical techniques (Jongman et al. 1995). These techniques will require coupling with population and ecosystem-based models for aid in the interpretation of stressor risk and alternative management actions (DeAngelis et al. 1998; Gentile et al. 2001). Several large-scale programs have been initiated such as the EaGLe program reported here. For example, Danz et al. (in press) developed a stratified experimental design for the development of environmental indicators in the Great Lakes coastal region. This design was based on the compilation of more than 200 data layers on stress information for 762 coastal units and the identification of gradients of stress for several coastal ecosystem types. Six different taxa (amphibians, birds, diatoms, fish, macroinvertebrates, vegetation) were randomly sampled across these stress gradients to detect differential responses. This work is still in progress but identifies the tremendous potential of spatially explicit public databases in the future development of environmental indicators. With the exponential increase in technological and computational capabilities including molecular techniques, remote sensing technology, modeling and statistical sophistication, data management and storage, and internet communication, analysis of biological communities as indicator signals of environmental condition and change will rapidly advance.
Conclusions
Coastal ecosystems have experienced tremendous stress from natural and anthropogenic influences for hundreds of years, and stress levels are projected to increase in the future. A new generation of ecologic indicators is needed to measure the condition, diagnose stressors, communicate condition to the public, assess potential future status, and evaluate management actions in our coastal regions. Among the many challenges, the development of these indicators will require improvements in our scientific understanding of stress–response relationships, relative contributions of multiple stressors, how stressors operate over different organizational and spatio–temporal scales, and how reference conditions are determined. Fortunately, there is an explosion of ideas and technology that can aid in the multidisciplinary advancement of indicators in our coastal waters.
Figure 1 Conceptual diagram of critical elements in indicator development and the ultimate identification of indicators of condition and change. The sequence reflects multiple stressors that are distributed over different spatio–temporal scales within the coastal environment. The white rectangular boxes represent major limitations in the development of indicators as discussed in this review.
Figure 2 (A) Chlorophyll and carotenoid photopigments are diagnostic of major estuarine phytoplankton groups. Since chlorophyll a is present in each of the groups, it is used to quantify total phytoplankton biomass. The individual carotenoids and some chlorophylls (e.g., chlorophyll b) can be used to distinguish and quantify individual phytoplankton functional groups, using the matrix factorization program ChemTax Mackey 1996). (B) Distributions, in time and space, in the Neuse River estuary of chlorophyll a, contributed by several phytoplankton functional groups dominating primary production between 1994 and 2000. Groups shown here are chlorophytes, cyanobacteria, and dinoflagellates. Values were derived using ChemTax for surface water at midestuarine mesohaline locations sampled by the MODMON program (Neuse River Estuary and Monitoring 2001). Biweekly data were temporally extrapolated along the axis of the estuary, from its freshwater head at New Bern, North Carolina (0 km), to a downstream mesohaline location near the entrance to Pamlico Sound. Freshwater discharge entering the estuary is also shown. The dates are shown of landfall of the four major hurricanes that have significantly affected flow and nutrient enrichment since mid-1996.
==== Refs
References
Anderson S Sadinski W Shugart L Brussard P Depledge M Ford T 1994 Genetic and molecular ecotoxicology: a research framework Environ Health Perspect 112 suppl 12 3 8 7713030
Bailey RC Day KE Norris RH Reynoldson TB 1995 Macroinvertebrate community structure and sediment bioassay results from nearshore areas of North American Great Lakes J Great Lakes Res 21 45 52
Bennett WA Moyle PB 1996. Where have the fishes gone? Factors producing fish declines in the Sacramento estuary. In: San Francisco Bay: the Ecosystem (Hollibaugh JT, ed). San Francisco:Pacific Division of the American Association for the Advancement of Science.
Boesch DF Burreson E Dennision W Houde E Kemp M Kennedy V 2001 Factors in the decline of coastal ecosystems Science 293 629 638 11474098
Brazner JC 1997 Regional, habitat, and human development influences on coastal wetlands and beach fish assemblages in Green Bay, Lake Michigan J Great Lakes Res 23 36 51
Bricker SB Clement CG Pirhalla DE Orlando SP Farrow DRG 1999. National Estuarine Eutrophication Assessment: Effects of Nutrient Enrichment in the Nation’s Estuaries. Silver Spring, MD:National Oceanic and Atmospheric Administration, National Ocean Service, Special Projects Office, and the National Centers for Coastal Ocean Science.
Caldow RWG Racey PA 2000 Large scale processes in ecology and hydrology J Appl Ecol 37 suppl 1 6 12
Costanza R Arge DR De Groot R Farber S Grasso M Hannon B 1997 The value of the world’s ecosystems services and natural capital Nature 387 253 260
Cottingham KL 2002 Tackling biocomplexity: the role of people, tools, and scale BioScience 52 793 799
Danz NP Regal RR Niemi GJ Brady VJ Hollenhorst T Johnson LB In press. Environmentally stratified sampling design for the development of Great Lakes indicators. Environ Monit Assess.
DeAngelis DL Gross LJ Huston MA Wolff WF Fleming DM Comiskey EJ 1998 Landscape modeling for Everglades ecosystem restoration Ecosystems 1 64 75
Detenbeck NE Galatowitsch SM Atkinson J Ball H 1999 Evaluating perturbations and developing restoration strategies for inland wetlands in the Great Lakes basin Wetlands 19 789 820
Dixit AS Dixit SS Smol JP 1992 Algal microfossils provide high temporal resolution of environmental trends Water Air Soil Pollut 62 75 87
Elofson K Folmer H Gren I-M 2003 Management of eutrophicated coastal ecosystems Ecol Econ 47 1 11
Environment Canada and U.S. EPA 2003 State of the Great Lakes 2003. EPA 905-R-03-004. Washington, DC:U.S. Environmental Protection Agency.
Field K Chem EC Dick LK Fuhrman J Griffith J Holden PA 2003 A comparative study of culture-independent, library-independent genotypic methods of fecal source tracking J Water Health 1 181 194 15382723
Gentile JH Harwell MA Cropper WP Jr Harwell CC DeAngelis D Davis S 2001 Ecological conceptual models: a framework and case study of ecosystem management for South Florida sustainability Sci Total Environ 274 231 253 11453300
Guerschman JP Paruelo JM Burke IC 2003 Land use impacts on the normalized difference vegetation index in temperate Argentina Ecol Appl 13 616 628
Harding LW Degobbis D Precali R 1999. Production and fate of phytoplankton: annual cycles and interannual variability. In: Ecosystems at the Land-Sea Margin: Drainage Basin to Coastal Sea, Vol 55, (Malone TC, Malej A, Harding LW Jr, eds). Washington, DC:American Geophysics Union, Coastal and Estuarine Studies, 131–172.
Harvell CD Kim K Burkholder JM Colwell RR Epstein PR Grimes J 1999 Diseases in the ocean: emerging pathogens, climate links, and anthropogenic factors Science 285 1505 1510 10498537
Hawkins CP Norris RH Gerritsen J Hughes RM Jackson SK Stevenson RJ 2000 Evaluation of landscape classifications for biological assessment of freshwater ecosystems; synthesis and recommendations N Am Benthol Soc 19 541 556
Herdendorf CE 1992 Lake Erie coastal wetlands: an overview J Great Lakes Res 18 533 551
Herlihy AT Stoddard JL Johnson CB 1998 The relationship between stream chemistry and watershed land cover data in the Mid-Atlantic region. U.S Water Air Soil Pollut 105 377 386
Hershey AE Lima AR Niemi GJ Regal RR 1998 Effects of Bacillus thuringiensis israelensis (BTI) and methoprene on nontarget macroinvertebrates in Minnesota wetlands Ecol Appl 8 41 60
Hobbie JE ed. 2000. Estuarine Sciences: A Synthetic Approach to Research and Practice. Washington, DC:Island Press.
Hobbs RJ 1998. Managing ecological systems and processes. In: Ecological Scale: Theory and Applications (Peterson DL, Parker VT, eds). New York:Columbia University Press, 459–484.
Howarth R Anderson D Cloern J Elfring C Hopkinson C Lapointe B 2000. Nutrient pollution of coastal rivers, bays, and seas. Issues in Ecology. No. 7. Washington, DC:Ecological Society of America.
Huggett RJ Kimerle RA Mehrle PM JrBergman HL eds. 1992. Biomarkers: Biochemical, Physiological, and Histological Markers of Anthropogenic Stress. Boca Raton, FL:Lewis Publishers.
Hughes RM 1995. Defining acceptable biological status by comparing with reference conditions. In: Biological Assessment and Criteria. Tools for Water Resource Planning and Decision Making (Davis WS, Simon TP, eds). Baco Raton, FL:Lewis Publishers, 31–47.
Innes JL 1998. Measuring environmental change. In: Ecological Scale: Theory and Applications (Peterson DL, Parker VT, eds). New York:Columbia University Press, 429–457.
Jackson JBC Kirby MX Berger WH Bjorndal KA Botsford LW Bourque BJ 2001 Historical overfishing and recent collapse of coastal ecosystems Science 293 629 638 11474098
Jackson RB Carpenter SR Dahm CN McKnight DM Naiman RJ Postel SL 2001. Water in a changing world. Issues in Ecology. No. 9. Washington DC:Ecological Society of America.
Johnson RK Goedkoop W 2002 Littoral macroinvertebrate communities: spatial scale and ecological relationships Freshwater Biol 47 1840 1854
Jongman RHG Ter Braak CJF Van Tongeren OFR 1995. Data Analysis in Community and Landscape Ecology. Cambridge, UK:Cambridge University Press.
Karr JR 1981 Assessment of biological integrity using fish communities Fisheries 6 21 27
Karr JR Chu EW 1999. Restoring Life in Running Waters: Better Biological Monitoring. Washington DC:Island Press.
Kentula ME Brooks RP Gwin SE Holland CC Sherman AD Sifneos JC 1992. Wetlands. An Approach to Improving Decision Making in Wetland Restoration and Creation. Washington, DC:Island Press.
Kuivila KM Foe CG 1995 Concentration, transport and biological effects of dormant spray pesticides in the San Francisco estuary, California Environ Toxicol Chem 14 1141 1150
Kutka FJ Bachman MD 1990 Acid sensitivity and water chemistry correlates of amphibian breeding ponds in northern Wisconsin, USA Hydrobiologia 208 153 160
Levin SA 1992 The problem of pattern and scale in ecology Ecology 73 1943 1967
Likens GE 1989. Long-term Studies in Ecology: Approaches and Alternatives. New York:Springer-Verlag.
Long ER 2000 Degraded sediment quality in U.S. estuaries: a review of magnitude and ecological implications Ecol Appl 10 338 349
Luoma SN Clements WH Gerritsen J Hatch A Jepson P Reynoldson T 2001. Separating stressor influences form environmental variability: eight case studies from aquatic and terrestrial ecosystems. In: Ecological Variability: Separating Natural from Anthropogenic Causes of Ecosystem Impairment (Baird DJ, Burton GA Jr, eds). Pensacola, FL:Society of Environmental Toxicology and Chemistry, 141–177.
MacDonald DD Ingersoll CG 2002. A Guidance Manual to Support the Assessment of Contaminated Sediments in Freshwater Ecosystems. Volume III. Interpretation of the Results of Sediment Quality Investigations. EPA-905-B02-001-C. Chicago:U.S. EPA Great Lakes National Program Office.
Mackey M Mackey D Higgins H Wright S 1996 CHEMTAX—a program for estimating class abundances from chemical markers: application to HPLC measurements of phytoplankton Mar Ecol Prog Ser 144 265 283
McCarthy JF Shugart LR eds. 1990. Biomarkers of Environmental Contamination. Chelsea, MI:Lewis Publishers.
McCarthy JJ Canziani OF Leary NA Dokken DJ White KS eds. 2001. Climate Change 2001: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Third Assessment Report of the Intergovernmental Panel on Climate Change. New York:Cambridge University Press.
McKenzie DH Hyatt DE McDonald VJ 1992. Ecological Indicators, Vol 2. New York:Elsevier Applied Science.
Millie DF Schofield OM Kirkpatrick GB Johnsen G Tester PA Vinyard BT 1997 Detection of harmful algal blooms using photopigment and absorption signatures: a case study of the Florida red-tide dinoflagellate, Gymnodinium breve Limnol Oceanogr 42 1240 1251
Neuse River Estuary Modeling and Monitoring (MODMON) 2001. Scientists Modeling and Monitoring Results. Chapel Hill, NC:North Carolina Department of Environment and Natural Resources and the University of North Carolina Water Resources Research Institute. Available: http://www.marine.unc.edu/neuse/modmon/ [accessed 3 May 2004].
Nichols FH Cloern JE Luoma SN Peterson DH 1986 The modification of an estuary Science 231 567 573 17750968
Niemi GJ Hershey AE Shannon L Hanowski JM Lima A Axler RP 1999 Ecological effects of mosquito control on zooplankton, insects, and birds Environ Toxicol Chem 18 549 559
Niemi GJ McDonald ME In press. Applications of ecological indicators. Annu Rev Ecol Evol Syst.
Nixon SW 1995 Coastal marine eutrophication: a definition, social causes, and future concerns Ophelia 41 199 219
NOAA 1998. Population: Distribution, Density and Growth, by Culliton TJ. NOAA’s State of the Coast Report. Silver Spring, MD:National Oceanic and Atmospheric Administration. Available: http://state_of_coast.noaa.gov/bulletins/html/pop_01/pop.html [accessed 3 June 2003].
Noss RF 1990 Indicators for monitoring biodiversity: a hierarchical approach Conserv Biol 4 355 364
NRC 1994. Review of EPA’s Environmental Monitoring and Assessment Program: Forest and Estuaries Component. Washington, DC:National Academy Press.
NRC 2000. Ecological Indicators for the Nation. Washington, DC:National Academy Press.
O’Connor RJ Walls TE Hughes RM 2000 Using multiple taxonomic groups to index the ecological condition of lakes Environ Monit Assess 61 207 228
Olsen AR Sedransk J Edwards D Gotway CA Liggett W Rathbun S 1999 Statistical issues for monitoring ecological and natural resources in the United States Environ Monit Assess 54 1 45
O’Neill RV Krummel JR Gardner RH Sugihara G Jackson B DeAngelis DL 1988 Indices of landscape pattern Landscape Ecol 1 153 162
Paerl HW 1997 Coastal eutrophication and harmful algal blooms: importance of atmospheric deposition and groundwater as new nitrogen and other nutrient sources Limnol Oceanogr 42 1154 1165
Paerl HW Dyble J Twomey L Pinckney JL Nelson J Kerkhof L 2002 Characterizing man-made and natural modifications of microbial diversity and activity in coastal ecosystems Anton van Leeuw 81 487 507
Paerl HW Pinckney JL Fear JM Peierls BL 1998 Ecosystem responses to internal and watershed organic matter loading: consequences for hypoxia in the eutrophying Neuse River estuary, North Carolina, USA Mar Ecol Prog Ser 166 17 25
Paerl HW Valdes LM Pinckney JL Piehler MF Dyble J Moisander PH 2003 Phytoplankton photopigments as indicators of estuarine and coastal eutrophication BioScience 53 953 964
Page HM 1995 Variation in the natural abundance of 15 N in the halophyte, Salicornia virginica , associated with ground-water subsidies of nitrogen in a southern California salt marsh Oecologia 104 181 188
Paul JF 2003 Development and applying an index of environmental integrity for the U.S. Mid-Atlantic region Environ Manage 67 175 185
Peierls BL Caraco NF Pace ML Cole JJ 1991 Human influence on river nitrogen Nature 350 386 387
Pilkey OH Coooper JAG 2004 Society and sea level rise Science 303 1781 1782 15031484
Pinckney JL Richardson TL Millie DF Paerl HW 2001 Application of photopigment biomarkers for quantifying microalgal community composition and in situ growth rates Org Geochem 32 585 595
Rabalais NN Turner RE Justice D Dortch Q Wiseman WJ Jr Gupta BK 1996 Nutrient changes in the Mississippi River and system responses on the adjacent continental shelf Estuaries 19 386 407
Rahel FJ 2000 Homogenization of fish faunas across the United States Science 288 854 856 10797007
Rakocinski CF Brown SS Gaston GR Heard RW Walker WW Summers JK 1997 Macrobenthic responses to natural and contaminant-related gradients in northern Gulf of Mexico estuaries Ecol Appl 7 1278 1298
Ray GC McCormic-Ray J 2004. Coastal-marine Conservation: Science and Policy. Malden, MA:Blackwell Publishing.
Rheinhardt RR Rheinhardt MC Brinson MM Faser KE Jr 1999 Application of reference data for assessing and restoring headwater ecosystems Ecol Restoration 7 241 251
Richardson K 1997 Harmful or exceptional phytoplankton blooms in the marine ecosystem Adv Mar Biol 31 302 385
Robbins CS Sauer JR Greenberg RS Droege S 1989 Population declines in North American birds that migrate to the Neotropics Proc Natl Acad Sci 86 7658 7662 2798430
Robinson SK Thompson FR III Donovan TM Whitehead DR Faaborg J 1995 Regional forest fragmentation and the nesting success of migratory birds Science 267 1987 1990 17770113
Rose KA Murphy CA Diamond SL Fuiman LA Thomas P 2003 Using nested models and laboratory data for predicting population effects of contaminants on fish: a step toward a bottom-up approach for establishing causality in field studies Hum Ecol Risk Assess 9 231 257
Ryther JH Dunstan W 1971 Nitrogen, phosphorus, and eutrophication in the coastal marine environment Science 171 1008 1112 4993386
Simon TP ed. 2003. Biological Response Signatures, Indicator Patterns Using Aquatic Communities. Boca Raton, FL:CRC Press.
Stegeman J Solow AR 2002 Environmental health and the coastal zone Environ Health Perspect 110 660 661
Stein BA Kutner LS Adams JS eds. 2000. Precious Heritage—the Status of Biodiversity in the United States. Oxford:Oxford University Press.
Sundareshwar PV Morris JT Koepfler EK Fornwalt B 2003 Phosophorus limitation of coastal ecosystem processes Science 299 563 565 12543975
Suter GW II Norton SB Cormier SM 2002 A methodology for inferring the causes of observed impairments in aquatic ecosystems Environ Toxicol Chem 21 1101 1111 12069293
U.S. EPA 1999. Total Maximum Daily Load (TMDL) Program. Washington DC:U.S. Environmental Protection Agency. Available: http://www.epa.gov/OWOW/tmdl [accessed 11 May 2004].
U.S. EPA 2001. Clean Water Action Plan: National Coastal Condition Report. EPA620-R-00-004. Washington, DC:U.S. Environmental Protection Agency. Available: http://www.cleanwater.gov [accessed 30 October 2003].
U.S. EPA 2002. Environmental Monitoring and Assessment Program Research Strategy. EPA 620/R-02/002. Research Triangle Park, NC:U.S. Environmental Protection Agency.
U.S. EPA 2003. A SAB Report: A Framework for Assessing and Reporting on Ecological Condition. EPA-SAB-EPEC-02-009A. Washington, DC:U.S. Environmental Protection Agency.
Van Dohla FM 2000 Marine algal toxins: origins, health effects, and their increased occurrence Environ Health Perspect 108 suppl 1 133 141 10698729
Vitousek PM Mooney HA Lubchenko J Mellilo JM 1997 Human domination of earth’s ecosystem Science 277 494 499
Whittier TR Paulsen SG Larsen DP Peterson SA Herlihy AT Kaufmann PR 2002 Indicators of ecological stress and their extent in the population of northeastern lakes: a regional assessment BioScience 52 235 247
Wolter PT White MA 2002 Recent forest cover type transitions and landscape structural changes in northeast Minnesota Landscape Ecol 17 133 155
|
15198917
|
PMC1247190
|
CC0
|
2021-01-04 23:39:45
|
no
|
Environ Health Perspect. 2004 Jun 11; 112(9):979-986
|
utf-8
|
Environ Health Perspect
| 2,004 |
10.1289/ehp.6903
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0112-00098710.1289/ehp.694115198918Research ArticleReviewsRecent Developments in Low-Level Lead Exposure and Intellectual Impairment in Children Koller Karin 1Brown Terry 1Spurgeon Anne 2Levy Len 11Medical Research Council Institute for Environment and Health, University of Leicester, Leicester, United Kingdom2Institute of Occupational Health, University of Birmingham, Birmingham, United KingdomAddress correspondence to K. Koller, MRC Institute for Environment and Health, University of Leicester, 94 Regent Rd., Leicester LE1 7DD, UK. Telephone: 44 0 116 223 1629. Fax: 44 0 116 223 1601. E-mail:
[email protected] authors are indebted to R. Canfield, Cornell University, Ithaca, New York, USA, for providing further details of his study.
We gratefully acknowledge the support of the U.K. Department for Environment, Food and Rural Affairs.
The authors declare they have no competing financial interests.
6 2004 28 4 2004 112 9 987 994 18 12 2003 28 4 2004 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. In the last decade children’s blood lead levels have fallen significantly in a number of countries, and current mean levels in developed countries are in the region of 3 μg/dL. Despite this reduction, childhood lead poisoning continues to be a major public health problem for certain at-risk groups of children, and concerns remain over the effects of lead on intellectual development in infants and children. The evidence for lowered cognitive ability in children exposed to lead has come largely from prospective epidemiologic studies. The current World Health Organization/Centers for Disease Control and Prevention blood level of concern reflects this and stands at 10 μg/dL. However, a recent study on a cohort of children whose lifetime peak blood levels were consistently < 10 μg/dL has extended the association of blood lead and intellectual impairment to lower levels of lead exposure and suggests there is no safety margin at existing exposures. Because of the importance of this finding, we reviewed this study in detail along with other recent developments in the field of low-level lead exposure and children’s cognitive development. We conclude that these findings are important scientifically, and efforts should continue to reduce childhood exposure. However, from a public health perspective, exposure to lead should be seen within the many other risk factors impacting on normal childhood development, in particular the influence of the learning environment itself. Current lead exposure accounts for a very small amount of variance in cognitive ability (1–4%), whereas social and parenting factors account for 40% or more.
childrencognitive functionintellectual impairmentIQlead exposure
==== Body
The effects of lead poisoning have been known since ancient times. In 200 BC the Greek physician Dioscorides observed that “lead makes the mind give way.” Until the beginning of the 20th century, lead poisoning was viewed largely as an occupational disease of adults. In the 1890s lead paint poisoning in children was first recognized, and childhood lead poisoning is now well documented and persists as a major public health problem throughout the world. Clinical features of acute lead poisoning include abdominal pain and neurologic symptoms of lead encephalopathy including headache and confusion. In severe cases renal failure and convulsions can occur (Lewis 1997), and extremely high levels may lead to coma and death (Meyer et al. 2003b). Features of chronic lead poisoning include behavioral changes, nephritis, and peripheral neuropathy [Lewis 1997; World Health Organization (WHO) 1995]. Children are more vulnerable to lead exposure for three reasons: young children are more at risk of ingesting environmental lead through normal mouthing behaviors (Lanphear et al. 2002), absorption from the gastrointestinal tract is higher in children than adults (Ziegler et al. 1978), and the developing nervous system is thought to be far more vulnerable to the toxic effects of lead than the mature brain (Lidsky and Schneider 2003).
Although there appears to be no dispute about the effects of high levels of lead, there has been uncertainty about the effects of low levels of lead exposure on children’s health. The debate has been particularly heated in the United States (Ferber 2002; Wakefield 2002), where data used to support laws and policies relating to lead exposure have become the subject of a number of lawsuits (Bellinger and Dietrich 2002; Mushak 2002; Needleman 2002; Nelson 2002; O’Dowd 2002; Pinder 2002). A special issue of Archives of Clinical Neuropsychology in 2001 was devoted to the topic of intelligence quotient (IQ) and low-level lead exposure in children. Five groups of scientists were invited to reply to an article by Kaufman (2001a) who posed the question “Do low levels of lead produce IQ loss in children?” (Brown 2001; Hebben 2001; Nation and Gleaves 2001; Needleman and Bellinger 2001; Wasserman and Factor-Litvak 2001). Kaufman argues that parental variables are far more important to a child’s cognitive development than is low-level lead exposure, and that the loss of a few IQ points (if true) is unlikely to have meaningful consequences for society (Kaufman 2001a, 2001b). In contrast, Needleman argues that lead-induced neurotoxicity has a causal role not only in cognitive loss but also in the subsequent development of juvenile delinquency and socially disruptive behavior (Needleman 1995; Needleman and Bellinger 2001; Needleman et al. 2002). These two positions represent the opposite ends of a spectrum of opinion on the relationship between low-level lead exposure and child development.
In contrast, debate in European countries has been muted with an overriding feeling that since the banning of leaded gasoline and lead-containing paints, lead exposure no longer poses a significant environmental threat to health. Publication of a study by Canfield and colleagues in 2003 (Canfield et al. 2003) challenged this view. Their study showed a dose-dependent decline in cognitive function in a cohort of children whose lifetime peak blood levels never rose above the current World Health Organization/Centers for Disease Control and Prevention (WHO/CDC) blood lead level of concern (10 μg/dL) and suggests there is no safety margin at existing exposures. Since its publication in April 2003, the Canfield study has been widely quoted and has extended the debate beyond the United States. With this in mind, the U.K. Department for Environment, Food and Rural Affairs commissioned the Medical Research Council Institute for Environment and Health to examine in detail the findings of Canfield and colleagues and to place their study within the context of other recent developments, not just in the area of low-level lead exposure but also in the wider context of normal childhood development. Our findings form the basis of this review.
Sources of Lead Exposure/Current Blood Lead Levels
The main sources of lead in children’s environments are diet, lead-based paint in older housing, lead in soil and dust from contaminated leaded paint and gasoline, or past and present mining and industrial activity (Mielke 2002; Mielke and Reagan 1998). Exposure from air and waterborne sources has been greatly reduced with the introduction of unleaded gasoline and the replacement of lead water pipes and water tanks with nonlead alternatives. However, lead in soil and dust continues to be a major source of exposure. Indoor floor dust accounts for approximately 50% of a young child’s total lead intake [Institute for Environment and Health (IEH) 1998]. Although dust is a major source of lead intake throughout the first 1–2 years of childhood, lead-contaminated window sills in older housing become an increasingly important source of lead as children become mobile and stand upright.
Blood lead levels peak in children at around 2 years of age, and hand-to-mouth behavior and pica (eating substances not normally eaten e.g., soil or paint chips) are significantly associated with elevated blood lead levels (Lanphear et al. 2002). Children typically ingest < 50 mg/day of soil on average (Stanek and Calabrese 1995). However, in the case of pica, this amount can be ≥ 5 g a day (Mielke and Reagan 1998), and some children have ingested 25–60 g during a single day (Calabrese et al. 1997). Indeed, from the point of view of risk assessment, Calabrese and colleagues urge that soil pica be seen “as an expected, although highly variable, activity in a normal population of young children, rather than an unusual activity in a small subset of the population.” Soil abatement and paint hazard remediation programs have attempted to reduce children’s exposures to lead and other heavy metals, with mixed outcomes (Elias and Gulson 2003; Lanphear et al. 2003).
Children’s blood lead concentrations have fallen substantially in a number of countries in the last few decades, including the United States, Australia, Mexico, Germany, Poland, Sweden, and the United Kingdom (Delves et al. 1996; IEH 1998; Jarosinska and Rogan 2003; Meyer et al. 2003a). By 1999 the geometric mean blood lead for U.S. children 1–5 years of age had fallen from 15 μg/dL in the late 1970s to 2.0 μg/dL. A survey of 774 Swedish children over the period 1995–2001 showed blood lead levels had stabilized at 2 μg/dL at 7–11 years of age (Strömberg et al. 2003). In the United Kingdom, blood lead levels of 584 children measured during 1995 in the Avon Longitudinal Study of Pregnancy and Childhood (ALSPAC) showed a geometric mean of 3.44 μg/dL at 2.5 years of age (Golding et al. 1998). Despite these falls in blood lead levels, childhood lead poisoning continues to be a major public health problem for certain groups of children, specifically low-income, urban, African-American children in the United States (Roberts et al. 2001), children suffering from abuse and neglect (Chung et al. 2001), children living in rural mining communities (Lynch et al. 2000), and children in developing countries (Falk 2003; Fewtrell et al. 2004).
Lowering of exposure guideline levels reflects concern over the growing body of evidence that low levels of lead exposure have subtle effects on the nervous system of children. Since 1971 there have been four reductions in the CDC guideline level above which children are considered to have an elevated lead level. This level currently stands at 10 μg/dL (0.483 μmol/L). In 1997 the CDC estimated that 4.4% of children in the United States 1–5 years of age have blood lead levels ≥ 10 μg/dL (Lynch et al. 2000). In a recent report of blood lead levels in children 6 months to 5 years of age living in New Orleans, Louisiana, USA, 29% had levels ≥ 10 μg/dL (Rabito et al. 2003). In Wuxi City, China, 27% of children 1–5 years of age had blood lead levels > 10 μg/dL (Gao et al. 2001), whereas in Johannesburg, South Africa, the blood lead levels of 78% of schoolchildren ≥ 10 μg/dL (Mathee et al. 2002) and in Dhaka, Bangladesh, 87% of children 4–12 years of age had blood lead levels > 10 μg/dL (Kaiser et al. 2001). In the United Kingdom, large-scale blood lead monitoring programs ceased in the late 1980s, and there is a paucity of recent data on blood lead levels in young children. The proportion of children with blood lead levels > 10 μg/dL ranged from 0.74 to 5% according to recent reports from three different regions of England (IEH 1998; Lewendon et al. 2001; O’Donohoe et al. 1998), and there is growing concern that significant numbers of children under 5 years of age remain at risk from lead exposure in the United Kingdom (Grigg 2004).
Cross-Sectional Studies
Cross-sectional studies form part of a worldwide effort to quantify the effects of lead exposure in children. The main limitation of such cross-sectional studies is that they measure blood lead at one specific time point only. Because the half-life of lead in blood approximates that of the erythrocyte (approximately 35 days), it is primarily an indicator of recent exposure. This is of particular importance with lead exposure, as blood lead levels peak in children at around 2 years of age.
We identified eight recent cross-sectional studies looking at the relationship between blood lead concentrations and children’s cognitive abilities: the large U.S. National Health and Nutrition Examination Survey (NHANES) III (Lanphear et al. 2000) and seven studies from six different countries [Croatia, Denmark, Saudi Arabia, Mexico, Pakistan, and Taiwan (Al-Saleh et al. 2001; Calderón et al. 2001; García-Vargas et al. 2002; Nielsen et al. 2000; Prpic-Majic et al. 2000; Rahman et al. 2002; Wang et al. 2002)]. All studies examined children 6 years of age or older (range 6–16 years) but differed in sample size (80–4,853) and the number of confounders considered. Mean blood lead levels ranged from 2.94 to 9.73 μg/dL. It was unfortunate that the very large NHANES III study (4,853 children) lacked data on two key confounders: home environment and maternal IQ. There was no consistent effect of blood lead levels on cognitive function across these studies, and taken together we believe they add little to the current debate on low lead exposure and its effect on cognition.
Prospective (Longitudinal) Studies
The evidence for lowered intellectual and cognitive ability in children exposed to lead comes largely from prospective epidemiologic studies of cohorts in Boston, Massachusetts, USA; Cincinnati and Cleveland, Ohio, USA; Port Pirie and Sydney, Australia; and Yugoslavia (Factor-Litvak et al. 1999). A number of these studies are still ongoing. The main focus of current debate centers on the difficulties of adjusting for confounders (covariables), which include socioeconomic status (SES), home environment, and genetic factors. SES is measured in a number of ways that generally involve an index derived from data on household income, parents’ education, employment status, occupation, and home ownership. Home environment is frequently measured using the Home Observation for Measurement of the Environment Inventory (HOME) index. This reflects the quality and quantity of emotional and cognitive stimulation and support in the home environment. The total score is the sum of a number of items, each scored as present (1) or absent (0), in various categories: parental responsivity, acceptance of child, organization of home environment, provision of play materials, parental involvement with child, and variety of stimulation.
Longitudinal studies have many advantages over cross-sectional studies: a) the time sequence of events can be assessed, b) they can provide information on a wide range of outcomes, and c) there is reduced recall and selection bias compared with case–control studies. Children’s intellectual capacities change with time, and therefore age-specific tests must be used (i.e., there is no single psychometric test that can cover the entire age range of interest). Unfortunately, in the five ongoing lead/IQ studies identified, a variety of cognitive test instruments were used, even for children of the same age, and no two studies adjusted for the same covariables. It is therefore not possible to directly compare results between these studies.
The Yugoslavia Prospective Lead Study was initiated in 1985. Pregnant women (n = 1,502) living in two towns in Yugoslavia were identified as having differing lead exposures. One town is on the site of a lead smelter, whereas the other (control) town lies 25 miles to the south. Maternal blood lead was measured at midpregnancy and at delivery, and child blood leads were determined at subsequent 6-month intervals. The report by Wasserman and colleagues (2000) includes all children (n = 390) having at least one assessment of intellectual functioning at 3, 4, 5, or 7 years of age with complete data on all covariates. Three normed and age-specific tests of cognitive function were used. This review is a reanalysis of data given in the authors’ full 1999 report, when the study was in its fourteenth year (Factor-Litvak et al. 1999), and examines whether there are critical time periods for the effects of lead exposure on IQ. Data analysis was performed after grouping observations into three exposure–change categories. An increase in postnatal blood lead was defined by a change of 50% or more relative to prenatal levels, with the postnatal period divided into early (0–2 years of age) or late (2–7 years of age). Average prenatal blood lead levels were 10 μg/dL (range 3–30 μg/dL) and average postnatal levels at 2–7 years of age were 17.4 μg/dL (range 6.6–49 μg/dL). The wide ranges are a reflection of pooling data from children living in two towns with very different exposure levels.
Both prenatal (p < 0.001) and postnatal (p < 0.05) exposure were independently and significantly negatively correlated with IQ, and no critical period of vulnerability was found. A 50% rise in prenatal blood lead levels was associated with a 1.07-point loss in IQ score [95% confidence interval (CI), 0.6–1.53], whereas a 50% increase in postnatal blood lead relative to prenatal levels was associated with a 2.82-point IQ loss (95% CI, 0.52–4.91). Because the analyses first controlled for prenatal blood lead, the postnatal change measure indicated a substantial change in exposure and was not a reflection of whether the mean postnatal blood lead was high or low. Covariates included in the regression analysis were quality of the home (HOME score); maternal age, intelligence, education and ethnicity; birth weight; and sex. Together these accounted for approximately 50% of the variance in IQ at 7 years of age; lifetime lead exposure accounted for 4.2% of the variance (Factor-Litvak et al. 1999).
The report by Schnaas and colleagues (2000) forms part of the Mexico City Prospective Lead Study of 436 children born after uncomplicated pregnancy. Intellectual function was measured using the McCarthy Scale, which provides a general index of intellectual ability (General Cognitive Index, GCI); subtests measure both cognitive and motor function. Complete data were obtained for 112 children followed at 6-month intervals between 3 and 5 years of age. Prenatal blood lead measures were recorded at intervals during pregnancy, at delivery, and in cord blood. Average postnatal blood lead levels were calculated for three time periods: 6–18 months, 24–36 months, and 42–54 months. Geometric mean blood lead concentrations were approximately 10 μg/dL during the study period. Covariates used in regression models were maternal IQ, child’s sex, Apgar score at 5 min, birth weight, birth rank order, maternal educational level and IQ, and family SES (no details given). The authors did not include HOME scores, claiming that HOME scores are highly correlated with maternal IQ and it was therefore sufficient to include only maternal IQ in the model. The article is methodologically very complex, with interactions measured between many variables. The central finding is that prenatal log-transformed blood lead levels are not associated with intellectual function, either within or between subjects, whereas postnatal lead levels were significantly correlated with intellectual function. The strength of the association between mean blood lead (6–18 months) and GCI increases with age up to 4 years of age, after which it becomes less strong and decreases toward zero. This study is one of only a few that attempts to examine in detail the temporal pattern of the association of lead levels and intellectual function.
Tong et al. (2000) provide an update on the 375 children born in the lead-smelting city of Port Pirie, South Australia, who have been followed from birth and had reached 11–13 years of age at the time of the study. Previous studies of this cohort had shown that blood lead concentration was negatively associated with cognitive performance, with girls more sensitive to the effects of lead at 2, 4, and 7 years of age (Baghurst et al. 1992; Tong et al. 1996, 1998). Geometric mean blood lead levels in this cohort increased from 8.3 μg/dL at birth to 21.2 μg/dL at 2 years of age and decreased to 7.9 μg/dL at 11–13 years of age. This study explores whether there is any effect modification between lead exposure and key sociodemographic factors on IQ [measured using the Wechsler Intelligence Scales for Children-Revised (WISC-R) instrument] at 11–13 years of age. A large number of covariates were measured. Sociodemographic factors included sex, maternal IQ, HOME scores, and SES (estimated by Daniel’s scale of prestige of occupations in Australia). The cohort was divided into three groups on the basis of lifetime average blood lead levels, with the lowest group < 12 μg/dL and the highest group > 17 μg/dL. The effect of sex became statistically insignificant at 11–13 years of age, and the authors speculate that this may have been due to attrition in numbers. (The original cohort comprised 723 children.) The impact of lead on IQ was more marked in children with lower SES, although this became non-significant after adjusting for covariates. The high-SES children performed significantly better in arithmetic and vocabulary WISC-R subscales than children from poor SES backgrounds. Adjusted regression coefficients showed boys lost 2.6 IQ points (95% CI, 2.9 to –8.0), whereas girls lost 7.4 IQ points (95% CI, –1.7 to –13.1) for each 2.7-fold increase in lifetime average blood lead level.
The study by Emory and colleagues (2003) examined 79 mother–infant pairs who represented an independent sample drawn from a larger population of more than 500 subjects in an ongoing study of lead exposure. Mothers came from an urban cohort of low-SES African Americans in Atlanta, Georgia, USA, and their infants were included in the study if they were born after uneventful pregnancies. Maternal blood lead was measured at 6 months’ gestation and before delivery and compared with infant memory at 7 months, assessed by the Fagan preferential-looking test. This study was noteworthy for its use of more sensitive calibration standards and continual verification reference samples to increase confidence in measuring very low blood lead levels (< 5 μg/dL). Mean maternal blood lead was 0.72 ± 0.86 μg/dL. Umbilical blood lead was measured, but no data were given in the article. Infant Fagan scores were classified as low, medium, or high risk of later mental retardation. Significant negative correlations between maternal blood lead and subsequent infant Fagan ratings were reported. These differences were not related to gestational age, birth weight, or age at testing nor were they related to mother’s education, although it was not stated how this was measured. Overall, these findings should be treated cautiously because of the small numbers in the low- and high-risk groupings, and the lack of detailed information on confounders. The authors acknowledge that their results require replication. This is an ongoing study, and it will be of interest to follow future publications.
The recent article by Canfield and colleagues (2003) is part of this continuing evidence base and relates to a cohort of 240 children born between July 1994 and January 1995, living in Rochester, New York, USA, and enrolled in the Rochester Longitudinal Lead Study. This study population is a nested cohort within a larger group of children and their families who took part in a 24-month randomized dust-control trial published in 1999 (Lanphear et al. 1999). The article by Canfield and colleagues reports on the results of blood lead concentrations measured at 6, 12, 18, 24, 36, 48, and 60 months of age, and IQ determined at 3 and 5 years of age using the Stanford-Binet Intelligence Scale. The relations between blood lead levels and IQ were estimated with a variety of models, with adjustments for nine prespecified covariables: child’s sex, birth weight, iron status, and home environment (HOME scale conducted by face-to-face interview and direct observation within the home) (Canfield R, personal communication); mother’s IQ, years of education, race, and tobacco use during pregnancy; and household income. Adjusting for number of siblings and birth order did not alter the model estimates or significance levels, and these covariates were therefore not included in the secondary analyses (Canfield R, personal communication).
The study reports a significant negative association (p = 0.004) between blood lead levels and IQ, with a 0.46-point decrease in IQ for each microgram per deciliter increase in lifetime average blood lead concentration (lifetime being equivalent to the child’s total exposure over 3 or 5 years). For the subsample of children whose maximal blood lead level remained below 10 μg/dL over the 5 years, the IQ loss associated with a given change in blood lead level was greater. In these 101 children, the study indicates a loss of 0.74 IQ points for each microgram per deciliter increase in lifetime blood concentration. The authors suggest a nonlinear relationship between children’s IQ scores and their blood lead concentration, with larger associations at lower lead concentrations. The importance of this study is that it extends the association of blood lead concentrations and intellectual impairment to concentrations below the current level of concern, which stands at 10 μg/dL (0.483 μmol/L), and implies that there is no safety margin at existing exposures.
To fully evaluate the results of the Canfield study, the experiences of their (nested) study cohort within the original dust-control trial must be considered. The Canfield cohort comprised 240 children from a larger group of 276 children and their families taking part in the dust-control trial. Families were eligible for the dust-control trial if they lived in the city of Rochester and had a child 5–7 months of age at the time of the baseline visit. Participants were identified using sequential lists of live births from three urban hospitals, and families were recruited by telephone. Families who agreed to participate were visited by a study team who carried out a baseline interview and collected a venous blood sample from the child. In addition, an experienced technician collected and analyzed dust samples at various indoor locations and measured lead content of painted surfaces inside and outside the home. This original cohort was randomly divided into an intervention group (n = 140) and a control group (n = 135). Families in the intervention group received cleaning equipment and up to eight visits by a dust-control advisor, although the length of time between visits was not specified. All families continued to be visited by the study team at 6-month intervals for blood sampling and environmental lead measurements by a technician (blinded to intervention status). In addition, at each of these home visits an interviewer (also blinded) conducted a face-to-face interview to identify, among other things, the type and frequency of cleaning and the last time cleaning was performed.
Over the 18-month dust-control study period, there was a 2.6-fold increase in blood lead levels in all children, but no difference in blood lead levels by intervention status (geometric mean level 2.85 μg/dL at 6 months, 7.55 μg/dL at 24 months). House dust lead levels declined sharply in both the intervention and control groups. Six months after the first baseline visit, dust lead levels in interior window sills and on floors had decreased by approximately 50% and continued to decline at a slower rate over the following year. The authors recognized several limitations of the study, including sampling the same location in each house, that is, the act of sampling itself may have introduced an artificial decline in dust lead levels. Another possibility was that the act of sampling altered the cleaning behavior of the control group families (the Hawthorne effect). To examine whether the regular visits and dust sampling introduced such an effect, birth certificate data were used to construct a matched negative-control group of 236 children. Children were matched by race, month of birth, and poverty level (measured by census block group characteristics). At 24 months of age the geometric mean blood lead levels were 7.3 μg/dL (95% CI, 6.6–8.2) in the intervention group, 7.8 μg/dL (95% CI, 6.9–8.7) in the control group, and 7.3 μg/dL ± 2.2 μg/dL (CI not given) in the matched negative controls. No Hawthorne effect was apparent.
If it is assumed that the matched negative-control group lived in homes with dust lead levels equivalent to those found in the study cohort before any interventions (baseline values), house dust lead levels do not appear to correlate with blood lead levels in the study children. This is not discussed in the 1999 article by Lanphear et al. (1999) but highlights the difficulties of accurately measuring lead levels in the personal environment of young children. It is interesting that data on house dust lead levels were not included in a follow-up report on dust control and blood lead levels when these children attained 48 months of age (Lanphear et al. 2000).
Canfield et al. (2003) analyze the original intervention and the control children as a single population, and it is pertinent to ask whether any of the interventions in the original dust-control study may have affected blood lead or IQ levels in their (nested) study group. Although there was no significant difference in blood lead between the two groups from 12–24 months of age, the intervention group mean blood levels were 5–7% lower than the controls. If dust control had altered the variability in blood lead levels, this could have affected the power of the study to look at associations with IQ. The second question to ask is, “Could the up to eight extra visits from the dust-control advisor have resulted in a more stimulating learning environment for children in the intervention group compared with controls?” These families had extra visits by one of two randomly assigned advisors, with the provision and replenishment of cleaning equipment and supplies (brooms, dustpans, sponge mops, buckets, gloves, and detergents). Mean IQs of study children and their mothers were below the national average, commensurate with sample demographics. The IQs of the children were normally distributed, whereas IQs of the mothers were slightly skewed because of a larger than expected number of observations in the 70–75 range (Canfield R, personal communication). Because of low sample numbers no significance should be attached to this finding. If these interventions enhanced the cognitive development of the children, they would have resulted in a shift in the relationship between IQ and blood lead levels, with a higher IQ for a given lifetime blood lead level in half of the study children. The effect of this would be to reduce the overall contribution of lead exposure to intellectual impairment. Therefore, in the context of the results of the Canfield study, the previous experiences of their study cohort might have biased the study findings toward the null, that is, attenuated the association between blood lead level and IQ.
Nonlinear mixed models were analyzed using the full range of blood lead values. Figure 1 illustrates the unadjusted lifetime average blood lead and IQ values. The authors state that the cluster of 10 children with low blood lead levels and high IQ “were not unduly influential in the statistical models,” and regression diagnostics did not identify any outliers in the data. Secondary analyses (using lifetime average blood lead levels) were carried out on the basis of observations with IQ scores < 110 (Canfield R, personal communication). For the full model this eliminated the 16 highest IQ scores. The overall linear regression coefficient for the remaining subgroup was –0.44 (p = 0.005), which is not significantly different from the coefficient of –0.46 for all 172 children. For the group of children with a peak blood lead concentration < 10 μg/dL, 15 observations were eliminated by using the IQ < 110 cutoff. In this case the linear regression coefficient was –1.07 (p = 0.038), which again is not significantly different from the coefficient of –1.37 for all children with a peak blood lead < 10 μg/dL. However, after eliminating these observations, the p-value decreased from 0.05 to 0.08 in the quadratic model.
It would be of interest to know a) if removal of data from the 140 children in the original dust-control intervention cohort alters the semiparametric analysis relationship given in Figure 1, and b) to which group the cluster of 10 children with high IQs (> 115) and low blood lead levels (< 5 μg/dL) were assigned in the original study. In addition, data from the nine children with the highest blood lead levels may have had a disproportionate influence on the final slope of the curve compared with subjects clustered around the average blood lead level, and further information on these nine children would also be of interest.
Cognitive function was assessed using an abbreviated Stanford-Binet Intelligence Scale (version IV) at 3 and 5 years of age, with a different examiner administering the test at each age. Results are expressed as the composite score. However, this test may not have been the most accurate measure of IQ for this cohort. The Stanford-Binet is heavily weighted on verbal skills and has been superseded by the Wechsler scales for this reason. Anyone who lacks English proficiency will do less well in this test, and children were correctly excluded from analysis if their parents lacked English proficiency. Overall, the study children had below-average Stanford-Binet scores (89.8 ± 11.4). However, the standard method for calculating the composite score excludes subtests with a raw score of zero, and thus overestimates IQ in those children achieving a zero in any subtest. The Stanford-Binet IV score at 3 years of age does not correlate well with Wechsler Preschool and Primary Scale of Intelligence (WPPSI) scores at 4–5 years of age, but correlation is significantly improved by considering the number of subtests the child did not perform at 3 years of age (Grunau et al. 2000). The power of this study would therefore have been increased if the children had been assessed using the WPPSI test or if the authors had considered the number of zero-scored subtests in their analysis. However, if it is assumed that children in the Canfield study achieving a zero in any subtest are those with below-average IQs, the overall effect would have been to introduce a differential error in the estimates of IQ, that is, overestimating the IQ scores of children with higher lead levels. This would have biased (toward the null) the estimate of the slope of the relationship between blood lead and IQ and would have reduced the nonlinearity observed in Figure 1.
In correspondence after publication of the Canfield study, Bellinger and Needleman (2003) reanalyzed data from their prospective Boston cohort study, focusing on 48 children whose blood lead levels never exceeded 10 μg/dL at birth, 6, 12, 18, 24, 57, or 120 months. The regression coefficient was greater (–1.56) than that derived from analyses of children with peak blood lead levels > 10 μg/dL (–0.58), that is, their results replicated those of Canfield and colleagues. This reproducibility is of particular interest because the Boston cohort (high SES, average IQ of 105 at 2–4 years of age, mean blood lead 6.5 μg/dL at 2 years of age) was in many respects very different from the Rochester cohort (low SES, average IQ of 90 at 3 years of age, mean blood lead 9.7 μg/dL at 2 years of age). The authors conclude that residual confounding probably accounts for at least some of the disparity between the regression coefficients above and below 10 μg/dL, and because of this “the precise shape of the dose–effect relation at lead levels below 10 μg/dL remains uncertain” (Bellinger and Needleman 2003).
In summary, of the three most recent longitudinal studies that measured prenatal lead exposure (average blood lead ranging from 1 to 11.5 μg/dL), two found a negative association with subsequent IQ, and one found no effect. In contrast, all five recent longitudinal studies that measured postnatal exposure (average blood lead levels ranging from 6 to 44 μg/dL) found significant associations with cognitive development, and this association was maintained after adjusting for a range of covariates including child’s sex and birth weight and parental/maternal IQ and years of education. The Port Pirie and Rochester studies considered the widest range of confounding factors and were the most robust methodologically. With the report of Canfield and colleagues and the recalculation of the Boston cohort results, these findings in nearly 1,300 children support an association between childhood lead exposure and subsequent cognitive impairment and extend the range of concern to children with lifetime average blood lead levels < 10 μg/dL.
Discussion
Epidemiologic studies are subject to two types of error: systematic and random. Systematic errors (or bias) are by far the most problematic as they are generally not measured and they do not decrease as the sample size increases. Key sources of bias include those associated with aspects of selection and the distortion of the cause–effect relation by confounding. Reasons for the controversy over the lead–IQ link include a) the large number of confounders that must be considered when measuring an effect on children’s intelligence; and b) the frequent finding that the more covariates included in regression models, the smaller the effect of blood lead on IQ becomes, although it remains in the same direction (WHO 1995). The most important confounders are SES, parental IQ, and the quality of the home environment. Other factors associated with both IQ and blood lead levels include sex, nutritional status, and parental smoking behavior.
Three studies shed light on the area of confounding (Needleman and Bellinger 2001; Tong and Lu 2001;Wasserman and Factor-Litvak 2001). Blood lead levels have been negatively and positively associated with SES. Because of the sociodemographics and geography of Boston, Massachusetts, USA, increased prenatal lead levels were found in children of higher SES, and after adjustment for covariates the association of lead with IQ loss increased (Needleman and Bellinger 2001). This effect was also seen in the Yugoslavia prospective lead study in which children living near a smelter were from higher SES backgrounds than those living in a nearby control town with lower lead exposure (Wasserman and Factor-Litvak 2001). A study on the identification of confounders in the Port Pirie cohort study (Tong and Lu 2001) found that the size of the relationship between blood lead levels and mean IQ scores decreased by up to 40% when adjustment was made for 4 confounders but by less than 10% when a further 10 confounders were added to the regression models. The four main covariates were SES, quality of home environment, maternal IQ, and parental smoking behavior. The 10 extra confounders, which had little effect individually, included age, sex, birth weight, birth rank, maternal age, number of siblings, and duration of breastfeeding (Tong and Lu 2001).
Bellinger (2000) has argued that factors such as SES and sex should not be viewed solely as confounders but as effect modifiers as well. Unlike confounding, effect modification is a true characteristic of the association between an exposure and its end point. An example is the association between alcohol consumption and blood pressure, which varies in size depending on the modifying effects of the age, sex, and smoking status of the individual. Using data from the Boston prospective lead study, Bellinger showed that children from the lower half of the social class distribution demonstrated a decrease in performance at lower blood lead levels than children in the upper half of the distribution. However, this protective effect of higher SES was lost in the group of children with the highest cord blood lead levels. The author’s hypothesis is that at a given exposure level the magnitude of the estimated effect varies depending on the individual’s location on the social class continuum (Bellinger 2000). This idea is not new. In 1984 Winneke and Krämer (1984) showed the protective effects of higher SES on visual-motor performance deficits in lead-exposed children and concluded,
the common practice of merely removing the effects of confounding factors, such as SES, appears doubtful. . . . In addition, some of the inconsistencies in this area of research might be due to differential sampling of subgroups of lead-exposed children characterized by different levels of psychosocial adversity.
Intuitively, Bellinger’s hypothesis is very attractive and provides a possible explanation for the variability between ostensibly similar studies. In the lead field in the past, study results have been deemed right or wrong, usually on the basis of how the issue of confounding was handled. If dose–effect relationships are not independent of other host characteristics, it will be necessary to model three (or more)-way interactions. However, most prospective studies are designed with only enough statistical power to detect main effects and do not have the power to detect effect modification in subgroups of the main cohort. Bellinger urges a move away from broad, population-based cohorts toward a greater use of focused sampling frames, which should include adequate numbers within specific subgroups (Bellinger 2000). The report on the Port Pirie cohort at 11–13 years of age supports the hypothesis that children from socially disadvantaged backgrounds are apparently more sensitive to the effects of lead than children from higher SES families (Tong et al. 2000).
The powerful influence of SES on developmental outcome has been elegantly demonstrated in a report on school-age children born to mothers with heroin dependency (Ornoy et al. 2001). The study followed children born to mothers with heroin dependency raised at home or adopted at a very young age. These children were compared with groups of control children with average SES, children raised in families with a heroin-dependent father, or children born in families with low SES and high environmental deprivation. The children with environmental deprivation or raised at home by parents with heroin dependency had the lowest intellectual achievements. The adopted children had normal scores on the verbal WISC-R and on the Bender test, as well as normal reading and arithmetic skills, although they had a higher rate of attention deficit hyperactivity disorder than control children. Ornoy later extended this work to include two other high-risk cohorts: children born to mothers with diabetes and children born prematurely with low birth weights. Again he was able to demonstrate that a good home environment had a strong influence on subsequent intellectual abilities but not on motor skills or attention span (Ornoy 2003).
Animal models using spatial learning in rats have shown the protective effect of an enriched environment on lead-induced neurotoxicity (Schneider et al. 2001). Of particular relevance is a recent report on rats exposed to low levels of lead during early development, that is, from birth to weaning at day 21. This exposure produced a lasting deficit in spatial learning that could be completely reversed by raising the rats in an enriched environment after weaning. This reversal was accompanied by nerve growth factor gene induction and recovery of deficits in hippocampal glutamate receptor gene expression (Guilarte et al. 2003).
Genetic predisposition can also affect vulnerability to lead-induced neurotoxicity; this area of research has recently been reviewed by Lidsky and Schneider (2003). Three genes are currently believed to play a role in lead neurotoxicity: the ALAD gene, which codes for δ-aminolevulinic acid dehydratase; the vitamin D receptor (VDR) gene; and the hemochromatosis gene coding for a defective protein known as HFE. There are two forms of the ALAD protein, ALAD1 and ALAD2; lead has a higher affinity for ALAD2. Preliminary evidence has shown adolescents with the ALAD1 phenotype are more resistant to the effects of lead on behavior and attention than ALAD2 individuals. There are at least two alleles (b and B) and three variants of the VDR genotype, and among adults occupationally exposed to lead, b individuals have higher lead levels in blood and bone. Mutated HFE protein is known to cause hemochromatosis, in which large quantities of iron are deposited in internal organs. Because lead can be incorporated into processes requiring iron, polymorphisms in HFE might be expected to influence lead absorption. It is likely that future epidemiologic studies will include analysis of ALAD status and possibly other biomarkers.
Generally, no single epidemiologic study should be treated as the sole source of convincing evidence. The weight of evidence for any causal link comes when a number of studies using similar or preferably different methodologies in different populations reveal the same finding. In the low-level lead–IQ link, the balance has come down in favor of an association, with the methodologically sound study by Canfield et al. (2003) indicating that these effects are seen at peak blood lead levels below 10 μg/dL. Having established a valid association, the use of a number of the Bradford Hill criteria can assist in making causal inferences: temporal relationship (Does lead exposure precede the effect on cognition?), biological plausibility (Are there neurotoxic mechanisms to explain the effect of lead on cognition?), and biological gradient and strength (Is there a dose–response relationship, and if so, how strong is it?).
Evidence is increasing for a temporal relationship. The finding that 4–5 years of age is the critical period for manifestation of earlier (postnatal) lead exposure (Schnaas et al. 2000) might explain the wide variability in effects reported in cross-sectional studies that only looked at children 6 years of age or older. Further support for the critical period comes from the finding by Rogan et al. (2001) that chelation therapy in lead-poisoned children has no beneficial effect when given at 4–6 years of age.
Mechanistically, no unifying theory explains the neurotoxicity of lead or how lead might affect cognition. The ability of lead to substitute for calcium is a common factor underlying many of its toxic actions, including apoptosis and influences on neurotransmitter storage and release, second messengers, cerebrovascular endothelial cells, and glial cells. A variety of mechanisms may be important, and these are summarized in recent reviews by De Gennaro (2002) and Lidsky and Schneider (2003). Lead activates calmodulin, calcineurin, and protein kinase C at very low doses (Deng and Poretz 2002; Kern and Audesirk 2000). Glutamate receptors are thought to be involved in mediation of learning and memory, and changes in N-methyl-d-aspartate glutamate receptor subunits are observed in animals that show cognitive deficits induced by exposure to lead (Lau et al. 2002; Nihei and Guilarte 2001). Lead-induced decreases in hippocampal neurotrophic factor gene expression in rats can be reversed by raising the animals in an enriched environment (Schneider et al. 2001).
Concerning dose–response relationships, IQ tests are blunt measures of neurologic status, and blood lead is at best only a crude index of lead-induced neurotoxicity. However, a negative association has been found across groups of children from a range of populations around the world. Visual-motor tests and tests of attention are designed to assess more limited cognitive domains than IQ tests, and it is of interest that more consistent decreases have been reported for these measures in cross-sectional studies (Al-Saleh et al. 2001; Calderón et al. 2001; Walkowiak et al. 1998) and prospective studies (Dietrich et al. 1993; Tong et al. 1996; Wasserman et al. 1997). The bluntness of IQ tests to measure cognitive function is underlined by a study on five children who underwent left temporal lobectomy for epilepsy. Each patient experienced significant language-related cognitive loss after surgery, and these losses were clinically evident in four of the five patients. However, IQ testing alone did not reliably identify these deficits. Only one child showed a loss of verbal IQ; the other four children showed increases in verbal IQ (Dlugos et al. 1999).
It is clear that blood lead levels have fallen significantly over the last 40 years. During the 1970s, childhood blood lead concentrations of 40 μg/dL were not unusual. The available evidence suggests that mean blood lead levels are now in the range 2–4 μg/dL in the United States and much of Europe. Despite this reduction in lead exposure, it could be argued that current baseline blood lead levels continue to constitute a global public health risk, as preindustrial humans are estimated to have had 100- to 1,000-fold lower blood lead levels than the population of today (Owen and Flegal 1998). With the recent evidence demonstrating an inverse association between blood lead levels and cognitive function in children exposed to low levels of lead, there is no safety margin at existing exposures. Clearly, efforts must continue to minimize childhood exposure. However, we would urge that these efforts be seen in perspective. The magnitude of the lead–IQ dose–response relationship is small on a population basis and should be set against the far greater combined effect of SES status and quality of the caregiving environment. It has been argued that, instead of “chasing after an ever-receding lead threshold,” attention and funds should be focused on “the more complex social ills that are associated with continued lead exposure in a small segment of the population” (Gee and McKay 2002). Current lead exposure accounts for a very small amount of variance in cognitive ability (1–4%), whereas covariates such as social and parenting factors account for 40% or more (Weiss 2000).
Figure 1 Intelligence quotient as a function of lifetime average blood lead concentration. Data were modified from Canfield et al. (2003).
==== Refs
References
Al-Saleh I Nester M DeVol E Shinwari N Munchari L Al-Shahria S 2001 Relationships between blood lead concentrations, intelligence, and academic achievement of Saudi Arabian schoolgirls Int J Hyg Environ Health 204 165 174 11759160
Baghurst PA McMichael AJ Wigg NR Vimpani GV Robertson EF Roberts RJ 1992 Environmental exposure to lead and children’s intelligence at the age of seven years. The Port Pirie Cohort Study N Engl J Med 327 1279 1284 1383818
Bellinger DC 2000 Effect modification in epidemiologic studies of low-level neurotoxicant exposures and health outcomes Neurotoxicol Teratol 22 133 140 10642122
Bellinger DC Dietrich KN 2002 Ethical challenges in conducting pediatric environmental health research [Introduction] Neurotoxicol Teratol 24 443 12127883
Bellinger DC Needleman HL 2003 Intellectual impairment and blood lead levels [Letter] N Engl J Med 249 5 500 502 12890850
Brown RT 2001 Behavioral teratology/toxicology: how do we know what we know? Arch Clin Neuropsychol 16 389 402 14590170
Calabrese EJ Stanek EJ James RC Roberts SM 1997 Soil ingestion: a concern for acute toxicity in children Environ Health Perspect 105 1354 1358 9405323
Calderón J Navarro ME Jimenez-Capdeville ME Santos-Diaz MA Golden A Rodriguez-Leyva I 2001 Exposure to arsenic and lead and neuropsychological development in Mexican children Environ Res 85 69 76 11161656
Canfield RL Henderson CRJ Cory-Slechta DA Cox C Jusko TA Lanphear BP 2003 Intellectual impairment in children with blood lead concentrations below 10 μg per deciliter N Engl J Med 348 1517 1526 12700371
Chung EK Webb D Clampet-Lundquist S Campbell C 2001 A comparison of elevated blood lead levels among children living in foster care, their siblings, and the general population Pediatrics 107 5 E81 11331731
De Gennaro LD 2002 Lead and the developing nervous system Growth Dev Aging 66 43 50 12212631
Delves HT Diaper SJ Oppert S Prescott-Clarke P Periam J Dong W 1996 Blood lead concentrations in United Kingdom have fallen substantially since 1984 Br Med J 313 883 884 8870600
Deng W Poretz RD 2002 Protein kinase C activation is required for the lead-induced inhibition of proliferation and differentiation of cultured obligodendroglial progenitor cells Brain Res 929 87 95 11852034
Dietrich KN Berger OG Succop PA 1993 Lead exposure and the motor development status of urban six-year-old children in the Cincinnati Prospective Study Pediatrics 91 301 307 7678702
Dlugos DJ Moss EM Duhaime AC Brooks-Kayal AR 1999 Language-related cognitive declines after left temporal lobectomy in children Pediatr Neurology 21 1 444 449
Elias RW Gulson B 2003 Overview of lead remediation effectiveness Sci Total Environ 303 1 13 12568760
Emory E Ansari Z Pattillo R Archibold E Chevalier J 2003 Maternal blood lead effects on infant intelligence at age 7 months Am J Obstet Gynecol 188 S26 S32 12712133
Factor-Litvak P Wasserman G Kline JK Graziano J 1999 The Yugoslavia Prospective Study of environmental lead exposure Environ Health Perspect 107 9 15 9872712
Falk H 2003 International environmental health for the pediatrician: case study of lead poisoning Pediatrics 112 259 264 12837919
Ferber D 2002 Toxicology. Overhaul of CDC panel revives lead safety debate Science 298 732 12399564
Fewtrell LJ Prüss-Ustün A Landrigan P Ayuso-Mateos JL 2004 Estimating the global burden of disease of mild mental retardation and cardiovascular diseases from environmental lead exposure Environ Res 94 120 133 14757375
Gao W Li Z Kaufman RB Jones RL Wang Z Chen XZ 2001 Blood lead levels among children aged 1 to 5 years in Wuxi City, China Environ Res 87 1 11 19 11534960
García-Vargas GG Rubio-Andrade M Casillas E Vera-Aguilar E Cebrián ME 2002 Cognitive effects in lead exposed children from urban areas in the region Lagunera, Mexico [Abstract] Toxicologist 66 128
Gee A McKay C 2002 Childhood blood lead and cognition [Letter] J Toxicol Clin Toxicol 40 519 520 12217007
Golding J Smith M Delves HT Taylor H 1998. The ALSPAC study on lead in children. In: Recent UK Blood Lead Surveys, Report R9 (Gompertz D, ed). Leicester, UK:MRC Institute for Environment and Health, 35–39.
Grigg J 2004 Environment toxins; their impact on children’s health Arch Dis Child 89 244 250 14977703
Grunau RE Whitfield MF Petrie J 2000 Predicting IQ of biologically “at risk” children from age 3 to school entry: sensitivity and specificity of the Stanford-Binet Intelligence Scale IV J Dev Behav Pediatr 2 401 410 11132790
Guilarte TR Toscano CD McGlothan JL Weaver SA 2003 Environmental enrichment reverses cognitive and molecular deficits induced by developmental lead exposure Ann Neurol 53 50 56 12509847
Hebben N 2001 Low lead levels and neuropsychological assessment: let us not be misled [Commentary] Arch Clin Neuropsychol 16 353 357 14590167
IEH 1998. Recent Blood Lead Surveys. Report R9. Leicester, UK:MRC Institute for Environment and Health.
Jarosinska D Rogan WJ 2003 Preventing lead poisoning in children: can the US experience inform other countries? The case of Poland Cent Eur J Public Health 11 192 197 14768781
Kaiser R Henderson AK Daley WR Naughton M Khan MH Rahman M 2001 Blood lead levels of primary school children in Dhaka, Bangladesh Environ Health Perspect 109 6 563 566 11445508
Kaufman AS 2001a Do low levels of lead produce IQ loss in children? A careful examination of the literature Arch Clin Neuropsychol 16 303 341 14590165
Kaufman AS 2001b How dangerous are low (not moderate or high) doses of lead for children’s intellectual development? Arch Clin Neuropsychol 16 403 431 14590171
Kern M Audesirk G 2000 Stimulatory and inhibitory effects of inorganic lead on calcineurin Toxicology 150 171 178 10996673
Lanphear BP Dietrich KN Auinger P Cox C 2000 Cognitive deficits associated with blood lead concentrations < 10 μg/dL in US children and adolescents Public Health Rep 115 521 529 11354334
Lanphear BP Hornung R Ho M Howard CR Eberle S Knauf K 2002 Environmental lead exposure during early childhood J Pædiatr 140 40 47
Lanphear BP Howard C Eberly S Auinger P Kolassa J Weitzman M 1999 Primary prevention of childhood lead exposure: a randomized trial of dust control Pediatrics 103 772 777 10103301
Lanphear BP Succop P Roda S Henningsen G 2003 The effect of soil abatement on blood lead levels in children living near a former smelting and milling operation Public Health Rep 118 83 91 12690062
Lau WK Yeung CW Lui PW Cheung LH Poon NT Yung KKL 2002 Different trends in modulation of NMDAR1 and NMDAR2B gene expression in cultured cortical and hippocampal neurons after lead exposure Brain Res 932 10 24 11911857
Lewendon G Kinra S Nelder R Cronin T 2001 Should children with developmental and behavioural problems be routinely screened for lead? Arch Dis Child 85 286 288 11567935
Lewis RL 1997. Metals. In: Occupational and Environmental Medicine, 2nd ed. (LaDou J, ed). Stamford, CT:Appleton and Lange, 405–439.
Lidsky TI Schneider JS 2003 Lead neurotoxicity in children: basic mechanisms and clinical correlates Brain 126 5 19 12477693
Lynch RA Malcoe LH Skaggs VJ Kegler MC 2000 The relationship between residential lead exposures and elevated blood levels in a rural mining community J Environ Health 63 9 15
Mathee A von Schirnding YER Levin J Ismail A Huntley R Cantrell A 2002 A survey of blood lead levels among young Johannesburg school children Environ Res 90 3 181 184 12477462
Meyer I Hoelscher B Frye C Becker K Wichmann HE Heinrich J 2003a Temporal changes in blood lead levels of children in east Germany Int J Hyg Environ Health 206 3 181 192 12872526
Meyer PA McGeehin MA Falk H 2003b A global approach to childhood lead poisoning prevention Int J Hyg Environ Health 206 363 369 12971691
Mielke HW 2002 Research ethics in pediatric environmental health: lessons from lead [Commentary] Neurotoxicol Teratol 24 467 469 12127889
Mielke HW Reagan PL 1998 Soil is an important pathway of human lead exposure Environ Health Perspect 106 suppl 1 217 229 9539015
Mushak P 2002 Studies of pervasive toxic contaminants in children: staying the ethical course [Commentary] Neurotoxicol Teratol 24 463 465 12127888
Nation JR Gleaves DH 2001 Low-level lead exposure and intelligence in children Arch Clin Neuropsychol 16 375 388 14590169
Needleman HL 1995 Environmental lead and children’s intelligence. Studies included in the meta-analysis are not representative [Letter] Br Med J 310 1408 7787567
Needleman HL 2002 What is not found in the spreadsheets [Commentary] Neurotoxicol Teratol 24 459 461 12127887
Needleman HL Bellinger D 2001 Studies of lead exposure and the developing central nervous system: a reply to Kaufman Arch Clin Neuropsychol 16 359 374 14590168
Needleman HL McFarland C Ness RB Fienberg SE Tobin MJ 2002 Bone lead levels in adjudicated delinquents. A case control study Neurotoxicol Teratol 24 711 717 12460653
Nelson RM 2002 Appropriate risk exposure in environmental health research. The Kennedy-Krieger lead abatement study Neurotoxicol Teratol 24 445 449 12127884
Nielsen U Kamp JJ Grandjean P White RF 2000 Environmental lead exposure and neurodevelopmental outcome in Danish preschool children [Commentary] Neurotoxicology 21 896 897
Nihei MK Guilarte TR 2001 Molecular changes in glutamatergic synapses induced by Pb2+ : association with deficits of LTP and spatial learning Neurotoxicology 22 635 643 11770885
O’Donohoe J Chalkley S Richmond J Barltrop D 1998 Blood lead in U.K. children—time for a lower action level? Clin Sci 95 219 223 9680505
O’Dowd P 2002 Controversies regarding low blood lead level harm Med Health R I 85 345 348 12462867
Ornoy A 2003 The impact of intrauterine exposure versus postnatal environment in neurodevelopmental toxicity: long-term neurobehavioral studies in children at risk for developmental disorders Toxicol Lett 140–141 171 181
Ornoy A Segal J Bar-Hamburger R Greenbaum C 2001 Developmental outcome of school-age children born to mothers with heroin dependency: importance of environmental factors Dev Med Child Neurol 43 668 675 11665823
Owen BD Flegal AR 1998 Blood lead concentrations in marine mammals validate estimates of 102 - to 103 -fold increase in human blood lead concentrations Environ Res 78 2 134 139 9719617
Pinder L 2002 Commentary on the Kennedy Krieger Institute Lead Paint Repair and Maintenance Study Neurotoxicol Teratol 24 477 479 12127892
Prpic-Majic D Bobic J Simic D House DE Otto DA Jurasovic J 2000 Lead absorption and psychological function in Zagreb (Croatia) school children Neurotoxicol Teratol 22 347 356 10840178
Rabito FA Shorter C White LE 2003 Lead levels among children who live in public housing Epidemiology 14 263 268 12859025
Rahman A Maqbool E Zuberi HS 2002 Lead-associated deficits in stature, mental ability and behaviour in children in Karachi Ann Trop Paediatr 22 301 311 12530279
Roberts JR Reigart JR Ebeling M Hulsey TC 2001 Time required for blood lead levels to decline in nonchelated children J Toxicol Clin Toxicol 39 153 160 11407501
Rogan WJ Dietrich KN Ware JH Dockery DW Salganik M Radcliffe J 2001 The effect of chelation therapy with succimer on neuropsychological development in children exposed to lead N Engl J Med 344 1421 1426 11346806
Schnaas L Rothenberg SJ Perroni E Martinez S Hernández C Hernández RM 2000 Temporal pattern in the effect of postnatal blood lead level on intellectual development of young children Neurotoxicol Teratol 22 805 810 11120385
Schneider JS Lee MH Anderson DW Zuck L Lidsky TI 2001 Enriched environment during development is protective against lead-induced neurotoxicity Brain Res 896 48 55 11277972
Stanek EJ Calabrese EJ 1995 Daily estimates of soil ingestion in children Environ Health Perspect 103 276 285 7768230
Strömberg U Lundh T Schütz A Skerfving S 2003 Yearly measurements of blood lead in Swedish children since 1978: an update focusing on the petrol lead free period 1995–2001 Occup Environ Med 60 370 372 12709524
Tong S Baghurst P McMichael A Sawyer M Mudge J 1996 Lifetime exposure to environmental lead and children’s intelligence at 11–13 years: the Port Pirie cohort study Br Med J 312 1569 1575 8664666
Tong S Baghurst PA Sawyer MG Burns J McMichael AJ 1998 Declining blood levels and changes in cognitive function during childhood: the Port Pirie cohort study JAMA 280 1915 1919 9851476
Tong S Lu AY 2001 Identification of confounders in the assessment of the relationship between exposure and child development Ann Epidemiol 11 38 45 11164118
Tong S McMichael AJ Baghurst PA 2000 Interactions between environmental lead exposure and sociodemographic factors on cognitive development Arch Environ Health 55 330 335 11063408
Wakefield J 2002 The lead effect? Environ Health Perspect 110 A574 A580 12361937
Walkowiak J Altmann L Krämer U Sveinsson K Turfeld M Weishoff-Houben M 1998 Cognitive and sensorimotor functions in 6-year-old children in relation to lead and mercury levels: adjustment for intelligence and contrast sensitivity in computerized testing Neurotoxicol Teratol 20 511 521 9761589
Wang CL Chuang HY Ho CK Yang CH Tsai JL Wu TS 2002 Relationship between blood lead concentrations and learning achievement among primary school children in Taiwan Environ Res 89 12 18 12051780
Wasserman GA Factor-Litvak P 2001 Methodology, inference and causation: environmental lead exposure and childhood intelligence Arch Clin Neuropsychol 16 343 352 14590166
Wasserman GA Liu X Lolacono NJ 1997 Lead exposure and intelligence in 7-year-old children: The Yugoslavia Prospective Study Environ Health Perspect 105 956 962 9410739
Wasserman GA Liu X Popovac D Factor-Litvak P Kline J Waternaux C 2000 The Yugoslavia Prospective Lead Study: contributions of prenatal and postnatal lead exposure to early intelligence Neurotoxicol Teratol 22 811 818 11120386
Weiss B 2000 Vulnerability of children and the developing brain to neurotoxic hazards Environ Health Perspect 108 suppl 3 375 381 10852831
WHO 1995. Inorganic Lead. International Programme on Chemical Safety, Environmental Health Criteria 165. Geneva:World Health Organization. Available: http://www.inchem.org/documents/ehc/ehc/ehc165.htm [accessed 30 March 2004].
Winneke G Krämer U 1984 Neuropsychological effects of lead in children: interactions with social backgriound variables Neuropsychobiology 11 195 202 6472605
Ziegler EE Edwards BB Jensen RL 1978 Absorption and retention of lead by infants Pediatr Res 12 29 34 643372
|
15198918
|
PMC1247191
|
CC0
|
2021-01-04 23:36:23
|
no
|
Environ Health Perspect. 2004 Jun 28; 112(9):987-994
|
utf-8
|
Environ Health Perspect
| 2,004 |
10.1289/ehp.6941
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0112-00099510.1289/ehp.673615198919Mini-Monograph: Information SystemsHealth and Environment Information Systems for Exposure and Disease Mapping, and Risk Assessment Jarup Lars Small Area Health Statistics Unit, Department of Epidemiology and Public Health, Imperial College London, London, United KingdomAddress correspondence to L. Jarup, Small Area Health Statistics Unit, Department of Epidemiology and Public Health, Imperial College London, St. Mary’s Campus, 16 South Wharf Rd., London W2 1PF, United Kingdom. Telephone: 44 20 7594 3337. Fax: 44 20 7402 8837. E-mail:
[email protected] European Health and Environment Information System project is funded by a grant from the European Union, DG Public Health. The Small Area Health Statistics Unit is funded by a grant from the Department of Health, Department for Environment, Food and Rural Affairs, Environment Agency, Health and Safety Executive, Scottish Executive, Welsh Assembly Government, and Northern Ireland Department of Health, Social Services and Public Safety.
The views expressed in this publication are those of the author and not necessarily those of the funders.
The author declares he has no competing financial interests.
6 2004 15 4 2004 112 9 995 997 12 9 2003 15 12 2003 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. A large number of chemicals are used on a regular basis in modern society. Thousands of new chemicals are added each year, many of which may have toxic properties constituting potential health hazards. Rapid assessment of the risk associated with the use of these chemicals is therefore essential to protect people from exposure to potentially harmful substances. Exposures to chemicals (and physical agents) are typically unevenly distributed geographically as well as temporally. Disease occurrence also shows geographically varying patterns. Geographic information systems (GIS) may be used to produce maps of exposure and/or disease to reveal spatial patterns. Exposure mapping using advanced GIS modeling may enhance exposure assessment in environmental epidemiology studies. Disease maps can be valuable tools in risk assessment to explore changes in disease patterns potentially associated with changes in environmental exposures. Spatial variations in risk and trends related to distance from pollution sources may be studied using software tools such as the Rapid Inquiry Facility, developed by the U.K. Small Area Health Statistics Unit and enhanced in the European Health and Environment Information System project, for an initial quick evaluation of any potential health hazards associated with an environmental pollutant.
disease mappingexposure assessmentGIShealth and environment information systemsrisk assessmentspatial epidemiology
==== Body
Geographic information systems (GIS) are being used increasingly in environmental epidemiology, adding a further dimension to risk assessment (Vine et al. 1997).
A thorough exposure assessment is vitally important to risk assessment. Because measurements of environmental contaminants are expensive, few geographic monitoring locations are normally used. For example, air pollutants are usually measured in one or a few places in a city or a region, even though air pollution is not evenly distributed. Consequently, modeled pollution data are commonly used to assess exposure.
Accurate and detailed health data are equally important for risk assessment. Mortality and cancer incidence are often well reported and of good quality, whereas other health data (such as congenital anomalies and hospital admissions) may be underascertained and of uneven quality. The reporting of a health outcome may vary between geographic regions and over time. Thus, although geographic patterns of disease may be used to infer possible associations between environmental pollutants and health effects, such patterns may also reflect differences in health data recording.
Overlaying maps of exposure and populations may define populations at risk. However, linking of exposure and disease is highly dependent on the accuracy of exposure assessment as well as the time elapsed between initial exposure and disease (the latency time). The longer the latency time, the more difficult it will be to associate exposure with disease because of changes in exposure over time and/or population changes due to migration.
In spite of the difficulties of dealing with spatial data, risk assessment may benefit greatly from a spatial approach, as demonstrated by the articles in this mini-monograph. Indeed, it has been argued that “risk professionals will not mislead by presenting maps—they mislead by not presenting maps” (Hargrove et al. 1996).
Exposure Mapping
The role of GIS in exposure assessment is discussed in detail by Nuckols et al. (2004). As pointed out by the authors, it is important to keep in mind the definition of exposure. A person is considered exposed to an environmental agent if the agent in question has been in contact with a body surface. Nevertheless, sometimes exposure has occurred merely because the supposedly exposed population is living or has lived in a contaminated environment, without demonstrating that contact between the pollutants and the population has occurred. Such careless use of the term “exposure” should be avoided.
For example, soil may be contaminated with chemical waste such as heavy metals and other persistent substances, and exposure may indeed occur if the soil is used for growing vegetables (Staessen et al. 1994) or if there is a leakage of chemicals to groundwater, which may pollute drinking water wells. GIS have been used to map soil contamination, particularly, heavy metals. However, most residents in the contaminated area will not be exposed, and thus such GIS-derived data should not be used indiscriminately to assess exposure.
GIS are commonly used for assessing exposure to air pollution. There is almost always contact between the air pollutant and the human respiratory system, and thus exposure will indeed take place. However, individual exposure may vary greatly depending on many different circumstances such as the proportion of time spent indoors and outdoors, respectively.
Several studies have been performed around point sources of pollution, such as industrial plants, using circular areas at different distances from the source to define exposure zones (e.g., Aylin et al. 2001; Wilkinson et al. 1997). This approach has proven useful although it provides a rather crude estimate of exposure that does not consider, for example, meteorologic conditions or topography
Most studies of the relationship between traffic-related pollution and respiratory disease have used distance from roads as the exposure indicator (e.g., Hoek et al. 2002; Wilkinson et al. 1999). More recently, regression or dispersion modeling using GIS has been used to assess air pollution exposure (e.g., Bellander et al. 2001; Brauer et al. 2003; Briggs et al. 2000).
Similarly, Poulstrup and Larsen (2004) used dispersion modeling to define populations exposed to airborne dioxin around industrial plants in Denmark.
Verkasalo et al. (2004) used distance from a contaminated river as a proxy for exposure to dioxin. The authors note that the use of a nonspecific surrogate measure for exposure may have introduced considerable measurement error or confounding by correlated exposures. Nevertheless, they found the approach useful for an initial assessment of a potential environmental health hazard.
Disease Mapping
Disease mapping may be used to identify possible disease clusters, to define and monitor epidemics, to provide baseline data on health patterns, and to show changes in disease patterns over time. Disease mapping may also be useful for initial exploration of relationships between exposure and disease, particularly, acute health effects.
Maps of cancer incidence and mortality, computed on a global scale by the International Agency for Research of Cancer, are readily interpretable (Ferlay et al. 2001). Other large-scale maps such as the atlases of cancer mortality produced by the U.S. National Cancer Institute are also relatively easy to interpret (Cancer Mortality Maps & Graphs 1999).
Small-area maps of disease are much more difficult to produce and interpret in a meaningful way. A recent study of prostate cancer did not show any marked geographic variability in incidence at a small-area scale, arguing against a geographically varying, etiologically strong environmental risk factor (Jarup et al. 2002a). However, caution needs to be exercised in the interpretation because of factors such as latency time and migration.
Risk Assessment
Health and environment information systems based on GIS may be useful in the risk assessment process (for exposure assessment, for disease mapping, for assessing health risks associated with point sources of pollution, and for estimating the numbers of people at risk). The user should be aware of both the strengths and weaknesses associated with this approach.
Studies of variations in risk with distance from pollution sources such as industrial plants (Aylin et al. 2001) or landfill sites (Elliott et al. 2001; Jarup et al. 2002b) have been relatively common in environmental epidemiology.
Attempts to assess risk by overlaying maps of exposure and disease, given the (in)accuracy of the exposure estimates, latency periods, and migration problems, are likely to be misleading and should be avoided.
A main advantage in using GIS for exposure assessment is the possibility of modeling exposure geographically so that individual exposure may be estimated without the need for time-consuming and expensive measurements. Modeling uncertainties must be considered, especially when applied to large areas in particular, as the spatial resolution and coverage of environmental data are often poor.
GIS techniques can be used to estimate number of expected cases in a population potentially exposed to air pollution, by combining dispersion models with demographic data to produce estimates of the number of people exposed to certain levels of air pollution. Existing data on exposure–response relationships can then be used to compute the number of expected cases at each exposure level. Exposure–response relationships are currently available for only a few pollutants (e.g., PM10) for which good exposure models are not available, whereas data on exposure–response relationships are sparse for air pollutants, which can be more readily modeled (e.g., nitrogen dioxide)
Disease maps can be useful in risk assessment by defining a baseline pattern of disease that could be followed up by continued mapping of disease occurrence over time to explore potential changes in disease patterns that may be associated with changes in environmental exposures.
Disease mapping is used increasingly to describe variations of disease (most commonly cancer) between regions (Buntinx et al. 2003; Jarup et al. 2002a). Disease mapping may be very valuable as a means of assessing geographic differences in health, but several pitfalls need to be considered.
Arbitrary boundaries (usually administrative areas) are often used to map diseases. If the results are sensitive to change in boundaries, it is obvious that caution should be exercised when interpreting apparent associations between environmental exposures and health effects. This problem has been termed the modifiable area unit problem (Openshaw 1984) and is fundamental in all attempts to map disease using aggregated statistics.
It is clear that administrative boundaries may not be ideal for mapping health outcomes and that choice of boundaries may have a major influence on the results. The use of arbitrary but uniform boundaries such as grid squares may reduce these problems to some extent.
Standardized mortality (morbidity) ratios (SMRs) are commonly used as estimates of the relative risk. A criticism sometimes raised against mapping SMRs is that SMRs are not directly comparable, as they are not based on the same standard population (Julious et al. 2001). This is theoretically correct, but in practice, comparisons of SMRs between geographical areas will be misleading only if the age and sex structure of the populations are extremely disparate (Goldman and Brender 2000), which very rarely occurs in practice. The imprecision of alternative statistical estimates such as directly standardized rate ratios, when calculated on small area scale, is a far more serious problem (Jarup and Best 2003).
Small Area Health Statistics Unit
The U.K. Small Area Health Statistics Unit (SAHSU) was established in 1987 after a recommendation of an inquiry into the incidence of leukemia in children and young adults near the Sellafield nuclear plant.
The primary purpose of SAHSU is to assess environmental health risks using routinely collected health statistics data. SAHSU has developed a tool, the Rapid Inquiry Facility (RIF), for rapid assessment of environmental health hazards. The RIF produces estimated relative risks for any given condition for the population within defined areas around a point source, relative to the population in a local reference region (Aylin et al. 1999). The system creates maps to illustrate disease variation across small areas, as well as smoothed small-area maps that account for sampling variability in the observed data, to aid interpretation of the results.
Performing substantial research studies is an integral part of SAHSU work, making efficient use of routinely collected data and enhancing these data sets in the process. In recent years special attention has been given to exposure assessment, using advanced GIS techniques to define exposed populations (e.g., Elliott et al. 2001). State-of-the-art statistical methods are of crucial importance for studies of small-area variations in health, and thus there is an ongoing development of such methods within SAHSU (Richardson et al. 2004). Several aspects of spatial epidemiology have recently been reported by SAHSU in a comprehensive book (Elliott et al. 2000).
European Health and Environment Information System
The European Health and Environment Information System (EUROHEIS ) project was launched in 1999 to improve the understanding of the links between environmental exposures, health outcome, and risk through development of integrated information systems for rapid assessment of relationships between the environment and health at a geographic level.
The European Commission funded EUROHEIS for 3 years (2000–2003). During the first year of the project, the feasibility of implementing an analysis tool in different European countries, based on the SAHSU RIF, was examined. In the Nordic countries (Denmark, Finland, Sweden), health and population data are available at an individual level. In the United Kingdom and the Netherlands, data are available at postcode level, each post-code comprising approximately 15 households in the United Kingdom, and an average of 17 addresses in the Netherlands. The data resolution in Italy and Spain is much lower (municipality level, varying from a few hundred to several thousand people or more). In spite of the differences in data resolution between countries, the feasibility study showed that RIF implementation should be possible in all partner countries, and further development of the RIF was carried out to facilitate the transfer.
During the second year of the project, the RIF was installed in Spain and Sweden, confirming that it was feasible to run the system in countries with widely different data resolution. A similar system was already in place in Finland, whereas a modified version was used in Denmark, taking full advantage of the Danish personal identification system to follow up populations exposed to environmental agents.
In the third year of the project, the usefulness of the system in answering questions concerning environmental health risks was demonstrated through a series of case studies carried out within partner countries. In Denmark, cancer occurrence in a population exposed to airborne dioxins was studied using modeled dioxin emissions, which defined exposure at various distances around the plant (Poulstrup and Larsen 2004). The Finnish partner investigated cancer risk possibly associated with exposure to chlorophenols and polychlorinated dibenzo-p-dioxins and poly-chlorinated dibenzofurans emanating from a heavily contaminated river (Verksalo et al. 2004), and in Spain, the RIF was used to study relationships between drinking water hardness and cardiovascular and cerebrovascular mortality, using health and water quality data at municipality level (Ferrandiz et al. 2004).
To disseminate the EUROHEIS project results and to stimulate interest for spatial epidemiology, an international conference was held in Sweden in March 2003, attracting approximately 100 delegates, mainly from Europe but also from Peru, India, Israel, Canada, and the United States. Although the conference focused on the role of EUROHEIS in environmental health, it also comprised a wide range of presentations on exposure assessment, statistical methods, and other aspects of spatial epidemiology. This mini-monograph contains a selection of papers based on presentations at the conference.
Future Developments
Using concentric circles to identify exposed populations usually leads to a bias of the relative risk toward the null. A better approach would be to use data on wind direction and speed, and temperature as well as local topography to model the pollutant dispersion (Williams and Ogston 2002). The modeled data preferably should be validated by monitoring environmental media (air, water, soil). The potential to incorporate such model output (e.g., from the Atmospheric Dispersion Modeling System) in the RIF would greatly enhance the software.
Maps of relative risks show point estimates almost exclusively and do not consider the uncertainty in the relative risk estimates. Even if confidence intervals for the relative risk estimates are given in accompanying tables, it should be recognized that maps are far more powerful than tables for conveying information about geographic variations in risk. Therefore, techniques need to be developed to also map information about uncertainty in risk estimates in a way that is easy to interpret. A possible solution when using Bayesian smoothing methods is to map the posterior probability of the relative risk of any area exceeding a prespecified threshold (Jarup et al. 2002a; Jarup and Best 2003).
Exposure is commonly based on current environmental data, whereas information on historical exposure tends to be sparse. Many study areas are also prone to extensive migration, and most chronic diseases have long latency times. Therefore, methods should be developed to explore the effects on risk due to migration, as suggested in the article by Elliott and Wartenberg (2004). Some countries such as Denmark may use their unique personal identification system to follow up exposed populations to get more accurate estimates of the relative risk.
Future developments also include translation of the RIF software into a more user-friendly system, making it easily available to additional countries, and facilitation of dissemination via the Internet. Further methodologic progress will make it possible to input data from dispersion (and similar) models to enhance exposure assessment and to develop data export facilities to make further, more advanced statistical analysis possible when needed.
This article is part of the mini-monograph “Health and Environment Information Systems for Exposure and Disease Mapping, and Risk Assessment.”
==== Refs
References
Aylin P Bottle A Wakefield J Jarup L Elliott P 2001 Proximity to coke works and hospital admissions for respiratory and cardiovascular disease in England and Wales Thorax 56 228 233 11182017
Aylin P Maheswaran R Wakefield J Cockings S Jarup L Arnold R 1999 A national facility for small area disease mapping and rapid initial assessment of apparent disease clusters around a point source: the UK Small Area Health Statistics Unit J Public Health Med 21 289 298 10528956
Bellander T Jonson T Gustavsson P Pershagen G Järup L 2001 Using geographic information systems to assess individual historical exposure to air pollution from traffic and house heating in Stockholm Environ Health Perspect 109 633 639 11445519
Brauer M Hoek G van Vliet P Meliefste K Fischer P Gehring U 2003 Estimating long-term average particulate air pollution concentrations: application of traffic indicators and geographic information systems Epidemiology 14 228 239 12606891
Briggs DJ de Hoogh C Gulliver J Wills J Elliott P Kingham S 2000 A regression-based method for mapping traffic-related air pollution: application and testing in four contrasting urban environments Sci Total Environ 15 151 167 10843339
Buntinx F Geys H Lousbergh D Broeders G Cloes E Dhollander D 2003 Geographical differences in cancer incidence in the Belgian province of Limburg Eur J Cancer 39 2058 2072 12957461
Cancer Mortality Maps & Graphs 1999. Geographic Patterns and Time Trends of Cancer Death Rates in the U.S. Bethesda, MD:National Cancer Institute. Available: http://www3.cancer.gov/atlasplus/type.html [accessed 28 November 2003].
Elliott P Briggs D Morris S de Hoogh C Hurt C Kold Jensen T 2001 Risk of adverse birth outcomes in populations living near landfill sites Br Med J 323 363 368 11509424
Elliott P Wakefield JC Best NG Briggs D eds. 2000. Spatial Epidemiology: Methods and Application. Oxford:Oxford University Press.
Elliott P Wartenberg D 2004 Spatial epidemiology: current approaches and future challenges Environ Health Perspect 112 998 1006 15198920
Ferlay J Bray F Pisani P Parkin DM 2001. GLOBOCAN 2000: Cancer Incidence, Mortality and Prevalence Worldwide, Version 1.0. IARC CancerBase No. 5. Lyon:IARC Press.
Ferrandiz J Abellán JJ Gómez-Rubio V Lopez-Quílez A Sanmartín P Abellán C 2004 Spatial analysis of the relationship between mortality from cardiovascular and cerebrovascular disease and drinking water hardness Environ Health Perspect 112 1037 1044 15198925
Goldman DA Brender JD 2000 Are standardized mortality ratios valid for public health data analysis? Stat Med 19 1081 1088 10790681
Hargrove WW Levine DA Miller MR Coleman PR Pack DL Durfee RC 1996. GIS and Risk Assessment: A Fruitful Combination. Available: http://gis.esri.com/library/user-conf/proc96/TO50/PAP028/P28.HTM [accessed 17 May 2004].
Hoek G Brunekreef B Goldbohm S Fischer P van den Brandt PA 2002 Association between mortality and indicators of traffic-related air pollution in the Netherlands: a cohort study Lancet 360 1203 1209 12401246
Jarup L Best N 2003 Editorial comment on geographical differences in cancer incidence in the Belgian Province of Limburg by Bruntinx and colleagues Eur J Cancer 39 1973 1975 12957450
Jarup L Best N Toledano MB Wakefield J Elliott P 2002a Geographical epidemiology of prostate cancer in Great Britain Int J Cancer 97 695 699 11807800
Jarup L Briggs D de Hoogh C Morris S Hurt C Lewin A 2002b Cancer risks in populations living near landfill sites in Great Britain Br J Cancer 86 1732 1736 12087458
Julious SA Nicholl J George S 2001 Why do we continue to use standardized mortality ratios for small area comparisons? J Public Health Med 23 40 46 11315692
Nuckols JR Ward MH Jarup L 2004 Using geographic information systems for exposure assessment in environmental epidemiology studies Environ Health Perspect 112 1007 1015 15198921
Openshaw S 1984. The Modifiable Areal Unit Problem. Concepts and Techniques in Modern Geography (CATMOG) 38. Norwich, UK:Geo Books.
Poulstrup A Hansen HL 2004 Use of GIS and exposure modeling as tools in a study of cancer incidence in a population exposed to airborne dioxin Environ Health Perspect 112 1032 1036 15198924
Richardson S Thomson A Best N Elliott P 2004 Interpreting posterior relative risk estimates in disease-mapping studies Environ Health Perspect 112 1016 1025 15198922
Staessen JA Lauwerys RR Ide G Roels HA Vyncke G Amery A 1994 Renal function and historical environmental cadmium pollution from zinc smelters Lancet 343 1523 1527 7911869
Verkasalo PK Kokki E Pukkala E Vartiainen T Kiviranta H Penttinen A 2004 Cancer risk near a polluted river in Finland Environ Health Perspect 112 1026 1031 15198923
Vine MF Degnan D Hanchette C 1997 Geographic information systems: their use in environmental epidemiologic research Environ Health Perspect 105 598 605 9288494
Wilkinson P Elliott P Grundy C Shaddick G Thakrar B Walls P 1999 Case-control study of hospital admission with asthma in children aged 5–14 years: relation with road traffic in north west London Thorax 54 1070 1074 10567625
Wilkinson P Thakrar B Shaddick G Stevenson S Pattenden S Landon M 1997 Cancer incidence and mortality around the Pan Britannica Industries pesticide factory, Waltham Abbey Occup Environ Med 54 101 107 9072017
Williams FL Ogston SA 2002 Identifying populations at risk from environmental contamination from point sources Occup Environ Med 59 2 8 11836461
|
15198919
|
PMC1247192
|
CC0
|
2021-01-04 23:39:44
|
no
|
Environ Health Perspect. 2004 Jun 15; 112(9):995-997
|
utf-8
|
Environ Health Perspect
| 2,004 |
10.1289/ehp.6736
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0112-00099810.1289/ehp.673515198920Mini-Monograph: Information SystemsSpatial Epidemiology: Current Approaches and Future Challenges Elliott Paul 1Wartenberg Daniel 21Small Area Health Statistics Unit, Department of Epidemiology and Public Health, Imperial College London, London, United Kingdom2Environmental and Occupational Health Sciences Institute and The Cancer Institute of New Jersey, University of Medicine and Dentistry of New Jersey, Robert Wood Johnson Medical School, Piscataway, New Jersey, USAAddress correspondence to P. Elliott, Small Area Health Statistics Unit, Department of Epidemiology and Public Health, Imperial College London, Faculty of Medicine, St. Mary’s Campus, Norfolk Place, London W2 1PG, United Kingdom. Telephone: 44 0 20 75943328. Fax: 44 0 20 7262 1034. E-mail:
[email protected] Small Area Health Statistics Unit is funded by a grant from the Department of Health, Department of the Environment, Food and Rural Affairs, Environment Agency, Health and Safety Executive, Scottish Executive, Welsh Assembly Government, and Northern Ireland Department of Health, Social Services and Public Safety. This research was also supported by grants R01 CA92693 from the National Cancer Institute and U61/ATU272387 from the Agency for Toxic Substances and Disease Registry, Centers for Disease Control and Prevention, to D.W.
The views expressed in this publication are those of the authors and not necessarily those of the funding bodies.
The authors declare they have no competing financial interests.
6 2004 15 4 2004 112 9 998 1006 12 9 2003 15 4 2004 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Spatial epidemiology is the description and analysis of geographic variations in disease with respect to demographic, environmental, behavioral, socioeconomic, genetic, and infectious risk factors. We focus on small-area analyses, encompassing disease mapping, geographic correlation studies, disease clusters, and clustering. Advances in geographic information systems, statistical methodology, and availability of high-resolution, geographically referenced health and environmental quality data have created unprecedented new opportunities to investigate environmental and other factors in explaining local geographic variations in disease. They also present new challenges. Problems include the large random component that may predominate disease rates across small areas. Though this can be dealt with appropriately using Bayesian statistics to provide smooth estimates of disease risks, sensitivity to detect areas at high risk is limited when expected numbers of cases are small. Potential biases and confounding, particularly due to socioeconomic factors, and a detailed understanding of data quality are important. Data errors can result in large apparent disease excess in a locality. Disease cluster reports often arise nonsystematically because of media, physician, or public concern. One ready means of investigating such concerns is the replication of analyses in different areas based on routine data, as is done in the United Kingdom through the Small Area Health Statistics Unit (and increasingly in other European countries, e.g., through the European Health and Environment Information System collaboration). In the future, developments in exposure modeling and mapping, enhanced study designs, and new methods of surveillance of large health databases promise to improve our ability to understand the complex relationships of environment to health.
disease clustersdisease mappingenvironmental pollutionepidemiologygeographic studiesmethods
==== Body
Spatial epidemiology is the description and analysis of geographically indexed health data with respect to demographic, environmental, behavioral, socioeconomic, genetic, and infectious risk factors. It is part of a long tradition of geographic analyses dating back to the 1800s when maps of disease rates in different countries began to emerge to characterize the spread and possible causes of outbreaks of infectious diseases such as yellow fever and cholera (Walter 2000). Over the ensuing decades, it grew in com.plexity, sophistication, and utility. Spatial epidemiology extends the rich tradition of ecologic studies that use explanations of the distribution of diseases in different places to better understand the etiology of disease (Doll 1980; Keys 1980). In this article we focus principally on small-area analyses of chronic, noninfectious diseases, where there is considerable current interest within the field of spatial epidemiology.
Recent advances in data availability and analytic methods have created new opportunities for investigators to improve on the traditional reporting of disease at national or regional scale by studying variations in disease occurrence rates at a local (small-area) scale (Walter 2000). Such investigations may include locally relevant health risk factor data such as exposures to local sources of environmental pollution and the distribution of locally varying socioeconomic and behavioral factors. They also present new challenges because as the scale of the investigation becomes narrowed to a particular small area or group of areas, the reduced size of the population at risk leads to small numbers of events and unstable risk estimates (Olsen et al. 1996). Furthermore, because of the small population, such studies are more susceptible to errors or local variations in the quality of both the health (numerator) and the population (denominator) data than studies conducted over larger areas. At the broader scale, purely local variations in data quality are likely to largely cancel out, whereas at the small-area scale, these variations could lead to serious biases if not detected. Finally, small-area studies (like other types of epidemiologic inquiry) are susceptible to confounding, which can result in spurious exposure–disease associations. In the small-area case, this is particularly so with respect to socioeconomic variables. People and communities tend to cluster in space in systematic ways that may be highly predictive of disease risk. For example, people of high socioeconomic status tend to live near others with high incomes and in areas with better housing and schooling than those in lower-income areas. Individuals with higher incomes tend to have more favorable risk factor profiles (e.g., they are more likely to be nonsmokers, take more leisure-time exercise, and eat more favorable diets) and as a consequence, have better health (Smith et al. 1996a, 1996b). Such spatially organized socioeconomic effects can have important influence on the rates of disease observed in small areas (Dolk et al. 1995). They may also be associated with the siting (or absence) of sources of environmental pollution, as “environmental (in)justice” dictates that poorer people in poorer areas are often more likely to be exposed to the effects of pollution (Corburn 2002).
We note that an in-depth and individual-based approach might investigate how individuals interact with their environment and how these interactions affect health. This could address, for example, why people with higher incomes take more leisure-time exercise. Is it because they have a local environment more enticing, have the financial resources to engage in specific activities, have jobs that afford them more leisure time, or pursue more leisure-time activities for other reasons? Such questions may have an important spatial component. However, we see these as second-order issues beyond the scope of this article.
We now briefly consider the analytic framework for carrying out spatial analyses and the types of studies commonly undertaken. We then focus on a number of challenges that face the practitioner of spatial epidemiology, including issues of data availability and quality, confidentiality, exposure assessment, exposure mapping, and study design.
Analytic Framework
In considering an analytic framework for spatial epidemiologic analyses (Elliott et al. 2000b), we first distinguish between point and area data. Each of the population, environmental exposure, and health data may be associated with a point, or exact spatial location such as a street address (occurrence data), or an area, a defined spatial region such as a community, of which it is representative (aggregate summaries, e.g., count data). Data from a variety of points (e.g., residence, workplace, hobby locations) may give the closest link to an assumed biologic model in which the average disease risk of an individual will reflect individual characteristics such as age, sex, and genetic factors (e.g., predisposition, susceptibility, immune or toxicologic response capability); lifestyle variables, such as smoking and diet; and exposure to environmental pollutants. The lifestyle and exposure factors may depend on the ways that the individual interacts with the environment as she/he moves through both time and space, which itself depends on the range of daily activities, type and location of residence, workplace, travel and migration patterns, habits, behaviors, and so on. Together with individual susceptibility factors, these may determine biological dose. For many environmental exposures, the parameter of interest may be cumulative lifetime dose, the maximum short-term dose, or even the cumulative dose above some threshold. For example, in carcinogenesis, the parameter may be the dose at some critical point in the multistage pathway underlying cancer formation (Moolgavkar 1999). For other outcomes, exposure to a single, high (toxic) dose may trigger an adverse response, as with chloracne after exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) from the Seveso accident in northern Italy (Caramaschi et al. 1981). The effects from ionizing radiation, on the other hand, are thought to reflect cumulative lifetime exposure, a more problematic metric for spatial epidemiology, although recent research suggests that the maximum rate of exposure mediates the effects (Cardis et al. 2001).
Case–control and cohort studies can give a relatively close approximation to the biologic model in investigating environmental health issues because both individual person characteristics and exposures are studied in the individual environment. Case–control studies provide point data for cases and a set of controls. They are prone to selection and other biases, are moderately expensive and time-consuming to carry out, and are not feasible in all situations. Cohort studies, although not subject to selection bias, are prone to other biases, including losses to follow-up, and are generally more expensive and time-consuming to carry out than case–control studies. Exploratory studies using aggregate data, such as geographic correlation studies, offer an alternative approach for generating, prioritizing, and analyzing data to address specific hypotheses of disease etiology and causation. Although they too are prone to biases and misclassification (Elliott and Wakefield 2000), they are generally easier, quicker, and less expensive to conduct than case–control or cohort studies. One example of this approach is with use of a dedicated system such as that developed by the Small Area Health Statistics Unit (SAHSU) in the United Kingdom (Elliott et al. 1992b); this has recently been adopted in other European countries as part of the European Health and Environment Information System (EUROHEIS) collaboration (EUROHEIS 2003). If these exploratory and other studies generate sufficient evidence in support of specific hypotheses, case–control and/or cohort studies can then be used to test these hypotheses with use of purpose-collected individual-level data.
Types of Spatial Epidemiologic Inquiry
Spatial epidemiology at small-area scale can be divided into three main areas:
disease mapping
geographic correlation studies
clustering, disease clusters, and surveillance.
We note that the above grouping is artificial. For example, depending on scale, disease mapping may provide information on individual disease clusters and more generally on disease clustering. A point source of exposure may give rise to a localized excess of cases that might be detected on a disease map, whereas geographic correlation studies share much in common with disease-mapping studies (with addition of one or more potential explanatory variables), and the statistical models used are often similar. Each of the above main types of inquiry is now considered in turn.
Disease Mapping
As noted earlier, disease maps have a long history. A survey in 1991 identified 49 international, national, and regional disease atlases (Walter and Birnie 1991). An early example was the work of Stocks, who described variations in cancer mortality across counties of England and Wales (Stocks 1936, 1937, 1939). More recent examples include an atlas of cancer incidence in England and Wales (Swerdlow and dos Santos Silva 1993) and an all-causes mortality atlas (Pickle et al. 1996) and separate cancer mortality atlas (Devesa et al. 1999) for the United States. Disease maps provide a rapid visual summary of complex geographic information and may identify subtle patterns in the data that are missed in tabular presentations. They are used variously for descriptive purposes, to generate hypotheses as to etiology, for surveillance to highlight areas at apparently high risk, and to aid policy formation and resource allocation. They are also useful to help place specific disease clusters and results of point-source studies in proper context (Wilkinson et al. 1997).
Disease maps typically show standardized mortality or morbidity (e.g., incidence) ratios (SMRs) for geographic areas such as countries, counties, or districts. The rate in area i is estimated by the standardized mortality (or morbidity) ratio (SMRi), calculated as Oi /Ei, where Oi is the observed number of deaths or incident cases of disease in the area (assumed to follow an independent Poisson distribution). Ei is the expected number of cases (calculated by applying age- and sex-specific death or disease rates to population counts for the area). The SMR thus defined is based on indirect standardization. Some authors advocate direct standardization, as it involves adjustment to a common standard (Julious et al. 2001). In our own experience, the two methods nearly always give near-identical results.
Although disease maps have both visual and intuitive appeal, caution is required in interpretation, as apparent patterns can be created or lost artifactually depending on how the mapped variable is depicted (e.g., the number and boundaries of the categories) and the geographic scale or resolution. The choice of colors for displaying data can also affect interpretation (Brewer and Pickle 2002; Smans and Esteve 1992). Maps of the same data drawn at different scales of resolution can result in very different visual patterns (Monmonier 1997). Figure 1, for example, from a study of childhood lead poisoning, shows maps at three different scales (U.S. census block group, ZIP codes, and counties) of the percentage of homes built before 1950 (a major risk factor for childhood lead overexposure) in New Jersey based on U.S. census data reported at the block group level of resolution. When aggregated by geopolitical boundaries, regional values are overweighted (geographically) by more compact, more urban ones that typically have more older housing, often obscuring important information in less-populated rural regions.
When constructing maps, users must select both the size of units and the method to aggregate units to highlight the features of interest. Homogeneity within aggregate groups is important for meaningful interpretation. Different scales and different aggregation strategies can lead to different but equally valid maps that emphasize different features of the data. In the geography literature, this is called the modifiable area unit problem (Openshaw 1984). Although generally the aim is to choose geographic units that are as small as possible, the choice is often dictated by the availability of data, and because of sparse data, there will often be a tradeoff between homogeneity within small geographic units and precision of risk estimates.
Variation in rates across the map may reflect differences in the quality of data, for example, in the diagnosis, classification, or reporting of disease (Best and Wakefield 1999), rather than true differences in disease rates. Furthermore, the digital boundaries identifying the geographic units, and the geographic linkages between the various data within a geographic information system (GIS) may contain errors, including errors in the assignment of geocodes (postcodes) (Briggs and Elliott 1995). Clearly these may lead to errors in the resultant maps. Data quality for denominator (population at risk) data, although often overlooked, can also be a problem. Inaccurate estimates can change the appearance of mapped patterns and complicate map comparisons, especially for areas with small populations. When calculating SMRs for intercensual years, investigators use different interpolation algorithms, which can lead to differences in denominators and rates. For example, in a study of cancer incidence in Dalgety Bay, Scotland, risks based on census data were overestimated because there had been rapid population growth in the area since the previous census (Black et al. 1994).
Recent focus on small-area mapping studies, where typically the unit of analysis has a population of 5,000 or less (such as census tracts in the United States or electoral wards in the United Kingdom), introduces an extra source of variability into the map because of random variation. Typically, sparsely populated areas with few (or zero) cases can generate extreme values of the SMR, as the variance of the SMR is inversely related to Ei and small populations will have large variability in the estimated rates. As these sparsely populated areas are often bigger than densely populated areas (because the administrative geography depends on population size), they tend to dominate the map visually even though they produce the least-precise risk estimates (Elliott et al. 1995). Methods based on Bayesian statistics (Clayton and Kaldor 1987) have been used to remove part of the random component from the map to give smoothed estimates of relative risk in each area. Such estimates are a compromise between the local value of the SMR and either the mean value for the map as a whole, or some local mean. Smoothing is greatest for the least-stable estimates (i.e., where Ei is small).
Figure 2 is an example of a small-area mapping study of adult leukemia incidence in the West Midlands region of England, 1974–1986 (Olsen et al. 1996). Each small area on the map is an electoral ward, which as noted above has a population of approximately 5,000 on average. The smallest wards, with the largest populations and hence the most stable risk estimates, are located toward the center of the map in and around the Birmingham conurbation. Figure 2A shows the age- and sex-adjusted SMRs based on the observed and expected values in each area, whereas Figure 2B shows the smoothed SMR, with smoothing to the overall mean using empirical Bayes methods. The unsmoothed map has considerable apparent variability, with more than 3-fold variation across the map. Many of the extreme values (both low and high) are found in the periphery of the map, that is, in the rural areas distant from the Birmingham conurbation. After smoothing, the map appears much flatter, and all the extremes are removed.
Although map smoothing on average produce a more stable and realistic map, an important issue is the extent to which disease excesses in any truly high-risk areas (especially those more sparsely populated) might be smoothed away. The degree of smoothing will determine the tradeoff between high sensitivity (truly high-risk areas correctly identified) and high specificity (areas without excess risk correctly identified). This tradeoff is important, as a sensitive but nonspecific measure will generate many false positive findings, whereas a specific but nonsensitive measure will miss areas with high risk. Richardson et al. (2004) have investigated the properties of commonly used map-smoothing techniques using a series of realistic scenarios to simulate possible patterns in the disease map. They conclude that unless the relative risk is of the order of 2 to 3 and expected numbers in the geographic unit are at least 5 (or for relative risks of order 2, expected numbers are at least 20), then the map-smoothing methods are likely to perform poorly in terms of their abilities to detect areas with true excess. This is important in designing appropriately powered investigations and in managing expectations as to what can be achieved with sparse data.
Geographic Correlation Studies
In geographic correlation studies, the aim is to examine geographic variations across population groups in exposure to environmental variables (which may be measured in air, water, or soil), socioeconomic and demographic measures (such as race and income), or lifestyle factors (such as smoking and diet) in relation to health outcomes measured on a geographic (ecologic) scale. This approach often takes advantage of data that are routinely available and can be used to investigate natural experiments where the exposure has a physical basis (e.g, soil, water) (Richardson and Monfort 2000). In addition, the effect of exposure measurement error is reduced by averaging across groups. However, geographic correlation is affected by the problems of disease-mapping studies noted above, together with the added complication of correlation with one or more explanatory variables. Such studies are often thought of as hypothesis-generating, as the unit of observation is the geographic group rather than the individual and associations observed at the group level do not necessarily hold at the individual level—the so-called ecologic fallacy (Piantadosi et al. 1988). For this reason, observations at the ecologic scale will usually need validation and replication at the individual level, for example, through cohort, case–control studies or possibly randomized, controlled prevention or intervention trials (such as lead chelation studies). Nonetheless, ecologic studies of this kind have been pivotal in developing and exploring major hypotheses of public health importance, for example, the linking of malignant hepatoma (which has very high incidence in Asian populations) with hepatitis B infection (Beasley 1988) and the seminal work of Keys and colleagues in elucidating the role of saturated fat in the etiology of coronary heart disease (Keys 1980).
The development of the first cancer mortality atlases in the United States in the mid-1970s (Mason et al. 1975, 1976) showed distinctive patterns of variation of different cancers and led to a series of informal correlational studies. Based on the patterns of high risk that appeared to correspond to specific activities, behaviors, or environmental exposures, investigators postulated specific hypotheses (Blot and Fraumeni 1982; Fraumeni 1988; Hoover et al. 1975; Mason 1976) that were later investigated through case–control studies. Although not all of these studies confirmed the geographically generated hypotheses, investigation of a regional excess of oral cavity and pharynx cancer among women revealed the previously unknown risk of smokeless tobacco use (Blot and Fraumeni 1977; Winn et al. 1981). Investigation of a regional excess of sinonasal cancer was consistent with studies in other countries showing risks associated with working in the furniture industry (Blot and Fraumeni 1977; Brinton et al. 1976, 1977, 1984, 1985), and study of local lung cancer excess was associated with residence near or employment in the arsenic industry (Blot and Fraumeni 1975, 1994).
Geographic correlation studies are also carried out at a more local or small-area scale, where the problem of ecologic bias may be lessened as the analysis is closer to the level of the individual. For example, Staessen et al. (1999) examined the relationship between environmental exposure to cadmium and bone density in 10 districts in Belgium (including 6 that bordered on three zinc smelters). Shaper et al. (1980) investigated the relationship between water hardness and cardiovascular disease in towns in Great Britain, while Maheswaran et al. (1999) assessed in particular the role of magnesium in the water supply in relation to mortality from acute myocardial infarction. The last of these studies used water zones in northwest England (each water zone serves up to 50,000 people) as the unit of analysis. For some environmental exposures, such as non-ionizing radiation from overhead power lines, the potential harmful effects may operate over a very small distance (up to 50–100 m from the power line), so only a highly localized or individual-based study can investigate the issue (Feychting and Ahlbom 1993; Olsen et al. 1993; Verkasalo et al. 1993).
One important issue merits brief mention here. Informal geographic correlation studies (or evaluations) are often conducted by non-scientists in their own communities or neighborhoods out of personal concern. When one suspects a local disease excess, or when oneself, a family member or friend is stricken with cancer, one often asks “Why? What did I or they do wrong? What is it about where I live or where I work that caused this tragedy?” This concern may cause one to seek an explanation or to consider local industries or sources of environmental pollution as the putative cause. In this process, an informal geographic correlation is being undertaken, insofar as the health event and putative environmental exposure have been juxtaposed. Most such evaluations do not provide useful etiologic clues, as neither the underlying variability in disease rates nor the post hoc nature of the association with sources of environmental pollution are properly accounted for.
Disease Clusters, Clustering, and Surveillance
Investigation of disease clusters and disease incidence near a point source usually assumes that the background risk surface is flat, against which a peak at the pollution source is being tested. If this is not the case and the background surface is bumpy, that is, there are peaks and troughs in the risk surface, this may indicate generalized or broad-scale clustering of the disease. (Clearly in this situation, the observation of a disease excess at a particular point may not be unusual.) This tendency for disease cases to occur in a nonrandom spatial pattern relative to the pattern of the noncases has a more robust statistical formulation than the investigation of disease clusters per se and may give clues as to etiology (Wakefield et al. 2000). For example, there is evidence of spatial clustering of Hodgkin disease (Alexander et al. 1989) that, along with other epidemiologic and laboratory evidence, has suggested a possible infectious etiology. The study of generalized clustering has much in common with disease mapping, and the same cautionary considerations apply, particularly concerning the quality of the underlying data.
Putative disease clusters may come to light because of media reports or be brought to the attention of the authorities by concerned individuals; as noted, often the apparent cluster will become linked with a local source of environmental pollution (Greenberg and Wartenberg 1991; Trumbo 2000). In general, this might be a point, line, or area source. Point sources include a chimney stack from an industrial site, a radio transmitter, mobile phone tower, and so forth. A line source refers to an extended linear source such as a road, river, or power line, and an area source may include industrial complexes, landfill sites, and other geographically defined areas such as water-supply zones (or watersheds). In practice, in the absence of detailed information concerning the extent of an industrial site or the locations within the site where emissions occur, area sources are often modeled as point sources. A recent study of landfill sites in the United Kingdom would be one example (Elliott et al. 2001). Although U.S. case–control studies have used similar exposure metrics, no extant systems allow similar, broad-based data assessments.
The term disease cluster is poorly defined but implies an excess of cases above some background rate bounded in time and space. These boundaries may be ill-defined, and so-called boundary shrinkage may occur, accentuating the apparent risk by focusing the investigation tightly on the cases making up the cluster.
The more narrowly the underlying population is defined, the less will be the number of expected cases, the greater will be the estimate of the excess rate, and often the more profound will be the statistical significance. (Olsen et al. 1996)
Despite the inherent problems, the local public health department may find itself compelled to respond, if only to allay public anxiety (Greenberg and Wartenberg 1991). Usually the initial assessment of the data will involve the following:
Detailed checking of the cases. This is an essential step, as the putative cluster may involve a disparate group of diagnoses, some double-counting (duplicate records) may occur, and some cases may be erroneously reported. One also must verify the location (or geocode) of each case, which can be difficult in some locales.
Definition of the boundaries in time and space so that a population denominator, by age and sex, can be constructed (usually from census records). Although accuracy is important, it is hard to validate the population data outside the census years, particularly as the areas get smaller.
Estimation of the expected numbers of cases based on age- and sex-specific background rates (e.g., obtained from published regional or national data).
Calculation of the SMR for the area.
Assessment of statistical significance (usually reported at p < 0.05) assuming a Poisson distribution for the occurrence of cases.
Communication of results to the public, providing context, plausibility, and plans for follow-up, if appropriate.
The process of obtaining the initial data outlined above can be extremely costly in both time and resources for local health department personnel, as data from several disparate sources must be assembled and brought together. In addition, for local health departments not familiar with the detailed methods and requirements of a major cluster investigation, inevitably there can be a steep learning curve. This might include familiarizing oneself with the specialist statistical methodologies of cluster investigation (beyond calculation of the SMR), as such methods are not part of the routine armory of the public health specialist (Elliott et al. 1995; Morris and Wakefield 2000; Waller and Lawson 1995). In the United Kingdom, a rapid inquiry facility (RIF) has been established within SAHSU to provide such analyses within a few working days for a particular area. This greatly facilitates the ability of a local public health department to respond quickly to reports of a putative disease excess in their area based on the available routine data. Areas can be defined by administrative geography such as electoral enumeration district (~ 400 individuals) or ward, by post-code (~ 13 households), or by map reference. The RIF includes routine national health and population data held in an Oracle database on its own dedicated computer system, with geographic linkages provided by a proprietary GIS (Aylin et al. 1999). The health records, including mortality, cancer incidence, hospital discharges, and congenital anomalies, all include the postcode, with geographic resolution of approximately 10–100 m. The RIF assembles the data and provides an SMR (with and without adjustment for socioeconomic variables) for the area of interest compared with regional or national rates. An unsmoothed and smoothed map (using empirical Bayes methods) are also produced, together with contextual maps of local landmarks, socioeconomic data, pollution sources, and so on. A version of the RIF has been made available to other European countries as part of the EUROHEIS consortium (EUROHEIS 2003). Although many state health departments in the United States routinely evaluate data in response to cluster inquiries, none currently has a comparable system dedicated to such activities.
Once a link between a putative disease cluster and a local source of environmental pollution has been put forward, it is extremely difficult to confirm or refute it without recourse to external data (e.g., from another area or time period). Because an informal process of data comparison (akin to multiple testing) has taken place (by the media, concerned individuals, etc.) in similar-sized localities elsewhere across the country, statistical testing in a formal sense is rendered invalid (Elliott and Wakefield 2000). Only disease occurrences at the high end of the distribution are highlighted. Diseases or areas with apparent low risk never come to the attention of the authorities. This informal process of multiple testing means that it is impossible to gauge the true significance (in a statistical sense) of an apparent disease excess in a particular locality. Many clusters, even where nominally statistically significant, will appear purely as a chance finding, particularly for rare events (such as most cancers). Conversely, some true disease excesses may be overlooked because of lack of systematic evaluation of the small-scale geographic pattern of disease incidence (Wartenberg 1995).
Local concerns about a disease cluster in a particular area must be sympathetically and sensitively handled but will not usually lead to formal study or any new etiologic insight (Drijver and Woudenberg 1999; Trumbo 2000). Indeed, against this background, it has been argued that individual cluster reports should not be investigated (Rothman 1990) unless there are sufficient numbers of cases (five or more) and risks in a particular area are very high (relative risk ≥ 20) (Neutra 1990).
Occasionally it will be necessary to carry out more detailed inquiry. Investigations have adopted either the case–control (e.g., Aschengrau et al. 1998; Infante-Rivard and Amre 2001; Morris and Knorr 1996; Mulder et al. 1994; Wrensch et al. 1999) or small-area (ecologic) approach (e.g., Berry and Bove 1997; Goldberg et al. 1995; Kokki et al. 2001; Lopez-Abente et al. 2001; Wilkinson et al. 1997). Where the routine health statistics appear to confirm suspicions of disease excess (notwithstanding the problems of multiple testing referred to above), then as indicated, examination of data for a different time period or area will be required. This allows the data to be tested within the usual statistical paradigm, as the initial observation generates a hypothesis that can then be tested on independent data. With a dedicated national system such as SAHSU in the United Kingdom, this can be done readily using the national database. Examples include national studies of cancer incidence near incinerators of waste solvents and oils after observations of excess incidence of cancer of the larynx near one such incinerator (Elliott et al. 1992a), and risk of leukemia and incidence of other cancers near TV and radio transmitters, after reports of a leukemia cluster near the Sutton Coldfield transmitter in the West Midlands, England (Dolk et al. 1997a, 1997b).
When the study is done because of a priori concerns about a source of environmental pollution rather than in response to a claim of disease excess in a particular area, the statistical framework is again more robust, as a hypothesis can be set up and tested in the usual way. Investigation may involve a number of or all such sources in the region or country. This increases statistical power and overcomes the problem of selection where one site, or a few sites, are chosen for study, perhaps because of suspicion of disease excess in the vicinity. However, it makes the possibly unrealistic assumption that the sources are similar with respect to their potential to cause environmental health problems, and high risk around one or two sources (which may have high levels of toxic releases into the environment) may be masked. In the United Kingdom, national studies undertaken a priori include cancer incidence near municipal incinerators (Elliott et al. 1996a), risk of hemopoietic cancers near oil refineries (Wilkinson et al. 1999), angiosarcoma of the liver near vinyl chloride plants (Elliott and Kleinschmidt 1997), and risk of congenital anomalies and various cancers near landfill sites (Elliott et al. 2001; Järup et al. 2002). In the Scandinavian countries, national studies of leukemia risk near power lines have been done that take advantage of the high-quality health and population registers available in those countries (Feychting and Ahlbom 1993; Olsen et al. 1993; Verkasalo et al. 1993).
Although national-scale small-area studies are unlikely on their own to establish causal links with the pollution source (unless the risk is very high), they do give a valuable answer to the public health question “If I live near polluting source X, am I (on average) at increased risk of disease?” and may indicate avenues for further inquiry such as studies of pathways of environmental exposure, biomarker studies, or case–control studies.
Cluster detection and surveillance.
Surveillance, or the systematic routine collection and analysis of health outcome data for disease prevention and control purposes (Thacker and Berkelman 1992), can be applied to the problem of disease clusters through the use of space, time, and space-time pattern detection methods (Kulldorff et al. 1997; Kulldorff 1997, 2001; Rogerson 1997, 2001; Rushton et al. 1996). This has been proposed as a more effective approach than ad hoc cluster studies for identifying local disease excesses and prioritizing them for follow-up investigations (Hardy et al. 1990; Wartenberg 1995). In contrast to the passive or reactive analysis of reported local disease excesses using systems like the RIF, surveillance offers the opportunity to provide proactive, early detection of raised incidence of disease even when there is no specific etiologic hypothesis. In addition to increasing the likelihood of identifying etiologic clusters, which may implicate behavioral, environmental contamination or other preventable risk factors, this approach could enable public health officials to identify potential problems earlier and conduct preliminary evaluations of nonetiologic situations that may be of concern to the public. In so doing, the officials would be able to respond to inquiries in a more thorough, consistent, scientific, and timely manner. This is in contrast to the current situation with disease clusters, already noted, where most potentially hazardous problems are investigated only after local residents, physicians, or others have brought them to the attention of health officials, often through political pressure or media publicity. A proactive identification system could also enable more timely interventions where warranted, ranging from education to increased screening to environmental cleanup, and more rapid assessment and possible resolution of community concerns when there is a valid, alternative explanation to the perception of a disease excess.
Proactive surveillance systems have been effective for disease prevention and control when applied to infectious disease outbreaks, occupational exposures and disease (Dubrow and Wegman 1983; Whorton et al. 1983; Williams et al. 1977), and adverse reactions to pharmaceuticals (Strom 2000) (often termed postmarket drug surveillance). Similar systems for the assessment of acute outbreaks have been developed and implemented in response to concerns about outbreaks from biological, chemical, or radiologic terrorism in which rapid, scientific assessment is essential for protecting the public health (Das et al. 2003; Gesteland et al. 2003; Platt et al. 2003).
Data quality issues are again important, as detecting apparent local clusters of disease may merely indicate areas with higher-quality data registration or perhaps areas of poor data quality where there are many duplicate registrations. Specificity is also a major issue, as, given the size of the database, the range of diseases, different age and sex strata, myriad definitions of areas of various sizes and configuration, and so forth, many false-positive clusters are bound to occur. For a surveillance program to be efficient and effective, researchers must provide methods for discrimination of true alarms, false alarms (false positives), and those situations that are less clear or equivocal, so that health department officials would not be obliged to follow up all apparent aberrations. One possible approach is to survey potential local sources of risk for the specific disease in question as is done currently for many cluster reports and respond only if there is an independent source of confirmatory or consistent environmental evidence. For those disease excesses for which there is a plausible, nonenvironmental explanation, clear and thoughtful communication to concerned communities based on solid scientific evidence could help dispel their urgent concerns.
For these reasons, in common with most public health departments, we do not currently advocate carrying out surveillance for chronic disease excesses as a matter of public health practice. We believe that this type of surveillance should not be put into practice until such time as the underlying data and methodologies provide a robust framework to support this activity, as would be the requirement for screening for other public health concerns. Nonetheless, we believe that development and evaluation of surveillance approaches is an important and priority area for future research on disease clusters.
Challenges
Data Availability and Quality
To carry out small-area studies using routine data sources, the basic data need to be made available, with high quality, and the inclusion of a geographically referenced code, such as the postcode in the United Kingdom or the census block or block group in the United States. Data should include (at the least) cancer registration as well as mortality, natality, and population data. Although natality and mortality data are a statutory requirement in developed countries, not all countries (including the United States) have a national cancer registry, reducing the ability to carry out studies of environmental health problems. In the United States, the Centers for Disease Control and Prevention (CDC) has established a program in environmental public health tracking, one component of which funds states to develop additional registries of health outcomes, such as asthma, for assessment of possible environmental etiologies (http://www.cdc.gov/nceh/tracking).
In purpose-designed case–control studies, detailed evaluation of the health data and assessment of the quality of the diagnostic information (for example, case note and histology review) are likely. In contrast, for spatial epidemiologic studies that rely on routine data sources, it is usually not possible to carry out detailed validation studies of the database. However, some assessment of the basic quality of the routine data is essential to inform their use in spatial analyses, and some limited validation of the cases might be undertaken (Elliott et al. 2000a). As already noted, the denominator data may contain substantial errors, particularly in the inter-censual years at small-area scale, and for the health event data there is always the potential for diagnostic error or misclassification, especially at older ages where diagnostic tests and postmortem examinations are carried out less frequently than at younger ages. Some events may be captured poorly, if at all, in routine registers (e.g., early abortions). For others, such as cancers, case registers may be subject to double counting and underregistration as well as diagnostic inaccuracies (Best and Wakefield 1999).
One type of relevant data not readily available in the United States or the United Kingdom is the history of residential locations. For longer-latency health outcomes, such as cancer incidence and many types of mortality, knowing the residential history of an individual would be far more useful for reconstructing exposure histories than his/her location/residence at time of diagnosis or death. Even for natality data, it has been shown in small studies in both the United States and the United Kingdom that between 20 and 25% of women change residences between date of conception and delivery (Khoury et al. 1988; Nelson 2003; Shaw and Malcoe 1991). However, as many move to nearby addresses (Nelson 2003), residential exposures may not change too much.
In contrast, the Scandinavian countries maintain historical registries of residences, and these have proved invaluable, as in the example already noted of constructing exposure histories to low-frequency electromagnetic fields from overhead power lines (Feychting and Ahlbom 1993; Olsen et al. 1993; Verkasalo et al. 1993). In the future, these types of data might become available in the United Kingdom through linkage to the National Health Service (NHS) number, although there are confidentiality issues concerning use of these data. In the United States, census data provide limited migration data to and from areal units, but typically data are not available for individuals. Although knowing when and where disease occurred is useful, knowing when and where prior exposures occurred is crucial for investigating etiology.
In the future, the increasing use and availability of computerization in medical care means that large new databases of morbidity, linked to individuals, may become available. Examples include general practitioner consultations in the United Kingdom, whereas in the United States there is particular interest in syndromic surveillance (e.g., Hartman et al. 2004). The quality of such data will need careful evaluation and no doubt will vary across specialties and medical practice and over time and space. Nonetheless, they promise exciting new opportunities for carrying out spatial epidemiologic inquiries using softer end points than those currently available, and hence potentially increasing the sensitivity of the methods to detect environmental health problems.
Data Protection and Confidentiality
The current climate of legislation in the United States and the European Union is providing greater recognition of the rights of individuals to confidentiality of personal data, including health data, and the need for consent for medical investigations. In 2003, the United States brought into force the Privacy Rule (Department of Health and Human Services 2002) arising from the Health Insurance Portability and Accountability Act of 1996 (1996) that further complicates this issue. This potentially impinges on the secondary use of routine data for epidemiology (including spatial epidemiologic studies) where the data were originally collected for other purposes (e.g., health care management or delivery), but consent for their use for medical research is impracticable. In the United Kingdom, recent legislation has made it possible to use such routinely collected data without consent if certain conditions and safeguards are met. It is imperative for the future of epidemiologic research that such uses of the data are allowed to continue, provided that appropriate safeguards are in place.
In addition, with the recent increase in availability of fine-scaled, geocoded data, there is a new concern about the confidentiality of blocks, neighborhoods, and communities. The ability to acquire data and map high rates of adverse outcomes, clusters, or areas with high levels of pollutants can cause concern and outrage and possibly influence property values. Yet, rules and principles of good practices for analysts and others are still in the formative stages. Providing researchers access to these data is necessary for this field of research to progress, but implementing appropriate controls for confidentiality and protection of data is essential to maintain the trust and support of the public.
Exposure Assessment, Exposure Mapping, and Study Design Issues
The quality of the exposure data has been regarded as the Achilles heel of environmental epidemiology. This holds true for spatial epidemiology, where distance is often used as a proxy for exposure to environmental pollutants, or some other geographic measure is used, for example, plume modeling (Nyberg et al. 2000). Although the availability of GIS has greatly enhanced the capability for spatial interpolation of exposure data (Briggs and Elliott 1995), the quality of the mapping depends on the accuracy and representativeness of the available input data, as well as the inherent validity of the interpolation method used.
Such approaches may provide valid first-order approximations to group or population exposure but may not capture individual exposure well nor allow for individual variations in absorption and susceptibility. Poorly measured exposure data can produce differential errors leading to systematic bias or result in random errors or imprecision, which (unless corrected) typically lead to bias toward no effect (Bernardinelli et al. 1997). More generally, such geographic methods of exposure assessment make a number of key assumptions that may limit their applicability in given situations (Elliott and Wakefield 1999). These include the following:
equating environmental exposure (i.e., external to the individual) with biologic (internal) dose
equating current exposure with past exposure
equating modeled estimates of exposure (including distance-based measures) with true exposure
equating exposure at a point (e.g., place of residence) with total personal exposure, that is, exposure integrated over the course of daily activities as the individual moves through the exposure field
equating group exposure and group exposure–disease relationships with individual exposure and relationships at the individual level, that is, ecologic fallacy (Piantadosi et al. 1988).
An important issue in geographic analyses is the extent that the population of the areal unit is homogenous, both with respect to the environmental exposure under investigation and potential confounders. Within-unit variability in these factors could lead to bias in risk estimates (Elliott and Wakefield 2000). Recently, interest has focused on semiecologic designs that combine data on the general population with individual-level survey data (Plummer and Clayton 1996). For example, the INTERSALT study, a cross-sectional study of over 10,000 people in 32 countries, assessed both individual and group effects. There was a positive cross-population association between average rise in blood pressure with age and average levels of salt intake (measured by urinary sodium excretion) across 52 population samples in 32 countries at the group level, reflecting broad-scale population differences, and a positive relationship between urinary sodium excretion and blood pressure at the individual level (Elliott et al. 1996b). In a mortality study of cohorts of individuals from six U.S. cities, a positive association of mortality with measures of particulate matter pollution was found across those cities, adjusting for averaged site (city) effects derived from smoking, socioeconomic factors, and other potential confounding data measured at individual level (Dockery et al. 1993).
In the future, developments in exposure biomarkers (Hulka et al. 1990) and molecular epidemiology should lead to improved exposure assessment methods with increased specificity and accuracy. Although it will not be feasible to apply these methods to large numbers of people, collection of such data on small subsamples of the population will aid in validation of the exposure model and provide information on within-area variability in the exposure data and potentially on confounders. This may reduce bias and provide improved risk estimates, and hence strengthen any causal inferences (Guthrie and Sheppard 2001).
One of the opportunities presented by GIS technology is the adaptation of traditional study designs to the spatial context. For example, one of the most vexing problems for epidemiologists occurs when both the disease and environmental exposure under investigation are rare. Both the case–control and the cohort approach are likely to be costly and/or difficult because of issues of representativeness and sample size. As an alternative, hybrid designs have been used: the nested case–control (Paddle 1981) or the case–cohort study (Kupper et al. 1975), or more complex approaches such as two-stage sampling with oversampling of both exposed and diseased individuals (Rothman and Greenland 1998). This, too, can be cumbersome and costly.
GIS technology may offer a more efficient and cost-effective solution, at least for exposures that can be readily characterized geographically (Wartenberg 1994). With this approach, a nested case–control or case–cohort study can be conducted within a large-scale population-based cohort by specifying a geographic subset of the cohort with high relative exposure, on average, for direct study. For example, epidemiologic studies of the possible association between exposure to magnetic fields and the incidence of childhood leukemia have been limited by the low prevalence of high exposures because the higher exposures are relatively rare and widely dispersed: less than 10% of children with exposures above even twice the average background, less than 3% above three times, and less than 1% above four times the average background exposures (Ahlbom et al. 2000; Greenland et al. 2000; Zaffanella 1993). Case–control studies have consequently ended up with few children with high exposures and no obvious high-exposure cohort. The resulting small quantitative difference between exposed and unexposed individuals in these studies has limited their sensitivity and ability to yield a consistent and conclusive result (Wartenberg 2001).
In a demonstration project, a cohort of children with a far higher likelihood of being exposed to high levels of magnetic fields was identified using a geographically defined population living within 0.5 miles of a high-voltage electric power transmission line (Wartenberg et al. 1993, 1997). Because of the relatively low population density in the entire study region (New York State), results were of limited sensitivity, though modification and improvements to this design approach look promising.
Conclusions
Advances in GIS and statistical methodology together with the availability of high-resolution, geographically referenced health databases present unprecedented new opportunities to investigate the environmental, social, and behavioral factors underlying geographic variations in disease rates at small-area scale. Such studies must be guided by good questions, excellent statistical methodology, and sound epidemiologic principles, including taking proper account of problems of data quality and the potential for bias and confounding. Spatial epidemiologic studies will become increasingly common in the future, both because of the instant visual appeal and wide availability of the new geographic techniques, and the desire for cleaner and healthier environments. With ongoing improvements in the data and methodologies, these studies will play an increasingly important role in our understanding of the complex relationships between environment and health.
This article is part of the mini-monograph “Health and Environment Information Systems for Exposure and Disease Mapping, and Risk Assessment.”
Figure 1 Percentage of homes built before 1950 in New Jersey based on U.S. census data reported at the block group level of resolution. The three maps depict the same data at three different scales: U.S. census block group, ZIP codes, and counties.
Figure 2 Adult leukemia by electoral ward in West Midlands Region, England, 1974–1986. (A). SMR; West Midlands = 1.0. (B) SMR after smoothing using empirical Bayes methods. Figure reproduced from Olsen et al. (1996), with permission of the BMJ Publishing Group.
==== Refs
References
Ahlbom A Day N Feychting M Roman E Skinner J Dockerty J 2000 A pooled analysis of magnetic fields and childhood leukemia Br J Cancer 83 692 698 10944614
Alexander FE Williams J McKinney PA Cartwright RA Ricketts TJ 1989 A specialist leukaemia/lymphoma registry in the UK. Part 2. clustering of Hodgkin’s disease Br J Cancer 60 948 952 2605104
Aschengrau A Paulu C Ozonoff D 1998 Tetrachloroethylene-contaminated drinking water and the risk of breast cancer Environ Health Perspect 106 suppl 4 947 953 9703477
Aylin P Maheswaran R Wakefield J Cockings S Jarup L Arnold R 1999 A national facility for small area disease mapping and rapid initial assessment of apparent disease clusters around a point source: the UK Small Area Health Statistics Unit J Public Health Med 21 289 298 10528956
Beasley RP 1988 Hepatitis B virus. The major aetiology of hepatocellular carcinoma Cancer 61 1942 1956 2834034
Bernardinelli L Pascutto C Best NG Gilks WR 1997 Disease mapping with errors in covariates Stat Med 16 741 752 9131762
Berry M Bove F 1997 Birth weight reduction associated with residence near a hazardous waste landfill Environ Health Perspect 105 856 861 9347901
Best NG Wakefield JC 1999 Accounting for inaccuracies in population counts and case registration in cancer mapping studies J R Stat Soc Ser A 162 363 382
Black RJ Sharp L Finlayson AR Hardness EF 1994 Cancer incidence in a population potentially exposed to radium-226 at Dalgety Bay Br J Cancer 69 140 143 8286196
Blot WJ Chow WH McLaughlin JK 1997 Wood dust and nasal cancer risk J Occup Environ Med 39 148 156 9048321
Blot WJ Fraumeni JF Jr 1975 Arsenical air pollution and lung cancer Lancet 2 142 144 49742
Blot WJ Fraumeni JF Jr 1977 Geographic patterns of oral cancer in the United States: etiologic implications J Chronic Dis 30 745 757 591603
Blot WJ Fraumeni J Jr 1982. Geographic epidemiology of cancer in the United States. In: Cancer Epidemiology and Prevention (Schottenfeld D, Fraumeni J Jr, eds). New York:W.B. Saunders, 179–193.
Blot WJ Fraumeni JF Jr 1994. Arsenic and lung cancer. In: Epidemiology of Lung Cancer (Samet JM, ed). New York:Marcel Dekker, 207–218.
Brewer CA Pickle L 2002 Comparison of methods for classifying epidemiological data on choropleth maps in series Ann Assoc Am Geogr 92 662 681
Briggs DJ Elliott P 1995 The use of geographical information systems in studies on environment and health World Health Stat Q 48 85 94 8585238
Brinton LA Blot WJ Becker JA Winn DM Browder JP Farmer JC Jr 1984 A case-control study of cancers of the nasal cavity and paranasal sinuses Am J Epidemiol 119 896 906 6731431
Brinton LA Blot WJ Fraumeni JF Jr 1985 Nasal cancer in the textile and clothing industries Br J Ind Med 42 469 474 4015994
Brinton LA Blot WJ Stone BJ Fraumeni JF Jr 1977 A death certificate analysis of nasal cancer among furniture workers in North Carolina Cancer Res 37 3473 3474 908001
Brinton LA Stone BJ Blot WJ Fraumeni JF Jr 1976 Nasal cancer in U.S. furniture industry countries Lancet 2 628 61362
Caramaschi F del Corno G Favaretti C Giambelluca SE Montesarchio E Fara GM 1981 Chloracne following environmental contamination by TCDD in Seveso, Italy Int J Epidemiol 10 135 143 6457009
Cardis E Richardson D Kesminiene A 2001 Radiation risk estimates in the beginning of the 21st century Health Phys 80 349 361 11281203
CDC 2003. Draft Framework for Evaluating Syndromic Surveillance Systems for Bioterrrorism Preparedness. Atlanta, GA:Centers for Disease Control and Prevention. Available at http://www.cdc.gov/epo/dphsi/syndromic/framework.htm [accessed 13 May 2004].
Clayton D Hills M 1993. Nested case-control studies. In: Statistical Methods in Epidemiology. Oxford:Oxford University Press, 329–335.
Clayton DG Kaldor J 1987 Empirical Bayes estimates of age-standardized relative risks for use in disease mapping Biometrics 43 671 681 3663823
Corburn J 2002 Environmental justice, local knowledge, and risk: the discourse of a community-based cumulative exposure assessment Environ Manage 29 451 466 12071497
Das D Weiss D Mostashari F Treadwell T McQuiston J Hutwagner L 2003 Enhanced drop-in syndromic surveillance in New York City following September 11, 2001 J Urban Health 80 2 suppl 1 i76 i88 12791782
Department of Health and Human Services. 2002 Standards for Privacy of Individually Identifiable Health Information; Final Rule Fed Reg 67 157 53182 53273
Devesa SS Grauman DG Blot WJ Pennello G Hoover RN Fraumeni JF Jr 1999. Atlas of Cancer Mortality in the United States, 1950–1994. Washington, DC:U.S. Government Printing Office.
Dockery DW Pope CA III Xu X Spengler JD Ware JH Ray ME 1993 An association between air pollution and mortality in six US cities N Engl J Med 329 1753 1759 8179653
Dolk H Elliott P Shaddick G Walls P Grundy C 1997a Cancer incidence near high power radio and TV transmitters in Great Britain. II. All transmitter sites Am J Epidemiol 145 10 17 8982017
Dolk H Mertens B Kleinschmidt I Walls P Shaddick G Elliott P 1995 A standardisation approach to the control of socioeconomic confounding in small area studies of environment and health J Epidemiol Community Health 49 suppl 2 S9 S14 8594139
Dolk H Shaddick G Walls P Grundy C Elliott P 1997b Cancer incidence near high power radio and TV transmitters in Great Britain. I: Sutton Coldfield transmitter Am J Epidemiol 145 1 9 8982016
Doll R 1980 The epidemiology of cancer Cancer 45 2475 2485 7378985
Drijver M Woudenberg F 1999 Cluster management and the role of concerned communities and the media Eur J Epidemiol 15 863 869 10608367
Dubrow R Wegman DH 1983 Setting priorities for occupational cancer research and control: synthesis of the results of occupational disease surveillance studies J Natl Cancer Inst 71 1123 1142 6581356
Elliott P Briggs D Morris S de Hoogh C Hurt C Jensen TK 2001 Risk of adverse birth outcomes in populations living near landfill sites Br Med J 323 363 368 11509424
Elliott P Eaton N Shaddick G Carter R 2000a Cancer incidence near municipal solid waste incinerators in Great Britain. Part 2: Histopathological and case-note review of primary liver cancer cases Br J Cancer 82 1103 1106 10737393
Elliott P Hills M Beresford J Kleinschmidt I Jolley D Pattenden S 1992a Incidence of cancer of the larynx and lung near incinerators of waste solvents and oils in Great Britain Lancet 339 854 858 1347867
Elliott P Kleinschmidt I 1997 Angiosarcoma of the liver in Great Britain in proximity to vinyl chloride sites Occup Environ Med 54 14 18 9072028
Elliott P Martuzzi M Shaddick G 1995 Spatial statistical methods in environmental epidemiology: a critique Stat Methods Med Res 4 137 159 7582202
Elliott P Shaddick G Kleinschmidt I Jolley D Walls P Beresford J 1996a Cancer incidence near municipal solid waste incinerators in Great Britain Br J Cancer 73 702 710 8605111
Elliott P Stamler J Nichols R Dyer AR Stamler R Kesteloot H 1996b Intersalt revisited: further analyses of 24 hour sodium excretion and blood pressure within and across populations. Intersalt Cooperative Research Group Br Med J 312 1249 1253 8634612
Elliott P Wakefield J 1999. Small area studies of environment and health. In: Statistics for the Environment 4: Health and the Environment (Barnett V, Turkman KF, Stein A, eds). Chichester, UK:John Wiley, 3–27.
Elliott P Wakefield JC 2000. Bias and confounding in spatial epidemiology. In: Spatial Epidemiology: Methods and Applications (Elliott P, Wakefield J, Best N, Briggs DJ, eds). Oxford:Oxford University Press, 68–84.
Elliott P Wakefield J Best N Briggs DJ 2000b. Spatial epidemiology: methods and applications. In: Spatial Epidemiology: Methods and Applications (Elliott P, Wakefield J, Best N, Briggs DJ, eds). Oxford:Oxford University Press, 3–14.
Elliott P Westlake AJ Kleinschmidt I Hills M Rodrigues L McGale P 1992b The Small Area Health Statistics Unit: a national facility for investigating health around point sources of environmental pollution in the United Kingdom J Epidemiol Community Health 46 345 349 1431704
EUROHEIS London:Imperial College. Available: http://www.med.ic.ac.uk/divisions/60/euroheis/homepage.htm [accessed 28 July 2003].
Feychting M Ahlbom A 1993 Magnetic fields and cancer in children residing near Swedish high-voltage power lines Am J Epidemiol 138 467 481 8213751
Fraumeni J Jr 1988. Etiologic insights from cancer mapping. In: Unusual Occurrences as Cluster to Cancer Etiology (Miller RW, Wantanabe S, Fraumeni J Jr, Sugimara T, Takayama S, Sugano H, eds). Tokyo:Japanese Scientific Society Press, 13–25.
Gesteland PH Gardner RM Tsui FC Espino JU Rolfs RT James BC 2003 Automated syndromic surveillance for the 2002 Winter Olympics J Am Med Inform Assoc 10 547 554 12925547
Goldberg MS al-Homsi N Goulet L Riberdy H 1995 Incidence of cancer among persons living near a municipal solid waste landfill site in Montreal, Quebec Arch Environ Health 50 416 424 8572719
Greenberg M Wartenberg D 1991 Communicating to an alarmed community about cancer clusters: a fifty state survey J Community Health 16 71 82 1856311
Greenland S Sheppard AR Kaune WT Poole C Kelsh MA 2000 A pooled analysis of magnetic fields, wire codes and childhood leukemia Epidemiology 11 624 634 11055621
Guthrie KA Sheppard L 2001 Overcoming biases and misconceptions in ecological studies J R Stat Soc Series A 164 141 154
Hardy RJ Schroder GD Cooper SP Buffler PA Prichard HM Crane M 1990 A surveillance system for assessing health effects from hazardous exposures Am J Epidemiol 132 S32 S42 2356834
Hartman J Mostashari F Olsiewski P Sosin D Available: http://www.syndromic.org [accessed 11 May 2004].
Health Insurance Portability and Accountability Act of 1996 1996. Public Law 104-191. U.S. Congress, 21 August 1996
Hoover R Mason TJ McKay FW Fraumeni JF Jr 1975 Cancer by county: new resource for etiologic clues Science 189 1005 1007 1220005
Hulka BS Wilcosky TC Griffith JD 1990. Biological Markers in Epidemiology. Oxford:Oxford University Press.
Infante-Rivard C Amre DK 2001 Congenital anomalies in children with acute lymphoblastic leukemia and in their family Int J Epidemiol 30 350 352 11369741
Järup L Briggs D De Hoogh C Morris S Hurt C Lewin A 2002 Cancer risks in populations living near landfill sites in Great Britain Br J Cancer 86 1732 1736 12087458
Julious SA Nicholl J George S 2001 Why do we continue to use standardized mortality ratios for small area comparisons? J Public Health Med 23 40 46 11315692
Keys A 1980. Seven Countries—A Multivariate Analysis of Death and Coronary Heart Disease. Boston:Harvard University Press.
Khoury M Stewart W Weinstein A Panny S Lindsay P Eisenberg M 1988 Residential mobility during pregnancy: implications for environmental teratogenesis J Clin Epidemiol 41 15 20 3335868
Kokki E Ranta J Penttinen A Pukkala E Pekkanen J 2001 Small area estimation of incidence of cancer around a known source of exposure with fine resolution data Occup Environ Med 58 315 320 11303080
Kulldorff M 1997 A spatial scan statistic Commun Stat Theory Meth 26 1481 1496
Kulldorff M 2001 Prospective time periodic geographical disease surveillance using a scan statistic J R Stat Soc Ser A 164 61 72
Kulldorff M Feuer EJ Miller B Freedman L 1997 Breast cancer clusters in the Northeast United States: a geographic analysis Am J Epidemiol 146 161 170 9230778
Kupper LL McMichael AJ Spirtas R 1975 A hybrid epidemiologic design useful in estimating relative risk J Am Stat Assoc 70 524 528
Lopez-Abente G Aragones N Pollan M 2001 Solid-tumor mortality in the vicinity of uranium cycle facilities and nuclear power plants in Spain Environ Health Perspect 109 721 729 11485872
Maheswaran R Morris S Falconer S Grossinho A Perry I Wakefield J 1999 Magnesium in drinking water supplies and mortality from acute myocardial infarction Heart 82 455 460 10490560
Mason TJ 1976 Geographic patterns of cancer risk: a means for identifying possible occupational factors Ann NY Acad Sci 271 370 376 1069526
Mason TJ McKay FW Hoover R Blot WJ Fraumeni JF Jr 1975. Atlas of Cancer Mortality for U.S. Counties, 1950–1969. Washington, DC:U.S. Government Printing Office.
Mason TJ McKay FW Hoover R Blot WJ Fraumeni JF Jr 1976. Atlas of Cancer Mortality among U.S. Nonwhites: 1950–1969. Washington, DC:U.S. Government Printing Office.
Monmonier M 1997. How to Lie with Maps. Chicago:The University of Chicago Press.
Moolgavkar SH 1999. Stochastic models for estimation and prediction of cancer risk. In: Statistics for the Environment. 4: Health and the Environment (Barnett V, Stein A, Turkman KF, eds). Chichester, UK:Wiley, 237–257.
Morris MS Knorr RS 1996 Adult leukemia and proximity-based surrogates for exposure to Pilgrim plants nuclear emissions Arch Environ Health 51 266 274 8757406
Morris SE Wakefield JC 2000. Assessment of disease risk in relation to a pre-specified source. In: Spatial Epidemiology: Methods and Applications (Elliott P, Wakefield J, Best N, Briggs DJ, eds). Oxford:Oxford University Press, 153–184.
Mulder YM Drijver M Kreis IA 1994 Case-control study on the association between a cluster of childhood haematopoietic malignancies and local environmental factors in Aalsmeer, The Netherlands J Epidemiol Community Health 48 161 165 8189171
Nelson P 2003. Geographical Epidemiology of Hypospadias: Small Area study of Birth Prevalence [PhD Thesis]. London:University of London.
Neutra RR 1990 Counterpoint from a cluster buster Am J Epidemiol 132 1 8 2356803
Nyberg F Gustavsson P Jarup L Bellander T Berglind N Jakobsson R 2000 Urban air pollution and lung cancer in Stockholm Epidemiology 11 487 495 10955399
Olsen JH Nielsen A Schulgen G 1993 Residence near high-voltage facilities and the risk of cancer in children Br Med J 307 891 895 8241850
Olsen SF Martuzzi M Elliott P 1996 Cluster analysis and disease mapping—why, when and how? Br Med J 313 863 866 8870578
Openshaw S 1984. The Modifiable Areal Unit Problem. Norwich, UK:Geo Books.
Paddle GM 1981. A strategy for the identification of carcinogens in a large, complex chemical company. In: Quantification of Occupational Cancer, Vol 9 (Peto R, Schneiderman M, eds). Cold Spring Harbor, NY:Cold Spring Harbor Laboratory.
Piantadosi S Byar DP Green SB 1988 The ecological fallacy Am J Epidemiol 127 893 904 3282433
Pickle LW Mungiole M Jones GK White AA 1996. Atlas of United States Mortality. Hyattsville, MD:National Center for Health Statistics.
Platt R Bocchino C Caldwell B Harmon R Kleinman K Lazarus R 2003 Syndromic surveillance using minimum transfer of identifiable data: the example of the National Bioterrorism Syndromic Surveillance Demonstration Program J Urban Health 80 2 suppl 1 i25 i31 12791776
Plummer M Clayton D 1996 Estimation of population exposure in ecological studies J R Stat Soc Ser B 58 113 126
Richardson S Montfort C 2000. Ecological correlation studies. In: Spatial Epidemiology: Methods and Applications (Elliott P, Wakefield J, Best N, Briggs D, eds). Oxford:Oxford University Press, 205–220.
Richardson S Thomson A Best N Elliott P 2004 Interpreting posterior relative risk estimates in disease-mapping studies Environ Health Perspect 112 1016 1025 15198922
Rogerson PA 1997 Surveillance systems for monitoring the development of spatial systems Stat Med 16 2081 2093 9308133
Rogerson PA 2001 Monitoring point patterns for the development of space-time clusters J R Stat Soc Ser A 164 87 96
Rothman KJ 1990 A sobering start for the cluster busters’ conference Am J Epidemiol 132 suppl 1 S6 S13 2356837
Rothman KJ Greenland S 1998. Modern Epidemiology. Philadelphia:Lippincott-Raven Publishers.
Rushton G Krishnamurty R Krishnamurti D Lolonis P Song H 1996 The spatial relationship between infant mortality and birth defect rates in a U.S. city Stat Med 15 1907 1919 8888483
Rushton G Lolonis P 1996 Exploratory spatial analysis of birth defect rates in an urban population Stat Med 15 717 726 9132899
Shaper AG Packham RF Pocock SJ 1980 The British regional heart study: cardiovascular mortality and water quality J Environ Pathol Toxicol 143 456 462
Shaw GM Malcoe LH 1991 Residential mobility during pregnancy for mothers of infants with or without congenital cardiac anomalies Arch Environ Health 45 310 312
Smans M Esteve J 1992. Practical approaches to disease mapping. In: Geographical and Environmental Epidemiology: Methods for Small-Area Studies (Elliott P, Cuzick J, English D, Stern R, eds). Oxford:Oxford University Press, 141–157.
Smith G Neaton JD Wentworth D Stamler R Stamler J 1996a Socioeconomic differentials in mortality risk among men screened for the Multiple Risk Factor Intervention Trial. I. White men Am J Public Health 86 486 496 8604778
Smith G Neaton JD Wentworth D Stamler R Stamler J 1996b Socioeconomic differentials in mortality risk among men screened for the Multiple Risk Factor Intervention Trial. II. Black men Am J Public Health 86 497 504 8604779
Staessen JA Roels HA Emelainov D Kuznetsova T Thijs L Vangronsveld J 1999 Environmental exposure to cadmium, forearm bone density, and risk of fractures: prospective population study Lancet 353 1140 1144 10209978
Stocks P 1936 Distribution in England and Wales of cancer of various sites Ann Rep Br Empire Cancer Campaign 13 239 280
Stocks P 1937 Distribution in England and Wales of cancer of various sites Annu Rep Br Empire Cancer Campaign 14 198 223
Stocks P 1939 Distribution in England and Wales of cancer of various sites Annu Rep Br Empire Cancer Campaign 16 308 343
Strom BL 2000. Pharmacoepidemiology. New York:John Wiley and Sons.
Swerdlow A dos Santos Silva I 1993. Atlas of Cancer Incidence in England and Wales 1968–85. Oxford:Oxford University Press.
Thacker SB Berkelman RL 1992. History of public health surveillance. In: Public Health Surveillance (Halperin W, Baker LE, Monson RR, eds). New York:Van Nostrand Reinhold, 1–15.
Trumbo CW 2000 Public requests for cancer cluster investigations: a survey of state health departments Am J Public Health 90 1300 1302 10937014
Verkasalo P Pukkala E Hongisto M Valjus J Jarvinen P Heikkila K 1993 Risk of cancer in Finnish children living close to power lines Br Med J 307 895 899 8241851
Wakefield JC Kelsall JE Morris SE 2000. Clustering, cluster detection, and spatial variation in risk. In: Spatial Epidemiology: Methods and Applications (Elliott P, Wakefield J, Best N, Briggs DJ, eds). Oxford:Oxford University Press, 128–152.
Waller LA Lawson AB 1995 the power of focused tests to detect disease clustering Stat Med 14 2291 2308 8711270
Walter SD 2000. Disease mapping: a historical perspective. In: Spatial Epidemiology: Methods and Applications (Elliott P, Wakefield J, Best N, Briggs DJ, eds). Oxford:Oxford University Press, 223–252.
Walter SD Birnie SE 1991 Mapping mortality and morbidity patterns: an international comparison Int J Epidemiol 20 678 689 1955252
Wartenberg D 1994. Use of Geographic Information Systems for Risk Screening and Epidemiology. (Andrews JS Jr, Frumkin H, Johnson BL, Mehlman MA, Xintaras C, Bucsela JA, eds). Princeton, NJ:Princeton Scientific Publishing, 853–859.
Wartenberg D 1995 Should we boost or bust cluster investigations [Editorial] Epidemiology 6 575 576 8589086
Wartenberg D 2001 Investigation disease clusters: why, when and how J R Stat Soc Ser A 164 13 22
Wartenberg D Greenberg M Lathrop R 1993 Identification and characterization of populations living near high voltage transmission lines: a pilot study Environ Health Perspect 101 626 631 8143596
Wartenberg D Greenberg M Lathrop R Manning R Brown S 1997. Using a geographic information system to identify populations living near high voltage electric power transmission lines in New York State. In: Proceedings of the International Symposium on Computer Mapping in Epidemiology and Environmental Health (Aangeenbrug RT, Leaverton PE, Mason TJ, Tobin GA, eds). Alexandria, VA:World Computer Graphics Foundation, 300–311.
Whorton MD Schulman J Larson SR Stubbs HA Austin D 1983 Feasibility of identifying high-risk occupations through tumor registries J Occup Med 25 657 660 6631564
Wilkinson P Thakrar B Shaddick G Stevenson S Pattenden S Landon M 1997 Cancer incidence and mortality around the Pan Britannica Industries pesticide factory, Waltham Abbey Occup Environ Med 54 101 107 9072017
Wilkinson P Thakrar B Walls P Landon M Falconer S Grundy C 1999 Lymphohaematopoietic malignancy around all industrial complexes that include major oil refineries in Great Britain Occup Environ Med 56 577 580 10615289
Williams RR Stegens NL Goldsmith JR 1977 Associations of cancer site and type with occupation and industry from the Third National Cancer Survey Interview J Natl Cancer Inst 59 1147 1185 903993
Winn DM Blot WJ Shy CM Pickle LW Toledo A Fraumeni JF Jr 1981 Snuff dipping and oral cancer among women in the southern United States N Engl J Med 304 745 749 7193288
Wrensch M Yost M Miike R Lee G Touchstone J 1999 Adult glioma in relation to residential power frequency electromagnetic field exposures in the San Francisco Bay area Epidemiology 10 523 527 10468425
Zaffanella L 1993. Survey of Residential Magnetic Field Sources. TR-102759 (2 vols). Palo Alto, CA:Electric Power Research Institute.
|
15198920
|
PMC1247193
|
CC0
|
2021-01-04 23:39:44
|
no
|
Environ Health Perspect. 2004 Jun 15; 112(9):998-1006
|
utf-8
|
Environ Health Perspect
| 2,004 |
10.1289/ehp.6735
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0112-00100710.1289/ehp.673815198921Mini-Monograph: Information SystemsUsing Geographic Information Systems for Exposure Assessment in Environmental Epidemiology Studies Nuckols John R. 1Ward Mary H. 2Jarup Lars 31Department of Environmental and Radiological Health Sciences, Colorado State University, Fort Collins, Colorado, USA2Occupational and Environmental Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Bethesda, Maryland, USA3Small Area Health Statistics Unit, Department of Epidemiology and Public Health, Imperial College London, London, United KingdomAddress correspondence to J.R. Nuckols, 125 EHB, Department of Environmental and Radiological Health Sciences, Colorado State University, Fort Collins, CO 80523 USA. Telephone: (970) 491-7295. Fax: (970) 491-2940. E-mail:
[email protected] acknowledgement is given to S. Weigel for her contribution to the text used in the section on geospatial sciences, and to P. Stewart (OEEB-NCI) for her input on the exposure assessment process.
Preparation of this article was funded in part by an intergovernmental personnel agreement between OEEB-NCI-NIH-DHHS and Colorado State University.
The authors declare they have no competing financial interests.
6 2004 15 4 2004 112 9 1007 1015 12 9 2003 25 3 2004 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Geographic information systems (GIS) are being used with increasing frequency in environmental epidemiology studies. Reported applications include locating the study population by geocoding addresses (assigning mapping coordinates), using proximity analysis of contaminant source as a surrogate for exposure, and integrating environmental monitoring data into the analysis of the health outcomes. Although most of these studies have been ecologic in design, some have used GIS in estimating environmental levels of a contaminant at the individual level and to design exposure metrics for use in epidemiologic studies. In this article we discuss fundamentals of three scientific disciplines instrumental to using GIS in exposure assessment for epidemiologic studies: geospatial science, environmental science, and epidemiology. We also explore how a GIS can be used to accomplish several steps in the exposure assessment process. These steps include defining the study population, identifying source and potential routes of exposure, estimating environmental levels of target contaminants, and estimating personal exposures. We present and discuss examples for the first three steps. We discuss potential use of GIS and global positioning systems (GPS) in the last step. On the basis of our findings, we conclude that the use of GIS in exposure assessment for environmental epidemiology studies is not only feasible but can enhance the understanding of the association between contaminants in our environment and disease.
environmental epidemiologyexposure assessmentgeographic information systems
==== Body
Environmental epidemiology is an area of epidemiology concerned with the study of associations between environmental exposures and health outcomes, with the purpose of further understanding the etiology of disease. The term “environment” implies a spatial context. Thus, the study of interactions between humans and their environment requires spatial information and analysis. Geographic information system (GIS) software allows environmental and epidemiologic data to be stored, analyzed, and displayed spatially. The logical structure and functionality of a GIS are shown in Figure 1 (Falbo et al. 1991). Data collection can be accomplished by importing tabular or digital data that are referenced with map coordinates defining their geographic position. The data are entered into a database where they are stored as a map with a specified theme (termed “data layer”). Tabular (attribute) data corresponding to the theme can be stored with each data layer. Analytical functions within the software can be used to process and manipulate both map and attribute data through linkages established within the GIS. Two types of output are common: tabular (summary data, statistics, reports) and cartographic (maps, map files, and map overlays). Several publications describe the structure and functionality of a GIS more thoroughly (Chrisman 2002; DeMers 2000). Vine et al. (1997) provide an overview of the use of specific functions in GIS software that could be useful in environmental epidemiologic research. Beyea and Hatch (1999) provide an in-depth discussion of geographic modeling for exposure assessment in environmental epidemiology, as well as an extensive literature review. Briggs and Elliot (1995) provide a review of spatial analysis and mapping in environmental health.
GIS have been used at different levels of sophistication in environmental epidemiology studies. These uses range from simply locating the study population by geocoding addresses (assigning mapping coordinates) to using proximity to contaminant source as a surrogate for exposure (Bell et al. 2001; Comba et al. 2003; Langholz et al. 2002; Xiang et al. 2000) to integrating environmental monitoring data into the analysis of the health outcomes (Floret et al. 2003; Gallagher et al. 1998; Reynolds et al. 2002a, 2002b, 2003). However, most of the latter studies have been ecologic in design; relatively few studies have used GIS in estimating environmental levels of a contaminant at the individual level (Nyberg et al. 2000; Reif et al. 2003; Rogers et al. 2000). A number of studies have used GIS to design exposure metrics for use in epidemiologic studies (Bellander et al. 2001; Brody et al. 2002; Cicero-Fernandez et al. 2001; Gunier et al. 2001; Inserra et al. 2002; Kohli et al. 1997; Rull and Ritz 2003; Swartz et al. 2003; Ward et al. 2000). Although yet to be applied in the context of an epidemiologic analysis, several studies have investigated the use of GIS in estimating activity patterns of the study population for potential linkage to environmental data to refine personal exposure estimates (Elgethun et al. 2003; Phillips et al. 2001). Similarly, the use of GIS in spatial statistics for linking exposure and health data in the context of epidemiologic analysis is a growing field of research (Ali et al. 2002; Christakos and Serre 2000; Elliott et al. 2001). This article is a discussion of the fundamentals of the scientific disciplines required to use GIS in exposure assessment for epidemiologic studies and explores how a GIS can be used to accomplish several steps in the exposure assessment process (those shaded blue in Figure 2). Specifically these steps are a) defining the study population, b) identifying source and potential routes of exposure, c) estimating environmental levels of target contaminants, and d ) estimating personal exposures.
Fundamentals of GIS Application in Exposure Assessment
Using GIS in exposure assessment for epidemiologic studies requires knowledge and expertise in at least three core scientific areas: geospatial sciences, environmental sciences, and epidemiology.
Geospatial Science
For a GIS to accurately represent occurrences on the earth’s surface, the location of data must be reliable, accurate, and pertinent (Falbo et al. 1991). Geospatial science is the systematic study of geographic variables relating to, occupying, or having the character of space. Fundamental elements of geospatial sciences relevant to GIS applications in exposure assessment include data representation, scale, and accuracy. Data representation is the format of the unit of analysis used in the GIS. The most commonly used representations of space in a GIS are the raster and vector data models. In the raster model, grid cells serve as the basic units of analysis. An example would be pixels of remotely sensed imagery from satellite imagery. The vector model uses points, lines, or polygons based on continuous geometry of space to represent data. Other, more specialized data models are available in most GIS software. For example, the triangulated irregular network (TIN) model provides an efficient means of representing elevation data often used for terrain analysis. GIS software contain algorithms for translating between formats, for example, raster → vector, vector → raster, point → TIN, although some error may be introduced by these data transformation processes. More complete information on data models can be found in textbooks such as those by Chrisman (2002) and DeMers (2000).
Selection of scale is perhaps the most important factor in creating and analyzing GIS databases for exposure assessment and epidemiology. The following is a list of definitions of the the scaling factors most likely to be encountered in an epidemiology study:
Cartographic scale: Traditional map scale ratio relates the size of a feature on the ground to the size of a feature on the map. This is the scale normally listed on a road map. Scale selection results in the amount of detail including roads, water bodies, and land use patterns.
Geographic extent: Refers to the size of the study area. For example, a study can be regional scale or global scale. The extent of the study area and/or its subsets can affect the analysis results (e.g., different results might be obtained when looking at cancer incidence in one state or province versus nationwide).
Spatial resolution: Refers to the grain, or smallest, unit that is distinguishable. Map data at different scales will allow for resolution of different objects. For example, a house site represented on a 1:24,000 scale map would not appear on a 1:100,000 scale map. In remotely sensed imagery, resolution is directly related to the pixel size, the area on the ground from which the radiances are integrated. Lower resolution pixel (1 km2) data may be less useful than higher resolution pixel Landsat data (30 m2) for some environmental health studies.
Operational scale: Refers to the scale at which the process of interest occurs. For example, contaminant transport may occur at a small or large scale. Processes can be resolution dependent, that is, they can be detected at one scale but not another.
Homogeneity and heterogeneity of spatial data are affected by scale, and the scale chosen may affect the ability of the study to detect a relationship between the environmental exposure and the health outcome. This issue is similar to the modifiable areal unit problem, a term introduced by Openshaw and Taylor (1979) that has long been recognized as an issue in the analysis of aggregated data such as disease incidence rates and census enumeration (Fotheringham and Wong 1991; Holt et al. 1996). For example, studies of disease incidence reported at the county level require the environmental data to be aggregated to an exposure metric at the same resolution. Such aggregation may obscure intracounty variation in exposure (operational scale) and thus the relationship between the target contaminant and the disease.
Accuracy can be defined as how well the GIS data represent reality in terms of positional, attribute, and temporal accuracy. Positional accuracy relates to the agreement between data representation in the GIS and actual location of the data, or “ground truth.” Attribute accuracy is a measure of how well information linked to the data representation format is correct (e.g., is the line segment tagged with the correct street information?). Temporal accuracy concerns the appropriateness of using a particular snapshot or snapshots of time for a particular GIS-based analysis or modeling effort. For example, temporal accuracy would reflect how well using a single-year crop map would reflect proximity to pesticide use for exposure assessment of a particular disease outcome. Errors in GIS can be categorized as source errors or processing errors. Source errors relate to the accuracy of the data per se, that is, the differences between the data in the GIS and reality. For example, geocoding is often used to estimate the location of residences and pollutant sources; however, the positional error generated at this first step in the exposure assessment process is rarely evaluated. A study by Krieger et al. (2001) compared geocoding firms and found widely varying geocoding success rates as well as large differences in the accuracy of census tract assignment. The positional accuracy of geocoded addresses in epidemiology studies was evaluated in a breast cancer case–control study in western New York (Bonner et al. 2003) and in a non-Hodgkin lymphoma case–control study in Iowa (Ward et al., in press). The positional errors were comparable in the two studies; the majority of homes were geocoded to within 100 meters of their location determined by GPS. However, positional errors were greater for homes outside the large metropolitan areas (Bonner et al. 2003), and rural addresses in Iowa had a median positional error of around 200 meters (Ward et al. submitted).
Processing errors can be introduced into the database as a result of GIS-based analysis and modeling. For each layer of data combined in a GIS analysis, additional uncertainty in the analysis process will be introduced because of error propagation. Beyea and Hatch (1999) provide an in-depth discussion of uncertainty in GIS-based exposure modeling.
Environmental Science
Environmental science is the systematic study of the complex of physical, chemical, and biotic factors that act upon on an organism or an ecologic community and ultimately determine its form and survival. It can include circumstances, objects, or conditions by which an organism or community is surrounded and the aggregate of social and cultural conditions that influence the life of an individual or community. Fundamental elements of environmental science relevant to GIS applications in exposure assessment include measurement data and predictive algorithms for fate and transport of chemical compounds in the environment.
Environmental science studies rely heavily on measurement data of the factors that influence life. Institutions in almost every country in the world, such as the U.S. Environmental Protection Agency (U.S. EPA), have been established with a primary mission of collecting and analyzing environmental samples to understand the impact of these factors on the health of the earth’s ecosystem. As a result, an abundance of measurement data concerning the chemical composition of air and water resources is available to environmental epidemiology studies. A basic principle in environmental sciences is that measurement data should be used within the bounds of the purpose for which the sample was collected. Often this purpose is to define regional or systematic trends in environmental quality at a scale and resolution that may not be adequate for epidemiologic studies, especially studies of individuals. For example, public water utilities operating in the United States with a service population > 10,000 are required by federal law to report levels of certain byproducts of the disinfection process to the U.S. EPA. Most utilities meet this requirement by taking four samples at different locations in their water distribution system every 3 months. Although this sampling design may be sufficient to indicate compliance with the law, it may not be sufficient to adequately encompass the spatial and temporal variability in exposure necessary to classify exposure to individuals using the water.
Environmental scientists often use computer-based simulation models to supplement measurement data in environmental studies. These models are generally composed of mathematic algorithms designed to predict interaction between, and effect of the complex factors on, an organism or ecologic community. The models can be stochastic (based on statistical probability) or deterministic (based on physical processes). In either case the models are dependent on measurement data for calibration of the predictive algorithms and validation of the predicted results. A fundamental rule in environmental modeling is not to transfer use of a model from one geographic region to another without validating it with measurement data from the new study area. Often such model transfer will require recalibration of the model as well. It is also a general rule in environmental modeling to reserve a statistically sufficient portion of available measurement data for model validation. Caution should also be employed in using a model at a spatial scale or temporal pattern for which it was not designed. A number of textbooks address environmental science and modeling (Clark 1996; Crawford-Brown 2001).
“Geophysical plausibility” is the term we have coined for use in application of environmental science to exposure assessment for epidemiology. In simplest terms this axiom would dictate that an association between a contaminant source and exposure to an organism or ecologic community cannot exist unless there is a plausible geophysical route of transport for the contaminant between the source and the receptor. For example, assume we are conducting a study of drinking water as the sole source of exposure to a specific contaminant and a disease outcome. If a landfill is leaching the contaminant into a groundwater resource (aquifer) in our study area, but our study population has always used another water supply source with no geophysical connectivity to the aquifer, it is implausible that the contaminant from the landfill is causing the adverse health outcome through a drinking water route of exposure. This axiom is particularly relevant in the use of GIS-based processing functions (e.g., kriging on measurement data) to develop exposure estimates in environmental epidemiology studies.
Epidemiology
The fundamental guidelines for the design of an environmental epidemiology study are relevant whether or not GIS technology is being used for exposure assessment. A well-designed epidemiologic study takes into account potential confounding factors, including other exposures that may co-occur with the exposure of interest. The study should be designed to have adequate power to detect an association between the exposure and health outcome and to evaluate exposure–response relationships. For many environmental exposures the anticipated magnitude of the association with disease is likely to be modest, therefore a careful evaluation of the expected prevalence of exposure is critical to determining adequate study power.
A GIS can be used to evaluate the population potentially exposed and to determine if there is likely to be adequate variation in exposure across a study area. Wartenberg et al. (1993) used a GIS to develop an automated method for identifying populations living near high-voltage lines for the purpose of evaluating childhood leukemia and electromagnetic radiation. Another example is the use of a GIS to link disease registry information with public water supply monitoring and location data to determine potential study areas for evaluating the relation between disinfection byproducts exposure and adverse reproductive outcomes and cancer (Raucher et al. 2000).
The epidemiologic study should have the capability to evaluate the exposure in relation to an appropriate latency for the disease and to evaluate critical time windows of exposure. One limitation of a GIS is that mapped data often represent only one snapshot in time. However, several recent efforts have used GIS to reconstruct historical exposure to pesticides (Brody et al. 2002) and drinking water contaminants (Swartz et al. 2003) over a period of decades for a study of breast cancer on Cape Cod, Massachusetts. A study of fetal death in California (Bell et al. 2001) used an exposure metric based on agricultural pesticide use near the mother’s residence during specific time periods during the pregnancy.
Misclassification of exposure is of particular concern in environmental epidemiology studies because of the challenges in estimating exposure to environmental contaminants, which can occur across multiple locations and often at low levels. Exposure errors in time–series studies can occur as a continuum of measurement errors between classic-type errors and Berkson errors, as has been presented in detail by Zeger et al. (2000) regarding air pollution and health. Each type of error has different effects on the estimation of risk. Berkson error occurs when the exposure metric is at the population level, and individual exposures vary because of different activity patterns. An example of a population-level or aggregate exposure metric is the assignment of air pollutant levels from a stationary air monitor to the population living in the vicinity of the monitor. Berkson error does not lead to bias in the risk estimate although the variance of the risk estimate is increased (Zeger et al. 2000).
In a classic error model the exposure metric used in an epidemiologic study is measured with error and is an imperfect surrogate for the true exposure. If misclassification of exposure is nondifferential in terms of the health outcome, the effect is generally to bias risk estimates toward the null, thus potentially missing true associations (Copeland et al. 1977; Flegal et al. 1986). To evaluate the degree of misclassification that may occur in an epidemiologic study, it is important to consider the sensitivity and specificity of the exposure metric employed. Sensitivity is the ability of an exposure metric to correctly classify as exposed those who are truly exposed. Specificity is the ability of the metric to correctly classify as unexposed those who are unexposed. Most epidemiologists do not formally assess the validity of their exposure metric before a study is launched; however, small reductions in sensitivity and/or specificity of the exposure metric can have substantial effects on the estimates of risk. When the true prevalence of exposure is low (e.g., less than 10%) small reductions in specificity cause substantial reductions in the risk estimates, whereas reductions in sensitivity have smaller effects. When the exposure is common in the study population, the sensitivity of the exposure metric becomes more important (Stewart and Correa-Villasenor 1991).
A common metric used in studies employing GIS is the proximity between a pollutant source and a residence. Simple proximity metrics are likely to overestimate the population truly exposed (high sensitivity but low specificity). If those truly exposed represent only a small percent of the study population, there will be substantial attenuation of the risk estimate if a true risk exists. Rull and Ritz (2003) compared several methods of classifying a study population in California on the basis of agricultural pesticide use reported by the California Pesticide Use Reporting (CPUR) database (http://www.cdpr.ca.gov/). The prevalence of exposure differed substantially depending on the metric used. They assumed that a metric that accounted for the location of crop fields more accurately represented true exposures and this metric resulted in lower exposure prevalence compared with a metric based on the CPUR database alone. In a simulation study they demonstrated that the reduced specificity of the CPUR metric resulted in substantial attenuation of risk estimates.
Using GIS to Define the Study Population in an Epidemiologic Study
When epidemiologists select a study population, they are, by default, defining a system boundary for the exposure assessment process. This system boundary is an important element of source-receptor modeling approaches that may be used in the exposure assessment process. Location data for the study population are typically a set of geopolitical units (census enumeration unit boundaries) or the actual residences of the study population. Both of these data types can be represented using functions common to most GIS software. Usually, the subjects are identified from health registries or other records that identify individual cases or disease rates in a geographic area. Examples include cancer registry data, hospital records of a particular disease outcome, or death certificate data. Many of these data are now stored digitally, and an increasing percentage are also georeferenced so that transfer to a GIS database is possible. Controls are identified and located by the epidemiologist, often by frequency-matching characteristics of each case subject that are relevant to disease etiology, including age and sex. Controls are usually selected from the same general geographic region, which should represent the base population from which the cases arise.
Example: Classification of Populations near Landfill Sites (Elliott et al. 2001)
Public concern has been raised that living near a landfill site may be hazardous to health. In particular, several U.S. and U.K. studies have shown excess risk of birth anomalies in populations living near landfill sites (Dolk et al. 1998; Fielder et al. 2000; Vrijheid 2000). To investigate potential risk of adverse birth outcomes associated with landfill sites in Great Britain, investigators had access to an extensive data set of current and previously opened landfill sites provided by the environmental protection agencies in Great Britain. Data were incorporated in a GIS, resulting in a database containing 19,196 landfill sites in England, Wales, and Scotland. Detailed data on boundaries were unavailable for most sites, and therefore point locations had to be used. Site centroids were given for a majority of sites. The location of the site gateway at the time of reporting was used for the remainder. Geocoded data were supplied for landfill site locations but were of low accuracy (often rounded to 1,000 m), and area data were inadequate for most sites. Landfill site areas also changed considerably over time. Postcodes, which were used to define the location of cases and births, only approximated the place of residence. When researchers tried to intersect location of landfill(s) and residences of study subjects, they found that landfill sites are often highly clustered, so that individual postcodes may lie close to as many as 30 or more sites. Given that study subjects may be exposed to several landfill sites, distance from the nearest landfill site was not regarded as a meaningful proxy for exposure. As a compromise between the need for spatial precision and the limited accuracy of the data, a 2-km zone was constructed around each site (Figure 3), giving a resolution similar to or higher than that of previous studies (Dolk et al. 1998; Fielder et al. 2000) and at the likely limit of dispersion for landfill emissions (WHO 2001). The reference population comprised people living more than 2 km from all known landfill sites during the study period. Availability of landfills and health outcome data were restricted to the study period from 1983 to 1998.
Because health data were available only to 1998 and because of concerns about the quality of the early landfill data, 9,631 sites that closed before 1982 or opened after 1997 were excluded (allowing a 1-year lag period for the birth outcomes), as were landfill sites for which there were inadequate data. The remaining 9,565 sites included 774 sites for special (hazardous) waste, 7,803 for nonspecial waste, and 988 handling unknown types.
The study was the largest performed on possible associations between residence near landfill sites and adverse birth outcomes. A GIS-based approach was necessary because of the large number of landfill sites included in the study; individual investigations of several thousand landfill sites would have been practically difficult and prohibitively expensive. The most striking finding was that approximately 80% of the British population live within 2 km of a landfill site. This also imposed unique challenges for the epidemiologic study design, given that 80% of the study population was potentially exposed and only 20% could be used as a reference. In most environmental epidemiology studies, the situation is the opposite in that the prevalence of those potentially exposed is much lower. This high prevalence of potential exposure had implications for the statistical analysis, as the usual reference rates after stratification by known confounders would not be estimated with the negligible error normally associated with such studies. Despite this, the reference area included over 2 million births over the study period. To guard against overinterpretation, 99% (rather than the more commonly used 95%) confidence intervals around the relative risk (RR) estimates were computed.
The authors were aware of the relative inaccuracy of postcodes (used to define the location of cases and births), as these give only an approximation of place of residence, accurate to 10–100 m in urban areas but > 1 km in some rural areas. Furthermore, it is well known that postcodes are afflicted with several other problems: they may change over time, be terminated, or even recycled. However, such problems affect only a small minority (approximately 1%) of U.K. post-codes. Thus, given the size of the study (national rather than local), this is not a major problem. For further details the reader is referred to the original article (Elliott et al. 2001).
Using GIS to Identify Source and Potential Routes of Exposure in an Epidemiologic Study
The exposure or agent of interest in an environmental epidemiology study may be a chemical (a single compound or, rarely, a mixture) or physical agents (particulates, radiation, noise). Once the agent is identified, a GIS can be instrumental in identifying sources and potential routes of exposure. Source identification is a function of the occurrence of the target agent in a specified environmental medium (air, water, food, dust, etc.). Identifying the sources enables assessment of the likelihood of exposure across the study population and provides data on the route of exposure information necessary for calculating personal exposure.
Example: Neurobehavioral Effects of Exposure to Trichloroethylene through a Municipal Water Supply (Reif et al. 2003)
The basis for this study was initially a cross-sectional study of exposure to a number of chemicals with documented release in a community adjacent to a Superfund waste site, the Rocky Mountain Arsenal (RMA) near Denver, Colorado, USA. Study participants were randomly selected from an area within 1.61 km (1 mile) that abutted to the north, northwest, and west boundaries of the site, where fugitive chemicals had been detected in ground and surface waters, sediments, and soils (Figure 4). A total of 585 persons who had lived at their current residence for at least 2 years were eligible for the study; 472 participated. Results of the initial study warranted a second study, conducted in 1991, during which the researchers interviewed and conducted neurobehavioral testing of 204 adults originally identified by the first study (ATSDR 1996). Results of the 1991 study showed a trend toward an increased prevalence of neurologic disorders and adverse reproductive outcomes, particularly in the area north/northwest of the RMA, compared with communities at a greater distance from RMA, presumed to be unexposed to the site. However, the researchers again relied on proximity to the RMA as a surrogate for exposure, and there was evidence that this may have resulted in nondifferential misclassification of exposure, which tends to drive the effect estimate or relative risk toward the null value (Copeland et al. 1977). The researchers initiated a revised exposure assessment using a GIS-based analysis of fate and transport of chemicals in the groundwater regimen hydraulically downgradient from the RMA site. The researchers selected trichloroethylene (TCE) as the marker contaminant for the exposure assessment because of its neurotoxi-cologic properties, and because it had been detected in water supply wells in the study area. The researchers constructed an operable MODFLOW (U.S. Geological Survey, Reston, Virginia, USA) simulation model that accurately reflected hydraulic characteristics of groundwater regime in the study area and used a GIS to develop input variables to the model, including source location of TCE on the RMA site. However, the researchers could not validate TCE levels measured in water wells used by the local water district (LWD), where 90% of the study population resided. The researchers expanded the geographic extent of their study area, and determined that the source of TCE in the groundwater was from multiple hazardous waste sites, including some located outside the original study area. Once the primary source was properly identified, the researchers confirmed the measurement results of TCE in the LWD supply wells by the groundwater model. TCE levels in the wells were then used as input to a hydraulic and water quality simulation model, EPANET (Rossman 1994), to predict TCE levels in the distribution system of the LWD. The researchers used GIS to geocode the study population, develop input data for the simulation model, and assign individual exposure to TCE by linking results of the model to the census block group of residence (Figure 4). The study with the refined exposure assessment found a stronger association of risk for neurobehavioral disorders in the study population than was found in the 1991 study, in which exposure was based primarily on proximity to a source of chemical contamination, including TCE. The study demonstrates that GIS-based technology can be used to refine exposure for epidemiologic investigations, improving sensitivity and specificity beyond a simple proximity metric. It also demonstrates the effect that selection of operational scale can have on exposure assessment in an epidemiology study. The operation of the water distribution system could not be discerned when proximate census blocks were used as a surrogate for exposure.
Using GIS to Estimate Environmental Levels of Target Contaminants in an Epidemiologic Study
Exposure is a function of the concentration of target contaminant in the environment of the study population. The optimal method for quantifying levels of the target agent is the measurement of the environmental media associated with each potential route of exposure during the critical time period for exposure. However, rarely is there an opportunity to make such measurements. Alternatively, predicted environmental levels of the target agent can be estimated using source-receptor modeling. Computer-based models designed to predict levels of contaminants due to point sources (smokestack) or nonpoint sources (drift from aerial spraying of pesticides) are available. Often these predictions are used as a surrogate for exposure in the epidemiologic analysis. In either case, validation of the estimates is important to understand the results of the epidemiologic study. Validation is often overlooked in the exposure assessment process. It is also important that the environment depicted in the modeling period be representative of the environment during the exposure period necessary for the epidemiologic study. Generally, the degree to which validation can be accomplished is a function of measurement data available for the time period of interest. Most source-receptor models require some measurements for constructing (calibrating) the predictive algorithms.
Example: The Lung Cancer in Stockholm Study (Bellander et al. 2001; Nyberg et al. 2000)
A population-based case–control study, the Lung Cancer in Stockholm Study (LUCAS), was designed to investigate whether urban air pollution increases lung cancer risk. Previous studies had commonly used crude surrogates for individual exposure, limiting the power of detecting any risk associated with air pollution. The LUCAS study used advanced modeling techniques to assess individual exposure for relevant time periods (several decades before diagnosis). Detailed emission data, dispersion models, and GIS were used to assess historical exposure to several components of ambient air pollution. The study base consisted of all men 40–75 years of age who lived in Stockholm County at any time between 1985 and 1990 and who had lived in the county since 1950, with a maximum of 5 years of residence outside the county. A total of 1,042 lung cancer cases diagnosed between 1985 and 1990 were included, as well as 2,364 controls. Information on residence from 1955 to the end of follow-up for each individual, 1990–1995, was collected using a questionnaire. Nitrogen oxides (NOx and NO2) and sulfur dioxide (SO2) were chosen as indicators of air pollution from road traffic and residential heating, respectively.
Ambient air concentrations of NOx , NO2, and SO2 were assessed throughout the study area for three points in time (1960, 1970, and 1980) using reconstructed emission data for these index pollutants together with dispersion modeling (Figure 5). The modeled NO2 estimates for 1980 were validated with available measurement data. Linear intra- and extrapolation were used to obtain annual estimates for the remainder of the exposure period (1955–1990). Individual addresses were geocoded with an estimated error of < 100 m for 90% of the addresses. Annual air pollution estimates were then linked to residence coordinates, yielding cumulative residential exposure indices for each individual. There was a wide range of individual long-term average exposure, with an 11-fold interindividual difference in NO2 and an 18-fold difference in SO2
The detailed individual exposure assessment made it possible to assess relative risk potentially associated with road traffic. Average traffic-related NO2 exposure over 30 years was associated with a relative risk of 1.4 and a 95% confidence interval 1.0, 2.0 for the top decile of exposure, adjusted for tobacco smoking, socioeconomic status (SES), residential radon, and occupational exposures, and taking into consideration a latency period of 20 years (Nyberg et al. 2000). The signifi-cance of these results was recognized in an accompanying editorial as being the first study that had used this advanced exposure assessment, making the detailed analysis possible (Rothman and Cann 2000).
The results indicate that GIS can be useful for exposure assessment in environmental epidemiology studies, provided that detailed geographically related exposure data are available for relevant time periods.
Using GIS to Estimate Personal Exposure in an Epidemiologic Study
A key issue in exposure assessment is how well an exposure metric estimates exposure to the individual. Exposure has been defined as “the contact of a chemical, physical, or biological agent with the outer boundary of an organism” (Berglund et al. 2002). Exposure is a function of concentration and time: “An event that occurs when there is contact at a boundary between a human and the environment with a contaminant of a specific concentration for an interval of time” (NRC 1991). Thus, in the context of exposure assessment for an epidemiologic study, it is important to distinguish between environmental concentration, exposure concentration, and dose. The environmental concentration of an agent refers to its presence in a particular carrier medium [for example, polycyclic aromatic hydrocarbons (PAH) in ambient air], expressed in quantitative terms (for example, micrograms per cubic meter). Similarly, the exposure concentration of an agent refers to its presence in its carrier medium at the point of contact (for example, PAH in breathing zone air) expressed in quantitative terms (for example, micrograms per cubic meter). Finally, the dose refers to the amount of a pollutant that actually enters the human body, i.e., is taken up through absorption barriers. A number of variables can influence the exposure and dose. These include physiologic factors such as age, sex, physical condition, disease, and genetics, as well as exposure factors related to human behavior and activities (e.g., the amount of time spent commuting to work each day), and contact rates (e.g., the amount of drinking water ingested per day). In epidemiologic studies, environmental concentration will often be used as a surrogate for both exposure concentration and dose.
We could not find an example of the use of GIS to estimate personal exposure for an epidemiologic study. In our review of the literature, questionnaire data were generally used as a surrogate for deriving personal exposure. Only recently have researchers started using GIS to study activity patterns in a study population, which conceivably could be linked to environmental data for exposure assessment. Phillips et al. (2001) reported on a test of GPS data recorders as a means of validating time-location data recorded in study diaries of a subset of participants enrolled in the Oklahoma Urban Air Toxics Study. Elgethun et al. (2003) describe the development and testing of a data-logging GPS unit designed to be integrating into clothing. Both studies concluded that GPS units could be useful in developing time–location information for use in exposure assessment. GPS is a satellite-based technology composed of a system of satellites encircling earth and emitting a radio frequency detectable by GPS receivers. GPS receivers are designed to use this information and calculate coordinates of the receiver location. Precision of these coordinates can vary based on receiver design and signal quality. Phillips et al. (2001) reported precision of about 10 m for most readings, whereas Elgethun et al. (2003) reported mean root mean square error of 3.2 m outdoors and 5.8 m indoor in positional accuracy for two GPS units tested. This level of precision should be sufficient for most studies attempting to link location of a participant with a particular environmental setting where contaminant monitoring or modeling data are available for linkage using a GIS. A major advantage of the technology, as reported by Phillips et al. (2001), was not only that its use confirmed all reported trips over a 12- to 23-hr monitoring period, but it provided time–location data on travel events not recorded in the participant diary.
Both Phillips et al. (2001) and Elgethun et al. (2003) reported limitations of the technology as a sole source of space–time data for an exposure assessment study. Both studies found the reception of the satellite signals to be adversely impacted by shielding from buildings of certain materials (concrete, steel), electrical power stations, and to some extent vehicle body panels. Signal blockage continues to be an issue with GPS today. Phillips et al. (2001) also reported extensive failure including battery failure, data-logging failure, and data storage limitations, which resulted in capturing only about 30% of the total monitoring time attempted in 25 trials. Elgethun et al. (2003) reported reception efficiencies of 79% outdoors, 20% in homes, 12% in vehicles, and 6–9% in schools and businesses. These findings indicate that although GIS using GPS technology hold promise in terms of integrating study population activity data with measured or predicted levels of environmental contaminants in the exposure assessment process, their use is still very much in the developmental research stage for use in epidemiology studies.
Discussion
Our findings indicate that GIS can greatly enhance epidemiologic research in terms of definition of source and routes of potential exposure and estimation of environmental levels of target contaminants in the exposure assessment process. We found over 15 studies published since 1998 that describe the successful use of GIS for one or more of these purposes. Across all of these studies, there was consensus that the use of GIS was instrumental in achieving optimal exposure assessment. In our example studies, GIS improved resolution of the source of potential exposure (Elliott et al. 2001; Reif et al. 2003), identified the most likely route of exposure (Reif et al. 2003), and estimated levels of target contaminants for use in estimating exposure to the study population (Nyberg et al. 2000; Reif et al. 2003). Our examples of environmental epidemiology studies using GIS also emphasize the importance of interdisciplinary study teams.
GIS have been used to evaluate environmental justice issues, usually by linking information about potential sources of environmental pollutants to census information on sociodemographic characteristics of a population (Perlin et al. 2001; Waller et al. 1999). However, only recently have GIS been used in the design of environmental epidemiology studies. Each example in our article demonstrates that GIS can (and perhaps should) be used in the early planning stages of an environmental epidemiology study to help locate a potential study population with a wide range of exposure. The statistical power of an epidemiologic study and the precision of the risk estimates are optimized when the study population includes adequate numbers of those with both high and low exposures. An example of how GIS have been used to identify a study population with a range of exposures is a feasibility study of childhood leukemia and electromagnetic radiation from power transmission lines in New Jersey (Wartenberg et al. 1993). A GIS was used to identify the population living close to transmission lines and a comparison population farther away. Demographic information was evaluated for both the exposed and unexposed populations to determine potential confounding factors. Other examples include the use of GIS for surveillance and study of lead poisoning from residential exposures (Roberts et al. 2003; Wartenberg 1992).
The increasing availability of environmental databases in a geographic format (Paulu et al. 1995), including the location of industrial sites and releases (Toxic Release Inventory Program 2004), should make it feasible to incorporate these potential exposure data into epidemiologic studies. For example, in a recently started cross-sectional study on potential adverse health effects (primarily hypertension) of airport-related noise exposure, study populations are being selected using modeled noise contours around the participating airports (European Commission 2003). Such models are particularly applicabile in the selection of study populations exposed to different levels of the pollutants under study, using a cross-sectional or cohort study approach. A case–control design, in which cases are selected from, for example, hospital data or cancer registries, will usually have a predefined area (hospital catchment or cancer registry area); thus, preexisting exposure information may be less relevant in the study population selection. However, exposure information can be used to delimit the study area within the bounds of the catchment area or disease registry. For example, AWWA (2000) demonstrated the feasibility of linking environmental monitoring data with birth and cancer registry data to identify optimal geographic locations for epidemiologic studies of by-products of chlorination in public water supplies in the United States. GIS also have potential uses in the selection of controls for an epidemiologic study, as they are usually randomly selected from the same geographic area as the cases. As frequency matching (on age and sex) is commonly applied for study efficiency reasons, GIS could also be used for further frequency matching on SES, where areas are classified according to a georeferenced SES index.
There are, of course, a number of caveats regarding use of GIS for exposure assessment in environmental epidemiologic studies. We reviewed fundamental principles of three scientific disciplines critical to such applications: geospatial science, environmental science, and epidemiology. Axiomatic themes from each of these scientific disciplines should be adhered to in any case, but they are particularly relevant when using a GIS. These themes include accuracy and validity of data (raw and calculated), appropriate selection of mathematic formulas and models, and scientific plausibility. The application of these axiomatic themes can be very different across the scientific disciplines, which reinforces the need for multidisciplinary teams in conducting environmental epidemiology studies. For example, researchers in each of the disciplines are trained in determining the accuracy and precision of measurement data. However, only the geospatial scientist or geographer is generally trained to rectify geographic data so that two or more GIS-based data layers such as health outcome and environmental data can be merged and the resulting data layer used to determine the association more accurately. Similarly, only the epidemiologist is likely to be trained to search for and identify other data layers that, if omitted from the test of association, could confound the results.
Use of measured environmental data and mathematic algorithms for estimating contaminant levels in exposure assessment is another area requiring specialized expertise in most cases. Since the advent of the computer age, packaged software has become more and more prevalent for such applications, but the old modeler adage “garbage in, garbage out” is perpetual truth. Even with the color maps produced using a GIS, “mapped garbage” is still “garbage.” In this article we propose several fundamental principles of environmental science and modeling that should be adhered to when using GIS in exposure assessment for epidemiology studies. Perhaps the most important of these principals can be captured by the term “validation.” In each of our example studies, environmental data were used to develop an exposure metric for use in epidemiology. The data used were collected for other purposes, commonly for administrative or regulatory use. These studies demonstrate the range of measurement data quality and degree of validation that may be possible from relatively low (Elliott et al. 2001) to high (Nyberg et al. 2000). They also demonstrate the likely consequences across this range in terms of risk estimates in an epidemiology study. In Elliott et al. (2001), a database on landfill sites was obtained from the environmental protection agencies, which collected the data from site operators in the licensing process. Thus, data that would have been useful for exposure assessment were not readily available (e.g., volumes and types of waste actually received at the landfill sites, measurement data for specific chemicals being released into the environment, or the extent of contamination). Instead, the likely limit of dispersion for landfill emissions (2 km) was estimated based on published information and used as an exposure boundary around each site, degree of hazard for exposure was derived from the type of license held by the operator, and the epidemiologic analysis assumed a common relative risk for all landfill sites. The researchers did not validate these exposure metrics. It is likely that sites licensed to carry special (hazardous) waste did not necessarily do so, and that sites licensed to carry nonspecial waste actually did carry some hazardous waste as well. The resulting exposure misclassification was most likely nondifferential, which could result in a bias risk estimate toward the null (Copeland et al. 1977). The findings of the study, small excess risks for some birth outcomes after exposure to landfills, seem to verify this conclusion.
The study reported by Reif et al. (2003) concerning TCE and neurobehavioral demonstrated that improvement in exposure assessment techniques “refined exposure . . . with adequate specificity to reveal adverse effects [of TCE] in the nervous system.” In that study, the researchers refined exposure assessment by replacing a proximity metric such as the one used in Elliott et al. (2001) with exposure predictions based on validated environmental measurements (TCE levels in groundwater at source wells for a municipal water system) and validated transport modeling (water pressure and volume in the municipal water system) during the exposure period for the study. However, data were not available to validate predicted TCE levels at study participants’ residences.
In the final example study that we reviewed, Bellander et al. (2001) had sufficient source emission and environmental measurement data to calibrate and validate predicted levels of NO2 in the environment of Stockholm, Sweden, for at least a portion of the exposure period in an epidemiologic study of lung cancer (1955–1990). They also validated their predicted location of residence in Stockholm for each participant in the study by cross-checking results using external geocoding service companies. The resolution and precision of this exposure assessment process resulted in the capability to detect a wide range of individual long-term average exposure and to detect risk of lung cancer to average traffic level exposure to NO2 within a 95% confidence limit. The procedures and results of these studies clearly indicate the need for expertise in environmental science and related disciplines in epidemiologic studies involving pollutant emissions.
Conclusion
In summary, we have reviewed the recent literature on the use of GIS in exposure assessment for environmental epidemiology and described principles and applications of three core scientific disciplines needed, in our opinion, to successfully implement such studies: geospatial science, environmental science, and epidemiology. This by no means preempts the need for other scientific disciplines in the execution of such studies. In particular, statistics is a core science that would benefit every study, and other disciplines should be included based on the focus and objective of the study. Based on our findings, we offer the following conclusions:
The use of GIS in exposure assessment for environmental epidemiology studies is not only feasible but can enhance the understanding of the association between contaminants in our environment and disease.
A good environmental epidemiology study design should aim to maximize exposure contrasts and thus study population selection should be based on an a priori conception of the geographic distribution of exposures in the study area whenever possible (even if crude). For this purpose, GIS-based exposure mapping can be useful, given that geo-referenced data are available at a relevant scale.
It is preferable in an environmental epidemiology study to estimate and validate levels of the agent (contaminant) of interest in the environment of the study population. These levels are the basis for estimating personal exposure and dose and for classifying exposure across a study population. GIS and related technology (source/receptor model; environmental simulation models) can improve accuracy in identifying source and route of potential exposure in a study area and in estimating levels of target contaminants.
When environmental levels of the agent (contaminant) of interest in the environment of the study population cannot be measured or accurately predicted, GIS provide the optimal technology for using proximity to contaminant source in an environmental epidemiology study. It is well established as a viable tool in ecologic study design.
GIS and related technologies such as the GPS are useful for providing precise locations of study participant residences and other stationary data. Research is needed on how to integrate this use of the technology with epidemiologic questionnaire and environmental data for exposure assessment.
Environmental epidemiology studies require interdisciplinary expertise and adherence to the fundamental principles of geospatial science, environmental science, and epidemiology.
This article is part of the mini-monograph “Health and Environment Information Systems for Exposure and Disease Mapping, and Risk Assessment.”
Figure 1 Structure and functionality of a GIS.
Figure 2 Exposure assessment process. Steps for which use of GIS is discussed in this article are highlighted in blue.
Figure 3 Distribution of landfill sites in the Great Britain, buffered to 2 km, with an inset showing details of the buffer zones in pink (SAHSU 2001). The high density of sites in many areas results in considerable overlap of the buffer zones used to define exposures, and thus means that many areas are classified as exposed from a number of different landfill sites (see inset).
Figure 4 Exposure zone in original RMA study (ATSDR 1996) and refined resolution of predicted exposure to TCE by census block as reported by Reif et al. (2003).
Figure 5 Modeled ambient air concentrations of NO2 emissions from all sources (1980 data) using reconstructed emission data for this index pollutant together with dispersion modeling (Bellander et al. 2001).
==== Refs
References
Ali M Emch M Donnay JP 2002 Spatial filtering using a raster geographic information system: methods for scaling health and environmental data Health Place 8 85 92 11943581
ATSDR. 1996. Reproductive, Neurobehavioral, and other Disorders in Communities Surrounding the Rocky Mountain Arsenal. A Report by the Agency for Toxic Substances and Disease Registry and Colorado State University. Publ no PB96-178058. Springfield, VA:National Technical Information Service.
AWWA 2000. Identifying Geographic Regions for Potential Future Epidemiology Research on Disinfection Byproducts. Rep no 341. American Water Works Research Foundation. Denver, CO:American Water Works Association.
Bell EM Hertz-Picciotto I Beaumont J 2001 A case-control study of pesticides and fetal death due to congenital anomalies Epidemiology 12 148 156 11246574
Bellander T Jonson T Gustavsson P Pershagen G Järup L 2001 Using geographic information systems to assess individual historical exposure to air pollution from traffic and house heating in Stockholm Environ Health Perspect 109 633 639 11445519
Berglund M Elinder CG Järup L 2002. Human Exposure Assessment: An Introduction. World Health Organization. Available: http://www.imm.ki.se/PDF/HEAboken.pdf [accessed 12 December 2003]
Beyea J Hatch M 1999 Geographic exposure modeling: a valuable extension of geographic information systems for use in environmental epidemiology Environ Health Perspect 107 suppl 1 181 190 10229717
Bonner MR Han D Nie J Rogerson P Vena JE Freudenheim JL 2003 Positional accuracy of geocoded addresses in epidemiologic research Epidemiology 14 404 412
Briggs DJ Elliot P 1995 The use of geographical information systems in studies on environment and health World Health Stat Q 48 85 94 8585238
Brody JG Vorhees DJ Melly SJ Swedis SR Drivas PJ Rudel RA 2002 Using GIS and historical records to reconstruct residential exposure to large-scale pesticide application J Expo Anal Environ Epidemiol 12 64 80 11859434
Chrisman NR 2002. Exploring Geographic Information Systems. 2nd ed. New York:Wiley Publishing.
Christakos G Serre ML 2000 Spatiotemporal analysis of environmental exposure-health effect associations J Expo Anal Environ Epidemiol 10 168 187 10791598
Cicero-Fernandez P Torres V Rosales A Cesar H Dorland K Munoz I 2001 Evaluation of human exposure to ambient PM10 in the metropolitan area of Mexico City using a GIS-based methodology J Air Waste Manag Assoc 51 1586 1593 11720106
Clark MM 1996. Transport Modeling for Environmental Engineers and Scientists. New York:Wiley Publishing.
Comba P Ascoli V Belli S Benedetti M Gatti L Ricci P 2003 Risk of soft tissue sarcomas and residence in the neighborhood of an incinerator Occup Environ Med 60 680 683 12937191
Copeland KT Checkoway H Holbrook RH McMichael AJ 1977 Bias due to misclassification in the estimate of relative risk Am J Epidemiol 105 488 495 871121
Crawford-Brown DJ 2001. Mathematical Methods of Environmental Risk Modeling. Boston:Kluwer Publishing.
DeMers MN 2000. Fundamentals of Geographic Information Systems. 2nd ed. New York:Wiley Publishing.
Dolk H Vrijheid M Armstrong B Abramsky L Bianchi F Garne E 1998 Risk of congenital anomalies near hazardous-waste landfill sites in Europe: the EUROHAZCON study Lancet 352 423 427 9708749
Elgethun K Fenske RA Yost MG Palcisko GJ 2003 Time-location analysis for exposure assessment studies of children using a novel global positioning system instrument Environ Health Perspect 111 115 122 12515689
Elliott P Briggs D Morris S de Hoogh C Hurt C Kold Jensen T 2001 Risk of adverse birth outcomes in populations living near landfill sites Br Med J 323 363 368 11509424
European Commission 2003. Hypertension and Exposure to Noise near Airports Project. Brussels:The European Commission. Available: http://www.hyena.eu.com [accessed 2 March 2004].
Falbo DL Queen LP Blinn CR 1991. Introduction to Data Analysis Using Geographic Information Systems. Report No. FO-05740. University of Minnesota Extension Service. St Paul, MN. Available: http://www.extension.umn.edu/distribution/naturalresources/DD5740.html#i [accessed 2 March 2004].
Fielder HM Poon-King CM Palmer SR Moss N Coleman G 2000 Assessment of impact on health of residents living near the Nanty-Gwyddon landfill site: retrospective analysis Br Med J 320 19 22 10617518
Flegal KM Brownie C Haas JD 1986 The effects of exposure misclassification on estimates of relative risks Am J Epidemiol 123 736 751 3953551
Floret N Mauny F Challier B Arveux P Cahn J-Y Viel J-F 2003 Dioxin emissions from a solid waste incinerator and risk of non-Hodgkin’s lymphoma Epidemiology 14 392 398 12843761
Fotheringham AS Wong DWS 1991 The modifiable area unit problem in multivariate analysis Environ Plan A 23 1025 1044
Gallagher MD Nuckols JR Stallones L Savitz DA 1998 Exposure to trihalomethanes and adverse pregnancy outcomes in Colorado J Epidemiol 9 484 489
Gunier RB Harnly ME Reynolds P Hertz A Von Behren J 2001 Agricultural pesticide use in California: pesticide prioritization, use densities, and population distributions for a childhood cancer study Environ Health Perspect 109 1071 1078 11689348
Holt D Steel DG Tranmer M Wrigley N 1996 Aggregation and ecological effects in geographically based data Geogr Anal 28 244 261
Inserra S Phifer B Pierson R Campagna D 2002 Community-based exposure estimate for hydrogen sulfide J Expo Anal Environ Epidemiol 12 124 129 11965529
Kohli S Sahlen K Lofman O Sivertun A Foldevi M Trell E 1997 Individuals living in areas with high background radon: a GIS method to identify populations at risk Comput Methods Programs Biomed 53 105 112 9186047
Krieger N Waterman P Lemieux K Zierler S Hogan JW 2001 On the wrong side of the tracts? Evaluating the accuracy of geocoding in public health research Am J Public Health 91 1114 1116 11441740
Langholz B Ebi KL Duncan CT Peters JM London SJ 2002 Traffic density and the risk of childhood leukemia in a Los Angeles case-control study Ann Epidemiol 12 482 487 12377426
National Research Council 1991. Human Exposure Assessment for Airborne Pollutants. Advances and Opportunities. Washington, DC:National Academy Press.
Nyberg F Gustavsson P Järup L Bellander T Berglind N Jakobsson R 2000 Urban air pollution and lung cancer in Stockholm Epidemiology 11 587 595
Openshaw S Taylor P 1979. A million or so correlation coefficients: three experiments on the modifiable area unit problem. In: Statistical Applications in the Spatial Sciences (Wrigley N, ed). London:Pion, 127–144.
Paulu CA Ozonoff DM Coogan P Wartenberg D 1995 Making environmental data accessible for public health aims. The Massachusetts Environmental Database project Public Health Rep 110 776 783 8570834
Perlin SA Wong D Sexton K 2001 Residential proximity to industrial sources of air pollution: interrelationships among race, poverty, and age J Air Waste Manag Assoc 51 406 421 11266104
Phillips ML Hall TA Nurtan AE Lynch R Johnson DL 2001 Use of global positioning system technology to track subject’s location during environmental exposure sampling J Expo Anal Environ Epidemio1 1 207 215
Raucher RS Nuckols JR Krapfl H Weigel SW 2000. Identifying Geographic Regions for Potential Future Epidemiology Research on Disinfection Byproducts. Report no 341. Denver, CO:American Water Works Association.
Reif JS Burch JB Nuckols JR Metzger L Anger WK 2003 Neurobehavioral effects of exposure to trichloroethylene through a municipal water supply Environ Res 9 248 258 14615234
Reynolds P Von Behren J Gunier RB Goldberg DE Hertz A Harnly ME 2002a Childhood cancer and agricultural pesticide use: an ecologic study in California Environ Health Perspect 110 319 324 11882484
Reynolds P Von Behren J Gunier RB Goldberg DE Hertz A Smith D 2002b Traffic patterns and childhood cancer incidence rates in California, United States Cancer Causes Control 13 665 673 12296514
Reynolds P Von Behren J Gunier RB Goldberg DE Hertz A Smith D 2003 Childhood cancer incidence rates and hazardous air pollutants in California: an exploratory analysis Environ Health Perspect 111 663 668 12676632
Roberts JR Hulsey TC Curtis GB Reigart JR 2003 Using geographic information systems to assess risk for elevated blood lead levels in children Public Health Rep 118 221 229 12766217
Rogers JF Thompson SJ Addy CL McKeown RE Cowen DJ Decoufle P 2000 Association of very low birth weight with exposures to environmental sulfur dioxide and total suspended solids Am J Epidemiol 151 602 613 10733042
Rossman LA 1994. EPANET Users Manual, Version 1.1. Cincinnati, OH:U.S. Environmental Protection Agency.
Rothman KJ Cann CI 2000 The smoking gun? Epidemiology 11 485 486 10955398
Rull RP Ritz B 2003 Historical pesticide exposure in California using pesticide use reports and land-use surveys: an assessment of misclassification error and bias Environ Health Perspect 111 1582 1859 14527836
SAHSU 2001. A Geographical Analysis of Populations Living around Landfill Sites. Technical Report. London:Small Area Health Statistics Unit, Imperial College.
Stewart W Correa-Villasenor A 1991 False positive exposure errors and low exposure prevalence in community-based case-control studies Appl Occup Environ Hyg 6 534 540
Swartz CH Rudel RA Kachajian JR Brody JG 2003 Historical reconstruction of wastewater and land use impacts to groundwater used for public drinking water: exposure assessment using chemical data and GIS J Expo Anal Environ Epidemiol 13 403 416 12973368
Toxic Release Inventory (TRI) Program 2004. Washington, DC:U.S. Environmental Protection Agency. Available: http://www.epa.gov/tri [accessed 2 March 2004].
Vine MF Degnan D Hanchette C 1997 Geographic information systems: their use in environmental epidemiologic research Environ Health Perspect 105 6 598 605 9288494
Vrijheid M 2000 Health effects of residence near hazardous waste landfill sites: a review of epidemiologic literature Environ Health Perspect 108 101 112 10698726
Waller LA Louis TA Carlin BP 1999 Environmental justice and statistical summaries of differences in exposure distributions J Expo Anal Environ Epidemiol 9 56 65 10189627
Ward MH Nuckols JR Weigel SJ Maxwell SK Cantor KP Miller RS 2000 Identifying populations potentially exposed to agricultural pesticides using remote sensing and a geographic information system Environ Health Perspect 108 5 12 10622770
Wartenberg D 1992 Screening for lead exposure using a geographic information system Environ Res 59 310 317 1464284
Wartenberg D Greenberg M Lathrop R 1993 Identification and characterization of populations living near high-voltage transmission lines: a pilot study Environ Health Perspect 101 626 632 8143596
WHO 2001. Methods of Assessing Risk to Health from Exposure to Hazards Released from Waste Landfills. Report from WHO meeting 10–12 April 2000, Lodz, Poland. Bilthoven, the Netherlands:World Health Organization Regional Office for Europe, European Centre for Environment and Health.
Xiang H Nuckols JR Stallones L 2000 A geographic information assessment of birth weight and crop production patterns around mother’s residence Environ Res 82 160 167 10662530
Zeger SL Thomas D Dominici F Samet JM Schwartz J Dockery D 2000 Exposure measurement error in time-series studies of air pollution: concepts and consequences Environ Health Perspect 108 419 426 10811568
|
15198921
|
PMC1247194
|
CC0
|
2021-01-04 23:36:23
|
no
|
Environ Health Perspect. 2004 Jun 15; 112(9):1007-1015
|
utf-8
|
Environ Health Perspect
| 2,004 |
10.1289/ehp.6738
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0112-00101610.1289/ehp.674015198922Mini-Monograph: Information SystemsInterpreting Posterior Relative Risk Estimates in Disease-Mapping Studies Richardson Sylvia Thomson Andrew Best Nicky Elliott Paul Small Area Health Statistics Unit, Department of Epidemiology and Public Health, Imperial College Faculty of Medicine, Imperial College London, Norfolk Place, London, United KingdomAddress correspondence to S. Richardson, Department of Epidemiology and Public Health, Imperial College Faculty of Medicine, Imperial College London, Norfolk Place, London, W2 1PG, United Kingdom. Telephone: 44 0 207 594 3336. Fax: 44 0 207 402 2150. E-mail:
[email protected] thank P. Green for stimulating discussions and for providing the computer code of the MIX model.
The U.K. Small Area Health Statistics Unit is funded by the Department of Health, Department of the Environment, Food and Rural Affairs, Environment Agency, Health and Safety Executive, Scottish Executive, National Assembly for Wales, and the Northern Ireland Assembly.
The authors declare they have no competing financial interests.
6 2004 15 4 2004 112 9 1016 1025 12 9 2003 2 3 2004 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. There is currently much interest in conducting spatial analyses of health outcomes at the small-area scale. This requires sophisticated statistical techniques, usually involving Bayesian models, to smooth the underlying risk estimates because the data are typically sparse. However, questions have been raised about the performance of these models for recovering the “true” risk surface, about the influence of the prior structure specified, and about the amount of smoothing of the risks that is actually performed. We describe a comprehensive simulation study designed to address these questions. Our results show that Bayesian disease-mapping models are essentially conservative, with high specificity even in situations with very sparse data but low sensitivity if the raised-risk areas have only a moderate (< 2-fold) excess or are not based on substantial expected counts (> 50 per area). Semiparametric spatial mixture models typically produce less smoothing than their conditional autoregressive counterpart when there is sufficient information in the data (moderate-size expected count and/or high true excess risk). Sensitivity may be improved by exploiting the whole posterior distribution to try to detect true raised-risk areas rather than just reporting and mapping the mean posterior relative risk. For the widely used conditional autoregressive model, we show that a decision rule based on computing the probability that the relative risk is above 1 with a cutoff between 70 and 80% gives a specific rule with reasonable sensitivity for a range of scenarios having moderate expected counts (~ 20) and excess risks (~1.5- to 2-fold). Larger (3-fold) excess risks are detected almost certainly using this rule, even when based on small expected counts, although the mean of the posterior distribution is typically smoothed to about half the true value.
Bayesian hierarchical modelscancer mappingenvironmental epidemiologysensitivitysmall-area studiesspatial smoothingspecificity
==== Body
Spatial analyses of health outcomes have long been recognized in the epidemiologic literature as playing a specific and important role in description and analysis. In particular, they can highlight sources of heterogeneity underlying spatial patterns in the health outcomes and consequently are able to suggest important public health determinants or etiologic clues. A good example of geographic epidemiology is the seminal monograph by Doll (1980), which described some of the first hypotheses concerning the influence of environment and lifestyle characteristics on cancer mortality and discussed how these arose from studying the geographic distribution of various cancers. These early studies were usually performed on a large geographic scale, using mostly international or regional comparisons.
Recently, the availability of local geographically indexed health and population data, together with advances in computing and geographic information systems, has encouraged the analysis of health data on a small geographic scale (Elliott et al. 2000). The motivation is the increased interpretability of small-scale studies, as they are in principle less susceptible to the component of ecologic bias created by the within-area heterogeneity of exposure or other determinants. They are also better able to detect highly localized effects such as those related to industrial pollution in the vicinity. Conversely, small-scale studies require more sophisticated statistical analysis techniques than, for example, an analysis between countries, because the data are typically sparse with low (even zero) counts of events in many of the small areas. Further, frequently there is evidence of overdispersion of the counts with respect to the Poisson model as well as spatial patterns indicating some dependence between the counts in neighboring areas.
Faced with these nonstandard characteristics, statistical models have been developed to address these issues and make best use of small-area health data. In connection with generic developments in a flexible modeling strategy using the paradigm of Bayesian hierarchical models, hierarchical disease-mapping models based on conditional autoregressions (CAR) were proposed in the 1990s through the work of Besag et al. (1991), Clayton and Bernardinelli (1992), and Clayton et al. (1993). These CAR models are now commonly used both by statisticians and epidemiologists, and their implementation is facilitated by existing software such as WinBUGS (Spiegelhalter et al. 2002). Alternative semiparametric formulations to CAR have also been proposed recently (Denison and Holmes 2001; Green and Richardson 2002; Knorr-Held and Rasser 2000) to model more heterogeneous risk surfaces and particularly to allow for potential discontinuities in the risk. The main characteristic of all these models is to provide some shrinkage and spatial smoothing of the raw relative risk estimates that otherwise would be computed separately in each area. Such shrinkage gives a more stable estimate of the pattern of underlying risk of disease than that provided by the raw estimates. The pattern of the raw risks, strongly influenced by the size of the population at risk, leads to a noisy and blurred picture of the true unobserved risks.
Within the disease-mapping paradigm, questions have been raised about the performance of these models in recovering the true risk surface, the influence of the prior structure specified, and the amount of smoothing of the risks actually performed by these models. In other words, it is important to understand thoroughly the sensitivity (ability to detect true patterns of heterogeneity of risk) and the specificity (ability to discard false patterns created by noise) of Bayesian disease-mapping models. This is the focus of this article. This understanding is crucial for interpretation of any specific disease pattern derived through the use of such models. Such a calibration study cannot be performed on real data because it relies on knowing the true underlying pattern of risk. We have thus conducted an extensive simulation study where the generated data patterns are close to those found in typical disease-mapping studies. We report here the main conclusions that can be drawn.
Let us stress that we are not placing ourselves in the context of cluster detection methods based on so-called point data, that is, data where the precise geographic location of all the cases (and controls) is known. These methods, which have been reviewed in a number of monographs or special issues (e.g., Alexander and Boyle 1996) are typically used on a localized scale, mostly to study the spatial distribution of cases around a point source or different patterns of randomness or clustering of the cases in relation to those of controls. Here we are concerned with methods for describing the overall spatial pattern of cases aggregated over small areas and the interpretation of the residual variations once the Poisson noise has been smoothed out by the disease-mapping models. Several simulation exercises to study different aspects of the performance of disease-mapping models have been reported recently. For example, Lawson et al. (2000) compared a number of models that could be used for disease mapping by goodness of fit criteria that included correlation between the simulated map and the smoothed map and a Bayesian information criterion. They concluded that the version of the CAR model proposed by Besag et al. (1991) [Besag, Yorke, and Mollié (BYM) model] was the most robust model among those with spatial structure. We also consider the BYM model here and compare its performance with two models not considered by Lawson et al. (2000): a version of the BYM model that is more robust to outliers and hence may be better able to detect abrupt changes in the spatial pattern of risk, and a spatial mixture model proposed by Green and Richardson (2002) (see the section on Bayesian disease-mapping models for details). Our study also specifically addresses the smoothing of high-risk areas and further use of the posterior distribution of the relative risks for detecting areas with excess risk—issues not considered by Lawson et al. (2000). Jarup et al. (2002) reported the results of a small simulation study similar to ours; we chose the same study area and expected counts to carry out our comprehensive exercise.
In the next section we describe the simulation setup and the Bayesian disease-mapping methods to be compared. We then discuss the interpretation of the means of the posterior distribution of the relative risk estimates typically reported in disease-mapping studies and displayed as summary maps, and we illustrate quantitatively how much smoothing is performed. We then discuss how information on the whole posterior distribution of the relative risks can be better exploited to discriminate between areas displaying higher risk and areas with relative risk close to background level. We conclude with a short discussion that emphasizes the importance of interpreting any results from a disease-mapping exercise in the context of the size of expected counts and the potential spatial structure of the risks.
Materials and Methods
The basic setup of disease-mapping studies is as follows: The number of cases of a particular disease Yi occurring in area Ai is recorded, where the set of areas {Ai},i = 1, 2,…,n represents a partition of the region under study. For each area Ai, the expected number of cases Ei is also computed using reference rates for the disease incidence (or mortality) and the sociodemographic strata (with respect to age, sex, and perhaps socioeconomic characteristics) where census data are available.
The distribution of the counts Yi is typically assumed to come from a Poisson distribution, as the diseases usually considered in such studies are rare and this distribution gives a good approximation to the underlying binomial distribution that would hold for each risk stratum. The local variability of the counts is thus modeled as follows:
The parameter of interest is θi, the relative risk that quantifies whether the area i has a higher (θi > 1) or lower (θi < 1) occurrence of cases than that expected from the reference rates. It is this parameter that we are trying to estimate to quantify the heterogeneity of the risk and to highlight unusual patterns of risks.
Data Generation
The spatial structure used throughout the simulations is that of the 532 wards in the county of Yorkshire, England. Wards are administrative areas in the United Kingdom, with a total population of approximately 5,000 on average. We base our expected counts Ei on those calculated by Jarup et al. 2002 for prostate cancer in males 45–64 years of age over the period from 1975 to 1991. We then simulate three spatial patterns of increased risks. For each pattern, we examine three magnitudes for the elevated risks. We also examine how the inference is changed if the expected counts are multiplicatively increased by a scale factor (SF) varying from 2 to 10.
Three spatial patterns for areas of elevated risk were chosen. The choice of patterns was intended to span a spectrum ranging from a scenario with single isolated areas with elevated risks (the hardest test case for any smoothing method) to a scenario with a number of larger clusters of several contiguous areas with elevated risks (a situation with a substantial amount of heterogeneity). In all cases the elevated areas were selected in turn at random from the set of areas with the required expected counts. In the Simu 1 and Simu 3 cases, once an area was selected, a buffer of neighboring areas with background risk (excluded thereafter from the random selection) was placed around it to produce the required pattern of isolated high-risk clusters. The three generated patterns were defined as follows:
Simu 1: five isolated single wards with expected counts ranging from 0.8 to 7.3 corresponding, respectively, to the 10th, 25th, 50th, 75th, and 90th percentiles of the distribution of the expected counts
Simu 2: a group of contiguous wards representing 1% of the total expected counts. In effect, this chosen 1% cluster grouped four wards with fairly comparable expected counts ranging from 3.6 to 7.0, giving an average expected count per ward of 5.4 over the four wards
Simu 3: a situation with high heterogeneity comprising 20 such 1% clusters that are nonoverlapping.
Note that for Simu 3, the twenty 1% clusters each have a total expected count close to 17 but a large disparity in terms of numbers of constitute areas: 10 clusters had 2 or 3 areas, whereas 8 clusters had more than 8 areas, up to a maximum of 18 areas. Correspondingly, the expected counts in each of the wards in the clusters ranged from 0.3 for some wards in the 18-area cluster to 12 for the cluster with 2 areas. Simu 3 thus corresponds to a realistic situation of heterogeneity of risk where both small clusters with high expected counts, for example, typically a populated area, and large clusters each with small expected counts, for example, in rural areas, are present. This high degree of heterogeneity has to be considered when interpreting the results for the Simu 3 case where an average over all the 20 clusters is presented. Note also that contrary to the Simu 2 case, about half the background areas in Simu 3 have a neighbor that belongs to one of the 20 clusters. In each case, apart from the elevated risk areas described above, all other areas are called background areas.
For each spatial pattern in Simu 1 and Simu 2, counts Yi were generated as follows: Counts in all background areas were generated from a Poisson distribution with mean Ei. For all the other areas, an elevated relative risk with magnitude θi > 1 was used and counts were simulated as Poisson variables with mean θiEi . The simulation was repeated for three values of θi (1.5, 2, and 3) and for different SFs that multiply the expected counts Ei for all areas. Thus, results reported, for example, for an area with E = 1.92, θ = 2, and SF = 4, correspond to counts generated from a Poisson with mean 15.36 (2 × 4 × 1.92).
For Simu 3 a slightly different procedure for generating the cases was used to ensure that Σ Yi = Σ Ei (Appendix A). Note that for Simu 1 and Simu 2, the simulation procedure meant only that ΣYi ≈ ΣEi. This corresponds, for instance, to an epidemiologic situation where expected counts Ei are calculated based on an external reference rate. However, Simu 3 uses internal reference rates because otherwise ΣYi would have been much larger than Ei , which could distort the overall risk estimates. The multinomial procedure used in Simu 3 and detailed in Appendix A implies that, in effect, the multiplicative contrast between areas of elevated risk and background areas is still 3, 2, and 1.5, respectively, but the corresponding relative risks in each area (denoted * θ i ) relative to the internal (i.e., study region average) reference rates are now 2.1, 1.65, and 1.35 for the elevated areas and 0.7, 0.82, and 0.9 for the background areas.
To allow for sampling variability, each simulation case was replicated 100 times. The results presented are averaged over these 100 replications. A total of 36 simulation scenarios were investigated, corresponding to three spatial patterns (Simu 1, 2, and 3) × three different magnitudes of elevated risk (θ = 3, 2, and 1.5) × 4 SFs for the expected counts Ei (SF = 1, 2, 4, and 10).
Bayesian Disease-Mapping Models
Bayesian disease-mapping models treat the relative risks {θi } as random variables and specify a distribution for them. This part of the model is crucial, as the distributional assumptions thus made allow borrowing of information across the areas. The distribution specified is referred to as the second hierarchical level of the model to distinguish it from the first-level distribution specified in equation 1 that pertains to the random sampling variability of the observed counts about their local mean. It is at this second level that the spatial dependence between the relative risks is introduced. This spatial dependence is represented by means of a prescribed neighborhood graph that defines the set of neighbors (denoted by ∂i) for each area i.
The most commonly used parametric model at the second level of the hierarchy is the CAR model. This specifies the distribution of the log relative risks vi = log(θi ) by
where σ2 is an unknown variance parameter, and &vmacr;i
= Σj∈∂ivj/ni , where ni is the number of neighbors of area i. Thus, essentially the log relative risk in one area is influenced by the average log relative risk of its neighbors, with variability characterized by a conditional variance σ2/ni
This CAR model makes a strong spatial assumption and has only one free parameter linked to the conditional variance σ 2. To increase flexibility, Besag et al. (1991) recommend modeling log (θi) as the sum of a CAR process and an unstructured exchangeable component δi
~ N(0,τ2), i = 1, …,n independently:
This is the BYM model introduced by Besag et al. (1991) that we referred to earlier. We use this model as a benchmark, as its use in disease-mapping studies has been widespread since 1991.
The Gaussian distribution used in the CAR specification above induces a high level of smoothness. In the same 1991 article, Besag et al. (1991) discussed an alternative specification using the heavier-tailed, double-exponential distribution rather than the Gaussian distribution in Equation 2. In effect, this is similar to performing a median-based local smoothing (or L1 norm) rather than a mean-based smoothing, thus allowing more abrupt changes in the geographical pattern of risk. We will refer to this model as L1-BYM.
With any such parametric specification, the amount of smoothing performed (e.g., controlled by the parameters σ2 and τ2) is affected globally by all the areas and is not adaptive. Concerns that such parametric models could oversmooth have led several authors to develop semiparametric spatial models that replace the continuously varying spatial distribution for {θi} by discrete allocation or partition models. Such models allow discontinuities in the risk surface and make fewer distributional assumptions. Partition models that allow a variable number of clusters have been proposed by Denison and Holmes (2001) and Knorr-Held and Rasser (2000).
In this article we investigate the performance of a related spatial mixture model recently proposed by Green and Richardson (2002) that we refer to as MIX. This model leads to good estimation of the relative risks compared with the BYM model for a variety of cases of discontinuities of the risk surface. The idea underlying the MIX model is to replace a continuous model for θi by a mixture model that uses a variable number of risk classes and a spatially correlated allocation model to distribute each area to a class. By averaging over a large number of possible configurations, the marginal distribution of the relative risk is nevertheless smooth. To be precise, it is assumed that θi
= θZi , where Zi, i = 1, 2,…,n are allocation variables taking values in 1, 2,…,k and θj, j = 1,2,…,k are the values of the relative risks that characterize the k different components or risk classes. To have maximum flexibility, the number of components k of the mixture is treated as unknown. Given k, the allocations Zi follow a spatially correlated process, the Potts model, which has been used in image processing and other spatial applications and involves a positive interaction parameter ψ (similar to an autocorrelation parameter) that influences the degree of spatial dependence of the allocations. Specifically, the allocation of an area to a risk component will be favored probabilistically by the number of neighbors currently attributed to that component scaled multiplicatively by ψ. In this way the prior knowledge that areas close by tend to have similar risks can be reflected through the allocation structure. The interaction parameter ψ is treated as unknown and jointly estimated with the number of components and their associated risk. The MIX model can adapt to various patterns of risk and model discontinuities by creating a new risk class if there is sufficient information in the data to warrant this. Further details on the specification of the model are given in Green and Richardson (2002). Thus, in the comparison described later, we have implemented one reference model BYM and two alternative models, the parametric L1-BYM and the semiparametric MIX model.
Implementation
Bayesian inference is based on the joint posterior distribution of all parameters given the data. In our case this joint distribution is mathematically intractable and is simulated using the framework of Markov chain Monte Carlo techniques now commonly used in Bayesian analyses (Gilks et al. 1996). All parameters involved in the models described above, for example, the variances σ2 or τ2 or the interaction parameter ψ, are given prior distributions at a third level of the hierarchy. Implementation of the BYM and L1-BYM was carried out using the free software WinBUGS (Spiegelhalter et al. 2002). Implementation of the MIX model was carried out using a purpose-built Fortran code.
Results
How Smooth Are the Posterior Means?
The results of a Bayesian disease-mapping analysis are typically presented in the form of a map displaying a point estimate (usually the mean or median of the posterior distribution) of the relative risk for each area. To interpret such maps, one needs to understand the extent to which the statistical model is able to smooth the risk estimates to eliminate random noise while at the same time avoiding oversmoothing that might flatten any true variations in risk. To address this issue, we consider the two aspects separately: a) do the Bayesian methods provide adequate smoothing of the background rates, and b) to what extent is the posterior mean estimate different from the background risk in the small number of areas simulated with a true elevated risk?
In all the cases simulated, we found substantial shrinkage of the relative risk estimates for the background rates. This is well illustrated in Figure 1, which displays raw and smooth estimates for all the background areas of Simu 2 and an SF of 1 or 4. Note that when SF = 1, the histogram of the raw standardized mortality or morbidity ratio (SMR) estimates is very dispersed (Figure 1A), with a range of 0–11, and shows a skewed distribution. Clearly, mapping the raw SMRs would present a misleading picture of the risk pattern, whereas any of the three Bayesian models give posterior mean relative risk estimates for the background areas that are well centered on 1 (Figure 1B–D), with just a few areas having estimates outside the 0.9–1.1 range. When the expected counts are higher (SF = 4), the histogram of the raw SMRs is less spread but still substantially overdispersed, whereas those corresponding to the three models are even more concentrated on 1 than when SF = 1 (Figure 1F–H). Thus the false patterns created by the Poisson noise are adequately smoothed out by all the disease-mapping models.
Details of the performance of the BYM model in estimating the relative risk of the high-risk areas are presented in Table 1, with findings for L1-BYM and MIX shown in Tables 2 and 3, respectively. Overall, for the BYM model, a great deal of smoothing of the relative risks is apparent. For the isolated areas in Simu 1, one can see that relative risks of 1.5 in any single area are smoothed away, even in the most favorable case of an area with expected counts of 70 (90% area SF = 10). When the simulated relative risk is 2, the posterior mean risk estimate is above 1.2 only when the expected count is around 50 or more (e.g., 75% area with SF = 10). Relative risks of 3 are smoothed to about half their values when the expected counts are around 10 (e.g., 25% area with SF = 10 or 75% area with SF = 2). Comparison of Simu 2 with Simu 1 (75% area) shows that having a cluster of high-risk areas rather than a single area with elevated risk slightly decreases the amount of smoothing for the same average expected count. Again, this is apparent in the many-cluster situation of Simu 3, where even though the true θ*i are smaller, the relative risk estimates are higher than those for Simu 2.
Overall, the performance of the L1-BYM model (Table 2) is similar to that of the BYM model. However, as expected, the L1-BYM model effects a little less smoothing in cases of large expected counts or high relative risk estimates. For Simu 3 the estimates are nearly identical to those of the BYM model. Thus, simply changing the distributional assumptions in the autoregressive specifications results in only a small modification in the estimates.
The results for the MIX model given in Table 3 show a different pattern than those for the BYM or L1-BYM. For Simu 1 and an elevated relative risk of 1.5, strong smoothing toward 1 is apparent as for BYM. However, for Simu 2, posterior mean relative risks become higher than 1.2 for the largest SF. At the other end of the spectrum, relative risks of 3 are well estimated with posterior means above 2.5 as soon as the expected count is above 10 either for single areas (e.g., 50% area with SF = 4) or for the 1% clustered areas with SF = 2. These results are in accordance with the nature of the MIX model. When there is sufficient evidence in the data to create a group of areas with higher risk, the posterior mean risks for the areas in this group are well estimated and close to the simulated values. Otherwise, all areas are allocated to the background category and smoothed toward 1.
Having many heterogeneous clusters as in Simu 3 does not improve the MIX performance as much as that of BYM. Because of the more diffuse nature of some of the clusters, more areas in the background are randomly included in the group of areas with higher risk. Thus, the MIX model still has a mode close to the true relative risk, but the histogram of the mean posterior risks for all the high-risk areas has a longer left-hand tail than in the Simu 2 scenario (Figure 2).
The difference in performance of the three models is further illustrated in Figure 3, which displays, for the three models, box plots of the posterior mean estimates of the relative risk in the raised-risk areas over the 100 replicates for Simu 2 with true relative risks of 3 and 2. When the true relative risk is 3, the MIX model is clearly performing better than the other two models, whereas for a relative risk of 2 and the lowest SF, the MIX model is the model that produces the most smoothing.
Interpreting the Posterior Distribution of the Risk
Mapping the posterior mean relative risk as discussed previously does not make full use of the output of the Bayesian analysis that provides, for each area, samples from the whole posterior distribution of the relative risk. Mapping the probability that a relative risk is greater than a specified threshold of interest has been proposed by several authors [e.g., Clayton and Bernardinelli (1992)]. We carry this further and investigate the performance of decision rules for classifying an area Ai as having an increased risk based on how much of the posterior distribution of θi exceeds a reference threshold. Figure 4 presents an example of the posterior distribution of the relative risk for such an area. The shaded proportion corresponds to the posterior probability that θ > 1. To be precise, to classify any area as having an elevated risk, we define the decision rule D(c, R0), which depends on a cutoff probability c and a reference threshold R0 such that area Ai is classified as having an elevated risk according to D(c, R0) ↔ Prob(θi > R0) > c. The appropriate rules to investigate will depend on the shape of the posterior distribution of θi for the elevated areas. We first discuss rules adapted to the autoregressive BYM and L1-BYM models. For these two models we have seen that, in general, the mean of the posterior distribution of θi in the raised-risk areas is greater than 1 but rarely above 1.5 in many of the scenarios investigated. Thus, it seems sensible to take R0 = 1 as a reference threshold. We would also expect the bulk of the posterior distribution to be shifted above 1 for these areas, suggesting that cutoff probabilities well above 0.5 are indicated. In the first instance, we choose c = 0.8. Thus, for the BYM and L1-BYM models, we report results corresponding to the decision rule D(0.8, 1). See Appendix B for a detailed justification of this choice of value of c and the performance of different decision rules.
In contrast, we have seen that the mean of the posterior distribution of θi for raised-risk areas for the MIX model is closer to the true value for many scenarios, and there is clear indication that the upper tail of this distribution can be well above 1. Furthermore, the spread of this distribution is less than the corresponding one for the BYM or L1-BYM models, as noted by Green and Richardson (2002). The choice of threshold is thus more crucial for this model, making it harder to find an appropriate decision rule. After some exploratory analyses of the simple clusters in Simu 1 and Simu 2, we found that a suitable decision rule for the MIX model in these two scenarios is to choose R0 = 1.5. For such a high threshold, one would expect that it is enough for a small fraction (e.g., 5 or 10%) of the posterior distribution of θi to be above 1.5 to indicate that an area has elevated risk. Thus, for the MIX model we report results corresponding to the decision rule D(0.05, 1.5).
Two types of errors are associated with any decision rule: a) a false-positive result, that is, declaring an area as having elevated risk when in fact its underlying true rate equals the background level (an error also traditionally referred to as type I error or lack of specificity); and b) a false-negative result, that is, declaring an area to be in the background when in fact its underlying rate is elevated (an error also referred to as type II error or lack of sensitivity). In epidemiology, performances are discussed either by reporting these error rates or their complementary quantities that measure the success rates of the decision rule. The two goals of disease mapping can be summarized as follows: not to overinterpret excesses arising by chance, that is, to minimize the false-positive rate but to detect patterns of true heterogeneity, that is, to maximize the sensitivity. We thus choose to report these two easily interpretable quantities. To be precise, for any decision rule D(c, R0), we compute
the false-positive rate (or 1 – specificity), that is, the proportion of background areas falsely declared elevated by the decision rule D(c, R0)
the sensitivity (or 1 – false-negative rate), that is, the proportion of areas generated with elevated rates correctly declared elevated by the decision rule D(c, R0).
It is clear that there must be a compromise between these two goals: a stricter rule (i.e., one with a higher value of c or R0 or both) reduces the false-positive rate but also decreases the sensitivity and thus increases the false-negative rate. Thus, to judge the performance of any decision rule, one has to consider both types of errors, not necessarily equally weighted. See Appendix B for an illustration of the implication of different weighting on the overall performance of the decision rule.
Table 4 summarizes the probabilities of false-positive results for the three models. For BYM and L1-BYM, the probabilities stay below 10% with no discernible pattern for Simu 1 and Simu 2. The error rates are clearly smaller and around 3% for Simu 3. In this scenario, the background relative risk is shifted below 1, so a decision rule with R0 = 1 is, in effect, a more stringent rule than in the case of Simu 1 and Simu 2 where the background relative risks are close to 1. For the MIX model, the false-positive rates are quite low for Simu 1 and Simu 2 and stay mostly below 3%. However, as shown in the last line of Table 4, these rates have greatly increased for the Simu 3 scenario, indicating that the decision rule D(0.05, 1.5) is no longer appropriate in this heterogeneous context. The heterogeneity creates a lot of uncertainty, with some background areas being grouped with nearby high-risk areas; consequently, the rule D(0.05, 1.5) is not stringent (specific) enough. Thus, we have investigated a series of rules D(c, 1.5) for c = 0.1–0.4 for the MIX model in the Simu 3 scenario. As c increases, the probability of false positive decreases; for D(0.4, 1.5), the probability is, on average, around 3% and always below 7% (Table 5).
Concerning the detection of truly increased relative risks and sensitivity, we first discuss the results for the BYM and L1-BYM models. As expected from the posterior means shown in Tables 1 or 2, the ability to detect true increased risk areas is limited when the increase is only of the order of 1.5. If one takes as a guideline the cases where the detection of true positive is 50% or more, Tables 6 and 7 show that this sensitivity is reached for an expected count of around 50 in the case of a single isolated area and around 20 for the 1% cluster scenario. This shows that for rare diseases and small areas, there is little chance of detecting increased risks of around 1.5 while adequately controlling the false-positive rate.
True relative risks of 2 are detected with at least 75% probability when expected counts are between 10 and 20 per area, depending on the spatial structure of the risk surface, whereas true relative risks of 3 are detected almost certainly when expected counts per area are 5 or more. There is no clear pattern of difference between the results for BYM and L1-BYM; overall, the sensitivity is similar. For Simu 3 we see that the sensitivity is lower than for the other simulation scenarios with equivalent expected counts (as were the rates of false positive in Table 4), in line with the true relative risks being closer to 1 than for Simu 1 and Simu 2. Hence, the decision rule D(0.8, 1) is more specific but less sensitive in this scenario. In situations comprising a large degree of heterogeneity akin to Simu 3, it thus might be advantageous to consider alternative rules, even if the rate of false positive is less well controlled. For example, in the case of a true relative risk (θ) = 1.65 and SF = 4, the use of rule D(0.7, 1) for the BYM model leads to a higher probability of false positive (6% compared with the 3% shown in Table 4). However, the corresponding gain in sensitivity is more than 10%, with the probability of detecting a true positive increasing to 82% compared with 71% when using the rule D(0.8, 1) (Table 5). Nevertheless, even with this relaxed and more sensitive rule, the chance of detecting a true relative risk as small as 1.3 is only around 50% if the SF is 4 (i.e., average cluster with total expected count around 80). On the other hand, true relative risks of around 2 are detected with high probability as soon as the SF is 2 (which corresponds, on average, to a cluster with total expected count of 40).
The contrasting behavior of the MIX model is again apparent in Table 8 when one compares the results for the θ =1.5 scenario with the other columns. For Simu 1 and Simu 2 the sensitivity is generally below that of the BYM model and especially when the true relative risk is 1.5; single clusters with θ = 1.5 are simply not detected. In the 1% cluster case expected counts of at least 20 (10) are necessary to be over 95% certain of detecting a true relative risk of 2 (3) (Table 8). Note that the results of the last line of Table 8 should be discounted in view of the high probability of false-positive results corresponding to this scenario (Simu 3) for the D(0.05, 1.5) rule shown in Table 4. Thus, it is apparent that for the MIX model, it is hard to calibrate a good decision rule appropriate for a variety of spatial patterns of elevated risk. In Table 5 we summarize the results corresponding to the decision rule D(0.4, 1.5), which offers a reasonable compromise between keeping the rate of false positives below 7% and an acceptable detection rate of true clusters. With this rule true relative risks of 1.65 with an SF of 2 (i.e., average cluster with total expected count slightly under 40) or larger have more than a 50% chance of being detected, and true relative risks of around 2 are nearly always detected. However, this model does not detect a true relative risk as small as 1.3.
Discussion
This comprehensive simulation study highlights some important points to be considered in interpreting any disease-mapping exercise based on hierarchical Bayesian procedures. First, the necessary control of false positives is indeed achieved using any of the models described. However, this is accompanied by a strong smoothing effect that renders the detection of localized increases in risk nearly impossible if these are not based on large (3-fold or more) excess risks or, in the case of more moderate (2-fold) excess risks, substantial expected counts of approximately 50 or more. Thus, in any study it is important to report the range of expected counts across the map and to calibrate any conclusions regarding the relative risks with respect to these expected counts.
In general Bayes procedures offer a tradeoff between bias and variance reduction of the estimates. Particularly in cases where the sample size is small, they produce a set of point estimates that have good properties in terms of minimizing squared error loss (Carlin and Louis 2000). This variance reduction is attained through borrowing information resulting from the adopted hierarchical structure, leading to Bayes point estimates shrunk toward a value related to the distribution of all the units included in the hierarchical structure. The effect of shrinkage is thus dependent on the prior structure that has been assumed and conditional on the latter being close to the true model in some sense. Consequently, different prior structures will lead to different shrinkage. Note that the desirable properties of the estimates thus obtained will depend on the ultimate goal of the estimation exercise. If producing a set of point estimates of the relative risk is the aim, then posterior means of the relative risk are best in squared error loss terms. However, if the goal is to estimate the histogram or the ranks of the area relative risks, different loss function should be considered. The desirability and difficulty of simultaneously achieving these triple goals has been discussed by Shen and Louis (1998) and has been illustrated in spatial case studies by Conlon and Louis (1999) and Stern and Cressie (1999). In our study, we focus on the goal of estimating the overall spatial pattern of risk, which involves producing and interpreting a set of point estimates that will not only give a good indication of the presence of heterogeneity in the relative risks but also highlight where on the map this heterogeneity arises and whether this is linked to isolated high- and/or low-risk areas or to more general spatial aggregation of areas of similar high or low risk. Inference about the latter will depend on the sensitivity and specificity of the posterior risk estimates, as discussed in this article. If the goal is purely the testing of heterogeneity, other methods could be used, such as the Potthoff-Whittinghill test or scan statistics [see Wakefield et al. (2000) for review] that test for particular prespecified patterns of overdispersion. Conversely, if the aim is a local study around a point source, then again, the disease-mapping framework is not appropriate, and focused models that make use of the additional information about the location of the putative cluster of high risk are required (Morris and Wakefield 2000).
We have shown that besides reporting and mapping the mean posterior relative risk, the whole posterior distribution can be usefully exploited to try to detect true raised-risk areas. For the BYM model, decision rules based on computing the probability that the relative risk is above 1 with a cutoff between 70 and 80% gives a specific rule. With this type of rule an average expected count of 20 in each of the raised-risk areas leads to a 50% chance of detecting a true relative risk of 1.5, but at least a 75% chance if the true relative risk is 2. For the same scenarios, the posterior mean relative risks are 1.05 and 1.23, respectively, showing that the posterior probabilities rather than the mean posterior relative risks are crucial for interpreting results from the BYM model. On the other hand, 3-fold increases in the relative risk are detected almost certainly with average expected counts of only 5 per area, although the mean of the posterior distribution is typically smoothed to about half the true excess. Note that the performance of the BYM model does improve when the risk is raised in a small group of contiguous areas with similar expected counts rather than in a single area because of the way spatial correlation is taken into account in these models.
We found no notable difference in performance between the BYM model, which uses a Gaussian distribution, and the L1 BYM version, which uses a heavier-tailed, double-exponential distribution. This finding is in agreement with that of an earlier simulation study (Best et al. 1999) that compared these two models. However, there were some clear differences between the BYM models and the spatial allocation model MIX. The performance of the latter model is characterized by an all-or-none feature in the sense that it tends to allocate the true raised-risk areas to either an elevated risk group or to a background group, depending on how much uncertainty is present in the data. If the information from the data is sufficient (i.e., moderate-size expected counts and/or high true excess risks) the MIX model is able to separate the raised-risk and background areas quite well, producing considerably less smoothing of the raised-risk estimates than BYM. When the information in the data is sparse, uncertainty in the groupings leads to more smoothing than the BYM. This type of dichotomy makes any decision rule exploiting the posterior distribution of the relative risks hard to calibrate and less useful than for the BYM model. The MIX model is best used for providing estimates of the underlying magnitude of the relative risks if those are clearly raised rather than as a tool for detecting the presence of areas with excess risk in a decision rule context.
Conclusion
We have quantified to what extent some usual and some more recently developed Bayesian disease-mapping models are conservative, in the sense that they have low sensitivity for detecting raised-risk areas that have only a small excess risk but that, conversely, any identified patterns of elevated risk are, on the whole, specific. We would view this amount of conservatism as a positive feature, as we wish to avoid false alarms when investigating spatial variation in disease risk. However, the magnitude of the risk in any areas identified as raised is likely to be considerably underestimated, and it is worth investigating a range of spatial priors that produce different amounts of smoothing. Given that most environmental risks are small, it is clear that such methods are seriously underpowered to detect them. This represents a major limitation of the small-area disease-mapping approach, although exploiting the full posterior distribution of the relative risk estimates using the decision rules proposed here can improve the discrimination between areas with background and elevated rates. For localized excesses where the geographic source of the risk can be hypothesized, these methods are not appropriate, and focused tests should be used instead. Future applications of small-area disease-mapping methods should therefore consider carefully the tradeoff between size of the areas, size of the expected counts, and the anticipated magnitude and spatial structure of the putative risks. Recently proposed multivariate extensions of Bayesian disease-mapping models (e.g., Gelfand and Vounatsou 2003; Knorr-Held and Best 2001) also deserve further consideration, as they may lead to improved power by enabling risk estimates to borrow information across multiple diseases that share similar etiologies as well as across areas.
This aticle is part of the mini-monograph “Health and Environment Information Systems for Exposure and Disease Mapping, and Risk Assessment.”
Appendix A
Generation of the observed cases for Simu 3 using the multinomial distribution.
For Simu 3, the number of cases for each of the 532 areas is generated using the multinomial distribution as follows:
where N is the total number of cases in the study region and is set equal (to the nearest integer) to the sum of the expected counts across all 532 areas. Hence N = 1,732 for the SF = 1 scenario and appropriate multiples of this for the other SFs.
The parameter θi represents the relative risk in area i relative to some nominal external reference rate. However, the constraint Σi
Yi
= N = Σi
Ei imposed by the multinomial sampling effectively rescales the true relative risk in each area to be
The interpretation of θ*
i is the relative risk in area i relative to the average risk in the study region.
Appendix B
Tradeoff between false-positive and false-negative rates for different decision rules.
Figure B1 shows three different loss functions representing weighted tradeoffs between the two types of errors: false positive and false negative, associated with the D(c, 1) decision rule for detecting raised-risk areas using the BYM model, plotted against cutoff c. Defining as in the text the false-negative rate to be the probability of failing to detect a true raised risk (i.e., 1 – sensitivity), and the false-positive rate to be the probability of false detection of a background area as corresponding to a raised risk (i.e., 1 – specificity), the three loss functions used are as follows:
Figure B1 Variation of the total error rate (loss function) as function of the cutoff probability c for different weighting of the two types of errors (false positive and false negative). Results shown are for the Simu 2 using the BYM model.
with each error being equally weighted.
where we weight the false negative error as twice as bad as the lack of specificity.
where we weight the lack of specificity as twice as bad as the false negative.
We wish to choose c to minimize the losses, and the graphs show that, on average, a value of around 0.7–0.8 is appropriate. Note that the plots in Figure B1 are based on Simu 2 with SF = 2 or 4. For a small number of other scenarios (mainly with SF = 1), a value of c < 0.7 was needed to minimize the loss. However, for consistency, we have used the same value of c (= 0.8) for all the BYM and L1-BYM results presented in this article.
Figure 1 Histograms of the raw SMRs (A,E ) and posterior means of the relative risks (B–H) for all the background areas of Simu 2 derived by each of the three models. Note that the crosses on the x-axes indicate the minimum and maximum values obtained. SF indicates the scale factor used for the expected values.
Figure 2 Histograms comparing the distribution of the posterior means of the relative risks estimated by the BYM or MIX models for the high-risk areas of Simu 2 or Simu 3 using a scale factor of 4 for the expected values and a true relative risk (marked by the vertical line on each plot) of θ = 2 (Simu 2) or θ *1=1.65 (Simu 3).
Figure 3 Box plots of the posterior means of the relative risks estimated by the three models for the high-risk areas of Simu 2 as a function of the scaling factor.
Figure 4 Posterior distribution of a relative risk θ, with shaded area indicating Prob(θ > 1).
Table 1 Posterior mean relative risk estimates for the raised-risk areas for the BYM model (average over replicate data sets).
SF = 1
SF = 2
SF = 4
SF = 10
Raised-risk area θ = 1.5 θ = 2 θ = 3 θ = 1.5 θ = 2 θ = 3 θ = 1.5 θ = 2 θ = 3 θ = 1.5 θ = 2 θ = 3
Simu 1
10% area (E = 0.84) 1.01 1.02 1.06 1.01 1.02 1.12 1.01 1.03 1.20 1.01 1.07 1.40
25% area (E = 1.10) 1.03 1.04 1.10 1.00 1.03 1.15 1.01 1.05 1.28 1.02 1.09 1.52
50% area (E = 1.92) 1.02 1.05 1.15 1.00 1.05 1.28 1.02 1.08 1.46 1.03 1.16 1.79
75% area (E = 5.37) 1.03 1.05 1.31 1.03 1.07 1.55 1.04 1.12 1.86 1.05 1.33 2.35
90% area (E = 7.38) 1.03 1.07 1.34 1.03 1.10 1.62 1.04 1.15 2.07 1.07 1.40 2.47
Simu 2
1% cluster (Ē = 5.42) 1.04 1.08 1.45 1.04 1.14 1.76 1.05 1.23 2.11 1.09 1.45 2.43
Simu 3 θ* = 1.35 θ* = 1.65 θ* = 2.1 θ* = 1.35 θ* = 1.65 θ* = 2.1 θ* = 1.35 θ* = 1.65 θ* = 2.1 θ* = 1.35 θ* = 1.65 θ* = 2.1
20 × 1% clusters (Ē range: 0.77–11.6) 1.04 1.23 1.63 1.07 1.3 1.74 1.12 1.38 1.84 1.19 1.48 1.95
Table 2 Posterior mean relative risk estimates for the raised-risk areas for the L1-BYM model (average over replicate data sets).
SF = 1
SF = 2
SF = 4
SF = 10
Raised-risk area θ = 1.5 θ = 2 θ = 3 θ = 1.5 θ = 2 θ = 3 θ = 1.5 θ = 2 θ = 3 θ = 1.5 θ = 2 θ = 3
Simu 1
10% area (E = 0.84) 1.01 1.02 1.05 1.01 1.02 1.12 1.01 1.02 1.16 1.01 1.07 1.21
25% area (E = 1.10) 1.01 1.03 1.11 1.00 1.04 1.15 1.00 1.06 1.24 1.03 1.09 1.35
50% area (E = 1.92) 1.01 1.03 1.16 1.00 1.05 1.28 1.01 1.08 1.55 1.03 1.17 2.22
75% area (E = 5.37) 1.02 1.05 1.32 1.03 1.08 1.56 1.03 1.13 1.98 1.05 1.35 2.67
90% area (E = 7.38) 1.04 1.07 1.48 1.03 1.13 1.93 1.05 1.25 2.43 1.08 1.60 2.72
Simu 2
1% cluster (Ē = 5.42) 1.04 1.08 1.45 1.04 1.14 1.76 1.05 1.23 2.11 1.09 1.45 2.43
Simu 3 θ* = 1.35 θ* = 1.65 θ* = 2.1 θ* = 1.35 θ* = 1.65 θ* = 2.1 θ* = 1.35 θ* = 1.65 θ* = 2.1 θ* = 1.35 θ* = 1.65 θ* = 2.1
20 × 1% clusters (Ē range: 0.77–11.6) 1.04 1.22 1.61 1.07 1.29 1.74 1.12 1.38 1.85 1.19 1.49 1.97
Table 3 Posterior mean relative risk estimates for the raised-risk areas for the MIX model (average over replicate data sets).
SF = 1
SF = 2
SF = 4
SF = 10
Raised-risk area θ = 1.5 θ = 2 θ = 3 θ = 1.5 θ = 2 θ = 3 θ = 1.5 θ = 2 θ = 3 θ = 1.5 θ = 2 θ = 3
Simu 1
10% area (E = 0.84) 1.00 1.01 1.02 1.00 1.02 1.27 1.00 1.01 1.53 1.01 1.10 2.50
25% area (E = 1.10) 1.00 1.02 1.09 1.00 1.01 1.17 1.00 1.05 1.80 1.01 1.22 2.67
50% area (E = 1.92) 1.00 1.02 1.25 1.00 1.04 1.88 1.00 1.23 2.78 1.02 1.72 3.02
75% area (E = 5.37) 1.00 1.03 1.57 1.00 1.07 2.44 1.01 1.42 2.91 1.04 1.87 3.02
90% area (E = 7.38) 1.00 1.03 1.60 1.01 1.09 2.46 1.01 1.49 2.91 1.06 1.89 3.02
Simu 2
1% cluster (Ē = 5.42) 1.02 1.06 1.98 1.01 1.25 2.66 1.03 1.72 2.92 1.21 1.92 2.98
Simu 3 θ* = 1.35 θ* = 1.65 θ* = 2.1 θ* = 1.35 θ* = 1.65 θ* = 2.1 θ* = 1.35 θ* = 1.65 θ* = 2.1 θ* = 1.35 θ* = 1.65 θ* = 2.1
20 × 1% clusters (Ē range: 0.77–11.6) 1.02 1.19 1.55 1.05 1.31 1.64 1.12 1.44 1.81 1.31 1.55 2.06
Table 4 False-positive rates (1 – specificity) for the three models.a
SF = 1
SF = 2
SF = 4
SF = 10
Background θ = 1.5 θ = 2 θ = 3 θ = 1.5 θ = 2 θ = 3 θ = 1.5 θ = 2 θ = 3 θ = 1.5 θ = 2 θ = 3
BYM
Simu 1 0.08 0.10 0.05 0.04 0.06 0.04 0.03 0.08 0.06 0.03 0.05 0.08
Simu 2 0.07 0.06 0.06 0.05 0.05 0.06 0.05 0.05 0.07 0.04 0.08 0.10
Simu 3b 0.02 0.03 0.02 0.02 0.03 0.02 0.03 0.03 0.01 0.03 0.02 0.01
L1-BYM
Simu 1 0.05 0.09 0.06 0.06 0.10 0.05 0.03 0.06 0.06 0.05 0.05 0.08
Simu 2 0.07 0.09 0.06 0.05 0.07 0.06 0.05 0.06 0.06 0.04 0.07 0.08
Simu 3b 0.04 0.03 0.02 0.02 0.03 0.02 0.03 0.03 0.02 0.03 0.02 0.01
MIX
Simu 1 0.00 0.04 0.00 0.01 0.04 0.00 0.03 0.02 0.00 0.02 0.00 0.08
Simu 2 0.00 0.01 0.11 0.00 0.04 0.04 0.00 0.06 0.01 0.01 0.02 0.00
Simu 3b 0.02 0.51 0.44 0.02 0.52 0.25 0.01 0.33 0.12 0.00 0.14 0.03
aDecision rules are D(0.8, 1) for BYM and L1-BYM and D(0.05, 1.5) for MIX.
bFor Simu 3, θ* = 1.35, 1.65, or 2.1 instead of θ = 1.5, 2, or 3, respectively.
Table 5 Simu 3: performance of the BYM and MIX models under alternative decision rules.
SF = 1
SF = 2
SF = 4
SF = 10
θ* = 1.35 θ* = 1.65 θ* = 2.1 θ* = 1.35 θ* = 1.65 θ* = 2.1 θ* = 1.35 θ* = 1.65 θ* = 2.1 θ* = 1.35 θ* = 1.65 θ* = 2.1
BYM – D(0.7, 1)
Probability (false detection) 0.10 0.07 0.05 0.07 0.07 0.04 0.08 0.06 0.03 0.08 0.05 0.02
Probability (true detection) 0.23 0.51 0.71 0.36 0.68 0.84 0.56 0.82 0.93 0.81 0.95 0.99
MIX – D(0.4,1.5)
Probability (false detection) 0.00 0.03 0.07 0.00 0.06 0.05 0.00 0.07 0.03 0.00 0.03 0.01
Probability (true detection) 0.00 0.23 0.76 0.00 0.62 0.88 0.00 0.84 0.93 0.00 0.93 0.98
Table 6 Sensitivity (1 – false-negative rate) for the BYM model.a
SF = 1
SF = 2
SF = 4
SF = 10
Raised-risk area θ = 1.5 θ = 2 θ = 3 θ = 1.5 θ = 2 θ = 3 θ = 1.5 θ = 2 θ = 3 θ = 1.5 θ = 2 θ = 3
Simu 1
10% area (E = 0.84) 0.08 0.06 0.08 0.04 0.02 0.36 0 0.06 0.68 0.02 0.42 0.98
25% area (E = 1.10) 0.36 0.48 0.38 0.20 0.24 0.36 0.20 0.50 0.82 0.28 0.54 1
50% area (E = 1.92) 0.32 0.48 0.40 0.16 0.32 0.66 0.24 0.66 0.98 0.30 0.96 1
75% area (E = 5.37) 0.08 0.30 0.74 0.12 0.52 0.98 0.22 0.76 1 0.66 1 1
90% area (E = 7.38) 0.12 0.22 0.74 0.10 0.64 0.98 0.34 0.88 1 0.88 1 1
Simu 2
1% cluster (Ē = 5.42) 0.18 0.42 0.95 0.30 0.74 1 0.53 0.97 1 0.90 1 1
Simu 3 θ* = 1.35 θ* = 1.65 θ* = 2.1 θ* = 1.35 θ* = 1.65 θ* = 2.1 θ* = 1.35 θ* = 1.65 θ* = 2.1 θ* = 1.35 θ* = 1.65 θ* = 2.1
20 × 1% clusters (Ē range: 0.77–11.6) 0.09 0.34 0.56 0.17 0.51 0.74 0.37 0.71 0.88 0.66 0.90 0.94
aDecision rule is D(0.8, 1).
Table 7 Probability of true detection (sensitivity) for the L1-BYM model.a
SF = 1
SF = 2
SF = 4
SF = 10
Raised-risk area θ= 1.5 θ= 2 θ= 3 θ= 1.5 θ= 2 θ= 3 θ= 1.5 θ= 2 θ= 3 θ= 1.5 θ= 2 θ= 3
Simu 1
10% area (E = 0.84) 0.02 0.04 0.04 0.04 0.08 0.32 0.02 0.02 0.54 0.04 0.28 0.54
25% area (E = 1.10) 0.26 0.34 0.38 0.24 0.38 0.40 0.16 0.46 0.88 0.44 0.52 0.98
50% area (E = 1.92) 0.28 0.38 0.42 0.30 0.42 0.66 0.26 0.56 0.96 0.44 0.86 1
75% area (E = 5.37) 0.08 0.24 0.74 0.06 0.50 0.94 0.20 0.78 1 0.68 1 1
90% area (E = 7.38) 0.16 0.22 0.76 0.10 0.68 0.98 0.24 0.90 1 0.86 1 1
Simu 2
1% cluster (Ē = 5.42) 0.17 0.35 0.91 0.23 0.64 1 0.39 0.95 1 0.85 1 1
Simu 3 θ* = 1.35 θ* = 1.65 θ* = 2.1 θ* = 1.35 θ* = 1.65 θ* = 2.1 θ* = 1.35 θ* = 1.65 θ* = 2.1 θ* = 1.35 θ* = 1.65 θ* = 2.1
20 × 1% clusters (Ē range: 0.77–11.6) 0.10 0.31 0.55 0.16 0.48 0.75 0.35 0.70 0.89 0.65 0.90 0.98
aDecision rule is D(0.8, 1).
Table 8 Probability of true detection (sensitivity) for the MIX model.a
SF = 1
SF = 2
SF = 4
SF = 10
Raised-risk area θ= 1.5 θ= 2 θ= 3 θ= 1.5 θ= 2 θ= 3 θ= 1.5 θ= 2 θ= 3 θ= 1.5 θ= 2 θ= 3
Simu 1
10% area (E = 0.84) 0 0 0.05 0 0.02 0.35 0 0.04 0.56 0 0.31 0.54
25% area (E = 1.10) 0 0.02 0.20 0 0.01 0.30 0 0.16 0.72 0.06 0.53 0.98
50% area (E = 1.92) 0 0.02 0.33 0 0.10 0.77 0 0.51 0.98 0.05 0.94 1
75% area (E = 5.37) 0 0.02 0.51 0 0.18 0.90 0 0.67 0.99 0.10 0.98 1
90% area (E = 7.38) 0 0.05 0.55 0 0.19 0.93 0 0.68 0.99 0.14 0.98 1
Simu 2
1% cluster (Ē = 5.42) 0.02 0.10 0.86 0.01 0.46 0.99 0.05 0.95 1 0.47 1.00 1.00
Simu 3 θ* = 1.35 θ* = 1.65 θ* = 2.1 θ* = 1.35 θ* = 1.65 θ* = 2.1 θ* = 1.35 θ* = 1.65 θ* = 2.1 θ* = 1.35 θ* = 1.65 θ* = 2.1
20 × 1% clusters (Ē range: 0.77–11.6) 0.04 0.85 0.99 0.04 0.99 0.99 0.06 0.99 0.99 0.0 0.99 1.00
aDecision rule is D(0.5, 1.5).
==== Refs
References
Alexander FE Boyle P eds1996 Methods for Investigating Localised Clustering of Disease IARC Sci Publ 135 1 247 9103929
Besag J York J Mollié A 1991 Bayesian image restoration with applications in spatial statistics Ann Inst Math Stat 43 1 20
Best NG Arnol RA Thomas A Waller LA Conlon EM 1999. Bayesian models for spatially correlated disease and exposure data (with discussion). In: Bayesian Statistics 6 (Bernardo JM, Berger JO, Dawid AP, Smith AFM, eds). Oxford:Oxford University Press, 131–156.
Carlin BP Louis A 2001. Bayes and Empirical Bayes Methods. 2nd ed. London:Chapman and Hall.
Clayton D Bernardinelli L 1992. Bayesian methods for mapping disease risk. In: Geographical and Environment Epidemiology: Methods for Small Area Studies (Elliott P, Cuzick J, English D, Stern R, eds). Oxford:Oxford University Press, 205–220.
Clayton D Bernardinelli L Montomoli C 1993 Spatial correlation in ecological analysis Int J Epidemiol 22 1193 1202 8144305
Conlon EM Louis TA 1999. Addressing multiple goals in evaluating region-specific risk using Bayesian methods. In: Disease Mapping and Risk Assessment for Public Health (Lawson A, Biggeri A, Bohning D, Lesaffre E, Viel JF, Bertollini R, eds). Chichester, UK:John Wiley & Sons, 31–47.
Denison DGT Holmes CC 2001 Bayesian partitioning for estimating disease risk Biometrics 57 143 149 11252589
Doll R 1980 The epidemiology of cancer Cancer 45 2475 2485 7378985
Elliott P Wakefield JC Best NG Briggs DJ 2000. Spatial Epidemiology: Methods and Applications. Oxford:Oxford University Press.
Gelfand A Vounatsou P 2003 Proper multivariate conditional autoregressive models for spatial data analysis Biostatistics 4 11 15 12925327
Gilks WR Richardson S Spiegelhalter DJ 1996. Markov Chain Monte Carlo in Practice. London:Chapman and Hall.
Green PJ Richardson S 2002 Hidden Markov models and disease mapping J Am Stat Assoc 97 1055 1070
Jarup L Best N Toledanao M Wakefield J Elliott P 2002 Geographical epidemiology of prostate cancer in Great Britain Int J Cancer 97 695 699 11807800
Knorr-Held L Best NG 2001 A shared component model for detecting joint and selective clustering of two diseases J R Stat Soc [Ser A] 164 73 85
Knorr-Held L Rasser G 2000 Bayesian detection of clusters and discontinuities in disease maps Biometrics 56 13 21 10783772
Lawson AB Biggeri AB Boehning D Lesaffre E Viel JF Clark A 2000 Disease mapping models: an empirical evaluation Stat Med 19 2217 2241 10960849
Morris SE Wakefield JC 2000.Assessment of disease risk in relation to a pre-specified source. In: Spatial Epidemiology: Methods and Applications (Elliott P, Wakefield JC, Best NG, Briggs DJ, eds). Oxford:Oxford University Press.
Shen W Louis TA 1998 Triple-goal estimates in two-stage hierarchical models J R Statist Soc [Ser B] 60 455 471
Spiegelhalter DJ Thomas A Best NG Lunn DJ 2002. WinBUGS: Bayesian Inference Using Gibbs Sampling Manual, Version 1.4. London:Imperial College; Cambridge, UK:MRC Biostatistics Unit. Available: http://www.mrc-bsu.cam.ac.uk/bugs [accessed 17 January 2004].
Stern H Cressie N 1999. Inference for extremes in disease mapping. In: Disease Mapping and Risk Assessment for Public Health (Lawson A, Biggeri A, Bohning D, Lesaffre E, Viel JF, Bertollini R, eds). Chichester, UK:John Wiley & Sons, 63–84.
Wakefield JC Kelsall JE Morris SE 2000. Clustering, cluster detection, and spatial variation in risk. In: Spatial Epidemiology: Methods and Applications (Elliott P, Wakefield JC, Best NG, Briggs DJ, eds). Oxford:Oxford University Press, 128–152.
|
15198922
|
PMC1247195
|
CC0
|
2021-01-04 23:39:20
|
no
|
Environ Health Perspect. 2004 Jun 15; 112(9):1016-1025
|
utf-8
|
Environ Health Perspect
| 2,004 |
10.1289/ehp.6740
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0112-00102610.1289/ehp.674115198923Mini-Monograph: Information SystemsCancer Risk Near a Polluted River in Finland Verkasalo Pia K. 1Kokki Esa 1Pukkala Eero 2Vartiainen Terttu 1Kiviranta Hannu 1Penttinen Antti 3Pekkanen Juha 11Department of Environmental Health, National Public Health Institute, Kuopio, Finland2Finnish Cancer Registry, Institute for Statistical and Epidemiological Cancer Research, Helsinki, Finland3Department of Mathematics and Statistics, University of Jyväskylä, Jyväskylä, FinlandAddress correspondence to P.K. Verkasalo, Department of Environmental Health, National Public Health Institute, P.O. Box 95 (Neulanie-mentie 4), FIN-70701, Kuopio, Finland. Telephone: 358 17 201 481. Fax: 358 17 201 265. E-mail:
[email protected] thank our collaborators in the European Health and Environment Information System for Disease and Exposure Mapping and Risk Assessment (EUROHEIS) project for sharing their expertise in fruitful discussions.
This research was supported by grants EU SI2.291820 (2000CVG2-605) and SI2.329122 (2001CVG2-604) from the European Commission Health and Consumer Protection Directorate-General; grant 52876 from the Academy of Finland; the Ellen and Artturi Nyyssönen Foundation; and the Paavo Koistinen Foundation.
The authors declare they have no competing financial interests.
6 2004 15 4 2004 112 9 1026 1031 12 9 2003 13 4 2004 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. The River Kymijoki in southern Finland is heavily polluted with polychlorinated dibenzo-p-dioxins and dibenzofurans and may pose a health threat to local residents, especially farmers. In this study we investigated cancer risk in people living near the river (< 20.0 km) in 1980. We used a geographic information system, which stores registry data, in 500 m × 500 m grid squares, from the Population Register Centre, Statistics Finland, and Finnish Cancer Registry. From 1981 to 2000, cancer incidence in all people (N = 188,884) and in farmers (n = 11,132) residing in the study area was at the level expected based on national rates. Relative risks for total cancer and 27 cancer subtypes were calculated by distance of individuals to the river in 1980 (reference: 5.0–19.9 km, 1.0–4.9 km, < 1.0 km), adjusting for sex, age, time period, socioeconomic status, and distance of individuals to the sea. The respective relative risks for total cancer were 1.00, 1.09 [95% confidence interval (CI), 1.04–1.13], and 1.04 (95% CI, 0.99–1.09) among all residents, and 1.00, 0.99 (95% CI, 0.85–1.15), and 1.13 (95% CI, 0.97–1.32) among farmers. A statistically significant increase was observed for basal cell carcinoma of the skin (not included in total cancers) in all residents < 5.0 km. Several other common cancers, including cancers of the breast, uterine cervix, gallbladder, and nervous system, showed slightly elevated risk estimates at < 5.0 km from the river. Despite the limitations of exposure assessment, we cannot exclude the possibility that residence near the river may have contributed to a small increase in cancer risk, especially among farmers.
cancerdioxinsepidemiologyGISPCDDPCDFrecord linkage
==== Body
The River Kymijoki is one of the largest rivers in southern Finland, with nearly 190,000 people living < 20.0 km from its shoreline (Figure 1). The river flows south to the Gulf of Finland, which is a part of the Baltic Sea surrounded by nine European countries. The effluents from several pulp and paper mills as well as from manufacturing of chloro alkali chemicals—in particular, a chlorophenol fungicide, Ky-5, in one factory—heavily loaded the river between the 1950s and the 1990s. The discharge of these compounds decreased during the 1990s after improvements in methods of pulp bleaching and effluent treatment and the ceasing of production in 1984 of Ky-5. However, the river sediments still contain high levels of the persistent and toxic environmental pollutants polychlorinated dibenzo-p-dioxins (PCDDs) and poly-chlorinated dibenzofurans (PCDFs) (Finnish Environment Institute 2003; Paasivirta 1996; Verta et al. 1999). The surface sediment levels of PCDD/Fs are between 0.5 and 350 ng/g in dry weight as toxic equivalents and thus are among the highest sediment levels observed worldwide. Elevated PCDD/F concentrations have also been measured in sediments of the Gulf of Finland (Isosaari et al. 2002), in fish caught from the River Kymijoki and the Gulf of Finland (Korhonen et al. 2001), and in fishermen living in the delta area (Kiviranta et al. 1999, 2002; Korhonen et al. 2001).
The most toxic congener, 2,3,7,8-tetra-chlorodibenzo-p-dioxin (2,3,7,8-TCDD), has also been classified by the International Agency for Research on Cancer (IARC) as “carcinogenic to humans” on the basis of sufficient evidence from animal and limited evidence from human studies (IARC 1997). For the other PCDD/Fs, there is inadequate or limited evidence of carcinogenicity from animal studies, and practically no studies have been conducted in humans. Overall, the strongest epidemiologic evidence for the carcinogenicity of 2,3,7,8-TCDD is for all cancers combined rather than for any specific site. The literature suggests an increase of 40% at most, deriving primarily from studies of occupational cohorts with mixed exposures (Kogevinas 2000; Kogevinas et al. 1997) and the industrial accident in Seveso, Italy (Bertazzi et al. 2001; Warner et al. 2002).
In this study we investigated cancer risk in people living near the River Kymijoki (< 20.0 km) using small-area statistics on health (SMASH) system designed for investigations of cancer risk near geographically defined exposure sources in Finland (Kokki et al. 2001). We assumed that PCDD/Fs are mobilized from the river surface sediments and reach nearby residents via the food chain (e.g., by consumption of locally caught fish). We hypothesized that cancer risk increases with decreasing distance to the river. Furthermore, we hypothesized that farmers show a higher risk than most other people, because farmers are more likely to be exposed to river water because of their lifestyles and/or because comparisons within a defined population group are less likely to be confounded.
Materials and Methods
Small-Area Statistics on Health System
The SMASH system has previously been used to investigate cancer risk near geographically defined exposure sources in Finland (Kokki et al. 2001, 2002; Pekkanen et al. 1995). It is a geographic information system (GIS) developed through a collaboration of the Department of Environmental Health, National Public Health Institute, Finland, and the Finnish Cancer Registry. The system runs on ArcView GIS, version 3.2 (Environmental Systems Research Institute Inc., Redlands, CA, USA) and stores nationwide registry data, in 500 m × 500 m grid squares, from the Population Register Centre, Statistics Finland, and the Finnish Cancer Registry. Data include population counts by age, sex, socioeconomic status (SES), and location coordinate of residence for 1980 and all cancer cases from 1981 to 2000).
All three source registries contain nationwide data with good quality and coverage. The Finnish Cancer Registry, founded in 1952, receives information on all known cases of cancer from hospitals, pathological and hematologic laboratories, and practicing physicians. A validation study showed that over 99% of all malignant cancers are registered by the Finnish Cancer Registry (Teppo et al. 1994). In 1999, cancer diagnoses were based on histologic confirmation in 94.6% of cases and solely on death certificates in 0.9% of cases (Finnish Cancer Registry 2003). A total of 27 cancers were selected to be studied. They were classified traditionally according to the International Classification of Diseases, 7th revision, [World Health Organization (WHO) 1995] modified by the Finnish Cancer Registry and include the most common cancer types and others that are of special interest in the case of PCDD/Fs. Basal cell carcinomas (BCCs) of the skin were not included in the total numbers because there are large variations in the BCC rates by hospital catchment area, suggesting that many cases may remain undetected. Nervous system tumors denote tumors of the central as well as the peripheral nervous system. Extranodal non-Hodgkin lymphomas were classified according to their primary site. The original data sets were linked using personal identification numbers unique to every resident in Finland. The data were available in 500 m × 500 m grid squares and were further aggregated according to our hypothesis on geographic reference to the river.
Exposure Assessment
The study population was defined as all people (farmers in particular) living within 20.0 km from the River Kymijoki (i.e., in a 500 m × 500 m grid square at least partially located within 20.0 km from the river shoreline) on 31 December 1980. The correct registration of the place of residence (97% of Finns surveyed actually lived in the same building as that recorded in the registry) (Statistics Finland 1994) and the accurate geocoding of the latitude and longitude of the central points of each residence (± 10 m) ensure the correct spatial registration of cases and reference population relative to exposure sources of interest.
To allow comparisons within the study population, the study area was divided into nine subareas according to increasing distance to the river downstream from the factory producing Ky-5 (< 1.0 km; 1.0–4.9 km; 5.0–19.9 km), and according to increasing distance to the sea (< 20.0 km; 20.0–39.9 km; 40.0–59.9 km) (Figure 1). The cut points were selected a priori to distinguish varying exposure levels but remain, however, hypothetical. According to our primary hypothesis, the people and especially the farmers living nearest the river were suspected to be at the highest risk for cancer risk. The distance to sea variable was intended to measure pollution along the river flow on the north–south axis. However, its meaning is somewhat speculative. For example, many fish samples have been more heavily contaminated with PCDD/Fs close to the Gulf of Finland, but conversely, the surface sediments reach their peak levels near the factory producing Ky-5. SMASH was used to organize geographically defined data sets. The data sets were then entered into the SAS OnlineDoc statistical software, version 8 (SAS Institute Inc., Cary, NC, USA 1999) for estimation of cancer risk.
Statistical Analyses
We assessed the risk for total cancer and 27 selected cancer types for all people, and separately for farmers, living near the river on 31 December 1980. All variables were classified according to the situation in 1980 and available in 500 m × 500 m grid squares. The total number of inhabited grid squares in 1980 was 197,520 for all of Finland and 4,687 for the study area.
For each grid square in Finland, numbers of subjects (population at risk in 1980) and observed cancers were counted by sex, age (5 years of age groups), time period (1981–1990–1991–2000), and SES (upper-level clerical workers, lower-level clerical workers, skilled workers, unskilled workers, farmers, unknown). For the study area, we counted numbers of subjects and observed cases of cancer according to distance between river and residence (< 1.0 km, 1.0–4.9 km, 5.0–19.9 km) and according to distance between sea and residence (< 20.0 km, 20.0–39.9 km, 40.0–59.9 km). We estimated reference incidence rates separately for total Finnish population and Finnish farmers, dividing the number of new cases of cancer by the population at risk in 1980, by sex, age, time period, and SES (in analyses of all people but not farmers). For the study area, we calculated expected numbers of cancers as the number of subjects multiplied by reference incidence rate for that cancer by sex, age, time period, SES, distance to sea, and distance to river.
Standardized incidence ratios (SIRs) were calculated by dividing the observed number of cases by the expected number of cases. SIRs were counted overall and by sex, age, SES (in analyses of all residents but not farmers), time period, distance to sea, and distance to river.
For distance to river comparisons, we used Poisson regression main-effect models for the observed numbers of cases in 3 × 3 contingency tables, where the classification is based on distance to river (three categories) and distance to sea (three categories). Logarithmically transformed expected numbers, formed from the reference population, were used as offset variables. We assumed that the sex, age, and SES, together with geographic effects related to river and sea, address the spatial variation in the data properly and give an interpretation in terms of distances. We plotted the residuals from the models for total cancer among all people and farmers and from models for BCCs among all people by distance to river and distance to sea. The model fits well with the data at the level of aggregation used. Detailed spatial analysis would be possible in theory (e.g., with Poisson regression with a correlated random component) (Best 1999) but would require partition of the study area into finer units and would likely be uninformative because of the small numbers.
We obtained maximum likelihood estimates for relative risk (RR) using PROC GENMOD in statistical software SAS (SAS Institute Inc.). The 95% confidence intervals (CIs) for SIRs are based on Poisson variation around expected values; confidence intervals for relative risk are two-sided Wald CIs. Statistical significance was set at p < 0.05.
Results
In total, 188,884 people were living closer than 20.0 km from the River Kymijoki in 1980 (Table 1). Of these, 83% were < 60 years of age, 6% were farmers, 53% were living < 20.0 km from the sea, and 27% were living < 1.0 km from the river.
Risk in All Residents
A total of 14,242 cases of cancer were diagnosed among the study cohort between 1981 and 2000. The incidence of total cancer in all residents was very similar to the general population risk (SIR = 0.99; 95% CI, 0.98 –1.01) (Table 2). Similarly, when studied by sex, age, or time period, the risk for total cancer differed no more than 3% from the general population risk. There was a subtle increase in the risk for total cancer in those living < 20.0 km and decreases in those living farther away from the Gulf of Finland. The SIRs for the 27 cancer subtypes studied were between 0.76 and 1.21 when comparing all residents with general population (Table 3). Statistically significant risk increases were observed for skin cancers.
The SES-adjusted relative risks for total cancer were 1.04 for those living < 1.0 km (95% CI, 0.99–1.09) and 1.09 for those living 1.0–4.9 km from the river (95% CI, 1.04–1.13), compared with those living 5.0–19.9 km from the river (Table 2). Overall, the SES-adjusted relative risks for total cancer suggested subtle increases between 1 and 15%, when analyzed by background variables. The relative risks were slightly higher for those living 1.0–4.9 km from the river than for those living < 1.0 km from the river. For those living 1.0–4.9 km from the river, relative risks for total cancer were statistically or marginally significantly increased in all subgroups but one (Table 2).
As for cancer subtypes, statistically significant risk increases were observed for BCC in those living < 1.0 km from the river and for cancers of the uterine cervix and corpus, breast, and lung and for BCCs in those living 1.0–4.9 from the river (Table 3). Several other cancer types also showed elevated risk estimates.
Risk in Farmers
Between 1981 and 2000, a total of 1,143 cases of cancer were diagnosed among farmers living in the study area in 1980. The incidence of total cancer in farmers living in the study area did not differ statistically significantly from the incidence in all farmers in the country (RR = 0.96; 95% CI, 0.91–1.02) (Table 4). However, the risk was slightly decreased in men and in those living 40.0–59.9 km from the sea. A statistically significant risk increase was observed for liver cancer, and statistically or marginally significant risk decreases were observed for cancers of the stomach and lung.
The relative risk for total cancer in farmers was highest for those living < 1.0 km from the river (RR = 1.13; 95% CI, 0.97–1.32) (Table 4). The relative risks for total cancer in farmers living < 1.0 km from the river showed increases between 8 and 54% for all categories, although statistically significant increases were not detected. The highest estimate for relative risk (54% increase) was for those < 45 years of age at baseline. No consistent risk increases were observed for farmers living 1.0–4.9 km from the river.
No statistically significant risk changes by distance to river were observed for any of the 24 cancer subtypes for which the models converged (Table 5). However, for farmers living < 1.0 km from the river, the risk estimates for 14 subtypes were elevated by > 5%; the risk estimates for 4 were within 5% from reference; and those for 8 were decreased by > 5%. The respective numbers were 13, 2, and 9 for farmers living 1.0–4.9 km from the river.
Discussion
Small-Area Statistics on Health System
SMASH has been a useful tool in assessing cancer risks in freely selected areas in Finland (Kokki et al. 2001, 2002; Pekkanen et al. 1995). The high quality of nationwide registries on population and cancer (Teppo et al. 1994) also provided an excellent opportunity for the present study. The accurate geocoding of places of residence (± 10 m) contrasts SMASH with systems developed in many other countries (Aylin et al. 1999; National Cancer Institute 2003; National Center for Health Statistics 2003). Adjustment for SES was important, as socioeconomically determined lifestyle variations in risk can easily be attributed to environmental pollutants. In addition, the ability to use the most representative reference population (e.g., comparing farmers with farmers) further reduced the potential effects of confounding due to factors not related to the local environment. On the other hand, limitations of the methodology include the estimated denominators of the risk estimates (based on number of subjects in each 500 m × 500 m grid square in 1980), the small numbers for many cancer subtypes, and most importantly, the nonspecificity of exposure assessment.
Exposure Assessment
In this study, exposure assessment was based solely on the place of residence at one point in time. In other words, we calculated the distance between residence and river shoreline but had no specific measure for PCDD/F exposure. To our knowledge, there are no previous GIS studies that have examined disease risks along a river. However, similar methodologies have been used to study risks close to other line-shaped features such as roads (Harrison et al. 1999), railways (Dickinson et al. 2003), and power lines (Feychting and Ahlbom 1993; Verkasalo et al. 1993). In PCDD/F epidemiology, GIS-based methodologies have previously been applied to detect cancer clusters around a municipal waste incinerator with high PCDD/F emissions (Viel et al. 2000) and to model airborne exposures to PCDD/Fs (Cohen et al. 2002; Floret et al. 2003; Stellman et al. 2003).
Possible health threats related to individuals living near this polluted river are an important issue for both decision makers and the general public. However, the use of a nonspecific surrogate measure for exposure may have introduced considerable measurement error or confounding by correlated exposures. To be considered a confounder, this other (correlating) exposure must be associated with individuals living near the river, and it would also have to show an association with increased risk of total cancer.
During the first half of the study period (as well as during several decades before that), the River Kymijoki was severely loaded with effluents from pulp bleaching and chloro alkali and Ky-5 manufacture, resulting in high environmental levels of polychlorinated phenols, catechols, guaiacols, PCDD/Fs, diphenyl ethers, and mercury (Paasivirta 1996). Of these pollutants, 2,3,7,8-TCDD has shown perhaps the strongest association with increased cancer risk (classified into group 1 by IARC) (IARC 2003). However, < 0.5% of total PCDD/F levels, measured as toxic equivalents, was explained by 2,3,7,8-TCDD (Vartiainen T, unpublished data). Other pollutants such as polychlorophenols (IARC group 2B: “possibly carcinogenic to humans”) and methyl mercury (IARC group 2B) may also be linked with increases in specific cancer subtypes. In practice this means that alternative or simultaneous effects of correlating environmental exposures cannot be excluded. Similarly, the possibility of a chance effect, residual confounding by some SES-related lifestyle factor, or confounding by some unidentified factor cannot be ruled out.
Regional Variation in Cancer
Total cancer incidence in all people living < 20.0 km from the River Kymijoki was at the level expected based on the general population, whereas some particular cancer subtypes showed small increases or decreases in risk. In many cases observed cancer patterns may reflect commonly known reasons for regional variation in cancer.
For example, total cancer incidence in people living < 20.0 km from the Gulf of Finland was slightly increased compared with general population incidence but reflected the incidence in the town of Kotka (data not shown). In addition, the cancer pattern in the 20.0-km zone reflected increased SIRs for cancers of the bladder, pancreas, and skin (but no change for cancers of the stomach, lung, breast, and prostate) in Kotka (data not shown). This is no surprise, because 58% of the population living < 20.0 km from the sea lived precisely in the town of Kotka. Conversely, increased risk for bladder cancer, for example, has been associated with chlorination by-products (Koivusalo et al. 1997), which occur in high levels in the local municipal drinking water (Vartiainen T, unpublished data). Such exposures can prevent detection of an association between living close to the river and increased cancer risk (if such an association exists).
In this study the 23% increase for BCCs in people living < 20.0 km from the Gulf of Finland (data not shown) may reflect the generally high detection rates for BCC in the local hospital catchment area (30% higher incidence in men and 26% higher incidence in women compared with national average rates between 1995 and 1999; calculated based on the Finnish Cancer Registry data). These examples suggest that one should probably place more emphasis on local rather than on countrywide reference populations while using a GIS-based approach.
Increased Risk?
In this study we found that cancer incidence in all people as well as in farmers living close to the River Kymijoki was at the level expected based on national rates. Among all people and farmers living < 1.00 km from the river, however, the SES-adjusted relative risks for total cancer were consistently > 1.00 (statistically nonsignificant), whether analyzed by sex, age, time period, or distance to sea. The lowest estimate for relative risk was 1.01 (for all residents ≥ 60 years of age at baseline); the highest estimate was 1.54 (for farmers < 45 years of age at baseline). The relative risks for farmers were generally higher than the relative risks for all residents. The relative risks for all people were also elevated <1.0–4.9 km from the river.
The magnitude of the effect was thus smaller than the effects described in earlier studies of the occupationally exposed PCDD/ PCDF cohorts (40% increase) (Kogevinas 2000; Kogevinas et al. 1997) and in the study of the Seveso cohort (30% increase) (Bertazzi et al. 2001). However, occupational cohorts tend to show higher risks than general population cohorts.
In principle, the suggestive increase in a broad spectrum of cancers is compatible with the consensus that the strongest epidemiologic evidence for the carcinogenicity of 2,3,7,8-TCDD is for all cancers combined, rather than for any specific site (IARC 1997). Traditionally, there are two clear examples of agents that cause an increase in cancers at many sites: tobacco (Baron and Rohan 1996) and ionizing radiation (Boice et al. 1996). Both, however, also show clearly elevated risks for some specific cancer subtypes. In the case of PCDD/Fs, it is not clear whether some specific cancer subtypes are more strongly associated with the exposure than other subtypes.
We observed a statistically significant risk increase for BCCs among all residents living < 1.0 km from the river. We also observed increases for cancers of the uterine cervix and corpus, breast, and lung, and BCCs among those living 1.0–4.9 from the river. No statistically significant risk changes occurred in cancer subtypes among farmers. However, several rather common cancers showed somewhat elevated risk estimates. The subtypes with suggestive risk increases among all residents and among farmers include the cancers of the uterine cervix and corpus, breast, and gallbladder. Among farmers living < 1 km from the river, on the other hand, suggestive risk increases of at least 50% were observed for cancers of the thyroid, uterine cervix, ovary, gallbladder, rectum, and breast, Hodgkin disease, and non-melanoma of the skin.
Cancers of the reproductive, endocrine, and hematopoietic systems and soft tissue sarcoma have traditionally been of interest to PCDD/F researchers. Our increased risk estimate for breast cancer in women was compatible with other studies suggesting an increase in breast cancer (Warner et al. 2002). Although this study included very few exposed cancers of the hematopoietic system, another GIS-based study has reported 2.3-fold risk increase in non-Hodgkin lymphoma due to PCDD/F emissions from a solid waste incinerator (Floret et al. 2003). Another GIS study examined the spatial distribution of sarcomas and non-Hodgkin lymphomas around a municipal solid waste incinerator with high emission levels of PCDD/Fs, identifying highly significant clusters around the incinerator (Viel et al. 2000). On the other hand, it is also worth noticing that we observed an increased risk estimate for lung cancer. Studies of the occupationally exposed PCDD/Fs cohorts (Kogevinas 2001) and the Seveso cohort (Bertazzi et al. 2001), but not the Swedish Baltic Sea fishermen cohort (Svensson et al. 1995), have reported risk increases for lung cancer.
Conclusions
This study cannot exclude the possibility that residence near the River Kymijoki may have contributed to a subtle increase in the risk of total cancer, especially among farmers. The limitations of the available data and analytical methods must be recognized. It is also vital to appreciate that this is a small area (ecologic) study, where exposure assessment is based solely on place of residence, and the possible biologic pathway is not clear. Thus, this study can provide only first approximations of risks and tell only a little about causality.
This article is part of the mini-monograph “Health and Environment Information Systems for Exposure and Disease Mapping, and Risk Assessment.”
Figure 1 Exposure zones around the River Kymijoki. Reproduced with permission of the National Land Survey of Finland.
Table 1 Distribution of people living < 20.0 km from the River Kymijoki in 1980 according to sex, age, SES, distance to sea, and distance to river.
Variable No. (%)
Sex
Men 91,687 (48.5)
Women 97,197 (51.5)
Age (years)
< 15 37,013 (19.6)
15–29 44,974 (23.8)
30–44 41,621 (22.1)
45–59 33,704 (17.8)
≥60 31,572 (16.7)
SES
Upper-level clerical workers 22,463 (11.9)
Lower-level clerical workers 40,004 (21.2)
Skilled workers 20,275 (10.7)
Unskilled workers 56,012 (29.7)
Farmers 11,132 (5.9)
Unknown 38,998 (20.6)
Distance to the Gulf of Finland (km)
< 20.0 100,276 (53.1)
20.0–39.9 34,007 (18.0)
40.0–59.9 54,601 (28.9)
Distance to the River Kymijoki (km)
< 1.0 51,723 (27.4)
1.0–4.9 82,243 (43.5)
5.0–19.9 54,918 (29.1)
Total 188,884 (100.0)
Table 2 Risk for total cancer between 1981 and 2000 among all people living < 20.0 km from the River Kymijoki in 1980.
Distance to the River Kymijoki (km)
< 1.0
1.0–4.9
5.0–19.9
All people < 20.0
Variable Obs RR 95% CI Obs RR 95% CI Obs RR Obs SIRa 95% CI
All sites 3,866 1.04 0.99–1.09 6,338 1.09 1.04–1.13 4,038 1.00 14,242 0.99 0.98–1.01
Sex
Men 1,819 1.04 0.97–1.11 2,979 1.10 1.03–1.16 1,970 1.00 6,768 0.97 0.95–0.99
Women 2,047 1.04 0.97–1.10 3,359 1.08 1.02–1.14 2,068 1.00 7,474 1.01 0.99–1.04
Age (years)
< 45 793 1.04 0.94–1.15 1,351 1.10 1.00–1.20 811 1.00 2,955 0.98 0.95–1.02
45–59 1,347 1.07 0.99–1.16 2,277 1.11 1.04–1.19 1,359 1.00 4,983 0.99 0.97–1.02
≥60 1,726 1.01 0.95–1.08 2,710 1.07 1.00–1.13 1,868 1.00 6,304 0.97 0.97–1.02
Time period
1981–1990 1,738 1.04 0.97–1.11 2,746 1.07 1.01–1.14 1,825 1.00 6,309 0.97 0.95–1.00
1991–2000 2,128 1.03 0.97–1.10 3,592 1.10 1.04–1.16 2,213 1.00 7,933 1.01 0.98–1.03
Distance to the Gulf of Finland (km)
< 20.0 2,140 1.03 0.97–1.09 3,175 1.07 1.02–1.12 2,816 1.00 8,131 1.04 1.02–1.07
20.0–39.9 795 1.02 0.92–1.14 952 1.06 0.96–1.18 510 1.00 2,257 0.92 0.88–0.96
40.0–59.9 931 1.09 0.98–1.20 2,211 1.15 1.06–1.26 712 1.00 3,854 0.93 0.91–0.96
Obs, observed number of cases.
aAll people in Finland were used as reference for SIRs.
Table 3 Risks for cancer subtypes between 1981 and 2000 among all people living < 20.0 km from the River Kymijoki in 1980.
Distance to the River Kymijoki (km)
< 1.0
1.0–4.9
5.0–19.9
All people < 20.0
Primary site Obs RR 95% CI Obs RR 95% CI Obs RR Obs SIRa 95% CI
Larynx, epiglottis 31 1.64 0.93–2.89 35 1.15 0.66–2.00 20 1.00 86 0.88 0.70–1.09
Cervix uteri 39 1.61 0.97–2.67 61 1.65 1.03–2.64 25 1.00 125 1.02 0.85–1.22
Gallbladder, bile ducts 60 1.45 0.98–2.15 83 1.34 0.93–1.94 44 1.00 187 0.90 0.77–1.03
Corpus uteri 128 1.20 0.93–1.54 220 1.31 1.05–1.65 116 1.00 464 1.06 0.96–1.16
Lip 34 1.14 0.72–1.81 44 0.99 0.64–1.53 38 1.00 116 1.06 0.88–1.27
Skin, BCCb 1,000 1.13 1.03–1.24 1,566 1.13 1.04–1.23 953 1.00 3,519 1.16 1.13–1.20
Ovary 104 1.13 0.85–1.49 139 0.94 0.73–1.23 98 1.00 341 1.02 0.91–1.13
Breast 554 1.12 0.99–1.26 915 1.15 1.03–1.28 512 1.00 1,981 0.97 0.93–1.01
Nervous system 133 1.12 0.87–1.43 232 1.22 0.98–1.53 124 1.00 489 1.00 0.91–1.09
Rectum, rectosigmoid 162 1.09 0.88–1.36 272 1.16 0.95–1.41 156 1.00 590 1.05 0.97–1.14
Stomach 220 1.08 0.90–1.31 352 1.12 0.95–1.33 224 1.00 796 0.97 0.90–1.03
Skin, nonmelanoma 165 1.08 0.87–1.34 241 1.05 0.87–1.28 176 1.00 582 1.21 1.11–1.31
Multiple myeloma 45 1.08 0.72–1.63 76 1.18 0.81–1.70 47 1.00 168 0.88 0.76–1.03
Bladder, ureter, urethra 143 1.05 0.83–1.32 241 1.14 0.93–1.40 153 1.00 537 0.97 0.89–1.06
Non-Hodgkin lymphoma 102 1.03 0.78–1.35 164 1.04 0.81–1.33 104 1.00 370 1.02 0.92–1.13
Prostate 399 1.02 0.89–1.17 587 0.98 0.87–1.11 444 1.00 1,430 0.97 0.92–1.02
Lung, trachea 396 1.00 0.87–1.14 709 1.14 1.01–1.28 433 1.00 1,538 0.90 0.86–0.95
Liver 51 1.00 0.68–1.49 84 1.05 0.74–1.50 50 1.00 185 1.01 0.87–1.17
Kidney 144 0.98 0.78–1.23 231 0.99 0.81–1.22 159 1.00 534 1.06 0.97–1.15
Thyroid gland 45 0.97 0.64–1.45 75 1.00 0.69–1.44 49 1.00 169 0.76 0.65–0.88
Soft tissues 25 0.97 0.56–1.67 49 1.23 0.77–1.98 28 1.00 102 1.07 0.87–1.30
Skin, melanoma 103 0.93 0.72–1.22 204 1.17 0.93–1.47 119 1.00 426 1.12 1.01–1.23
Pancreas 145 0.91 0.73–1.14 257 1.05 0.87–1.28 171 1.00 573 1.07 0.99–1.16
Testis 14 0.91 0.44–1.85 27 1.10 0.59–2.04 17 1.00 58 1.16 0.88–1.49
Colon 221 0.88 0.73–1.05 390 0.99 0.84–1.16 269 1.00 880 1.05 0.98–1.12
Leukemia 72 0.84 0.62–1.15 141 1.06 0.82–1.38 96 1.00 309 0.96 0.86–1.08
Hodgkin disease 14 0.65 0.33–1.26 38 1.11 0.66–1.97 24 1.00 76 0.87 0.68–1.08
Obs, observed number of cases.
aAll people in Finland were used as reference for SIRs.
bBCCs of the skin are not included in the total numbers.
Table 4 Risk for total cancer between 1981 and 2000 among farmers living < 20.0 km from the River Kymijoki in 1980.
Distance to the River Kymijoki (km)
< 1.0
1.0–4.9
5.0–19.9
All farmers < 20.0
Variable Obs RR 95% CI Obs RR 95% CI Obs RR Obs SIRa 95% CI
All sites 209 1.13 0.97–1.32 230 0.99 0.85–1.15 704 1.00 1,143 0.96 0.91–1.02
Sex
Men 112 1.10 0.89–1.36 123 0.95 0.77–1.16 404 1.00 639 0.92 0.85–1.00
Women 97 1.16 0.92–1.46 107 1.04 0.83–1.29 300 1.00 504 1.01 0.93–1.11
Age (years)
< 45 27 1.54 0.98–2.42 24 1.09 0.69–1.74 70 1.00 121 0.97 0.80–1.15
45–59 68 1.08 0.82–1.42 78 0.99 0.77–1.28 246 1.00 392 0.97 0.88–1.07
≥60 114 1.08 0.87–1.33 128 0.97 0.79–1.18 388 1.00 630 0.95 0.88–1.03
Time period
1981–1990 100 1.15 0.92–1.45 115 1.08 0.87–1.34 322 1.00 537 0.96 0.88–1.05
1991–2000 109 1.11 0.89–1.38 115 0.91 0.74–1.13 382 1.00 606 0.96 0.88–1.04
Distance to the Gulf of Finland (km)
< 20.0 112 1.11 0.89–1.38 113 1.16 0.93–1.44 279 1.00 504 1.01 0.92–1.10
20.0–39.9 67 1.11 0.85–1.46 81 0.90 0.70–1.16 244 1.00 392 0.99 0.90–1.10
40.0–59.9 30 1.33 0.91–1.96 36 0.80 0.56–1.14 181 1.00 247 0.83 0.73–0.94
Obs, observed number of cases.
aAll farmers in Finland were used as reference for SIRs.
Table 5 Risks for cancer subtypes in 1981–2000 among farmers living < 20.0 km from the River Kymijoki in 1980.
Distance to the River Kymijoki (km)
< 1.0
1.0–4.9
5.0–19.9
All farmers < 20.0
Primary site Obs RR 95% CI Obs RR 95% CI Obs RR Obs SIRa 95% CI
Thyroid gland 3 2.29 0.54–9.69 3 1.70 0.41–7.14 5 1.00 11 0.79 0.39–1.41
Hodgkin disease 2 2.20 0.36–13.42 2 2.02 0.34–12.13 3 1.00 7 1.06 0.42–2.18
Cervix uteri 1 2.04 0.18–22.72 2 2.68 0.38–19.08 2 1.00 5 0.75 0.24–1.76
Ovary 9 1.83 0.81–4.12 2 0.32 0.07–1.39 18 1.00 29 1.24 0.83–1.78
Skin, nonmelanoma 15 1.72 0.92–3.19 5 0.47 0.18–1.21 32 1.00 52 1.04 0.78–1.37
Gallbladder, bile ducts 3 1.72 0.43–6.97 3 1.46 0.36–5.86 6 1.00 12 0.72 0.37–1.26
Rectum, rectosigmoid 13 1.65 0.86–3.18 7 0.69 0.30–1.56 31 1.00 51 1.07 0.80–1.41
Breast 23 1.54 0.94–2.52 24 1.28 0.79–2.07 55 1.00 102 0.92 0.75–1.11
Nervous system 8 1.36 0.60–3.07 11 1.48 0.72–3.06 22 1.00 41 1.29 0.92–1.75
Corpus uteri 8 1.27 0.56–2.87 8 1.01 0.45–2.26 23 1.00 39 1.17 0.83–1.60
Skin, BCCb 56 1.26 0.93–1.72 43 0.80 0.57–1.12 160 1.00 259 1.07 0.94–1.21
Stomach 12 1.22 0.63–2.35 17 1.37 0.77–2.43 38 1.00 67 0.79 0.61–1.00
Lung, trachea 19 1.13 0.68–1.89 26 1.18 0.75–1.85 70 1.00 115 0.78 0.64–0.93
Pancreas 9 1.07 0.51–2.27 11 1.05 0.53–2.10 31 1.00 51 1.14 0.85–1.50
Liver 4 1.01 0.33–3.08 6 1.19 0.46–3.08 15 1.00 25 2.17 1.41–3.21
Prostate 27 0.99 0.65–1.52 32 0.92 0.62–1.37 107 1.00 166 0.99 0.84–1.15
Bladder, ureter, urethra 7 0.91 0.39–2.08 6 0.63 0.26–1.52 30 1.00 43 0.86 0.62–1.16
Kidney 7 0.87 0.38–1.98 13 1.31 0.68–2.52 30 1.00 50 1.24 0.92–1.63
Leukemia 5 0.87 0.33–2.34 9 1.33 0.61–2.91 21 1.00 35 1.19 0.83–1.66
Non-Hodgkin lymphoma 4 0.70 0.24–2.04 3 0.40 0.12–1.35 22 1.00 29 0.99 0.66–1.42
Colon 5 0.45 0.18–1.15 10 0.73 0.36–1.46 40 1.00 55 0.88 0.66–1.15
Skin, melanoma 2 0.39 0.09–1.96 5 0.78 0.29–2.10 19 1.00 26 0.88 0.58–1.29
Multiple myeloma 1 0.39 0.05–3.10 3 1.09 0.29–4.11 8 1.00 12 0.66 0.34–1.16
Soft tissues 3 NC NC 0 NC NC 7 NC 10 1.29 0.62–2.37
Testis 1 NC NC 0 NC NC 2 NC 3 1.09 0.22–3.18
Larynx, epiglottis 2 NC NC 0 NC NC 4 NC 6 0.80 0.29–1.73
Lip 3 NC NC 4 NC NC 4 NC 11 0.59 0.59–1.06
Abbreviations: —, models for these cancer subtypes did not converge; NC, models for these cancer subtypes did not converge; Obs, observed number of cases.
aAll farmers in Finland were used as reference for SIRs.
bBCCs of the skin are not included in total numbers.
==== Refs
References
Aylin P Maheswaran R Wakefield J Cockings S Jarup L Arnold R 1999 A national facility for small area disease mapping and rapid initial assessment of apparent disease clusters around a point source: the UK Small Area Health Statistics Unit J Public Health Med 21 289 298 10528956
Baron JA Rohan TE 1996. Tobacco. In: Cancer Epidemiology and Prevention (Schottenfeld D, Fraumeni JFJ, eds). Oxford:Oxford University Press, 269–289.
Bertazzi PA Consonni D Bachetti S Rubagotti M Baccarelli A Zocchetti C 2001 Health effects of dioxin exposure: a 20-year mortality study Am J Epidemiol 153 1031 1044 11390319
Best N 1999. Bayesian ecological modelling. In: Disease Mapping and Risk Assessment for Public Health (Lawson A, Biggeri A, Böhning D, Lesaffre E, Viel J-F, Bertollini R, eds). Chichester, UK:John Wiley & Sons, Ltd, 193–201.
Boice JDJ Land CE Preston DL 1996. Ionizing radiation. In: Cancer Epidemiology and Prevention (Schottenfeld D, Fraumeni JFJ, eds). Oxford:Oxford University Press, 319–354.
Cohen MD Draxler RR Artz R Commoner B Bartlett P Cooney P 2002 Modeling the atmospheric transport and deposition of PCDD/F to the Great Lakes Environ Sci Technol 36 4831 4845 12487307
Dickinson HO Hammal DM Dummer TJB Parker L Bithell JF 2003 Childhood leukaemia and non-Hodgkin’s lymphoma in relation to proximity to railways Br J Cancer 88 695 698 12618876
Feychting M Ahlbom A 1993 Magnetic fields and cancer in children residing near Swedish high-voltage power lines Am J Epidemiol 138 467 481 8213751
Finnish Cancer Registry 2003. Cancer Statistics for Finland. Helsinki:National Research and Development Centre for Welfare and Health. Available: http://www.cancerregistry.fi/ [accessed 23 July 2003].
Finnish Environment Institute 2003. Quality of Finnish Surface Waters in 1994–1997. Helsinki:Finnish Environment Institute. Available: http://www.vyh.fi/eng/environ/state/waterquo/map9497.htm [accessed 5 August 2003].
Floret N Mauny F Challier B Arveux P Cahn JY Viel JF 2003 Dioxin emissions from a solid waste incinerator and risk of non-Hodgkin lymphoma Epidemiology 14 392 398 12843761
Harrison RM Leung PL Somervaille L Smith R Gilman E 1999 Analysis of incidence of childhood cancer in the West Midlands of the United Kingdom in relation to proximity to main roads and petrol stations Occup Environ Med 56 774 780 10658564
IARC 1997 Polychlorinated Dibenzo-para dioxins and Polychlorinated Dibenzofurans IARC Monogr Eval Carcinog Risks Hum 69 1 666 9379504
IARC 2003. IARC Cancer Databases and Other Resources. Lyon:International Agency for Research on Cancer. Available: http://www.iarc.fr/ [accessed 1 August 2003].
Isosaari P Kankaanpaa H Mattila J Kiviranta H Verta M Salo S 2002 Spatial distribution and temporal accumulation of polychlorinated dibenzo-p -dioxins, dibenzofurans, and biphenyls in the Gulf of Finland Environ Sci Technol 36 2560 2565 12099450
Kiviranta H Vartiainen T Tuomisto J 2002 Polychlorinated dibenzo-p -dioxins, dibenzofurans, and biphenyls in fishermen in Finland Environ Health Perspect 110 355 361 11940453
Kiviranta H Verta M Korhonen M Vartiainen T 1999 Human exposure to Ky-5 originating PCDD/Fs in Kymijoki area, Finland Organohal Comp 44 169 172
Kogevinas M 2000 Studies of cancer in humans Food Addit Contam 17 317 324 10912245
Kogevinas M 2001 Human health effects of dioxins: cancer, reproductive and endocrine system effects Hum Reprod Update 7 331 339 11392380
Kogevinas M Becher H Benn T Bertazzi PA Boffetta P Bueno-de-Mesquita HB 1997 Cancer mortality in workers exposed to phenoxy herbicides, chlorophenols, and dioxins. An expanded and updated international cohort study Am J Epidemiol 145 1061 1075 9199536
Koivusalo M Pukkala E Vartiainen T Jaakkola JJ Hakulinen T 1997 Drinking water chlorination and cancer—a historical cohort study in Finland Cancer Causes Control 8 192 200 9134243
Kokki E Pukkala E Verkasalo PK Pekkanen J 2002. Small area statistics on health (SMASH): a system for rapid investigations of cancer in Finland. In: GIS for Emergency Preparedness and Health Risk Reduction (Briggs DJ, Forer P, Järup L, Stern R, eds). Dordrecht, the Netherlands:Kluwer Academic Publishers, 255–266.
Kokki E Ranta J Penttinen A Pukkala E Pekkanen J 2001 Small area estimation of incidence of cancer around a known source of exposure with fine resolution data Occup Environ Med 58 315 320 11303080
Korhonen M Verta M Lehtoranta J Kiviranta H Vartiainen T 2001 Concentrations of polychlorinated dibenzo-p -dioxins and furans in fish downstream from a Ky-5 manufacturing Chemosphere 43 587 593 11372842
National Cancer Institute 2003. Geographic Information System for Health (GIS-H). A New Tool for Research on Breast Cancer on Long Island. Bethesda, MD:National Cancer Institute. Available: http://www.healthgis-li.com/ [accessed 5 August 2003].
National Center for Health Statistics 2003. GIS and Public Health. Hyattsville, MD:National Center for Health Statistics. Available: http://www.cdc.gov/nchs/gis.htm [accessed 5 August 2003].
Paasivirta J 1996. Organoklooriyhdisteet Kymijoella 1990-luvulla [in Finnish]. Helsinki:Suomen ympäristökeskus. Suomen ympäristökeskuksen monistesarja 18.
Pekkanen J Pukkala E Vahteristo M Vartiainen T 1995 Cancer incidence around an oil refinery as an example of a small area study based on map coordinates Environ Res 71 128 134 8977621
Statistics Finland 1994. Evaluation study of census 1990. In: Population Census, Vol 9. Helsinki:Statistics Finland, 98–100.
Stellman JM Stellman SD Weber T Tomasallo C Stellman AB Christian R Jr 2003 A geographic information system for characterizing exposure to Agent Orange and other herbicides in Vietnam Environ Health Perspect 111 321 328 12611661
Svensson BG Mikoczy Z Stromberg U Hagmar L 1995 Mortality and cancer incidence among Swedish fishermen with a high dietary intake of persistent organochlorine compounds Scand J Work Environ Health 21 106 115 7618056
Teppo L Pukkala E Lehtonen M 1994 Data quality and quality control of a population-based cancer registry. Experience in Finland Acta Oncol 33 365 369 8018367
Verkasalo PK Pukkala E Hongisto MY Valjus JE Jarvinen PJ Heikkila KV 1993 Risk of cancer in Finnish children living close to power lines Br Med J 307 895 899 8241851
Verta M Ahtiainen H Hämäläinen H Jussila H Järvinen O Kiviranta H 1999 Organoyhdisteet ja raskasmetallit Kymijoen sedimentissä: esiintyminen, kulkeutuminen, vaikutukset ja terveysriskit [in Finnish] Suomen ympäristökeskuksen julkaisu 334 12 33
Viel JF Arveux P Baverel J Cahn JY 2000 Soft-tissue sarcoma and non-Hodgkin’s lymphoma clusters around a municipal solid waste incinerator with high dioxin emission levels Am J Epidemiol 152 13 19 10901325
Warner M Eskenazi B Mocarelli P Gerthoux PM Samuels S Needham L 2002 Serum dioxin concentrations and breast cancer risk in the Seveso Women’s Health Study Environ Health Perspect 110 625 628 12117637
|
15198923
|
PMC1247196
|
CC0
|
2021-01-04 23:36:23
|
no
|
Environ Health Perspect. 2004 Jun 15; 112(9):1026-1031
|
utf-8
|
Environ Health Perspect
| 2,004 |
10.1289/ehp.6741
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0112-00103210.1289/ehp.673915198924Mini-Monograph: Information SystemsUse of GIS and Exposure Modeling as Tools in a Study of Cancer Incidence in a Population Exposed to Airborne Dioxin Poulstrup A. Hansen H.L. Regional Public Health Office, Vejle, DenmarkAddress correspondence to A. Poulstrup, Vedelsgade 17A, DK 7100 Vejle, Denmark. Telephone: 45 75 82 37 99. Fax: 45 75 72 35 64. E-mail:
[email protected] study was supported by grants from the Danish Ministry of Internal Affairs and Health, Environmental Cancer Research Programme (383-38-2001); the European Commission, Health and Consumer Protection Directorate-General [SI2.329122 (2001CVG2-604)]; and the National Survey and Cadastre, Denmark. The study obtained approval in all aspects of the research from the Danish Data Protection Agency.
The authors declare they have no competing financial interests.
6 2004 15 4 2004 112 9 1032 1036 15 9 2003 2 2 2004 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. In environmental health research there is a recognized need to develop improved epidemiologic and statistical methods for rapid assessment of relationships between environment and health. Exposure assessment is identified as a major challenge needing attention. In this study an exposure simulation model was used to delimit almost exactly in space and time an urban population exposed to airborne dioxin. A geographic information system (GIS) was used as the electronic environment in which to link the exposure model with the demographic, migration, and cancer data of the exposed population. This information is available in Denmark on an individual basis. Standardized incidence ratios (SIRs) for both men and women in 10-year age bands were calculated for three different exposure areas. Migration patterns were outlined. SIRs showed no excess of cancer incidences during the time span chosen (13 years; 1986–1998) in the whole exposed area or in the medium or higher polluted areas. The exposure model appeared very useful in selection of the appropriate exposure areas. The integration of the model in a GIS together with individual data on addresses, sex, age, migration, and information from routine health statistics (Danish Cancer Registry) proved its usefulness in demarking the exposed population and identifying the cancers related to that population. Less than one-third of the study population lived at the same address after 13 years of observation, and only half were still residing in the area, indicating migration of people as a major misclassification.
air pollutioncancerdioxinenvironmental epidemiologyexposure modelGIShealth registersmigration
==== Body
A major challenge for investigators in environmental epidemiology is to correctly identify populations at risk from exposure to environmental contaminants. To date, three methods have been used to identify the populations at risk from point sources of air pollution: physical monitoring, environmental monitoring, or mathematical modeling (Williams and Ogston 2002). This article is a discussion of the utilization of a computerized air pollution model, normally used by the environmental protection authorities for assessing pollution values (immissions), and the putative offense of legally set thresholds of emission. To test the model for its appropriateness as an improved tool for assessment of exposure, an actual case was used of known dioxin air pollution in an urban area.
In the town of Kolding in the southern part of the Jutland peninsula, Denmark, three outlets of dioxin were identified. All three emitted dioxin into the air through their chimneys. The dioxin consisted mainly of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD, or dioxin). One outlet in particular, an aluminum recycling plant, was found on two occasions—early November and early December 2000—to have emitted large quantities of dioxins, up to 180 ng/m3/hr. All three plants had been operating for years. The main culprit was the aluminum recycling plant, which had been in continuous operation since 1970 with an almost unchanged method of production and output.
Our plan was to layer the computer-simulated exposure model in a geographic information system (GIS) and use the simulated immission concentrations to more accurately demarcate the exposed population. The information on addresses, vital statistics, migration, and cancer of the population of Denmark or any subset was available on individuals and on the delineated population in this study. This information was layered into the same GIS environment, enabling a more exact identification of the exposed population in both space and time.
All malignant cancers were used as the health indicators of the exposed population to assess eventual negative health outcomes caused by the dioxin pollution. TCDD is a major environmental carcinogen causing various types of cancers (IARC 1997).
Materials and Methods
The air pollution simulation model used in Denmark to assess hourly immissions of airborne pollutants is a Gaussian air dispersion model based on emission data of the actual pollutant(s) and time series of meteorological data such as wind speed, wind direction, wind temperature, rain, snow, number of stacks, their heights, surrounding buildings, and surrounding terrain. The model [OML, Operationel Meteorologiske Luftkvalitets-modeller (Danish)] has been widely validated both in Europe and North America and is reliable in predicting hourly immissions of one or more airborne pollutants (DMU 2001). Only two measurements of dioxin were available as input source, and both were obtained in November and December 2000. The GIS used was the software package ArcGis version 8.12 (ESRI, Atlanta, GA, USA).
All current and past addresses in Denmark since 1999 were geocoded with Universal Transverse Mercator coordinates with a precision of a few meters and subsequently layered into the GIS. The Address Project is described elsewhere (Briggs et al. 2002). By linkage of all individuals to these addresses using the unique Central Population Register (CPR) number (10-digit number in the Civil Registration System), the GIS eventually contained all the following information in addition to the addresses of each individual: date of birth, sex, migration (into, out of, and around the study area), and date of death (Figure 1). Each green spot represents an address and a table of the mentioned attributes.
The CPR contains data on more than 7 million people who are or have been residents in Denmark since 1968. The key to the register is the personal identification number, the CPR number, which is a unique 10-digit number that all residents in Denmark are assigned at birth or when immigrating. In addition to the present address of all residents, the CPR also contains historical addresses and the dates on which the individual moved to and from that address. If a person dies, disappears, or takes up residence abroad, this is also recorded as having moved away from the address. The CPR, its structure, updating, and other details are described elsewhere (Briggs et al. 2002).
All health registers in Denmark use the CPR number as entry key, which makes it easy to merge health data into a GIS where CPR has been incorporated. In this study all malignant cancer (except skin cancer) was used as the health outcome indicator. By merging the Danish Cancer Register with CPR data, the necessary cancer incidence information was retrieved. Details on Danish health registers have been described elsewhere (Briggs et al. 2002), particularly in the Danish Cancer Register (Storm 1991).
In this study, 1986–1998 was the time span chosen for analyses. The first year, 1986, was chosen as a compromise between the arduous work of geocoding historical addresses and the cost of this operation versus having a sufficient number of years with cancer data for analyses. The main dioxin producer, the aluminum recycling plant, became operational in 1970, which was sufficient time from start of production (and pollution) to account for the induction and latency time of developing (eventual) cancer in the surrounding population. The end of study year, 1998, was chosen because that was the last year with obtainable cancer data at the time a request was sent to the Danish Cancer Register.
In this study the exposure simulation model, OML, was used to demarcate three zones relevant for studying cancer development related to the dioxin exposure:
Zone 1 encompassed the whole residential area identified to be exposed to dioxin.
Zone 2 included the area identified to be exposed to 3.5 ng dioxin/m3/hr or higher. Zone 2 is part of zone 1.
Zone 3 included the inner and highest exposed area with estimated dioxin immissions of 4.5 ng dioxin/m3/hr or higher. Zone 3 is part of zones 2 and 3.
These zones and the selection are illustrated in Figure 2, where the immission concentration band borders are used to demarcate the three zones. In Figure 3 the malignant cancers have been linked and overlayered, appearing as yellow dots. For confidentiality reasons, any single, outstanding cancer case has been obscured so that only clusters of cancers (aggregated over 13 years) are visible.
The following criteria were used to select the study population:
Individuals were included if they lived in the area between 1986 and 1998. Individuals were included in a calendar year if they moved into the area on or after 1 January in the same year.
Cancer cases were included if the year of diagnosis was in or later than the year the individual moved into the area (i.e., cancer cases with year of diagnosis before migration into the area were excluded).
Skin cancer diagnoses (ICD-7 code 191; National Board of Health 2003) were excluded because they comprise a considerable number of rather harmless cancers.
The reference population chosen was the Danish population in the same period, 1986–1998. It was desirable but not financially possible to obtain a reference area similar to the study area.
Study Population
At the start of follow-up, 1 January 1986, 15,404 individuals resided in the study area. During the next 13 years, between 2,069 and 3,470 individuals moved into the area each year; a total of 46,392 different individuals resided in the area during the 13-year follow-up period. During the study period the population gradually increased to 20,217 individuals by the end of 1998. Among the 46,392 persons who lived in the study area from 1986 to 1998, 3,205 individuals were newborns who had their first-ever address in the area.
Among the 15,404 individuals residing in the study area 1 January 1986, 7,758 (50.4%) were still living in the study area at the end of 1998, and 4,799 (31.2%) had not changed their address. Figure 4 illustrates this development. Among those who were < 10 years of age on 1 January 1986, 57% were still residing in the area 13 years later, whereas only 30% of those 10–20 years of age still lived in the area at the end of 1998. This figure was 43% for the group 21–30 years of age; 65% for the group 31–40 years of age, and increased to 71% for the group 41–50 years of age. The same figure gradually decreased for the group 51–60 years of age to 65% for those who remained in the area until the end of the study period, and rapidly decreased for older age groups.
The 46,392 individuals who lived in the study area had a total of 75,437 periods of residence. Among the same 46,392 individuals, 61.5% had one address, 21.5% had two different addresses, and 17% had three or more addresses within the study area during 1986–1998.
On 1 January 1986, 50.3% of the females and 55.8% of the male residents in the study area were < 40 years of age. The corresponding figures for the whole population of Denmark in 1986 were 54.0% for male residents and 58.3% for females.
Statistical Methods
For each year from 1986 through 1998, information on the number of eligible residents and cancers from the study area (zones 1–3) were retrieved through the GIS model and cumulated into nine 10-year age bands (0–9, 10–19, 20–29 . . . ≥80) stratified by sex.
The Danish population during the same period was used as the reference population. Number of cancers was derived from the Danish Cancer Register and population data were from the Bureau of Statistics (Danmarks Statistik). These data were similarly grouped into cumulated 10-year age bands stratified by sex. Years of risk were calculated using the number of residents in each calendar year in the study area (“living in and not moved out”) and in the general population, respectively. Each resident in a given year was counted as 1 year of exposure. The number of expected cases of cancer was calculated based on the total number of person-years for each 10-year age category multiplied by the cancer rate of the Danish men and women, respectively, during the same period. The standardized incidence ratio (SIR)—the ratio between observed and expected numbers—was calculated with 95% confidence limits (95% CL) assuming a Poisson distribution of the cancer cases. Normal distribution for observed cancers was assumed when figures were above 100.
Results
The method of using an air pollution simulation model to identify exposure and exposed population was operational, and the subsequent incorporation into a GIS environment integrating individual statistics of address, vital statistics, and cancer created no severe technical problems.
Results of the statistical analyses are presented in Table 1. Only a single age band in zone 1 had confidence limits above 1.0. No excess of cancer in the study area during 1986–1998 could be demonstrated. The study population was anticipated to be geographically stable, but this appeared not to be true, with only one-third of the original residents still living in the area at the end of the study period.
Discussion
The OML, a commercial product (Danish Environmental Protection Agency 1997), is used widely by environmental regulatory bodies in Denmark to assess immission values of airborne pollutants. This product proved useful to visualize exposure in a GIS milieu to outline the research area. The incorporation of the model in GIS presented no serious technical problems.
However, the OML output, like with most models, is no better than the quality of the input data, and in this case only two dioxin measurements from the chimney smoke were available. In 2000, when the environmental authorities discovered grossly excessive emissions (180 ng dioxin/m3/hr) with a legal threshold of 1 ng dioxin/m3/hr, the aluminum recycling plant immediately started injecting active carbon and chalk into the smoke-cooling process, hence reducing the content of dioxin to far below thresholds. The emission of dioxin has been reduced further since then. The official threshold was lowered in 2001 to 0.1 ng dioxin/m3/hr, following EU regulations. So the first and only data available were two measurements in autumn 2000, which do not allow for extensive conclusions on the amount of airborne dioxin dispersed to the adjacent surroundings.
Airborne dioxin alone is adsorbed onto plants, trees, vegetables, and soil but is easily washed away by rain. A soil examination in the exposed area in the summer 2001 produced no evidence of a major contamination of the area (Vejle Amt 2001).
A major study on environmental and hereditably caused cancers (Lichtenstein et al. 2000) concluded that genetic factors make only a minor contribution to development of sporadic cancer, with environmental factors being the major contributor.
Airborne dioxin is presumably absorbed in the lungs, making up 75% of the total content. European average dioxin concentrations range between 0.01 and 0.4 pg/m3, which translated into a Danish situation for an average adult is an intake via the lungs of 0.2–8 pg dioxin/day. Dioxin via the airways is not the only entrance into the body; intake via food is assumed to constitute as much as 15 pg a day (2.44 pg/kg body weight (bw)/day; 70 kg) (Danish Environmental Protection Agency 1997).
If people in Kolding have had concentrations of airborne dioxin in their ambient environment for many years in the range of the measured values of 100–200 ng dioxin/m3/hr that produce inhalation concentrations in the range of 1–6 pg dioxin/m3/hr, then using the above estimate would have caused a daily extra intake of 20 pg dioxin in the least polluted area and up to 120 pg in the highest polluted area (zone 3).
Tolerable daily intake is 5 pg/kg bw in Denmark (Danish Environmental Protection Agency 1997). In the Netherlands authorities recommend figures be lowered to 1 pg/kg bw (Health Council of the Netherlands 1996). An extra intake of up to 120 pg dioxin/day for an adult would entail a substantial extra burden for the body of a well-known carcinogen.
In the Kolding case no one knows whether actual emissions over the years have been even higher (or lower) than the measured values, meaning that the measured values could just as well have been in the lower range of the actual pollution of dioxin. However, the information above on dioxin in soil in the exposure area (Vejle Amt 2001), together with the fact that no excess cancers were detected in any of the years under study, in any of the zones, in any age group or any sex group, indicate that no major pollution of the study area with airborne dioxin has taken place over the years. As the peak dioxin values were detected in late 2000, any later consequence on cancer development will not be detectable until later. The relevant authorities have decided to continuously scrutinize the cancer data of the area in years to come.
Four years of latency has been chosen as a very conservative restriction to allow for any early effect. Most likely the latency, at least for adults, is much longer.
A planned follow-up of the present study is a search in the Danish Cancer Register for cancers diagnosed outside the study area among previous residents of the study area.
The tool we have developed has its limitations. Most environmental exposures in a modern industrial society stem from food or are widely present in the environment, for example, exhaust from vehicles. Fewer are present in well-defined geographical areas, and few are strong enough to have any significant impact on health. These factors limit the opportunity to investigate environmental health relationships using spatial analytical methods, and inhibit the types of problems that can be addressed.
The Address Project offers new and unique possibilities for performing studies of relationships between environmental exposures and health of the population in Denmark. These studies might be based on a range of different study designs (Aylin et al. 1999; Elliott et al. 1992). Because of the ability to track individuals over time, retrospective, space–time studies are possible. In each case, the detailed address-based data now available and the ability to link data files are likely to enhance these studies.
In this study we decided to use the knowledge of the migration of the population to apply two restrictions: to include only individuals who had actually stayed in the area and to include only the cancer cases that were diagnosed after the individual had moved into (or after their birth in) the area.
Further restrictions could have been implemented, but this would have implicated the choice of a reference area with a population where the same restrictions could be made. The actual restrictions that were applied based on the individual migration data available are an improvement in dealing with this misclassification problem in epidemiologic studies and indicate the vast opportunities in the system.
There are, however, several important limitations on what can be expected even from a fully developed system, and several issues concerning choice of study design need to be carefully considered.
First, it is important to base such studies on defined and plausible hypotheses about the relationships being examined. Possible exposure pathways also need to be identified. Without these preconditions, results are likely to be difficult to interpret, at best, or are even misleading.
Second, by considering environment, one is concerned with more than just the soil on which we walk, the water we drink, and the air we inhale. Environment is also what we eat and wear, what we smoke, where we work and relax, and from a biological point of view, it is more likely that the causes of death and diseases may be found here rather than geographically varying pollution in the ambient environment.
In addition to these theoretical considerations, a number of other limitations must be recognized. For example, most geographically based studies assume that populations are static and that exposures occur in fixed locations (usually the place of residence). Obviously, this is not true. People are highly mobile, both in terms of short-term activities (e.g., daily travel to work) and long-term migration. Rates of migration in a population may be high. Therefore, knowledge about where people work or have worked is essential if all misclassification is to be ruled out.
The analysis of the stability of the population in the study area disclosed a high mobility. Less than one-third lived at the same address after 13 years of observation, and only half were still residents in the study area. A high degree of mobility within the study area was also found. The chosen study area is an ordinary mixed residential and industrial suburb, and the observed mobility of the population is likely to be representative of similar areas in Denmark. In ecologic studies, information on exposure and the exposed individuals is of vital importance. Such studies will therefore be highly susceptible to the fact that only a relatively small proportion of the study population remains in the area during a prolonged exposure in the local environment. A further improvement in exposure assessment would be to measure actual at-risk time for each individual. This was not done in this study although it is possible within the model and with the Danish data sets. We hope to perform such a study in the future.
In an extensive analysis of geographic exposure modeling and its usefulness in environmental epidemiology (Beyea and Hatch 1999), the authors emphasized the importance of considering all uncertainty aspects when making the models: type and quantity of pollutants, their pathways into surroundings, exposed population, and time of pollution.
The tested GIS with linkage of addresses and individual health information gives new opportunities for high-quality, small-area health studies in a wide range of situations. When fully developed and covering the whole of Denmark, it will create a useful tool both for administrators, planners, and public health offices as well as researchers. As with all such systems, however, it is crucial to recognize the limitations of the system and to apply it only where appropriate. The geographical stability of the study population is especially crucial to address, describe, and include in the exposure assessment. Otherwise, this misclassification may totally distort the true picture.
This article is part of the mini-monograph “Health and Environment Information Systems for Exposure and Disease Mapping, and Risk Assessment.”
Figure 1 Map of Kolding Town with dioxin source in red and address points in green .
Figure 2 Computer-simulated exposure of dioxin from three sources (red) are layered onto the electronic map (GIS) and seen as different colored bands, with highest dioxin immissions in bright red and lowest in faint green. The immission concentration band borders (blue) are used to demarcate the three zones used for analyses of cancer development.
Figure 3 Demarcation of zone 3 and the addresses (and individuals) in blue within the polygon. Cancers diagnosed among the individuals in zone 3 during 1986–1998 are marked in yellow (overlayed on the blue dots).
Figure 4 Migration of residents living in the study area in January 1986 from the area from 1986 to the end of 1998.
Table 1 Cumulated cancer incidence data from 1986 to 1998 for three zones.
Age group (years) Men (N)a Men (n)b Expected (n)c RR 95% CL Women (N)a Women (n)b Expected (n)c RR 95% CL
Zone 1: whole study area (estimated exposure > 0 pg dioxin/m3/hr)
0–9 18.420 4 3.21 1.25 0.34, 3.19 16.871 3 2.57 1.17 0.24, 3.41
10–19 18.609 2 3.46 0.58 0.07, 2.09 18.990 7 2.64 2.65 1.07, 5.47
20–29 33.031 12 14.78 0.81 0.42, 1.42 33.448 11 13.98 0.79 0.39, 1.41
30–39 23.072 12 17.74 0.68 0.35, 1.18 21.994 18 27.14 0.66 0.39, 1.05
40–49 18.870 32 33.02 0.97 0.66, 1.37 19.357 59 66.47 0.89 0.68, 1.15
50–59 14.161 54 70.45 0.77 0.58, 1.00 15.113 106 97.75 1.08 0.71, 1.55
60–69 10.395 143 132.29 1.08 0.91, 1.27 12.930 138 148.67 0.93 0.64, 1.27
70–79 7.647 156 179.30 0.87 0.74, 1.02 11.560 176 171.68 1.03 0.74, 1.36
80+ 3.421 65 95.76 0.68 0.52, 0.87 7.178 103 115.11 0.89 0.58, 1.28
Total 147.626 480 550.01 0.87 0.80, 0.95 157.441 621 646.01 0.82 0.82, 1.12
Zone 2 (estimated exposure > 3.5 pg dioxin/m3/hr)
0–9 4.913 0 0.86 0.00 0.00, 4.31 4.485 2 0.68 2.92 0.35, 10.56
10–19 4.945 0 0.92 0.00 0.00, 4.01 4.723 2 0.66 3.05 0.37, 11.01
20–29 5.286 4 2.37 1.69 0.46, 4.33 5.387 3 2.25 1.33 0.27, 3.89
30–39 4.678 5 3.60 1.39 0.45, 3.24 5.335 2 6.58 0.30 0.04, 1.10
40–49 4.366 9 7.64 1.18 0.54, 2.24 5.053 14 17.35 0.81 0.44, 1.35
50–59 3.548 15 17.65 0.85 0.48, 1.40 3.792 29 24.53 1.18 0.79, 1.70
60–69 2.565 40 32.64 1.23 0.88, 1.67 3.141 35 36.12 0.97 0.68, 1.35
70–79 1.649 45 38.66 1.16 0.85, 1.56 2.596 35 38.55 0.91 0.63, 1.26
80+ 600 13 16.79 0.77 0.41, 1.32 1.368 19 21.94 0.87 0.52, 1.35
Total 32.550 131 121.13 1.08 0.41, 1.32 35.850 141 148.66 0.95 0.66, 1.30
Zone 3 (estimated exposure > 4.5 pg dioxin/m3/hr)
0–9 1.824 0 0.32 0.00 0.00, 11.62 1.492 0 0.23 0.00 0.00, 23.28
10–19 1.746 0 0.32 0.00 0.00, 11.36 1.543 1 0.21 4.66 0.12, 25.99
20–29 1.630 1 0.73 1.37 0.03, 7.64 1.740 0 0.73 0.00 0.00, 7.28
30–39 1.703 3 1.31 2.29 0.47, 6.70 1.993 1 2.46 0.41 0.01, 2.27
40–49 1.558 4 2.73 1.47 0.40, 3.76 1.943 6 6.67 0.90 0.33, 1.96
50–59 1.498 5 7.45 0.67 0.22, 1.57 1.667 12 10.78 1.11 0.58, 1.94
60–69 1.132 15 14.41 1.04 0.58, 1.72 1.586 16 18.24 0.88 0.50, 1.42
70–79 855 19 20.05 0.95 0.57, 1.48 1.295 12 19.23 0.62 0.32, 1.09
80+ 292 4 8.17 0.49 0.13, 1.25 462 5 7.41 0.67 0.22, 1.57
Total 12.238 51 55.49 0.92 0.68, 1.21 13.721 53 65.96 0.80 0.60, 1.05
Abbreviations: 95% CL, 95% confidence limit; RR, relative risk.
aBackground population.
bNumber of cancer incidents.
cExpected numbers of cancer cases.
==== Refs
References
Aylin P Maheswaran R Wakefield J Cockings S Järup L Arnold R 1999 A national facility for small area disease mapping and rapid initial assessment of apparent disease clusters around a point source: the UK Small Area Health Statistics Unit J Public Health Med 21 289 298 10528956
Beyea J Hatch M 1999 Geographic exposure modeling: a valuable extension of geographic information systems for use in environmental epidemiology Environ Health Perspect 107 suppl 1 181 190 10229717
Briggs D Forer P Järup L Stern R eds. 2002. GIS for Emergency Preparedness and Health Risk Reduction. NATO Science Series. IV. Earth and Environmental Sciences, Vol 11. Dordrecht, the Netherlands:Kluwer Academic Publishers.
Danish Environmental Protection Agency 1997. Dioxins. Working Report no 50. Copenhagen:Danish Environmental Protection Agency.
DMU. National Environmental Research Institute 2001. Beskrivelse af OML-Modellens Versioner [in Danish]. Available [in English]: http://www.oml.dmu.dk [accessed 26 January 2004].
Elliott P Cuzick J English D Stern R (eds). 1992. Geographical and Environmental Epidemiology: Methods for Small-area Studies. Oxford:Oxford University Press.
Health Council of the Netherlands Dioxins. Publ no. 1996/10E. 1996. Rijswijk, the Netherlands:Health Council of the Netherlands.
IARC 1997 Polychlorinated Dibenzo-para -dioxins and Polychlorinated Dibenzofurans IARC Monogr Eval Carcinog Risks Hum 69 1 666 9379504
Lichtenstein P Holm NV Verkasalo PK Iliadou A Kaprio J Koskenvuo M 2000 Environmental and heritable factors in the causation of cancer N Engl J Med 343 78 85 10891514
National Board of Health 2003. Cancer Incidence in Denmark 1999. Randers, Denmark:Buchs Grafiske.
Storm HH 1991 The Danish Cancer Registry, a self-reporting national cancer registration system with elements of active data collection. In: Principles and Methods IARC Sci Publ 95 220 236 1894325
Vejle Amt 2001. Analysis of dioxin in specimens of soil and plants, collected around Gotthard Aluminum, Kolding. [Translated from Danish.] Sagsnummer 17.187. Vejle County, Denmark:Environmental Protection Department.
Williams FLR Ogston SA 2002 Identifying populations at risk from environmental contamination from point sources Occup Environ Med 59 2 8 11836461
|
15198924
|
PMC1247197
|
CC0
|
2021-01-04 23:39:44
|
no
|
Environ Health Perspect. 2004 Jun 15; 112(9):1032-1036
|
utf-8
|
Environ Health Perspect
| 2,004 |
10.1289/ehp.6739
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0112-00103710.1289/ehp.673715198925Mini-Monograph: Information SystemsSpatial Analysis of the Relationship between Mortality from Cardiovascular and Cerebrovascular Disease and Drinking Water Hardness Ferrándiz Juan 1*Abellán Juan J. 12Gómez-Rubio Virgilio 1López-Quílez Antonio 1Sanmartín Pilar 3Abellán Carlos 4Martínez-Beneito Miguel A. 4Melchor Inmaculada 4Vanaclocha Hermelinda 4Zurriaga Óscar 4Ballester Ferrán 5Gil José M. 5Pérez-Hoyos Santiago 5Ocaña Ricardo 61Departamento d’Estadística i Investigació Operativa, Universitat de València, Valencia, Spain2Instituto Valenciano de Estadística, Generalitat Valenciana, Valencia, Spain3Departamento de Matemática Aplicada y Estadística, Universidad Politécnica de Cartagena, Cartagena, Spain4Departamento de Epidemiología, Dirección General de Salud Pública, Generalitat Valenciana, Valencia, Spain5Unidad de Epidemiología y Estadística, Escuela Valenciana de Estudios para la Salud, Generalitat Valenciana, Valencia, Spain6Escuela Andaluza de Salud Pública, Consejería de Salud, Junta de Andalucía, Granada, SpainAddress correspondence to A. López-Quílez, Dept. d’Estadística i Investigació Operativa, Universitat de València, Dr. Moliner 50, Burjassot E-46100, Spain. Telephone: 34 963543792. Fax: 34 963544735. E-mail:
[email protected]*The authors express their high regard and gratitude to Juan Ferrándiz, who died during the revision process of this article. Juan was the lead investigator of the Spanish EUROHEIS group and the lead investigator for all the people involved in the project.
This research was funded partially by the Dirección General de Salud Pública, the Escuela Valenciana de Estudios para la Salud (grant IVESP99/066), and the EUROHEIS project (grant SI2.329122, 2001CVG2-604).
The authors declare they have no competing financial interests.
6 2004 15 4 2004 112 9 1037 1044 12 9 2003 31 3 2004 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Previously published scientific papers have reported a negative correlation between drinking water hardness and cardiovascular mortality. Some ecologic and case–control studies suggest the protective effect of calcium and magnesium concentration in drinking water. In this article we present an analysis of this protective relationship in 538 municipalities of Comunidad Valenciana (Spain) from 1991–1998. We used the Spanish version of the Rapid Inquiry Facility (RIF) developed under the European Environment and Health Information System (EUROHEIS) research project. The strategy of analysis used in our study conforms to the exploratory nature of the RIF that is used as a tool to obtain quick and flexible insight into epidemiologic surveillance problems. This article describes the use of the RIF to explore possible associations between disease indicators and environmental factors. We used exposure analysis to assess the effect of both protective factors—calcium and magnesium—on mortality from cerebrovascular (ICD-9 430–438) and ischemic heart (ICD-9 410–414) diseases. This study provides statistical evidence of the relationship between mortality from cardiovascular diseases and hardness of drinking water. This relationship is stronger in cerebrovascular disease than in ischemic heart disease, is more pronounced for women than for men, and is more apparent with magnesium than with calcium concentration levels. Nevertheless, the protective nature of these two factors is not clearly established. Our results suggest the possibility of protectiveness but cannot be claimed as conclusive. The weak effects of these covariates make it difficult to separate them from the influence of socioeconomic and environmental factors. We have also performed disease mapping of standardized mortality ratios to detect clusters of municipalities with high risk. Further standardization by levels of calcium and magnesium in drinking water shows changes in the maps when we remove the effect of these covariates.
environmental epidemiologygeographic information systemshierarchical spatial modelsrelative riskspatial smoothing
==== Body
Cardiovascular disease is the primary cause of mortality in developed countries, with the exception of Japan. In Comunidad Valenciana, an autonomous region in Spain with approximately 4 million inhabitants, cardiovascular disease accounts for 35% of total female mortality and 46% of total male mortality (Melchor et al. 1998). Geographic variability within Comunidad Valenciana has been documented (Ferrándiz et al. 2000, 2002; Nolasco et al. 1992).
Previously published scientific articles have reported a negative correlation between drinking water hardness and cardiovascular mortality. Some results obtained in ecologic studies (Cradford et al. 1971; Lacey and Shaper 1984; Pocock et al. 1980) suggest that high levels of drinking water hardness (i.e., high concentrations of calcium and magnesium) are protective against cardiovascular diseases, mainly against ischemic heart disease.
Several surveys based on individual cases (Hall and Jungner 1993; Van der Vijver et al. 1992) have not confirmed the protective effect of calcium. Nevertheless, Rylander et al. (1991) and Yang (1998) suggest the beneficial effect of magnesium against coronary heart disease mortality as well as against cerebrovascular mortality. The results of these studies, obtained at the aggregated level, have been partially corroborated by case–control studies (Rubenowitz et al. 1996, 2000). These case–control studies, conducted in 18 southern municipalities in Sweden, found a protective effect of magnesium against acute myocardial infarction mortality but failed to find this effect for the total incidence in men.
A recent report by Marx and Neutra (1997) on the relationship between ischemic disease and magnesium in drinking water presented an analysis of several ecologic studies showing contradictory results, perhaps because the studies were not sufficiently specific to find the associations the investigators were exploring. The authors concluded their report by recommending that further studies be conducted to evaluate the apparent benefit of drinking water with high magnesium concentration.
A key issue to be addressed is the hypothetical temporal sequence between exposure and adverse health effects. It is unclear when to measure these factors, as there is no clear latency period. Some authors (Rubenowitz et al. 1996) have indicated that 1 year is sufficient to produce observable magnesium effects. However, other authors pointed out that longer periods of observations are needed (Marx and Neutra 1997).
More recently, Ferrandiz et al. (2003) studied the relationship between cerebrovascular mortality and calcium and magnesium concentrations in drinking water in 262 municipalities of the Valencia province in Spain from 1990 to 1995. They found a decreasing temporal trend, suggesting the cumulative effect of this beneficial factor, although this assertion needs further research.
Our research extends this last study, taking advantage of the Spanish Rapid Inquiry Facility (RIF). The RIF is an analytical tool for quick assessment that is applied to the data gathered in an information system developed within the European Health and Environment Information System (EUROHEIS) project and allows exposure analysis with covariates (Gómez et al. 2002).
First, we enlarged the period studied to 1991–1998. Second, we included all municipalities of Comunidad Valenciana, not just those belonging to the province of Valencia, thus enlarging our study from 263 to 538 municipalities. This provides a wider range of values for the factors being studied. Third, we considered ischemic [ICD-9 410-414 (World Health Organization 1978)] as well as cerebrovascular [ICD-9 430–438; (World Health Organization 1978)] diseases. Finally, data on drinking water hardness have also been completed and updated as a result of efforts to build a comprehensive environmental database inside the RIF.
Material and Methods
Data
One of the primary advantages of the RIF is the comprehensive database built into it. Mortality/morbidity, demographic, socioeconomic, and environmental data are assembled in a georeferenced system, allowing geographic representations of these phenomena. All data used in our study came from this source.
When constructing and updating the RIF database, mortality counts were obtained from the mortality registry of the Dirección General de Salud Pública. These numbers correspond to residents of Comunidad Valenciana and include deaths of those residents that occurred in or out of the region.
Similarly, environmental data such as those on the quality of drinking water were obtained from the Servicio de Calidad de Aguas at the Conselleria de Medio Ambiente. This agency has been analyzing public water supplies on an annual basis since 1989, although the frequency of the measurements varies between municipalities and some data are still missing. The average number of measurements in each municipality was 5.6 from 1991 to 1998. Fitting these data into the RIF database required statistical imputation of their values. Bayesian analysis of spatiotemporal models was used to perform this task as described in Abellán et al. (2003). Nevertheless, calcium and magnesium concentrations in drinking water were stable during the study, thereby minimizing the influence of the imputation methodology on the results of the analysis. Any missing value at a location was estimated by the average of the nearest 5 years at the same location. This simple procedure was sufficient in previous exploratory analyses.
Finally, demographic and socioeconomic data were provided by the Valencian Statistical Institute, where municipal statistics are updated regularly.
Exposure Analysis
We performed exposure analysis within the RIF by defining regions (bands) composed of geographical units sharing similar levels of the risk factor under study.
These degrees of risk can be based on distance to a putative origin of risk (as in point source analysis) or on values of some environmental variables, as in our study. For each of the calcium and magnesium concentrations, we defined five bands, using as cut points the quintiles of their respective distributions on the 538 municipalities studied. To achieve uniqueness of these bands during the period studied, we used the values corresponding to 1991, the first year of our study. This choice was based, among other reasons, on a special program of water quality measurement used by the regional authorities that year. These regional authorities used a methodology common to all of the municipalities in Comunidad Valenciana.
Table 1 shows the values of calcium and magnesium defining those bands, the number of municipalities in each, and the percentage of total population of the region. The unevenness shown by these values is due to multiple ties in some of the cut points as well as to the variability of population sizes of these 538 municipalities. Populated municipalities make appreciable contributions to the band where they are allocated.
Assessing Effects of Risk Factors
We performed statistical analysis using the estimation of the relative risk of each band i by the corresponding standardized mortality ratio SMRi = Oi/Ei of observed (O) to expected (E) mortality counts. In the computation of the expected counts Ei, standardization by age groups was performed separately for each sex, as well as by levels of a deprivation index based on three municipal indicators: the ratios of unemployed individuals, the proportion of illiterates among individuals > 10 years of age, and number of vehicles per individual inhabitants (Arias et al. 1993). This deprivation index has been incorporated in the RIF as a new field attached to each municipality register. As a comparison region for each municipality, deprivation index standardization uses the band of the covariate that it belongs to and not the whole region of study. Thus, the model incorporates and controls the fact that risk could not be the same at different levels of the covariate.
In our study we performed indirect intrinsic standardization, using the population of the whole region as the reference population for each of the periods studied. The standardization procedure implicitly assumes that the expected rate in each stratum of a region is equal to the product of the relative risk of the region and a common mortality rate of this stratum. This is called the proportionality assumption (Wakefield et al. 2000), which must be checked to obtain valid conclusions. We performed a linear fit to the strata-specific rates of each municipality, and we did not observe clear departures from the linear assumption (Ferrándiz et al. 2003).
The output provided by the RIF includes the SMRs and their 95% confidence intervals in each band for every sex group.
Significant relative risks, that is, those for which confidence intervals do not include the value 1, are highlighted in the RIF output to facilitate their detection by visual inspection. Thus, we can identify those levels of the studied factors that correspond to unusually high or low relative risks.
Relying on this band-by-band inspection to identify the influence of a risk factor as significant has a statistical drawback. Because we are performing multiple tests to obtain a unified conclusion, we encounter the problem of simultaneous inference; that is, we risk identifying the global effect of an irrelevant factor as significant with a probability much higher than the nominal 5% level of each test. Thus, we have to protect against this global type I error by increasing the confidence level of our intervals or by performing a global test of homogeneity of bands before accepting any individual significant result.
This second alternative seems easier from the output of the RIF. It provides the observed Oi and expected Ei counts so that we can perform a χ2test of homogeneity of the number of (n) bands by computing the statistic
to be compared with the quantiles of the χ2 distribution with n – 1 degrees of freedom. In Equation 1, r is the ratio of total observed to total expected cases in the entire region, the maximum likelihood estimator of the common relative risk under the assumption of homogeneity of bands and Poisson-distributed counts.
Handling Multiple Covariates
Standardizing mortality/morbidity rates by levels of a covariate as we have with age groups and deprivation index is a way of filtering its influence to allow the resulting statistics to be free from its effects. The remaining variability, if any, will be due to sources other than this covariate.
Covariate analysis, an option available within the RIF environment, performs this task. Once we have stipulated the desired bands of the covariate under study, the RIF computes the relevant statistics of each band, as described in the preceding section. Then we can ask the program to build a new index with these levels to standardize rates in future studies. [See Gómez et al. (2002) for computational details.]
In each analysis we performed within the RIF, we can compare results obtained before and after standardization by levels of a covariate. For example, we want to know if calcium concentration in drinking water is a relevant covariate once we have considered the magnesium concentration. Thus, we have compared bands defined from calcium levels after standardization by levels of magnesium. Heterogeneity of these bands will indicate that calcium provides relevant information beyond that supplied by magnesium. Furthermore, comparison of calcium bands before and after standardizing by levels of magnesium will illustrate the interaction of both factors.
Disease Mapping
One main objective of epidemiologic surveillance tasks is the detection of regions that have unusually high risk. Disease mapping is a powerful tool designed to this end, especially when we are dealing with environmental risk factors. Because environmental phenomena are linked to geography, the influence of these risk factors can be detected by geographic representations of relative risks. [See Lawson and Williams (2001) for an introductory text and Lawson et al. (1999) for a deeper insight.]
Disease mapping deals typically with small geographic units. If the influence of hidden environmental factors extends over several units, mortality/morbidity counts will be correlated. Therefore, to analyze these units we need statistical models allowing for spatial correlation.
Furthermore, the small populations attached to these geographic units produce unstable estimates of relative risks, thus requiring more robust statistical methods.
The RIF addresses both problems by resorting to the empirical Bayes analysis of a hierarchical Poisson-gamma model similar to that of Clayton and Kaldor (1987). Computational details are described in the statistical appendix of Aylin et al. (1999).
From a surveillance perspective, we want to determine if removing the effects of a covariate changes the geographical pattern of relative risks. To this end we can perform disease mapping before and after standardization by levels of a covariate. By comparing the resulting maps, we can verify whether high-risk regions move to lower levels of risk or if they remain high, indicating that factors other than this covariate are still affecting population health status. There could be hidden factors not included in the study. The geographic pattern can help us determine the nature of these hidden factors.
Results
To delimit the size of the studied phenomena, we first considered the annual rates per 100,000 inhabitants for the whole region. The annual rates for cerebrovascular disease from 1991 to 994 are 153.37 for women and 114.26 for men. From 1995 to 1998 these rates are 129.03 and 97.17, respectively.
The annual rates for ischemic heart disease per 100,000 inhabitants are 80.13 for women and 121.46 for men from 1991 to 1994, whereas they are 86.41 and 126.10, respectively, from 1995 to 1998.
Cerebrovascular disease rates are higher in women than in men; for ischemic heart disease the rates are higher in men. Comparing both periods, we observe a decrease in the rate of cerebrovascular disease and an increase in the rate of ischemic heart disease.
The subsequent analysis focuses on relative risks rather than on the rates and is based on the routine output of the RIF.
Exposure Analysis
Figures 1 and 2 are a comparison of the relative risk of bands defined from calcium and magnesium concentration levels. They display the SMRs and the 95% confidence intervals from the output obtained with the RIF. Figure 1 illustrates cerebrovascular mortality and Figure 2 illustrates ischemic heart mortality. All SMRs for these figures have been computed after standardization by age and deprivation index.
For each disease we have constructed four plots according to sex and covariate. In each band, both 1991–1994 and 1995–1998 are represented side by side for better comparison of temporal variation.
The horizontal line at SMR = 1 allows quick recognition of those intervals not containing this particular value, that is, those intervals that were not significant because the corresponding band shows a significantly high or low relative risk. Because we have performed intrinsic indirect standardization, distance from the SMR = 1 indicates a difference with respect to the average behavior of the whole region. The presence of significant confidence intervals is a clear sign of the heterogeneity of the bands.
As we discussed in “Material and Methods,” this information has to be complemented with testing the homogeneity of the bands. The resulting chi-square statistic G and corresponding p-values are displayed in Table 2 under the headings G and p-value.
Magnesium after Calcium and Calcium after Magnesium
To see the additional effect of each covariate once the other has been taken into account, we have repeated the analysis of the preceding section. This time, however, SMRs have been computed after standardization by the covariate not explicitly present in the exposure analysis. Consequently, Figure 3 has to be compared with Figure 1 and Figure 4 with Figure 2. The corresponding homogeneity tests appear in Table 2 under the headings G and p-value.
From these comparisons we can see that trends are similar in general but that confidence intervals become less significant. Many more intervals intersect the horizontal line SMR = 1 when we standardize by the covariate not present in the exposure analysis. This loss of significance is apparent as well from columns p-value and p-value of Table 2. We verify there that p-values increase notably from from first column to second column, indicating that the hidden covariate contributes to the heterogeneity between bands. Conversely, small p-values suggest the covariate that defines the bands still provides useful information beyond that of the covariate used in standardization.
Disease Mapping
Rapid Inquiry Facility output gives tabulated SMRs for all municipalities. Shown for each sex group (males, females, and males + females) for each municipality are observed and expected number of cases, the corresponding SMR and the 95% confidence interval, and the smoothed estimation of this SMR based on the empirical Bayes procedure mentioned in preceding sections. These rows are duplicated to show standardized and nonstandardized results.
Because we are working with 538 municipalities and two diseases, the textual output is more than 500 pages for each of the studied covariates. Although interesting for detailed consultation purposes, it does not fit in the reduced space of a scientific paper. Maps better summarize these results. They facilitate the capture of essential aspects of health status. However, the entire set of maps for the present study is excessive, and we will restrict ourselves to some of the most illustrative results.
Figure 5 presents disease mapping of total (males + females) cerebrovascular mortality for 1991–1998. Figure 5A and C represent smoothed municipal relative risks. Figure 5B and D distinguish between significantly high, significantly low, and nonsignificant 95% confidence intervals of SMRs, as the value SMR = 1 is below, above, or inside the interval. Thus, we have an estimate of the relative risk (Figure 5A,C) jointly with a measure of our confidence that the value represents a real risk and is not being produced by mere chance (Figure 5B,D).
In Figure 5A and B, SMRs have been standardized by age, sex, and deprivation index. In Figure 5C and D, standardization has included magnesium levels as well.
Discussion
Effects of Calcium and Magnesium
The results displayed in Table 2 suggest a relationship between calcium and magnesium and the data on mortality from cerebrovascular and ischemic diseases. According to the p-value, testing homogeneity of bands shows clear evidence of this association for cerebrovascular disease in women. All p-values are below 0.0001 for both periods studied and both covariates. Evidence of this association is not as strong for this same disease in men, because there is no clear significant result from 1991 to 1994 with magnesium and from 1995 to 1998 with calcium.
For ischemic heart disease significant heterogeneous results are not achieved if the threshold is set to 0.001 to declare a p-value significant. Nevertheless, the p-values are quite small, with most between 0.001 and 0.05. A cautious conclusion could be not to discard the possibility of this association without further consideration.
We can examine the nature of those relationships. Focusing on the plot for women and magnesium in Figure 1D (p-value = 1.12 × 10–7), the one most significant is in Table 2, where we can see a descending trend with increasing levels of magnesium from bands 1 to 4. Band 5 breaks this trend, giving a U-shaped aspect to this plot.
A similar pattern can be seen for ischemic heart mortality in women and magnesium levels, although in this case heterogeneity has not been so significant (p-value = 0.031).
Regarding the interaction between calcium and magnesium, Table 2 reveals clearly how the effects of both covariates are partially confounded because they produce in each other a loss of significance when used in the previous standardization. The correlation coefficient for both covariates is 0.59. In this situation it is difficult to assess the independent effect of each one, and it is best to refer to the effect of hardness of drinking water.
In summary, we can say that this study provides statistical evidence of a relationship between mortality from cardiovascular diseases and hardness of drinking water. This relationship is stronger in cerebrovascular disease than in ischemic heart disease, is more pronounced in women than in men, and is more apparent with magnesium than with calcium. Nevertheless, the protective nature of these two factors is not clearly established. Although the results obtained suggest this possibility, they are not conclusive because of the irregular trend in the series of confidence intervals and because many of the results are not significant. Hidden socioeconomic and environmental factors not controlled with the deprivation index or the studied covariates may remain. As suggested by a reviewer, these could be caused by an ecologic bias associated with this region.
Temporal Trend
We have paired the confidence intervals corresponding to 1991–1994 and 1995–1998 for each band in every plot in Figures 1–4. Direct inspection of these charts reveals the stability of the SMRs during the entire period studied. The majority of these pairs have a large intersection, with both intervals sharing a large portion of their range of values.
Although one may have the impression that small decreases predominate in these sets of paired intervals, the overlapping areas are so important that the evidence of temporal variation is negligible.
Spatial Distribution of Risk
No spatial trend is apparent from maps presented in Figure 5, but several clusters of different sizes are scattered over the entire region. For illustrative purposes we focused on two particular regions, which are circled in the figure.
The upper circle shows a cluster of municipalities with high SMR and significant confidence intervals. This is obvious in the map of smoothed SMRs standardized by age, sex, and deprivation index (Figure 5A) and in the map with significance of confidence intervals (Figure 5B). The lower circle shows another less extreme cluster.
When we include magnesium levels in the standardization calculus, we remove the effect of this covariate in some sense. When we compare Figure 5A and B with Figure 5C and D, we can see the effect of this removal. For example, the upper circle shows that this change produces even more municipalities with significant confidence intervals than before (so that the situation is worse than previously thought). In the lower circles, the opposite is true. Some municipalities decrease the significance of their SMRs (so that their previous high relative risk has been partially explained by their level of magnesium).
Some Methodological Issues
The strategy of analysis followed in this study conforms to the exploratory nature of the RIF as a tool to get quick and flexible insight into epidemiologic surveillance problems.
One primary concern with the type of exposure analysis described here is the sensitivity of results to the number and cut points of exposure bands. An advisable practice is to try various configurations of these bands. We still lack a clear recommendation about this topic.
In our study we used three, five, and seven bands in each of the 12 studies (disease–covariate–sex combinations). Table 3 shows results obtained in one of these comparisons—cerebrovascular mortality in women with bands defined from magnesium levels. Figure 6 displays the SMRs and their 95% confidence intervals from the output obtained with three, five, and seven bands. We can see that although concrete numerical results vary, the general conclusions remain.
Considering these 12 comparisons of different band settings, we found that three bands tend to produce less significant heterogeneous results than with five and seven bands, whereas there is little difference between these last settings. Therefore, we have presented results with five bands.
Conclusions
Water in the Comunidad Valenciana is very hard. Because of the infrequent occurrence, water with > 200 mg/L calcium and water containing > 10 mg/L magnesium are considered rare by the World Health Organization (1996). Nearly 20% of Comunidad Valenciana municipalities and population have water supplies with calcium concentrations > 200 mg/L. The case of magnesium is more striking—90% of the water supply contains magnesium concentrations > 10 mg/L. This could make identification of dose–response patterns and comparison between our results and those from other studies difficult. Nevertheless, we observed considerable consistency in a detailed analysis of the results obtained in the most recent studies of this issue.
In other studies of calcium, the rank of calcium concentration distribution in water is closer than in our study. For example, in the case–control study performed in Sweden (Rubenowitz et al. 1996), the interval is 22–225 mg/L, and in a recent ecologic study conducted in France (Marque et al. 2003), the calcium concentration interval is 9–146 mg/L. As in our study, mortality results for cardiovascular noncerebrovascular diseases are less clear than those for cerebrovascular diseases. Conversely, the case–control study performed in Sweden shows a nonmonotonic U-shaped relationship between calcium and heart attack risk mortality, as the authors obtained odds ratio < 1 in intermediate calcium concentrations between 34 and 81 mg/L (Rubenowitz et al. 1996). In other words, they found a relationship between heart attack risk mortality calcium levels that correspond to the second level of our distribution, in which we found a significant relationship, with SMR < 1 at concentrations between 65 and 89 mg/L. In a later study in Sweden, the results for calcium were not conclusive (Rubenowitz et al. 2000). Nevertheless, in the French study, the shape of the relationship between calcium and cardiovascular mortality presents a clear biologic gradient, with less mortality risk at higher calcium concentrations in water (Marque et al. 2003).
On the other hand, our results support the protective effect assumption of magnesium and the mortality risk due to cardiovascular diseases (Rubenowitz et al. 1996, 2000). In addition, the results of the study performed in France (Marque et al. 2003) show a U-shaped relation between magnesium concentration in drinking water and cerebrovascular mortality, with lower risk in intermediate values of magnesium, in the same way described in our study.
Briefly, the results of our study in Valencia support the assumption of association between magnesium and mortality risk due to cardiovascular diseases. However, results for calcium are less clear. The current lack of studies and the ecologic nature and limitations of the exposure valuation used suggest that these study results should be explored further with more suitable designs. This could be achieved in the moving cohorts studies framework that addresses the role of different nutrients and other factors in cardiovascular health.
This article is part of the mini-monograph “Health and Environment Information Systems for Exposure and Disease Mapping, and Risk Assessment.”
Figure 1 95% Confidence intervals and means of SMRs for cerebrovascular mortality of males (A,C) and females (B,D) in bands defined for calcium (A,B) and magnesium (C,D). B, band.
Figure 2 95% Confidence intervals and means of SMRs for ischemic heart mortality of males (A,C) and females (B,D) in bands defined for calcium (A,B) and magnesium (C,D). B, band.
Figure 3 95% Confidence intervals and means of SMRs for cerebrovascular mortality of males (A,C) and females (B,D) in bands defined from calcium (A,B) and magnesium (C,D). B, band.
Figure 4 95% Confidence intervals and means of SMRs for ischemic heart mortality of males (A,C) and females (B,D) in bands defined for calcium (A,B) and magnesium (C,D), after standardization by calcium C,D) and magnesium (A,B). B, band.
Figure 5 Disease mapping of total cerebrovascular mortality for the whole period: smoothed SMRs (A,C) and significance of 95% confidence intervals (B,D) after standardization by age, sex, and deprivation index (A,B) and further standardization by Mg (C,D). Inf, infinity. Municipalities illustrating the change of risk level when adjusting for the covariate are circled in blue.
Figure 6 Comparing settings of magnesium bands for cerebrovascular mortality in women. Output obtained with (A) three bands, (B) five bands, and (C) seven bands. B, band.
Table 1 Bands defined in terms of calcium and magnesium concentrations.
Covariate Band threshold (mg/L) Number of muncipalities Percentage of population
Calcium [12, 65] 109 13.00
]65, 89] 110 12.89
]89, 112] 128 14.93
]112, 136] 84 39.02
]136, 480] 107 20.15
Magnesium [1, 14] 120 8.23
]14, 23] 96 9.78
]23, 34] 129 25.00
]34, 43] 87 38.01
]43, 117] 106 18.94
Table 2 Testing homogeneity of bands.
Disease Covariate Period Sex Ga p-Valuea Gb p-Valueb
Cerebrovascular Calcium 1991–1994 Males 24.38 6.70 × 10–5 14.88 0.0049
Females 30.75 3.44 × 10–6 5.46 0.2436
1995–1998 Males 8.02 9.06 × 10–2 4.85 0.3028
Females 27.07 1.92 × 10–5 4.87 0.3008
Magnesium 1991–1994 Males 10.16 3.79 × 10–2 6.87 0.1427
Females 21.65 2.35 × 10–4 1.71 0.7883
1995–1998 Males 22.89 1.33 × 10–4 16.46 0.0025
Females 38.00 1.12 × 10–7 15.78 0.0033
Ischemic Calcium 1991–1994 Males 8.85 6.49 × 10–2 8.58 0.0725
Females 11.73 1.95 × 10–2 6.77 0.1488
1995–1998 Males 13.28 9.98 × 10–3 14.76 0.0052
Females 15.14 4.49 × 10–3 10.98 0.0268
Magnesium 1991–1994 Males 10.75 2.96 × 10–2 7.16 0.1275
Females 10.64 3.10 × 10–2 2.80 0.5918
1995–1998 Males 10.62 3.11 × 10–2 8.63 0.0710
Females 8.02 9.08 × 10–2 6.98 0.1368
G, chi-square statistic defined by Equation 1.
aAfter standardization by age and deprivation index.
bAfter further standardization by levels of the other covariate.
Table 3 Testing homogeneity of bands for cerebrovascular mortality in women with 3, 5, and 7 magnesium bands.
Bands (n) Period G p-Value
3 1991–1994 34.830 2.74 × 10–8
3 1995–1998 52.580 3.82 × 10–12
5 1991–1994 21.650 2.35 × 10–4
5 1995–1998 38.000 1.12 × 10–7
7 1991–1994 38.950 7.32 × 10–7
7 1995–1998 49.920 4.88 × 10–9
G, chi-square statistic defined by Equation 1.
==== Refs
References
Abellán C Ferrándiz J López A 2003. Spatio temporal imputation in environmental data sets. In: Proceedings from EUROHEIS/SAHSU Conference 2003, 30–31 March 2003, Östersund, Sweden (Jarup L, ed). London:Imperial College London, 74–78.
Arias A Rebagliato M Palumbo MA Bellver R Ashton J Colomer C 1993 Desigualdades en salud en Barcelona y Valencia Med Clin 100 281 287
Aylin P Maheswaran R Wakefield J Cockings S Jarup L Arnold R 1999 A national facility for small area disease mapping and rapid initial assessment of apparent disease clusters around a point source: the UK Small Area Health Statistics Unit J Public Health Med 21 3 289 298 10528956
Clayton D Kaldor J 1987 Empirical Bayes estimates of age-standardized relative risks for use in disease mapping Biometrics 43 671 681 3663823
Cradford MD Gardner MJ Morris JN 1971 Changes in water hardness and local death-rates Lancet 14 327 330
Ferrándiz J Abellán JJ López A Sanmartín P Vanaclocha H Zurriaga O 2002. Geographical distribution of the cardiovascular mortality in Comunidad Valenciana (Spain). In: GIS for Emergency Preparedness and Health Risk Reduction (Briggs D, Forer P, Jarup L, Stern R, eds). Dordrecht, the Netherlands:Kluwer Academic Publishers, 267–282.
Ferrándiz J López A Sanmartín P Vanaclocha H Zurriaga O Abellán JJ 2000. Atlas of Cardiovascular Mortality in the Comunidad Valenciana (1987–1996). Available: http://dgsp.san.gva.es/isum/epidemiologia/estudios/atlas.htm [accessed 15 Mar 2004].
Ferrándiz J López-Quílez A Gómez-Rubio V Sanmartín P Martínez-Beneito M Melchor I 2003 Statistical relationship between hardness of drinking water and cerebrovascular mortality in Valencia: a comparison of spatiotemporal models Environmetrics 14 5 491 510
Gómez V Ferrándiz J López A 2002. Implementation of Covariate Studies in an Epidemiological G.I.S. Technical Rep 15-2002. Valencia, Spain:Dept. d’Estadística i I.O., Universitat de València. Available: http://matheron.uv.es/investigar/tr15-02.ps [accessed 15 Mar 2004].
Hall P Jungner I 1993 Hard drinking water and ischemic heart disease: calcium, bloodlipids, and acute myocardial infarcts J Med Syst 17 227 281 8254269
Lacey RF Shaper AG 1984 Changes in water hardness and cardiovascular death rates Int J Epidemiol 13 18 24 6698699
Lawson A Biggeri A Böhning D Lesaffre E Viel JF Bertollini R eds. 1999. Disease Mapping and Risk Assessment for Public Health. Chichester, UK:Wiley.
Lawson AB Williams FLR 2001. An Introductory Guide to Disease Mapping. Chichester, UK:Wiley.
Marque S Jacqmin-Gadda H Dartigues JF Commenges D 2003 Cardiovascular mortality and calcium and magnesium in drinking water: an ecological study in elderly people Eur J Epidemiol 18 305 309 12803370
Marx A Neutra RR 1997 Magnesium in drinking water and ischemic heart disease Epidemiol Rev 19 258 272 9494787
Melchor I Garcia-Senchermes C Martínez P Valero S Company C Vanaclocha H 1998. Estadísticas de Mortalidad de la Comunidad Valenciana 1995. Valencia, Spain:Conselleria de Sanitat, Generalitat Valenciana.
Nolasco A Orts R Pérez-Hoyos S Benavides FG Godoy C Librero J 1992. Análisis Geográfico de la Mortalidad en la Comunidad Valenciana (1981–84). Vol 21. Valencia, Spain:Generalitat Valenciana.
Pocock SJ Shaper AG Cook DG Packmahan RF Lacey RF Powell P 1980 British Regional Heart Study: geographic variations in cardiovascular mortality, and the role of water quality Br Med J 5 1243 1249 7388489
Rubenowitz E Axelsson G Rylander R 1996 Magnesium in drinking water and death from acute myocardial infarction Am J Epidemiol 143 456 462 8610660
Rubenowitz E Molin I Axelsson G Rylander R 2000 Magnesium in drinking water in relation to morbidity and mortality from acute myocardial infarction Epidemiology 11 416 421 10874548
Rylander R Bonevik H Rubenowitz E 1991 Magnesium and calcium in drinking water and cardiovascular mortality Scand J Work Environ Health 17 91 94 2047817
Van der Vijver LPL Van der Waal MAE Weterings KGC Dekkere JM Schouten EG Kok FJ 1992 Calcium intake and 28-year cardiovascular and coronary artery disease and low hardness of drinking water Int J Epidemiol 21 36 39 1544755
Wakefield JC Best NG Waller L 2000. Bayesian approaches to disease mapping. In: Spatial Epidemiology: Methods and Applications (Elliot P, Wakefield JC, Best NG, Briggs DJ, eds). Oxford:Oxford University Press, 104–127.
World Health Organization 1978. International Classification of Diseases, Ninth Revision. Basic Tabulation List with Alphabetical Index. Geneva:World Health Organization.
World Health Organization 1996. Guidelines for Drinking-Water Quality. 2nd ed. Vol 2. Geneva:World Health Organization.
Yang CY 1998 Calcium and magnesium in drinking water and risk of death from cerebrovascular disease Stroke 29 411 414 9472882
|
15198925
|
PMC1247198
|
CC0
|
2021-01-04 23:39:21
|
no
|
Environ Health Perspect. 2004 Jun 15; 112(9):1037-1044
|
utf-8
|
Environ Health Perspect
| 2,004 |
10.1289/ehp.6737
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0112-a00543PerspectivesCorrectionBackward Estimation of Exposure to Organochlorines using Repeated Measurements Karmaus Wilfried 7 2004 112 10 A543 A543 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
Karmaus et al. detected errors in their article “Backward Estimation of Exposure to Organochlorines using Repeated Measurements” [
Environ Health Perspect 112:710–716 (2004)]. In Table 1, values in the equations for the proposed regression model were incorrect. The correct equations for Table 1 are as follows:
For estimation of 1970s values from 1980s values,
For estimation of 1980s values from 1990s values
Also, in Figure 3, there should not be a measurement for the year 2010. The authors appologize for the errors.
|
0
|
PMC1247392
|
CC0
|
2021-01-04 23:40:08
|
no
|
Environ Health Perspect. 2004 Jul; 112(10):A543
|
utf-8
|
Environ Health Perspect
| 2,004 |
nan
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences 10.1289/ehp.7028ehp0112-00113715289156ResearchCommentariesThe Need to Decide If All Estrogens Are Intrinsically Similar Moggs Jonathan G. Ashby John Tinwell Helen Lim Fei Ling Moore David J. Kimber Ian Orphanides George Syngenta CTL, Alderley Park, Cheshire, United KingdomAddress correspondence to J.G. Moggs, Syngenta CTL, Alderley Park, Cheshire, SK10 4TJ UK. Telephone: 44-1625-519315. Fax: 44-1625-585715. E-mail:
[email protected] thank A. Soames for the histopathology data, I. Kupershmidt and E. Hunter (Silicon Genetics) for advice on statistical analysis of microarray data, and T. Barlow (Food Standards Agency) for critical comments.
This work was partially supported by the U.K. Food Standards Agency.
The authors declare they have no competing financial interests.
8 2004 19 5 2004 112 11 1137 1142 12 2 2004 19 5 2004 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. We used gene expression profiling to investigate whether the molecular effects induced by estrogens of different provenance are intrinsically similar. In this article we show that the physiologic estrogen 17β-estradiol, the phytoestrogen genistein, and the synthetic estrogen diethylstilbestrol alter the expression of the same 179 genes in the intact immature mouse uterus under conditions where each chemical has produced an equivalent gravimetric and histologic uterotrophic effect, using the standard 3-day assay protocol. Data are also presented indicating the limitations associated with comparison of gene expression profiles for different chemicals at times before the uterotrophic effects are fully realized. We conclude that the case has yet to be made for regarding synthetic estrogens as presenting a unique human hazard compared with phytoestrogens and physiologic estrogens.
diethylstilbestrolestrogengene expressiongenisteinmicroarrayphytoestrogentoxicogenomicsuterus
==== Body
The question of whether phytoestrogens and synthetic estrogens are toxicologically similar, or intrinsically different, presents a challenge to all involved in human hazard and risk assessments. Although there is a general concern that exposure to nanogram or microgram amounts of environmental estrogens may be associated with adverse health effects, in the public mind there is a widespread belief that foods and dietary supplements containing milligram quantities of phytoestrogens confer only health benefits. An implicit distinction therefore seems to have been drawn between synthetic and plant-derived estrogens—a belief sustained in the public mind by the assumption that natural is good and synthetic is bad—but an untested and potentially misleading notion for those involved with science-based human hazard/risk assessments.
Phytoestrogens and synthetic estrogens are generally considered separately in the literature. For example, Howdeshell et al. (1999) suggested a possible association between the advance in first estrus observed in mice exposed in utero to 2.4 μg/kg of the synthetic environmental estrogen bisphenol A and reports of an increased incidence of hypospadias in boys (Paulozzi et al. 1997) and the earlier sexual maturation of girls (Herman-Giddens et al. 1997)—the implication being that synthetic estrogens present a greater hazard than the much higher levels of phytoestrogens being consumed by those same children. In contrast, there are reports of an increased incidence of hypospadias in boys born to vegetarians (North and Golding 2000), of alterations in the menstrual cycle (Cassidy et al. 1994), and of reduced breast cancer incidences (Messina 1999) among women eating diets rich in phytoestrogens. Support for these epidemiologic observations comes from experimental studies indicating that advances in sexual development in rodents can be induced by their exposure to phytoestrogens (Casanova et al. 1999; Cassidy and Faughnan 2000; Safe et al. 2002). In contrast to these separate lines of inquiry, Newbold and colleagues have evaluated potential similarities between natural and synthetic estrogens. In seminal studies, they demonstrated that neonatal exposure of female mice to equipotent uterotrophic doses of the phytoestrogen genistein (GEN; Figure 1) or the synthetic estrogen diethylstilbestrol (DES) leads to an identical incidence of uterine adenomas at 18 months of age (Newbold et al. 2001). However, in attempting to draw parallels, or distinctions, between phytoestrogens and synthetic estrogens, it is imperative to consider growing awareness of the complexity of estrogen signaling pathway and the pleuripotential biologic activities of most organic chemicals—irrespective of their origin.
Estrogen signaling in mammalian cells is primarily mediated at the molecular level by two members of the nuclear receptor superfamily—estrogen receptors alpha (ER-α) and beta (ER-β). Ligand-activated ER-α and ER-β function as transcription factors, in conjunction with numerous coregulatory proteins, in order to activate or repress the transcription of ER-responsive genes (Hall et al. 2001; Moggs and Orphanides 2001). There is considerable variation in the binding affinity of ER-α and ER-β among different estrogens (Kuiper et al. 1998). In the case of the chemicals studied here, the physiologic estrogen 17β-estradiol (E2) and DES bind with a similar affinity to ER-α and ER-β, whereas GEN binds with approximately 20-fold higher affinity to ER-β than to ER-β (Kuiper et al. 1998). Concerning nonhormonal properties of the test chemicals (most of which have only be defined in vitro), GEN inhibits a range of enzymes, including tyrosine kinases (Akiyama et al. 1987), nitric oxide synthase (Duarte et al. 1997), and topoisomerase II (Okura et al. 1988), and also decreases calcium-channel activity (Potier and Rovira 1999), lipid peroxidation (Arora et al. 1998), and diacylglycerol synthesis (Dean et al. 1989). Likewise, DES is reported to induce aneuploidy in mammalian cells (Aardema et al. 1998) and to bind to rat liver DNA (Williams et al. 1993). More recently, some phytoestrogens were reported to inhibit the aromatase-mediated conversion of testosterone to E2
in vitro (Almstrup et al. 2002), and equol, the major circulating estrogenic metabolite associated with the dietary ingestion of phytoestrogens, is reported to selectively sequester dihydrotestosterone and thereby to act as a functional antiandrogen in vivo (Lund et al. 2004).
In order to advance understanding in this area, we decided to compare the genes expressed in the immature mouse uterus when it had grown in response to treatment with the estrogens E2, DES, and GEN. The immature mouse uterus was selected for our analysis because it is a major estrogen-responsive organ and forms the basis for a reference assay of estrogenic activity (Owens and Ashby 2002), including carcinogenesis (Newbold et al. 2001). Furthermore, it expresses both ER-α and ER-β (Weihua et al. 2000) and the androgen receptor (Frasor et al. 2003). We initially conducted a global analysis of gene expression in the mouse uterus at 1, 2, 4, 8, 24, 48, and 72 hr after exposure to a single high dose of either GEN (250 mg/kg) or E2 (400 μg/kg). These single high doses yielded a sustained uterotrophic response over 72 hr (Figure 2A) and were selected to avoid the complex transcriptional program that may result from the standard uterotrophic assay exposure regime in which each test compound is dosed by repeated administration on 3 consecutive days (Odum et al. 1997). Groups of 10 sexually immature mice [Alpk:APfCD-1; 19/20 days of age; maintained on RM1 diet (Special Diets Services Ltd., Witham, Essex, UK)] received a single subcutaneous injection of each compound or the test vehicle [arachis oil (AO); 5 mL/kg], and uterine RNA was isolated and pooled by group at each of the seven time points to determine gene expression levels among the 12,488 mouse genes represented on the Affymetrix MG-U74Av2 GeneChip (Affymetrix, High Wycombe, UK). Transcript profiling was performed using MG-U74Av2 GeneChip and Microarray Analysis Suite 5.0 (Affymetrix). Normalization and hierarchical clustering were performed with GeneSpring 6.0 (Silicon Genetics, Redwood City, CA, USA). MIAME (Minimum Information About a Microarray Experiment)-compliant microarray data are available as supplementary information and submitted to the Gene Expression Omnibus (GEO) database (GEO 2004). These data were analyzed using unsupervised hierarchical clustering and yielded temporal relationships between the expression profiles of 3,450 genes that were either up- or down-regulated (> 1.5-fold) by E2 and/or GEN (Figure 2B). Each chemical induced a similar, multistage transcriptional response (Figure 2B), although it is noteworthy that we observed variations in the magnitude and timing of both early (e.g., c-fos) and late (e.g., lactotransferrin) ER-responsive genes during the uterotrophic responses induced by E2 and GEN (Figure 2C).
A detailed description of the molecular functions of the genes affected, together with their association with physiologic changes during uterine growth, has been reported (Orphanides et al. 2003) and will be described in more detail in a future publication (Moggs et al., unpublished data).
These observations suggest that GEN does not induce “off-target” ER-independent transcriptional responses, that is, those associated with the properties of GEN other than estrogenicity. Furthermore, there was no evidence for the topoisomerase II–inhibiting properties of GEN in the bone marrow of the present mice despite demonstration of the sensitivity of that tissue to the potent micronucleus-inducing activity of the topoisomerase II inhibitor etoposide (data not shown). Together, these data led us to question whether a synthetic estrogen such as DES would also induce similar transcriptional responses in the immature mouse uterus.
In order to avoid temporal vagaries in gene expression (e.g., Figure 2C), we decided to anchor our transcript profiling data to the phenotype of the grown uterus by employing equipotent uterotrophic doses of E2, GEN, and DES. We compared the global gene expression profiles in the uteri of intact immature mice stimulated with three daily low doses of either GEN, DES, or E2, with an exposure regimen the same as that used in a standard 3-day uterotrophic assay (Odum et al. 1997). The route of administration and the doses of GEN and DES used were as described by Newbold et al. (2001) in their equivalent-outcome carcinogenicity bioassays of these two chemicals. Three independent replicates of four groups of sexually immature mice (Alpk:APfCD-1; 19/20 days of age; maintained on RM1 diet) received three daily subcutaneous injections of GEN (50 mg/kg), E2 (2.5 μg/kg), or DES (2 μg/kg). Control animals received the vehicle, AO (5 mL/kg). These doses elicited similar uterotrophic responses (72 hr after the initial dose; Figure 3A, Table 1) and identical histologic changes in the uteri of the treated animals (Table 1). Uterine RNA was isolated and pooled for each of the 12 groups and analyzed for changes in gene expression levels using the same Affymetrix microarray of 12,488 mouse genes. The data were analyzed using two independent statistical methods. First, unsupervised hierarchical clustering defined the global relationships (Euclidean distances) between the 12 gene expression profiles (Figure 3B). The three control groups clustered under one node, whereas the chemical treatment groups formed a separate node of compound-independent clusters, indicating equal similarity within and between the transcriptional responses induced by the three estrogens (Figure 3B). One-way analysis of variance (ANOVA), with Bonferroni (Holm 1979) correction (familywise error rate < 0.05) to minimize false positives, identified 179 genes where expression levels were altered by one or more chemical treatments (Figure 3C). Remarkably, Tukey post hoc testing revealed that all of these genes were affected in all nine compound treatment groups.
Table 2 highlights the high degree of similarity between the transcriptional responses to each of the three estrogens. These include established estrogen-responsive genes such as lactotransferrin, complement component 3, c-fos, small proline-rich protein 2A, and keratoepithelin (Hewitt et al. 2003; Naciff et al. 2003), together with many genes that have not previously been associated with estrogenicity (Table 2).
Although these three estrogens can alter the expression of some genes with different magnitudes [e.g., peptidyl arginine deiminase II is up-regulated to a lesser extent by E2 (1.86-fold ± 0.27) relative to GEN (9.11-fold ± 0.33) and DES (5.15-fold ± 1.53); Table 2], the present data show that the same genes are affected during equivalent uterotrophic responses. Previous studies have revealed both similarities and differences between transcriptional responses induced at a single time point after exposure to E2 and DES in the uteri of immature ovariectomized mice (Watanabe et al. 2003) and after exposure to either GEN, bisphenol A, or 17α-ethynyl estradiol in the reproductive tract of intact adult rats (Naciff et al. 2002). We suggest that these reported differences most probably arise from dose-dependent variations in the magnitude and kinetics of gene expression (Figure 2C), rather than from the operation of distinct mechanisms of estrogenic action.
Our data indicate that estrogens of differing provenance may have in common the potential for both beneficial and adverse health effects. This highlights the need for an holistic approach to hazard assessment wherein preconceptions are replaced by an objective assessment of the likely perturbations of physiologic functions caused by combined exposures to physiologic, synthetic, and plant-derived estrogens. This need is reinforced by data showing that plasma concentrations of isoflavones in infants fed soy formula are approximately 200 times higher than for those fed human milk (Setchell et al. 1997), by the estimated daily intake of approximately 29 mg of phytoestrogens for individuals taking dietary supplements (Committee on Toxicity of Chemicals in Food, Consumer Products and the Environment 2003), and by the demonstration that estrogens of different provenance can act additively in the rodent uterus (Tinwell and Ashby 2004).
Figure 1 Chemical structure of GEN, E2, and DES.
Figure 2 Induction of very similar multistage transcriptional responses in the mouse uterus by E2 and GEN. (A) Blotted uterine weights (mean ± SD) of sexually immature mice (n = 10/group) at different times after a single subcutaneous dose of E2 (400 μg/kg), GEN (250 mg/kg), or AO (control; 5 mL/kg). See text for details of experiments. (B) Temporal expression profiles of 3,450 genes up-regulated or repressed (> 1.5-fold) by either E2 (400 μg/kg) or GEN (250 mg/kg) at one or more of seven different time points. The magnitude of altered gene expression (fold change vs. time-matched vehicle control) is indicated by color; genes are grouped according to similarity of their temporal expression profiles (Pearson correlation-based hierarchical clustering). (C) Northern blot analysis of temporal expression pattern of early (c-fos) and late (lactotransferrin) estrogen-responsive genes; the fold induction of gene expression relative to time-matched vehicle controls was calculated after data were normalized to the expression of the control gene RPB1 (accession number NM_009089).
*p < 0.05;
**p < 0.01; two-sided Student t-test.
Figure 3 Equivalence of biologic responses induced in the mouse uterus by E2 (E), GEN (G), and DES (D). (A) Blotted uterine weights (mean ± SD) of three independent replicate groups (1–3) of sexually immature mice (n = 4/group) after three daily subcutaneous injections of either GEN (50 mg/kg), E2 (2.5 μg/kg), DES (2 μg/kg), or AO [control (C); 5 mL/kg]. (B) Unsupervised Euclidean-distance–based hierarchical clustering of 4,134 expressed genes. (C) Near-identical gene expression profiles induced by the three estrogens 72 hr after equipotent uterotrophic doses. Significant changes in gene expression induced by one or more of the three estrogens were identified by one-way ANOVA (parametric test, assuming equal variance). The magnitude of altered gene expression (fold change vs. vehicle control) is indicated by color.
Table 1 Blotted uterine weights and endometrial and epithelial cell heights (mean ± SD) after exposure to E2, GEN, or DES for 3 consecutive days.a
Cell height (μm)
Compound Dose (per kg) Blotted uterine weight (mg) Endometrium Epithelium
AO 5 mL 13.0 ± 2.4 159.0 ± 23.1 (11) 11.4 ± 1.1
E2 2.5 μg 45.3 ± 8.6* 246.1 ± 52.4* (9) 23.3 ± 1.4*
GEN 50 mg 39.8 ± 5.3* 273.7 ± 63.3* (12) 23.7 ± 3.1*
DES 2.0 μg 49.8 ± 13.0* 273.2 ± 55.9* (10) 22.6 ± 4.0*
There were 12 animals/group, but not all of the histopathology samples were suitable for analyses; numbers in parentheses indicate the number of animals per group from which the histology data were generated.
a Data were assessed for statistical significance using a two-sided Student t-test:
* p < 0.01.
Table 2 Quantitative data for 179 differentially expressed genes (from Figure 3C) regulated in the mouse uterus by all three estrogens (E2, GEN, and DES).a
Fold change in expression (mean ± SD)
Gene name GenBank accession no. E2 GEN DES
Up-regulated genes
Solute carrier family 9a3r1 U74079 1.8 ± 0.01 2.0 ± 0.1 2.0 ± 0.2
Keratin complex 2–8 X15662 2.6 ± 0.2 3.1 ± 0.2 3.1 ± 0.3
Laminin beta 3 U43298 4.3 ± 0.1 5.5 ± 1.1 5.3 ± 0.7
Claudin 7 AF087825 4.5 ± 0.5 6.5 ± 1.0 5.8 ± 0.6
bHLH-Zip transcription factor U49507 2.6 ± 0.3 3.1 ± 0.3 2.9 ± 0.1
RIKEN cDNA 1200008D14 AW208938 3.0 ± 0.3 3.5 ± 0.1 3.3 ± 0.3
Basic HLH-domain containing, class B2 Y07836 5.9 ± 1.0 6.6 ± 0.9 6.6 ± 0.8
RIKEN cDNA 9930104H07 AW122310 3.0 ± 0.3 3.2 ± 0.4 3.3 ± 0.1
Fucosyltransferase 2 AF064792 27.5 ± 1.2 34.6 ± 8.5 36.7 ± 5.5
Deleted in polyposis 1 U28168 1.8 ± 0.1 2.0 ± 0.02 2.0 ± 0.1
Microsomal glutathione S-transferase 3 AI843448 2.9 ± 0.2 3.3 ± 0.6 3.3 ± 0.1
Tumor-associated Ca signal transducer 2 Y08830 4.0 ± 0.3 4.6 ± 0.9 4.6 ± 0.3
Calpain 5 Y10656 5.5 ± 0.4 6.3 ± 1.0 6.6 ± 0.6
Mitochondrial creatine kinase Z13969 9.7 ± 1.1 12.2 ± 2.1 13.1 ± 1.8
ATPase 6v1a1 AW123765 2.0 ± 0.1 2.1 ± 0.2 2.1 ± 0.2
Tumor-associated Ca signal transducer 2 AI563854 8.0 ± 0.4 9.2 ± 1.0 8.5 ± 0.4
Lymphocyte antigen 6 complex, locus A X04653 7.8 ± 0.9 8.8 ± 0.3 8.5 ± 0.4
Chloride channel calcium-activated 3 AV373378 26.4 ± 3.4 26.7 ± 1.0 26.2 ± 3.9
Small proline-rich protein 2I AJ005567 23.9 ± 1.5 24.7 ± 1.3 23.6 ± 1.6
Oncoprotein induced transcript 1 AA615075 19.0 ± 3.1 20.0 ± 1.2 18.9 ± 2.5
Small proline-rich protein 2F AJ005564 59.8 ± 8.4 65.8 ± 1.1 60.6 ± 2.6
Small proline-rich protein 2E AJ005563 12.0 ± 1.0 12.9 ± 0.8 12.1 ± 0.9
Mucin 1 M84683 8.3 ± 0.6 8.6 ± 0.3 8.5 ± 0.5
Lipoocalin 2 X81627 150.3 ± 15.0 175.7 ± 10.5 162.8 ± 6.5
RIKEN cDNA 2210409B01 AF109906 3.5 ± 0.6 4.0 ± 0.3 3.8 ± 0.8
Interferon-activated gene 202A M31418 7.9 ± 1.0 9.8 ± 2.5 8.8 ± 0.7
Nuclear ankyrin-repeat protein AA614971 3.7 ± 0.6 4.3 ± 0.7 4.1 ± 0.9
RIKEN cDNA 5730469M10 AI850090 22.0 ± 5.8 30.5 ± 9.1 27.0 ± 6.5
RIKEN cDNA 1110034C02 AI837104 1.5 ± 0.1 1.6 ± 0.1 1.6 ± 0.03
IMAGE cDNA 4988271 AV373294 8.0 ± 2.5 10.6 ± 1.6 9.2 ± 1.1
RIKEN cDNA 5730493B19 AW122413 12.7 ± 0.3 19.0 ± 4.1 15.7 ± 0.9
Peptidoglycan recognition protein AV092014 13.4 ± 1.5 18.3 ± 3.0 14.5 ± 2.1
Inhibin beta-B X69620 13.6 ± 3.1 19.4 ± 4.7 16.0 ± 1.4
CEA-related cell adhesion molecule 2 AF101164 11.9 ± 1.6 17.8 ± 5.3 14.2 ± 1.7
Keratin complex 1–19 M36120 4.4 ± 0.4 5.5 ± 1.1 4.8 ± 0.5
CEA-related cell adhesion molecule 1 M77196 15.9 ± 2.4 23.9 ± 5.9 19.0 ± 3.8
SRC family-associated phosphoprotein 2 AB014485 2.7 ± 0.04 3.2 ± 0.4 2.9 ± 0.3
Peptidoglycan recognition protein AF076482 7.7 ± 1.9 10.4 ± 2.7 9.0 ± 2.0
CEA-related cell adhesion molecule 1 M77196 19.5 ± 3.9 30.4 ± 8.8 22.2 ± 2.7
CEA-related cell adhesion molecule 1 X67279 6.4 ± 0.7 8.4 ± 1.1 7.1 ± 1.1
Spermidine N1-acetyl transferase L10244 8.3 ± 0.9 11.2 ± 0.8 9.3 ± 0.6
RIKEN cDNA 0610007O07 AI851762 2.7 ± 0.1 3.0 ± 0.3 2.8 ± 0.1
Arginase 1 U51805 79.4 ± 9.8 131.9 ± 20.0 99.6 ± 14.5
Acetyl-coenzyme A synthetase 2 AW125884 2.2 ± 0.2 2.0 ± 0.1 2.2 ± 0.2
v-erb-b2 homolog 3 AI006228 3.4 ± 0.4 3.1 ± 0.6 3.4 ± 0.4
Phospholipase D3 AF026124 2.6 ± 0.2 2.4 ± 0.2 2.6 ± 0.2
RIKEN cDNA 0610031J06 AW122935 1.9 ± 0.1 1.8 ± 0.1 1.8 ± 0.1
Complement component 1q X58861 2.1 ± 0.1 2.0 ± 0.1 2.0 ± 0.2
Scotin AW123754 2.0 ± 0.1 2.0 ± 0.2 2.0 ± 0.2
CD24a antigen M58661 3.2 ± 0.1 3.1 ± 0.1 3.3 ± 0.3
Argininosuccinate synthetase 1 M31690 2.7 ± 0.3 2.7 ± 0.3 2.8 ± 0.4
ATPase 6v1a1 U13837 2.1 ± 0.1 2.1 ± 0.2 2.2 ± 0.2
Gelsolin-like actin-capping protein X54511 3.6 ± 0.5 3.7 ± 0.5 3.7 ± 0.2
Golgi phosphoprotein 2 AW125446 4.5 ± 0.5 4.6 ± 0.5 4.7 ± 0.1
Aldolase 1A Y00516 2.3 ± 0.2 2.3 ± 0.1 2.4 ± 0.1
Cathepsin L X06086 6.4 ± 0.8 6.3 ± 1.1 6.9 ± 0.4
CD14 antigen X13333 3.0 ± 0.1 2.8 ± 0.1 3.1 ± 0.2
Decay accelerating factor 2 L41365 4.0 ± 0.1 3.8 ± 0.8 3.8 ± 0.2
Actin-related protein 2/3 complex 1B AW212775 2.1 ± 0.2 2.1 ± 0.1 2.1 ± 0.2
Protective protein for β-galactosidase J05261 2.0 ± 0.1 2.0± 0.1 2.0 ± 0.1
Elastase 1 M27347 2.7 ± 0.1 2.5 ± 0.2 2.6 ± 0.1
Connexin 26 M81445 10.6 ± 1.0 9.8 ± 0.5 10.4 ± 0.8
Ceruloplasmin U49430 15.1 ± 2.8 15.0 ± 5.5 14.5 ± 2.1
Cathepsin H U06119 3.0 ± 0.2 3.0 ± 0.3 3.0 ± 0.3
Basigin Y16258 1.6 ± 0.1 1.5 ± 0.1 1.7 ± 0.1
Peptidylprolyl isomerase C–associated X67809 2.2 ± 0.2 2.2 ± 0.1 2.4 ± 0.3
Glutathione reductase 1 AI851983 2.3 ± 0.2 2.3 ± 0.1 2.6 ± 0.3
START domain–containing 3 X82457 1.5 ± 0.1 1.4 ± 0.03 1.5 ± 0.01
CD68 antigen X68273 4.6 ± 0.6 4.2 ± 0.6 5.0 ± 0.7
RIKEN cDNA E030027H19 AW211760 2.7 ± 0.3 2.7 ± 0.2 2.9 ± 0.1
cDNA sequence BC004044 AI461767 3.1 ± 0.2 3.4 ± 0.1 3.8 ± 0.5
E74-like factor 3 AF016294 5.1 ± 0.8 5.9 ± 0.5 6.5 ± 0.5
Glutathione S-transferase omega 1 AI843119 5.0 ± 1.1 4.6 ± 0.6 4.1 ± 0.7
Interferon-stimulated protein 20 AW122677 4.2 ± 0.1 4.3 ± 0.7 3.4 ± 0.4
Clusterin D14077 3.6 ± 0.7 3.9 ± 0.9 3.4 ± 0.4
Galectin 3 X16834 7.4 ± 1.3 8.7 ± 0.7 6.9 ± 0.4
Small proline-rich protein 2Ab AJ005559 51.1 ± 4.0 78.3 ± 15.7 44.2 ± 5.2
Complement component 3b K02782 14.8 ± 1.8 18.8 ± 1.0 14.8 ± 0.8
Small proline-rich protein 2C AJ005561 220.3± 31.0 340.5 ± 37.1 214.1 ± 41.1
Small proline-rich protein 2G AJ005565 9.4 ± 0.9 11.0 ± 0.3 9.6 ± 0.6
Prominin AF039663 3.5 ± 0.5 3.6 ± 0.6 3.4 ± 0.3
Lactotransferrinb J03298 88.7 ± 18.4 99.2 ± 13.9 76.9 ± 21.8
Carbonic anhydrase 2 M25944 7.9 ± 0.5 8.2 ± 0.9 7.3 ± 0.4
Complement component factor I U47810 36.5 ± 4.4 38.4 ± 5.6 32.9 ± 4.2
Mannosidase 2alphaB1 U87240 2.0 ± 0.2 2.0 ± 0.1 1.9 ± 0.1
Small proline-rich protein 2B AJ005560 32.9 ± 3.7 39.2 ± 1.9 30.7 ± 5.0
Small proline-rich protein 2Ab AJ005559 269.8 ± 23.7 329.1 ± 42.9 59.1 ± 40.8
RIKEN cDNA 5830413E08 AI849939 3.3 ± 0.4 3.3 ± 0.5 3.0 ± 0.3
RIKEN cDNA 1110029F20 AW125508 4.1 ± 0.1 4.1 ± 0.4 3.7 ± 0.1
Annexin A3 AJ001633 2.7 ± 0.5 4.2 ± 0.7 3.2 ± 0.6
Peptidase 4 U51014 2.0 ± 0.1 2.9 ± 0.3 2.3 ± 0.2
Laminin gamma 2 U43327 6.3 ± 1.3 17.3 ± 6.8 10.2 ± 1.0
Ubiquitin-like 3 AW120725 1.5 ± 0.1 1.8 ± 0.1 1.7 ± 0.03
Urate oxidase M27695 23.8 ± 9.5 143.9 ± 62.9 43.8 ±17.7
Amiloride binding protein 1 AI197481 3.5 ± 1.0 10.1 ± 0.8 6.0 ± 1.2
Keratin complex 1–19 AU040563 4.5 ± 1.0 7.2 ± 1.0 5.6 ± 0.4
Activated leukocyte cell adhesion molecule L25274 3.6 ± 0.8 5.1 ± 0.9 4.4 ± 0.5
CCAAT/enhancer binding protein β M61007 2.3 ± 0.1 2.8 ± 0.3 2.6 ± 0.1
Peptidyl arginine deiminase, type I AB013848 8.6 ± 0.7 15.8 ± 3.1 12.0 ± 1.7
Enolase 1 α AI841389 2.5 ± 0.3 3.2 ± 0.5 3.0 ± 0.3
p53 apoptosis effector related to Pmp22 AI854029 2.9 ± 0.3 4.1 ± 0.8 3.7 ± 0.5
β-Glucuronidase M19279 1.9 ± 0.1 2.3 ± 0.2 2.2 ± 0.1
Leucine-rich α-2-glycoprotein AW230891 9.3 ± 1.1 17.6 ± 4.1 14.5 ± 2.6
Quiescin Q6 AW123556 3.7 ± 0.2 5.5 ± 1.2 4.8 ± 0.7
GADD45a U00937 1.9 ± 0.2 2.6 ± 0.3 2.3 ± 0.1
Alkaline phosphatase 2 J02980 9.2 ± 0.4 22.6 ± 6.2 15.8 ± 2.0
Immediate early response 3 X67644 5.5 ± 0.8 10.8 ± 2.2 8.9 ± 2.0
Progressive ankylosis AW049351 2.2 ± 0.1 2.9 ± 0.4 2.8 ± 0.4
RAS p21 protein activator 4 AA163960 6.8 ± 0.9 14.2 ± 2.5 12.1 ± 1.7
Tumor-associated calcium signal transducer 1 M76124 2.1 ± 0.2 2.7 ± 0.3 2.6 ± 0.2
Hydroxysteroid (17-beta) dehydrogenase 11 AA822174 1.9 ± 0.1 2.3 ± 0.3 2.4 ± 0.1
Platelet-activating factor acetylhydrolase 1ba1 U57746 1.9 ± 0.1 2.2 ± 0.2 2.3 ± 0.1
Branched chain aminotransferase 1 U42443 2.4 ± 0.2 3.4 ± 0.2 3.4 ± 0.01
RIKEN cDNA 2400004E04 AI846720 1.7 ± 0.1 2.4 ± 0.2 2.3 ± 0.2
Myeloblastosis oncogene M12848 2.8 ± 0.4 5.6 ± 1.1 5.0 ± 0.2
K+ conductance calcium-activated channel N4 AF042487 3.1 ± 0.2 4.2 ± 0.6 3.5 ± 0.8
ATPase 6v1b2 AI843029 1.7 ± 0.1 1.8 ± 0.1 1.8 ± 0.1
Cystic fibrosis transmembrane regulator M60493 3.4 ± 0.5 4.5 ± 0.8 3.9 ± 0.3
RIKEN cDNA 1110008P14 AI839839 4.3 ± 0.4 6.0 ± 1.0 5.1 ± 0.2
Fused toes Z67963 2.6 ± 0.2 3.2 ± 0.3 2.8 ± 0.1
Solute carrier family 39a8 AW124340 3.5 ± 0.5 4.5 ± 0.7 3.8 ± 0.3
Cytochrome b-561 AI846517 2.2 ± 0.2 2.5 ± 0.2 2.3 ± 0.2
Secreted phosphoprotein 1 X13986 30.2 ± 3.3 47.4 ± 7.4 31.1 ± 2.7
Ion transport regulator Fxyd3 X93038 5.3 ± 0.4 6.8 ± 1.1 5.5 ± 0.6
Janus kinase 3 L40172 2.1 ± 0.2 2.5 ± 0.2 2.2 ± 0.2
Cytochrome b-245alpha AW046124 2.9 ± 0.4 3.6 ± 0.4 2.9 ± 0.2
RIKEN cDNA A430096B05 AI465965 6.3 ± 1.0 8.6 ± 0.03 6.3 ± 0.6
Small proline-rich protein 2J AJ005568 8.6 ± 2.1 13.8 ± 3.2 8.5 ± 0.7
Cathepsin B M65270 2.2 ± 0.1 2.6 ± 0.2 2.2 ± 0.1
RIKEN cDNA 1600025H15 AI842734 2.2 ± 0.1 2.7 ± 0.3 2.2 ± 0.2
c-fos oncogeneb V00727 3.2 ± 0.4 4.7 ± 0.9 3.7 ± 0.8
Guanine nucleotide binding protein γ5 AI843937 1.6 ± 0.1 1.8 ± 0.1 1.6 ± 0.03
Serine palmitoyltransferase lc2 U27455 1.6 ± 0.1 2.0 ± 0.2 1.7 ± 0.1
Cystatin B U59807 1.5 ± 0.1 1.7 ± 0.1 1.5 ± 0.02
Villin 2 X60671 1.9 ± 0.2 2.4 ± 0.3 1.9 ± 01
RIKEN cDNA 0610010O12 AI849011 1.9 ± 0.1 2.5 ± 0.3 1.9 ± 0.03
Matrix metalloproteinase 7 L36244 47.8 ± 18.6 208.6 ± 83.3 48.2 ± 11.9
RIKEN cDNA 4930422J18 AV376312 2.0 ± 0.3 2.9 ± 0.3 2.0 ± 0.2
RIKEN cDNA 1700017B05 AW049360 1.6 ± 0.1 2.0 ± 0.1 1.6 ± 0.1
Galactosidase beta 1 M57734 1.8 ± 0.1 2.0 ± 0.1 1.7 ± 0.1
Cathepsin C U74683 2.6 ± 0.1 3.3 ± 0.2 2.3 ± 0.3
Interferon-stimulated protein 15 X56602 3.6 ± 0.6 2.0 ± 0.2 3.8 ± 0.7
MAP kinase–interacting kinase 2 Y11092 1.8 ± 0.1 1.4 ± 0.1 1.9 ± 0.1
Glutathione S-transferase theta 2 X98056 3.5 ± 0.4 2.9 ± 0.4 3.6 ± 0.2
Gene name accession no. E2 GEN DES
Homeobox B6 M18401 1.5 ± 0.02 1.5 ± 0.1 1.6 ± 0.03
Procollagen VIalpha 3 AF064749 2.1 ± 0.2 1.9 ± 0.2 2.1 ± 0.02
Interferon regulatory factor 7 U73037 11.7 ± 0.9 8.1 ± 0.7 12.6 ± 1.5
Scavenger receptor class B2 AB008553 2.7 ± 0.1 2.4 ± 0.3 2.6 ± 0.2
Polyimmunoglobulin receptor AB001489 8.1 ± 0.7 6.2 ± 0.8 7.9 ± 1.0
Proteasome subunit β10 Y10875 2.1 ± 0.04 1.9 ± 0.04 2.1 ± 0.1
RIKEN cDNA 0610010E05 AV312736 2.9 ± 0.3 2.5 ± 0.2 2.7 ± 0.4
RIKEN cDNA 0610010E05 AI854839 3.7 ± 0.5 3.0 ± 0.1 3.4 ± 0.3
Xanthine dehydrogenase X75129 12.2 ± 2.2 8.9 ± 1.2 10.1 ± 1.5
Prominin AF039663 3.5 ± 0.2 2.9 ± 0.2 3.1 ± 0.3
Interferon-induced protein IFIT1 U43084 17.5 ± 2.9 9.9 ± 1.5 14.0 ± 2.1
Interferon-induced protein IFIT3 U43086 8.1 ± 2.3 4.7 ± 0.2 6.7 ± 0.6
Proteasome subunit β8 U22033 2.0 ± 0.1 1.8 ± 0.2 2.1 ± 0.1
RIKEN cDNA 1600023A02 AW121336 1.9 ± 0.1 1.7 ± 0.1 2.0 ± 0.04
Small proline-rich protein 1A AF057156 11.1 ± 3.7 8.6 ± 0.8 14.7 ± 0.8
MAP kinase-interacting kinase 2 AI845732 2.0 ± 0.1 1.7 ± 0.1 2.0 ± 0.2
Lymphocyte antigen 6 complex, locus E U47737 2.0 ± 0.01 1.7 ± 0.1 2.0 ± 0.1
Guanylate nucleotide binding protein 2 AJ007970 3.0 ± 0.1 2.0 ± 0.2 2.6 ± 0.1
Peptidyl arginine deiminase, type IIb D16580 1.9 ± 0.3 9.1 ± 0.3 5.2 ± 1.5
Down-regulated genes
Solute carrier family 29a1 AI838274 2.0 ± 0.2 2.9 ± 0.3 2.5 ± 0.1
Lymphocyte specific 1 D49691 1.6 ± 0.1 2.3 ± 0.2 1.8 ± 0.1
Claudin 5 U82758 2.0 ± 0.2 2.8 ± 0.1 2.7 ± 0.5
Potassium channel td12 AI842065 1.6 ± 0.04 2.0 ± 0.04 2.1 ± 0.1
Zinc finger homeobox 1a D76432 1.5 ± 0.1 1.7 ± 0.1 1.8 ± 0.01
Monoamine oxidase A AI848045 2.3 ± 0.2 2.7 ± 0.2 2.5 ± 0.4
Histidine decarboxylase X57437 4.8 ± 0.8 7.1 ± 0.6 5.4 ± 0.8
α-2 Adrenergic receptor M97516 3.0 ± 0.3 4.2 ± 1.1 3.6 ± 0.3
Transcription factor 21 AF035717 1.8 ± 0.2 2.2 ± 0.2 2.0 ± 0.1
Homeobox D8 X56561 2.2 ± 0.1 2.8 ± 0.03 2.5 ± 0.2
Carboxypeptidase X2 AF017639 4.1 ± 0.7 5.2 ± 1.0 4.2 ± 0.2
RIKEN cDNA A230106A15 AI848841 3.8 ± 0.2 4.7 ± 0.5 4.2 ± 0.7
Reduced expression 3 AA790008 3.1 ± 0.2 3.5 ± 0.3 3.2 ± 0.4
TGF-β binding protein 4 AA838868 1.8 ± 0.1 2.1 ± 0.1 1.8 ± 0.2
Keratoepithelinb L19932 11.5 ± 2.5 12.6 ± 0.9 9.7 ± 3.6
GLI-Kruppel family member GLI AB025922 11.6 ± 0.8 12.2 ± 2.6 8.2 ± 2.6
Abbreviations: CEA, carcinoembrionary antigen; SRC, steroid receptor coactivator; TGF, transforming growth factor.
a Gene names (derived from the NetAffx database; Liu et al. 2003), GenBank accession numbers (GenBank 2004), and mean (± SD) fold induction/repression of gene expression are shown in the same order as the gene cluster in Figure 3C.
b Genes mentioned in the text.
==== Refs
References
Aardema MJ Albertini S Arni P Henderson LM Kirsch-Volders M Mackay JM 1998 Aneuploidy: a report of an ECETOC task force Mutat Res 410 3 79 9587424
Akiyama T Ishida J Nakagawa S Ogawara H Watanabe S Itoh N 1987 Genistein, a specific inhibitor of tyrosine-specific protein kinases J Biol Chem 262 5592 5595 3106339
Almstrup K Fernandez MF Petersen JH Olea N Skakkebaek NE Leffers H 2002 Dual effects of phytoestrogens result in U-shaped dose–response curves Environ Health Perspect 110 743 748 12153753
Arora A Nair MG Strasburg GM 1998 Antioxidant activities of isoflavones and their biological metabolites in a liposomal system Arch Biochem Biophys 356 133 141 9705203
Casanova M You L Gaido KW Archibeque-Engle S Janszen DB Heck HA 1999 Developmental effects of dietary phytoestrogens in Sprague-Dawley rats and interactions of genistein and daidzein with rat estrogen receptors α and β in vitro Toxicol Sci 51 236 244 10543025
Cassidy A Bingham S Setchell KDR 1994 Biological effects of a diet of soy protein rich in isoflavones on the menstrual cycle of premenopausal women Am J Clin Nutr 60 333 340 8074062
Cassidy A Faughnan M 2000 Phyto-oestrogens through the life cycle Proc Nutr Soc 59 489 496 10997682
Committee on Toxicity of Chemicals in Food, Consumer Products and the Environment 2003. Phytoestrogens and Health. London:Food Standards Agency. Available: http://www.foodstandards.gov.uk/multimedia/pdfs/phytoreport0503 [accessed 2 June 2004].
Dean NM Kanemitsu M Boynton AL 1989 Effects of the tyrosine-kinase inhibitor genistein on DNA synthesis and phospholipid-derived second messenger generation in mouse 10T1/2 fibroblasts and rat liver T51B cells Biochem Biophys Res Commun 165 795 801 2532009
Duarte J Ocete MA Perez-Vizcaino F Zarzuelo A Tamargo J 1997 Effect of tyrosine kinase and tyrosine phosphatase inhibitors on aortic contraction and induction of nitric oxide synthase Eur J Pharmacol 338 25 33 9408000
Frasor J Barnett DH Danes JM Hess R Parlow AF Katzenellenbogen BS 2003 Response-specific and ligand dose-dependent modulation of estrogen receptor (ER) α activity by ERβ in the uterus Endocrinology 144 3159 3166 12810572
Genbank 2004. Bethesda, MD:National Center for Biotechnology Information, National Library of Medicine. Available: http://www.ncbi.nlm.nih.gov/Genbank/GenbankOverview.html [accessed 2 June 2004].
GEO 2004. Gene Expression Omnibus Homepage. Bethesda, MD:National Center for Biotechnology Information, National Library of Medicine. Available: http://www.ncbi.nlm.nih.gov/geo/ [accessed 2 June 2004].
Hall JM Couse JF Korach KS 2001 The multifaceted mechanisms of estradiol and estrogen receptor signaling J Biol Chem 276 36869 36872 11459850
Herman-Giddens ME Slora EJ Wasserman RC Bourdony CJ Bhapkar MV Koch GG 1997 Secondary sexual characteristics and menses in young girls seen in office practice: a study from the Pediatric Research in Office Settings network Pediatrics 99 505 512 9093289
Hewitt SC Deroo BJ Hansen K Collins J Grissom S Afshari CA 2003 Estrogen receptor-dependent genomic responses in the uterus mirror the biphasic physiological response to estrogen Mol Endocrinol 17 2070 2083 12893882
Holm S 1979 A simple sequentially rejective Bonferroni test procedure Scand J Stat 6 65 70
Howdeshell KL Hotchkiss AK Thayer KA Vandenbergh JG vom Saal FS 1999 Exposure to bisphenol A advances puberty Nature 401 763 764 10548101
Kuiper GG Lemmen JG Carlsson B Corton JC Safe SH van der Saag PT 1998 Interaction of estrogenic chemicals and phytoestrogens with estrogen receptor β Endocrinology 139 4252 4263 9751507
Liu G Loraine AE Shigeta R Cline M Cheng J Valmeekam V 2003 NetAffx: Affymetrix probesets and annotations Nucleic Acids Res 31 82 86 12519953
Lund TD Munson DJ Haldy ME Setchell KD Lephart ED Handa RJ 2004 Equol is a novel anti-androgen that inhibits prostate growth and hormone feedback Biol Reprod 70 1188 1195 14681200
Messina MJ 1999 Legumes and soybeans: overview of their nutritional profiles and health effects Am J Clin Nutr 70 439S 450S 10479216
Moggs JG Orphanides G 2001 Estrogen receptors: orchestrators of pleiotropic cellular responses EMBO Rep 9 775 781 11559590
Naciff JM Jump ML Torontali SM Carr GJ Tiesman JP Overmann GJ 2002 Gene expression profile induced by 17α-ethynyl estradiol, bisphenol A, and genistein in the developing female reproductive system of the rat Toxicol Sci 68 184 199 12075121
Naciff JM Overmann GJ Torontali SM Carr GJ Tiesman JP Richardson BD 2003 Gene expression profile induced by 17α-ethynyl estradiol in the prepubertal female reproductive system of the rat Toxicol Sci 72 314 330 12655037
Newbold RR Banks EP Bullock B Jefferson WN 2001 Uterine adenocarcinoma in mice treated neonatally with genistein Cancer Res 61 4325 4328 11389053
North K Golding J 2000 A maternal vegetarian diet in pregnancy is associated with hypospadias Br J Urol Int 85 107 113
Odum J Lefevre PA Tittensor S Paton D Routledge ED Beresford NA The rodent uterotrophic assay: critical protocol features, studies with nonyl phenols, and comparison with a yeast estrogenicity assay Regul Toxicol Pharmacol 25 176 188 9185893
Okura A Arakawa H Oka H Yoshinari T Monden Y 1988 Effect of genistein on topoisomerase activity and on the growth of [Val 12]Ha-ras -transformed NIH 3T3 cells Biochem Biophys Res Commun 157 183 189 2848517
Orphanides G Moggs JG Spurway T Tinwell H Ashby J Kimber I 2003 Use of gene expression profiling to understand the transcriptional programme associated with estrogen-induced uterine growth: implications for the use of surrogate molecular markers in toxicology [Abstract] Toxicol Sci 72 12
Owens JW Ashby J 2002 Critical review and evaluation of the uterotrophic bioassay for the identification of possible estrogen agonists and antagonists: in support of the validation of the OECD uterotrophic protocols for the laboratory rodent. Organisation for Economic Co-operation and Development Crit Rev Toxicol 32 445 520 12487363
Paulozzi LJ Erickson JD Jackson RJ 1997 Hypospadias: trends in two US surveillance systems Pediatrics 100 831 834 9346983
Potier B Rovira C 1999 Protein tyrosine kinase inhibitors reduce high-voltage activating calcium currents in CA1 pyramidal neurones from rat hippocampal slices Brain Res 816 587 597 9878884
Safe SH Pallaroni L Yoon K Gaido K Ross S McDonnell D 2002 Problems for risk assessment of endocrine-active estrogenic compounds Environ Health Perspect 110 925 929 12634121
Setchell KDR Zimmer-Nechemias L Cai J Heubi JE 1997 Exposure of infants to phyto-oestrogens from soy-based infant formula Lancet 350 23 27 9217716
Tinwell H Ashby J 2004 Sensitivity of the immature rat uterotrophic assay to mixtures of estrogens Environ Health Perspect 112 575 582 15064164
Watanabe H Suzuki A Kobayashi M Lubahn DB Handa H Iguchi T 2003 Similarities and differences in uterine gene expression patterns caused by treatment with physiological and non-physiological estrogens J Mol Endocrinol 3 487 497 14664709
Weihua Z Saji S Mäkinen S Cheng G Jensen EV Warner M 2000 Estrogen receptor (ER) β, a modulator of ERα in the uterus Proc Natl Acad Sci USA 97 5936 5941 10823946
Williams GM Iatropoulos M Cheung R Radi L Wang CX 1993 Diethylstilbestrol liver carcinogenicity and modification of DNA in rats Cancer Lett 68 193 198 8443792
|
15289156
|
PMC1247471
|
CC0
|
2021-01-04 23:40:08
|
no
|
Environ Health Perspect. 2004 Aug 19; 112(11):1137-1142
|
utf-8
|
Environ Health Perspect
| 2,004 |
10.1289/ehp.7028
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences 10.1289/ehp.7122ehp0112-00114315289157ResearchCommentariesUsing Human Disease Outbreaks as a Guide to Multilevel Ecosystem Interventions Cook Angus Jardine Andrew Weinstein Philip School of Population Health, University of Western Australia, Perth, AustraliaAddress correspondence to A. Cook, School of Population Health, M431, University of Western Australia, 35 Stirling Highway, Crawley, 6009 Western Australia. Telephone: 61-8-6488-7804. Fax: 61-8-6488-1188. E-mail:
[email protected] authors declare they have no competing financial interests.
8 2004 27 5 2004 112 11 1143 1146 26 3 2004 27 5 2004 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Human health often depends on environmental variables and is generally subject to widespread and comprehensive surveillance. Compared with other available measures of ecosystem health, human disease incidence may be one of the most useful and practical bioindicators for the often elusive gauge of ecologic well-being. We argue that many subtle ecosystem disruptions are often identified only as a result of detailed epidemiologic investigations after an anomalous increase in human disease incidence detected by routine surveillance mechanisms. Incidence rates for vector-mediated diseases (e.g., arboviral illnesses) and direct zoonoses (e.g., hantaviruses) are particularly appropriate as bioindicators to identify underlying ecosystem disturbances. Outbreak data not only have the potential to act as a pivotal warning system for ecosystem disruption, but may also be used to identify interventions for the preservation of ecologic health. With this approach, appropriate ecologically based strategies for remediation can be introduced at an earlier stage than would be possible based solely on environmental monitoring, thereby reducing the level of “ecosystem distress” as well as resultant disease burden in humans. This concept is discussed using local, regional, and global examples, thereby introducing the concept of multilevel ecosystem interventions.
bioindicatorsdisease controldisease outbreaksecologic managementecosystem healthsurveillance
==== Body
During the construction of the Panama Canal in the 1880s, continuous outbreaks of yellow fever killed > 5,500 workers, or > 6% of the workforce. The most immediate cause of those outbreaks was, then as now, obvious and was captured in the records of artist-to-be Paul Gauguin, who was then a digger with the French Canal Company: “At night I am devoured by mosquitoes” (Harrison 1978). The ultimate cause of these outbreaks was more complex, however, involving disruptions to both environment and society, mediated by a range of political and economic drivers.
Models have been developed to describe the way in which such interacting disruptions influence health (Parkes and Weinstein 2004), but from a biophysical perspective one of the more constructive lines of analysis is directed toward the disruption of the immediate ecosystem. During the Panama Canal construction, it is obvious (with the wisdom of ecologic hindsight) that replacing a rainforest environment with an urban/industrial environment offers the opportunity for container-breeding vector mosquitoes to proliferate and to transmit disease at a scale never before encountered in an affected area. By today’s standards, this disease outbreak and the associated ecosystem disturbance might seem to have followed a relatively obvious path. Nevertheless, despite a greater contemporary understanding of microbiologic and ecologic dynamics, insults to the environment occurring even in modern times are often discovered only as a result of detailed outbreak investigation.
Measurable bioindicators of ecosystem health were first described in detail by Rapport et al. (1985). These include changes in nutrient cycling, decreased species diversity as a result of decreasing habitat diversity, retrogression (a reversal of the normal process of species succession as the ecologic community is simplified), and increased fluctuations in population size. Presence of disease also explicitly formed one of the bioindicators, and it was suggested that increased disease incidence among plant, animal, and human populations would manifest as the fabric of the ecosystem begins to deteriorate and natural buffering and protective mechanisms break down.
The intrinsic link between ecosystem health and human disease (especially vector-mediated disease) has been discussed in a number of previous publications (Cassis 1998; Chivian 2001; Epstein 1995; Forget and Lebel 2001; Haines et al. 2000; McMichael 1997; Nielsen 2001; VanLeeuwen et al. 1999; Waltner-Toews 2001). These authors have noted that ecosystem health is heavily influenced by human activities and that, vice versa, human health depends on proper ecosystem functioning. Reflecting this close relationship, it has been suggested that disease incidence within a human population can be used as a bioindicator or “yardstick” of the health of the ecosystem of which the community is a part (Rapport 1999).
We concur with this concept and advance it two steps further by contending a) that human disease incidence is in fact one of the most useful and practical bioindicators of the health of an ecosystem and b) that using human health as a bioindicator in this way can assist in guiding rapid and appropriate ecosystem interventions. A major advantage in using disease outbreaks as bioindicators of even subtle ecosystem disruptions is that the health of human populations is generally subject to more widespread and accurate surveillance than is ecosystem health (Spiegel and Yassi 1997). Many sources of data, such as data obtained from disease registries, infectious disease notification systems, and hospitalizations, provide ongoing measurement and monitoring of human communities. It should be emphasized that we are not advocating that information on human suffering should simply be used to better preserve the environment. Our approach very much supports dual end points: early and appropriate minimization of ecologic degradation in its own right, with the major consequence that this is the pathway by which we will preserve the public health for communities living in these environments.
The incidence data most useful in signaling underlying ecosystem processes relate to vector-mediated diseases (e.g., arboviral illnesses), direct zoonoses (e.g., hantaviruses), and infections that appear to transcend simple transmission categories [e.g., viruses that were zoonotic but “transformed” to direct anthroponoses, such as SARS (severe acute respiratory virus) and HIV (human immunodeficiency virus)]. A number of direct anthroponoses (i.e., disease spread by direct human-to-human transmission, such as measles, polio, and chlamydia), reflect human dynamics such as crowding and sexual contact, so their roles as ecosystem bioindicators are less likely to be pertinent. However, ecologic disruption may act as an indirect or partial determinant even for some of these infections (e.g., as in the case of cholera transmission; Tauxe et al. 1994).
As we outline in the examples below, many stresses and disruptions to natural ecosystem functioning are identified only as a result of detailed epidemiologic investigations, which in turn follow an increase in human disease incidence detected by routine surveillance. By identifying ecosystem disruptions that affect human health using this outbreak-based approach, appropriate strategies for intervention and remediation can be introduced at an earlier stage than would be possible based solely on environmental monitoring. Although it is still possible to detect ecosystem disruption using traditional environmentally based bioindicators of ecosystem health, this generally requires a detailed and complex investigation in terms of the cost and feasibility of obtaining and analyzing valid and consistent data (Patil et al. 2001; Rapport et al. 1995). The interpretation of ecologic indicators obtained can also be problematic, and the results may be difficult to convey to policy makers and a broader public (Schaeffer 1996). To illustrate our argument, we discuss examples of disease outbreaks that have led to the identification of ecosystem disruptions and the appropriate corresponding ecosystem interventions at local, regional, and global levels using the comprehensive framework provided by the Millennium Ecosystem Assessment (2003) as a basis.
Ecosystem Interventions for Local Outbreaks
The paradigm that draws together human outbreaks (as identified by, e.g., disease surveillance data) and environmental disruption will first be discussed in relation to local ecologic transformations. Among the best-documented examples is deforestation, which is often accompanied by ecologic simplification toward either monoculture or subsistence agriculture. The outbreaks of monkeypox in Zaire and hantaviruses in the Americas have both acted as clear bioindicators for disruption of the local distribution of natural vegetation. The clearance and replacement of complex rainforest, such as through slash-and-burn clearing practices, have encouraged a massive proliferation of small animals (e.g., rodents and squirrels) that act as vectors for both of these viral diseases (Glass et al. 2000; Khodakevich et al. 1988). Compared with the cases of human disease that arose as a result, few other routinely obtained and readily interpretable bioindicators were able to alert the authorities and general public to the extent of the underlying ecologic process. Ecosystem interventions that were readily suggested were to limit the removal of forest and impingement of human communities on the habitats of virus-carrying mammals.
Another example of the link between localized outbreaks as an indicator for ecosystem health is provided by Ross River virus (RRV) patterns in Australia. RRV is the most common arboviral infection in Australia—with 52,053 laboratory-diagnosed cases reported from when reporting began in 1992 until the end of 2003 (Communicable Diseases Network Australia 2004)—and is characterized by traditional rheumatic joint manifestations, rash and constitutional effects, and more recently described presentations including glomerulonephritis (Selden and Cameron 1996). Distinct seasonal epidemic activity is observed in northern tropical regions during summer months when rainfall is highest. In some areas virus activity may persist year-round, but winter rainfall in tropical regions is generally insufficient to support vector breeding and thus limits transmission during the dry season (Russell 2002).
This typical disease pattern in the tropical northeast Kimberley region of western Australia has undergone recent changes. Unusual dry-season cases of RRV disease led to suspicions that ongoing development of the Ord River Irrigation Area had disrupted the local natural ecosystem to the extent that mosquitoes were now able to breed year-round. A subsequent entomologic investigation during August (usually the driest month of the year) confirmed that mosquito breeding was indeed occurring in the dry season (Jardine et al., in press). Significantly larger numbers of adult and larval Culex annulirostris, an important vector of RRV and a range of other arboviruses in Australia, were collected within the irrigation area compared with nonirrigated reference areas.
These disruptions to the dry-season ecology of the mosquito fauna in the area were therefore detected only as a result of an investigation sparked by an unusual outbreak of human disease. Routine mosquito surveys, which might have acted as an alternative bioindicator, are simply not carried out in the dry season. These findings suggest that appropriate ecosystem strategies to reduce breeding of disease vector mosquitoes should focus primarily on restoring local hydrology to reduce potential mosquito breeding habitats. In particular, the ability of mosquitoes to breed year-round means control activities must be ongoing and not restricted to a few months during the peak of the wet season (Jardine et al., in press).
Ecosystem Interventions for Regional Outbreaks
Illness patterns at a wider, regional level may also relate to and act as telltale signs for disturbances of usual ecologic processes. The recent epidemic of new variant form of Creutzfeldt–Jakob disease from beef consumption led to the identification of a little-recognized ecologic anomaly, with investigations into the source of the disease revealing that animals that are naturally herbivorous—beef cattle—were being transformed into carnivores through the introduction of meat and bone meal in their feed (Wilesmith et al. 1988, 1991, 1992). For regions that engage in such practices, the early intervention suggested was to reverse this food-chain anomaly by banning material extracted from other mammals in cattle feed (Nathanson et al. 1997).
A more complex food-chain disruption was highlighted by the increased incidence of Lyme disease in the northeastern United States during the last two decades. The “emergence” of this disease led to detailed ecologic studies of vector ticks in the genus Ixodes, which transmit the pathogen Borrelia burgdorferi from white-footed mice to humans. In a healthy ecosystem, a variety of small mammals are available for these ticks to feed on, and most of these hosts do not carry Borrelia spirochetes. Thus, the proportion of infected ticks is small, and the probability of Borrelia transmission to humans is low. However, with the growing disruption of regional ecosystems, the diversity of small mammals decreased and was replaced by burgeoning numbers of white-footed mice (LoGiudice et al. 2003). This mouse species, which multiplied and invaded the niches vacated by more sensitive animals, in turn became the more common host for the ticks. To complicate matters further, more mature ticks feed on deer, which have in turn proliferated because of the removal of major predators (wolves) from the food chain in these areas. The net result is a larger tick population with a higher percentage of infected reservoir species, all more likely to infect the increasing numbers of humans impinging on a once pristine regional ecosystem. An appropriate regional intervention suggested is restoration of biodiversity in such ecosystems, which would reduce the abnormal proliferation of white-footed mice, deer, ticks, and reservoirs for Borrelia (Wilson 2002).
The link between regional outbreaks and ecosystem change is further illustrated by the periodic emergence of ciguatera fish poisoning. Ciguatera, linked to toxic marine dinoflagellates (Gambierdicus toxicus), is a syndrome characterized by acute gastroenteritis and neurologic symptoms (including inverted temperature perception, an odd symptom whereby cold objects appear hot to touch and vice versa). The severity of poisoning ranges from imperceptibly mild to rapidly lethal, and there is generally a history of cases having consumed tropical reef fish. In Pacific Island countries that rely on fish as a major source of protein, ciguatera poisoning is the cause of a significant disease burden (Laurent et al. 1993), and anecdotal evidence suggested that this burden increased through the late 1990s. Although subsequent investigations demonstrated a correlation between sea surface temperature and the number of cases, particularly on islands strongly influenced by El Niño climatic conditions (Hales et al. 1999), dinoflagellate proliferation is probably most enhanced by physical disturbances to coral reef ecosystems. When reefs are blasted (e.g., for the coral trade) or suffocated (by runoff from deforested hillsides), massive coral death occurs, creating extensive substrates for the growth of macroalgae (Kohler and Kohler 1992). It is on the surface of these macroalgae that ciguatera-causing dinoflagellates grow, and damaged coral ecosystems may therefore be particularly productive of toxic fish. Hales et al. (1999) based their study into the underlying ecosystem disturbances accounting for ciguatera on data collected as part of routine health surveillance; no equivalent environmental monitoring data were available on the “health” of coral reef ecosystems around the Pacific. Again, the findings also suggested possible regional solutions, including remediation of selected coral reef ecosystems, such as by reforestation of hillsides to limit runoff and avoidance of blasting, especially on those islands where most ciguatera cases are occurring.
Ecosystem Interventions for Global Outbreaks
In particular circumstances, the consequences of ecosystem impingement and disruption may become apparent on a global scale. One pathogen whose emergence, with devastating consequences, was driven partly by ecosystem distress is HIV/AIDS. This retrovirus has overwhelmed the communities of many countries, including those that were already highly socially and economically vulnerable (Tinker 1988). It took some years for the origins of the virus to become apparent, but most now believe that it originated from simian reservoirs: probably chimpanzees for HIV-1 and sooty mangabeys for HIV-2 (Gao et al. 1999). The stage of transferal to human populations most likely occurred with the practice of using primates as a food source. Indeed, before the disease expanded to such a devastating level, neither the scale nor implications of “bush meat” practices were fully acknowledged (Tutin 2000). Such dietary practices have a potential capacity to transmit other retroviruses, of which at least 20 simian forms have been identified (Dalgleish and Weiss 1999). Indeed, Wolfe et al. (2004) have confirmed zoonotic infections with simian foamy virus in residents of central African forests who reported direct contact with blood and body fluids of wild nonhuman primates.
The implications for preserving ecosystem health suggested by the example of the global HIV/AIDS outbreak are clear. One response to the threat is to reduce any further risk of simian retroviral transmission by responding to the ecologic disruptions that HIV/AIDS brought to light. However, to minimize the infective risk of simian retroviral infections to the general population, remedial measures must occur in those original environments from which emergence occurred: rainforests or other habitats in which primates thrive (Bisong 1999). In other words, the optimal ecosystem interventions required to limit retroviral spread are quite geographically remote from most of the global population even though they comprise the majority of people who would probably suffer the consequences of a further HIV-like outbreak. Most options to further reduce rainforest penetration in the remote parts of other continents would entail a considerable degree of operational complexity and impinge on the sovereign rights of countries to manage their own ecosystems (often in the face of profound poverty). Ecosystem health—and ultimately, human health—might be served only by simultaneously addressing the socioeconomic deprivation that drives forest clearances and consumption of primates and encouraging alternative sources of cropping and land management (Stephens et al. 2002).
The opportunity for multiple approaches to ecosystem intervention is clearly evident for the arboviral disease dengue fever. The occurrence of one or more of the dengue serotypes across most tropical regions of the world (Wilson and Chen 2002) reflects ecosystem disturbances at multiple levels in a manner that few traditional bioindicators could capture. Unlike the threat posed by simian retroviruses, which may respond to local actions to reduce bushmeat contact, multiple ecosystem interventions are suggested for dengue control that operate at numerous, often overlapping levels: locally, to remove artificial breeding habitats for the Aedes mosquito vectors that are provided by containers (Knudsen 1995; Moore et al. 1990; Tauil 2001); regionally, to limit the disruption of waterways (which encourage stagnation and high nutrient loads) (Forattini et al. 2001); and globally, to minimize the effects of global warming that encourages mosquito breeding for longer durations at a wider range of latitudes (Chan et al. 1999; Hales et al. 2002; Hopp and Foley 2003). The climatic instability associated with the warming trend may also drive excess rainfall and flooding in many areas, thus again providing ideal breeding sites. Thus, information about outbreaks of dengue fever, reliably monitored in many countries, can inform an integrated approach—operating at three levels—to the management ecosystem disturbances.
Discussion
We live in an era of emerging and reemerging infectious disease attributable to ecosystem disruptions (Weinhold 2004), a phase that has been termed the third epidemiologic transition (Barrett et al. 1998; McMichael 1993). Given the threats to health in this modern milieu, understanding and assessing the links between anthropogenic pressure on ecosystems, human health, and ecosystem structure and functioning are vitally important (Koren and Crawford-Brown 2004). Although currently “there is no simple solution to a quantitative and quick assessment of ecosystem health” (Ramade 1995), we contend that human disease surveillance (particularly notification systems for infectious disease) at local, regional, and global levels is often a readily available and accurately recorded bioindicator that could be used for such purposes. Monitoring of disease events is more widespread, accurate, and subject to ongoing quality assurance than many of the “indicators of ecosystem health” that have been proposed in the past (Spiegel and Yassi 1997), which are often difficult to routinely measure and which require intensive investigation and complex analysis (Rapport et al. 1995). A similar argument could be mounted for other such diffuse ecosystem measures such as “vitality,” “vigor,” and “resilience” (Mageau et al. 1995). Despite their conceptual appeal, these indicators do not lend themselves to routine assessment or the rapid development of possible intervention strategies.
It is important to note, however, that our advocacy of using outbreak data for the purposes outlined in this article does not suggest that such information necessarily should be used as a direct substitute for alternative ecologic measurements. Nor do we imply that conclusions drawn from epidemiologic analysis somehow invalidate those derived from other systems of ecologic monitoring. Many ecologic measures pertain to the health of other (i.e., nonhuman) organisms or systems or may act below the threshold by which the overt appearance of infectious disease in humans may occur. Some environmental agents also operate to cause disease in other or more gradual mechanisms, as in the case of carcinogens or teratogens.
Rather than relying solely on human disease incidence as a bioindicator, we acknowledge that in many situations standard measures of ecosystem health may be entirely synergistic and complementary to outbreak data (Rapport 1999). For example, one sampling strategy that could be successfully integrated with the use of outbreak data is monitoring the abundance and distribution of synanthropes and other organisms that act as intermediaries for human disease. For example, rodent, mosquito, and algal populations not only reflect the potential for transmission, but in themselves may function as integrative indicators of ecosystems function.
Furthermore, there are limitations in the use of outbreak data as a measure of ecosystem disruption that must be recognized. First, communicable disease monitoring and surveillance are clearly less useful options in regions with a low human population density (e.g., circumpolar regions) or where incidence data are erratically obtained, unreliable, or simply not collected. Second, many diseases show underlying variation independent of ecosystem disruption. For example, increases may relate solely to seasonality or other cyclical patterns (e.g., measles outbreaks secondary to human immunity dynamics). Although in such situations the effects of transformed ecosystem may be absent, it is important to consider that ecologic disturbances may also overlie or distort baseline fluctuations (e.g., through acting to enhance “normal” mosquito breeding seasons). As discussed, diseases transmitted principally by direct human-to-human contact (e.g., measles, varicella) are less likely to be affected by ecologic change. However, the degree to which environmental changes contribute or underlie even direct anthropologic disease patterns is becoming increasingly apparent, especially those linked to compromised water supplies and poor sanitation (as illustrated by the relationships between climate change, flooding, and diseases such as cholera and dysentery; e.g., McMichael 1997).
Conclusion
Burger and Gochfeld (2001) highlight the need for development of bioindicators that can be used for the integrated assessment of both ecologic and human health, and that these must be easily measured and understood, be cost-effective, and have direct societal relevance to gain long-term support. It would appear that human disease incidence meets all of these requirements and, despite certain inherent limitations, can be used for early identification of ecologic disruption.
This process facilitates early intervention, which in turn can decrease the level of “ecosystem distress” and the resultant disease burden in humans. Human disease surveillance pathways could therefore help define areas at ecologic risk (Weinstein et al. 1994). This process would capitalize on an existing health infrastructure that must remain intact in any case if our societies are to maintain the public health gains of the last century. The local, regional, and global levels of our approach will also encourage a renewed perspective of environmental problems. Policy makers and public health officials are often inclined to consider the underlying drivers of ecologic and human health in local or, at best, regional terms. This truncated approach is becoming less and less relevant to current problems. To understand many current environmental health issues, our multilevel approach endorses a move toward a multilevel paradigm, such as those promulgated by the Millennium Ecosystem Assessment (2003) and authors such as Aron and Patz (2001).
The desired goal of the multilevel approach to outbreak data is for both ecologic and human health to be enhanced. This discussion provides further evidence of the undesirability of artificially separating the well-being and viability of communities from that of the biosphere. Outbreak data can act as a pivotal warning system for ecosystem injury and may also be used to guide logical interventions for the simultaneous preservation of ecologic and human health at the local, regional, and global levels. Our recommendation is to acknowledge and exploit the strengths of using human disease surveillance for these purposes.
==== Refs
References
Aron JL Patz J 2001. Ecosystem Change and Public Health: A Global Perspective. Baltimore, MD:Johns Hopkins University Press.
Barrett R Kuzawa CW McDade T Armelagos GJ 1998 Emerging and re-emerging infectious diseases: the third epidemiologic transition Annu Rev Anthropol 27 247 271
Bisong FE 1999 Deforestation and the erosion of biodiversity in the Cross River State rainforest, south-eastern Nigeria Global J Pure Appl Sci 5 241 250
Burger J Gochfeld M 2001 On developing bioindicators for human and ecological health Environ Monit Assess 66 23 46 11214446
Cassis G 1998 Biodiversity loss: a human health issue Med J Aust 169 568 569 9887893
Chan NY Ebi KL Smith F Wilson TF Smith AE 1999 Assessment framework for climate change and infectious diseases Environ Health Perspect 107 329 337 10210687
Chivian E 2001 Environment and health. 7. Species loss and ecosystem disruption—the implications for human health CMAJ 164 66 69 11202670
Communicable Diseases Network Australia 2004. National Notifiable Diseases Surveillance System. Canberra ACT:Communicable Diseases Network Australia. Available: http://www1.health.gov.au/cda/Source/Rpt_4_sel.cfm [accessed 15 March 2004].
Dalgleish A Weiss RA 1999. HIV and the New Viruses. 2nd ed. London:Academic Press.
Epstein PR 1995 Emerging diseases and ecosystem instability: new threats to public health Am J Public Health 85 168 172 7856775
Forattini OP Kakitani I Ueno HM 2001 Emergence of Aedes albopictus in artificial containers Rev Saude Publica 35 456 460 11723517
Forget G Lebel J 2001 An ecosystem approach to human health Int J Occup Environ Health 7 S3 S38 11387989
Gao F Bailes E Robertson DL Chen Y Rodenburg CM Michael SF 1999 Origin of HIV-1 in the chimpanzee Pan troglodytes Nature 397 436 441 9989410
Glass G Cheek J Patz JA Shields TM Doyle TJ Thoroughman DA 2000 Predicting high risk areas for hantavirus pulmonary syndrome with remotely sensed data: the Four Corners outbreak, 1993 J Emerg Infect Dis 6 239 246
Haines A McMichael AJ Epstein PR 2000 Environment and health. 2. Global climate change and health Can Med Assoc J 163 729 734 11022589
Hales S de Wet N Maindonald J Woodward A 2002 A potential effect of population and climate changes on global distribution of dengue fever: an empirical model Lancet 360 830 834 12243917
Hales S Weinstein P Woodward A 1999 Ciguatera fish poisoning, El Niño, and Pacific sea surface temperatures Ecosyst Health 5 20 25
Harrison G 1978. Mosquitoes, Malaria and Man. New York:E.P. Dutton.
Hopp MJ Foley J 2003 Worldwide fluctuations in dengue fever cases related to climate variability Clim Res 25 85 94
Jardine A Lindsay MDD Heyworth J Weinstein P In press. Dry-season mosquito breeding associated with irrigation in the north-east Kimberley region of Western Australia: potential impact on mosquito-borne disease transmission. EcoHealth.
Khodakevich L Jezek Z Messinger D 1988 Monkeypox virus: ecology and public health significance Bull WHO 66 747 752 2853010
Knudsen AB 1995 Global distribution and continuing spread of Aedes albopictus Parassitologia 37 91 97 8778670
Kohler ST Kohler CC 1992 Dead bleached coral provides new surfaces for dinoflagellates implicated in ciguatera fish poisonings Environ Biol Fish 35 413 416
Koren HS Crawford-Brown D 2004 A framework for the integration of ecosystem and human health in public policy: two case studies with infectious agents Environ Res 95 92 105 15068935
Laurent D Bourdy G Amade P Cabalion P Bourret D 1993. La Gratte ou Ciguatera: Ses Remèdes Traditionnels dans le Pacifique Sud [in French]. Paris:ORSTOM.
LoGiudice K Ostfeld RS Schmidt KA Keesing F 2003 The ecology of infectious disease: effects of host diversity and community composition on Lyme disease risk Proc Natl Acad Sci USA 100 567 571 12525705
Mageau MT Costanza R Ulanowicz RE 1995 The development and initial testing of a quantitative assessment of ecosystem health Ecosyst Health 1 201 213
McMichael AJ 1993. Planetary Overload. Cambridge, UK:Cambridge University Press.
McMichael AJ 1997 Healthy world, healthy people People Planet 6 6 9 12321048
Millennium Ecosystem Assessment 2003. Ecosystems and Human Well-being: A Framework for Assessment. Washington, DC:World Resources Institute.
Moore CG Francy DB Eliason DA Bailey RE Campos EG 1990 Aedes albopictus and other container-inhabiting mosquitoes in the United States: results of an eight-city survey J Am Mosq Control Assoc 6 173 178 2370523
Nathanson N Wilesmith J Griot C 1997 Bovine spongiform encephalopathy (BSE): causes and consequences of a common source epidemic Am J Epidemiol 145 959 969 9169904
Nielsen NO 2001 Ecosystem approaches to human health Cad Saude Publica 17 suppl 69 75 11426267
Parkes M Weinstein P 2004. An ecosystems approach to environmental health. In: Environmental Health in Australia and New Zealand (Cromar N, Cameron S, Fallowfield H, eds). Melbourne, Australia:Oxford University Press, 45–65.
Patil GP Brooks RP Myers WL Rapport DJ Taillie C 2001 Ecosystem health and its measurement at landscape scale: toward the next generation of quantitative assessments Ecosyst Health 7 307 316
Ramade F 1995. Qualitative and quantitative criteria defining a “healthy” ecosystem. In: Evaluating and Monitoring the Health of Large-Scale Ecosystems (Rapport DJ, Gaudet CL, Calow P, eds). NATO ASI Series I: Global Environmental Change, Vol 28. Berlin:Springer-Verlag, 43–61.
Rapport DJ 1999 Epidemiology and ecosystem health: natural bridges Ecosyst Health 5 174 180
Rapport DJ Gaudet CL Calow P eds. 1995. Evaluating and Monitoring the Health of Large-Scale Ecosystems. NATO ASI Series I: Global Environmental Change, Vol 28. Berlin:Springer-Verlag.
Rapport DJ Regier HA Hutchinson TC 1985 Ecosystem behavior under stress Am Nat 125 617 640
Russell RC 2002 Ross River virus: ecology and distribution Annu Rev Entomol 47 1 31 11729067
Schaeffer DJ 1996 Diagnosing ecosystem health Ecotoxicol Environ Saf 34 1 18 34 8793317
Selden SM Cameron AS 1996 Changing epidemiology of Ross River virus disease in South Australia Med J Aust 165 313 317 8862330
Spiegel J Yassi A 1997 The use of health indicators in environmental assessment J Med Syst 21 275 289 9507404
Stephens PA Frey-Roos FA Sutherland WJ 2002 Sustainable exploitation of social species: a test and comparison of models J Appl Ecol 39 629 642
Tauil PL 2001 Urbanization and dengue ecology Cad Saude Publica 17 suppl 99 102 11426270
Tauxe R Seminario L Tapia R Libel M 1994. The Latin American epidemic. In: V. cholerae and Cholera: Molecular to Global Persepctives (Wachsmuth K, Blake PA, Olsvok O, eds). Washington, DC:American Society of Microbiology, 237–292.
Tinker J 1988 AIDS in developing countries Iss Sci Tech 4 1 7
Tutin CE 2000 Ecology and social organization of African tropical forest primates: aid in understanding retrovirus transmission Bull Soc Pathol Exot Filiales 93 157 161
VanLeeuwen JA Waltner-Toews D Abernathy T Smit B 1999 Evolving models of human health toward an ecosystem context Ecosyst Health 5 204 219
Waltner-Toews D 2001 An ecosystem approach to health and its applications to tropical and emerging diseases Cad Saude Publica 17 suppl 7 22 11426254
Weinhold B 2004 Infectious disease: the human costs of our environmental errors Environ Health Perspect 112 A32 A39 14698949
Weinstein P Worswick D Macintyre I Cameron AS 1994 Human sentinels for arbovirus surveillance and regional risk classification in South Australia Med J Aust 160 494 499 8170425
Wilesmith JW Ryan JB Atkinson MJ 1991 Bovine spongiform encephalopathy: epidemiological studies on the origin Vet Rec 128 199 203 1823120
Wilesmith JW Ryan JB Hueston WD 1992 Bovine spongiform encephalopathy: case-control studies of calf feeding practices and meat and bonemeal inclusion in proprietary concentrates Res Vet Sci 52 325 331 1620965
Wilesmith JW Wells GA Cranwell MP Ryan JB 1988 Bovine spongiform encephalopathy: epidemiological studies Vet Rec 123 638 644 3218047
Wilson ME 2002 Prevention of tick-borne diseases Med Clin North Am 86 219 238 11982299
Wilson ME Chen LH 2002 Dengue in the Americas Dengue Bull 26 44 61
Wolfe ND Switzer WM Carr JK Bhullar VB Shanmugam V Tamoufe U 2004 Naturally acquired simian retrovirus infections in central African hunters Lancet 363 9413 932 937 15043960
|
15289157
|
PMC1247472
|
CC0
|
2021-01-04 23:40:09
|
no
|
Environ Health Perspect. 2004 Aug 27; 112(11):1143-1146
|
utf-8
|
Environ Health Perspect
| 2,004 |
10.1289/ehp.7122
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences 10.1289/ehp.6705ehp0112-00114715289158ResearchArticlesDeterminants of Bone and Blood Lead Levels among Minorities Living in the Boston Area Lin Charles 1Kim Rokho 234Tsaih Shirng-Wern 5Sparrow David 36Hu Howard 341School of Medicine, University of California at San Francisco, San Francisco, California, USA2Occupational Health Surveillance Program, Massachusetts Department of Public Health, Boston, Massachusetts, USA3Channing Laboratory, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts, USA4Occupational Health Program, Department of Environmental Health, Harvard School of Public Health, Boston, Massachusetts, USA5Statistical Genetics Group, Jackson Laboratories, Bar Harbor, Maine, USA6Normative Aging Study, Department of Veterans Affairs Medical Center, Boston, Massachusetts, USAAddress correspondence to C. Lin, 1565 5th Ave. #202, San Francisco, CA 94122 USA. Telephone: (415) 665-2940. Fax: (415) 665-2940. E-mail:
[email protected]. Address reprint requests to H. Hu, Landmark Center East, 3-110A, 401 Park Dr., Boston, MA 02215 USA. Telephone: (617) 384-8968. E-mail: [email protected] thank D. Burger, F. Milder, S. Harcourt, R. Heldman, G. Barbella, S. Oliveira, T. Luu, G. Fleischaker, M. Barr, L. Hennessey, and S. Datta for their assistance.
Support was provided by National Institute of Environmental Health Sciences (NIEHS) grants ES 05257-06A1 and P42-ES05947 (with funding from the U.S. Environmental Protection Agency), NIEHS Occupational and Environmental Health Center grant 2 P30 ES00002, and National Institutes of Health (NIH) 1P20 MD000501. Subjects were evaluated in the outpatient Clinical Research Center of the Brigham and Women’s Hospital with support from NIH grant NCRR GCRC M01RR02635. The K X-ray fluorescence instrument used in this work was developed by ABIOMED, Inc. (Danvers, MA) with support from NIH grant SBIR 2R44 ES03918-02.
The authors declare they have no competing financial interests.
8 2004 3 5 2004 112 11 1147 1151 27 8 2003 3 5 2004 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. We measured blood and bone lead levels among minority individuals who live in some of Boston’s neighborhoods with high minority representation. Compared with samples of predominantly white subjects we had studied before, the 84 volunteers in this study (33:67 male:female ratio; 31–72 years of age) had similar educational, occupational, and smoking profiles and mean blood, tibia, and patella lead levels (3 μg/dL, 11.9 μg/g, and 14.2 μg/g, respectively) that were also similar. The slopes of the univariate regressions of blood, tibia, and patella lead versus age were 0.10 μg/dL/year (p < 0.001), 0.45 μg/g/year (p < 0.001), and 0.73 μg/g/year (p < 0.001), respectively. Analyses of smoothing curves and regression lines for tibia and patella lead suggested an inflection point at 55 years of age, with slopes for subjects ≥ 55 years of age that were not only steeper than those of younger subjects but also substantially steeper than those observed for individuals > 55 years of age in studies of predominantly white participants. This apparent racial disparity at older ages may be related to differences in historic occupational and/or environmental exposures, or possibly the lower rates of bone turnover that are known to occur in postmenopausal black women. The higher levels of lead accumulation seen in this age group are of concern because such levels have been shown in other studies to predict elevated risks of chronic disease such as hypertension and cognitive dysfunction. Additional research on bone lead levels in minorities and their socioeconomic and racial determinants is needed.
blacksblood leadbone leadminority groupsoccupationssmokingX-ray fluorescence
==== Body
Research has suggested that lead toxicity may disproportionately affect minority groups. (Bailey et al. 1994; Rothenberg et al. 1999). Despite substantial declines in blood lead levels in the general U.S. population, a substantial body of research, including data from the Third National Health and Nutrition Examination Survey (NHANES III), shows that African Americans continue to have higher blood lead levels than do whites (Brody et al. 1994; Lanphear et al. 1996; Mahaffey et al. 1982; Pirkle et al. 1998).
Most studies analyzing racial differences in lead toxicity have focused on blood lead as a biomarker. Although blood lead mostly provides an accurate measure of recent lead exposure, evidence has been growing to indicate that this biomarker does not adequately reflect an individual’s health risk due to cumulative lead exposure (Hu et al. 1998). In adults, about 95% of the total body lead burden is stored in the skeleton (Barry and Mossman 1970), and the half-life of lead in bone ranges from years to decades (Rabinowitz 1991). With a half-life of up to 25 years (Rabinowitz et al. 1976), bone lead is a biologic marker of cumulative lead exposure over many years and may better predict the effects of lead toxicity that arise from chronic low to moderate exposure, such as hypertension (Cheng et al. 2001; Glenn et al. 2003; Hu et al. 1996a; Korrick et al. 1999; Lee et al. 2001).
Sociodemographic rather than genetic factors, including low income and residence in older housing, have been attributed to the higher blood lead levels seen in black children (Pirkle et al. 1998). Although low income and education have also been correlated with higher blood and bone lead levels among white males (Elreedy et al. 1999; Hu et al. 1996b), minority groups are disproportionately affected. In recent data from the Normative Aging Study, nonwhite blue-collar workers had significantly higher blood and patella lead did than white blue-collar workers, suggesting an interaction between occupational exposures and race/ethnicity (Elmarsafawy et al. 2002). However, the generalizability of this study is limited by its size, because nonwhites comprised < 2% of the total sample population. To build on this work, in the Community Lead Study we focused on analyzing the bone lead levels of an exclusively minority sample.
Materials and Methods
Study subjects.
Subjects were recruited from the pool of subjects who participated in a study funded by the National Institutes of Health (NIH; “Impact of Sleep-Disordered Breathing in Older Adults,” NIH HL51075; principal investigator, D. Sparrow; 1 July 1994–30 June 1997) that had significant minority and female involvement. These subjects had been initially recruited via solicitation letters sent to residents in Boston, Massachusetts, census tracts in the Jamaica Plain neighborhood with high minority representation. Additional subjects for our study were drawn from the Roxbury, Dorchester, and Jamaica Plain neighborhoods—which also have high minority representation—through participant referrals to family members and friends.
Letters introducing the study and demographic and consent forms were sent to potential subjects. Only minority subjects ≥ 35 years of age were accepted. Willing and eligible participants were invited to the Brigham and Women’s Hospital outpatient clinic in Boston, where a fresh whole blood specimen was collected for lead measurement and where K X-ray fluorescence (KXRF) bone lead measurements were taken. Blood and bone lead measurements were taken between 1999 and 2000 for participating subjects. Participants who completed the study were reimbursed for their time and effort.
The human research committees of the Brigham and Women’s Hospital and the Department of Veterans Affairs Medical Center in Boston approved the research project. Written informed consent was obtained from all participants.
Blood lead measurement.
Blood for lead measurements was collected in 7-mL trace-metal–free tubes (Becton-Dickinson Co., Bedford, MA) containing EDTA and sent for analysis to ESA Laboratories, Inc. (Chelmsford, MA). The ESA Laboratories blood lead analysis protocol and quality control and quality assurance specifications are described elsewhere (Hu et al. 1996b).
KXRF bone lead measurement.
An ABIOMED KXRF instrument (ABIOMED, Inc., Danvers, MA) was used to take bone lead measurements of each subject’s midtibial shaft and patella. The physical principles, technical specifications, validation, and quality control procedures of this (Burger et al. 1990; Hu et al. 1990, 1994) and similar KXRF instruments (Ellis et al. 1987; Somervaille et al. 1985) have been described in detail elsewhere.
Briefly, KXRF uses a 109Cd gamma-ray source to induce fluorescence from the target tissue. The emitted photons are then detected, counted, and arrayed on a spectrum (Hu et al. 1989). The net lead signal is determined after Compton background counts are subtracted by a linear least-squares algorithm.
For each subject, 30-min measurements were taken at the midshaft of the tibia and patella after each region had been washed with a 70% solution of isopropyl alcohol. The KXRF beam collimator was sited perpendicular to the flat bone surface for the tibia and patella.
Statistical analyses.
We used Stata version 7.0 (Stata Corporation, College Station, TX) and S-Plus version 6.1 (Insightful Corporation, Seattle, WA) for database management and statistical analysis. The quality of the KXRF measurements was preserved by discarding tibia and patella lead values with associated measurement-uncertainty estimates of > 10 μg/g and > 15 μg/g, respectively. Negative tibia and patella measurements were retained to minimize bias and increase efficiency of comparing bone lead levels among different populations (Kim et al. 1995).
We created final education categories after collapsing comparable educational levels that had similar blood and bone lead data. Subjects with technical school training and college education were pooled together, as were students with graduate and professional schooling. For race, the 69 black participants comprised one category and the 15 other subjects, who were Hispanic, Asian, and American Indian, were classified as “other.” For job type, we classified retired subjects as white collar or blue collar based on their previous occupation. For example, doctors, lawyers, engineers, and so on, were categorized as white collar, whereas technicians, repairmen, carpenters, and so forth, were categorized as blue collar. We adopted a complete classification list of professions which has been published elsewhere by Elmarsafawy et al. (2002).
We examined blood, patella, and tibia lead levels across categories of age, race, education, smoking status, alcohol consumption, and job type. Simple linear regression of blood, tibia, and patella lead level versus age was performed over the entire age range. We performed graphic evaluation by locally weighted scatter plot smoothing (Lowess) to verify and select an inflection point for both biomarkers (Cleveland et al. 1976). Separate regression analyses were performed on subjects younger and older than this cutoff point, and male and female data were analyzed separately and together.
Multiple linear regression models were constructed to predict blood, tibia, and patella lead. Age, sex, race, educational level, alcohol consumption, cumulative smoking, and job type—variables known to be associated with these biomarkers—were forced into all models. Interaction terms of black race with blue-collar work, black race with male sex, and male sex with blue-collar work were tested for significance.
Results
A total of 108 subjects participated in this study, coming from the Jamaica Plain, Roxbury, and Dorchester neighborhoods in Boston: 86 black, 7 Hispanic, 3 American Indian, 2 Asian, and 10 other or unknown. Response rates to mailings in the parent study (< 10%) and the present study (< 10%) made our sample largely one of convenience. A final population of 84 subjects were included in the present analyses after we excluded 14 subjects for tibia (n = 13) and patella (n = 1) lead values associated with measurement uncertainty estimates of > 10 μg/g and > 15 μg/g, respectively, and 10 subjects who were missing covariate values for education (n = 6) and/or job type (n = 4). Comparisons between the 84 included subjects and 24 excluded subjects revealed no meaningful differences with regard to blood, tibia, or patella lead or age, race, education, pack-years of smoking, alcohol consumption, or job type.
Our sample of 84 subjects had a mean age of 50 years (range, 31–77 years), and 56 (67%) subjects were female (Table 1). The proportions of subjects whose education was limited to high school or lower (48%), who had a history of smoking (57%), and who worked in blue-collar jobs (33%) were not too dissimilar from the proportions we observed among the predominantly white subjects participating in the Normative Aging Study (47, 68, and 41%, respectively; Hu et al. 1996b; Elmarsafawy et al. 2002).
The mean and median blood lead levels of the present sample of 84 subjects were 3.0 μg/dL and 2.2 μg/dL, respectively. The mean ± SD for tibia lead was 11.9 ± 11.0 μg/g, and for patella lead, 14.2 ± 15.3 μg/g. In simple regression models, the slope coefficients of blood lead, and patella lead versus age were 0.10 μg/dL/year (p < 0.001), 0.45 μg/g/year (p < 0.001), and 0.73 μg/g/year (p < 0.001), respectively. Further analysis with smoothing plots indicated that the associations between age and bone lead biomarkers were nonlinear. In general, the univariate regression slopes of tibia and patella lead versus age were greater among subjects ≥ 55 years of age than among those < 55 years of age (Figures 1–3). A simple linear regression of patella lead versus age followed an average slope of 0.2 μg/g/year up to 55 years and then inflected upward to increase at 0.83 μg/g/year in subjects ≥ 55 years of age (data not shown). Likewise, tibia lead increased at a rate of 0.15 μg/g/year in participants < 55 years of age and at a rate of 0.69 μg/g/year ≥ 55 years of age (Figure 4). The differences in regression coefficients between the two age groups were statistically significant. Because of limited published data on patella lead, only the regression coefficients of tibia lead are compared with those observed in other studies in Figure 4 (Hu et al. 1990; Hu et al. 1996b; Kosnett et al. 1994; Roy et al. 1997).
For blood and patella lead, males and females < 55 years of age had similar rates of lead accumulation with increasing age; in contrast, among subjects ≥ 55 years of age, males had higher rates of accumulation. For tibia lead, a similar trend was observed, but the disparity among those ≥ 55 years was smaller.
In multiple regression models with independent variables that included age, sex, race, pack-years of smoking, drinking, educational levels, and occupation, age was the most significant predictor for blood, tibia, and patella lead (Table 2). A history of smoking > 20 pack-years predicted a 7.2 μg/g/year increase in tibia lead with borderline significance (p < 0.10). Having a blue-collar occupation significantly predicted an 8.02 μg/g increase in patella lead (p < 0.05). When interaction terms between black race and blue-collar work, black race and male sex, and male sex and blue-collar occupation were tested as predictors of lead biomarkers, they were insignificant (data not shown).
Discussion
The blood and bone lead levels we observed in this study indicate that this minority sample had lead exposure similar to that of the general population. The relatively low levels of blood lead (mean, 3.0 μg/dL) parallel those reported for individuals 20–74 years of age in the 1988–1991 NHANES III (mean, 3.0 μg/dL) (Pirkle et al. 1994). Studies of community-exposed, predominantly white subjects of similar age had tibia lead levels (Gamblin et al. 1994; Kosnett et al. 1994; Somervaille et al. 1988) and patella lead levels (Korrick et al. 2002) comparable with those observed in this study.
As seen in other studies of general population samples, age was the predominate correlate of tibia lead (Gamblin et al. 1994; Kosnett et al. 1994; McNeill et al. 2000) and patella lead (Hu et al. 1996b; Korrick et al. 2002), accounting for nearly half of the variability in multivariate regressions of both biomarkers. The strong association between age and bone lead probably reflects exposure to different levels of environmental lead over time, that is, the birth cohort effect previously reported (Kim et al. 1997). Univariate smoothing curves and simple regression models of tibia and patella lead versus age showed a smaller slope among subjects < 55 years of age, an inflection point at 55 years, and a greater slope at ≥ 55 years (Figures 2 and 3). These results may reflect a trend similar to that observed previously for tibia lead among community-exposed white men (Kosnett et al. 1994). We did not attempt a nonlinear model, which would likely have overfitted data from our small sample size.
Compared with age-related increases in bone lead of β = 0.31 μg/g/year observed by Hu et al. (1990) and β = 0.38 μg/g/year observed by Kosnett et al. (1994), tibia lead in our study increased at a lower rate (β = 0.15 μg/g/year) for subjects < 55 years of age. This discrepancy may be due to the differences in age range among these three studies. Study subjects from both the Hu et al. (1990) and Kosnett et al. (1994) studies, with ranges of 21–58 years and 20–55 years, respectively, spanned a wider age range, including younger individuals in their third decade with higher growth and bone formation rates.
In contrast, among subjects ≥ 55 years of age, tibia lead increased at a greater rate (β = 0.69 μg/g/year) than that measured in previous Normative Aging Study research (β = 0.38 μg/g/year; Hu et al. 1996b) (Figure 4). It is possible that genetic differences in bone turnover account for some of the observed discrepancy in those subjects ≥ 55 years of age. Two studies (Aloia et al. 1998; Luckey et al. 1996) have shown that postmenopausal black women have lower rates of bone turnover than do postmenopausal white women. Decreased formation of new bone in a more recent lower-lead environment would result in higher relative bone lead levels. The fact that women comprise two-thirds of our subjects may explain why the tibia lead accumulation rate in subjects ≥ 55 years of age was higher than that observed elsewhere (Hu et al. 1996b).
The steeper bone lead–age relationship in older individuals may also be due to increased environmental exposures among retirees. Boston’s prevalence of old housing has been tied to greater lead exposure via dust inhalation from lead paint in children (Bailey et al. 1994) and via ingestion of water contaminated from lead plumbing in adults (Potula et al. 1999). Housing age and lead amounts in tap water were not measured in this study, but given the fact that approximately 50% of housing in Boston was built before 1950 (Massachusetts Department of Public Health 1999), it seems likely that retirees who spend more time at home are at increased risk for environmental lead exposure.
When analyzed by sex, males exhibited a higher rate of blood and patella lead accumulation among subjects ≥ 55 years of age. This sex-related difference may be due to increased bone remodeling in postmenopausal women, supporting a trend previously reported by Hertz-Picciotto et al. (2000) and Rothenberg et al. (1994). As old bone is replaced by new bone matrix formed in a more recent lower-lead environment, women may experience a relative decrease in bone lead concentration. The same trend is blunted in tibia lead and reflects its lower bone turnover rate and thus decreased sensitivity to the onset of menopause.
Although participants were not asked if they had occupational lead exposure, in multivariate analyses blue-collar work by itself was a significant determinant of patella lead (p < 0.05). In previous research of a sample of white men who were not employed in lead-related industries, we found that tibia and patella lead levels were higher in those employed in blue-collar jobs (Elmarsafawy et al. 2002). Because the same classification criteria for white-collar and blue-collar jobs were used in this study, this analysis provides some support that occupational lead exposure is a risk factor for these minority individuals. Unlike our findings in the Normative Aging Study (Hu et al. 1996b), low education was not a significant predictor of blood or bone lead; however, individuals in our sample were relatively well educated, with > 50% having had some college education and two-thirds working in white-collar jobs, which may have limited our ability to discern the influence education as a proxy of social class.
The main limitations of this study stem from our relatively limited sample size as well as potential biases related to our subject recruitment. Although our subjects came from some of Boston’s high-minority-representation communities, they were volunteers who had participated in previous research and who essentially comprised a convenience sample. In addition, our models of bone lead were only able to explain up to 24% of their variance—a figure that is similar to those found in other studies such as the Normative Aging Study (Hu et al. 1996b); thus, there are likely many unmeasured factors that could explain differences in bone lead distributions. Nevertheless, comparisons of our data with those from the predominantly white subjects participating in the Normative Aging Study and other populations is instructive, particularly because their distributions of the main factors determined to be predictors of bone lead in the general population—education, smoking, and occupational status—are similar. In so doing, our finding of tibia bone lead levels in subjects > 55 years of age that were higher than those seen in whites of similar age is concerning because elevated bone lead levels have been clearly implicated as a risk factor for chronic disease such as hypertension (Cheng et al. 2001; Glenn et al. 2003; Hu et al. 1996a; Korrick et al. 1999; Lee et al. 2001). Clearly, more research is needed to examine bone and blood lead levels in minority groups, with an increased emphasis on those who may have lower educational and occupational status and who may therefore be at greatest risk for chronic lead exposure and its impact on health.
Figure 1 Scatter plots and smoothed lines of blood lead levels (n = 84) versus age in community-exposed minority subjects, Community Lead Study, Boston, Massachusetts, 1999–2000.
Figure 2 Scatter plots and smoothed lines of tibia lead levels (n = 84) versus age in community-exposed minority subjects, Community Lead Study, Boston, Massachusetts, 1999–2000.
Figure 3 Scatter plots and smoothed lines of patella lead levels (n = 84) versus age in community-exposed minority subjects, Community Lead Study, Boston, Massachusetts, 1999–2000.
Figure 4 Comparison of regression lines of tibia bone lead versus age between community-exposed minority subjects from the Community Lead Study, Boston, Massachusetts, 1999–2000, and other studies of predominantly white participants.
Table 1 Lead biomarker levels (mean ± SD) among Community Lead Study subgroups, Boston, Massachusetts, 1999–2000.
Variable No. Blood lead (μg/dL) Tibia lead (μg/g) Age-adjusted tibia lead (μg/g) Patella lead (μg/g) Age-adjusted patella lead (μg/g)
Age (years)
< 45 28 2.0 ± 1.2 8.3 ± 8.4 8.9 ± 14.3
46–60 41 2.8 ± 1.7 10.8 ± 11.5 11.8 ± 11.4
61–75 15 5.3 ± 3.2# 21.7 ± 8.6# 30.9 ± 15.7#
Sex
Female 56 2.7 ± 2.2 11.8 ± 11.9 12.0 ± 11.0 13.8 ± 15.0 14.1 ± 13.5
Male 28 3.6 ± 2.2* 12.1 ± 9.0 11.7 ± 7.9 15.0 ± 16.1 14.3 ± 13.3
Race
Black 69 3.0 ± 2.3 12.9 ± 11.2 12.7 ± 10.4 14.6 ± 15.4 14.4 ± 12.9
Other 15 2.9 ± 2.0 7.3 ± 8.8* 7.9* ± 7.2 12.3 ± 14.9 13.3 ± 15.9
Education
High school dropout 14 2.6 ± 2.2 9.4 ± 6.3 11.3 ± 7.4 12.9 ± 10.0 16.0 ± 10.1
High school graduate 26 3.6 ± 2.9 13.7 ± 12.9 12.9 ± 10.8 16.9 ± 20.3 15.7 ± 16.4
Technical school, college 31 2.7 ± 1.4 10.8 ± 9.2 10.8 ± 8.9 12.2 ± 12.0 12.2 ± 11.0
Graduate school, professional 13 3.0 ± 2.1 13.6 ± 14.8 13.0 ± 13.8 15.2 ± 16.0 14.1 ± 15.8
Smoking (pack-years)
0 36 2.7 ± 1.9 9.9 ± 11.1 9.9 ± 11.0 12.7 ± 14.9 12.7 ± 14.6
1–19 38 3.0 ± 2.1 11.9 ± 10.3 12.4 ± 8.7 12.5 ± 14.4 13.3 ± 11.5
≥ 20 10 4.1 ± 3.3 19.2 ± 11.3* 17.5 ± 9.5* 26.0 ± 16.1** 23.2 ± 13.1*
Consuming ≥ 2 alcoholic drinks/day
No 78 2.9 ± 2.2 12.0 ± 11.2 11.9 ± 10.2 14.0 ± 15.6 13.8 ± 13.6
Yes 6 3.7 ± 2.5 10.0 ± 7.9 12.0 ± 8.9 16.3 ± 11.3 19.5 ± 9.8
Job type
White collar/mixed 56 2.8 ± 2.1 11.3 ± 11.7 11.1 ± 10.9 12.0 ± 15.3 11.8 ± 13.3
Blue collar 28 3.4 ± 2.3 13.1 ± 9.6 13.4 ± 8.1 18.6 ± 14.5* 19.1 ± 12.5**
* p < 0.1,
** p < 0.05, and
# p < 0.01 by analysis of variance.
Table 2 Multiple regression of lead biomarker levels in the Community Lead Study, Boston, Massachusetts, 1999–2000.
Blood lead (μg/dL)
Tibia lead (μg/g)
Patella lead (μg/g)
Characteristic β 95% CI β 95% CI β 95% CI
Age (years) 0.10 0.05–0.14 0.42 0.19–0.65 0.72 0.42–1.02
Male sex 0.59 −0.46–1.65 −1.75 −7.13–3.63 −4.04 −11.08–3.00
Black racea −0.11 −1.28–1.07 4.04 −1.92–10.00 −0.29 −8.09–7.51
Educationb
High school graduate 0.56 −0.92–2.03 −0.52 −8.03–7.00 −0.17 −9.99–9.66
Technical school, college 0.07 −1.35–1.50 −0.82 −8.08–6.44 −1.40 −10.90–8.10
Graduate school, professional 0.16 −1.47–1.78 2.04 −6.23–10.32 0.38 −10.45–11.21
Smoking (pack-years)c
0–20 0.10 −0.89–1.09 2.67 −2.37–7.72 −0.53 −7.13–6.07
≥ 20 0.52 −1.07–2.11 7.18 −0.92–15.29 8.62 −1.98–19.22
Currently consuming 0.97 −0.85–2.80 −1.41 −10.72–7.90 1.12 −11.06–13.31
≥ 2 alcoholic drinks/day
Blue-collar occupationd 0.22 −0.92–1.35 2.10 −3.67–7.87 8.02 0.47–15.58
Total model adjusted R2 0.18 0.15 0.24
CI, confidence interval.
a Compared with other minorities.
b Baseline: high school dropout.
c Compared with 0 pack-years of smoking.
d Compared with white-collar occupations.
==== Refs
References
Aloia JF Mikhail M Pagan CD Arunachalam A Yeh JK Flaster E 1998 Biochemical and hormonal variables in black and white women matched for age and weight J Lab Clin Med 132 383 389 9823932
Bailey AJ Sargent JD Goodman DC Freeman J Brown MJ 1994 Poisoned landscapes: the epidemiology of environmental lead exposure in Massachusetts children 1990–1991 Soc Sci Med 39 757 766 7973872
Barry PS Mossman DB 1970 Lead concentrations in human tissues Br J Ind Med 27 339 351 5488693
Brody DJ Pirkle JL Kramer RA Flegal KM Matte TD Gunter EW 1994 Blood lead levels in the US population. Phase 1 of the Third National Health and Nutrition Examination Survey (NHANES III, 1988 to 1991) JAMA 272 277 283 8028140
Burger DE Milder FL Morsillo PR Adams BB Hu H 1990 Automated bone lead analysis by K-X-ray fluorescence for the clinical environment Basic Life Sci 55 287 292 2088281
Cheng Y Schwartz J Sparrow D Aro A Weiss ST Hu H 2001 Bone lead and blood lead levels in relation to baseline blood pressure and the prospective development of hypertension: the Normative Aging Study Am J Epidemiol 153 164 171 11159162
Cleveland WS Kleiner B Warner JL 1976 Robust statistical methods and photochemical air pollution data J Air Pollut Control Assoc 26 36 38 1244365
Ellis KJ Yasumura S Morgan WD 1987. In Vivo Body Composition Studies. London:Institute of Physical Sciences in Medicine.
Elmarsafawy SF Tsaih SW Korrick S Dickey JH Sparrow D Aro A 2002 Occupational determinants of bone and blood lead levels in middle aged and elderly men from the general community: the Normative Aging Study Am J Ind Med 42 38 49 12111689
Elreedy S Krieger N Ryan PB Sparrow D Weiss ST Hu H 1999 Relations between individual and neighborhood-based measures of socioeconomic position and bone lead concentrations among community-exposed men: the Normative Aging Study Am J Epidemiol 150 129 141 10412957
Gamblin C Gordon CL Muir DC Chettle DR Webber CE 1994 In vivo measurements of bone lead content in residents of southern Ontario Appl Radiat Isot 45 1035 1038 7994288
Glenn BS Stewart WF Links JM Todd AC Schwartz BS 2003 The longitudinal association of lead with blood pressure Epidemiology 14 30 36 12500043
Hertz-Picciotto I Schramm M Watt-Morse M Chantala K Anderson J Osterloh J 2000 Patterns and determinants of blood lead during pregnancy Am J Epidemiol 152 829 837 11085394
Hu H Aro A Payton M Korrick S Sparrow D Weiss ST 1996a The relationship of bone and blood lead to hypertension. The Normative Aging Study JAMA 275 1171 1176 8609684
Hu H Milder FL Burger DE 1989 X-ray fluorescence: issues surrounding the application of a new tool for measuring burden of lead Environ Res 49 295 317 2753011
Hu H Milder FL Burger DE 1990 X-ray fluorescence measurements of lead burden in subjects with low-level community lead exposure Arch Environ Health 45 335 341 2270952
Hu H Payton M Korrick S Aro A Sparrow D Weiss ST 1996b Determinants of bone and blood lead levels among community-exposed middle-aged to elderly men. The Normative Aging Study Am J Epidemiol 144 749 759 8857824
Hu H Rabinowitz M Smith D 1998 Bone lead as a biological marker in epidemiologic studies of chronic toxicity: conceptual paradigms Environ Health Perspect 106 1 8 9417769
Hu H Watanabe H Payton M Korrick S Rotnitzky A 1994 The relationship between bone lead and hemoglobin JAMA 272 1512 1517 7966843
Kim R Aro A Rotnitzky A Amarasiriwardena C Hu H 1995 K X-ray fluorescence measurements of bone lead concentration: the analysis of low-level data Phys Med Biol 40 1475 1485 8532760
Kim R Landrigan C Mossmann P Sparrow D Hu H 1997 Age and secular trends in bone lead levels in middle-aged and elderly men: three-year longitudinal follow-up in the Normative Aging Study Am J Epidemiol 146 586 591 9326437
Korrick SA Hunter DJ Rotnitzky A Hu H Speizer FE 1999 Lead and hypertension in a sample of middle-aged women Am J Public Health 89 330 335 10076481
Korrick SA Schwartz J Tsaih SW Hunter DJ Aro A Rosner B 2002 Correlates of bone and blood lead levels among middle-aged and elderly women Am J Epidemiol 156 335 343 12181103
Kosnett MJ Becker CE Osterloh JD Kelly TJ Pasta DJ 1994 Factors influencing bone lead concentration in a suburban community assessed by noninvasive K X-ray fluorescence JAMA 271 197 203 8277545
Lanphear BP Weitzman M Eberly S 1996 Racial differences in urban children’s environmental exposures to lead Am J Public Health 86 1460 1463 8876521
Lee BK Lee GS Stewart WF Ahn KD Simon D Kelsey KT 2001 Associations of blood pressure and hypertension with lead dose measures and polymorphisms in the vitamin D receptor and δ-aminolevulinic acid dehydratase genes Environ Health Perspect 109 383 389 11335187
Luckey MM Wallenstein S Lapinski R Meier DE 1996 A prospective study of bone loss in African-American and white women—a clinical research center study J Clin Endocrinol Metab 81 2948 2956 8768857
Mahaffey KR Annest JL Roberts J Murphy RS 1982 National estimates of blood lead levels: United States, 1976–1980: association with selected demographic and socioeconomic factors N Engl J Med 307 573 579 7110203
Massachusetts Department of Public Health, Childhood Lead Poisoning Prevention Program 1999. Massachusetts’ Fight against Childhood Lead Poisoning: Updated Trends, 1993–1998. Boston:Massachusetts Department of Public Health, Childhood Lead Poisoning Prevention Program.
McNeill FE Stokes L Brito JA Chettle DR Kaye WE 2000 109 Cd K X ray fluorescence measurements of tibial lead content in young adults exposed to lead in early childhood Occup Environ Med 57 465 471 10854499
Pirkle JL Brody DJ Gunter EW Kramer RA Paschal DC Flegal KM 1994 The decline in blood lead levels in the United States. The National Health and Nutrition Examination Surveys (NHANES) JAMA 272 284 291 8028141
Pirkle JL Kaufmann RB Brody DJ Hickman T Gunter EW Paschal DC 1998 Exposure of the U.S. population to lead, 1991–1994 Environ Health Perspect 106 745 750 9799191
Potula V Serrano J Sparrow D Hu H 1999 Relationship of lead in drinking water to bone lead levels twenty years later in Boston men: the Normative Aging Study J Occup Environ Med 41 349 355 10337604
Rabinowitz MB 1991 Toxicokinetics of bone lead Environ Health Perspect 91 33 37 2040248
Rabinowitz MB Wetherill GW Kopple JD 1976 Kinetic analysis of lead metabolism in healthy humans J Clin Invest 58 260 270 783195
Rothenberg SJ Karchmer S Schnaas L Perroni E Zea F Fernandez Alba J 1994 Changes in serial blood lead levels during pregnancy Environ Health Perspect 102 876 880 9644197
Rothenberg SJ Manalo M Jiang J Khan F Cuellar R Reyes S 1999 Maternal blood lead level during pregnancy in south central Los Angeles Arch Environ Health 54 151 157 10444035
Roy MM Gordon CL Beaumont LF Chettle DR Webber CE 1997 Further experience with bone lead content measurements in residents of southern Ontario Appl Radiat Isot 48 391 396 9116655
Somervaille LJ Chettle DR Scott MC 1985 In vivo measurement of lead in bone using X-ray fluorescence Phys Med Biol 30 929 943 4048276
Somervaille LJ Chettle DR Scott MC Tennant DR McKiernan MJ Skilbeck A 1988 In vivo tibia lead measurements as an index of cumulative exposure in occupationally exposed subjects Br J Ind Med 45 174 181 3348993
|
15289158
|
PMC1247473
|
CC0
|
2021-01-04 23:40:10
|
no
|
Environ Health Perspect. 2004 Aug 3; 112(11):1147-1151
|
utf-8
|
Environ Health Perspect
| 2,004 |
10.1289/ehp.6705
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences 10.1289/ehp.6871ehp0112-00115215289159ResearchArticlesAge-Related Differences in Susceptibility to Carcinogenesis: A Quantitative Analysis of Empirical Animal Bioassay Data Hattis Dale 1Goble Robert 1Russ Abel 1Chu Margaret 2Ericson Jen 11George Perkins Marsh Institute, Clark University, Worcester, Massachusetts, USA2Office of Research and Development, U.S. Environmental Protection Agency, Washington, DC, USAAddress correspondence to D. Hattis, George Perkins Marsh Institute, Clark University, 950 Main St., Worcester, MA 01610 USA. Telephone: (508) 751-4603. Fax: (508) 751-4600. E-mail:
[email protected] thank H. Barton (U.S. EPA) for suggesting some additions and modifications to the data originally published by the U.S. EPA (2003). We also thank W. Setzer and P. White (U.S. EPA) for their helpful review of technical statistical issues.
This research was supported by a cooperative agreement with the U.S. EPA (CR 829746-01).
The conclusions are those of the authors and do not necessarily reflect the views of the U.S. EPA.
The authors declare they have no competing financial interests.
8 2004 12 5 2004 112 11 1152 1158 19 11 2003 12 5 2004 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. In revising cancer risk assessment guidelines, the U.S. Environmental Protection Agency (EPA) analyzed animal cancer bioassay data over different periods of life. In this article, we report an improved analysis of these data (supplemented with some chemical carcinogenesis observations not included in the U.S. EPA’s original analysis) and animal bioassay studies of ionizing radiation. We use likelihood methods to avoid excluding cases where no tumors were observed in specific groups. We express dosage for animals of different weights on a metabolically consistent basis (concentration in air or food, or per unit body weight to the three-quarters power). Finally, we use a system of dummy variables to represent exposures during fetal, preweaning, and weaning–60-day postnatal periods, yielding separate estimates of relative sensitivity per day of dosing in these intervals. Central estimate results indicate a 5- to 60-fold increased carcinogenic sensitivity in the birth–weaning period per dose ÷ (body weight0.75-day) for mutagenic carcinogens and a somewhat smaller increase—centered about 5-fold—for radiation carcinogenesis per gray. Effects were greater in males than in females. We found a similar increased sensitivity in the fetal period for direct-acting nitrosoureas, but no such increased fetal sensitivity was detected for carcinogens requiring metabolic activation. For the birth–weaning period, we found an increased sensitivity for direct administration to the pups similar to that found for indirect exposure via lactation. Radiation experiments indicated that carcinogenic sensitivity is not constant through the “adult” period, but the dosage delivered in 12- to 21-month-old animals appears a few-fold less effective than the comparable dosage delivered in young adults (90–105 days of age).
carcinogenesisfetalionizing radiationmutagenic chemicalsrisk assessmentstatistical analysissusceptibility
==== Body
Standard animal cancer bioassays were designed as a qualitative screen for carcinogenic activity. In this context, it is easy to see how the additional difficulties of dosing at early life stages might have been considered to provide an only modest incremental return of qualitative hazard identification information compared with the extra effort and complexity of assuring adequate and comparable delivery of test substances over a full lifetime of exposure, from conception through adulthood. Therefore, conventional animal cancer bioassay studies conducted by the U.S. National Toxicology Program (NTP) and elsewhere have been designed to start dosing in early adulthood—usually 6–8 weeks of age in mice and rats (NTP 1993, 1999).
Over the last couple of decades, however, animal bioassay results have been routinely used as a basis for quantitative projections of potential cancer risks for populations exposed over a full lifetime, from conception through death. Moreover, the results of such risk projections are routinely used to arrive at a variety of types of determinations needed for practical decisions, for example:
How extensive is the cleanup that is needed at hazardous waste sites to achieve risks that are below X incidence of harm with Z confidence? [Hattis and Anderson 1999; U.S. Environmental Protection Agency (EPA) 2001]
What health prevention benefits should be expected from reducing exposures by various amounts for toxicants in ambient air, drinking water, and foods subjected to the chemical transformations from different methods of cooking? Do the incremental benefits of specific intervention measures justify their costs, when compared with available alternatives? [National Research Council 2002; Office of Management and Budget (OMB) 2000].
In the current revision of cancer risk assessment guidelines by the U.S. EPA (2003a), a question has arisen about whether human exposures during early life stages—during adolescence and before—should be attached any greater weight in risk projections than exposures during adulthood that are analogous to exposures represented in conventional animal bioassay testing. After reviewing an extensive set of nonconventional animal bioassay testing results, the U.S. EPA (2003b) concluded that there was appreciable evidence that juvenile exposures to mutagenic carcinogens conferred greater risks per day of dosing than do exposures during adulthood. The U.S. EPA proposed that for mutagenic chemicals, exposures in the first 2 years of life should be assumed to be 10 times as potent as exposures in adulthood. A similar 3-fold increase in expected risk was proposed for assessments of the effects of exposures between 2 and 15 years of age.
Both the age cutoffs used in this proposal and the extent of the assumed increase in sensitivity relative to adults were the products of relatively informal analyses of the assembled database. There was no analysis of data for carcinogenesis after transplacental exposure in the fetal period, and there was no distinction between preadult exposures before versus after weaning. Moreover, comparisons were done based on juvenile:adult ratios of raw cancer incidence (the fraction of animals observed to develop tumors) for comparably dosed animals. This potentially introduced distortions of two types: first, there was no allowance for tumor multiplicity (more than one effective tumor generation event per animal) in animal groups where a large fraction of the animals developed tumors, and second, the ratio analysis necessarily excluded data sets in which no tumors were observed in adult animals. In this article, we somewhat expand the database assembled by the U.S. EPA (2003b), and we present a more formal statistically weighted analysis of relative cancer potency in terms of cancer transformations per animal per unit dose for animals in different age groups, scaled to the highest experimental dose used either in adult animals or (if no fully adult animals were tested) the oldest age group of animals included in the experiment. We also derived separate summary relative potency estimates for the fetal, birth–weaning (approximately 21 days in rodents), and weaning–60-day periods. Where dosage spans multiple age groups, we used dummy variables to represent the observed tumor risk as the sum of cancer contributions from dosing in different periods. The data were analyzed in a series of subsets (mutagenic vs. nonmutagenic chemicals vs. radiation; male vs. female; liver vs. nonliver) to show how the results depend on various factors.
Description of the Databases
An overview of the data is presented in Table 1. Experimental results described in detail by the U.S. EPA (2003b) were corrected in a few cases and supplemented as follows:
We added esophageal tumors for diethylnitrosamine (DEN; Peto et al. 1984); liver but not esophageal tumors from this article were included in the U.S. EPA analysis (U.S. EPA 2003b). Additionally, we added control observations reported by Peto et al. (1991).
The exposure time was corrected for some vinyl chloride groups; we also included additional control and comparison group information for 52-week exposures described by Maltoni et al. (1984).
We consolidated 6,000 and 10,000 ppm exposure groups for vinyl chloride; both of these are far greater than saturating levels for the metabolic activation of this chemical. Results for control (zero-dose) groups were also consolidated in several cases.
We added the results of a major single-dose study of N-nitrosomethylurea by Terracini et al. (1976) and data from several reports on carcinogenesis from ionizing radiation in rats and mice (Cahill et al. 1975; Castanera et al. 1971; Di Majo et al. 1990; Knowles 1985; Sasaki 1991).
We deleted groups that did not show defined observations for controls (numbers of animals tested and numbers with tumors).
Data for two nonmutagenic chemicals (DDT and dieldrin) were eliminated from the analysis because of the complexity of the dosing protocol used. In these experiments, some groups were given gavage exposures, some direct dietary exposures, and some both in sequence. This rendered unambiguous calculations of comparable dosages for the different groups difficult.
The principal analyses maintain the subdivisions between continuous-dosing protocols (in which dosing was maintained at a given rate for a defined period) versus discrete-dosing experiments (in which only a single dose, or up to four single doses were given to the animals at defined ages).
The full databases as well as models used for the statistical analyses of continuous, discrete, and radiation dosing data are available on our website (Hattis 2004).
Modeling Methods
Dosimetric conversions.
The assessment of comparable dosimetry for animals in different life stages has been a substantial issue in discussions of the analysis of these data. For various experiments in the original U.S. EPA listing (U.S. EPA 2003b), doses are quoted in terms of a concentration in an environmental medium (parts per million in diet or water or air to the individual for exposures after weaning, and to the mother in the case of fetal and birth–weaning exposures); in other cases, doses that were directly administered to animals via intraperitoneal or other injections were originally expressed in terms of micrograms per kilogram body weight or similar units. For entry into our analysis, we left the doses expressed in terms of environmental media concentrations unchanged, but we transformed the doses expressed as micrograms per kilogram body weight into micrograms/(kilogram body weight)0.75 by multiplying by estimated individual body weights to the one-quarter power. [Body weights for this purpose were taken from Nomura (1976) for mice and from the NTP (1999) and Zhang et al. (2001) for rats.] The aim of this transformation was to use a dose metric that (to the extent possible with available information short of physiologically based toxicokinetic modeling) is expected to be approximately proportional to internal daily average systemic concentrations of the parent compounds or putative active metabolites for continuous dosing, or area under the concentration–time curve (AUC) for discrete dosing.
The basis for this approach is similar to the principal current basis for dosimetric conversions for interspecies projections of cancer risks: that risks are assumed to be similar across species if the internal time-integrated concentrations of active metabolites are similar across species. Similarity of internal time-integrated concentrations is assessed with the aid of observations that both bulk uptake and elimination processes tend to scale across species with metabolic rates—approximately in proportion to body weight to the three-quarters power (Boxenbaum 1982; Federal Council for Science, Engineering and Technology 1992; Travis and White 1988; Travis et al. 1990). We have recently found that a similar transformation reconciles clearance rates of drugs across age groups in humans—at least after a period of severely deficient clearance in the first few months of infancy. Table 2, documenting this result, is based on a new regression analysis of human data for pharmaceuticals and methods that have been previously described (Ginsberg et al. 2002; Hattis et al. 2003). We have not located a comparable set of in vivo clearance observations in rats or mice. The literature does contain several reports that indicate depressed liver-metabolizing activity in the neonatal period based on in vitro measurements of the activity of some liver enzymes (Basu et al. 1971; Macleod et al. 1972) and differences between the sexes in the maturation of metabolizing capabilities (with generally greater activity observed in males). To assess the possible influence of a neonatal deficit of either activating or detoxifying activity on our findings, in the “Results” we include comparative analyses of the single-dose data for apparent relative sensitivity at narrowly defined time windows—contrasting day 1 after birth with later periods before and after weaning. We performed these comparisons for the two carcinogens that are thought to be direct acting (not requiring metabolic activation) and for those that putatively need metabolic activation before directly DNA-reactive substances are generated. We also assessed differences in apparent life-stage–related sensitivity between the sexes.
For ionizing radiation exposures, we have chosen to leave the doses in units of absorbed energy—rads or grays. If the oxidative products generated by radiation are the actual carcinogenic agents, and if these are predominantly destroyed by metabolism-dependent processes that operate at rates that scale with metabolic rates, it is possible that achieving comparable integrated dose × time levels of the active agents might require the same (body weight)0.75 conversions as used for chemicals. Making such a transformation would tend to decrease the time-integrated dosage for the younger post-natal animals and therefore would tend to increase the assessed sensitivity per dose relative to adult exposures. As it happens, such a transformation would have brought the radiation results more closely into alignment with the results for mutagenic chemicals.
Equation fit and statistical optimization.
One basic difference between our methodology and that used for these data by the U.S. EPA (2003b) is a transformation of the raw observations of tumor incidence in different groups into the estimated number of tumor transformations per animal. This corrects for the fact that researchers cannot usually distinguish between cases where one or more than one tumor was induced in a particular organ within a specific animal (or where more than one tumor would have been induced at the site studied had the animal lived to the end of the observation period). To accomplish this, we use the same Poisson transformation that has been traditionally used for the multistage and related statistical models of carcinogenesis.
The Poisson distribution is appropriate for processes that occur as the result of independent events where the number of possible events occurring in a particular unit of observation is unlimited. Our use of the Poisson distribution in this case derives from the basic fact that tumors start in individual cells (Fialkow 1997; Knudson, 1973, 1977). Each tumor is conceived to be an independent event arising as the result of the completion of the last stage mutation in one stem cell out of many other susceptible stem cells in a particular organ. It should be noted that this last-stage event will not generally have occurred during the preadult life stages that are the focus of our analysis, but the effects of these early life exposures will manifest as incremental tumors that occur during the life-long period of observation of the animals.
Fraction of animals with tumors
where m is the tumor transformations per animal at the studied site. Solving for m:
Because most of the experiments use only a single dose of carcinogen for each age group, no more sophisticated multistage treatment of tumor dose response is possible with these data. Given this, relative cancer transformation rates in different age groups in comparison with adult animals were estimated by fitting the continuous data to the following equation:
Fraction with tumors
where B is the group background transformations per animal; A is the group adult transformations per animal at the highest adult dose rate; a is the fraction of the adult period with dosing at the maximum adult rate (this term reflects an adjustment where a group received less than the full adult dosing rate); f is the fraction of the fetal period with dosing at the maximum adult rate (also adjusted for dose rate as needed); F is the fetal:adult sensitivity ratio; c is the fraction of the birth–weaning period with dosing at the maximum adult rate (also adjusted for dose rate as needed); C is the birth–weaning:adult sensitivity ratio; w is the fraction of the weaning–60-day period with dosing at the maximum adult rate (also adjusted for dose rate as needed); and W is the weaning–60-day:adult sensitivity ratio.
In Equation 4, the terms designated with lowercase letters represent the input dosing and tumor response data for each group of dosed animals or controls. Where continuous daily dosing occurred over only part of a life stage, we entered the fraction of the life stage where dosing occurred. Similarly, where dosing for a particular group occurred at a fraction of the maximal rate given to adults, that fraction was entered as input data. This model form treats contributions to ultimate cancer transformation events from different life stages as additive.
The equation has two types of estimated parameters (designated with upper case letters). First, A and B are used only within specific experiments (a particular tumor type associated with exposure to a particular chemical in a particular animal group). By contrast, the three remaining “generic” parameters (F, C, and W) are estimated based on the results of all the dose groups for all chemicals and animals included in a particular run that contained some dosing within each life stage, compared with controls. Thus, for these generic parameters, the results represent summary central estimates [and upper (UCL) and lower confidence limits (LCL)] for all chemicals, tumor types, species (rats and mice), and other characteristics of the included experimental data. In light of this, in the “Results” we present alternative sets of estimates designed to explore the influence of sex, mutagenic character, tumor site, and other characteristics on the assessments of differences in susceptibility among life stages. Finally, because the doses used in the model fitting were expressed in terms of dose ÷ (body weight)0.75, the units of the relative sensitivity parameters should similarly be understood to be
Estimates of the uppercase terms were derived by minimizing the “deviance” between observed and model predicted data points, as described by Haas (1994) and McCullagh and Nelder (1989): For nonzero numbers of tumors in a particular group the “deviance” is
where k is the number of dose groups; Ni is the number of animals with tumors in group i; Ti is the total number of animals in group i; πi is the model-predicted proportion of animals with tumors in group i; and π0i is Ni/Ti.
This deviance-minimization optimization was accomplished in Microsoft Excel spreadsheets using the “solver” facility (Microsoft Corporation, Redmond, WA). Haas (1994) also provided procedures for deriving profile-likelihood-based confidence intervals (Venzon and Moolgavkar 1988) for these fitted parameters based on the chi-square statistic. For each confidence interval estimate, all parameters other than the one being assessed were allowed to vary. Thus, the upper and lower 95% confidence limits for the birth–weaning:adult sensitivity estimates reflect possible uncertainties in all the group background transformations per animal, group adult transformations per animal, and the sensitivities of fetal and weaning–60-day life stages relative to adults. A similar approach was used for the discrete dosing data and for the combined continuous and discrete data by dividing the doses by the estimated numbers of days in each dosing period (8 days for the fetal dosing period, 21 days for the birth–weaning life stage, 39 days for the weaning–60-day life stage, and 663 days for the adult period).
Results and Discussion: Relative Sensitivity of Different Life Stages in Animals
Before considering the age-related differential sensitivity results for continuous versus discrete dosing in detail, it is worth noting that they may be reflecting somewhat different factors. The continuous dosing results:
Include enzyme induction effects, if any
Inherently reflect a dilution of any fluctuations in short-term sensitivity caused by, for example, waves of cell proliferation in specific organs in narrow time windows
Possibly present fewer complications from high-dose kinetic and dynamic nonlinearities
Have somewhat more straightforward implications for adaptation of traditional chronic dosing assessments.
On the other hand, the results from experiments where dosing was administered at discrete times:
Almost always exclude direct enzyme induction effects
Are capable of revealing short-term sensitivity fluctuations, to the extent that these occur
Are likely to be done at somewhat higher dose rates, with some increase in potential complications from high-dose nonlinearities
Have more straightforward implications for assessment of risks from acute exposure events.
Results for overall continuous chemical, discrete chemical, and radiation dosing data sets.
Table 3 shows the results of fitting the continuous and discrete dosing data as a whole, together with similar results for radiation exposures. In all three sets of data, the birth–weaning period is suggested to be the most sensitive per day of dosing, followed by the fetal period and the weaning–60-day period. Each independent data set yields a central estimate of the birth–weaning sensitivity that is about 5- to 10-fold greater than the sensitivity per day of dosing in adulthood, with doses expressed per body weight0.75.
Mutagenic versus nonmutagenic chemicals.
In the case of the continuous dosing data, some of the chemicals were classified by the U.S. EPA (2003b) as mutagenic, and some not. (All of the chemicals with discrete dosing data, and ionizing radiation, are mutagenic.) Table 4 shows the continuous dosing results broken out for mutagens versus nonmutagens. In contrast with the mutagens, for nonmutagenic carcinogens none of the age groups manifest significantly greater sensitivity than is seen for adults (defined as 1 in these tables). It should also be noted that separating out the nonmutagens leaves the mutagenic compounds showing significantly more birth–weaning period sensitivity than is seen for either the discrete-dosing chemical data or the radiation observations.
Male versus female animals.
Tables 5–7 show the contrast between results in male versus female animals for continuously dosed mutagens, mutagenic chemicals delivered in discrete doses, and radiation experiments, respectively. The differences appear most prominent for the continuous dosing data (Table 5), where males seem to have much larger increases in sensitivity relative to adults for the fetal and birth–weaning life stages, and by contrast, females show a large increase in sensitivity for the weaning–60-day period. Considerable reserve is in order in interpreting the latter result, however, in the light of the slender database available for the continuous dosing analysis (only 3 chemicals and 16 dose groups for each sex) and the fact that neither the larger set of discrete-dosing data (Table 6) nor the radiation-dosing data (Table 7, based on fetal and weaning–60-day stages only) exhibits a similar enhanced female relative sensitivity for the weaning–60-day period, compared with males.
One way of weighing the different observations from continuous versus discrete chemical dosing experiments is to combine the two sets of results into a single model for analysis. The results of such a combination for male and female life-stage relative sensitivity ratios are shown in Table 8. The combined data tend to reinforce the suggestion that there are male–female differences in age-related sensitivity patterns but fail to sustain the initial suggestion from the continuous dosing data of an increase in the sensitivity for females in the weaning–60-day period relative to adults. On the other hand, the combined data do indicate an increased sensitivity for this period in males. The combined data for the fetal and birth–weaning periods indicate much more prominent excess sensitivity relative to adults in males than in females.
Distributional form for the statistical uncertainties in estimated life stage/adult sensitivities.
Figures 1 and 2 show lognormal probability plots (Hattis and Burmaster 1994) of the statistical uncertainty distributions for the life stage:adult sensitivity ratios for the male and female combined discrete and continuous dosing data for mutagenic carcinogens. In this type of plot, correspondence of the points to the fitted line is an indicator of the fit of a log-normal distribution to the statistical uncertainties in central estimate life stage:adult sensitivity ratios. (The Z-score that makes up the x-axis is the number of standard errors above or below the median of the normal distribution log10 transformed values.) It can be seen that the uncertainty distributions are well described by the lognormal fits. We stress that these plots are of confidence limits on the aggregate central tendency results for all chemicals in the covered groups. The uncertainties in estimates for individual chemicals are being analyzed separately (Hattis et al., unpublished data), together with implications for human risk for a particular mutagenic chemical.
Rats versus mice.
Table 9 shows comparative results for life-stage–specific relative tumor sensitivities in rats versus mice for the combined discrete and continuous dosing experiments. There is a suggestion that the rat data may indicate somewhat larger effects relative to adults for the fetal and weaning–60-day life stages; however, the 95% confidence limits overlap. In the light of the very limited numbers of chemicals with relevant observations for rats, there should be no strong inference that the suggested rat/mouse differences are real.
Direct-acting carcinogens versus agents requiring metabolic activation.
All but two of the mutagenic carcinogens covered in the database are thought to require metabolic activation to produce DNA-reactive agents (U.S. EPA 2003). The two exceptions are the nitrosoureas—methyl- and ethylnitrosourea. Comparing life stage:adult sensitivity results for the metabolically activated versus direct-acting compounds can shed light on whether the previous results, including the relevant dosimetry, are likely to have been appreciably distorted by immaturity of metabolic activating systems in the neonatal period.
Table 10 shows the relevant comparison using our standard breakdown of life stages, based on the single-dose data. The results indicate a clear difference in fetal sensitivity for direct-acting versus metabolically activated compounds. As might have been expected, there is, if anything, less carcinogenic susceptibility in the fetal period for metabolically activated compounds, whereas the fetal life stage shows 5- to 25-fold greater sensitivity than adults for the direct-acting nitrosoureas.
Table 11 shows the results of using a finer breakdown of time periods, made possible by the focus on data resulting from direct dosing at discrete times. Beyond the fetal period, there is no apparent difference in the pattern of relative sensitivity with age between the nitrosoureas and the metabolically activated carcinogens. In both cases, relative sensitivity peaks near birth and declines progressively thereafter until it reaches about double the adult sensitivity at day 21. Beyond the fetal period, there is thus no indication of a perinatal deficit in metabolic activating activity for this set of carcinogens.
Direct dosing in the birth–weaning period versus dosing via lactation.
Another important dosimetric issue is whether the lactational exposures resulting from primary dietary exposure to maternal animals are in fact equivalent to doses directly administered to pups during the birth–weaning period. Table 12 shows the results of separate estimations of the relative tumor susceptibility for direct versus lactational exposure for the combined set of continuous and discrete dosing experiments. The data show that no diminution in birth–weaning sensitivity is indicated for lactational exposures compared with direct administration of known doses. If anything, the lactational exposures appear somewhat more potent than direct administration per unit of estimated external exposure, although the 95% confidence limits overlap. One possible interpretation of this result, if repeated, is that some of the bolus doses given in the direct administration experiments may have partially saturated metabolic activation pathways, leading to less effective dose of DNA-reactive metabolites per unit exposure than when similar materials are administered more slowly via milk.
Radiation results for different times during the “adult” period.
The “adult” comparison groups for the discrete chemical dosing experiments generally were exposed in early adulthood—within 4–6 months of age. By contrast, the radiation experiments include groups extending to much older ages—up to 16–18 months. As shown in Table 13, these data indicate a considerable reduction in sensitivity for radiogenic cancer induction with advancing age.
Liver tumors versus tumors in other organs.
As indicated in Table 1, many of the tumors studied in these rodent experiments come from the liver, particularly for the continuous dosing studies. We have found that, in general, life-stage–specific enhancements of sensitivity seem to be greater for the liver than for the lung, but life-stage–specific excesses in sensitivity are still apparent for the aggregate of nonliver, nonlung organs (Hattis 2004).
Toward quantitative applications in human health risk assessment.
On a qualitative level, this analysis provides more detailed understanding and confidence in the fact that there is an increased early-life sensitivity for mutagenic carcinogens—reinforcing the conclusions drawn by the U.S. EPA (2003b). The next step toward applying these data for quantitative human risk assessment is to develop time/age mapping between rodents and people. What ages in people approximately correspond to the rodent fetal, birth–weaning, and weaning–60-day periods studied in this analysis? We are developing a preliminary mapping based on the times at which rodents and people attain various fractions of the average body weights they have at sexual maturity (Hattis et al., unpublished data). In this second article we also use a Monte-Carlo model–based distributional analysis of the combined uncertainties in a) the central estimates of life-stage–related differences in carcinogenesis susceptibility, as derived in this article; b) the chemical-to-chemical variation in the life-stage–related susceptibility estimates; and c) the rodent/human time mapping uncertainty. Quantitative assessment of these three uncertainties together is needed for full distributional analyses of cancer risks for exposures in early life stages.
Figure 1 Lognormal plots of likelihood-based uncertainty distributions for cancer transformations per daily dose for various life stages for mutagenic chemicals (relative to comparable exposures of adults) for combined discrete and continuous dosing experiments in females. Log(birth–weaning/adult): y = 0.646 + 0.0785x; R2 = 1.000. Log(fetal/adult): y = 0.246 + 0.134x; R2 = 1.000. Log(weaning–60 days/adult): y = 0.0880 + 0.124x; R2 = 0.999.
Figure 2 Lognormal plots of likelihood-based uncertainty distributions for cancer transformations per daily dose for various life stages for mutagenic chemicals (relative to comparable exposures of adults) for combined discrete and continuous dosing experiments in males. Log(birth–weaning/adult): y = 1.76 + 0.113x; R2 = 0.999. Log(fetal/adult): y = 1.41 + 0.132x; R2 = 1.000. Log(weaning–60 days/adult): y = 0.705 + 0.133x; R2 = 0.999.
Table 1 Overall description of the databases.
Dose groups with exposures in specific life stages (no. of animals × tumor-site observations)
Dosing protocol No. of chemicals or radiation types Total dose groups Control groups Fetal Birth–weaning Weaning–60 days Adult (≥ 60 days)
Continuous 9 (5 mutagenic)a 151b (103 liver) 29 (2,562) 14 (820) 62 (3,071) 62 (6,128) 85 (7,544)
Discrete (1–4×) 6 (all mutagenic)c 274b (90 liver) 45 (2,926) 8 (290) 117d (4,681) 85d (3,596) 37 (979)
Radiation 4e 138 (42 liver) 21 (4,283) 18 (1,323) 18 (1,744) 18 (1,529) 63 (3,668)
In some experiments, tumor observations were reported separately for two or more anatomical sites (e.g., liver and stomach). In these cases, the numbers reported here count the same individual animals more than once.
a The chemicals classified as mutagenic were benzidine, benzo(a)pyrene, DEN, safrole, and vinyl chloride; the chemicals classified as not mutagenic were amitrole, diphenylhydantoin, ethylene thiourea, and polybrominated biphenyls.
b The numbers of groups do not add to the total because some groups had dosing in more than one life stage.
c Benzo(a)pyrene, DEN, dimethylbenzanthracene, ethylnitrosourea, methylnitrosourea, and urethane.
d Sixty-six groups were dosed on the first day after birth, 69 groups received exposures between days 1 and 21, 19 groups were dosed on day 21, and 68 groups were dosed between days 22 and 60; this finer breakdown is presented in the expanded-time analysis of the single-dose data in Table 11. The sum of these numbers exceeds the total because some groups received dosing in more than one of these more finely divided time categories.
e The ionizing radiation exposures were from 137Cs gamma rays, X rays, neutrons, and internal beta rays resulting from the injection of tritiated water.
Table 2 Geometric mean ratiosa of child/adult clearance/body weight and (clearance/body weight0.75): regression results from 104 data groups for 27 drugs for humans in various age ranges.
Form for expressing total body clearance Premature neonates Full-term neonates 1 week–2 months 2–6 months 6 months–2 years 2–12 years 12–18 years
Mg/kg body weight 0.52a (0.43–0.63) 0.66 (0.61–0.73) 0.77 (0.71–0.84) 1.21 (1.06–1.39) 1.71 (1.52–1.92) 1.42 (1.31–1.53) 0.97 (0.78–1.2)
Mg/(kg body weight)0.75b 0.23 (0.19–0.28) 0.31 (0.28–0.34) 0.38 (0.35–0.42) 0.68 (0.59–0.78) 1.03 (0.91–1.17) 1.08 (1.00–1.17) 0.93 (0.74–1.17)
Data in parentheses indicate the ± 1 SE range.
a These data are the antilogs of the B coefficients that result from fitting the equation: log(mean clearance) = B0 (intercept) + B1 × (1 or 0 for chemical 1) + B2 × (1 or 0 for chemical 2) + … + Ba × (1 or 0 for age group 1) + Bb × (1 or 0 for age group 2) + …. A more complete description of the underlying data and methodology has been reported by Ginsberg et al. (2002), Hattis et al. (2003), and Hattis (2004).
b Input clearance/(kg body weight)0.75 data for the regression results reported in this line were calculated from clearance/body weight data by multiplying by group mean estimated body weights0.25. For children ≥ 2 years of age, body weights for this transformation were estimated using the formulas described by Hattis et al. (2003), averaged for both sexes. Body weights of 2.5 and 3.5 kg were assumed for premature and full-term neonates < 1 week of age, respectively, and a log-linear interpolation was made between 3.5 kg at age 1 week and 6.3 kg at 2 months for groups with mean ages in that interval.
Table 3 Summary of results from fitting cancer bioassay data: relative susceptibility of different life stages per day of dosing.
Dosing type and age group Maximum likelihood estimate 95% LCL 95% UCL
All continuous chemical dosing experimentsa
Fetal period (8 days beginning on GD12) 4.9 0.5 9.3
Birth–weaning (21 days) 8.7 6.5 10.8
Weaning–60-days (39 days) 0.000 0.000 0.24
All discrete chemical dosing experimentsb
Fetal period (8 days beginning GD12) 5.1 3.6 8.5
Birth–weaning (21 days) 10.5 7.2 16.2
Weaning–60-days (39 days) 1.51 1.03 2.31
All ionizing radiation dosing experimentsc
Fetal period (8 days beginning GD12) 3.5 2.2 5.7
Birth–weaning (21 days) 5.3 3.9 8.3
Weaning–60-days (39 days) 2.4 1.8 3.4
GD, gestation day. Data are maximum likelihood estimates and confidence limits of cancer inductions per dose/(body weight0.75-day) relative to comparably dosed adults.
a Based on a total of 151 group tumor incidence observations for nine chemicals.
b Based on a total of 274 group tumor incidence observations for six chemicals.
c Based on a total of 138 group tumor incidence observations for four radiation types.
Table 4 Comparative results for continuous dosing of chemicals classified as mutagenic versus those classified as nonmutagenic (U.S. EPA 2003b): relative susceptibility of different life stages per day of dosing.
Mutagenicity class and age group Maximum likelihood estimate 95% LCL 95% UCL
Chemicals classified by the U.S. EPA as mutagenica
Fetal period 8.4 3.5 15.5
Birth–weaning 24 17.1 34
Weaning–60-days 3.7 0.0 9.1
Chemicals classified by the U.S. EPA as nonmutagenicb
Fetal period 0.0 0.0 17.4
Birth–weaning 3.0 0.0 4.7
Weaning–60-days 0.0 0.0 2.0
Data are maximum likelihood estimates and confidence limits of cancer inductions per dose/(body weight0.75-day) relative to comparably dosed adults.
a Five compounds, 43 tumor incidence observations.
b Four compounds, 108 tumor incidence observations in animal groups.
Table 5 Comparative results for male versus female animals for mutagenic chemicals given in continuous dosing experiments.
Sex and age group Maximum likelihood estimate 95% LCL 95% UCL
Male
Fetal period 35 16.5 72
Birth–weaning 133 80 245
Weaning–60-days 0.0 0.0 9.7
Female
Fetal period 2.3 0.24 9.7
Birth–weaning 3.4 1.1 8.4
Weaning–60-days 41 18 98
Data are maximum likelihood estimates and confidence limits of cancer inductions per dose/(body weight0.75-day) relative to comparably dosed adults, for continuous dosing for chemicals classified by the U.S. EPA (2003b) as mutagenic (three compounds, 16 tumor incidence observations).
Table 6 Comparative results for male versus female animals for mutagenic chemicals given in discrete dosing experiments.
Sex and age group Maximum likelihood estimate 95% LCL 95% UCL
Male animals
Fetal period 5.7 3.5 11.1
Birth–weaning 11.1 6.6 19.5
Weaning–60-days 1.58 0.99 2.6
Female animals
Fetal period 4.4 2.1 10.2
Birth–weaning 9.7 5.6 20
Weaning–60-days 1.45 0.75 3.2
Data are maximum likelihood estimate and confidence limits of cancer inductions per dose/(body weight0.75-day) relative to comparably dosed adults, for discrete dosing for chemicals classified by the U.S. EPA (2003b) as mutagenic (six compounds, 137 tumor incidence observations).
Table 7 Comparative results for male versus female animals for radiation dosing experiments.
Sex and age group Maximum likelihood estimate 95% LCL 95% UCL
Male animalsa
Fetal period 7.4 3.2 43
Birth–weaning No data No data No data
Weaning–60-days 2.3 1.6 3.3
Female animalsb
Fetal period 2.7 1.5 5.4
Birth–weaning 4.7 3.4 8.7
Weaning–60-days 2.4 1.4 4.6
Data are maximum likelihood estimates and confidence limits of cancer inductions per dose in rads or grays relative to comparably dosed adults.
a Sixty-six tumor incidence observations for two forms of radiation (X rays and neutrons).
b Sixty-nine tumor incidence observations for three forms of radiation (gamma rays, neutrons, and internal exposure to beta rays from tritiated water).
Table 8 Comparative results for male versus female animals for mutagenic chemicals: analysis of combined data from continuous and discrete dosing experiments.
Sex and age group Maximum likelihood estimate 95% LCL 95% UCL Arithmetic mean
Male animals
Fetal period 25 15.6 42 27
Birth–weaning 57 38 90 59
Weaning–60-days 5.0 3.1 8.6 5.3
Female animals
Fetal period 1.77 1.05 2.9 1.83
Birth–weaning 4.4 3.3 6.0 4.5
Weaning–60-days 0.82 0.50 1.29 0.85
Data are maximum likelihood estimates and confidence limits of cancer inductions per dose/(body weight0.75-day) relative to comparably dosed adults (nine compounds, 153 tumor incidence observations).
Table 9 Comparative results for mice versus rats in combined discrete plus continuous dosing experiments.
Species and age group Maximum likelihood estimate 95% LCL 95% UCL
Micea
Fetal period 6.5 4.2 9.9
Birth–weaning 17.7 13.2 24
Weaning–60-days 2.3 1.53 3.3
Ratsb
Fetal period 18.9 8.3 45
Birth–weaning 21 11.7 38
Weaning–60-days 3.9 1.94 7.3
Data are maximum likelihood estimates and confidence limits of cancer inductions per dose/(body weight0.75-day) relative to comparably dosed adults: discrete plus continuous dosing for chemicals classified by the U.S. EPA (2003b) as mutagenic.
a Eight compounds, 265 tumor incidence observations.
b Four compounds, 44 tumor incidence observations.
Table 10 Comparative results for discrete dosing of chemicals for direct-acting nitrosoureas versus other mutagenic carcinogens thought to require metabolic activation to DNA-reactive compounds: standard breakdown of life stages.
Metabolism class and age group Maximum likelihood estimate 95% LCL 95% UCL
Direct-acting mutagenic carcinogensa
Fetal period 11.6 5.4 25
Birth–weaning 10.2 5.1 21
Weaning–60-days 2.7 1.37 5.6
Metabolically activated mutagenic carcinogensb
Fetal period 0.21 0.01 0.90
Birth–weaning 15.0 8.4 33
Weaning–60-days 1.24 0.76 2.3
Data are maximum likelihood estimates and confidence limits of cancer inductions per dose/(body weight0.75-day) relative to comparably dosed adults.
a Ethylnitrosourea and methylnitrosourea (108 tumor incidence observations).
b Benzo(a)pyrene, diethylnitrosamine, dimethylbenzanthracene, and urethane (166 tumor incidence observations in animal groups).
Table 11 Comparative results for discrete dosing of chemicals for direct-acting nitrosoureas versus other mutagenic carcinogens thought to require metabolic activation to DNA-reactive compounds: expanded breakdown of ages.
Metabolism class and age group Maximum likelihood estimate 95% LCL 95% UCL
Direct acting mutagenic carcinogensa
Fetal period 4.4 2.0 12.4
Day 1 6.2 3.6 18.0
Other birth–weaning (except 1 or 21 days) 3.7 1.8 10.0
Day 21 2.2 1.44 4.9
> 21 weaning–60-days 0.92 0.38 2.7
Metabolically activated mutagenic carcinogensb
Fetal period 0.13 0.01 0.52
Day 1 17.3 10.0 36
Other birth–weaning (except 1 or 21 days) 10.7 6.2 22
Day 21 1.9 1.06 3.7
> 21 weaning–60-days 0.87 0.54 1.52
Data are maximum likelihood estimate and confidence limits of cancer inductions per dose/(body weight0.75-day) relative to comparably dosed adults.
a Ethylnitrosourea and methylnitrosourea (108 tumor incidence observations).
b Benzo(a)pyrene, diethylnitrosamine, dimethylbenzanthracene, and urethane (166 tumor incidence observations in animal groups).
Table 12 Effect of separate estimation of relative sensitivity in the birth–weaning period for lactational exposures versus direct administration: combined continuous and discrete dosing data for nine mutagenic carcinogens (317 tumor incidence observations).
Dosing mode and age group Maximum likelihood estimate 95% LCL 95% UCL
Fetal period 6.0 5.5 8.8
Birth–weaning direct 11.6 8.5 16.1
Birth–weaning lactational 21.4 15.3 30
Weaning–60-days 1.70 0.77 2.4
Data are maximum likelihood estimates and confidence limits of cancer inductions per dose/(body weight0.75-day) relative to comparably dosed adults: discrete + continuous dosing for chemicals classified by the U.S. EPA (2003b) as mutagenic.
Table 13 Relative sensitivity for radiation-related carcinogenesis indicated by an expanded breakdown of adult age groups: all ionizing radiation dosing experiments (based on a total of 138 group tumor incidence observations for four radiation types).
Age group Maximum likelihood estimate 95% LCL 95% UCL
Fetal period 2.1 1.3 3.4
Birth–weaning 3.1 2.2 4.8
Weaning–60-days 1.5 1.1 2.1
6–12 months 0.32 0.00 0.69
Elderly (19–21 months) 0.36 0.19 0.60
Data are maximum likelihood estimates and confidence limits of cancer inductions per rads or grays relative to young adults (90–105 days).
==== Refs
References
Basu TK Dickerson JWT Parke DVW 1971 Effect of development on the activity of microsomal drug-metabolizing enzymes in rat liver Biochem J 124 19 24 5126471
Boxenbaum H 1982 Interspecies scaling, allometry, physiological time, and the ground plan of pharmacokinetics J Pharmacokinet Biopharm 10 2 201 227 7120049
Cahill DF Wright JF Godbold JH Ward JM Laskey JW Tompkins EA 1975 Neoplastic and life-span effects of chronic exposure to tritium. II. Rats exposed in utero J Natl Cancer Inst 55 1165 1169 1206742
Castanera TJ Jones DC Kimeldorf DJ Rosen VJ 1971 The effect of age at exposure to a sublethal dose of fast neutrons on tumorigenesis in the male rat Cancer Res 31 1543 1549 5121658
Di Majo V Coppola M Rebessi S Covelli V 1990 Age-related susceptibility of mouse liver to induction of tumors by neutrons Radiat Res 124 227 234 2247603
Federal Coordinating Council for Science, Engineering and Technology 1992 Draft report: a cross-species scaling factor for carcinogen risk assessment based on equivalence of mg/kg3/4 /day Fed Reg 57 24152 24173
Fialkow PJ 1977. Clonal origin and stem cell evolution of human tumors. In: Genetics of Human Cancer (Mulvihill JJ, Miller RW, Fraumeni JF, eds). New York:Raven Press, 439–452.
Ginsberg G Hattis D Sonawane B Russ A Banati P Kozlak M 2002 Evaluation of child/adult pharmacokinetic differences from a database derived from the therapeutic drug literature Toxicol Sci 66 185 200 11896285
Haas CN 1994 Dose response analysis using spreadsheets Risk Anal 14 1097 1100 7846318
Hattis D 2004. Human Interindividual Variability Homepage. Available: http://www2.clarku.edu/faculty/dhattis [accessed 15 June 2004].
Hattis D Anderson E 1999 What should be the implications of uncertainty, variability, and inherent “biases”/”conservatism” for risk management decision making? Risk Anal 19 95 107
Hattis D Burmaster DE 1994 Assessment of variability and uncertainty distributions for practical risk analyses Risk Anal 14 713 730
Hattis D Ginsberg G Sonawane B Smolenski S Russ A Kozlak M 2003 Differences in pharmacokinetics between children and adults—II. Children’s variability in drug elimination half-lives and in some parameters needed for physiologically-based pharmacokinetic modeling Risk Anal 23 117 142 12635728
Knowles JF 1985 Changing sensitivity of neonatal rats to tumorigenic effects of N -nitroso-N -ethylurea and X-radiation, given singly or combined J Natl Cancer Inst 74 853 857 3886991
Knudson AG 1973 Mutation and human cancer Adv Cancer Res 17 317 352
Knudson AG 1977 Genetics and etiology of human cancer Adv Hum Genet 8 1 66 20760
MacLeod SM Renton KW Eade NR 1972 Development of hepatic microsomal drug-oxidizing enzymes in immature male and female rats J Pharmacol Exp Ther 183 489 498 4344247
Maltoni C Lefemine G Ciliberti A Cotti G Carretti D 1984. Experimental research on vinyl chloride carcinogenesis. In: Archives of Research on Industrial Carcinogenesis, Vol 2 (Maltoni C, Mehlman MA, eds). Princeton, NJ:Princeton Scientific Publishers, 1–533.
McCullagh P Nelder JA 1989. General Linear Models. 2nd ed. London:Chapman and Hall.
National Research Council 2002. Estimating the Public Health Benefits of Proposed Air Pollution Regulations. Washington, DC:National Academy Press.
Nomura T 1976 Comparison of tumour susceptibility among various organs of foetal, young and adult ICR/Jcl mice Br J Cancer 33 521 533 179560
NTP 1993. Toxicology and Carcinogenesis Studies of 1,3-Butadiene (CAS No. 77-47-4) in B6C3F1 Mice (Inhalation Studies). TR-434. Research Triangle Park, NC:National Toxicology Program.
NTP 1999. Toxicology and Carcinogenesis Studies of Pentachlorophenol (CAS No. 87-86-5) in F344/N Rats (Feed Studies). TR-483. Research Triangle Park, NC:National Toxicology Program.
OMB 2000. Guidelines to Standardize Measures of Costs and Benefits and the Format of Accounting Statements. Memorandum from Jacob J. Lew, 22 March 2000. Washington, DC:Office of Management and Budget. Available: http://www.whitehouse.gov/omb/memoranda/m00-08.pdf [accessed 11 May 2004].
Peto R Gray R Brantom P Grasso P 1984 Nitrosamine carcinogenesis in 5120 rodents: chronic administration of sixteen different concentrations of NDEA, NDMA, NPYR and NPIP in the water of 4440 inbred rats, with parallel studies on NDEA alone of the effect of age of starting (3, 6 or 20 weeks) and of species (rats, mice, or hamsters) IARC Sci Publ 57 627 665 6533057
Peto R Gray R Brantom P Grasso P 1991 Effects on 4080 rats of chronic ingestion of N -nitrosodiethylamine or N -nitrosodimethylamine: a detailed dose-response study Cancer Res 51 6415 6451 1933906
Sasaki S 1991 Influence of the age of mice at exposure to radiation on life-shortening and carcinogenesis J Radiat Res 32 suppl 2 73 85 1823369
Terracini B Testa MC Cabral JB Rossi L 1976 The roles of age at treatment and dose in carcinogenesis in C3HfDp mice with a single administration of N -nitroso-N -methylurea Br J Cancer 33 427 439 1268033
Travis CC White RK 1988 Interspecific scaling of toxicity data Risk Anal 8 119 125 3375501
Travis CC White RK Ward RC 1990 Interspecies extrapolation of pharmacokinetics J Theor Biol 142 285 304 2338826
U.S. EPA 2001. Risk Assessment Guidance for Superfund. Vol III, Part A. Process for Conducting Probabilistic Risk Assessment. EPA/540-R-02-002. Washington, DC:U.S. Environmental Protection Agency.
U.S. EPA 2003a. Draft Final Guidelines for Carcinogen Risk Assessment (External Review Draft). Washington, DC:U.S. Environmental Protection Agency, Risk Assessment Forum. Available: http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=55445 [accessed 15 June 2004].
U.S. EPA 2003b. Supplemental Guidance for Assessing Cancer Susceptibility from Early-Life Exposure to Carcinogens (External Review Draft). Washington, DC:U.S. Environmental Protection Agency, Risk Assessment Forum. Available: http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=55446 [accessed 11 May 2004].
Venzon DJ Moolgavkar SH 1988 A method for computing profile-likelihood-based confidence intervals Appl Stat 37 87 94
Zhang S Leske DA Lanier WL Holmes JM 2001 Postnatal growth retardation exacerbates acidosis-induced retinopathy in the neonatal rat Curr Eye Res 22 2 133 139 11402390
|
15289159
|
PMC1247474
|
CC0
|
2021-01-04 23:40:10
|
no
|
Environ Health Perspect. 2004 Aug 12; 112(11):1152-1158
|
utf-8
|
Environ Health Perspect
| 2,004 |
10.1289/ehp.6871
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences 10.1289/ehp.6961ehp0112-00115915289160ResearchArticlesBehavioral Alterations in Response to Fear-Provoking Stimuli and Tranylcypromine Induced by Perinatal Exposure to Bisphenol A and Nonylphenol in Male Rats Negishi Takayuki 12Kawasaki Katsuyoshi 23Suzaki Shingo 1Maeda Haruna 1Ishii Yoshiyuki 1Kyuwa Shigeru 1Kuroda Yoichiro 24Yoshikawa Yasuhiro 121Department of Biomedical Science, Graduate School of Agricultural and Life Sciences, University of Tokyo, Tokyo, Japan2Core Research for Evolutional Science and Technology, Japan Science and Technology Agency, Saitama, Japan3Department of Psychology, Hoshi University, Tokyo, Japan4Department of Molecular and Cellular Neurobiology, Tokyo Metropolitan Institute for Neuroscience, Tokyo, JapanAddress correspondence to T. Negishi, Department of Biomedical Science, Graduate School of Agricultural and Life Sciences, University of Tokyo, 1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-8657, Japan. Telephone: 81-3-5841-5037. Fax. 81-3-5841-8186. E-mail:
[email protected] authors declare they have no competing financial interests.
8 2004 26 5 2004 112 11 1159 1164 12 1 2004 26 5 2004 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. The purpose of this study was to examine whether perinatal exposure to two major environmental endocrine-disrupting chemicals, bisphenol A (BPA; 0.1 mg/kg/day orally) and nonylphenol [NP; 0.1 mg/kg/day (low dose) and 10 mg/kg/day (high dose) orally] daily from gestational day 3 to postnatal day 20 (transplacental and lactational exposures) would lead to behavioral alterations in the male offspring of F344 rats. Neither BPA nor NP exposure affected behavioral characteristics in an open-field test (8 weeks of age), in a measurement of spontaneous motor activity (12 weeks of age), or in an elevated plus-maze test (14 weeks of age). A passive avoidance test (13 weeks of age) showed that both BPA- and NP-treated offspring tended to delay entry into a dark compartment. An active avoidance test at 15 weeks of age revealed that BPA-treated offspring showed significantly fewer avoidance responses and low-dose NP-treated offspring exhibited slightly fewer avoidance responses. Furthermore, BPA-treated offspring significantly increased the number of failures to avoid electrical unconditioned stimuli within 5-sec electrical shock presentation compared with the control offspring. In a monoamine-disruption test using 5 mg/kg (intraperitoneal) tranylcypromine (Tcy), a monoamine oxidase inhibitor, both BPA-treated and low-dose NP-treated offspring at 22–24 weeks of age failed to show a significant increment in locomotion in response to Tcy, whereas control and high-dose NP-treated offspring significantly increased locomotion behavior after Tcy injection. In addition, when only saline was injected during a monoamine-disruption test, low-dose NP-treated offspring showed frequent rearing compared with the control offspring. The present results indicate that perinatal low-dose BPA or NP exposure irreversibly influenced the reception of fear-provoking stimuli (e.g., electrical shock), as well as monoaminergic neural pathways.
behaviorbisphenol Afearlearningmonoaminenonylphenol
==== Body
Recently, there has been increasing concern about the exposure of the developing fetus to environmental endocrine-disrupting chemicals (EDCs). The disruption of cognitive function and various behavioral traits due to EDC exposure has been suspected (Schantz and Widholm 2001) because the development of the central nervous system (CNS) is highly regulated by endogenous hormones directly, including gonadal hormones, and by hormonally regulated events that occur early in development. The main purpose of this study was to examine whether perinatal exposure to two well-known environmental EDCs, bisphenol A (BPA) and nonylphenol (NP), can lead to behavioral alterations in the male offspring of F344 rats.
BPA (4,4′-isopropylidene-2-diphenol) is a high-production4volume chemical used in the manufacture of polycarbonate plastics, epoxy resins, and polyester resins. Worldwide production of BPA has increased and will most likely continue to increase in the future (Staples et al. 1998). Human exposure to BPA can occur via BPA-containing products included in certain baby bottles, food containers, resin-based food can linings, and dental sealants. Previous reports revealed that BPA had estrogenic (Gaido et al. 1997; Laws et al. 2000), and antiandrogenic (Sohoni and Sumpter 1998) activity in in vitro and in vivo assays. More recent studies have also identified BPA as an antiandrogen by a yeast two-hybrid system (Lee et al. 2003) and as an antagonist to thyroid hormone activity (Moriyama et al. 2002). These various activities of BPA might exert complicated adverse effects on CNS development because endogenous hormones at appropriate levels at certain limited developmental stages are essential for normal CNS development.
NP (4-nonylphenol) is another environmental EDC with week estrogenic activity (White et al. 1994). NP is used as an additive or surfactant in the manufacture of plastics, and it is a degradation product of nonylphenol polyethoxylates, which are widely used. NP has been shown to have equal or even more estrogenic activity than BPA in in vitro and in vivo assays (Laws et al. 2000). Although weak androgenic NP activity was identified by Sohoni and Sumpter (1998), a more recent study using a yeast two-hybrid system revealed the antiandrogenic effects of NP (Lee et al. 2003). NP may thus have different effects according to the experimental conditions of each assay system. It is therefore also possible that NP exerts a variety of activities under in vivo conditions.
There have been a number of reports suggesting the adverse effects of perinatal exposure to BPA on various behavioral traits in laboratory rodents. In mice, exposure to BPA during fetal development was shown to alter maternal behavior (Palanza et al. 2002) and enhance a methamphetamine-induced abuse state (Suzuki et al. 2003). In rats, alteration of sociosexual behavior (Farabollini et al. 2002), play behavior (Dessi-Fulgheri et al. 2002), and impulsive behavior (Adriani et al. 2003); reduced response to amphetamines (Adriani et al. 2003); and reduced behavioral sexual differentiation (Kubo et al. 2001, 2003) have been demonstrated after perinatal BPA exposure. In our previous report using F344 rats (Negishi et al. 2003b), perinatal exposure to 4 mg/kg/day BPA significantly affected the appropriate avoidance responses of offspring at 8 weeks of age in a shuttle-box avoidance test, suggesting that some alteration took place in response to fear-provoking stimuli; these responses are furthermore known to be controlled by the monoaminergic system (Gingrich 2002; Giorgi et al. 2003; Inoue et al. 1994). When these results are taken together, it appears that perinatal exposure to BPA can interfere with the development of monoaminergic systems, which might in turn be responsible for subtle behavioral changes.
In contrast, only a few studies have reported the effects of perinatal NP exposure on the behavioral traits of the offspring of experimental animals. Ferguson et al. (2000) demonstrated toxicity of NP to mothers and offspring, but found no alterations in open-field activity and running wheel activity in offspring after perinatal NP exposure. Latendresse et al. (2001) reported that the intake of a sodium solution was increased in offspring perinatally treated with NP (2,000 ppm in diet; > 200 mg/kg/day), but instead of focusing on CNS alterations, the authors’ focus was the relationship between this increased intake and the renal toxicity of NP. However, the possibility remains that NP at a much lower (≤ 10 mg/kg/day) dose alters certain cognitive functions and/or fine behavioral characteristics, including the response to fear-provoking stimuli, but without being associated with general motor dysfunction.
In the present study, we examined the adverse effects of low-dose (0.1 mg/kg/day) perinatal BPA or NP exposure on behavioral characteristics. To this end, we performed a series of behavioral tests: an open-field test, a measurement of spontaneous activity during a dark phase, a step-through passive avoidance test, an elevated plus-maze test, and a two-way shuttle-box avoidance test. In addition, in order to evaluate suspected alterations in the monoaminergic system, we investigated behavioral responses to tranylcypromine (Tcy), a monoamine oxidase inhibitor.
Materials and Methods
Animals and treatments.
Male and female F344/N rats were purchased from SLC (Sizuoka, Japan). The animals were maintained under controlled temperature (24 ± 1°C) and humidity (55 ± 5%), on a 12-hr light (09:00–21:00 hr):12-hr dark (21:00–09:00 hr) cycle. Food and water were freely available. After acclimatization for 1 week, female rats were placed with males. Vaginal smears were examined daily: a sperm-positive smear determined gestational day (GD)0. After detection, the pregnant dams were housed individually and were randomly assigned to an exposure condition (n = 10–11/condition). The dams were orally exposed to BPA (0.1 mg/kg/day; Tokyo Kasei Kogyo, Tokyo, Japan) or NP (0.1 or 10 mg/kg/day; Tokyo Kasei Kogyo) dissolved in corn oil, or to corn oil alone (vehicle control; 2 mL/kg/day) from GD3 until postnatal day (PND)20. Oral administrations of BPA and NP were performed by gavage. Because animals were trained to receive the feeding needle before mating, this procedure was not stressful. The dams were examined for clinical signs of toxicity and were weighed daily before dosing. After parturition (PND0), the pups were counted, weighed, and assigned to groups of six pups per litter, maintaining equivalent sex distributions when possible. Pups remained with their biological mother. Offspring were weighed and the body weights recorded on PND0, 3, 7, 10, 14, and 21 and again at 8 and 13 weeks of age. We included the mean weight of each littermate in the statistical analysis. Male pups were marked with ink for identification; On PND21, the marked male pups were gathered from different litters and housed together according to treatment group (7–8/cage). The dams were anesthetized with diethyl ether and then sacrificed by exsanguination; the body weights and organ weights (liver, kidney, spleen, and thymus) were then recorded.
We randomly selected one male pup per litter to undergo a series of behavioral tests (n = 9–10/group). The remaining male pups in the litter were subjected to the measurement of organ weights (liver, kidney, spleen, thymus, brain, and testis) at weaning (PND21) or at 8 weeks of age. Although male rats were usually housed in groups according to experimental treatment, rats were housed individually for some behavioral tests. At the end of each behavioral test, rats were again housed in a group according to treatment. In this study, we excluded female pups from behavioral tests because the estrous cycle in mature females affects various behavioral characteristics. When using female animals, it is important to consider the estrous cycle in evaluating the results of behavioral tests that require several consecutive days. This study was approved by the Animal Care and Use Committee of the Graduate School of Agricultural and Life Sciences, University of Tokyo.
Open-field behavior test.
At 8 weeks of age, animals were subjected to an open-field test. Each subject was housed individually for 24 hr before the test. The open-field apparatus was a rectangular field (56 × 39 cm); none of the animals was familiar with this apparatus. Open-field behavior was recorded for 5 min by a video camera positioned above the apparatus during the dark phase (21:30–23:00 hr) under a low white light; responses were automatically analyzed by a computer-assisted system, which classified observed behavior as locomotion, rearing, or “other” behaviors.
Spontaneous motor activity.
We measured spontaneous motor activity in 12-week-old male offspring using a Supermex (Muromachi Kikai, Tokyo, Japan) (Masuo et al. 1997). The Supermex consisted of a sensor monitor, which was mounted above the cage to detect changes in heat across multiple zones of the cage through an array of Fresnel lenses. The body heat radiated by the animal was detected with the sensor head of the monitor, which contained paired infrared pyroelectric detectors. In this manner, the system allowed the monitoring and counting of all spontaneous movements. Each animal was housed individually in the experimental cage—a differently arranged housing cage with food and water freely available—for 24 hr before the measurement to become accustomed to this experimental condition; spontaneous activity was then measured for about 12 hr (starting at 21:00 hr). All counts were automatically totaled and recorded in 2-min intervals. We defined “immobile time” as 2 min with no signal (count = 0).
Passive avoidance test.
We conducted the step-through passive avoidance test when the animals were 13 weeks of age. The test was carried out during the light phase (13:00–17:00 hr), and each animal was housed individually during the test. The passive avoidance apparatus consisted of light and dark compartments. The first time each animal was placed in this apparatus, an electric foot shock (0.25 mA, 3 sec) was delivered to the animal through the grid floor just after the animal had completely left the light compartment for the dark compartment. We recorded the latency period required before each animal entered the dark compartment after having been placed in the light compartment. Twenty-four hours later, a retention trial (with no shock) was performed, and the latency period before entering the dark compartment was recorded. In addition to the traditional measure, we recorded the frequency and percentage of duration of poking into the dark compartment until the animal completely entered the dark compartment in the retention trial. If an animal failed to enter the dark compartment within 20 min, the test was terminated.
Elevated plus-maze test.
The elevated plus-maze apparatus consisted of two open arms (50 × 10 cm) and two closed arms (50 × 10 cm, with 50-cm high walls) extending from a central square platform (10 × 10 cm); arms were arranged so that those of the same type were opposite each other. The apparatus was elevated 60 cm above the floor. At 14 weeks of age, each animal was placed in the central square facing an open arm during the dark phase (21:30–23:00 hr). We then recorded standard spatiotemporal factors for 5 min (i.e., the frequency of entries into the open arms and the closed arms was recorded, whereby “arm entry” was defined as moving the head into an open arm).
Active avoidance test.
At 15 weeks of age, animals were subjected to an active avoidance test in a two-way shuttle-box (Muromachi Kikai, Tokyo, Japan) consisting of two compartments connected to each other by a hole in the wall; this test was carried out during the light phase (13:00–17:00 hr). Each animal was housed individually through the active avoidance test. Each animal was allowed to become accustomed to the shuttle-box apparatus for 5 min before every session; the animal was then subjected to 25 daily trials/session of avoidance conditioning in four consecutive sessions (acquisition test). For each trial, a 5-sec conditioned stimulus (CS), consisting of a buzzer and light, was followed by a 5-sec unconditioned stimulus (UCS), which included a scrambled shock of 0.2 mA delivered through the floor grid. In addition, on the day after the fourth session, each rat performed the extinction test, which is basically the avoidance text without the UCS. Each trial was separated by variable intertrial intervals (10–90 sec between trials; total of 1,250 sec/session). During the acquisition tests in sessions 1–4 and the extinction test, we recorded the percentage of correct avoidance responses, in which the animals moved to the other compartment of the shuttle box within a 5-sec CS in each block of 25 trials. To evaluate further behavioral characteristics in this procedure, we recorded the percentage of failures to avoid the stimulus within 5-sec UCS and the latency periods associated with both the CS and UCS throughout the four acquisition sessions.
Monoamine-disruption test.
Disturbances of the monoaminergic system in the CNS were induced by a single intraperitoneal (i.p.) injection of Tcy (trans-2-phenylcyclopropyl-amine hydrochloride; Sigma-Aldrich, St. Louis, MO, USA). At 22–24 weeks of age, BPA- or NP-treated male offspring were subjected to the monoamine-disruption test. Before the monoamine-disruption tests, we determined the optimal dose of Tcy for the monoamine-disruption test in a different set of male F344 rats (n = 15) at 9 weeks of age. We injected (i.p.) Tcy solution in 0.9% saline at 0, 2, 5, and 10 mg/mL (1 mL/kg) at 16:00 hr (n = 4, 4, 4, and 3, respectively) and then measured spontaneous motor activity as described above. We confirmed that animals treated with 5 mg/kg Tcy showed a high increment of activity at 21:30 hr, when open-field behavior was recorded.
Saline challenge.
On the first day of the monoamine-disruption test, we injected 0.9% saline (1 mL/kg; i.p.) as a vehicle into each rat; 5.5 hr after the injection, we observed and recorded the behavior of the animals in the open-field apparatus for 4 min.
Tcy challenge.
On the day after the saline challenge, we injected 5 mg/kg Tcy (i.p.) into the same animal and recorded open-field behavior for 4 min, as described for the saline challenge. Behavioral analyses in the monoamine-disruption test were performed as described for the open-field test.
Statistical analyses.
We conducted statistical analyses using StatView, Version 5.0 (SAS Institute, Cary, NC, USA). We analyzed the effects of perinatal BPA or NP exposure on maternal body weight increase and the body weight of male offspring by analysis of variance (ANOVA) with one between-subject factor (treatment) and one repeated-measures factor (days). The number of total, male, and female pups;, the organ weights of dams at weaning; and the organ weights of male offspring at PND21 and at 8 weeks of age were analyzed by one-way ANOVA. Behavioral measurements were analyzed by one-way ANOVA, except for the percentages of correct avoidance in the shuttle-box avoidance test, which were assessed by repeated measures of ANOVA over days (sessions). In the analysis of latency in the passive avoidance test, data processed through logarithmic transformations were used for the ANOVA because of their significantly inappropriate distributions with respect to the normal distributions. In each statistical analysis, the effects of BPA and NP exposure were analyzed with respect to the control in the same ANOVA. When the ANOVA produced significant results, we then performed the post hoc Fisher’s protected least-significant difference test for comparisons between groups. The significance level for all tests was set at p < 0.05.
Results
Maternal toxicity and reproductive results.
Oral exposure to BPA or NP showed no statistically significant effect on maternal body weight increase during pregnancy and lactation or on the number of total, male, and female pups (data not shown). All dams in this study delivered their offspring on GD22. There was no significant effect of 40-day exposure to BPA or NP on either body weight or organ weights (data not shown).
Development of male offspring.
Perinatal exposure to BPA or NP had no significant effect on either body weight gain or organ weights of male offspring on PND21 or at 8 weeks of age (data not shown). No male offspring died during the course of the study (> 25 weeks of age).
Open-field test.
In the open-field test, neither BPA nor NP exposure significantly affected the percentage of locomotion [F(3,32) = 0.271, p > 0.5] or the number of rearings [F(3,32) = 0.189, p > 0.5; data not shown].
Spontaneous motor activity.
To assess general motor activity under nonstress conditions, we recorded spontaneous motor activity of male offspring. Neither BPA nor NP exposure had any effect on the rhythm of activity, the total counts of activity [F(3,32) = 0.554, p > 0.5], or the immobile time [F(3,35) = 0.078, p > 0.5] during the 12-hr dark phase (data not shown).
Passive avoidance test.
On shock-presenting day of the passive avoidance test, the subjects readily entered the dark compartment (< 30 sec). During the retention trial 24 hr after shock presentation, ANOVA [F(7,63) = 12.174, p < 0.001] and multiple comparisons revealed that the subjects showed significant hesitation (p < 0.01) to enter the dark compartment compared with the short latency during shock presentation in all of the experimental groups (Figure 1A). Although both BPA- and NP-treated groups tended to remain in the light compartment longer than the control offspring, there was no significant difference in latency periods during the retention trial among the experimental groups. Neither the frequency of poking into the dark [F(3,28) = 1.166, p > 0.1; Figure 1B] nor the percentage of duration of poking into the dark [F(3,28) = 1.919, p > 0.1; Figure 1C] during the retention trial was affected by chemical exposure.
Elevated plus-maze test.
Neither BPA nor NP exposure significantly altered the frequency of entering the open arms [F(3,27) = 0.571, p > 0.5] or the closed arms [F(3,27) = 0.139, p > 0.5] of the elevated plus-maze test, although the frequency of entering the open arms was slightly higher in the BPA-treated group than in the controls (data not shown).
Active avoidance test.
BPA and low-dose NP exposure significantly affected the avoidance responses of the male offspring in the active avoidance test. Repeated-measures one-way ANOVA showed a significant effect of chemical exposure [F(3,35) = 5.724, p < 0.01] and number of sessions [F(3,105) = 107.322, p < 0.0001], as well as a significant interaction between chemical exposure and the number of sessions [F(9,105) = 3.536, p < 0.001]. One-way ANOVAs and post hoc multiple comparisons for each session indicated significantly fewer avoidance responses in BPA-treated offspring at the first, second, and third sessions than in the control offspring (Figure 2A). Low-dose NP-treated offspring showed a lower avoidance rate than the control offspring, but only in the first session (Figure 2B). In the fifth extinction session (i.e., without electrical shocks as the UCS), the BPA- and NP-treated offspring showed slightly less correct avoidance behavior than the control offspring, although the effect of chemical exposure, as determined by one-way ANOVA, was not statistically significant [F(3,35) = 0.571, p = 0.104]. One-way ANOVA indicated a significant effect of chemical exposure and that the frequency of failure of avoidance within 5 sec of shock presentation—in which one-way ANOVA indicated a significant effect of chemical exposure [F(3,35) = 3.700, p < 0.05]—in BPA-treated offspring was significantly higher (p < 0.001) than that in the control offspring; low-dose NP-treated offspring showed a similar tendency (Figure 2C). We found no significant effect of chemical exposure on the mean of the latency periods associated with CS [F(3,35) = 0.722, p > 0.5; Figure 2D] and UCS [F(3,35) = 1.186, p > 0.1; Figure 2E] in 100 trials of four sessions.
Monoamine-disruption test.
Tcy injection led to a large, slow increase in motor activity at 5 and 10 mg/kg compared with the saline control, although 2 mg/kg Tcy did not induce an increase (Figure 3A). We confirmed that the animals administered 5 mg/kg Tcy showed a high increase of activity 5.5 hr after injection, which corresponded to the schedule for the monoamine-disruption test. One-way ANOVA and post hoc tests about locomotion behavior [F(7,58) = 2.498, p < 0.05] and the number of rearing behaviors [F(7,58) = 9.629, p < 0.01] yielded the following results. In the monoamine-disruption test, control and high-dose NP-treated offspring showed a significant increase in locomotion behavior resulting from Tcy injection (p < 0.01; Figure 3B). However, BPA-treated or low-dose NP-treated offspring failed to show a clear increase in locomotion (p > 0.1). Tcy also caused significant decreases in the number of rearing behaviors in all experimental groups (Figure 3C) in the monoamine-disruption test. When only saline was administered, low-dose NP-treated offspring showed a significantly increased number of rearing behaviors, compared with those of the control offspring in the saline challenge; in addition, the BPA-treated offspring also appeared to have a similar tendency (p < 0.1), which was abolished by monoamine disruption by Tcy (Figure 3C).
Discussion
In the present study, we carried out a series of behavioral tests and demonstrated the subtle and complex functional effects of perinatal exposure to BPA and NP on the behavior of male rat offspring. Behavioral alterations by perinatal exposure to BPA and NP were detected only in specific challenges involving fear-provoking stimuli and pharmacologic disruption of monoaminergic system, whereas spontaneous explorative behavior and responses to novelty were not affected by the exposure to these chemicals.
Evaluation of the toxicity of BPA as well as that of NP on maternal body weight, parturition, maternal organ weights at weaning (PND21), and general development (body weight and organ weights) of the offspring confirmed that there were no adverse effects of BPA at 0.1 mg/kg/day or of NP at 0.1 and 10 mg/kg/day, which was consistent with the findings of previous reports (Ferguson et al. 2000; Kwon et al. 2000).
In the present study, we detected no statistically significant alterations by BPA or NP in the spontaneous activity and behaviors of rats in the open-field test at 8 weeks of age and in the elevated plus-maze test at 14 weeks of age. This suggests that these chemical exposures induce no severe abnormalities in general behavior.
In the passive avoidance test, offspring perinatally exposed to BPA or NP seemed to be more sensitive to fear-inducing shock than were the control offspring, which might have led to the somewhat stronger retention in the chemical-treated groups; however, such changes were statistically ambiguous because of the large individual differences in the experimental conditions used in this study.
In the active avoidance test, BPA and low-dose NP showed clear or partial adverse effects on behavior, respectively. In particular, BPA-treated offspring may have been less able to learn than the control offspring in terms of causality. Low-dose NP-treated offspring were also affected to some extent. Although the possibility remains that BPA-treated offspring were more insensitive to electrical shock than were the control offspring, the slight elongation of the latency period in the passive avoidance test would have excluded this possibility. If BPA-treated offspring had found the electrical stimuli less fear-provoking and/or painful, they would have entered the dark compartment more quickly than the control offspring in the passive avoidance test. It would also be unlikely that less learning took place as a result of motor dysfunction or sensory abnormality because there was no alteration in the spontaneous activity during the dark phase and in the duration of locomotion in the open-field test, as well as in the latency periods associated with CS and UCS in the active avoidance test. When electrical shock was presented as a UCS, BPA-treated offspring failed to enter the opposite compartment within 5 sec more frequently than the control offspring, and NP-treated offspring showed a similar tendency. BPA-treated offspring tended to stiffen in the corner of the box during the UCS, and these animals appeared to stop avoiding the UCS more often than did the control offspring, as determined by direct observation (data not shown). It is possible that excessive fear of the UCS would interfere with the smooth progression of avoidance learning. Perinatal BPA exposure may render male offspring exceedingly vulnerable to intolerable levels of fear. Interestingly, this hypothesis may be supported by a previous study (Aloisi et al. 2002), which indicated that perinatal BPA exposure increased the sensitivity of the central neural pathways for nociception in male offspring. Farabollini and colleagues reported the details of various behavioral changes observed due to perinatal exposure to BPA at 0.04 or 0.4 mg/kg/day in rats (Adriani et al. 2003; Aloisi et al. 2002; Dessi-Fulgheri et al. 2002; Farabollini et al. 1999, 2002). In the present study we also provided new evidence of the behavioral adverse effects of perinatal exposure to BPA at a low dose of 0.1 mg/kg/day on electrical UCS-related responses in an active avoidance test.
In the monoamine-disruption test, Tcy-induced increases in locomotion were significantly less marked in BPA-treated and low-dose NP-treated offspring, but not in the high-dose NP-treated offspring, compared with the control offspring. This is the first study reporting behavioral alterations due to perinatal BPA and NP exposure shown by responses to the disruption of monoaminergic systems, with Tcy having clear and straightforward pharmacologic effects as a monoamine oxidase inhibitor. Previous studies have reported changes after BPA exposure in a mouse model of psycho-stimulant abuse (methamphetamine) (Suzuki et al. 2003) and in a rat model of an increment in activity by amphetamine (Adriani et al. 2003), which have complicated pharmacologic effects on CNS function. It is possible that BPA-treated and low-dose NP-treated offspring might be insensitive to monoamines overflowing in excess into the extrasynaptic space. Such animals might show abnormal expressions of each type of monoamine receptor (dopamine, serotonin, and noradrenaline receptors, including the subtypes of each receptor class) or monoamine oxidase in certain region(s) of the CNS. Further investigations considering each of the monoamergic systems (dopaminergic, serotoninergic, and noradrenalinergic) are likely to produce more insight into the mechanism of traces induced by perinatal BPA and NP exposure in the CNS. Although monoamine disruption by Tcy significantly reduced the number of rearing behaviors, the response to Tcy was not influenced by perinatal chemical exposure. There was a discrepancy between the results of behavior in the open-field apparatus at 8 weeks of age and > 22 weeks of age. When the animals were > 22 weeks of age, the observed significant increase in the number of rearing behaviors suggested that low-dose NP-treated offspring might have experienced less anxiety than the control offspring in the open-field apparatus. The results furthermore suggested that these behavioral alterations caused by perinatal low-dose NP exposure might appear only at a stage of advanced age, that is, at a time when rats are relatively slow in their movements and rarely show rearing compared with juveniles. Further investigations will be required in this regard. In any case, the effective dose of NP from a neurobehavioral standpoint is much lower than the dose associated with general physical toxicity, as observed in the case of BPA in our previous study of that substance (Negishi et al. 2003b).
We cannot address differences in sexes in the present study because we limited this behavioral study to the male offspring; our primary goal was to detect behavioral alteration by perinatal BPA exposure in male offspring. However, studies in rats by Kubo et al. (2001, 2003) have demonstrated that perinatal exposure to BPA removed the differences between the sexes in the volume of locus ceruleus and in behavioral characteristics. In addition, some studies have reported sex differences in the effects of BPA (Dessi-Fulgheri et al. 2002; Farabollini et al. 1999, 2002). Further experiments using the active avoidance test and the monoamine-disruption test on both male and female offspring would be informative.
In the present study, one animal per litter sequentially underwent all of the behavioral tests. It is possible that an experience in an earlier behavioral test influenced the results of the subsequent behavioral tests. For example, a painful experience immediately after exploratory behavior in the passive avoidance test might interfere with the behavioral propensity in the elevated plus-maze test. However, we believed that using sequential behavioral tests in the same animal would not obstruct the evaluation of effects of perinatal exposure to chemicals because all of the animals in the four treatment groups experienced the stimuli.
In summary, perinatal exposure to BPA and NP, both at 0.1 mg/kg/day, affected the extent of shock-related behavior and affected responses to the disruption of the mono-aminergic system, although the direct mechanisms of these alterations remain unclear at present. Moreover, the neurobehavioral toxicity of both BPA and NP may be out of proportion with the in vitro and in vivo estrogenic potency of these compounds determined by certain simple assay systems. It may be useful to consider other potencies and/or metabolites of BPA and NP (Moriyama et al. 2002; Yoshihara et al. 2001) in addition to their weak estrogenic activity. We suggest that there may be a causal relationship between behavioral alterations in response to fear-provoking stimuli and abnormality in the mono-aminergic system because both dopamine and serotonin play important roles in the processing of fear-provoking and/or stressful stimuli in the CNS (Gingrich 2002; Giorgi et al. 2003; Inoue et al. 1994). In our recent study using primary cultured neurons (Negishi et al. 2003a), BPA and NP inhibited staurosporine-induced neuronal cell death, interfering with caspase-3 activation. BPA and NP may, in this manner, disrupt programmed neuronal cell death during development, which would irreversibly lead to an abnormal neural network—including the monoaminergic system—and cause behavioral abnormalities in adulthood.
Conclusion
We conclude that perinatal BPA and NP exposure, even at slightly higher doses than those associated with environmental exposure in humans, had adverse behavioral effects on rats, especially when the animals were forced to avoid fear-provoking stimuli such as electrical shocks. Perinatal exposure to BPA and NP disrupted the reception of intolerable stress, which may be due to the alterations in monoaminergic system.
Figure 1 Effect of perinatal exposure (mean ± SE) to BPA or NP (mg/kg/day) on the behavioral characteristics in a passive avoidance test (n = 8/group). Abbreviations: Post, latency during the retention trial; Pre, latency on shock-presenting day. (A) The latency period until the animals completely entered the dark compartment. (B) The frequency of poking into a dark box until complete entrance. (C) The percent duration of poking into a dark box until complete entrance.
**p < 0.01 compared with Pre for same treatment.
Figure 2 Effects of perinatal exposure to BPA or NP (mg/kg/day) on behavioral characteristics in a shuttle-box avoidance test (mean ± SE; n = 9–10/group). Avoidance learning curves of male offspring perinatally exposed to BPA (A) or NP (B). (C) Percentage of failure of avoidance when an electrical shock was presented for 5 sec among 100 trials of four sessions. Length of latency period associated with a CS (D) and a UCS (E) in four sessions.
*p < 0.05;
**p < 0.01; and #p < 0.001 compared with control.
Figure 3 Effects of perinatal exposure to BPA or NP (mg/kg/day) on behavior in the monoamine-disruption test 5.5 hr after Tcy treatment. NS, not significant. (A) Locomotor activity (mean ± SE) after a single injection with 2, 5, or 10 mg/kg/day Tcy (n = 3–4/group). (B) Effect of perinatal BPA and NP on Tcy (5 mg/kg)-induced increases in locomotion behavior in an open-field apparatus (mean ± SE; n = 7–9/group). (C) Effects of perinatal BPA and NP on Tcy (5 mg/kg)-induced suppression of rearing in an open-field apparatus (mean ± SE; n = 7–9/group). aThe bracket indicates the comparison of offspring exposed to BPA and low-dose NP in the saline challenge.
*p < 0.05.
**p < 0.01 compared with saline control.
==== Refs
References
Adriani W Seta DD Dess-Fulgheri F Farabollini F Laviola G 2003 Altered profiles of spontaneous novelty seeking, impulsive behavior, and response to d -amphetamine in rats perinatally exposed to bisphenol A Environ Health Perspect 111 395 401 12676589
Aloisi AM Della Seta D Rendo C Ceccarelli I Scaramuzzino A Farabollini F 2002 Exposure to the estrogenic pollutant bisphenol A affects pain behavior induced by subcutaneous formalin injection in male and female rats Brain Res 937 1 7 12020856
Dessi-Fulgheri F Porrini S Farabollini F 2002 Effects of perinatal exposure to bisphenol A on play behavior of female and male juvenile rats Environ Health Perspect 110 suppl 3 403 407 12060836
Farabollini F Porrini S Della Seta D Bianchi F Dessi-Fulgheri F 2002 Effects of perinatal exposure to bisphenol A on socio-sexual behavior of female and male rats Environ Health Perspect 110 suppl 3 409 414 12060837
Farabollini F Porrini S Dessi-Fulgherit F 1999 Perinatal exposure to the estrogenic pollutant bisphenol A affects behavior in male and female rats Pharmacol Biochem Behav 64 687 694 10593191
Ferguson SA Flynn KM Delclos KB Newbold RR 2000 Maternal and offspring toxicity but few sexually dimorphic behavioral alterations result from nonylphenol exposure Neurotoxicol Teratol 22 583 591 10974597
Gaido KW Leonard LS Lovell S Gould JC Babai D Portier CJ 1997 Evaluation of chemicals with endocrine modulating activity in a yeast-based steroid hormone receptor gene transcription assay Toxicol Appl Pharmacol 143 205 212 9073609
Gingrich JA 2002 Mutational analysis of the serotonergic system: recent findings using knockout mice Curr Drug Target CNS Neurol Disord 1 449 465
Giorgi O Piras G Lecca D Hansson S Driscoll P Corda MG 2003 Differential neurochemical properties of central serotonergic transmission in Roman high- and low-avoidance rats J Neurochem 86 422 431 12871583
Inoue T Tsuchiya K Koyama T 1994 Regional changes in dopamine and serotonin activation with various intensity of physical and psychological stress in the rat brain Pharmacol Biochem Behav 49 911 920 7886107
Kubo K Arai O Ogata R Omura M Hori T Aou S 2001 Exposure to bisphenol A during the fetal and suckling periods disrupts sexual differentiation of the locus coeruleus and of behavior in the rat Neurosci Lett 304 73 76 11335058
Kubo K Arai O Omura M Watanabe R Ogata R Aou S 2003 Low dose effects of bisphenol A on sexual differentiation of the brain and behavior in rats Neurosci Res 45 345 356 12631470
Kwon S Stedman DB Elswick BA Cattley RC Welsch F 2000 Pubertal development and reproductive functions of Crl:CD BR Sprague-Dawley rats exposed to bisphenol A during prenatal and postnatal development Toxicol Sci 55 399 406 10828273
Latendresse JR Newbold RR Weis CC Delclos KB 2001 Polycystic kidney disease induced in F(1) Sprague-Dawley rats fed para -nonylphenol in a soy-free, casein-containing diet Toxicol Sci 62 140 147 11399801
Laws SC Carey SA Ferrell JM Bodman GJ Cooper RL 2000 Estrogenic activity of octylphenol, nonylphenol, bisphenol A and methoxychlor in rats Toxicol Sci 54 154 167 10746942
Lee HJ Chattopadhyay S Gong EY Ahn RS Lee K 2003 Antiandrogenic effects of bisphenol A and nonylphenol on the function of androgen receptor Toxicol Sci 75 40 46 12805653
Masuo Y Matsumoto Y Morita S Noguchi J 1997 A novel method for counting spontaneous motor activity in the rat Brain Res Brain Res Protoc 1 321 326 9384810
Moriyama K Tagami T Akamizu T Usui T Saijo M Kanamoto N 2002 Thyroid hormone action is disrupted by bisphenol A as an antagonist J Clin Endocrinol Metab 87 5185 5190 12414890
Negishi T Ishii Y Kyuwa S Kuroda Y Yoshikawa Y 2003a Inhibition of staurosporine-induced neuronal cell death by bisphenol A and nonylphenol in primary cultured rat hippocampal and cortical neurons Neurosci Lett 353 99 102 14664910
Negishi T Kawasaki K Takatori A Ishii Y Kyuwa S Kuroda Y 2003b Effects of perinatal exposure to bisphenol A on the behavior of offspring in F344 rats Environ Toxicol Pharmacol 14 99 108 21782668
Palanza PL Howdeshell KL Parmigiani S vom Saal FS 2002 Exposure to a low dose of bisphenol A during fetal life or in adulthood alters maternal behavior in mice Environ Health Perspect 110 suppl 3 415 422 12060838
Schantz SL Widholm JJ 2001 Cognitive effects of endocrine-disrupting chemicals in animals Environ Health Perspect 109 1197 1206 11748026
Sohoni P Sumpter JP 1998 Several environmental oestrogens are also anti-androgens J Endocrinol 158 327 339 9846162
Staples CA Dorn PB Klecka GM O’Block ST Harris LR 1998 A review of the environmental fate, effects, and exposures of bisphenol A Chemosphere 36 2149 2173 9566294
Suzuki T Mizuo K Nakazawa H Funae Y Fushiki S Fukushima S 2003 Prenatal and neonatal exposure to bisphenol-A enhances the central dopamine D1 receptor-mediated action in mice: enhancement of the methamphetamine-induced abuse state Neuroscience 117 639 644 12617968
White R Jobling S Hoare SA Sumpter JP Parker MG 1994 Environmentally persistent alkylphenolic compounds are estrogenic Endocrinology 135 175 182 8013351
Yoshihara S Makishima M Suzuki N Ohta S 2001 Metabolic activation of bisphenol A by rat liver S9 fraction Toxicol Sci 62 221 227 11452134
|
15289160
|
PMC1247475
|
CC0
|
2021-01-04 23:40:10
|
no
|
Environ Health Perspect. 2004 Aug 26; 112(11):1159-1164
|
utf-8
|
Environ Health Perspect
| 2,004 |
10.1289/ehp.6961
|
oa_comm
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.