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
Arthritis Res TherArthritis Research & Therapy1478-63541478-6362BioMed Central London ar18191627768010.1186/ar1819Research ArticleTumour necrosis factor-α stimulates dehydroepiandrosterone metabolism in human fibroblast-like synoviocytes: a role for nuclear factor-κB and activator protein-1 in the regulation of expression of cytochrome p450 enzyme 7b Dulos John [email protected] Allard [email protected] Annemieke [email protected] Cobi [email protected] Annemieke [email protected] Department of Pharmacology, Section Autoimmunity, N.V. Organon, Oss, The Netherlands2 Laboratory for Psychoneuroimmunology, University Medical Center Utrecht, Utrecht, The Netherlands2005 15 9 2005 7 6 R1271 R1280 17 5 2005 27 6 2005 4 8 2005 11 8 2005 Copyright © 2005 Dulos 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.
Glucocorticoids have successfully been used in the treatment of rheumatoid arthritis. Data suggest that 7α-hydroxy-dehydroepiandrosterone (7α-OH-DHEA), an immunostimulating metabolite of dehydroepiandrosterone, can block glucocorticoid-induced immune suppression. Formation of 7α-OH-DHEA is catalyzed by activity of cytochrome p450 enzyme 7b (Cyp7b). Recently, we reported that tumour necrosis factor (TNF)-α, IL-1α, IL-1β and IL-17 enhance Cyp7b mRNA expression and induce a concomitant increase in the formation of 7α-OH-DHEA by fibroblast-like synoviocytes (FLS) from rheumatoid arthritis patients. The aim of this study was to elucidate which signal transduction pathway is involved in the TNF-α-mediated induction of Cyp7b activity in FLS. We studied the effects of inhibitors of different signal transduction pathways on Cyp7b activity in FLS by measuring Cyp7b mRNA expression using reverse transcription PCR and by measuring the formation of 7α-OH-DHEA. We applied SN50, an inhibitor of nuclear translocation of transcription factors (i.e. activator protein-1 [AP-1] and nuclear factor-κB [NF-κB]); PSI, a proteasome inhibitor that prevents IκB degradation and thereby NF-κB release; SP600125, a c-Jun N-terminal kinase (JNK) inhibitor; and the mitogen-activated protein kinase inhibitors PD98059 (extracellular signal-regulated kinase) and SB203580 (p38). Cyp7b is constitutively expressed in RA FLS and can be activated in response to TNF-α. SN50 and PSI prevented the TNF-α-induced increase in Cyp7b activity, whereas the mitogen-activated protein kinase inhibitors PD98059 and SB203580 had no effect. In addition, inhibition of Cyp7b mRNA expression and activity was observed with SN50, PSI and SP600125, suggesting that NF-κB and AP-1 induce Cyp7b transcription. These findings suggest that NF-κB and AP-1 are involved in the TNF-α-enhanced formation of the dehydroepiandrosterone metabolite 7α-OH-DHEA. Our results are in accordance with presence of AP-1 and NF-κB binding sites in the Cyp7b promoter.
==== Body
Introduction
Rheumatoid arthritis (RA) is a chronic inflammatory disease characterized by hyperplasia of fibroblast-like synoviocytes (FLS), which is regarded to be important in cartilage and bone erosion [1]. Steroids such as dehydroepiandrosterone (DHEA), glucocorticoids, androgens and oestrogens have been shown to modulate the disease process in RA [2]. Several authors have suggested that the natural, abundantly present steroid DHEA may have immunostimulating effects [3,4]. Further data indicate that the 7α-hydroxy-dehydroepiandrosterone (7α-OH-DHEA) metabolite of DHEA, rather than DHEA itself, is responsible for these immunostimulating effects [5,6]. In several studies 7α-OH-DHEA was found to stimulate the immune system both in vitro and in vivo, and it has been suggested that 7α-OH-DHEA acts as an antiglucocorticoid [6,7].
The conversion of DHEA into 7α-OH-DHEA is catalyzed by cytochrome p450 enzyme 7b (Cyp7b) [8]. Because of the reported immunostimulating effects of 7α-OH-DHEA, we previously investigated the relation between Cyp7b activity and arthritis. We showed that the severity of murine collagen-induced arthritis was associated with an increase in Cyp7b activity and Cyp7b mRNA level in synovial biopsies [9].
Recently, we reported that Cyp7b mRNA expression and Cyp7b activity are present in FLS from patients with RA [10]. In addition, expression of Cyp7b in RA FLS was enhanced after in vitro treatment of these cells with tumour necrosis factor (TNF)-α, IL-1α, IL-1β and IL-17 [10]. TNF-α is abundantly produced in inflamed joints and is known to play a crucial role in the pathogenesis of RA [11]. Therefore, in the present study we used TNF-α to investigate which signal transduction pathway is involved in the TNF-α-mediated increase in Cyp7b activity in human FLS. Signaling pathways that mediate the effects of TNF-α include mitogen-activated protein kinases (MAPKs) and nuclear factor-κB (NF-κB) [12]. Three MAPK families have been implicated to play a role in RA, including extracellular signal (mitogenic)-regulated protein kinase (ERK)1/2; the stress-activated protein kinases, also called c-Jun NH2-terminal kinases (JNKs); and the p38 MAPKs [13]. The JNK pathway is of interest because of its capacity to phosphorylate the amino acids serine-63 and -73 on the c-Jun activation domain, which is a component of activator protein-1 (AP-1). AP-1 transcription factors consist of homodimers and heterodimers of the Jun and Fos family [14]. Apart from MAPKs, TNF-α activates nuclear translocation of NF-κB, which plays a central role in inflammatory diseases such as RA through induction of transcription of proinflammatory genes [15]. NF-κB is retained in the cytosol of nonstimulated cells by a noncovalent interaction with IκB. Upon stimulation by TNF-α, IκB is degraded and NF-κB is released and translocated to the nucleus inducing inflammatory gene expression [15].
Previous studies implicated a role for TNF receptor I in the regulation of Cyp7b activity [10], but these studies were inconclusive regarding the role played by TNF receptor II in regulation of Cyp7b activity. Thus, in order to study which signaling pathways are involved in TNF-α-induced Cyp7b activity, we used different inhibitors with relevance to TNF receptor signaling. SN50 was initially described as an inhibitor of nuclear translocation of NF-κB. However, in addition to its effect on NF-κB, SN50 blocks nuclear translocation of the AP-1 transcription factor [16,17]. For that purpose, the effect of SP600125 – a recently described inhibitor of JNK – on Cyp7b mRNA expression and activity was assessed [16]. The proteasome inhibitor PSI prevents degradation of IκB and thereby indirectly prevents NF-κB nuclear translocation [18]. To determine a possible role for MAPKs other than JNK in the TNF-α-induced Cyp7b activity, the ERK1/2 inhibitor PD98059 and the p38 inhibitor SB203580 were used.
In the present study we report that NF-κB and AP-1, but not ERK1/2 and p38, are probably involved in TNF-α-stimulated formation of 7α-OH-DHEA.
Materials and methods
Fibroblast-like synoviocytes
FLS cell lines were developed from synovial biopsies obtained from RA patients, after informed consent had been granted. All patients fulfilled the 1987 American College of Rheumatology criteria [19]. FLS were phenotyped as CD55+ synovial fibroblasts, as described previously [20]. Briefly, the synovial tissue was minced and digested for 2 hours with 1 mg/ml collagenase A in Dulbecco's modified Eagle's medium (DMEM) at 37°C. The tissue homogenate was filtered through a fine sieve (200 μm), washed and cultured overnight in synoviocyte medium (Tebu-Bio, Heerhugowaard, The Netherlands) in 5% carbon dioxide and 37°C to allow separation of adherent cells from the nonadherent cell population. Nonadherent cells were separated and adherent cells were cultured further in synoviocyte medium. The cells morphologically presenting as FLS were used between passages 2 and 17 in the experiments.
Antibodies and reagents
The anti-NF-κB-p65 was from Signal Transductions (Becton & Dickinson, Woerden, The Netherlands), and the biotinylated anti-mouse IgG antibody was from Brunschwig Chemie (Amsterdam, The Netherlands). TNF-α was bought from Peprotech (Tebu-Bio, Heerhugowaard, The Netherlands). The p38 MAPK inhibitor SB203580 and the ERK1/2-MAPK kinase (MEK)-1 inhibitor PD98059 were from Omnilabo (Breda, The Netherlands), dissolved in dimethylsulfoxide or methanol and used as controls. The proteasome inhibitor PSI and the JNK inhibitor SP600125 were purchased at Omnilabo (Breda, The Netherlands) and dissolved in dimethylsulfoxide. The SN50 peptide (Biomol, Plymouth, USA) was dissolved in DMEM/Ham's F-12 medium.
Measurement of TNF-α-induced Cyp7b activity in fibroblast-like synoviocytes
In order to arrest cell growth, synoviocyte medium was replaced by DMEM/Ham's F-12 medium with 10% foetal calf serum (FCS) and the FLS were cultured for another 3 days in a 24-well plate (Greiner, Alphen a/d Rijn, The Netherlands). FLS were preincubated in the presence or absence of SN50 for 2 hours, or PSI, SP600125, SB203580, or PD98059 for 1 hour in 2% charcoal-treated (depleted from steroids) FCS. Charcoal-treated FCS were prepared by suspending charcoal (Norit A) in Tris buffer. The suspension was then centrifuged for 10 min at 8.000 N/kg, the supernatant was removed and FCS added to the residue. This suspension was stirred for 30 min at 45°C and the charcoal was removed by centrifugation for 10 min at 8.000 N/kg. The supernatant was sterilized by membrane filtration using filters of pore sizes 0.8 and 0.2 μm successively. Following heat inactivation, FCS was stored at -20°C until use.
FLS were incubated with or without TNF-α and 1,2,6,7-[3H]-DHEA (1.5 × 10E-8 mol/l: NEN Life Science Products, Boston, MA, USA) for 24 hours. Steroid-containing medium (1 ml) was passed over a C18 Solid Phase Extraction cartridge (Sopachem, Wageningen, The Netherlands) to determine the conversion of 1,2,6,7-[3H]-DHEA into 3H-labelled 7α-OH-DHEA as a measure of Cyp7b activity. Steroids were eluted from the column with methanol. Next, 3H-labelled 7α-OH-DHEA and 3H-labelled DHEA were measured using high-performance liquid chromatography. The amount of 7α-OH-DHEA is expressed as the percentage of 3H-labelled 7α-OH-DHEA of the total amount of 3H-label measured. Recoveries after extraction were in the range 85–95%, and identification of 7α-OH-DHEA was confirmed by Gas Chromatography-Mass spectrometry GC-MS (data not shown).
Detection of 7α-OH-DHEA levels by radioimmunoassay
To determine 7α-OH-DHEA levels in FLS, a radioimmunoassay was performed using antiserum against 7α-OH-DHEA. The 7α-OH-DHEA metabolite is formed by the activity of the enzyme Cyp7b. The radioimmunoassay was performed at the Institute of Endocrinology at Prague (Czech Republic) in cooperation with Dr R Hampl [21]. In brief, FLS were preincubated in the presence or absence of SN50 for 2 hours or PSI, SP600125, SB203580, or PD98059 for 1 hour in 2% charcoal-treated (depleted from steroids) FCS. Thereafter, FLS were incubated with or without TNF-α and 1.5 × 10E-8 mol/l DHEA (Diosynth, Oss, The Netherlands) for 24 hours. Extraction was carried out using diethyl ether. Diethyl ether extracts containing 7α-OH-DHEA and 7β-OH-DHEA were evaporated under nitrogen, and the dry residue was dissolved in assay buffer and measured using radioimmunoassay as previously described [21].
Immunohistochemistry of fibroblast-like synoviocytes
FLS were grown on chamber slides (Nalge Nunc International; Fisher Emergo, Landsmeer, The Netherlands) and preincubated for 2 hours in the presence or absence of SN50 (100 μg/ml or 200 μg/ml) and thereafter stimulated for 30 min with TNF-α (0.5 ng/ml). After washing with phosphate-buffered saline (PBS), cells were fixed in methanol for 10 min and dried. The samples were blocked with buffer containing 2% normal goat serum, 2% human serum, and 2% serum albumin in PBS/0.01% Triton X-100 (PBS/T) for 30 min. Cells were then incubated with anti-NF-κB p65 antibody in the same buffer for 1 hour at ambient temperature. After washing with PBS-T, the FLS were incubated for 45 min with biotinylated anti-mouse IgG. After washing, cells were incubated for 30 min with avidin-biotin-peroxidase (Brunschwig Chemie, Amsterdam, The Netherlands). Following washing, the substrate was incubated for 10 min with enhanced diaminobenzidine in stable peroxidase buffer (Pierce; Perbio Science, Etten-Leur, The Netherlands). Following extensive washing in milli-Q water and dehydration, coverslips were placed with Entellan (Merck, Amsterdam, The Netherlands) mounting medium. Slides were visually analyzed under a Nikon Alphaphot-2 microscope (Uvikon, Bunnik, The Netherlands).
Cyp7b mRNA levels in fibroblast-like synoviocytes
FLS were preincubated with 200 μg/ml SN50 and then incubated in the presence or absence of TNF-α (0.5 ng/ml) for 6 hours. Cells were washed with PBS and total RNA was extracted with RNAzol (Campro, Veenendaal, The Netherlands). cDNA synthesis was done according to the manufacturer's protocol using random hexamer primers (Pharmacia, Woerden, The Netherlands) and reverse transcriptase (Pharmacia). For reverse transcription PCR, human Cyp7b sense (GTCCTGGAGAAATATTATGTGCAG) and antisense (CGCACACAGTAGTCCCCGG) primers were used. For GAPDH we used CCCTTCATTGACCTCAACTACATGG (sense) and GGTCCACCACCCTGTTGCTGTAGCC (antisense) as primers. Reverse transcription PCR was carried out using an Applied Biosystems (Nieuwerkerk a/d ijssel, The Netherlands) thermo cycler with an anneal temperature of 53°C.
Computer analysis of the Cyp7b promoter region
The promoter sequence of the human Cyp7b gene was identified and exported from the Ensembl database (vs19.34b.2; 9 February 2004) using the MartView export function. As promoter region, -1,000 to +100 nucleotides were selected in relation to the transcription start site. Promoter analysis for transcription factor binding sites was performed using the GEMSLauncher version 3.6 from Genomatrix and MatInspector professional release 7 [22]. Core and matrix similarity settings were 0.75 and optimized -0.03, respectively. The transcription factor family matrices V$AP1F, V$NFAT, V$NFKB and V$STAT were used.
Results
SN50 inhibited TNF-α-stimulated Cyp7b expression and activity
An FLS cell-line (SCRO.14.SF), obtained from a synovial biopsy from an RA patient, was used to study the effect of SN50 on the TNF-α-induced Cyp7b activity. SN50 (200 μg/ml) significantly reduced basal Cyp7b activity (Fig. 1a). Importantly, the increase in Cyp7b activity following stimulation of the cells with TNF-α was dose-dependently inhibited by SN50 (Fig. 1a).
To further substantiate this finding, five other FLS cell lines generated from RA synovial biopsies obtained from different RA patients were stimulated with TNF-α with or without the dose of 200 μg/ml SN50. DHEA was metabolized into 7α-OH-DHEA in all five untreated FLS cell lines used (Fig. 1b). TNF-α induced a significant increase in Cyp7b activity in all FLS used. When SN50 was applied in combination with TNF-α, conversion of DHEA into 7α-OH-DHEA was significantly inhibited in four out of five FLS cell lines.
To investigate whether the effect of SN50 interfered at the level of Cyp7b activity or expression, we also analyzed the influence of SN50 on the TNF-α-induced increase in Cyp7b mRNA expression in the SCRO.14.SF cell line. A weak signal for Cyp7b mRNA was observed in untreated FLS (Fig. 1c). When stimulated with TNF-α, a marked increase in Cyp7b mRNA level was observed. Incubation of FLS with SN50 almost completely prevented the TNF-α-induced increase in Cyp7b mRNA expression (Fig. 1c).
Studies were performed to investigate whether SN50 indeed inhibits transport of NF-κB to the nucleus. In untreated FLS, NF-κB is localized in the cytoplasm (Fig. 1d). Incubation of FLS with TNF-α strongly increased the presence of NF-κB in the nucleus. This nuclear translocation of NF-κB was inhibited by SN50 (Fig. 1d).
PSI inhibited the TNF-α-induced increase in Cyp7b activity
In subsequent experiments we examined the effect of PSI, a proteasome inhibitor that is known to prevent IκB degradation and thereby activation of NF-κB, on TNF-α-induced Cyp7b activation in the FLS cell line. PSI (1 × 10E-6 mol/l) significantly decreased Cyp7b activity in nonstimulated FLS. Moreover, PSI prevented the increase in Cyp7b activity following incubation with TNF-α (Fig. 2). The combined results with SN50 and PSI imply an involvement of NF-κB in TNF-α-induced Cyp7b activity.
MAPK inhibition did not affect the TNF-α-induced increase in Cyp7b activity
We further investigated a putative role for MAPKs in the TNF-α-induced increase in Cyp7b activity by using the MEK1 inhibitor PD98059 and the p38 inhibitor SB203580.
The p38 inhibitor (SB203580) did not affect Cyp7b activity in nonstimulated cells (Fig. 3). Also, following TNF-α stimulation no effect of SB203580 on the increase in Cyp7b activity was observed. Similarly, incubation of nonstimulated FLS with the MEK1/ERK1/2 inhibitor (PD98059) did not affect Cyp7b activity. Only at a high concentration (1 × 10E-5 mol/l) did application of PD98059 result in a small but statistically significant inhibition of TNF-α-induced increase in Cyp7b activity. The combination of SB203580 and PD98059 at high concentrations, similar to PD98059 alone, also exhibited a small but significant decrease in TNF-α-induced Cyp7b activity (Fig. 3). Similar findings were obtained using five additional RA FLS cell lines; a small inhibitory effect of the p38 inhibitor SB203580 at high concentration (1 × 10-5 mol/l) was observed in one cell line out of five after stimulation with TNF-α. In none of the five cell lines did we observe any effect on the TNF-α-induced increase in Cyp7b activity using 1 × 10-5 mol/l PD98059 (data not shown). From these results it is concluded that p38 and ERK1/2 do not appear to play a role in regulating Cyp7b activity.
Regulation of Cyp7b mRNA expression and activity in fibroblast-like synoviocytes
Previous studies implicated a role for TNF receptor I in regulating Cyp7b activity [10]. Because the TNF receptor I couples to AP-1 via the JNK pathway, we investigated the effect of the recently described JNK inhibitor SP600125 [17]. In addition, we analyzed the effect of NF-κB and MAPK inhibitors on TNF-α-induced Cyp7b mRNA expression. A weak Cyp7b mRNA signal was found in untreated FLS (Fig. 4a). Treatment of FLS with TNF-α resulted in an increase in Cyp7b mRNA expression. Moreover, SN50 prevented the increase in Cyp7b mRNA expression following incubation with TNF-α. Furthermore, the proteasome inhibitor PSI, which is known to prevent IκB degradation, blocked the TNF-α-induced Cyp7b mRNA expression. In addition, the JNK inhibitor SP600125 prevented the TNF-α-induced Cyp7b mRNA expression, which further substantiates a role for AP-1 in TNF-α-induced Cyp7b expression. Use of the MAPK inhibitors PD98059 and SB203580 did not result in convincing changes in TNF-α-induced Cyp7b mRNA expression.
We then determined Cyp7b enzymatic activity in FLS through the detection of 7α-OH-DHEA. Presence of TNF-α in the cultures resulted in increased Cyp7b activity compared with baseline (Fig. 4b). We subsequently analyzed the effect on TNF-α stimulation of the presence or absence of PSI, SN50, SP600125, PD98059 and or SB203580. TNF-α in combination with PSI, SN50, or SP600125 significantly decreased the Cyp7b activity to basal 7α-OH-DHEA levels (Fig. 4b). In contrast, addition of PD98059 or SB203580 did not significantly affect the TNF-α-induced increase in Cyp7b activity. The absence of an effect of the MAPK inhibitors PD98059 and SB203580 on TNF-α-induced Cyp7b activity is in accordance with our findings at the level of Cyp7b mRNA expression.
Presence of NF-κB and AP-1 binding sites within the Cyp7b promoter
Analysis of the proximal region of the Cyp7b promoter revealed nucleotide sequences that correspond to putative binding sites for NF-κB, AP-1, nuclear factor of activated T cells (NFAT), and signal transducer and activator of transcription (STAT)1 (Fig. 5). The presence of putative bindings sites for NF-κB and AP-1 within the Cyp7b promoter are in accordance with the findings in this report that NF-κB and AP-1 are involved in the TNF-α-enhanced Cyp7b activity.
Discussion
The findings of the present study suggest involvement of AP-1 and NF-κB, but not of p38 or ERK1/2, in the TNF-α-enhanced formation of the immunostimulating 7α-OH-DHEA.
We and others [23]. showed that, upon stimulation of cells with TNF-α, NF-κB translocates from the cytoplasm to the nucleus. As expected, translocation of NF-κB to the nucleus was inhibited by SN50. In addition, SN50 blocks the TNF-α-induced increases in Cyp7b activity and Cyp7b mRNA level, which suggests transcriptional involvement of NF-κB and/or other transcription factors such as AP-1 in TNF-α-induced Cyp7b activation. Initial reports suggested that SN50 is a specific inhibitor of NF-κB activation. However, Torgerson and coworkers [23] reported that SN50 blocks the nuclear translocation of the transcription factors AP-1, NFAT and STAT1 in Jurkat T cells stimulated with IFN-γ or phorbol myristate acetate (PMA) as well.
To determine whether STAT1 could be involved in TNF-α-induced Cyp7b activity, we analyzed the proximal region of the Cyp7b promoter for putative binding sites of STAT1, which revealed such sites in this region. It should be appreciated, however, that STAT1 is mainly activated by IFN-γ. Also, Cyp7b is not regulated by IFN-γ, as described previously [10]. Therefore, it is unlikely that STAT1 is involved in Cyp7b activity regulation.
There is evidence that the dose of SN50 determines the specificity of the inhibitor [16]. Therefore, it is likely that the doses of SN50 we used (100–200 μg/ml) can block both translocation of NF-κB and translocation of AP-1 to the nucleus [24]. Indeed, we observed inhibition of TNF-α-induced NF-κB nuclear translocation concomitantly with an inhibition of TNF-α-induced Cyp7b activity by SN50. In order to investigate a role for AP-1, we used the JNK inhibitor SP600125 [17]. The results demonstrate an involvement of the AP-1 complex in the TNF-α-induced Cyp7b expression and activity in FLS from RA patients. An involvement of NF-κB and AP-1 in the TNF-α-induced Cyp7b activity is in accordance with the presence of putative NF-κB and AP-1 binding sites within the Cyp7b promoter.
Our findings are consistent with data reported by Wu and coworkers [25] with respect to the presence of putative binding sites for NF-κB within the Cyp7b promoter. In contrast to our analysis, those authors [25] did not identify putative AP-1 binding sites, which could be due to the use of the default setting for the matrix score in MartView. However, other approaches are needed to substantiate further the role played by NF-κB and AP-1 in the TNF-α-induced increase in Cyp7B expression. This may be done by analysis of the Cyp7b promoter in a promoter reporter construct, with mutation of the putative NF-κB and AP-1 response elements. Moreover, the use of the siRNA technology could contribute to our understanding of the importance of NF-κB in the TNF-α-induced DHEA metabolism in human FLS.
Because the anti-glucocorticoid 7α-OH-DHEA, which is produced by the activity of the enzyme Cyp7b, might have stimulatory effects on the inflammatory process, studies with administration of 7α-OH-DHEA in animal models with susceptibility for arthritis are needed to elucidate the mechanism by which 7α-OH-DHEA influences the development of inflammatory processes. In this respect, it would be of interest to investigate whether inflammation is reduced in Cyp7b knockout mice, which do not express 7α-OH-DHEA. In addition, intra-articular delivery of 7α-OH-DHEA and/or Cyp7b expression systems should add to our understanding of the role played by Cyp7b in the arthritic process.
The inhibitory effect of PSI on the TNF-α-induced upregulation of Cyp7b activity is also in accordance with a role for NF-κB in regulating Cyp7b activity. Although it has not been described in the original studies of the action of PSI [18], we cannot exclude the possibility that inhibition of proteasome activity by PSI may interfere in other signal transduction pathways that are independent of NF-κB [26].
In this paper we show that inhibitors of the ERK1/2 and p38 signalling pathways did not convincingly affect Cyp7b mRNA expression and enzymatic activity in RA FLS following stimulation with TNF-α. Barchowsky and coworkers [27] also reported that there is no role for MAPKs after TNF-α stimulation of collagenase I expression in rabbit synovial fibroblasts. However, previous studies have reported activation of ERK1/2 and p38 in several cell lines, including synovial fibroblasts, after incubation with TNF-α [28]. We observed that, in contrast to TNF-α-induced Cyp7b activity, the MEK1/ERK1/2 inhibitor PD98059 and p38 inhibitor SB203580 reduced the TNF-α-induced IL-6 production in several RA FLS tested (data not shown). These results indicate that the inhibitors were active and can inhibit other effects of TNF-α, but they do not play a role in regulation of Cyp7b activity by TNF-α. Furthermore, it cannot be excluded that other MAPK isoforms such as ERK5, ERK7, p38γ and p38δ are regulated by TNF-α as well in the RA FLS used [29].
Conclusion
Our data suggest that there is a role for both NF-κB and AP-1 in regulating the expression and activity of Cyp7b (Fig. 6), which strengthens the rationale for specific inhibition of these pathways in arthritis.
Abbreviations
AP-1 = activator protein-1; Cyp7b = cytochrome p450 enzyme 7b; DHEA = dehydroepiandrosterone; DMEM = Dulbecco's modified Eagle's medium; ERK = extracellular signal-regulated kinase; FCS = foetal calf serum; FLS = fibroblast-like synoviocytes; IFN = interferon; IL = interleukin; JNK = c-Jun N-terminal kinase; MAPK = mitogen-activated protein kinase; MEK = mitogen-activated protein kinase kinase; NFAT = nuclear factor of activated T cells ; NF-κB = nuclear factor-κB; 7α-OH-DHEA = 7α-hydroxy-dehydroepiandrosterone; PBS = phosphate-buffered saline; PCR = polymerase chain reaction; PMA = phorbol myristate acetate; RA = rheumatoid arthritis; STAT = signal transducer and activator of transcription; TNF = tumor necrosis factor.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
JD was principle investigator, and designed most of the studies, carried out most of the assays and wrote the manuscript. AK (Allard Kaptein) helped in conceiving the study and helped to draft the manuscript. AK (Annemieke Kavelaars) and CH were involved in drafting and revising the article. AB helped in conceiving the study, helped to draft the manuscript and was the senior scientist responsible for the work. All authors read and approved the final manuscript.
Acknowledgements
We thank Dr E Bos for critical reading of the manuscript and M Toker and N Bisseling for photographic reproductions. C Meeuwisse is acknowledged for performing computer analysis of the Cyp7b promoter. Dr R Hampl is acknowledged for the performance of the radioimmunoassay analysis.
Figures and Tables
Figure 1 Cyp7b activity and mRNA expression is inhibited by SN50 in fibroblast-like synoviocytes. (a) Human fibroblast-like synoviocytes (FLS; SCRO.14.SF, passages 10–12) were plated at 1 × 105 cells/well in a 24-well plate and preincubated in the presence or absence (-) of the SN50 inhibitor for 2 hours. Thereafter, the cells were incubated with (solid bars) or without (open bars) tumour necrosis factor (TNF)-α for another 24 hours with 1.5 × 10-8 mol/l 3H-dehydroepiandrosterone (DHEA). The formation of [3H]-7α-hydroxy-dehydroepiandrosterone (7α-OH-DHEA) from [3H]-DHEA, representing cytochrome p450 enzyme 7b (Cyp7b) activity, was determined by high-performance liquid chromatography. The amount of 7α-OH-DHEA is expressed as the percentage [3H]-7α-OH-DHEA of the total amount of [3H]-label measured. Results are expressed as the mean ± standard error of the mean of triplicate samples. The data are representative of two independent experiments. *P < 0.005 (Student's t-test). (b) Human FLS (1 × 105 cells/well) were isolated from five different rheumatoid arthritis patient biopsies. Cells (1 × 105/well) were incubated in the presence and absence of TNF-α and in the presence of SN50 for 2 hours, as described in Materials and methods. Results are representative for one of the two independent experiments. SCRO.12.SF passage 2, SCRO.11.SF passage 3, SCRO.03.SF passage 8, SCRO.01.SF passage 6 and SCRO.08.SF passage 4 were used. *P < 0.005 versus TNF-α (Student's t-test). (c) FLS (SCRO.14.SF; passages 10–12) were incubated for 6 hours with 0.5 ng/ml TNF-α, SN50 200 μg/ml plus 0.5 ng/ml TNF-α, or incubated with medium control (-). Reverse transcription PCR was done using GAPDH and Cyp7b specific primers (35 cycles). The data are representative of two independent experiments. (d) FLS fibroblasts (SCRO.14.SF; passages 10–12) were grown on chamber slides. Cells were incubated for 2 hours in the presence or absence of 200 μg/ml SN50 before incubation for 30 min in the presence or absence of TNF-α (0.5 ng/ml). Immunoperoxidase staining was carried out with an antibody against nuclear factor-κB (NF-κB)p65 conjugated to peroxidase. Data are representative for three independent experiments.
Figure 2 PSI inhibits the TNF-α-induced increase in 7α-OH-DHEA. Human rheumatoid arthritis (RA) fibroblast-like synoviocytes (FLS; SCRO.14.SF, 1 × 105 cells/well; passages 10–12) were preincubated in the presence or absence (-) of the PSI inhibitor for 1 hour. Thereafter, the cells were incubated with (solid bars) or without (open bars) tumour necrosis factor (TNF)-α for another 24 hours with 1.5 × 10-8 mol/l 3H-DHEA, as described in Materials and methods. Data are expressed as mean ± standard error of the mean and are representative of four independent experiments. *P < 0.0005. 7α-OH-DHEA = 7α-hydroxy-dehydroepiandrosterone.
Figure 3 The effect of the MAPK inhibitors PD98059 or SB203580 on TNF-α-induced Cyp7b activity. (a) Fibroblast-like synoviocytes (FLS; SCRO.14.SF, passages 8–12) were incubated for 1 hour in the presence or absence (-) of the mitogen-activated protein kinase (MAPK) kinase (MEK)1 inhibitor PD98059 (PD) or the p38 inhibitor SB203580 (SB). Thereafter, cells were incubated in the presence or absence of 0.5 ng/ml tumour necrosis factor (TNF)-α plus 1.5 × 10-8 mol/l [3H]-dehydroepiandrosterone (DHEA) for 24 hours and processed using high-performance liquid chromatography. The amount of 7α-hydroxy-dehydroepiandrosterone (7α-OH-DHEA) is expressed as the percentage [3H]-7α-OH-DHEA of the total amount of [3H]-label measured. Results are expressed as the mean ± standard error of the mean of triplicate sample. Data are representative of three independent experiments. *P < 0.05 versus TNF-α (Student's t-test). (b) The data from panel a (three independent experiments) are combined for the highest inhibitor concentrations. PD98059 and SB23580 were dissolved in methanol (MeOH) and dimethylsulfoxide (DMSO), respectively, and used as controls. *P < 0.05 (Student's t-test). Cyp7b = cytochrome p450 enzyme 7b.
Figure 4 Regulation of Cyp7b mRNA expression and activity in FLS. (a) Human fibroblast-like synoviocytes (FLS; STSF.388, passages 8–10) were incubated for 6 hours with medium control (-), 0.5 ng/ml tumour necrosis factor (TNF)-α, TNF-α combined with the proteasome inhibitor (PSI) 1 × 10E-6 mol/l, SN50 200 μg/ml, SP600125 (SP) 1 × 10E-5 mol/l, PD98059 (PD) 1 × 10E-5 mol/l or SB23580 (SB) 1 × 10E-5 mol/l, as described in Materials and methods. RNA was isolated and cDNA was made and used for reverse transcription with GAPDH and cytochrome p450 enzyme 7b (Cyp7b) specific primers. The ratio of Cyp7b to GAPDH mRNA expression was 1.4 (TNF-α alone), 8 × 10-6 (PSI + TNF-α), 1.3 × 10-6 (SN50 + TNF-α), 0.5 × 10-6 (SP + TNF-α), 0.3 (PD + TNF-α) and 0.2 (SB + TNF-α). Data are representative of two independent experiments. (b) FLS (STSF.388, passages 8–10) were preincubated in the presence or absence (medium control [-]) of SN50 for 2 hours or SP600125 (SP), PD98059 (PD) or SB23580 (SB) for 1 hour. Thereafter, FLS were incubated in the presence or absence of 0.5 ng/ml TNF-α plus 1.5 × 10-8 mol/l dehydroepiandrosterone (DHEA) for 24 hours and processed for radioimmunoassay detection of 7α-hydroxy-dehydroepiandrosterone (7α-OH-DHEA). Results are expressed as the mean ± standard error of the mean of triplicate samples. Data are representative for two independent experiments. *P < 0.005 versus TNF-α (Student's t-test).
Figure 5 NF-κB and AP-1 binding sites within the Cyp7b promoter. Putative binding sites for selected transcription factor family matrices were identified using the MartView export function. Sequences for putative binding sites are underlined. *, transcription start side; -, presence of the transcription binding site on the minus DNA strand; AP-1, activator protein-1; Cyp7b, cytochrome p450 enzyme 7b; NFAT, nuclear factor of activated T cells; NF-κB, nuclear factor-κB; STAT, signal transducer and activator of transcription.
Figure 6 Simplified diagram of the proposed signalling events leading to Cyp7b gene transcription in synovial fibroblasts. Using inhibitors of the mitogen-activated protein kinase (MAPK) kinase (MEK)1/extracellular signal (mitogenic)-regulated kinase (ERK)1/2 pathway (i.e. PD98059), the p38 MAPK pathway (i.e. SB203580), the c-Jun-NH2-terminal kinase (JNK) pathway (i.e. SP600125), the IκB/nuclear factor-κB (NF-κB) pathway (i.e. PSI; dashed line) and the NF-κB/activator protein (AP)-1 pathway SN50, it was established that the NF-κB and AP-1 pathway is relevant to Cyp7b activity. All experiments were performed using synovial fibroblasts derived from patients with rheumatoid arthritis. Cyp7b, cytochrome p450 enzyme 7b; IκBα, inhibitor of NF-κB; IKK, IκB kinase complex (composed of three subunits – IKKα, IKKβ, and IKKγ [NEMO]; RelA (p65) and NF-κB1 [p50/p105] are subunits of NF-κB); RIP, receptor interacting protein; TNF, tumour necrosis factor; TNFR, TNF-α receptor; TRADD, TNF receptor associated death domain; TRAF, TNF receptor associated factor.
==== Refs
Zvaifler NJ Firestein GS Pannus and pannocytes. Alternative models of joint destruction in rheumatoid arthritis Arthritis Rheum 1994 37 783 789 8003048
Straub RH Cutolo M Involvement of the hypothalamic-pituitary-adrenal/gonadal axis and the peripheral nervous system in rheumatoid arthritis: viewpoint based on a systemic pathogenetic role Arthritis Rheum 2001 44 493 507 11263762 10.1002/1529-0131(200103)44:3<493::AID-ANR95>3.0.CO;2-U
Bradlow HL Murphy J Byrne JJ Immunological properties of dehydroepiandrosterone, its conjugates, and metabolites Ann N Y Acad Sci 1999 876 91 101 10415598
Daynes RA Dudley DJ Araneo BA Regulation of murine lymphokine production in vivo. II. Dehydroepiandrosterone is a natural enhancer of interleukin 2 synthesis by helper T cells Eur J Immunol 1990 20 793 802 2140789
Morfin R Courchay G Pregnenolone and dehydroepiandrosterone as precursors of native 7-hydroxylated metabolites which increase the immune response in mice J Steroid Biochem Mol Biol 1994 50 91 100 8049138 10.1016/0960-0760(94)90176-7
Morfin R Involvement of steroids and cytochromes P(450) species in the triggering of immune defenses J Steroid Biochem Mol Biol 2002 80 273 290 11948012 10.1016/S0960-0760(02)00025-0
Lathe R Steroid and sterol 7-hydroxylation: ancient pathways Steroids 2002 67 967 977 12398993 10.1016/S0039-128X(02)00044-2
Rose KA Stapleton G Dott K Kieny MP Best R Schwarz M Russell DW Bjorkhem I Seckl J Lathe R Cyp7b, a novel brain cytochrome P450, catalyzes the synthesis of neurosteroids 7alpha-hydroxy dehydroepiandrosterone and 7alpha-hydroxy pregnenolone Proc Natl Acad Sci U S A 1997 94 4925 4930 9144166 10.1073/pnas.94.10.4925
Dulos J Verbraak E Bagchus WM Boots AM Kaptein A Severity of murine collagen-induced arthritis correlates with increased CYP7B activity: enhancement of dehydroepiandrosterone metabolism by interleukin-1β Arthritis Rheum 2004 50 3346 3353 15476247 10.1002/art.20509
Dulos J van der Vleuten MAJ Kavelaars A Heijnen CJ Boots AM CYP7B expression and activity in fibroblast-like synoviocytes from patients with rheumatoid arthritis: regulation by pro-inflammatory cytokines Arthritis Rheum 2005 52 770 778 15751070 10.1002/art.20950
Elliot MJ Maini RN Feldmann M Long-Fox A Charles P Katsikis P Brennan FM Walker J Bijl H Ghraveb J Treatment of rheumatoid arthritis with chimeric monoclonal antibodies to tumor necrosis factor alpha Arthritis Rheum 1993 36 1681 1690 8250987
Firestein GS Manning AM Signal transduction and transcription factors in rheumatic disease Arthritis Rheum 1999 42 609 621 10211874 10.1002/1529-0131(199904)42:4<609::AID-ANR3>3.0.CO;2-I
Chang L Karin M Mammalian MAP kinase signalling cascades Nature 2001 410 37 40 11242034 10.1038/35065000
Palanki MS Inhibitors of AP-1 and NF-kappa B mediated transcriptional activation: therapeutic potential in autoimmune diseases and structural diversity Curr Med Chem 2002 9 219 227 11860356
Li Q Verma IM NF-kappaB regulation in the immune system Nat Rev Immunol 2002 2 725 734 12360211 10.1038/nri910
Das J Chen CH Yang L Cohn L Ray P Ray A A critical role for NF-kappa B in GATA3 expression and TH2 differentiation in allergic airway inflammation Nat Immunol 2001 2 45 50 11135577 10.1038/83158
Hammaker DR Boyle DL Chabaud-Riou M Firestein GS Regulation of c-Jun N-terminal kinase by MEKK-2 and mitogen-activated protein kinase kinase kinases in rheumatoid arthritis J Immunol 2004 172 1612 1618 14734742
Haas M Page S Page M Neumann FJ Marx N Adam M Zieglar-Heitbrock HW Neumeier D Brand K Effect of proteasome inhibitors on monocytic IkappaB-alpha and -beta depletion, NF-kappaB activation, and cytokine production J Leukoc Biol 1998 63 395 404 9500529
Saraux A Berthelot JM Chales G Le Henaff C Thorel JB Hoang S Valls I Devauchelle V Martin A Baron D Ability of the American College of Rheumatology 1987 criteria to predict rheumatoid arthritis in patients with early arthritis and classification of these patients two years later Arthritis Rheum 2001 44 2485 2491 11710704 10.1002/1529-0131(200111)44:11<2485::AID-ART428>3.0.CO;2-S
Hamann J Wishaupt JO van Lier RA Smeets TJ Breedveld FC Tak PP Expression of the activation antigen CD97 and its ligand CD55 in rheumatoid synovial tissue Arthritis Rheum 1999 42 650 658 10211878 10.1002/1529-0131(199904)42:4<650::AID-ANR7>3.0.CO;2-S
Lapcik O Hampl R Hill M Starka L Immunoassay of 7-hydroxysteroids: 2. Radioimmunoassay of 7alpha-hydroxy-dehydroepiandrosterone J Steroid Biochem Mol Biol 1999 71 231 237 10704912 10.1016/S0960-0760(99)00145-4
Quandt K Frech K Karas H Wingender E Werner T MatInd and MatInspector: new fast and versatile tools for detection of consensus matches in nucleotide sequence data Nucleic Acids Res 1995 23 4878 4884 8532532
Torgerson TR Colosia AD Donahue JP Lin YZ Hawiger J Regulation of NF-kappa B, AP-1, NFAT, and STAT1 nuclear import in T lymphocytes by noninvasive delivery of peptide carrying the nuclear localization sequence of NF-kappa B p50 J Immunol 1998 161 6084 6092 9834092
Boothby M Specificity of sn50 for NF-kappa B? Nat Immunol 2001 2 471 472 11376325 10.1038/88652
Wu Z Martin KO Javitt NB Chiang JY Structure and functions of human oxysterol 7alpha-hydroxylase cDNAs and gene CYP7B1 J Lipid Res 1999 40 2195 2203 10588945
Yoshimura S Bondeson J Brennan FM Foxwell BM Feldmann M Role of NFkappaB in antigen presentation and development of regulatory T cells elucidated by treatment of dendritic cells with the proteasome inhibitor PSI Eur J Immunol 2001 31 1883 1893 11433385 10.1002/1521-4141(200106)31:6<1883::AID-IMMU1883>3.0.CO;2-V
Barchowsky A Frleta D Vincenti MP Okamoto T Integration of the NF-kappaB and mitogen-activated protein kinase/AP-1 pathways at the collagenase-1 promoter: divergence of IL-1 and TNF-dependent signal transduction in rabbit primary synovial fibroblasts Cytokine 2000 12 1469 1479 11023661 10.1006/cyto.2000.0743
Suzuki M Tetsuka T Yoshida S Watanabe N Kobayashi M Matsui N Okamoto T The role of p38 mitogen-activated protein kinase in IL-6 and IL-8 production from the TNF-alpha- or IL-1beta-stimulated rheumatoid synovial fibroblasts FEBS Lett 2000 465 23 27 10620700 10.1016/S0014-5793(99)01717-2
Johnson GL Lapadat R Mitogen-activated protein kinase pathways mediated by ERK, JNK, and p38 protein kinases Science 2002 298 1911 1912 12471242 10.1126/science.1072682
|
16277680
|
PMC1297571
|
CC BY
|
2021-01-04 16:02:40
|
no
|
Arthritis Res Ther. 2005 Sep 15; 7(6):R1271-R1280
|
utf-8
|
Arthritis Res Ther
| 2,005 |
10.1186/ar1819
|
oa_comm
|
==== Front
Arthritis Res TherArthritis Research & Therapy1478-63541478-6362BioMed Central London ar18211627767810.1186/ar1821Research ArticleInflammation causes tissue-specific depletion of vitamin B6 Chiang En-Pei [email protected] Donald E [email protected] Jacob [email protected] Gerard [email protected] Yi-Cheng 1Roubenoff Ronenn [email protected] Department of Food Science and Biotechnology, National Chung Hsing University, Taichung, Taiwan2 Comparative Biology Unit, Jean Mayer USDA Human Nutrition Research Center on Aging, Tufts University, Boston, MA, USA3 Vitamin Metabolism and Aging Laboratory (JS), New England Medical Center, Boston, MA, USA4 Biostatistics Unit (GD), New England Medical Center, Boston, MA, USA5 Nutrition, Exercise Physiology, and Sarcopenia Laboratory, New England Medical Center, Boston, MA, USA2005 13 9 2005 7 6 R1254 R1262 4 5 2005 4 7 2005 2 8 2005 15 8 2005 Copyright © 2005 Chiang 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.
Previously we observed strong and consistent associations between vitamin B6 status and several indicators of inflammation in patients with rheumatoid arthritis. Clinical indicators, including the disability score, the length of morning stiffness, and the degree of pain, and biochemical markers, including the erythrocyte sedimentation rate and C-reactive protein levels, were found to be inversely correlated with circulating vitamin B6 levels. Such strong associations imply that impaired vitamin B6 status in these patients results from inflammation. In the present study we examined whether inflammation directly alters vitamin B6 tissue contents and its excretion in vivo. A cross-sectional case-controlled human clinical trial was performed in parallel with experiments in an animal model of inflammation. Plasma and erythrocyte and pyridoxal 5'-phosphate concentrations, urinary 4-pyridoxic acid excretion, and the activity coefficient of erythrocyte aspartate aminotransferase were compared between patients and healthy subjects. Adjuvant arthritis was induced in rats for investigating hepatic and muscle contents as well as the urinary excretion of vitamin B6 during acute and chronic inflammation. Patients with rheumatoid arthritis had low plasma pyridoxal 5'-phosphate compared with healthy control subjects, but normal erythrocyte pyridoxal 5'-phosphate and urinary 4-pyridoxic acid excretion. Adjuvant arthritis in rats did not affect 4-pyridoxic acid excretion or muscle storage of pyridoxal 5'-phosphate, but it resulted in significantly lower pyridoxal 5'-phosphate levels in circulation and in liver during inflammation. Inflammation induced a tissue-specific depletion of vitamin B6. The low plasma pyridoxal 5'-phosphate levels seen in inflammation are unlikely to be due to insufficient intake or excessive vitamin B6 excretion. Possible causes of decreased levels of vitamin B6 are discussed.
==== Body
Introduction
Vitamin B6 deficiency results in adverse health consequences, including hyperhomocysteinemia [1] and possibly arteriosclerotic lesions [2]. We have reported that the degree of disease activity is associated with vitamin B6 indices in patients with rheumatoid arthritis [3,4]. Bates and colleagues reported suboptimal vitamin B6 status in inflammatory conditions and in the acute-phase response in the elderly population [5]. These observations have attracted attention partly because vitamin B6 deficiency and several markers of inflammation were both found to be independent risk factors for thrombosis [6,7]. Although several clinical trials and epidemiological studies have demonstrated the associations between vitamin B6 and inflammatory diseases, the association between vitamin B6 status and inflammatory markers has been contentious, and the cause-effect relationship between these two has not been elucidated.
Pyridoxine deficiency increased the degree of paw edema by 54% in a rat model of inflammation; it was therefore suggested that pyridoxine deficiency might enhance inflammation [8]. However, in healthy middle-aged adults, B vitamin status does not seem to be a strong correlate of circulating levels of inflammatory markers [9]. In contrast, a low level of circulating vitamin B6 was found to be associated with elevation of the inflammatory marker C-reactive protein independently of plasma homocysteine levels in the Framingham Heart Study cohort [10]. A recent study indicated that low plasma concentrations of pyridoxal 5'-phosphate are inversely related to major markers of inflammation and independently associated with increased coronary artery disease in the Italian population [11]. Decreased vitamin B6 status was also reported in patients after surgery and trauma [12]. In conditions with inflammation such as inflammatory bowel disease, low plasma levels of vitamin B6 are commonly found, especially in patients with active disease [13]. In a recent study we observed strong and consistent associations between vitamin B6 status and several indicators of inflammation in patients with rheumatoid arthritis [4]. Plasma pyridoxal 5'-phosphate was correlated with disease-related disability, morning stiffness, and degree of pain, C-reactive protein, serum albumin, and erythrocyte sedimentation rate [4].
The objective of the present study was to determine whether inflammation directly decreases the primary pools of vitamin B6 metabolism and storage, and to examine whether inflammation alters the excretion of vitamin B6 in vivo.
Materials and methods
Clinical trial
Study subjects
Thirty-three adults (aged at least 18 years) with rheumatoid arthritis were recruited from the Tufts University New England Medical Center Arthritis Center, Boston, as described previously [14]. Seventeen healthy control subjects, who did not differ in their age range or gender distribution from the subjects with rheumatoid arthritis, were recruited through advertisements in the greater Boston area. Study subjects were 18 to 80 years old. For the rheumatoid arthritis group, subjects had to fulfill the American College of Rheumatology criteria for rheumatoid arthritis [15]. The criteria for the classification of acute arthritis of rheumatoid include the following: (1) morning stiffness in and around the joints, lasting at least 1 hour before maximal improvement; (2) at least three joint areas simultaneously have had soft tissue swelling or fluid (not bony overgrowth alone) observed by a physician; (3) at least one area swollen of hand joints in a wrist, metacarpophalangeal, or proximal interphalangeal joint; (4) simultaneous involvement of the same joint areas (as defined in (2)) on both sides of the body; (5) rheumatoid nodules observed by a physician; (6) abnormal amounts of serum rheumatoid factor; (7) radiographic changes typical of rheumatoid arthritis on posteroanterior hand and wrist radiographs, which must include erosions or unequivocal bony decalcification localized in or most marked adjacent to the involved joints.
Written informed consent was obtained from all subjects in accordance with the regulations of the New England Medical Center/Tufts University Human Investigation Review Committee. Subjects with pregnancy, anemia (hemoglobin 10 mg/dL or lower), thrombocytopenia (platelet count 50,000/ μL or lower), abnormal serum hepatic transaminase (serum aspartate aminotransferase or alanine aminotransferase at least 50 IU/L), renal insufficiency (serum creatinine at least 1.5 mg/dL), diabetes, cancer or use of oral contraceptive were excluded. Thirteen of the 33 patients (39%) and 7 of the 17 controls (41%) were taking vitamin B6 or multivitamin supplements before enrollment. These subjects were asked to stop doing so for at least 2 months before their participation in the study.
Experimental protocol
This cross-sectional study was conducted in the New England Medical Center General Clinical Research Center. Before enrollment, each subject was examined by the study physician, and was screened by blood and urine analyses to ensure eligibility. Each subject was instructed to perform a 24-hour urine collection for measurement of 4-pyridoxic acid excretion. Patients taking methotrexate were asked to come in at least 24 hours after their weekly dose of the medicine to minimize any potential acute effect on laboratory outcomes. Urine specimens were kept at 4°C with no additive during the collection period. After completion of the 24-hour urine collection, subjects were asked to fast overnight for 12 hours. During the following morning, fasting blood was drawn into a tube containing EDTA for the determination of plasma pyridoxal 5'-phosphate, erythrocyte pyridoxal 5'-phosphate, erythrocyte aspartate aminotransferase activity coefficient, folate, and vitamin B12. Blood was also collected for hematology and chemistry analyses. Each patient's blood specimens were kept on ice and were centrifuged within 15 min of the blood draw.
Animal model of inflammation
Animal
Thirty-six female 3-month-old Lewis rats were obtained from the National Institutes of Health. Animals were fed with the AIN93M diet during a 1-week washout period and during the experimental period. All animals were kept in individual mesh cages and acclimated to a 12-hour day/night cycle. The study protocol was approved by the Animal Care and Use Committee of National Chung Hsing University and of the Human Nutrition Research Center on Aging at Tufts University.
Induction of arthritis
After the washout period, animals of the same age and gender were sorted by body weight and assigned sequentially to the adjuvant arthritis or control groups (Table 1). Adjuvant arthritis was induced at baseline by injecting a single dose of Mycobacterium butyricum in mineral oil (complete Freund's adjuvant; 200 μL per rat) [16] into the base of the tail at the baseline time-point. The age-matched, paired-fed control animals received a saline injection.
Pair-feeding protocol
After the induction of adjuvant arthritis, for each rat in the adjuvant arthritis group one control rat, matched for age and weight, received the same amount of food on the next day. This protocol minimized variations in body weight and in vitamin B6 consumption due to different dietary intake.
Sample collection
Fasting blood samples were collected from the orbital sinus vein of each rat under anesthesia at baseline. Plasma was separated immediately by centrifugation and stored at -80°C until analysis. Sixteen animals (eight control and eight arthritic rats) were killed 21 days after the adjuvant injection, reflecting the condition of acute inflammation. The rest were killed on day 42, reflecting the condition of chronic inflammation. Animals were killed by thoracotomy and exsanguinations after anesthesia. Blood, liver, and skeletal muscle were collected and stored at -80°C until analysis for pyridoxal 5'-phosphate concentration. On selected days (days reflecting baseline, peak inflammation, and chronic inflammation), each animal was kept in an individual metabolic cage, specifically designed for the 24-hour collection of urine. The excretion of urinary 4-pyridoxic acid and creatinine was subsequently measured.
Laboratory analyses
Blood hematology and chemistry analyses and urinalysis for human subjects were performed at the New England Medical Center Clinical Laboratory, Boston. For measurements of B vitamins, fasting blood was drawn from each human subject, and plasma was separated and stored at -80°C until analysis. Erythrocytes were washed three times with 0.9% saline and then an aliquot of packed erythrocytes was frozen before the measurement of erythrocyte aspartate aminotransferase [17]. The activity coefficient of erythrocyte aspartate aminotransferase was calculated by dividing pyridoxal 5'-phosphate-stimulated enzyme activity by the unstimulated activity. For pyridoxal 5'-phosphate analysis, the freshly washed erythrocytes were extracted with an equal volume of 10% (w/v) perchloroacetic acid. After centrifugation, the supernatants were stored at -70°C until analysis. Erythrocyte and plasma pyridoxal 5'-phosphate concentrations were assayed by a modification of the tyrosine decarboxylase enzymatic procedure of Camp and colleagues [18], in which a 20 μL aliquot of sample was precipitated with 80 μL of 5% (w/v) trichloroacetic acid for deproteinization. The erythrocyte pyridoxal 5'-phosphate results were expressed as nmol/L of packed erythrocytes at a hematocrit of 100%. The coefficient of variation (percentage of the mean) for the pyridoxal 5'-phosphate assay was 7.6% within assays and 5.7% between assays. Plasma folate, red blood cell folate, and plasma vitamin B12 were measured with Quantaphase II B12/Folate Radioassays (Bio-Rad; Hercules, CA). The weight of each 24-hour urine specimen was measured; aliquots were taken and stored at -20°C until analysis. No preservatives were added to the 4-pyridoxic acid or creatinine aliquots. 4-Pyridoxic acid was measured by high-performance liquid chromatography after urine had been mixed with an equal volume of 5% trichloroacetic acid for deproteinization [19]. The HPLC consisted of a Hitachi L-7100 intelligent pump connected to an L-2480 fluorometric detector. The ranges for plasma pyridoxal 5'-phosphate and urinary pyridoxic acid were in good agreement with published results. The range of erythrocyte pyridoxal 5'-phosphate concentrations in our subjects was similar to that in a previous study by Heiskanen and colleagues [20].
Statistical analysis
Data were plotted so that they could be examined for normality before statistical analyses. Data for plasma pyridoxal 5'-phosphate and red blood cell folate levels in humans were log-transformed to achieve normal distributions, and the geometric means with the antilogarithm of the 95% confidence intervals are presented as results. Student's t-tests for independent samples were performed to determine whether there was a difference between patients with rheumatoid arthritis and controls in hematology measures, blood chemistry analyses, and vitamin profiles. For animal experiments, pooled Student's t-tests were performed to determine whether there was a difference between arthritic rats and matched control animals. All statistical analyses were performed with Systat 9.0 for Windows™ (SPSS, Chicago, IL).
Results
Demographic information
Thirty-six patients with rheumatoid arthritis were recruited from a previous study [14]. In the present study, a further 17 healthy subjects were recruited as control subjects. Three subjects from the patient group dropped out because of inconvenience of the 24-hour urine collection. Clinical and demographic characteristics of the study subjects are shown in Table 2. There were no differences in age, height, or weight between patients and controls, indicating that the control subjects and the patients did not differ in their physical conditions except for the presence of rheumatoid arthritis. The average disease duration in patients with rheumatoid arthritis was 10.8 ± 6.7 (mean ± SD) years. While participating in the study, 21 patients were taking non-steroidal anti-inflammatory drugs, 18 were taking prednisone, 16 were taking methotrexate, and 5 were taking gold. All patients had been taking the same medications for the 2 months preceding their entry into the study.
Patients and controls did not differ in the levels of serum albumin, creatinine, aspartate aminotransferase, and alanine aminotransferase. Patients had higher serum alkaline phosphatase levels, white blood cell counts and erythrocyte sedimentation rates than controls; patients also had a trend toward a reduced hematocrit, a smaller number of red blood cells and a lower hemoglobin level than the healthy control subjects (Table 2). Serum alkaline phosphatase concentration was inversely correlated with plasma pyridoxal 5'-phosphate concentration in all subjects (Pearson's correlation, r = -0.35, P = 0.012). In addition, serum albumin was modestly correlated with plasma pyridoxal 5'-phosphate in the patients (Pearson's correlation, r = 0.37, P = 0.04) but not in the control subjects.
Vitamin B6 indices were altered in specific tissues during inflammation in humans with rheumatoid arthritis
In the human study, plasma pyridoxal 5'-phosphate concentrations were significantly lower in patients than in healthy subjects (with about 50% lower). This observation was comparable to our previous finding [3]. In contrast, no difference was found between patients and controls in erythrocyte pyridoxal 5'-phosphate or erythrocyte aspartate aminotransferase or 4-pyridoxic acid levels. No difference was found in concentrations of plasma folate, red blood cell folate or plasma vitamin B12 between patients and controls.
These results suggest that the lower vitamin B6 concentration in patients with rheumatoid arthritis is tissue-specific.
Induction and progression of adjuvant arthritis in animals
Adjuvant arthritis was induced as described in the Materials and methods section. Arthritis onset was on day 14; the rats injected with adjuvant showed arthritic reactions including swollen paws and hind legs. Inflammation reached its peak on day 21 after the adjuvant injection. Joint swelling and body weight reduction continued for a further 4 weeks after the onset of arthritis (Table 1). Animals in the control and adjuvant groups were well matched in body weight at baseline before the induction of adjuvant arthritis.
Adjuvant arthritis altered vitamin B6 contents in specific tissues in the Lewis rat model
At baseline before the adjuvant/saline injection, there was no difference between the adjuvant arthritis group and the saline-injected group in plasma concentrations of pyridoxal 5'-phosphate or urinary 4-pyridoxic acid excretion. These observations indicated that the animals were also well matched at baseline with regard to their vitamin B6 status. From then on, each control animal received the same amount of food as its experimental counterpart ingested during the previous 24 hours; this pair-feeding procedure minimized the impact of various vitamin intakes between the adjuvant-treated and the control animals. Adjuvant arthritis reached its peak 21 days after the injection. At peak inflammation, significantly lower levels of pyridoxal 5'-phosphate were found in the circulation and in liver in those arthritis rats, but muscle pyridoxal 5'-phosphate concentration seemed to be unaltered. A lower level of pyridoxal 5'-phosphate was also present in the circulation and in liver during chronic inflammation on day 42. However, prolonged inflammation did not alter the muscle content of pyridoxal 5'-phosphate. Plasma pyridoxal 5'-phosphate concentration was correlated with hepatic pyridoxal 5'-phosphate content during peak (Pearson's correlation, r = 0.51, P < 0.005) and chronic (r = 0.38, P < 0.04) inflammation.
Adjuvant arthritis does not increase the urinary excretion of vitamin B6 at peak inflammation or during the chronic phase of inflammation
Despite the significantly lower pyridoxal 5'-phosphate levels in plasma and liver at acute inflammation, we did not observe a significant change in urinary 4-pyridoxic acid or creatinine excretion, indicating that the low level of vitamin B6 in plasma or liver at acute inflammation was not caused by excessive excretion of this vitamin. The 24-hour urinary excretion of 4-pyridoxic acid was highly correlated with the 24-hour urinary creatinine excretion throughout the study, indicating that renal function could be a significant determinant of 4-pyridoxic acid excretion in these animals. At baseline, 24-hour urinary 4-pyridoxic acid excretion was highly correlated with 24-hour urinary creatinine excretion in all rats (Pearson's correlation, r = 0.85, P < 0.0001). The 24-hour urinary excretion of 4-pyridoxic acid was correlated with 24-hour urinary creatinine excretion on day 21 (in control animals, r = 0.615, P = 0.005; in arthritic animals, r = 0.617, P = 0.014) and on day 42 (in control animals, r = 0.78, P < 0.0001; in arthritic animals, r = 0.80, P < 0.0001).
Discussion
The results of these rat and human studies indicate that inflammation directly affects vitamin B6 metabolism differently in different tissues. Furthermore, the low vitamin B6 level is unlikely to be due to a decrease in food intake or the excessive excretion of vitamin B6. Adjuvant arthritis in Lewis rats is a useful animal model for studying vitamin B6 status during inflammation. Adjuvant arthritis decreased the pyridoxal 5'-phosphate pools in the circulation and liver, whereas it did not alter the pyridoxal 5'-phosphate pool in the skeletal muscle. Liver was studied because of its significant metabolic relevance and muscle was studied because it is the major store for vitamin B6. The lower plasma pyridoxal 5'-phosphate concentration in arthritic animals during inflammation was found to be in the physiological range seen in humans with rheumatoid arthritis. Despite the limited number of human subjects, the difference between patients and controls in plasma pyridoxal 5'-phosphate concentration was significant. The average plasma pyridoxal 5'-phosphate concentration in patients with rheumatoid arthritis was about 55% of the level seen in the healthy controls. The mean pyridoxal 5'-phosphate concentration in rats with adjuvant arthritis was about 53% of the controls at peak inflammation on day 21. Rat adjuvant arthritis reflected the altered plasma pyridoxal 5'-phosphate and it is a potential model for studying vitamin B6 status during inflammation.
Lumeng and colleagues suggested that plasma pyridoxal 5'-phosphate concentration reflects vitamin B6 status in the liver in healthy humans [21]. Kinetic studies in rats also indicate that changes in plasma pyridoxal 5'-phosphate content primarily reflect changes in the relatively small, but metabolically relevant and more rapidly exchanging, liver pool (as compared with muscle) [22]. However it was not clear whether this is true during inflammation. The reduced plasma pyridoxal 5'-phosphate level in our study implies that vitamin B6 status in the liver in these patients was altered, because we previously found good correlations between circulating pyridoxal 5'-phosphate level and vitamin B6 functional status measured by methionine load and tryptophan load in these patients [14]. Results from adjuvant arthritis were in agreement with this postulation. In the rat arthritis model, both plasma and hepatic pyridoxal 5'-phosphate concentrations were lower (Table 3). Furthermore, plasma pyridoxal 5'-phosphate concentration was correlated with hepatic pyridoxal 5'-phosphate content. These results suggest that the lower circulating pyridoxal 5'-phosphate levels observed in rheumatoid arthritis could reflect a decrease in hepatic pyridoxal 5'-phosphate pools, and plasma pyridoxal 5'-phosphate is a good indicator of liver B6 status during inflammation.
Our data imply that there are distinct metabolic roles for plasma and erythrocytes in vitamin B6 metabolism during inflammation, and that the impact of inflammation on vitamin B6 is tissue specific. In human subjects, despite the significantly lower pyridoxal 5'-phosphate in plasma (and possibly in liver), erythrocyte pyridoxal 5'-phosphate seemed to be adequate in patients with rheumatoid arthritis, because no difference was found between patients and healthy controls in the erythrocyte pyridoxal 5'-phosphate level or the activity coefficient of erythrocyte aspartate aminotransferase (Table 4). These observations are in agreement with the findings by Talwar and colleagues, which showed that pyridoxal 5'-phosphate decreases in plasma but not erythrocytes during systemic inflammatory response [23].
Data from our animal model imply localized vitamin B6 depletion during inflammation. Before the present study it was not clear how the pyridoxal 5'-phosphate pool in muscle might react to an inflammatory process. In rats with adjuvant arthritis, hepatic pyridoxal 5'-phosphate content was decreased, whereas muscle pyridoxal 5'-phosphate content remained unaltered, suggesting localized vitamin B6 deficiency during inflammation.
Skeletal muscle seems to be less sensitive to vitamin B6 deficiency in humans. In young healthy males receiving a defined diet restricted in vitamin B6, the muscle content of vitamin B6 is relatively resistant to vitamin B6 deficiency, whereas plasma pyridoxal 5'-phosphate is more sensitive to dietary vitamin B6 depletion [24]. We conclude that liver and muscle have distinctive roles as the body undergoes metabolic changes; skeletal muscle, the body's major storage site of vitamin B6, may turn over very slowly during inflammation.
Low vitamin B6 status is unlikely to be due to lower intake or excessive excretion
Dietary intake is known to be a major determinant of vitamin B6 status. The arthritic and control rats showed decreases in plasma pyridoxal 5'-phosphate from the baseline levels. This was partly due to a decrease in overall food intake in both groups. However, the different vitamin B6 status observed between animals with adjuvant arthritis and control animals in the present study was not caused by different food intake between the two groups. Food intake of individual rats in the arthritis group was recorded daily, then a one-to-one match (each rat with adjuvant arthritis had its own weight-matched, saline-injected control) in food intake was arranged. This pair-feeding regimen in our animal experiments minimized the confounding effects of anorexia on the measures of vitamin B6.
Despite the significantly lower plasma pyridoxal 5'-phosphate in patients, 24-hour urinary 4-pyridoxic acid excretion in patients with rheumatoid arthritis did not differ from that of the healthy control subjects (Table 4). The low circulating pyridoxal 5'-phosphate level seen in these patients therefore did not result from excessive catabolism of vitamin B6. This is in agreement with the observation in our animal model. The 24-hour urinary excretion of 4-pyridoxic acid did not differ between control and rats with adjuvant arthritis, despite lower pyridoxal 5'-phosphate levels in plasma and liver in the adjuvant arthritic rats. To summarize these observations, the abnormal vitamin B6 status in rheumatoid arthritis results from the inflammatory process, and it is unlikely that it resulted from insufficient intake or excessive excretion of vitamin B6.
Potential factors involved in the compartmentalization of pyridoxal 5'-phosphate during inflammation
In healthy populations, the variance in plasma pyridoxal 5'-phosphate can be explained to a great extent by vitamin intake, serum albumin, and alkaline phosphatase. The later two are physiological variables directly related to pyridoxal 5'-phosphate metabolism. In an elderly Dutch population it was reported that a combination of vitamin B6 intake, alkaline phosphatase, alcohol consumption, and albumin accounted for 30 to 40% variance in plasma pyridoxal 5'-phosphate [25]. Serum albumin is an acute-phase reactant that decreases during the flaring of active arthritis [26]. As the major protein for pyridoxal 5'-phosphate transport in the circulation, albumin might protect pyridoxal 5'-phosphate from hydrolysis [27]. In the present study, serum albumin was found to be correlated with plasma pyridoxal 5'-phosphate in patients whereas no such correlation was detected in the control subjects. Lower albumin levels in patients with more active arthritis may partly contribute to the lower pyridoxal 5'-phosphate level in these patients, although further study is needed for this postulation.
Many patients with arthritis have been reported to have elevated alkaline phosphatase [28], including those in the present study. Although still in the normal range, mean serum alkaline phosphatase levels in our patients were 26% elevated compared with healthy control subjects. Alkaline phosphatase hydrolyzes the phosphorylated form of vitamin B6 [29]; we therefore speculate that serum alkaline phosphatase could be another key determinant of the concentration of circulating vitamin B6 coenzyme during inflammation. Alkaline phosphatase has been shown to regulate extracellular levels of pyridoxal 5'-phosphate in humans [30,31], and abnormal vitamin B6 metabolism was found in alkaline phosphatase knock-out mice [32]. We found that the serum alkaline phosphatase level was inversely correlated with the plasma pyridoxal 5'-phosphate level in our subjects, which indirectly supports the above hypothesis. Compartmentalization of pyridoxal 5'-phosphate has been reported in the acute-phase response, such as the acute phase of myocardial infarction [33]. Because erythrocyte pyridoxal 5'-phosphate level seem to be normal whereas plasma and hepatic pyridoxal 5'-phosphate levels are significantly lower during inflammation, pyridoxal 5'-phosphate might be compartmentalized between tissues. The elevated alkaline phosphatase during inflammation may facilitate the mobilization and uptake of B6 vitamers, because vitamin B6 is taken up by tissues primarily in the form of pyridoxal.
In contrast, elevated serum alkaline phosphatase or reduced albumin did not provide a satisfactory explanation for the lower plasma pyridoxal 5'-phosphate level in rheumatoid arthritis, because the presence of disease remained a significant determinant of plasma pyridoxal 5'-phosphate concentrations after adjustment for serum alkaline phosphatase and albumin concentrations [34]. The low plasma pyridoxal 5'-phosphate level in patients with rheumatoid arthritis may also be attributed to elevated pyridoxal phosphatase activity during inflammation. It has been reported that the decrease in plasma pyridoxal 5'-phosphate characteristically seen in cirrhosis may be related to a substantial elevation of hepatic pyridoxal 5'-phosphate phosphatase activity [35]. McCarty hypothesized that the pro-inflammatory cytokine interleukin-6 might stimulate the activity of pyridoxal phosphatase in hepatocytes, in these patients, and the elevated enzyme may result in reduced plasma pyridoxal 5'-phosphate concentrations [36].
It remains uncertain whether the activity of pyridoxal 5'-phosphate phosphatase is altered in patients with arthritis, and this should be considered for future studies.
Conclusion
A lower pyridoxal 5'-phosphate concentration in the circulation may reflect the removal of vitamin B6 coenzymes from the circulation to meet the higher demands of certain tissues during inflammation. In the animal model of adjuvant arthritis, lower pyridoxal 5'-phosphate levels in liver implied that it was largely hepatic pyridoxal 5'-phosphate that was used during inflammation. Further studies investigating the kinetics and regulation of B6 vitamers and enzymes in different body compartments are merited.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
All authors made substantive intellectual contributions to the present study. EPC conceived of the study, acquired partial funding, performed the human and animal experiments – including study designs, coordination, biochemical analyses, data acquisition, analysis, and interpretation – and drafted the manuscript. DES participated in the design and procedures of animal experiments. JS participated in the design of the study and the acquisition of funding, and was involved in revising the manuscript critically for important intellectual content. GED participated in the design of the study and performed the statistical analysis. YCW performed the animal experiments and analyses of metabolites. RR conceived of the study, acquired funding, and performed all clinical assessments in study subjects, and revised the manuscript critically for important intellectual content. All authors read and approved the final manuscript.
Acknowledgements
The authors thank Bernadette Muldoon RN, Karin Kohin RD, and Sarah Olson RD for their assistance in recruiting, and thank Dr Pamela Bagley for general support and supervision. Thanks are also given to Marie Nadeau for technical assistance, and the staff at the Human Nutrition Research Center Nutrition Evaluation Laboratory and the Tufts New England Medical Center Clinical Laboratory for hematology and chemistry analyses; to the General Clinical Research Center nursing staff for assistance with the study procedures; and to our volunteers. This project has been supported in part by a grant from the National Science Council of Taiwan (Grant # NSC 94-2320-B005-009; to E-PC). E-PC is a recipient of Dissertation Award from the Arthritis Foundation in the US. This project was also supported by the US Department of Agriculture under cooperative agreement no. 58-1950-9-001. Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the authors and do not necessarily reflect the view of the US Department of Agriculture. This study was also supported in part by grant RR-00054 from the National Center for Research Resources, for the General Clinical Research Center, New England Medical Center and Tufts University School of Medicine (to RR).
Figures and Tables
Table 1 Changes in body weight in response to adjuvant arthritis in Lewis rats
Group Body weight (g)
Baseline Day 21 Day 42
Adjuvant arthritis 275.4 ± 85.3 228.9 ± 71.4 (-16.9%) 214.1 ± 60.7 (-22.3%)
Control 268.3 ± 86.3 259.1 ± 77.5 (-3.4%) 262.1 ± 73.0 (-2.3%)
Data are shown as means ± SD. Percentage changes in body weight at each time point (compared with baseline) are shown in parentheses. Day 21 represents the acute inflammation condition; day 42 represents the chronic inflammation condition.
Table 2 Characteristics of study subjects
Characteristic RA (n = 33) Controls (n = 17) P
Sex (F:M) 23:10 10:7 -
Age 53.5 ± 12.4 52.9 ± 15.7 NS
Height (cm) 167.3 ± 11.0 170.1 ± 8.6 NS
Weight (kg) 77.1 ± 21.1 74.9 ± 17.9 NS
Rheumatoid factor (IU/mL) 89.4 ± 76.1 < 20 <0.0001
Chemistry
Urinary creatinine (g/24 h) 1.0 ± 0.4 1.2 ± 0.3 NS
Serum urea nitrogen (mg/dL) 13.2 ± 4.3 15.6 ± 3.0 <0.05
Albumin (g/dL) 3.9 ± 0.4 4.0 ± 0.4 NS
Serum AST (IU/L) 19.4 ± 9.8 21.9 ± 11.2 NS
Serum ALT (IU/L) 21.4 ± 4.5 23.9 ± 7.5 NS
Alkaline phosphatase (IU/L) 72.2 ± 19.6 57.2 ± 14.3 <0.01
Hematology
White blood cell counts (/nL) 7.3 ± 3.0 5.5 ± 1.2 <0.01
Hematocrit (%) 38.0 ± 3.5 39.8 ± 3.1 0.06
Hemoglobin (g/dL) 13.0 ± 1.6 13.7 ± 1.2 0.08
Mean corpuscular volume (fL) 88.3 ± 4.1 87.6 ± 6.4 NS
ESR (mm/h) 30.8 ± 20.4 8.2 ± 2.8 <0.0001
Data are presented as means ± SD. AST, aspartate aminotransferase; ALT, alanine aminotransferase; ESR, erythrocyte sedimentation rate; RA, rheumatoid arthritis. Bold P values are statistically significant.
Table 3 Effects of adjuvant arthritis on B vitamin status during inflammation in rats
Component Adjuvant arthritis (n = 18) Control (n = 20) P
Baseline
Plasma pyridoxal 5'-phosphate (nmol/L) 793.9 ± 191.3 744.2 ± 145.4 0.89
Urinary 4-pyridoxic acid (μg/d) 30.8 ± 17.3 30.0 ± 18.9 0.89
Acute inflammation
Plasma pyridoxal 5'-phosphate (nmol/L) 252.0 ± 62.5 480.9 ± 144.0 <0.0001
Urinary 4-pyridoxic acid (μg/24 h) 27.7 ± 14.9 32.9 ± 17.2 0.249
Liver pyridoxal 5'-phosphate (nmol/g) 22.6 ± 1.7 27.4 ± 4.9 0.035
Skeletal muscle pyridoxal 5'-phosphate (nmol/g) 7.7 ± 1.8 7.2 ± 0.9 0.88
Chronic inflammation
Plasma pyridoxal 5'-phosphate (nmol/L) 324.2 ± 91.6 393.7 ± 143.5 0.033
Urinary 4-pyridoxic acid (μg/24 h) 31.1 ± 11.8 27.1 ± 14.2 0.39
Liver pyridoxal 5'-phosphate (nmol/g) 21.1 ± 3.0 31.7 ± 4.7 0.0001
Muscle pyridoxal 5'-phosphate (nmol/g) 8.7 ± 4.8 6.4 ± 2.5 0.907
Data are presented as means ± SD. Day 21 represents the acute inflammation condition; day 42 represents the chronic inflammation condition. Bold P values are statistically significant.
Table 4 Indices of vitamin B status in patients with rheumatoid arthritis and control subjects
Component Patients (n = 33) Controls (n = 17) P
Plasma pyridoxal 5'-phosphate (nmol/L)a 24.7 (19.5–31.1) 46.2 (35.3–60.3) 0.001
Red blood cell pyridoxal 5'-phosphate (nmol/L packed red blood cells) 39.7 (34.2–45.2) 33.1 (24.4–41.7) 0.182
αEAST 1.8 (1.8–1.9) 1.9 (1.8–2.1) 0.242
Urinary 4-pyridoxic acid (μg/day)a 0.8 (0.6–1.0) 1.1 (0.9–1.6) 0.08
Plasma folate (μg/L)a 10.2 (8.4–12.5) 9.5 (7.9–11.3) 0.185
Red blood cell folate (μg/L)a 302 (262–347) 274 (226–333) 0.404
Vitamin B12 (ng/L)a 434 (390–496) 398 (334–473) 0.371
Data are shown as geometric means and 95% confidence intervals. aData were log-transformed to achieve normality for statistical analysis. αEAST, erythrocyte aspartate aminotransferase activity coefficient. Bold P values are statistically significant.
==== Refs
Mason JB Miller JW The effects of vitamins B12, B6, and folate on blood homocysteine levels Ann N Y Acad Sci 1992 669 197 203 1444025
Rinehart JF Greenberg LD Vitamin B6 deficiency in the rhesus monkey; with particular reference to the occurrence of atherosclerosis, dental caries, and hepatic cirrhosis Am J Clin Nutr 1956 4 318 325 13339702
Roubenoff R Roubenoff RA Selhub J Nadeau MR Cannon JG Freeman LM Dinarello CA Rosenberg IH Abnormal vitamin B6 status in rheumatoid cachexia. Association with spontaneous tumor necrosis factor alpha production and markers of inflammation Arthritis Rheum 1995 38 105 109 7818558
Chiang EP Bagley PJ Selhub J Nadeau M Roubenoff R Abnormal vitamin B6 status is associated with severity of symptoms in patients with rheumatoid arthritis Am J Med 2003 114 283 287 12681455 10.1016/S0002-9343(02)01528-0
Bates CJ Pentieva KD Prentice A Mansoor MA Finch S Plasma pyridoxal phosphate and pyridoxic acid and their relationship to plasma homocysteine in a representative sample of British men and women aged 65 years and over Br J Nutr 1999 81 191 201 10434845
Cattaneo M Hyperhomocysteinemia: a risk factor for arterial and venous thrombotic disease Int J Clin Lab Res 1997 27 139 144 9352375
Ridker PM Hennekens CH Buring JE Rifai N C-reactive protein and other markers of inflammation in the prediction of cardiovascular disease in women N Engl J Med 2000 342 836 843 10733371 10.1056/NEJM200003233421202
Lakshmi R Lakshmi AV Divan PV Bamji MS Effect of riboflavin or pyridoxine deficiency on inflammatory response Indian J Biochem Biophys 1991 28 481 484 1812086
Folsom AR Desvarieux M Nieto FJ Boland LL Ballantyne CM Chambless LE B vitamin status and inflammatory markers Atherosclerosis 2003 169 169 174 12860264 10.1016/S0021-9150(03)00161-8
Friso S Jacques PF Wilson PW Rosenberg IH Selhub J Low circulating vitamin B6 is associated with elevation of the inflammation marker C-reactive protein independently of plasma homocysteine levels Circulation 2001 103 2788 2791 11401933
Friso S Girelli D Martinelli N Olivieri O Lotto V Bozzini C Pizzolo F Faccini G Beltrame F Corrocher R Low plasma vitamin B-6 concentrations and modulation of coronary artery disease risk Am J Clin Nutr 2004 79 992 998 15159228
Labadarios D Brink PA Weich HF Visser L Louw ME Shephard GS van Stuijvenberg ME Plasma vitamin A, E, C and B6 levels in myocardial infarction S Afr Med J 1987 71 561 563 3576402
Saibeni S Cattaneo M Vecchi M Zighetti ML Lecchi A Lombardi R Meucci G Spina L de Franchis R Low vitamin B6 plasma levels, a risk factor for thrombosis, in inflammatory bowel disease: role of inflammation and correlation with acute phase reactants Am J Gastroenterol 2003 98 112 117 12526945 10.1111/j.1572-0241.2003.07160.x
Chiang EP Bagley PJ Roubenoff R Nadeau M Selhub J Plasma pyridoxal 5'-phosphate concentration is correlated with functional vitamin B-6 indices in patients with rheumatoid arthritis and marginal vitamin B-6 status J Nutr 2003 133 1056 1059 12672918
Hochberg MC Chang RW Dwosh I Lindsey S Pincus T Wolfe F The American College of Rheumatology 1991 revised criteria for the classification of global functional status in rheumatoid arthritis Arthritis Rheum 1992 35 498 502 1575785
Paulini K Korner B Mohr W Sonntag W The effect of complete Freund-adjuvant on chronic proliferating inflammation in an experimental granuloma model Z Rheumatol 1976 35 123 131 1266421
Williams DG Methods for the estimation of three vitamin dependent red cell enzymes Clin Biochem 1976 9 252 255 1000784
Camp VM Chipponi J Faraj BA Radioenzymatic assay for direct measurement of plasma pyridoxal 5'-phosphate Clin Chem 1983 29 642 644 6831691
Gregory JF 3rdKirk JR Determination of urinary 4-pyridoxic acid using high performance liquid chromatography Am J Clin Nutr 1979 32 879 883 433815
Heiskanen K Siimes MA Perheentupa J Salmenpera L Reference ranges for erythrocyte pyridoxal 5'-phosphate concentration and the erythrocyte aspartate transaminase stimulation test in lactating mothers and their infants Am J Clin Nutr 1994 59 1297 1303 8198054
Lumeng L Lui A Li TK Plasma content of B6 vitamers and its relationship to hepatic vitamin B6 metabolism J Clin Invest 1980 66 688 695 7419716
Bode W Mocking JA van den Berg H Influence of age and sex on vitamin B-6 vitamer distribution and on vitamin B-6 metabolizing enzymes in Wistar rats J Nutr 1991 121 318 329 2002404
Talwar D Quasim T McMillan DC Kinsella J Williamson C O'Reilly DS Pyridoxal phosphate decreases in plasma but not erythrocytes during systemic inflammatory response Clin Chem 2003 49 515 518 12600973 10.1373/49.3.515
Coburn SP Ziegler PJ Costill DL Mahuren JD Fink WJ Schaltenbrand WE Pauly TA Pearson DR Conn PS Guilarte TR Response of vitamin B-6 content of muscle to changes in vitamin B-6 intake in men Am J Clin Nutr 1991 53 1436 1442 2035470
Brussaard JH Lowik MR van den BH Brants HA Bemelmans W Dietary and other determinants of vitamin B6 parameters Eur J Clin Nutr 1997 51 Suppl 3 S39 S45 9598767
Denko CW Gabriel P Serum proteins – transferrin, ceruloplasmin, albumin, alpha 1-acid glycoprotein, alpha 1-antitrypsin – in rheumatic disorders J Rheumatol 1979 6 664 672 316826
Lumeng L Brashear RE Li TK Pyridoxal 5'-phosphate in plasma: source, protein-binding, and cellular transport J Lab Clin Med 1974 84 334 343 4853981
Aida S Alkaline phosphatase isoenzyme activities in rheumatoid arthritis: hepatobiliary enzyme dissociation and relation to disease activity Ann Rheum Dis 1993 52 511 516 8102225
Lumeng L Li TK Characterization of the pyridoxal 5'-phosphate and pyridoxamine 5'-phosphate hydrolase activity in rat liver. Identity with alkaline phosphatase J Biol Chem 1975 250 8126 8131 240852
Whyte MP Mahuren JD Vrabel LA Coburn SP Markedly increased circulating pyridoxal-5'-phosphate levels in hypophosphatasia. Alkaline phosphatase acts in vitamin B6 metabolism J Clin Invest 1985 76 752 756 4031070
Anderson BB O'Brien H Griffin GE Mollin DL Hydrolysis of pyridoxal-5'-phosphate in plasma in conditions with raised alkaline phosphate Gut 1980 21 192 194 7399318
Narisawa S Wennberg C Millan JL Abnormal vitamin B6 metabolism in alkaline phosphatase knock-out mice causes multiple abnormalities, but not the impaired bone mineralization J Pathol 2001 193 125 133 11169525 10.1002/1096-9896(2000)9999:9999<::AID-PATH722>3.0.CO;2-Y
Vermaak WJ Barnard HC Van Dalen EM Potgieter GM Van Jaarsveld H Myburgh SJ Compartmentalization of pyridoxal-5'-phosphate during the acute phase of myocardial infarction Klin Wochenschr 1988 66 428 433 3398428 10.1007/BF01745511
Chiang EP Vitamin B6 and homocysteine metabolism in rheumatoid arthritis PhD thesis 2000 Tufts University, Nutritional Science Department
Merrill AH JrHenderson JM Wang E Codner MA Hollins B Millikan WJ Activities of the hepatic enzymes of vitamin B6 metabolism for patients with cirrhosis Am J Clin Nutr 1986 44 461 467 3020959
McCarty MF Increased homocyst(e)ine associated with smoking, chronic inflammation, and aging may reflect acute-phase induction of pyridoxal phosphatase activity Med Hypotheses 2000 55 289 293 11000053 10.1054/mehy.1999.1032
|
16277678
|
PMC1297572
|
CC BY
|
2021-01-04 16:02:40
|
no
|
Arthritis Res Ther. 2005 Sep 13; 7(6):R1254-R1262
|
utf-8
|
Arthritis Res Ther
| 2,005 |
10.1186/ar1821
|
oa_comm
|
==== Front
Arthritis Res TherArthritis Research & Therapy1478-63541478-6362BioMed Central London ar18241627768310.1186/ar1824Research ArticleAssociation between occupational exposure to mineral oil and rheumatoid arthritis: results from the Swedish EIRA case–control study Sverdrup Berit [email protected]ällberg Henrik [email protected] Camilla [email protected] Ingvar 4Padyukov Leonid [email protected] Lars [email protected] Lars [email protected] Epidemiological Investigation of Rheumatoid Arthritis study group 1 Rheumatology Unit, Department of Medicine, Karolinska Institutet/Karolinska Hospital, Stockholm, Sweden2 Rheumatology Unit, Eskilstuna Hospital, Eskilstuna, Sweden3 Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden4 Department of Occupational Medicine, National Institute for Working Life, Stockholm, Sweden5 Stockholm Center for Public Health, Stockholm County Council, Stockholm, Sweden2005 23 9 2005 7 6 R1296 R1303 28 6 2005 29 7 2005 18 8 2005 24 8 2005 Copyright © 2005 Sverdrup 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 aim of the present study was to investigate the association between exposure to mineral oil and the risk of developing rheumatoid arthritis (RA), and in addition to perform a separate analysis on the major subphenotypes for the disease; namely, rheumatoid factor (RF)-positive RA, RF-negative RA, anticitrulline-positive RA and anticitrulline-negative RA, respectively. A population-based case–control study of incident cases of RA was performed among the population aged 18–70 years in a defined area of Sweden during May 1996–December 2003. A case was defined as an individual from the study base who for the first time received a diagnosis of RA according to the American College of Rheumatology criteria of 1987. Controls were randomly selected from the study base with consideration taken for age, gender and residential area. Cases (n = 1,419) and controls (n = 1,674) answered an extensive questionnaire regarding lifestyle factors and occupational exposures, including different types of mineral oils. Sera from cases and controls were investigated for RF and anticitrulline antibodies.
Among men, exposure to any mineral oil was associated with a 30% increased relative risk of developing RA (relative risk = 1.3, 95% confidence interval = 1.0–1.7). When cases were subdivided into RF-positive RA and RF-negative RA, an increased risk was only observed for RF-positive RA (relative risk = 1.4, 95% confidence interval 1.0–2.0). When RA cases were subdivided according to the presence of anticitrulline antibodies, an increased risk associated with exposure to any mineral oil was observed only for anticitrulline-positive RA (relative risk = 1.6, 95% confidence interval = 1.1–2.2). Analysis of the interaction between oil exposure and the presence of HLA-DR shared epitope genes regarding the incidence of RA indicated that the increased risk associated with exposure to mineral oil was not related to the presence of shared epitope genotypes.
In conclusion, our study shows that exposure to mineral oil is associated with an increased risk to develop RF-positive RA and anticitrulline-positive RA, respectively. The findings are of particular interest since the same mineral oils can induce polyarthritis in rats.
==== Body
Introduction
Rheumatoid arthritis (RA) is a disease that is dependent on genetic factors as well as environmental factors, as seen from both concordance data in twins and from a number of epidemiological and genetic studies [1,2]. Whereas knowledge on the genetic basis of this disease is rapidly advancing [3,4], there is a scarcity of data on environmental agents that may cause arthritis [5-7]. In particular, very little information exists in humans on environmental risk factors with a capacity to induce arthritis in experimental arthritis systems.
Agents that are able to induce experimental arthritis in animals, particularly rodents, include a number of adjuvants originating from many different sources, including bacteria, yeast, viruses and mineral oils. Several models thus exist where rodents with certain genetic backgrounds develop arthritis after being exposed to nonimmunogenic adjuvants intracutaneously [8,9] or even percutaneously [10,11]. The exact mechanisms involved in the pathogenesis of these adjuvant arthritis models are still not completely understood, but we know that the adjuvants/mineral oils can activate cells within the lymph nodes without causing any simultaneous apparent inflammatory reaction in the skin [10]. Whether similar mechanisms (i.e. polyarthritis induced by simple adjuvants) are also operative in human arthritis is an open issue, although case reports exist on the occurrence of arthritis after immunization with Bacillus Calmette–Guerin [12,13], which is known to contain adjuvants able to cause arthritis in rodents [14].
In order to investigate the possible relationship between the occurrence of RA and the exposure to a series of different environmental agents, including simple adjuvants, we are currently performing a large population-based case–control study in Sweden using incident cases of RA. In the present report, we investigate the association between exposure to various mineral oils and the risk of developing RA.
Materials and methods
The present study is a population-based case–control study of incident cases of RA among the population aged 18–70 years living in a geographically defined area in the middle part and southern part of Sweden during the period May 1996–December 2003. Ethical permission was obtained from relevant ethical committees and all the participants (cases as well as controls) consented to contribute to the study.
Case identification
A case was defined as a person in the study base who for the first time received a diagnosis of RA according to the American College of Rheumatology criteria of 1987 [15]. As described previously [16], all potential cases were examined and diagnosed by a rheumatologist at the unit entering the case into the study. All rheumatology units linked to the general welfare system in the study area participated in the study, as well as almost all of the, very few, privately-run rheumatology units. In total there were 19 reporting clinics, 15 of which were 'Early Arthritis Clinics' [17]. At the start some centres also reported cases that did not satisfy the criteria in order to enable investigations of undifferentiated arthritis, but these subjects were eventually excluded from the study.
Analysis of rheumatoid factor (RF) was performed locally and reported as RF-positivity or RF-negativity. The RF levels were determined and the cutoff value was set to 20.
Selection of controls
For each potential case a control was randomly selected from the study base with consideration taken for age, gender and residential area. The selection of controls was conducted using the national population register, which is continuously updated. If a control declined to participate, was not traceable or reported having RA, a new control was selected using the same principles (see also [16]). Controls belonging to cases excluded due to not fulfilling the American College of Rheumatology criteria remained in the study.
Data collection
Information about environmental exposures was collected by an identical questionnaire given to the cases shortly after they had been informed about the RA diagnosis and sent by mail to the controls. All questionnaires were supposed to be answered at home.
Unanswered or incompletely answered questionnaires were completed by mail or by telephone by purpose-trained persons not connected to the rheumatology clinics. This was carried out in an identical way for the case and control groups. In total, 1,480 cases and 2,038 controls were identified. Of these, 1,419 cases (1,012 women and 407 men) and 1,674 controls (1,188 women and 486 men) participated in the study, giving a participation rate of 96% for cases and a participation rate of 82% for controls.
Exposure
The questionnaire contains questions within a wide spectrum regarding personal circumstances, including lifestyle factors, occupational exposures, health aspects, socioeconomic factors and demographic data. Specific questions were asked about occupational exposure to cutting oil, motor oil, form oil, hydraulic oil and asphalt, respectively. This enabled the classification of cases and controls with regard to ever having been occupationally exposed to each of these mineral oils, respectively. Subjects who reported exposure to any of these mineral oils were classified as exposed to any mineral oil.
For each case the time point at which symptoms giving suspicion of RA started was used as an estimation of the disease onset. The year in which this time point occurred was defined as the index year. The same index year was used for the corresponding control. Only data on exposures up to the index year have been analysed in the present study.
Detection of antibodies to citrulline-containing peptides
Antibodies to citrulline-containing peptides (anti-CP) were analysed with the Immunoscan-RA Mark2 ELISA test (see [18]). A level above 25 U/ml was regarded as positive according to instructions in the kit and validation at the clinical immunology laboratory in Uppsala. The kit uses cyclic citrullinated peptides as the substrate, and sera reacting positively with this kit are, in the present paper, defined as having significant antibody titres against citrullinated peptides (anti-CP+)
Genotyping
DNA from ethylenediaminetetraacetic acid blood was extracted using the sequence-specific primer PCR method [19]. Among the HLA-DRB1 genes, DRB1*01, DRB1*04 and DRB1*10 were defined as 'shared epitope (SE) genes' [3,4]. At the beginning of the study, individuals from part of the material (81 cases) were subtyped for identification of HLA-DRB1*01 and HLA-DRB1*04 alleles. We determined 89% frequency of the DRB1*0101 allele and 98% frequency of the DRB1*0401 + DRB1*0404 + DRB1*0405 + DRB1*0408 alleles. For practical reasons the genotyping was restricted to only DR low-resolution analysis.
Potential confounding factors
All results were adjusted for age and residential area according to the principle of control selection. In the analyses, age was categorised into 10 strata (18–24, 25–29, 30–34, 35–39, 40–44, 45–49, 50–54, 55–59, 60–64 and 65–70 years of age). Smoking and occupational class could also be considered as potential confounding factors. Smoking was categorised into two strata (never smokers and ever smokers) and occupational class was categorised into seven strata (unskilled manual workers, skilled manual workers, assistant non-manual employees, intermediate non-manual employees, higher non-manual employees, self-employed and farmers).
Statistical analysis
Subjects who had been exposed to different mineral oils were compared with subjects unexposed to any mineral oil with regard to the incidence of RF+ RA, RF- RA, anti-CP+ RA, anti-CP- RA and RA overall, respectively, by calculating the odds ratio with the 95% confidence interval (CI). We performed matched analyses as well as unmatched analyses of the data. Odds ratios were adjusted for potential confounding by means of conditional logistic regression in the matched analyses and by means of unconditional logistic regression in the unmatched analyses. We only present results from the unmatched analyses as these were in close agreement with those from the matched analyses but, in general, had higher precision. Odds ratios were interpreted as the relative risk (RR) because the study was population based. Results for women and men were analysed separately. Estimates of RR were adjusted for potential confounding from age, gender, residential area and smoking. Further adjustment for occupational class only marginally changed the estimates and was not retained in the final analyses.
The presence of HLA-DR SE genes is a risk factor for RF+ RA and anti-CP+ RA, but not for either RF- RA or anti-CP- RA [20,21]. It is thus of interest to investigate the possibility of a gene–environment interaction between SE genes and exposure to mineral oil regarding the incidence of RF+ RA and anti-CP+ RA, respectively. Interaction between the genotype and mineral oil was evaluated using departure from the additivity of effects as a criterion of interaction, as suggested by Rothman and colleagues [22]. To quantify the amount of interaction, the attributable proportion due to interaction was calculated together with the 95% CI [23]. The attributable proportion due to interaction, which takes a value between 0 and 1, is the proportion of the incidence among persons exposed to two interacting factors that is attributable to the interaction per se (i.e. reflecting their joint effect beyond the sum of their independent effects). A potential interaction between smoking and mineral oil was also evaluated. All analyses were performed using the Statistical Analysis System (version 8.2; SAS Institute, Stockholm, Sweden).
Results
Of a total of 1,419 cases in this study, 1,012 were women and 407 were men (mean age at inclusion of 50 and 53 years, respectively). A total 65.5% of the female cases and 66.3% of the male cases were RF+. The mean duration of disease at inclusion in the study was 10 months. Only men reported a substantial occupational exposure to mineral oils (in total 135 cases and 132 controls), with occupational exposure to mineral oils uncommon among women (21 cases and 21 controls). Only men were retained in the further analysis. Among these men, motor oil (84 cases and 84 controls) and hydraulic oil (83 cases and 72 controls) were the most common exposures.
Exposure to any mineral oil was associated with a 30% increased risk of developing RA (RR = 1.3, 95% CI = 1.0–1.7) (Table 1). When cases were subdivided into RF+ RA and RF- RA, an increased risk was only observed for RF+ RA (RR = 1.4, 95% CI = 1.0–2.0). The same pattern with a higher RR associated with RF+ RA was observed for all of the specific mineral oils. The RR of developing RF+ RA associated with exposure to hydraulic oil was 1.5 (95% CI = 1.0–2.3).
We also investigated the relationship between exposure to different oils and the incidence of anti-CP+ RA and anti-CP- RA, respectively (Table 2). Exposure to any mineral oil was associated with a 60% increased risk of anti-CP+ RA (RR = 1.6, 95% CI = 1.1–2.2). Higher RR associated with anti-CP+ RA as compared with anti-CP- RA was seen for all the specific mineral oils. The RR of developing anti-CP+ RA associated with exposure to hydraulic oil was 1.7 (95% CI = 1.1–2.6), and that for motor oil was 1.5 (95% CI = 1.0–2.3).
In the analysis, adjustment was made according to age, residential area and smoking. The results after adjustment for smoking were almost identical to those not adjusted for smoking (Tables 1 and 2).
The presence of HLA-DR SE genes is a risk factor for RF+ RA and anti-CP+ RA, but not for either RF- RA or anti-CP- RA [20,21]. When we analysed the possibility of a gene–environment interaction between SE genes and exposure to mineral oil regarding the incidence of RF+ RA and anti-CP+ RA, respectively, no strong evidence of such an interaction was found (Table 3 and Fig. 1). The attributable proportion due to the interaction between SE genes and mineral oil was 0.2 (95% CI = -0.2-0.6) regarding RF+ RA as well as regarding anti-CP+ RA.
When a potential interaction between smoking and mineral oils was analysed, a tendency towards such an interaction was noted but a firm conclusion was hampered by the small numbers (regarding anti-CP+ RA, the attributable proportion due to interaction was 0.5 [95% CI = -0.2-1.2]).
Discussion
According to the results of this study, males exposed to various mineral oils in their profession appear to have an increased risk of developing RA. This observation is of interest in the search for aetiological factors of importance for triggering RA, as exposure to the same kinds of oils have been shown to induce polyarthritis with large similarities to RA in rodents [8,9].
This study has the advantage of being a population-based case–control study using only incident cases of new diagnosed RA, fulfilling the American College of Rheumatology criteria, assessed by a specialist in rheumatology. All rheumatology units linked to the general welfare system in the study area reported cases to the study, as did privately-run rheumatology units [17]. Cases received their questionnaire in connection with inclusion into the study at each study centre (i.e. in connection with the time point of the RA diagnosis), whereas the controls received their questionnaires by mail. All questionnaires were supposed to be answered at home. Both cases and controls returned their questionnaire by mail to the study secretariat at the Karolinska Institutet. It is unlikely that the differential distribution process of the questionnaire to cases and to controls, respectively, led to any important difference regarding the quality of exposure information.
A possible disadvantage with a case–control study with retrospective collection of exposure data is the risk for misclassification of exposure due to a recall bias that differs between cases and controls. Only subjects that received a diagnosis of RA for the first time were included in order to reduce the risk for recall bias, the mean duration between the estimated disease onset and inclusion into the study was 10 months, and analyses of data of environmental exposures were only performed up to the index year. Bias due to change in habits, job or work exposure as a result of the disease was therefore probably limited. In order to investigate whether misclassification of exposure to mineral oils differed between cases and controls, an industrial hygienist performed an independent classification based on information regarding occupation and the branch of industry during the period of the stated exposure to mineral oils. The industrial hygienist was blinded with regard to the disease status of the individuals. According to the result of this investigation a similar proportion (15% and 20%) of cases and controls seemed to be false positive with regard to exposure to mineral oils. The assessment made by the hygienist should not be regarded as more qualified than the assessment made by the subjects. However, the marked and similar correspondence of the hygienist's assessment and the subjects' statements among cases and controls, respectively, suggests that differential misclassification of exposure has probably not biased the estimated RR to any great extent. The response rate in the study was high, with 96% for cases and 82% for controls, which limits risk for selection bias in this stage.
All results were adjusted for age and residential area according to the principle of control selection. In the analysis, we investigated the potential confounding from smoking. Adjustment for smoking only marginally changed the estimated RR (Tables 1 and 2). Hence, differences regarding smoking habits do not explain the observed association between exposure to mineral oils and the risk of RA. When a potential interaction between smoking and mineral oils was analysed, a tendency towards such an interaction was noted, but the observation was based on a small number.
According to the results of our study, mineral oils (in particular, hydraulic oil and motor oil) appear to be associated with a particular high risk of RF+ RA and anti-CP+ RA. Bearing in mind the relatively small number of exposed men, however, caution is warranted regarding any far-reaching conclusion about particular mineral oils.
Previously, to our knowledge, only one study has investigated the relationship between exposure to mineral oils and the development of RA [24]. This is notable, since mineral oils have been very well documented as arthritogenic agents in rodents [8]. The administration of mineral oils, therefore, both intracutaneously as one single injection and percutaneously in multiple exposures, has been shown to induce an erosive and RA-like polyarthritis in certain strains of rats, in particular the DA rat strain [10,11].
Our results confirm those of previous studies that the presence of HLA-DR SE genes is a risk factor for anti-CP+ RA and RF+ RA but not for either anti-CP- RA or RF- RA [20,21]. Analysis of a possible interaction between the SE genes and exposure to mineral oils did not reveal any significant interaction between exposure to mineral oil and the presence of HLA-DR SE genes. Although based on small numbers of observations, this suggests that mechanisms responsible for the association between mineral oil exposure and RA may be different from those responsible for the association between smoking and RA, where a pronounced interaction between smoking and the HLA-DR SE genes was observed [20,21].
The potential molecular pathogenesis of polyarthritis associated with exposure to mineral oil is difficult to speculate on. It is known in experimental animals, however, that oils and other adjuvants confer their activation of the immune system mainly in the lymph nodes, without leaving any signs of inflammation in the exposed skin [11,25-28]. Here, an initial activation of the innate immune system subsequently leads to the activation also of T-cell immunity, in such a way that the T cells can subsequently transfer the disease to naive animals [29]. It is also known in rodents that susceptibility to adjuvant-induced arthritis, including oil-induced disease, is highly dependent on the genetic constitution of the exposed animals [26]. It is thus possible that humans with a genetic constitution similar to the adjuvant-susceptible rodents would also have a particularly high susceptibility for mineral-oil associated arthritis. This question may soon be possible to investigate further if precise polymorphic genes associated with susceptibility to adjuvant arthritis in rodents are identified, and indications that such genes exist have already been provided [30].
It is thus possible to hypothesise that adjuvant stimulus of the innate immune system, taking place in genetically susceptible human beings, would trigger the activation of a cascade of events involving activated T lymphocytes, and that these events for as yet unknown reasons finally result in inflammatory joint disease. In this context it is of interest that the association between RA and exposure to mineral oil does not appear to be related to the presence or absence of HLA-DR SE genes in the mineral-exposed individuals. In conjunction with another environmental exposure – smoking – a major interaction was seen between smoking and the presence of HLA-DR SE. This interaction has been suggested to depend on the capacity of smoke to induce an aberrant citrullination of proteins in the lungs of long-term smokers, something that may trigger anti-CP immunity in individuals carrying HLA-DR SE genes [21].
The present demonstration of an exposure not linked to the presence of HLA-DR SE genes indicates a complex pattern of interactions between several environmental triggers, several genetic features and, eventually, several patterns of immunoreactivities in the pathogenesis of RA. If these different patterns in humans can be linked to different rodent models of arthritis – as is suggested from the findings in the present paper – we may be able to use knowledge of molecular pathogenesis and targeted therapies gained in different animal systems to develop a better understanding of both pathogenesis and treatment of relevant subgroups of the human disease.
Conclusion
The present study shows that exposure to an environmental agent capable of inducing an RA-like polyarthritis in rodents – mineral oil – is associated with an increased risk for RF+ RA and anti-CP+ RA in man. Further exploration of this finding may be of interest in elucidating whether other types of adjuvants, such as microbial agents and other occupational agents, can also act as arthritis-inducing agents in humans, and to further link the molecular pathogenesis of adjuvant-associated arthritis in rodents with adjuvant-induced arthritis in man.
Abbreviations
anti-CP = antibodies to citrulline-containing peptides; CI = confidence interval; EIRA = Epidemiological Investigation of Rheumatoid Arthritis; ELISA = enzyme-linked immunosorbent assay; PCR = polymerase chain reaction; RA = rheumatoid arthritis; RF = rheumatoid factor; RR = relative risk; SE = shared epitope.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
BS contributed to the design of the study, and the interpretation and writing of the manuscript. HK performed a major part of the biostatistics work and contributed to the interpretation of results and the writing of the manuscript. CB contributed to the design of the study, to the collection of the data, to statistical analysis and to the writing of the manuscript. IL contributed to the design of the study and to the interpretation of the results. LP had the main responsibility for the genetic analyses and contributed to the analysis and interpretation of the results. LA and LK were responsible for the overall design of the Epidemiological Investigation of Rheumatoid Arthritis study, for the analysis of data and for the final writing of the manuscript. All authors read and approved the final text before submission of the manuscript.
Acknowledgements
The authors wish to thank Marie-Louise Serra (Institute of Environmental Medicine at Karolinska Institutet, Stockholm, Sweden) for excellent assistance in the collection of data and to thank Lena Nise (Institute of Environmental Medicine at Karolinska Institutet, Stockholm, Sweden) for data analysis. The study was supported by grants from the Swedish Medical Research Council, from the Swedish Council for Working life and Social Research, from the King Gustaf V's 80-year foundation, from the Swedish Rheumatism Foundation, from Stockholm County Council, from the Söderberg Foundation and from the insurance company AFA.
The Epidemiological Investigation of Rheumatoid Artrhitis study group consists of: Ingeli Andréasson, Landvetter; Eva Baecklund, Akademiska Hospital; Ann Bengtsson and Thomas Skogh, Linköping Hospital; Johan Bratt and Ingiäld Hafström, Karolinska University Hospital, Huddinge; Jan Cedergren and Ethel Nilsson, Norrköping Hospital; Kjell Huddénius, Rheumatology Clinic in Stockholm City; Shirani Jayawardene, Bollnäs Hospital; Ann Knight, Hudiksvall Hospital; Ido Leden, Kristianstad Hospital; Thomas Lerndal and Göran Lindahl, Danderyd Hospital; Bengt Lindell, Kalmar Hospital; Christin Lindström and Gun Sandahl, Sophiahemmet; Björn Löfström, Katrineholm Hospital; Birgitta Nordmark, Karolinska University Hospital, Solna; Ingmar Petersson, Spenshult Hospital; Christoffer Schaufelberger, Sahlgrenska University Hospital; Patrik Stolt, Västerås Hospital; Berit Sverdrup, Eskilstuna Hospital; Olle Svernell, Västervik Hospital; and Tomas Weitoft, Gävle Hospital.
Figures and Tables
Figure 1 Relative risk of anti-CP-positive RA with mineral exposure and different expression of SE genes. The relative risk (RR) of developing rheumatoid arthritis (RA) positive for antibodies to citrulline-containing peptides (anti-CP) with mineral exposure and shared epitope (SE) genes compared with unexposed subjects with SE genes is 1.4, with 95% confidence interval = 0.8–2.4. The corresponding risk among subjects with no SE gene is 1.5, with 95% confidence interval = 0.6–3.9.
Table 1 Relative risk for developing RF-positive or RF-negative RA in men exposed to mineral oils
Outcome Oil Exposed cases/controls (n)a Relative riskb 95% confidence interval Relative riskc 95% confidence interval
Total RA Cutting fluid 36/39 1.1 0.7–1.9 1.1 0.7–1.8
Motor oil 84/84 1.2 0.9–1.8 1.2 0.9–1.8
Hydraulic oil 83/72 1.4 1.0–2.1 1.4 1.0–2.0
Form oil 25/24 1.3 0.7–2.5 1.3 0.7–2.4
Asphalt 13/14 1.3 0.6–2.8 1.2 0.6–2.7
Any mineral oil 135/132 1.3 1.0–1.7 1.3 1.0–1.7
RF+ RA Cutting fluid 29/39 1.4 0.8–2.5 1.4 0.8–2.4
Motor oil 60/84 1.4 0.9–2.1 1.4 0.9–2.1
Hydraulic oil 56/72 1.6 1.0–2.3 1.5 1.0–2.3
Form oil 20/24 1.7 0.9–3.3 1.7 0.9–3.2
Asphalt 9/14 1.4 0.6–3.5 1.4 0.6–3.4
Any mineral oil 96/132 1.5 1.0–2.0 1.4 1.0–2.0
RF- RA Cutting fluid 7/39 0.6 0.2–1.4 0.6 0.2–1.4
Motor oil 24/84 0.9 0.6–1.6 0.9 0.5–1.6
Hydraulic oil 27/72 1.2 0.7–2.1 1.2 0.7–2.1
Form oil 5/24 0.6 0.2–1.8 0.6 0.2–1.8
Asphalt 4/14 0.9 0.3–3.0 0.9 0.3–3.0
Any mineral oil 39/132 1.0 0.6–1.5 1.0 0.6–1.5
Relative risk and 95% confidence interval for developing rheumatoid factor (RF)-positive rheumatoid arthritis (RA), RF- RA and RA overall (total RA) for men 18–70 years old exposed to different kinds of mineral oils compared with unexposed men.
aRF status unknown for one unexposed case.
bAdjusted for age and residential area.
cAdjusted for age, residential area and smoking.
Table 2 Relative risk for developing anti-CP-positive RA or anti-CP-negative RA for men exposed to mineral oils
Outcome Oil Exposed cases/controls (n) Relative riska 95% confidence interval Relative riskb 95% confidence interval
Anti-CP+ RA Cutting fluid 28/39 1.6 0.9–2.7 1.5 0.8–2.6
Motor oil 57/84 1.5 1.0–2.2 1.5 1.0–2.3
Hydraulic oil 56/72 1.8 1.2–2.7 1.7 1.1–2.6
Form oil 16/24 1.5 0.7–2.9 1.4 0.7–2.9
Asphalt 10/14 1.6 0.7–3.9 1.5 0.6–3.7
Any mineral oil 93/132 1.6 1.1–2.2 1.6 1.1–2.2
Anti-CP- RA Cutting fluid 8/39 0.6 0.2–1.3 0.6 0.2–1.3
Motor oil 27/84 0.9 0.6–1.5 0.9 0.6–1.5
Hydraulic oil 27/72 1.0 0.6–1.7 1.0 0.6–1.7
Form oil 9/24 1.1 0.5–2.6 1.1 0.5–2.6
Asphalt 3/14 0.7 0.2–2.6 0.7 0.2–2.6
Any mineral oil 42/132 1.0 0.6–1.5 1.0 0.6–1.5
Relative risk and 95% confidence interval for developing antibodies to citrulline-containing peptides (anti-CP)-positive rheumatoid arthritis (RA) and anti-CP- RA for men 18–70 years old exposed to different kinds of mineral oils compared with unexposed men.
aAdjusted for age and residential area.
bAdjusted for age, residential area and smoking.
Table 3 Relative risk for RA with different combinations of mineral oil exposure and shared epitope genes
Outcome Exposure to mineral oil No shared epitope Any shared epitope
Cases/controls (n)a Relative riskb 95% confidence interval Cases/controls (n)a Relative riskb 95% confidence interval
RF+ RA No 28/89 1.0c 128/88 4.7 2.8–8.0
Yes 17/43 1.2 0.6–2.5 73/38 6.4 3.4–11.8
RF- RA No 32/89 1.0c 63/88 2.1 1.2–3.6
Yes 10/43 0.7 0.3–1.5 25/38 2.1 1.1–4.2
Total RA No 60/89 1.0c 191/88 3.2 2.1–4.9
Yes 27/43 0.9 0.5–1.7 98/38 3.9 2.3–6.6
Anti-CP+ RA No 18/89 1.0c 126/88 8.0 4.4–14.7
Yes 13/43 1.6 0.7–3.7 75/38 11.1 5.6–22.1
Anti-CP- RA No 42/89 1.0c 65/88 1.5 0.9–2.5
Yes 14/43 0.7 0.3–1.5 23/38 1.5 0.7–2.9
Total RA No 60/89 1.0c 191/88 3.2 2.1–4.9
Yes 27/43 0.9 0.5–1.7 98/38 3.9 2.3–6.6
Relative risk and 95% confidence interval for developing rheumatoid factor (RF)-positive rheumatoid arthritis (RA), RF- RA, antibodies to citrulline-containing peptides (anti-CP)-positive RA, anti-CP- RA and total RA for men with different combinations of exposure to mineral oil and shared epitope genes compared with men unexposed to any mineral oil and lacking shared epitope genes.
aData on the shared epitope are missing for 10 cases and 51 controls exposed to mineral oil, and for 21 cases and 176 controls unexposed to mineral oil (in comparison with Tables 1 and 2).
bRelative risk adjusted for age, residential area and smoking.
cReference group.
==== Refs
Rooney BK Silman AJ Epidemiology of the rheumatic diseases Curr Opin Rheumatol 1999 11 91 97 10319210 10.1097/00002281-199903000-00002
MacGregor AJ Snieder H Rigby AS Koskenvuo M Kaprio J Aho K Silman AJ Characterizing the quantitative genetic contribution to rheumatoid arthritis using data from twins Arthritis Rheum 2000 43 30 37 10643697 10.1002/1529-0131(200001)43:1<30::AID-ANR5>3.0.CO;2-B
Gregersen PK Silver J Winchester RJ The shared epitope hypothesis. An approach to understanding the molecular genetics of susceptibility to rheumatoid arthritis Arthritis Rheum 1987 30 1205 1213 2446635
Jawaheer D Gregersen PK Rheumatoid arthritis. The genetic components Rheum Dis Clin North Am 2002 28 1 15 11840692 10.1016/S0889-857X(03)00066-8
Symmons DP Bankhead CR Harrison BJ Brennan P Barrett EM Scott DG Silman AJ Blood transfusion, smoking and obesity as risk factors for the development of rheumatoid arthritis: results from a primary care-based incident case-control study in Norfolk, England Arthritis Rheum 1997 40 1955 1961 9365083
Uhlig T Hagen KB Kvien TK Current tobacco smoking, formal education and the risk of rheumatoid arthritis J Rheumatol 1999 26 47 54 9918239
Karlson EW Lee IM Cook NR Manson JE Buring JE Hennekens CH A retrospective cohort study of cigarette smoking and risk of rheumatoid arthritis in female health professionals Arthritis Rheum 1999 42 910 917 10323446 10.1002/1529-0131(199905)42:5<910::AID-ANR9>3.0.CO;2-D
Kleinau S Erlandsson H Holmdahl R Klareskog L Adjuvant oils induce arthritis in the DA rat. I. Characterization of the disease and evidence for an immunological involvement J Autoimmun 1991 4 871 880 1812893 10.1016/0896-8411(91)90050-M
Cannon GW Woods ML Clayton F Griffiths MM Induction of arthritis in DA rats by incomplete Freund's adjuvant J Rheumatol 1993 20 7 11 8441169
Kleinau S Erlandsson H Klareskog L Percutaneous exposure of adjuvant oil causes arthritis in DA rats Clin Exp Immunol 1994 96 281 284 8187335
Sverdrup B Klareskog L Kleinau S Common commercial cosmetic products induce arthritis in the DA rat Environ Health Perspect 1998 106 27 32 9417771
Torisu M Miyahara T Shinohara N Ohsato K Sonozaki H A new side effect of BCG immunotherapy – BCG-induced arthritis in man Cancer Immunol Immunother 1978 5 77 83 10.1007/BF00199980
Hughes RA Allard SA Maini RN Arthritis associated with adjuvant mycobacterial treatment for carcinoma of the bladder Ann Rheum Dis 1989 48 432 434 2786389
Waksman BH Pearson CM Sharp JT Studies of arthritis and other lesions induced in rats by injection of mycobacterial adjuvant. II. Evidence that the disease is a disseminated immunological response to exogenous antigen J Immunol 1960 85 403 417 13782608
Arnett FC Edworthy SM Bloch DA McShane DJ Fries JF Cooper NS Healey LA Kaplan SR Liang MH Luthra HS The American Association 1987 revised criteria for the classification of rheumatoid arthritis Arthritis Rheum 1988 31 315 324 3358796
Stolt P Bengtsson C Nordmark B Lindblad S Lundberg I Klareskog L Alfredsson L the EIRA study group Quantification of the influence of cigarette smoking on rheumatoid arthritis: results from a population based case–control study, using incident cases Ann Rheum Dis 2003 62 835 841 12922955 10.1136/ard.62.9.835
Klareskog L Nordmark B Lindblad S On the organization of an early arthritis clinic Best Pract Res Clin Rheumatol 2001 15 1 15 11358411 10.1053/berh.2000.0122
Rönnelid J Wick MC Lampa J Lindblad S Nordmark B Klareskog L van Vollenhoven R Longitudinal analysis of anti-citrullinated protein/peptide antibodies (anti-CP) during 5 years follow-up in early rheumatoid arthritis: anti-CP status is a stable phenotype that predicts worse disease activity and greater radiological progression Ann Rheum Dis 2005
Olerup O Zetterquist H HLA-DR typing by PCR amplification with sequence-specific primers (PCR-SSP) in 2 hours: an alternative to serological DR typing in clinical practice including donor-recipient matching in cadaveric transplantation Tissue Antigens 1992 39 225 235 1357775
Padyukov L Silva C Stolt P Alfredsson L Klareskog L A gene–environment interaction between smoking and shared epitope genes in HLA-DR provides a high risk of seropositive rheumatoid arthritis Arthritis Rheum 2004 50 3085 3092 15476204 10.1002/art.20553
Klareskog L Stolt P Lundberg K Källberg H Bengtsson C Grunewald J Rönnelid J Erlandsson-Harris H Ulfgren A-K Rantapää-Dahlqvist S the EIRA study group A new model for an etiology of RA. Smoking may trigger HLA-DR (SE)-restricted immune reactions to autoantigens modified by citrullination Arthritis Rheum 2005
Rothman KJ Greenland S Walker AM Concepts of interaction Am J Epidemiol 1980 112 467 470 7424895
Hosmer DW Lemeshow S Confidence interval estimation of interaction Epidemiology 1992 3 452 456 1391139
Olsson A Skogh T Wingren G Occupational determinants for rheumatoid arthritis Scand J Work Environ Health 2000 26 243 249 10901117
Kleinau S Dencker L Klareskog L Oil-induced arthritis in DA rats: tissue distribution of arthritogenic 14C-labelled hexadecane Int J Immunopharmacol 1995 17 393 401 7591363 10.1016/0192-0561(95)00020-3
Holmdahl R Lorentzen JC Lu S Olofsson P Wester L Holmberg J Pettersson U Arthritis induced in rats with nonimmunogenic adjuvants as models for rheumatoid arthritis Immunol Rev 2001 184 184 202 12086312 10.1034/j.1600-065x.2001.1840117.x
Holm BC Svelander L Bucht A Lorentzen JC The arthritogenic adjuvant squalene does not accumulate in joints, but gives rise to pathogenic cells in both draining and non-draining lymph nodes Clin Exp Immunol 2002 127 430 435 11966758 10.1046/j.1365-2249.2002.01783.x
Mussener A Klareskog L Lorentzen J Kleinau S TNFα dominates mRNA expression in lymphoid tissues of rats developing collagen- and oil-induced arthritis Scand J Immunol 1995 42 128 134 7631134
Svelander L Mussener A Erlandsson-Harris H Kleinau S Polyclonal Th1 cells transfer oil-induced arthritis Immunology 1997 91 260 265 9227326 10.1046/j.1365-2567.1997.00251.x
Barton A Eyre S Myerscough A Brintnell B Ward D Ollier WE Lorentzen JC Klareskog L Silman A John S Worthington J Arthritis Rheumatism Campaign National Repository High resolution linkage and association mapping identifies a novel rheumatoid arthritis susceptibility locus homologous to one linked to two rat models of inflammatory arthritis Hum Mol Genet 2001 10 1901 1906 11555626 10.1093/hmg/10.18.1901
|
16277683
|
PMC1297574
|
CC BY
|
2021-01-04 16:02:39
|
no
|
Arthritis Res Ther. 2005 Sep 23; 7(6):R1296-R1303
|
utf-8
|
Arthritis Res Ther
| 2,005 |
10.1186/ar1824
|
oa_comm
|
==== Front
Arthritis Res TherArthritis Research & Therapy1478-63541478-6362BioMed Central London ar18251627767910.1186/ar1825Research ArticleAerobic exercise and its impact on musculoskeletal pain in older adults: a 14 year prospective, longitudinal study Bruce Bonnie [email protected] James F [email protected] Deborah P [email protected] Stanford University, Department of Immunology/Rheumatology, Palo Alto, CA 94304, USA2 Health Economics, Genentech/MS 241A, South San Francisco, CA 94080, USA2005 19 9 2005 7 6 R1263 R1270 20 5 2005 29 6 2005 23 8 2005 24 8 2005 Copyright © 2005 Bruce 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.
We studied the long term impact of running and other aerobic exercise on musculoskeletal pain in a cohort of healthy aging male and female seniors who had been followed for 14 years. We conducted a prospective, longitudinal study in 866 Runners' Association members (n = 492) and community controls (n = 374). Subjects were also categorized as Ever-Runners (n = 565) and Never-Runners (n = 301) to include runners who had stopped running. Pain was the primary outcome measure and was assessed in annual surveys on a double-anchored visual analogue scale (0 to 100; 0 = no pain). Baseline differences between Runners' Association members and community controls and between Ever-Runners versus Never-Runners were compared using chi-square and t-tests. Statistical adjustments for age, body mass index (BMI), gender, health behaviors, history of arthritis and comorbid conditions were performed using generalized estimating equations. Runner's Association members were younger (62 versus 65 years, p < 0.05), had a lower BMI (22.9 versus 24.2, p < 0.05), and less arthritis (35% versus 41%, p > 0.05) than community controls. Runners' Association members averaged far more exercise minutes per week (314 versus 123, p < 0.05) and miles run per week (26 versus 2, p < 0.05) and tended to report more fractures (53% versus 47%, p > 0.05) than controls. Ever-Runners were younger (62 versus 66 years, p < 0.05), had lower BMI (23.0 versus 24.3, p < 0.05), and less arthritis (35% versus 43%, p < 0.05) than Never-Runners. Ever-Runners averaged more exercise minutes per week (291 versus 120, p < 0.05) and miles run per week (23 versus 1, p < 0.05) and reported a few more fractures (52% versus 48%, p > 0.05) than Never-Runners. Exercise was associated with significantly lower pain scores over time in the Runners' Association group after adjusting for gender, baseline BMI, and study attrition (p < 0.01). Similar differences were observed for Ever-Runners versus Never-Runners. Consistent exercise patterns over the long term in physically active seniors are associated with about 25% less musculoskeletal pain than reported by more sedentary controls, either by calendar year or by cumulative area-under-the-curve pain over average ages of 62 to 76 years.
==== Body
Introduction
The prevalence of older adults in the United States is growing at a substantial rate. By 2030, nearly one-fifth of Americans will be in their sixties or older [1], which will have a considerable impact on public health. Numerous epidemiological and clinical studies have established that older adults who participate in regular physical activity are healthier and have a better quality of life than those who are inactive [2-4]. Regular exercise has also been shown to reduce pain in patients with knee osteoarthritis [5,6] and to help prevent mechanical low back pain [7]. In contrast, inactivity has been associated with greater pain with injury and has been associated with lower bone density and muscle tone [8]. On the other hand, some aerobic activities, such as running, have been found to result in increased risk for stress or other fractures [9,10]. Recurring trauma to soft tissue resulting from excessive physical activity conceivably could increase pain and disability [11]. Few studies have addressed the relationship between aerobic exercise and the perception of pain with advancing age.
To study the effect of exercise on disability and pain, our group [10] had investigated the relationship of running and its impact on musculoskeletal pain and disability in cohorts of Runners' Association members and community controls and Ever-Runners and Never-Runners who were followed prospectively for six years. In that study, no increase in joint pain or stiffness with age was observed in subjects who exercised often and intensely compared with their more sedentary counterparts. Pain was reduced, however, at all time points by about 25% in the exercising group. In fact, there was a slight decrease in pain for women who exercised over time.
In this investigation, we have extended that research in those cohorts. We have evaluated the association of vigorous physical activity with pain with advancing age after 14 years of follow-up. We hypothesized that those who regularly participated in running or other aerobic activity would report less musculoskeletal pain rather than more over the long term than did their inactive counterparts.
Materials and methods
Subjects
Sample selection and data collection methodology have been detailed previously [10]. Subjects were drawn from two groups: the Fifty-Plus Runners' Association with members across the United States and a Stanford University community-based random sample from the Lipid Research Clinics Study (community controls) which provided access to a sample that was similar in age to the Runners' Association. In this analysis, all subjects with at least two annual questionnaires were included. A total of 961 men and women (538 Runners' Association members and 423 community controls) who met eligibility criteria of being at least 50 years old, had at least a high school education, and used English as their primary language were initially enrolled in 1984. A major re-recruitment effort in 1991 targeted subjects who had dropped out in the first years of the study. To attenuate self-selection bias and exercise effects due to the exercisers among the community controls, we also created groups of Ever-Runners and Never-Runners based on responses to the question at baseline: "Have you ever run for exercise for a period greater than one month?" The study was approved by the Stanford University Investigational Review Board, and each subject gave their informed consent.
Data collection
Each subject completed annual, mailed health assessment questionnaires [12,13]. The questionnaire includes items on medical history, health status, exercise habits, history of musculoskeletal injuries, health care utilization, and demographic variables, such as height and weight, smoking, and alcohol use.
Assessment of physical activity
Physical activity data were obtained from responses to the question: "How many minutes each week do you exercise vigorously (vigorous exercise will cause you to sweat, and your pulse, if taken, will be above 120). Include periods of rapid walking at work and in daily activities." Subjects indicated their participation in running, jogging, swimming, bicycling/stationary bike, aerobic dance/exercises, stair steppers, brisk walking, hiking/treadmill, racket sports, and other.
Assessment of pain
Annually since 1987, pain was assessed using a visual analog scale (VAS) where 0 = no pain and 100 = worst pain. From 1987 through 1989, subjects responded to the question "How bad has pain or stiffness been in the past week?" and marked their response on the VAS. In 1987, the VAS anchors were: 0 = no pain or stiffness; and 100 = very severe pain or stiffness. In 1988 and 1989, the VAS anchors were: 0 = no pain; and 100 = severe pain. Beginning in 1990, a closed-end stem about the presence or absence of pain was added. If the patient affirmed they had pain, then they rated their amount of pain on the VAS. If they responded "No", then their pain score was assigned a value of zero.
Statistical analysis
Differences between groups at first evaluation (baseline) were compared using chi-square and t-tests. Results are reported as mean (SE) or proportion. Longitudinal data were analyzed using generalized estimating equations (GEE) [14]. Separate analyses were conducted in which repeated measurements were coded by calendar year for questionnaire response and by age. Main-effect predictors were exercise group (Runners' Association/community controls), gender, baseline age, baseline body mass index (BMI; kg/m2), years of education, number of hospital days in the past year, and dichotomous variables for smoking, and history of arthritis, fractures, and cancer at baseline. Baseline values yt = 1987 were defined as weighted means [15],
For all analyses, exercise group and gender were combined to form a four-level classification factor. Two-way interactions of each predictor with this classification factor were also included. To reduce collinearity, each continuous predictor was dichotomized about its mean. This dichotomization also produced estimates of gender and exercise group main effects that were more meaningful.
Four analytic approaches were employed to help reduce the impact of possible self-selection bias. First, we conducted a separate analysis by Ever-Runner and Never-Runner classification as well as by Runners' Association and community controls groups (from original enrollment). This grouping expanded the cases to include individuals who self-selected to run at an earlier age and who stopped running because of pain or other reasons before entering the study. Second, we used covariate adjustment to account for baseline differences between groups. Direct standardization [16] was used to produce covariate-adjusted mean VAS pain by exercise group and by study year for statistically significant predictors that were identified by the above regression analyses. Data for the community controls group from the first year that VAS pain was observed (1987) served as the reference standard. In addition to providing adjustment for differences on covariates between Runners' Association and community controls groups, use of a standard taken at baseline also permitted adjustment for possible attrition bias. Third, we assessed attrition bias (because any differential attrition between groups could result in a secondary form of self-selection bias) by performing separate analyses limited to study subjects who completed all questionnaires in addition to all subjects (completers versus all). Finally, we used a longitudinal study design in which an adverse stimulus is expected to eventually result in a poor outcome regardless of initial self-selection bias if groups differ sufficiently in exposure to the stimulus. If running creates damage through accumulated trauma, then runners with about ten-fold the amount of exposure to such trauma should have increased pain over time, and any initial differences due to self-selection should narrow as the study progresses.
Results
In 1987, the first year that VAS pain was assessed, 811 subjects returned questionnaires (458 Runners' Association members; 353 community controls). The 1991 re-recruitment efforts increased study enrollment to 881 subjects (496 Runners' Association members; 385 community controls). Fifteen subjects were excluded from these analyses because classification data for Ever-Runner versus Never-Runner were not available. Over the study duration, subject retention averaged more than 95% on an annual basis (and 98% of living subjects each year) as shown in Fig. 1. The mean (SE) years of follow up for Runners' Association and community controls were 11.4 (0.17) and 10.1 (0.22) (p < 0.05 for difference), respectively, and for Ever-Runners and Never-Runners it was 11.4 (0.16) and 10.5 (0.25) (p < 0.05 for difference), respectively. Data for this study are based on 866 subjects (492 Runners' Association and 374 community controls), who were also grouped as Ever-Runners (n = 565) and as Never-Runners (n = 301), for whom data were available.
Demographic characteristics at the beginning of the study are presented for the four study groups in Table 1. Overall, subjects were similarly well educated, but Runners' Association members and Ever-Runners were statistically younger, had lower BMI and baseline pain scores, ran more miles, exercised more minutes per week, and smoked less (all p < 0.05) relative to community controls and Never-Runners. History of arthritis was lower in Runners' Association members and Ever-Runners than community controls and Never-Runners, but statistically significant only for Ever-Runners versus Never-Runners. In quintiles of baseline exercise minutes/week, less than a fourth (22%, n = 37) of Ever-Runners and less than a tenth (7%, n = 12) of Runners' Association members were inactive, exercising less than 70 minutes a week (data not shown). In contrast, at the highest quintile, 88% (n = 155) of Runners' Association members and 91% (n = 159) of the Ever-Runners exercised between 355 and 2,119 minutes/week, indicating that subjects in both of these groups were very physically active. At the end of the study period, the groups had maintained similar levels of exercise minutes/week.
Baseline demographic characteristics for the two sets of groups by gender are shown in Table 2. For both Runners' Association members and Ever-Runners, the greater majority of subjects were male (approximately 83%), whereas in community controls and Never-Runners the sex ratios were more evenly split (56% and 50%, respectively). In females, Runners' Association members were younger, weighed less, reported less pain, fewer hospital days during the past year, ran more miles and exercised more minutes a week (all p < 0.05) than female community controls. Female Runners' Association members were also better educated, smoked less, and drank less alcohol than their community control counterparts (all p > 0.05) and there was a higher proportion of females with a history of arthritis and a lower proportion of females with a history of fractures in the community controls and Never-Runners compared to their counterparts, but these differences were statistically indistinguishable. Characteristics of male Runners' Association members versus community controls followed similar patterns, although differences in VAS pain and hospital days were not statistically significant, whereas education, smoking and alcohol were.
Pain scores over time, adjusted for group, gender and baseline BMI, are presented in Fig. 2 for each study group (Runners' Association members, community controls, Ever-Runners, and Never-Runners). The statistically significant covariates, excluding time and group, were gender (p < 0.01), baseline BMI (p < 0.01), cigarette packs/day and number of hospital days (p = 0.02). For both comparison groups of runners (Runners' Association and Ever-Runners), pain scores remained significantly lower over time (p < 0.01) when compared with community controls or Never-Runners. The dip in scores between 1987 and 1991 is a result of the rephrasing and coding of the pain question as described earlier. Pain scores were consistently about 25% less in the exercising group throughout the period of observation.
Because gender was a significant covariate, pain scores over time are presented by gender in Fig. 3, adjusted for covariates. Significant covariates, excluding time and group, are fracture in past year (p = 0.026) and the presence of arthritis (p < 0.001). As observed previously, community controls have more pain over time; however, female controls have the greatest self-reported pain, with female and male controls reporting significantly more pain than either female runners (p = 0.048) or male runners (p = 0.004). Similar results were observed for VAS pain scores by gender for Never-Runners and Ever-Runners (data not shown).
To evaluate the impact of study attrition and the possibility that withdrawal from the study might be associated with increased pain, we repeated the analyses by group and gender for study completers only. There were 61 female runners, 253 male runners, 84 female community controls and 116 male community controls who completed all questionnaires. In addition to time, group, and gender, presence of arthritis (p < 0.001) and education years (p = 0.036) were significant covariates. Similar to results of analyses using all available data, reduced levels of pain for male and female runners were observed in completers, although the only statistically significant difference over time is between female runners and male and female controls (p < 0.05).
Our final analyses tested the extent to which exercise and pain were affected by increasing age (Fig. 4). As in previous analyses, a history of arthritis and fractures were significant covariates (p < 0.001). There were significant increases in pain scores for female and male community controls and runners with increasing age, although the rates of increase are relatively modest. Older female runners tended to have the greatest beneficial impact, although these associations were statistically equivalent to differences in male runners and controls (p = 0.51).
Discussion
This paper addresses the issue as to whether consistent vigorous exercise patterns over the long term are associated with greater or reduced musculoskeletal pain. In these cohorts, runners had substantially reduced pain levels compared with controls, which persisted over average ages of 62 to 76 years. Exercise was associated with a substantial and significant reduction in pain even after adjusting for gender, baseline BMI and attrition, and despite the fact that fractures, a significant predictor of pain, were slightly more common among runners. This relationship held as well when study completers only were evaluated.
Previous studies have indicated that men and women can differ in levels of self-reported pain and its importance [17-20]. In this study, male community controls reported less pain than their female counterparts. Female community controls and Never-Runners tended to report the highest levels of pain on average, whereas female runners appeared to receive the greatest benefit in reduced pain.
This study does not provide insight into the mechanisms that might underlie these results, although we have previously ruled out self-report bias between runners and non-runners by differential validation against spousal values [21]. A trend toward more frequent reports of a history of arthritis in controls could have played a role; fractures, however, were more common in runners and should have worked in the opposite direction. Other possible mechanisms include endorphin release, exercise protection against secondary fibromyalgia, increased resistance to musculoskeletal micro-injury, psychologically based increase in pain threshold, innately high pain threshold influencing decision to exercise vigorously, or other psychological mechanisms.
Musculoskeletal pain has also been shown to be associated with disability in older individuals and individuals with some chronic diseases [22,23]. But in a study of persons with rheumatoid arthritis, Ward and Leigh [22] noted that pain was a larger contributor to measurement of overall health status than physical disability and among both older male and female individuals. Leveille and colleagues [23] evaluated the presence of pain in women over 65 years of age and observed widespread musculoskeletal pain.
Conclusion
The primary finding from this investigation is that while pain does increase with age in subjects in all study groups, there was no progressive increase in musculoskeletal pain in older adults who participated in regular vigorous exercise, including running, compared with those who did not. Initial differences favoring exercisers were shown to be maintained over time. As pain and disability are linked, our findings add to the evidence that morbidity associated with aging can be reduced by participating in regular aerobic activity.
Abbreviations
BMI = body mass index; GEE = generalized estimating equations; VAS = visual analog scale.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
BB performed statistical analyses, interpretation of data, and drafting of the manuscript; JFF participated in study design, interpretation of data, and drafting of the manuscript; DL participated in study design, statistical analyses, interpretation of data, and drafting of the manuscript. All authors have read and approved the final manuscript.
Acknowledgements
This research was supported by a grant from the National Institutes of Health (5R01-AG15815) to Stanford University, Department of Immunology/Rheumatology (JFF, Principal Investigator).
Figures and Tables
Figure 1 Sample size over time for Runners' Association members and community controls.
Figure 2 Adjusted mean visual analog scale pain scores over time by study group. For both comparison groups of runners (Runners' Association members and Ever-Runners), pain scores remained significantly lower over time (p < 0.01) when compared with community controls or Never-Runners.
Figure 3 Adjusted mean visual analog scale pain scores over time by gender and study group. After adjusting for covariates, the community controls have more pain over time; however, female controls have the greatest self-reported pain, with female and male controls reporting significantly more pain than either female runners (p = 0.0048) or male runners (p = 0.004).
Figure 4 Adjusted mean visual analog scale pain scores by age, gender and study group. Significant increases in pain scores for female and male community controls and runners were found with increasing age, although rates of increase are relatively modest. Older female runners tend to have the greatest benefit, although these associations are statistically equivalent to differences in male and female runners and controls (p = 0.51).
Table 1 Baseline characteristics of study groups
Runners' Association (n = 492) Community controls (n = 374) Runners
Ever (n = 565) Never (n = 301)
Male (%) 83 56 84 50
Mean (SE)
Age (years) 61.6 (0.25)a 65.1 (0.36) 61.7 (0.25)a 65.6 (0.41)
Education (years) 16.6 (0.11) 16.7 (0.13) 16.7 (0.10) 16.5 (0.15)
BMI (kg/m2) 22.9 (0.11)a 24.2 (0.18) 23.0 (0.11)a 24.3 (0.21)
Pain (VAS 0–100; 0 = no pain) 20.7 (1.05)a 25.7 (1.27) 20.4 (0.98)a 26.6 (1.44)
Hospital days/past year 0.27 (0.05) 0.34 (0.08) 0.24 (0.05) 0.41(0.09)
Running miles (week) 25.5 (0.65)a 2.1 (0.30) 23.1 (0.64)a 1.06 (0.29)
Exercise minutes (week) 313.6 (9.2)a 123.2 (6.8) 290.8 (8.7)a 119.9 (7.2)
Cigarette (packs/day) 0.02 (0.01)a 0.08 (0.02) 0.01 (0.01)a 0.11 (0.02)
Alcohol (oz./day) 1.08 (0.06) 1.22 (0.06) 1.11 (0.05) 1.22 (0.07)
Subjects (%) with history of
Arthritis 35 41 35* 43
Fractures 53 47 52 48
Cancer 0.002 0.005 0.004 0.003
ap < 0.05 Runners' Association versus community controls or Ever-Runners versus Never-Runners. BMI, body mass index; VAS, visual analog scale.
Table 2 Baseline characteristics by study group and gender
Runners' Association Community Controls Ever-Runners Never-Runners
Female (n = 82) Male (n = 410) Female (n = 163) Male (n = 211) Female (n = 93) Male (n = 472) Female (n = 152) Male (n = 149)
Mean (SE)
Age (years) 59.9 (0.5)a 62.0 (0.3)a 65.5 (0.5) 64.6 (0.5) 60.3 (0.5)a 62.0 (0.3)a 65.7 (0.6) 65.5 (0.6)
Education (years) 16.3 (0.3) 16.6 (0.1)a 15.8 (0.2) 17.5 (0.2) 16.4 (0.3)a 16.8 (0.1) 15.7 (0.2) 17.3 (0.2)
BMI (kg/m2) 21.6 (0.2)a 23.2 (0.1)a 23.2 (0.3) 25.0 (0.2) 21.6 (0.2)a 23.3 (0.1)a 23.4 (0.3) 25.3 (0.3)
Pain (VAS 0–100; 0 = no pain) 20.9 (2.6)a 20.7 (1.2) 27.6 (2.1) 23.0 (1.6) 19.6 (2.4)a 20.5 (1.1) 28.8 (2.1) 24.3 (1.9)
Hospital days/past year 0.09 (0.1)a 0.31 (0.1) 0.45 (0.1) 0.25 (0.1) 0.08 (0.0)a 0.28 (0.1) 0.48 (0.2) 0.33 (0.1)
Running miles (week) 19.4 (1.3)a 26.9 (0.7)a 0.99 (0.4) 3.0 (0.4) 17.4 (1.34)a 24.3 (0.7)a 0.9 (0.4) 1.3 (0.4)
Exercise minutes (week) 310.8 (19.8)a 314.2 (10.3)a 115.8 (10.0) 129.0 (9.1) 283.9 (18.9)a 292.2 (9.7)a 118.2 (10.8) 121.6 (9.4)
Cigarette (packs/day) 0.09 (0.04) 0.01 (0.04)a 0.13 (0.0) 0.05 (0.0) 0.05 (0.0)a 0.01 (0.0)a 0.16 (0.0) 0.06 (0.0)
Alcohol (oz./day) 1.0 (0.1) 1.1 (0.1)a 1.0 (0.8) 1.4 (0.1) 0.98 (0.1) 1.14 (0.1)a 1.01 (0.1) 1.46 (0.1)
Subjects (%) with history of
Arthritis 45 33 47 36 42 33 49 36
Fractures 54 53 41 52 51 52 42 54
Cancer 0 0.002 0 0.009 0 0.004 0 0.007
ap < 0.05 male or female Runners' Association versus matched community controls or male or female Ever-Runners versus matched gender Never-Runners. BMI, body mass index; VAS, visual analog scale.
==== Refs
Centers for Disease Control and Prevention (CDC) Trends in aging: United States and worldwide MMWR Morb Mortal Wkly Rep 2003 52 101 104 12645839
Centers for Disease Control The State of Aging and Health in America 2004 Washington: Merck Institute of Aging and Health
Macera CA Hootman JM Sniezek JE Major public health benefits of physical activity Arthritis Rheum 2003 49 122 128 12579603 10.1002/art.10907
Pate RR Pratt M Blair SN Haskell WL Macera CA Bouchard C Buchner D Ettinger W Heath GW King AC 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
Brady TJ Kruger J Helmick CG Callahan LF Boutaugh ML Intervention programs for arthritis and other rheumatic diseases Health Educ Behav 2003 30 44 63 12564667 10.1177/1090198102239258
Ettinger WH JrBurns R Messier SP Applegate W Rejeski WJ Morgan T Shumaker S Berry MJ O'Toole M Monu J Craven T A randomized trial comparing aerobic exercise and resistance exercise with a health education program in older adults with knee osteoarthritis. The Fitness Arthritis and Seniors Trial (FAST) JAMA 1997 277 25 31 8980206 10.1001/jama.277.1.25
Vuori I Exercise and physical health: musculoskeletal health and functional capabilities Res Q Exerc Sport 1995 66 276 285 8775582
Taimela S Diederich C Hubsch M Heinricy M The role of physical exercise and inactivity in pain recurrence and absenteeism from work after active outpatient rehabilitation for recurrent or chronic low back pain: a follow-up study Spine 2000 25 1809 1816 10888950 10.1097/00007632-200007150-00012
Daffner RH Martinez S Gehweiler JA Stress fractures in runners JAMA 1982 247 1039 1041 7057582 10.1001/jama.247.7.1039
Fries JF Singh G Morfeld D O'Driscoll P Hubert H Relationship of running to musculoskeletal pain with age. A six-year longitudinal study Arthritis Rheum 1996 39 64 72 8546740
McAlindon TE Cooper C Kirwan JR Dieppe PA Determinants of disability in osteoarthritis of the knee Ann Rheum Dis 1993 52 258 262 8484690
Fries JF Spitz P Kraines RG Holman HR Measurement of patient outcome in arthritis Arthritis Rheum 1980 23 137 145 7362664
Bruce B Fries J The Stanford health assessment questionnaire (HAQ): a review of its history, issues, progress, and documentation J Rheumatol 2003 30 167 178 12508408
Liang KY Zeger SL Longitudinal data analysis using generalized linear models Biometrika 1986 73 13 22
Sen A Srivastava M Regression Analysis: Theory, Method and Application 1990 New York: Springer-Verlag
Rosner B Fundamentals of Biostatistics 1995 Belmont, CA: Wadsworth Publishing Co
Herr KA Garand L Assessment and measurement of pain in older adults Clin Geriatr Med 2001 17 457 478 11459715 10.1016/S0749-0690(05)70080-X
Kelly AM Does the clinically significant difference in visual analog scale pain scores vary with gender, age, or cause of pain? Acad Emerg Med 1998 5 1086 1090 9835471
Unruh AM Gender variations in clinical pain experience Pain 1996 65 123 167 8826503 10.1016/0304-3959(95)00214-6
Unruh AM Ritchie J Merskey H Does gender affect appraisal of pain and pain coping strategies? Clin J Pain 1999 15 31 40 10206565 10.1097/00002508-199903000-00006
Fries JF Singh G Morfeld D O'Driscoll P Hubert H Relationship of running to musculoskeletal pain with age. A six-year longitudinal study Arthritis Rheum 1996 39 64 72 8546740
Ward MM Leigh JP The relative importance of pain and functional disability to patients with rheumatoid arthritis J Rheumatol 1993 20 1494 1499 8164204
Leveille SG Ling S Hochberg MC Resnick HE Bandeen-Roche KJ Won A Guralnik JM Widespread musculoskeletal pain and the progression of disability in older disabled women Ann Intern Med 2001 135 1038 1046 11747382
|
16277679
|
PMC1297575
|
CC BY
|
2021-01-04 16:02:40
|
no
|
Arthritis Res Ther. 2005 Sep 19; 7(6):R1263-R1270
|
utf-8
|
Arthritis Res Ther
| 2,005 |
10.1186/ar1825
|
oa_comm
|
==== Front
Arthritis Res TherArthritis Research & Therapy1478-63541478-6362BioMed Central London ar18261627768110.1186/ar1826Research ArticleOpen label phase II trial of single, ascending doses of MRA in Caucasian children with severe systemic juvenile idiopathic arthritis: proof of principle of the efficacy of IL-6 receptor blockade in this type of arthritis and demonstration of prolonged clinical improvement Woo Patricia [email protected] Nicholas [email protected] Anne-Marie [email protected] Taunton [email protected] Valentina [email protected] Polly [email protected] Helena [email protected] David [email protected] Tadamitsu [email protected] Great Ormond Street Hospital, NHS Trust, London, UK2 Hopital Necker Enfants Malades, Paris, France3 University of Birmingham and Birmingham Children's Hospital, NHS Trust, Birmingham, UK4 Chugai Pharma Europe Ltd, London, UK5 Laboratory of Immune Regulation, Graduate School of Frontier Biosciences, Osaka University, Osaka, Japan2005 15 9 2005 7 6 R1281 R1288 13 5 2005 15 6 2005 3 8 2005 24 8 2005 Copyright © 2005 Woo 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.
Eighteen Caucasian (white, Middle East and Asian) children diagnosed by paediatric rheumatologists in the UK and France as having systemic juvenile idiopathic arthritis (sJIA) were enrolled in this open label, single dose trial. All patients had evidence of continued symptoms and disease activity for at least three months while receiving >0.2 mg/kg/day of prednisolone, or its equivalent, prior to recruitment. Twelve patients also received methotrexate (≤20 mg/m2/week). The patients were divided into three groups receiving 2, 4 or 8 mg/kg of MRA (tocilizumab) by intravenous infusion. No evidence of dose-limiting toxicity was observed and there were no dose-limiting safety issues. MRA appeared to be dramatically effective, with clinical and laboratory responses observed by 48 h post infusion, and these improvements continued well after serum MRA was undetectable. Eleven patients achieved the JIA definition of improvement (at least 3 of 6 core set criteria with a 30% improvement and no more than one worsened by 30%) and eight achieved ≥50% improvement. There were no observable differences with age. Clinical improvement in these children was observed for up to eight weeks, supporting the hypothesis that IL-6 is a key cytokine in the upregulation of genes crucial in the inflammation processes of sJIA, and the possibility of sequestration of MRA in the extra-vascular compartment needs to be considered.
==== Body
Introduction
Juvenile idiopathic arthritis (JIA) is a heterogeneous group of persistent arthritides of unknown origin occurring before 16 years of age [1]. One subset, systemic JIA (sJIA), is defined by the additional presence of debilitating fever, evanescent rash, hepatosplenomegaly, lymphadenopathy and serositis. Severe complications include osteoporosis, growth retardation, systemic amyloidosis and macrophage activation syndrome and are observed more frequently in patients with long-standing disease than in other JIA subsets. Patients with sJIA have a range of other prominent features, including marked elevation of erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP), leucocytosis with high neutrophil counts and thrombocytosis [2]. Ferritin concentrations are high and correlate with systemic disease activity [2,3]. Anaemia is microcytic and characterised by a marked defect in iron supply for erythropoiesis [4]. Virtually all children with sJIA are negative for antinuclear antibodies and rheumatoid factor [2,4]. The mean duration of active disease is five to six years in Caucasians, with disabling polyarthritis becoming prominent in up to 50% of patients, while the systemic features usually regress within three to four years [5]. Twenty three percent of these patients have poor outcomes as adults [6].
The polyarthritis of sJIA is often extraordinarily resistant to treatment. Steroids are used to control systemic symptoms, but do not alter long-term prognosis and are associated with severe side effects such as osteoporosis and growth failure, to which patients with sJIA appear particularly susceptible. Methotrexate, effective in other forms of paediatric polyarthritis, appears less effective in sJIA [7]. Etanercept has been shown to have limited efficacy, with a possible increased risk of macrophage activation syndrome [8,9]. There is no strong evidence of definite improvements with the use of cyclosporine, azathioprine, or cyclophosphamide [2]. Chlorambucil has been used to treat amyloidosis with an improvement in survival by inducing remission [2,10]: but, as with cyclophosphamide, there are concerns over its long-term use due to side effects [10]. Pilot data on anakinra suggest blocking IL-1 can be effective [11,12], but further placebo controlled/comparative trials are necessary. Moreover, as anakinra is a daily subcutaneous injection, which is well known to cause some discomfort and pain, it may not be widely tolerated by children. There is, therefore, a need for a more effective treatment for sJIA, based on an understanding of the underlying pathophysiology of the disease process that may alleviate the systemic features of sJIA, as well as prevent progression of joint damage.
There is a strong body of evidence that IL-6 production is particularly high in sJIA and that this is genetically determined by a variant of the gene encoding IL-6 in a significant proportion of patients [13,14]. In addition to the increase in serum IL-6, it has been found that there is a significant increase in soluble IL-6 receptor (sIL-6R) concentrations [15]. The large quantities of serum IL-6 present in IL-6/sIL-6R complexes have been found to be of particular biological relevance in vivo [16]. IL-6 levels correlate with disease activity, fever pattern and platelet counts, indicating an important role for IL-6 in the pathogenesis of sJIA [17-19]. Components of the IL-6 signalling pathway are, therefore, rational targets for the treatment of sJIA.
MRA (tocilizumab) is a humanized anti-human IL-6 receptor antibody of kappa-IgG1 subclass, developed collaboratively by Osaka University and Chugai Pharmaceutical Company Ltd (Japan). MRA is humanized by the grafting of complementarity determining regions of a mouse anti-human IL-6 receptor monoclonal antibody onto human IgG1 by recombinant DNA technology. It has been shown to compete for both the membrane-bound and soluble forms of human IL-6 receptor, thus inhibiting the binding of IL-6 to its receptor and its pro-inflammatory activity [20]. As IL-6 is postulated to play a pivotal role in the pathogenesis and disease activity of sJIA, MRA is being developed in this indication. An anti-IL-6 receptor monoclonal antibody, such as MRA, may have an important role to play, not only in relieving the patient of the systemic features of the disease, but also in preventing the progression of joint damage and reducing extra-articular complications of the disease.
The primary objectives of this trial were to determine the safety and tolerability of single, ascending doses of MRA in children with sJIA and to determine the pharmacokinetic profile of a range of single intravenous doses of MRA in children with sJIA, as recommended by the Medicines and Healthcare products Regulatory Agency, UK. The secondary objectives were to evaluate the efficacy of the range of single intravenous doses and to explore the potential for an antigenic response.
Materials and methods
Eighteen children (white, Middle East and Asian Caucasians), diagnosed with sJIA and having active disease for at least three months while receiving greater than 0.2 mg/kg/day of prednisolone or its equivalent, were enrolled from the UK and France. Diagnosis was defined using the International League of Associations for Rheumatology criteria [1]. Ethics committee approval and written informed consent were obtained. All patients had at least one active systemic feature, one active joint and the doses of nonsteroidal anti-inflammatory drugs and steroids were unchanged for at least two weeks before the infusion of MRA. Concurrent methotrexate therapy was allowed to a maximum of 20 mg/m2 weekly, stabilized for at least four weeks prior to administration of the study medication. Patients receiving other disease-modifying anti-rheumatic drugs went through a washout period (at least five half-lives of the drug) before inclusion. Children with a history of macrophage activation syndrome were excluded.
Patients were divided equally within two age ranges (2 to 5 and 6 to 18 years) to receive 2, 4 or 8 mg/kg of MRA by intravenous infusion. A safety review was conducted before progression to the next dose range. Clinical assessments were carried out before infusion, at 48 h post-infusion and weekly thereafter. The duration of follow up for the 2, 4 and 8 mg/kg groups was 4, 6 and 8 weeks, respectively. Clinical assessment was performed using the JIA core set criteria (parent global visual analogue scale (VAS), physician's global VAS, number of active joints, number of joints with limited range, the childhood health assessment questionnaire (CHAQ) and ESR) and the definition of improvement as defined by international consensus (3 or more core set criteria improved by 30% and no more than 1 core set criterion worsened by 30%) [21]. Improvements of 50% and 70% were defined as above, substituting 30 with 50 and 70, respectively. Systemic features were recorded, as previously described [7]. The Systemic Feature Score included an assessment of fever, rash, cervical lymphadenopathy, axillary lymphadenopathy, inguinal lymphadenopathy, hepatomegaly, splenomegaly and clinical evidence of serositis (pericarditis, pleuritis or peritonitis). In addition to any occurrence of fever at the assessment visit, the parents/caregivers were asked about the presence of fever in the 24 h period preceding the assessment visit as part of the Systemic Feature Score. The overall Systemic Feature Score could range from 0 to 8 and is the sum of the eight individual features, which were scored as either absent (0 points) or present (1 point).
Laboratory measurements included full blood count, liver (gamma-glutamyl transpeptodase (GGT), aspartate aminotransferace (AST), alanine aminotransferase (ALT), bilirubin) and kidney (urea, serum creatinine, albumin) function tests, serum CRP and ESR. These were performed in the accredited routine laboratories of the participating hospitals. Levels of serum MRA, IL-6, sIL-6R and anti-MRA antibodies were screened during the follow up period by contract laboratories of the sponsor. The sponsor of the trial was Chugai Pharma Europe Ltd. Due to the exploratory nature of the study, no formal sample size calculations were performed. The sample size was based on clinical judgment and practical considerations. No formal statistical analysis of any data was performed.
Results
Safety data
The baseline demographic data for the 18 patients enrolled is shown in Table 1. Fifteen patients reported 59 treatment-emergent adverse events and the majority were mild. The majority of events reported were classified as gastrointestinal disorders, abnormal investigations (including increases in hepatic enzymes), respiratory disorders and infection/infestations. Four patients reported nine drug-related adverse events. No serious bacterial infections were documented. One patient developed urticaria within 4 h of the end of the 2 mg/kg MRA infusion, lasting two weeks. Transient rises of ALT occurred in three patients who also received methotrexate, but all had above normal values of GGT and/or AST, lactic dehydrogenase (LDH), and ALT at screening, but were below the levels stated in the exclusion criteria. All but three patients had at least one low lymphocyte value, generally at one and two weeks after infusion; however, eight patients (44%) had lymphocyte counts below the local laboratories' normal ranges, (but above the level stated in the exclusion criteria) before infusion of MRA.
There were no withdrawals due to adverse events. Any adverse events were reported up to the end of the follow up period. Five serious adverse events were reported in four patients: chicken pox in one patient, contracted during an outbreak at his nursery; transient pancytopenia in one patient at week seven; disease flares requiring hospitalisation occurred in two patients after three and six weeks (one of these reported a second adverse event; a herpes simplex mouth ulcer that had previously occurred during flares of disease). There were two flares of sJIA (in patients that received 4 and 8 mg/kg MRA) that were considered to be possibly drug-related; one developed at two and the other at six weeks post-infusion.
Efficacy data
Fifteen patients were included in the efficacy analysis (2 mg/kg (n = 4), 4 mg/kg (n = 6), 8 mg/kg (n = 5)); three patients were excluded due to protocol violations. At baseline, the systemic feature scores were generally comparable across the three treatment groups (Table 2 shows the baseline values for all 18 patients). There was considerable variability in baseline values for the physician's global VAS, total number of active joints, total number of joints with limitation of movement and ESR, with the 8 mg/kg treatment group appearing to have a less active disease state at baseline when assessed according to these parameters.
A marked clinical response was noted at 48 h in all patients and in all three dose groups. Clinical improvement, as defined by Giannini et al. [21] (30% in at least 3 core sets and no more than one worsened by 30%), was achieved at week one in 11 patients. The clinical response in the 4 mg/kg group (n = 6) was the most dramatic, with four and two patients achieving 50% and 70% improvements, respectively, at week 1 (Fig. 1). The effect appeared to be more prolonged in the 4 and 8 mg/kg groups: four and three patients achieved 50% and 70% improvements, respectively, at week six. There was a trend towards a reduction in the mean number of joints with active disease/limited movement, although these parameters showed less of an improvement compared to the other core outcome variables. Moreover, parents reported dramatic changes in their child in terms of mobility and well-being (as measured by Parent's/Caregiver's or Patient's VAS of overall well-being (0 mm = very well, 100 mm = very poor). Age did not affect this response. The Total Systemic feature score also decreased within one week post infusion in all dose groups.
Clinical improvement was paralleled by a decrease in CRP levels (Fig. 2). There was a transient increase of mean CRP after weeks one or two, but mean levels were lower than baseline for the remainder of the follow up period (Fig. 3). The same profile was observed for ESR, but with a small time delay in response, the maximum decrease being observed at week one. There were also improvements in the haemoglobin, total white counts and serum albumin, with the peak improvements at one week after infusion. Interestingly, only three children required intervention with rescue therapy, such as prednisolone, even though CRP values tended to return to baseline. The treatment was not modified during the follow up period, except for two patients who required intravenous methylprednisolone at week six (MRA 4 mg/kg, and 8 mg/kg) and one requiring an increase in oral steroids from 5 to 10 mg 24 h after infusion (MRA 2 mg/kg). The clinical and biological improvements were maintained after MRA was undetectable in the serum.
Serum IL-6 and sIL-6R increased for two to four weeks after dosing, with both the magnitude and duration of this effect being dose-related. Following this initial increase, IL-6 levels were observed to decrease for the 4 and 8 mg/kg dose groups.
Discussion
Single dosing with MRA in these 18 European Caucasian children does not appear to cause any drug related serious infections during the infusion and follow-up period. A report of multiple dosing in a group of Japanese children with a similar disease did not reveal serious infections as a significant side effect over one year [22].
An increased incidence of infection, including serious life threatening infections, has been observed with other monoclonal antibodies active against components of the immune system. In this study, infections were generally mild in intensity and the incidence of infections did not increase at the higher dose levels. Only the herpes simplex mouth ulcer during a disease flare three weeks post infusion was considered to be possibly related to MRA, although herpes simplex mouth ulcer has been reported previously during a disease flare of this patient. Of the nine infections reported, six were associated with a lymphocyte count below the lower limit of normal; however, lymphopenia was present in the majority of patients following dosing and nine patients with low lymphocyte counts did not have a corresponding infection. Further studies will need to address the longer term safety of MRA in a larger patient population, as this pilot study was to prove the safety of the therapy in a single dose setting.
MRA appears to be dramatically effective clinically in children with sJIA following a single infusion. Eleven patients achieved a clinical improvement – a 30% improvement in the composite score for disease activity in JIA (as defined by Giannini et al. [21]) – with maximum effect seen at one week for all the dose ranges. Eight and three patients, respectively, achieved 50% and 70% improvements in the composite score of disease activity mostly in the higher dose groups. The maximum duration of benefit was within four to eight weeks after dosing with all efficacy endpoints. The serum levels of the biological markers of the acute phase response (CRP and ESR) also showed dramatic decrease from baseline; many decreased 10-fold and, in a few patients, the serum CRP normalised. The prolonged effect of a single dose supports the hypothesis that IL-6 is a key cytokine in the upregulation of genes crucial in the inflammation processes of sJIA. This effect is in sharp contrast to the experience with single infusions of 30 mg/kg of intravenous methyl prednisolone that, given previously in these children, all resulted in transient and smaller clinical and laboratory improvements, lasting a maximum of 48 h before return to baseline. Response to MRA also compares well with anti-tumour necrosis factor therapy. Two infusions of 10 mg/kg of infliximab one week apart in a 16 year old sJIA patient led to some beneficial effect on the fever for only five days and no change in the joint symptoms [23]. Subsequent surveys of response rates to anti-tumour necrosis factor therapy have also shown that, at the most, there is significant improvement in only 30% [8,9]. Blockade of IL-6 signalling within inflammatory and immune cells could lead to secondary changes that may account for the prolonged well-being experienced by many patients in this small open trial. Another possible explanation of this prolonged clinical improvement in symptoms is that MRA may remain longer in the extravascular compartment; this possibility will need to be explored. Further controlled studies will need to be done, with multiple dosing, to assess the short and longer term efficacy and safety of MRA.
A recent report by Yokota et al. [22] using clinical and laboratory parameters similar to this study showed a similar and marked response in Japanese patients with sJIA to multiple dosing with MRA. The disease known as sJIA in Japan is very similar to that reported in Caucasians, but not enough is known about the phenotypes and genotypes to make a direct extrapolation of drug response. So this study provides further evidence in Caucasians that IL-6 is a key cytokine in this disease and blockade of its signalling by a monoclonal antibody produces dramatic and prolonged clinical and laboratory improvements.
Conclusion
MRA given as a single intravenous infusion to severe sJIA patients was tolerated over the 2 to 8 mg/kg dose range studied. There was no evidence of dose-limiting toxicity and no dose-limiting safety issues were observed at any of the dose levels administered during the study.
Evidence of efficacy was observed at all dose levels studied, both in clinical and biological markers. The efficacy response appeared to be more limited at the lowest dose level (2 mg/kg). The overall clinical response appeared to continue after MRA was undetectable in serum and gradually decreased over a four to six week period following the infusion of MRA. These promising results with MRA in sJIA, a disease with an unmet medical need, will be further developed in phase III trials.
Abbreviations
ACR = American College of Rheumatology; ALT = alanine aminotransferase; AST = aspartate aminotransferace; CHAQ = childhood health assessment questionnaire; CRP = C-reactive protein; ESR = erythrocyte sedimentation rate; GGT = gamma-glutamyl transpeptodase; IL = interleukin; JIA = juvenile idiopathic arthritis; LDH = lactic dehydrogenase; sIL-6R = soluble IL-6 receptor; sJIA = systemic juvenile idiopathic arthritis; VAS = visual analogue scale.
Competing interests
The trial was sponsored by Chugai Pharma Europe Ltd. The authors would like to declare the following possible competing interests. NW, PL, HW and VL received funding from Chugai towards the cost of attending the PReS Scientific Meeting in Stresa during 2003. DT is a full-time employee of Chugai Pharma Europe Ltd.
Authors' contributions
All authors contributed to and approved the final draft of this manuscript. PW had significant input into the study design; ethics committee submissions and protocol amendments; supervised the conduct of the study on patients at The Great Ormond Street Hospital, London; followed 11 patients that were entered; was the chief investigator of the study; reviewed the results and drafted and finalized the submitted manuscript. NW recruited, assessed and managed the ongoing care of patients; contributed to study amendments; was involved in data analysis; presented early outcome data to an international conference. A-MP entered and followed six patients through the study; participated in the discussion of the results. TS participated in the data gathering and verification; discussion of the data; drafting of the manuscript. VL participated in the patient selection process with TS; medical management and monitoring including dosing and follow-up visits; data collection and processing. PL participated in the patient selection process with PW and NW; medical management and monitoring including MRA dosing and follow-up visits; data collection and processing. HW participated in the data collection and processing of this study. DT participated in the supervision of the clinical and data management aspects of this study as Medical Director at Chugai Pharma Europe Ltd. TK discovered IL-6; developed anti-IL-6 receptor therapies for Castleman's disease and rheumatoid arthritis; collaborated with PW to obtain MRA for a trial in Caucasian children with sJIA.
Acknowledgements
We wish to thank our patients, the Arthritis Research Campaign, UK, for HW’s support, Professors Yokota and Nishimoto for sharing their preliminary results on repeated administration of MRA in two Japanese patients at the beginning of this trial, and Dr Didier Halimi for his help in the preparation of the manuscript.
Figures and Tables
Figure 1 Cumulative number of patients improving from baseline in juvenile idiopathic arthritis core outcome variables (dotted lines show non-improvers); ACR, American College of Rheumatology.
Figure 2 Serum C-reactive protein (CRP; mg/l) and erythrocyte sedimentation rate (ESR) of individual patients for all dose groups from day 0, pre-infusion, until the end of the follow-up period for each dose group.
Figure 3 Mean changes from baseline (day 0, pre-infusion) in C-reactive protein (CRP; mg/L).
Table 1 Patient demographics for the safety population
MRA
2 mg/kg (n = 6) 4 mg/kg (n = 6) 8 mg/kg (n = 6)
Age (years)
Mean ± SD 6.3 ± 3.33 7.7 ± 5.13 4.8 ± 1.83
Median 6.5 6.5 5.0
Range 3, 10 3, 17 2, 7
Gender
Male (n) 3 6 1
Female (n) 3 0 5
Race
Caucasian (n) 6 6 6
Weight (kg)
Mean ± SD 23.47 ± 10.954 27.45 ± 14.479 19.63 ± 4.286
Median 17.95 24.05 19.90
Range 13.8, 38.0 16.2, 55.5 14.6, 23.6
Height (cm)
Mean ± SD 108.3 ± 14.07 117.5 ± 23.84 108 ± 12.92
Median 105.5 111.5 102.0
Range 90, 126 95, 159 94, 127
SD, standard deviation.
Table 2 Baseline characteristics of sJIA for the full analysis population
MRA
2 mg/kg (n = 4) 4 mg/kg (n = 6) 8 mg/kg (n = 5) Overall (n = 15)
VAS of disease activitya (mm)
Mean ± SD 60.3 ± 29.94 50.5 ± 22.10 35.3 ± 30.09 48.9 ± 26.49
Range 20, 92 17, 74 11, 73f 11, 92g
VAS of overall well-beinga (mm)
Mean ± SD 42.8 ± 29.28 56.5 ± 34.85 59.2 ± 10.57 53.7 ± 26.42
Range 5,76 20, 100 50, 75 5, 100
Total number of active jointsb
Mean ± SD 27.8 ± 9.54 23.2 ± 13.01 12.4 ± 13.15 20.8 ± 13.07
Range 15, 36 4, 38 4, 35 4, 38
Total number of joints with limitation of movementb
Mean ± SD 17.8 ± 10.28 15.2 ± 11.27 12.8 ± 13.10 15.1 ± 11.00
Range 7, 30 5, 35 5, 36 5, 36
ESR (mm/h)
Mean ± SD 114.3 ± 20.45 78.0 ± 25.85 60.0 ± 18.13f 83.2 ± 30.07g
Range 95, 140 30, 100 35, 78 30, 140
CRP (mg/l)
Mean ± SD 140.8 ± 90.08 119.0 ± 52.23 125.6 ± 39.48 NC
Range 59, 262 60, 189 89, 178
Functional abilityc
Mean ± SD 0.7 ± 0.68 1.2 ± 0.89 2.3 ± 0.36 1.4 ± 0.91
Range 0.03, 1.5 0.13, 2.38 1.75, 2.75 0.03, 2.75
Total Systemic Feature Scored
Mean ± SD 3.0 ± 1.41 3.2 ± 1.72 3.0 ± 2.45 3.1 ± 1.79
Range 2, 5 1, 6 0, 6 0, 6
aMeasured on an increasing scale of 0 to 100 mm. bMaximum of 81. cCHAQ: maximum of 3. dMaximum of 8. eN = 2;individual patient's data presented. fN = 4. gN = 14. CHAQ, childhood health assessment questionnaire; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; NC, not calculated; SD, standard deviation; VAS, visual analogue scale.
==== Refs
Petty RE Southwood TR Manners P Baum J Glass DN Goldenberg J He X Maldonado-Cocco J Orozco-Alcala J Prieur AM International League of Associations for Rheumatology classification of juvenile idiopathic arthritis: second revision, Edmonton, 2001 J Rheumatol 2004 31 390 392 14760812
Laxer RM Schneider R Isenberg DA, Maddison PJ, Woo P, Glass DN, Breedveld FC Systemic-onset juvenile chronic arthritis Oxford Textbook of Rheumatology 2004 Oxford: Oxford University Press 798 810
Pelkonen P Swanljung K Siimes MA Ferritinemia as an indicator of systemic disease activity in children with systemic juvenile rheumatoid arthritis Acta Paediatr Scand 1986 75 64 68 3953279
Cazzola M Ponchio L de Benedetti F Ravelli A Rosti V Beguin Y Invernizzi R Barosi G Martini A Defective iron supply for erythropoiesis and adequate endogenous erythropoietin production in the anemia associated with systemic-onset juvenile chronic arthritis Blood 1996 87 4824 4830 8639855
Woo P Wedderburn LR Juvenile chronic arthritis Lancet 1998 351 969 973 9734957
Spiegel LR Schneider R Lang BA Birdi N Silverman ED Laxer RM Stephens D Feldman BM Early predictors of poor functional outcome in systemic-onset juvenile rheumatoid arthritis: a multicenter cohort study Arthritis Rheum 2000 43 2402 2409 11083261 10.1002/1529-0131(200011)43:11<2402::AID-ANR5>3.0.CO;2-C
Woo P Southwood TR Prieur AM Dore CJ Grainger J David J Ryder C Hasson N Hall A Lemelle I Randomized, placebo-controlled, crossover trial of low-dose oral methotrexate in children with extended oligoarticular or systemic arthritis Arthritis Rheum 2000 43 1849 1857 10943876 10.1002/1529-0131(200008)43:8<1849::AID-ANR22>3.0.CO;2-F
Quartier P Taupin P Bourdeaut F Lemelle I Pillet P Bost M Sibilia J Kone-Paut I Gandon-Laloum S LeBideau M Efficacy of etanercept for the treatment of juvenile idiopathic arthritis according to the onset type Arthritis Rheum 2003 48 1093 1101 12687553 10.1002/art.10885
Ramanan AV Schneider R Macrophage activation syndrome following initiation of etanercept in a child with systemic onset juvenile rheumatoid arthritis J Rheumatol 2003 30 401 403 12563702
Savolainen HA Chlorambucil in severe juvenile chronic arthritis: longterm followup with special reference to amyloidosis J Rheumatol 1999 26 898 903 10229413
Irigoyen PI Olson J Hom C Ilowite NT Treatment of systemic onset juvenile rheumatoid arthritis with anakinra [abstract] Arthritis Rheum 2004 50 S437 S438
Henrickson M Efficacy of anakinra in refractory systemic arthritis [abstract] Arthritis Rheum 2004 50 S438
Ogilvie EM Fife MS Thompson SD Twine N Tsoras M Moroldo M Fisher SA Lewis CM Prieur AM Glass DN Woo P The -174G allele of the interleukin-6 gene confers susceptibility to systemic arthritis in children: a multicenter study using simplex and multiplex juvenile idiopathic arthritis families Arthritis Rheum 2003 48 3202 3206 14613283 10.1002/art.11300
Fishman D Faulds G Jeffery R Mohamed-Ali V Yudkin JS Humphries S Woo P The effect of novel polymorphisms in the interleukin-6 (IL-6) gene on IL-6 transcription and plasma IL-6 levels, and an association with systemic-onset juvenile chronic arthritis J Clin Invest 1998 102 1369 1376 9769329
Keul R Heinrich PC Muller-newen G Muller K Woo P A possible role for soluble IL-6 receptor in the pathogenesis of systemic onset juvenile chronic arthritis Cytokine 1998 10 729 734 9770335 10.1006/cyto.1997.0343
de Benedetti F Massa M Pignatti P Albani S Novick D Martini A Serum soluble interleukin 6 (IL-6) receptor and IL-6/soluble IL-6 receptor complex in systemic juvenile rheumatoid arthritis J Clin Invest 1994 93 2114 2119 8182142
de Benedetti F Massa M Robbioni P Ravelli A Burgio GR Martini A Correlation of serum interleukin-6 levels with joint involvement and thrombocytosis in systemic juvenile rheumatoid arthritis Arthritis Rheum 1991 34 1158 1163 1930333
Rooney M David J Symons J Di Giovine F Varsani H Woo P Inflammatory cytokine responses in juvenile chronic arthritis Br J Rheumatol 1995 34 454 460 7788176
Prieur AM Roux-Lombard P Dayer JM Dynamics of fever and the cytokine network in systemic juvenile arthritis Rev Rhum Engl Ed 1996 63 163 170 8731233
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
Giannini EH Ruperto N Ravelli A Lovell DJ Felson DT Martini A Preliminary definition of improvement in juvenile arthritis Arthritis Rheum 1997 40 1202 1209 9214419
Yokota S Miyamae T Imagawa T Iwata N Katakura S Mori M Woo P Nishimoto N Yoshizaki K Kishimoto T Therapeutic efficacy of humanized recombinant anti-interleukin-6 receptor antibody in children with systemic-onset juvenile idiopathic arthritis Arthritis Rheum 2005 52 818 825 15751095 10.1002/art.20944
Elliott MJ Woo P Charles P Long-Fox A Woody JN Maini RN Suppression of fever and the acute-phase response in a patient with juvenile chronic arthritis treated with monoclonal antibody to tumour necrosis factor-alpha (cA2) Br J Rheumatol 1997 36 589 593 9189062 10.1093/rheumatology/36.5.589
|
16277681
|
PMC1297576
|
CC BY
|
2021-01-04 16:02:39
|
no
|
Arthritis Res Ther. 2005 Sep 15; 7(6):R1281-R1288
|
utf-8
|
Arthritis Res Ther
| 2,005 |
10.1186/ar1826
|
oa_comm
|
==== Front
Arthritis Res TherArthritis Research & Therapy1478-63541478-6362BioMed Central London ar18271627768410.1186/ar1827Research ArticleTranscriptional profiles discriminate bone marrow-derived and synovium-derived mesenchymal stem cells Djouad Farida [email protected] Claire [email protected]äupl Thomas [email protected]é Gilles [email protected] Najiba [email protected] Pascale [email protected] Florence [email protected] François [email protected]ème Thierry [email protected] Jacques [email protected] Christian [email protected]ël Danièle [email protected] INSERM Unit 475, Montpellier, France2 Rheumatology, Charité Hospital, Berlin, Germany3 CNRS, UMR 5124, Montpellier, France4 Hormonal Biology Laboratory, St Vincent de Paul Hospital, Paris, France5 Immuno-Rhumatologie, Lapeyronie Hospital, Montpellier, France2005 20 9 2005 7 6 R1304 R1315 12 5 2005 18 7 2005 26 7 2005 24 8 2005 Copyright © 2005 Djouad 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.
Previous studies have reported that mesenchymal stem cells (MSC) may be isolated from the synovial membrane by the same protocol as that used for synovial fibroblast cultivation, suggesting that MSC correspond to a subset of the adherent cell population, as MSC from the stromal compartment of the bone marrow (BM). The aims of the present study were, first, to better characterize the MSC derived from the synovial membrane and, second, to compare systematically, in parallel, the MSC-containing cell populations isolated from BM and those derived from the synovium, using quantitative assays. Fluorescent-activated cell sorting analysis revealed that both populations were negative for CD14, CD34 and CD45 expression and that both displayed equal levels of CD44, CD73, CD90 and CD105, a phenotype currently known to be characteristic of BM-MSC. Comparable with BM-MSC, such MSC-like cells isolated from the synovial membrane were shown for the first time to suppress the T-cell response in a mixed lymphocyte reaction, and to express the enzyme indoleamine 2,3-dioxygenase activity to the same extent as BM-MSC, which is a possible mediator of this suppressive activity. Using quantitative RT-PCR these data show that MSC-like cells from the synovium and BM may be induced to chondrogenic differentiation and, to a lesser extent, to osteogenic differentiation, but the osteogenic capacities of the synovium-derived MSC were significantly reduced based on the expression of the markers tested (collagen type II and aggrecan or alkaline phosphatase and osteocalcin, respectively). Transcription profiles, determined with the Atlas Human Cytokine/Receptor Array, revealed discrimination between the MSC-like cells from the synovial membrane and the BM-MSC by 46 of 268 genes. In particular, activin A was shown to be one major upregulated factor, highly secreted by BM-MSC. Whether this reflects a different cellular phenotype, a different amount of MSC in the synovium-derived population compared with BM-MSC adherent cell populations or the impact of a different microenvironment remains to be determined. In conclusion, although the BM-derived and synovium-derived MSC shared similar phenotypic and functional properties, both their differentiation capacities and transcriptional profiles permit one to discriminate the cell populations according to their tissue origin.
==== Body
Introduction
Mesenchymal stem cells (MSC) are progenitor cells that have the potential to differentiate into lineages of mesenchymal tissues including cartilage, bone, muscle and fat. They were initially isolated from bone marrow (BM) and characterized by the expression of various cell surface markers [1,2]. MSC have more recently been obtained from adipose tissue, peripheral blood, cord blood, cartilage [3-6] and synovial tissue [7].
Identification of MSC in the synovium has raised speculations about their biological role in the normal or pathologic joint physiology. As MSC have a great potential to repair damaged tissues, they are likely to contribute to joint regeneration in arthritis. Indeed, MSC have been detected in the synovial fluid of patients with arthritis, with a higher prevalence in osteoarthritis (OA). In this OA, MSC may participate in the highly active process of regeneration due to the reactivation of endochondrial ossification in the advanced phase of the disease [8]. However, a significant reduction in the in vitro chondrogenic and adipogenic activities of MSC has been reported in patients with OA [9]. The authors suggest that changes in the differentiation profile of MSC account for the increase of bone density and loss of cartilage that are characteristics of OA. Recent data suggest a possible involvement of MSC in the pathophysiology of OA, but also in inflammatory arthritis [10]. In the study, the authors show that during the induction phase of collagen-induced arthritis, marrow-derived mesenchymal cells accumulate in the synovium preceding the clinical onset of arthritis and afflux of inflammatory cells [10]. Thus, although still to be demonstrated, MSC may play a pivotal role in the induction phase of arthritis by promoting the accumulation of immunocompetent cells into the joint.
To date, identification of MSC from the synovial membrane exclusively relies on their phenotypic characterization and on the assessment of their differentiation potential. MSC from the synovial membrane were shown to express various surface markers (CD9, CD10, CD13, CD44, CD54, CD55, CD90, CD105, CD166, D7-FIB) and to be negative for CD14, CD20, CD45 and CD133 by fluorescent-activated cell sorting (FACS) analysis [7,8,11]. A more detailed study involving molecular characterization of MSC from the synovial membrane by RT-PCR has revealed the expression of various matrix molecules, adhesion molecules, ligands, receptors and transcription factors [7]. Functional characterization of MSC from the synovial membrane has shown their multilineage potential as they are able to differentiate towards chondrocytes, osteoblasts, adipocytes and, to a lesser extent, towards myocytes [7].
Isolation of MSC from the synovium [7,11], mainly based on adhesion properties, relies on the technique used to isolate synovial fibroblasts, suggesting that only a subset of the cell population corresponds to the MSC. On the basis of the present knowledge of the biology of BM-derived MSC, we underwent parallel studies to phenotypically and functionally compare the MSC isolated from both tissues. Interestingly, using quantitative analyses, our results show that the potential of differentiation towards osteocytes were significantly reduced in synovium-derived MSC. The present study is the first to demonstrate that MSC from the synovial membrane share the same immunosuppressive features as BM-MSC because they are able to inhibit the T-cell proliferation in a mixed lymphocyte reaction (MLR) and to display indoleamine 2,3-dioxygenase (IDO) activity. Importantly, using macroarray technology we provide evidence that the transcriptional profiles could be used to discriminate the MSC by function of their tissue origin, Activin A being one major upregulated gene in BM-MSC.
Materials and methods
Cell culture
Human MSC cultures were established from BM aspirates of healthy donors or from OA patients and rheumatoid arthritis (RA) patients undergoing hip replacement surgery, after informed consent. The cell suspension was diluted in serum-free medium, filtered on a nylon membrane (Cell Strainer; Dutscher, Cergy, France) and centrifuged at 200 × g for 10 min at ambient temperature. Mononuclear cells were then plated at the density of 5 × 104 cells/cm2 in α-MEM, supplemented with 10% fetal bovine serum (Perbio Science France SAS, Brebières, France), 1 ng/ml basic fibroblast growth factor, 100 U/ml penicillin and 100 μg/ml streptomycin. When cultures reached near confluence, cells were detached with 0.05% trypsin and 0.53 mM ethylenediamine tetracetic acid, and were subsequently replated at the density of 1,000 cells/cm2. BM adherent cells were used between passage 2, when a homogeneous population of cells was microscopically observed, and passage 7. The median age of the BM-MSC samples was 53.13 ± 18.3 years, corresponding one-half to healthy donors and approximately one-quarter to RA patients and one-quarter to OA patients.
Human synovium-derived adherent cells were isolated from synovial tissues either post mortem (healthy donors) or at the time of surgical knee replacement for degenerative OA or RA. Synovial tissues were finely minced and digested with 0.2% collagenase in DMEM containing 10% fetal bovine serum, 100 U/ml penicillin and 100 μg/ml streptomycin (complete DMEM). Following overnight incubation at 37°C, cells were collected by centrifugation, washed once and filtered on a nylon membrane (Cell Strainer; Dutscher). The cell suspension was then plated in complete DMEM in a 75 cm2 flask and passaged when reaching near confluence according to previous report [7]. Synovial cells were used between passage 4 and passage 8. The median age of synovium-derived MSC samples was 54.6 ± 28.8 years, corresponding one-half to RA patients and approximately one-quarter to OA patients and one-quarter to healthy donors. Due to ethical considerations, BM-derived and synovium-derived cells were not obtained from the same patients.
Phenotypic characterization
For flow cytometry, cells were harvested by treatment with 0.05% trypsin and 0.53 mM ethylenediamine tetracetic acid, and were resuspended in PBS containing 0.1% BSA and 0.01% sodium azide. Cell aliquots (105 to 5 × 105 cells/100 μl) were incubated on ice with conjugated mAbs against CD14, CD34, CD44, CD45, CD73, CD90 and CD105 (BD Pharmingen, Le Pont de Claix, France) or conjugated isotypic controls. Flow cytometry was performed on a fluorescence-activated cell sorter (FACS Scan, BD Biosciences, Le Pont de Claix, France), and data were analyzed with the Cellquest software (BD Pharmingen).
For immunofluorescence analysis, cells were fixed with acetone:methanol (1:1), washed with PBS and incubated with the primary mAb specific for the human prolyl-4-hydroxylase (Dako, Trappes, France) at 1:50 dilution for 30 min at room temperature. Washed slides were then incubated with a secondary fluorescein isothiocyanate-conjugated goat anti-mouse antibody for 30 min in the dark. Fluorescence was visualized using a Zeiss standard microscope equipped with an AxioCam MRcamera (Carl Zeiss Vision, Le Pecq, France).
Induction of genes by interferons
Cells were cultured in the presence of 1,000 U/ml IFN-α, IFN-β or IFN-γ at 37°C for 6 or 48 hours in the case of IFN-γ. Expression of major histocompatibility complex (MHC) class I and class II molecules was detected by flow cytometry using mAbs specific for HLA-A, HLA-B, HLA-C and anti-HLA-DR molecules (W6.32 and L243 clones, respectively). Induction of the 6–16 gene rapidly induced by IFNs was recorded by quantitative real-time PCR as previously described [12].
Chondrogenic differentiation and osteogenic differentiation
Chondrogenic differentiation was induced by a 21-day culture in micropellet. Briefly, cells (2.5 × 105 cells) were pelleted by centrifugation in 15 ml conic tubes and cultured in BMP-2-conditioned chondrogenic medium. Osteogenesis was induced by culture at low density (1.5 × 104 cells in a 100-mm-diameter culture dish) for 21 days in BMP-2-conditioned osteogenic medium. The conditioned media were obtained after incubation of C9 cells at confluence for 48 hours in the presence of either chondrogenic medium or osteogenic medium. C9 cells are derived from the C3H10T1/2 murine MSC line, and they express 1,230 ng hBMP-2 per 24 hour/106 cells under the control of a TetOff promoter [13]. As the control, supernatants from C3H10T1/2 cells were unable to induce any cell differentiation (data not shown). The chondrogenic medium consisted of DMEM supplemented with 0.1 μM dexamethasone (Sigma, l'Isle d'Abeau, France), 0.17 mM ascorbic acid and 1% insulin-transferrin-sodium selenite media supplement (Sigma). The osteogenic medium consisted of DMEM medium supplemented with 10% FCS, 10 mM β-glycerophosphate (Sigma), 0.1 μM dexamethasone (Sigma) and 0.05 mM ascorbic acid (Sigma). The adipogenic differentiation potential for BM-derived and synovium-derived MSC has been checked according to a previously described protocol [14].
Real-time RT-PCR
Total RNA was extracted from cell micropellets using the RNeasy mini kit (Qiagen S.A., Courtaboeuf, France) and from cells in monolayers using the Promega SV Total RNA Isolation System protocol (Promega, Charbonnières-les-bains, France) as recommended by the suppliers. Total RNA was reverse transcribed using Multiscribe reverse transcriptase (Applied Biosystems, Courtaboeuf, France). The TaqMan gene expression arrays and the TaqMan Universal Master Mix were used according to the manufacturer's recommendations (Applied Biosystems). Measurement and analysis of gene expression were performed using the ABI Prism 7000 Sequence Detection System software (Applied Biosystems). Content of cDNA samples was normalized by subtracting the number of copies of the endogenous GAPDH reference gene from the number of copies of the target gene (ΔCt = Ct of target gene – Ct of GAPDH). Expression of the specific gene was calculated using the formula 2-(ΔCt).
Mixed lymphocyte reaction
MLRs were performed as previously described [15]. Briefly, splenocytes from BALB/c mice and DBA/1 mice were isolated and stimulator splenocytes were inhibited to proliferate by treatment with 50 μg/ml mitomycin C (Sigma) at 37°C for 45 min. Each responder cell population and each stimulator cell population was seeded in triplicate at the concentration of 105 cells/100 μl per well, in 96-well round-bottom plates (BD Biosciences, Le Pont de Claix, France). Synovium-derived and BM-derived adherent cells (105 cells) were added to the MLR to obtain a 300 μl final volume. After 3 days of incubation, 1 μCi/well [3H]thymidine was added overnight and thymidine incorporation was measured using a β-scintillation counter. Each experiment was performed at least three times.
IDO activity measurement
Cells were stimulated with IFN-γ (1,000 U/ml) and/or tumor necrosis factor alpha (TNF-α) (50 ng/ml) for 48 hours in DMEM supplemented with L-tryptophan (100 μg/ml). IDO enzyme activity was measured by tryptophan-to-kynurenine conversion with photometric determination of the kynurenine concentration in the supernatant as the readout, as previously reported [16]. Briefly, 160 μl cell supernatant were transferred to a 96-well culture plate and 10 μl of 30% trichloroacetic acid was added for 30 min at 50°C. After centrifugation, 100 μl supernatant was mixed with 100 μl freshly prepared Ehrlich's solution and the absorbance was read with a microplate reader at 450 nm.
Isolation of total RNA and cDNA hybridization
Total RNAs of adherent cells (four separate samples from healthy BM and four separate samples from healthy synovium between passage 4 and passage 6) were extracted using the RNeasy mini kit (Qiagen S.A.) according to the manufacturer's instructions. Radiolabeled cDNA was prepared from each RNA sample with the Atlas array kit (Clontech, Saint Quentin en Yvelines, France) by a reverse-transcription step in the presence of α-[32P]dATP. The radiolabeled samples were hybridized to the Human Cytokine/Receptor Atlas Nylon cDNA Expression Array (BD Biosciences). After stringent washes, membranes were scanned using a Phosphoimager (Amersham Pharmacia Biotech, Saclay, France).
Gene array analysis
Quantification was performed using the AtlasImage software (BD Biosciences). Data from each array were normalized by the median value to eliminate the variability due to the sample labeling or the exposure duration. The normalized median was arbitrarily given the value 150. Analysis was performed using the Cluster and TreeView hierarchical clustering software developed by Eisen and colleagues [17]. Two filters have been used: one filter aimed at retaining only genes expressed above the median value, and the second filter retained genes for which the difference between the maximum and minimum values was twice the median value. Data were log-transformed (log-base 2), and the genes were median centered and clustered by correlation average linkage clustering. The hierarchical clustering was visualized with TreeView.
Activin A quantification by ELISA
Total activin A was measured by means of a highly specific solid-phase enzyme-linked immunometric assay using reagents supplied by DSL-France (Cergy-Pontoise, France). The first antibody was an anti-βA-subunit monoclonal antibody immobilized on microplate wells. The second antibody was a biotinylated monoclonal antibody. In order to minimize matrix effects, the assay procedure was adapted to the culture medium: the assay standards were reconstituted with nonincubated medium, which was also used as a diluent. The assay had no detectable cross-reaction with inhibin A, follistatin, activin B or inhibin B. The dilution curves of high-level samples paralleled the standard curves. The sensitivity was <0.1 ng/ml and the inter-assay coefficient of variation was <10%.
Statistical analysis
Statistics were performed with the Student t test or an unpaired Mann-Whitney test as appropriate according to data distribution. All data were analyzed by the program Instat (Graphpad, San Diego, CA, USA).
Results
Phenotypic characterization
Adherent cells isolated from BM or synovial tissue were first characterized according to the expression of surface markers known to be expressed or absent on BM-MSC. By FACS analysis, we showed that more than 99% of these cells from both tissues were negative for the expression of CD14, CD34 and CD45 and were positive for CD44, CD73, CD90 and CD105 (Fig. 1a). Similar results were also obtained using human primary skin fibroblasts (data not shown). The fluorescence intensities for each marker were not statistically different between the two cell populations, suggesting a similar level of expression on both cell populations – except for the marker CD90 (P = 0.0311), which was higher on MSC from synovium. Because contradictory results were reported using the antibody specific for prolyl-4-hydroxylase to immunophenotype the MSC [18,19], we checked whether it could be useful to discriminate between MSC derived from BM or MSC derived from synovium. Although no expression was found on both cell populations by FACS analysis (data not shown), they were both positive in immunocytochemistry (Fig. 1b). Similarly, the skin fibroblast cells used as the control were also positive for this marker (data not shown). Indeed, using various markers, adherent cell populations isolated from the two tissues displayed a similar phenotype, commonly observed with MSC-containing cell populations derived from BM.
Expression of MHC class I and class II
BM-MSC are known to be positive for MHC class I molecules and to be negative for MHC class II molecules in basal culture conditions, both being upregulated following treatment with IFN-γ [1]. We confirmed these observations with the cells isolated from BM used in this study (Fig. 2a) and with the synovium-derived cells (data not shown) as already reported [20]. However, no data are available on the effect of IFN-α and IFN-β on the expression of MHC molecules on synoviocytes or BM-MSC. We showed that, similarly to IFN-γ, IFN-α and IFN-β significantly upregulated the MHC class I molecules in both cell populations. Only IFN-γ significantly induced the expression of class II molecules, and both the number of positive cells (56 ± 16% and 36 ± 17% for BM-derived and synovium-derived cells, respectively) and their mean fluorescence were increased (Fig. 2b). However, no statistically significant difference between cells isolated from the two tissues was observed.
We then investigated whether the 6–16 gene, which is a representative of the early type I IFN-stimulated genes, was also upregulated in adherent cells isolated from BM or synovium. The expression of the 6–16 gene was upregulated with IFN-α and IFN-β (Fig. 2c), and to a lesser extent with IFN-γ (data not shown), and a statistically reduced expression level was observed with cells isolated from the synovial membrane. Thus, for most of the markers examined, no significant difference between adherent cells from BM and synovium was observed in our culture conditions.
Differentiation capacities
Similarly to BM-MSC, synovial-derived MSC have been reported to differentiate into chondrocytes, osteocytes and adipocytes [7]. We compared the expression levels of specific markers following the induction of chondrogenesis or osteogenesis in vitro to accurately quantify the differentiation capacities of adherent cells obtained from the two tissues. The expression of two chondrogenic markers (collagen type II and aggrecan) and two osteogenic markers, (osteocalcin and alkaline phosphatase) was quantified relative to GAPDH by quantitative RT-PCR (Fig. 3a, b). A significant increase of the mRNA was observed on both cell populations and for all the differentiation markers tested. The mean increase of collagen type II and aggrecan was stronger in synovium-derived cells than in BM-derived cells but the differences were not significant. Notably, the collagen type II marker was induced in all cell samples isolated from BM whereas only five out of eight samples from the synovial tissue were positive after differentiation (data not shown). Inversely, the mean induction of osteocalcin and alkaline phosphatase was statistically higher in cells from BM than in cells from the synovium. In these conditions, the primary skin fibroblastic cells were unable to differentiate along the chondrogenic and osteogenic pathways (data not shown). Although a high heterogeneity in the induction of the various markers was observed between the samples, synovial cells displayed a reduced osteogenic capacity.
Immunosuppressive nature
We and others have previously shown that BM-MSC exhibit immunosuppressive properties able to inhibit the proliferation of T cells in a MLR [15,21]. However, nothing was known on the behavior of synoviocytes. Similarly to BM-MSC, we showed that MSC from the synovium were able to suppress the proliferative capacities of T cells in a MLR whereas primary skin fibroblasts display no suppressive properties (Fig. 4a). It has very recently been reported that this effect may be mediated by the induction of the IDO activity in BM-MSC [22]. We thus determined whether cells derived from the synovium or from BM may display a similar induction of IDO activity upon IFN activation. As shown in Fig. 4b, incubation of cells with IFN-γ but not with TNF-α resulted in a similar induction of IDO activity in the two cell populations.
Identification of genes differentially expressed
We then compared the cDNA expression profiles of these cell populations using the Atlas Human Cytokine/Receptor Array membrane. Radiolabeled cDNAs of four BM-derived cells versus four synovium-derived cells from healthy donors were hybridized on the membranes, which are composed of cDNAs from 268 genes (Fig. 5a, b). Data were analyzed using the Cluster and TreeView hierarchical clustering software, resulting in a color-coded gene expression representation: expression below the median is green, expression above the median is red, whereas the median expression across all samples is black. The software groups genes by similarities in their pattern of expression, and groups samples by similarities in their pattern of expression over all samples (Fig. 5c).
The analysis of the 268 genes thus ended with the classification of samples into two groups differentiating between the two tissue origins, BM or synovial tissue, suggesting that the differential expression of some genes should permit one to discriminate between the two cell populations (Table 1). The different gene expression patterns allowed us to focus on genes grouped in clusters whose expression was higher in either group. Among the genes that were overexpressed in cells isolated from BM, we checked for the expression of the βA subunit of inhibin or activin. The βA chains either homodimerize by disulfide bonding to form activin A or heterodimerize with inhibin α-chains to form inhibin A [23]. We thus used a specific ELISA developed to quantify the protein levels expressed by adherent cells isolated from the two tissues. Indeed, we specifically detected high levels of activin A in the supernatants from BM-derived cells and detected lower amounts from cells cultured from the synovial membrane (<1/10) (Fig. 5d).
Discussion
It is now established that MSC can be isolated from the synovial membrane [7] as well as from BM or other tissues (for a review, see [1]). The culture conditions, based in part on adherence to plastic, used to isolate MSC from the synovial membrane are similar to those used to obtain fibroblastic synoviocytes, suggesting that only a subset of cells in this synovium-derived cell population are stem cells, like MSC from BM-derived cultures. As no specific marker for MSC is presently available, their characterization relies essentially on their functional properties, but to date no quantitative data allow comparison between MSC isolated from various tissues. In the present study we have shown that BM-derived and synovium-derived adherent cell populations can be induced to differentiate towards chondrocytes and osteoblasts, but a significant fivefold to 10-fold reduction in the expression levels of the osteogenic markers was observed with the adherent cells from the synovium. Furthermore, we show that the transcriptional profiles permit one to discriminate between the cell populations isolated from BM or the synovium, and that activin A might be a useful marker since it is highly secreted by MSC from BM.
In agreement with other studies [8,11], we confirm that, together with skin fibroblasts, synovium-derived adherent cells express various markers known to be present on BM-MSC, such as CD44, CD90 and CD105. We now show that the cells isolated from the synovial membrane also express the CD73 marker, an ecto-5'-nucleotidase recognized by the SH3 antibody, and express prolyl-4-hydroxylase. Prolyl-4-hydroxylase was shown to be expressed by synoviocytes [18,24] but to be absent on BM-MSC using FACS analysis [18]. In the present study we were able to detect the protein in both cell populations by immunocytochemistry, whereas it was negative by FACS analysis, suggesting that this marker was not specific for synoviocytes. Expression of MHC class I molecules on BM-MSC and synovial fibroblasts has been reported, as well as the induction of MHC class II molecules upon treatment with IFN-γ [25,26]. However, although IFN-β has been shown to induce the expression of MHC class I and class II molecules on various cell types [27,28], no data were available on the potential role of IFN-α and IFN-β on synovial fibroblasts or BM-MSC. To our knowledge, this is the first report to show that IFN-α and IFN-β upregulate the MHC class I molecules and the 6–16 gene, which is one of the early responsive genes induced by IFN, but fail to induce the MHC class II molecules on both cell populations. Altogether, these data illustrate the high similarity between adherent cell populations isolated from BM and the synovium, based on the expression of various phenotypic markers known to be currently tested with MSC. This suggests that these markers correspond to molecules expressed on cells of mesenchymal origin as they are also detected on primary skin fibroblasts (data not shown) and points out the lack of specific markers.
The multilineage potential of MSC from the synovium has already been described [6,7]. These studies only described qualitative results based on histological and immunohistological staining or semiquantitative RT-PCR. In our study, we performed quantitative RT-PCR to indicate quantitative differences in the expression level of the markers specific for the differentiated states. Variability in gene expression was observed between samples, independently of gender, age or the status of the patient (normal, OA or RA) (data not shown). In previous studies, the multilineage potential of synovium-derived MSC was reported to be independent of donor age, passaging or cryopreservation [7]. We now report that synovium-derived cells expressed significantly lower levels of osteogenic marker mRNA after in-vitro-induced osteogenesis, whereas these cells tended to secrete higher amounts of chondrogenic marker mRNA after chondrogenesis induction, although the differences with BM-derived MSC were not statistically significant. These data may reflect a different cellular phenotype or a different amount of MSC inside the two cell populations. Indeed, in the cells isolated from the synovium tissue we measured 1.8 ± 1.4 colony-forming units in 104 plated cells, which is in the same range obtained with cells from BM (estimated to be 1 in 104 or 105 mononuclear cells) [29]. Alternatively, they may reflect a commitment of stem cells under the influence of the environmental parameters. The presence of progenitors already committed to the chondrogenic lineage may be in higher amounts in the synovial membrane, where they contribute to the homeostasis of the cartilage tissue that is in close contact. Conversely, the higher capacity of differentiation toward osteoblasts observed with cells isolated from BM may suggest a higher numbers of cells committed to the osteogenic lineage inside the BM. Availability of markers specific for the MSC or different stages of differentiation would help to answer this question.
Another functional characteristic of MSC is their capacity to inhibit the proliferation of T cells in a MLR [15]. We show here for the first time that synovium-derived cells not only suppress the proliferative activity of T cells, but also exhibit functional IDO activity upon stimulation with IFN-γ to the same extent as BM-MSC. IDO activity has recently been suggested to contribute to the T-cell suppressive mechanism in human MSC. IDO has been identified as a T-cell inhibitory effector pathway in professional antigen-presenting cells upon induction by IFN-γ and other proinflammatory molecules such as TNF-α. This enzyme catalyzes the conversion from tryptophan to kynurenine; because tryptophan is an essential amino acid, its depletion will impair protein synthesis, leading to inhibition of cell proliferation. Depletion of tryptophan has also been shown to lead to stabilization of IL-6 and IL-8 mRNA, resulting in increased IL-6 and IL-8 responses that were proposed to be implicated in enhanced inflammatory responses to bacterial challenges after a viral infection [30]. A comparison of patients with RA, OA, psoriatic arthritis and gout recorded the highest levels of IL-1β, IL-6, IL-8 and IDO as well as the lowest levels of tryptophan in RA synovial fluids, indicating stimulated cellular immune responses in RA patients [31,32]. Indeed, the possible dual activity of IDO in synoviocytes as well as in BM-MSC still needs to be elucidated. In conditions where TNF-α and IFN-γ are only poorly present, the IDO activity may lead to an immunosuppressive environment inside the joint favoring the inhibition of immune cell proliferation. In the inflammatory context, where TNF-α and IFN-γ are prominent, induction of proinflammatory cytokines may reverse the cytokine balance, leading to a reversion of the immunosuppressive capacity of MSC as we previously showed in the collagen-induced arthritis model of arthritis [33].
The presence of MSC in the synovial membrane addresses the question of their origin and function within the joint. MSC in the synovium may be recruited from the blood that enters the synovial tissue, as they are present in normal conditions and even in higher numbers in the case of injury [8] due to their recruitment from the other tissues where they reside. MSC may also come from the bone marrow, which is connected with the intra-articular space by channels, enlarged in RA [10,34]. The role of MSC is possibly related to their potential to repair tissues of mesodermal origin present inside the joint in the case of traumatic or pathologic injuries [10]. Another postulated role for MSC is their possible involvement in the early phases of osteoarticular diseases and, in particular, in RA [10]. Although MSC possess immunosuppressive capacities, we have previously shown that they are unable to display a benefit in the collagen-induced arthritis model because they lose this property in the presence of TNF-α [33]. Moreover, the increase of MHC class II expression on MSC upon IFN-γ stimulation may further contribute to the aggravation of the immune response. We thus may postulate that TNF-α is the key molecule at the onset of RA pathogenesis that induces or contributes to modifying the characteristics of the MSC, which then act to favor the accumulation of immunocompetent cells into the joint.
An important feature revealed in the present study is that cells isolated from BM and the synovial membrane could be distinguished by distinct gene expression profiles. Both populations are thus characterized by the differential expression of various genes, in particular activin A that is upregulated in BM-MSC. Activins and inhibins are members of the transforming growth factor beta super-family that play roles in skeletal development and bone morphogenesis [23]. Activin A is a multifunctional cytokine that regulates cell growth and differentiation, whose effects are diverse depending on the cell type. In synoviocytes, activin A has been reported to promote proliferation, to be induced by IL-1β and to be upregulated in OA and RA patients [35-37]. It is still unclear, however, whether activin A accelerates or inhibits RA autoimmunity and inflammation. The secretion of activin A in BM-MSC is induced by BMP-2, at least in vitro, and therefore has been suggested to be downstream BMP-2 in the differentiation program that results in skeletal development [38]. In the bone marrow, where MSC are under the influence of transforming growth factor beta and BMP molecules, the upregulation of activin A may be involved early at the beginning of the cascade of events promoting chondrogenic/osteogenic differentiation [39]. However, activin A has been recently shown to play a role in the maintenance of the pluripotency of human embryonic stem cells [40]. The maintenance of pluripotency of embryonic stem cells could involve Wnt signaling and could occur through a crosstalk between the transforming growth factor beta/activin and Wnt pathways [41]. Indeed, activin A may play a dual role according to the environmental parameters: proliferation or maintenance of pluripotency. The higher amounts of activin A produced by BM-MSC together with similar numbers of colony-forming units further suggest their higher multipotent potential.
In summary, the similarity between adherent cells cultured from BM and from synovial tissue suggests a common origin. The few discrepancies between cells may reflect the impact of the tissue environment on the properties of MSC. Thus, due to pathological conditions, reduced differentiation properties and reversion of immunosuppression of MSC have been reported. It will be of therapeutic interest to determine whether MSC originating from various tissue sources share the same features. In this respect, the demonstration that high levels of activin A are produced by BM-MSC may potentially be of relevance in arthritis and repair since it may be associated with the pluripotency of the cells. MSC isolated from various tissues not involved in the specific pathology may be an alternative and more suitable source of cells with fully functional features for tissue engineering.
Abbreviations
α-MEM = alpha-minimum essential medium; BM = bone marrow; BSA = bovine serum albumin; DMEM = Dulbecco's modified Eagle's medium; ELISA = enzyme-linked immunosorbent assay; FACS = fluorescent-activated cell sorting; FCS = fetal calf serum; IDO = indoleamine 2,3-dioxygenase; IFN = interferon; IL = interleukin; mAb = monoclonal antibody; MHC = major histocompatibility complex; MLR = mixed lymphocyte reaction; MSC = mesenchymal stem cells; OA = osteoarthritis; PBS = phosphate-buffered saline; PCR = polymerase chain reaction; RA = rheumatoid arthritis; RT = reverse transcriptase; TNF-α = tumor necrosis factor alpha.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
FD performed the majority of the experimental work and participated in the analysis of the data. CB participated in the cell culture work and the molecular analysis. TH procured the synovial fibroblasts from healthy donors. GU participated in the molecular analysis. NL performed the quantification of activin by ELISA. PL-P helped in the analysis of the data. FA helped in the analysis of the data. FC procured the samples of arthritic synovium. TR helped with the biostatistical analysis. JS participated in the analysis of the data. CJ participated in the design of the study and the analysis of the data. DN participated in the design of the study and the analysis of the data, and wrote the manuscript. All authors read and approved the final manuscript.
Figures and Tables
Figure 1 Immunophenotype of bone marrow-derived (left) and synovium-derived adherent cells (right). (a) Fluorescence-activated cell sorting of CD marker expression. Results are expressed as the mean fluorescence intensity ± standard error of the mean. One representative experiment out of six different samples of bone marrow mesenchymal stem cells (three normal, two osteoarthritis and one rheumatoid arthritis) and synovium mesenchymal stem cells (two rheumatoid arthritis, two osteoarthritis and two healthy) is shown. Control corresponds to the fluorescence due to the isotypic control. (b) Immunofluorescence staining on cells in a monolayer using the monoclonal antibody specific for human prolyl-4-hydroxylase.
Figure 2 Induction of HLA class I and class II expression by IFN in adherent cells. (a) One example of basal and IFN-γ-induced expression of HLA-A, HLA-B, HLA-C (left) and HLA-DR (right) molecules on mesenchymal stem cells from bone marrow (BM) by fluorescence-activated cell sorting analysis. (b) Expression of HLA-A, HLA-B, HLA-C (left) and HLA-DR (right) molecules by BM-derived cells (n = 5; two normal, two osteoarthritis and one rheumatoid arthritis) and synovium-derived cells (n = 7; three rheumatoid arthritis, two normal, two osteoarthritis) after induction with 1,000 U/ml IFN-α, IFN-β and IFN-γ, shown as the fold increase of the mean fluorescence intensity over the control ± standard error of the mean. (c) Detection of the 6–16 early response gene to IFN, assessed by quantitative RT-PCR, normalized to GAPDH mRNA (n = 3 for synovium-derived and n = 4 for BM-derived cell samples). Results display the fold increase of the 6–16 gene in IFN-induced samples over nontreated samples. * P < 0.05.
Figure 3 In-vitro-induced differentiation of bone marrow-derived and synovium-derived adherent cells. (a) Chondrogenic differentiation was evaluated after 21 days in micropellet culture. Expression of the specific markers for chondrogenesis (collagen type II [Col2] and aggrecan [Agg]) was determined by quantitative RT-PCR from the various samples: eight bone marrow (BM) mesenchymal stem cells (MSC) (four healthy, two osteoarthritis [OA], two rheumatoid arthritis [RA]) and eight synovium-MSC (four RA, two OA, two RA). (b) Osteogenic differentiation was evaluated after 21 days in monolayers. Expression of the osteogenic markers (osteocalcin [OC] and alkaline phosphatase [AP]) was determined by quantitative RT-PCR from the various samples: eight BM-MSC (four healthy, two OA, two RA) and eight synovium-MSC (four RA, two OA, two healthy). The expression levels were normalized on the basis of GAPDH expression, and the results are reported as ratios of the marker gene versus GAPDH using the formulae 2-ΔCt (×100) ± standard error of the mean. Statistics compared eight cell samples from BM and eight cell samples from the synovium. * P < 0.05, ** P < 0.01.
Figure 4 Immunosuppressive properties of bone marrow-derived and synovium-derived adherent cells. (a) Proliferative activity of T cells in a mixed lymphocyte reaction. Responding BALB/c splenocytes (105 cells) were incubated for 4 days with mitomycin-treated DBA/1 splenocytes (105) in the presence or absence of 105 adherent cells from bone marrow (BM) (n = 10; six healthy, two rheumatoid arthritis, two osteoarthritis) or synovium (n = 10; five rheumatoid arthritis, two osteoarthritis, three normal) or normal fibroblasts (fibro; n = 3). The proliferative response corresponding to the average counts per minute of triplicates of alloreactive T cells (allo) was assigned the value of 100% ± standard deviation (SD). (b) The indoleamine 2,3-dioxygenase (IDO) activity was detected as the tryptophan to kinurenin conversion measured at an optical density (OD) of 450 nm. Cells were cultured in the absence or presence of IFN-γ and/or tumor necrosis factor alpha (TNF-α) for 48 hours and the detection of kinurenin was measured by photometry. The IDO activity is expressed as the mean activity from four different samples of cells from each tissue ± SD.
Figure 5 Analysis of genes differentially expressed by normal bone marrow-derived and synovium-derived adherent cells. Total RNAs were extracted from cells at passages 4–6 obtained from healthy donors and reverse-transcribed to radiolabeled cDNAs before hybridization to gene array membranes (Atlas Human Cytokine/Receptor Array; BD Biosciences). Each gene is represented as duplicate spots. (a) One example of a representative array from cells isolated from bone marrow (BM) (n = 4). The arrow represents the activin/inhibin βA subunit. (b) One example of a representative array from synovial cells (n = 4). The arrow represents the activin/inhibin βA subunit. (c) Hierarchical clustering analysis. The dendrogram at the top represents the relationship of the samples according to the similarity in their gene expression profile. Arrows indicate some areas of genes that permit one to discriminate between samples. Red, high expression, green, low expression. (d) Quantification of the secreted activin A by specific ELISA. Results are expressed as the mean of secretion in 24-hour and 48-hour supernatants of four cell samples ± standard error of the mean.
Table 1 Genes whose mRNA expression is differentially expressed in bone marrow-derived or synovium-derived cells
Gene Fold changes versus day 0
Upregulated in bone marrow-derived cells
MIP-1β 40.5
IGF2 34.3
OX40 ligand 27.5
Neuregulin 26.8
Lymphotoxin-β 13.5
Acidic fibroblast growth factor 11.4
GAPDH 11.3
Lymphotoxin-α precursor 10.0
BIGH3 9.5
TRAIL 7.2
IFN-γ receptor 7.2
G-CSF 6.9
Glia-derived neurite-promoting factor (GDNPF) 6.4
IFN-α/β receptor β-subunit precursor 6.2
Hepatocyte growth factor 6.0
RPL13A 5.9
Platelet-activating factor receptor 5.6
Keratinocyte growth factor; FGF7 4.8
Endothelin receptor type A 4.5
IGF-binding protein 2 (IGFBP2) 4.1
Inhibin bA subunit precursor; activin bA subunit precursor 4.1
Apoptosis-related protein TFAR15 4.0
Related to receptor tyrosine kinase (RYK) 3.4
VEGFR1 3.3
IGF-binding protein 3 (IGFPB3) 3.0
Insulin-like growth factor I receptor (IGF1R) 2.9
Downregulated in bone marrow-derived cells
MIP2-α, GRO-β -26.2
IL-8 precursor -25.3
Tumor necrosis factor-inducible protein TSG-6 -13.8
βNGF -11.9
Granulocyte chemotactic protein (GCP2) -6.2
FLT3 ligand -5.1
Leukocyte IFN-inducible peptide -4.9
BMP-4 + BMP-2B -4.7
Activin receptor-like kinase 1 (Alk1); transforming growth factor beta superfamily receptor I (TSR1) -4.2
Tumor necrosis factor alpha -4.0
TRAIL receptor 2; death receptor 5 -3.8
BMP3B; GDF10 -3.6
Ciliary neurotropic factor receptor (CNTFR) -3.6
Tumor necrosis factor receptor (TNFR) -3.3
B-cell growth factor 1 precursos (BCGF1) -3.2
IL-13 precursor -3.1
Transforming growth factor beta -2.8
IL-6 -2.6
ERBB2 receptor protein-tyrosine kinase -2.5
IFN regulatory factor -2.3
Upregulation or downregulation refer to the fold increase higher than ratio 2 in bone marrow samples (n = 4) compared with synovial samples (n = 4), with a Mann-Whitney significance of 0.05.
==== Refs
Barry FP Murphy JM Mesenchymal stem cells: clinical applications and biological characterization Int J Biochem Cell Biol 2004 36 568 584 15010324 10.1016/j.biocel.2003.11.001
Pittenger MF Mackay AM Beck SC Jaiswal RK Douglas R Mosca JD Moorman MA Simonetti DW Craig S Marshak DR Multilineage potential of adult human mesenchymal stem cells Science 1999 284 143 147 10102814 10.1126/science.284.5411.143
Zuk PA Zhu M Mizuno H Huang J Futrell JW Katz AJ Benhaim P Lorenz HP Hedrick MH Multilineage cells from human adipose tissue: implications for cell-based therapies Tissue Eng 2001 7 211 228 11304456 10.1089/107632701300062859
Zvaifler NJ Marinova-Mutafchieva L Adams G Edwards CJ Moss J Burger JA Maini RN Mesenchymal precursor cells in the blood of normal individuals Arthritis Res 2000 2 477 488 11056678 10.1186/ar130
Lee OK Kuo TK Chen WM Lee KD Hsieh SL Chen TH Isolation of multipotent mesenchymal stem cells from umbilical cord blood Blood 2004 103 1669 1675 14576065 10.1182/blood-2003-05-1670
Alsalameh S Amin R Gemba T Lotz M Identification of mesenchymal progenitor cells in normal and osteoarthritic human articular cartilage Arthritis Rheum 2004 50 1522 1532 15146422 10.1002/art.20269
De Bari C Dell'Accio F Tylzanowski P Luyten FP Multipotent mesenchymal stem cells from adult human synovial membrane Arthritis Rheum 2001 44 1928 1942 11508446 10.1002/1529-0131(200108)44:8<1928::AID-ART331>3.0.CO;2-P
Jones EA English A Henshaw K Kinsey SE Markham AF Emery P McGonagle D Enumeration and phenotypic characterization of synovial fluid multipotential mesenchymal progenitor cells in inflammatory and degenerative arthritis Arthritis Rheum 2004 50 817 827 15022324 10.1002/art.20203
Murphy JM Dixon K Beck S Fabian D Feldman A Barry F Reduced chondrogenic and adipogenic activity of mesenchymal stem cells from patients with advanced osteoarthritis Arthritis Rheum 2002 46 704 713 11920406 10.1002/art.10118
Marinova-Mutafchieva L Williams RO Funa K Maini RN Zvaifler NJ Inflammation is preceded by tumor necrosis factor-dependent infiltration of mesenchymal cells in experimental arthritis Arthritis Rheum 2002 46 507 513 11840454 10.1002/art.10126
Fickert S Fiedler J Brenner RE Identification, quantification and isolation of mesenchymal progenitor cells from osteoarthritic synovium by fluorescence automated cell sorting Osteoarthritis Cartilage 2003 11 790 800 14609532 10.1016/S1063-4584(03)00167-5
Dondi E Rogge L Lutfalla G Uze G Pellegrini S Down-modulation of responses to type I IFN upon T cell activation J Immunol 2003 170 749 756 12517937
Moutsatsos IK Turgeman G Zhou S Kurkalli BG Pelled G Tzur L Kelley P Stumm N Mi S Müller R Exogenously regulated stem cell-mediated gene therapy for bone regeneration Mol Ther 2001 3 449 461 11319905 10.1006/mthe.2001.0291
Peister A Mellad JA Larson BL Hall BM Gibson LF Prockop DJ Adult stem cells from bone marrow (MSCs) isolated from different strains of inbred mice vary in surface epitopes, rates of proliferation, and differentiation potential Blood 2004 103 1662 1668 14592819 10.1182/blood-2003-09-3070
Djouad F Plence P Bony C Tropel P Apparailly F Sany J Noel D Jorgensen C Immunosuppressive effect of mesenchymal stem cells favors tumor growth in allogeneic animals Blood 2003 102 3837 3844 12881305 10.1182/blood-2003-04-1193
Daubener W Hucke C Seidel K Hadding U MacKenzie CR Interleukin-1 inhibits gamma interferon-induced bacteriostasis in human uroepithelial cells Infect Immun 1999 67 5615 5620 10531207
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
Potian JA Aviv H Ponzio NM Harrison JS Rameshwar P Veto-like activity of mesenchymal stem cells: functional discrimination between cellular responses to alloantigens and recall antigens J Immunol 2003 171 3426 3434 14500637
Campagnoli C Roberts IA Kumar S Bennett PR Bellantuono I Fisk NM Identification of mesenchymal stem/progenitor cells in human first-trimester fetal blood, liver, and bone marrow Blood 2001 98 2396 2402 11588036 10.1182/blood.V98.8.2396
Alvaro-Gracia JM Zvaifler NJ Firestein GS Cytokines in chronic inflammatory arthritis. V. Mutual antagonism between interferon-gamma and tumor necrosis factor-alpha on HLA-DR expression, proliferation, collagenase production, and granulocyte macrophage colony-stimulating factor production by rheumatoid arthritis synoviocytes J Clin Invest 1990 86 1790 1798 2174906
Di Nicola M Carlo-Stella C Magni M Milanesi M Longoni PD Matteucci P Grisanti S Gianni AM Human bone marrow stromal cells suppress T-lymphocyte proliferation induced by cellular or nonspecific mitogenic stimuli Blood 2002 99 3838 3843 11986244 10.1182/blood.V99.10.3838
Meisel R Zibert A Laryea M Gobel U Daubener W Dilloo D Human bone marrow stromal cells inhibit allogeneic T-cell responses by indoleamine 2,3-dioxygenase-mediated tryptophan degradation Blood 2004 103 4619 4621 15001472 10.1182/blood-2003-11-3909
Ball EM Risbridger GP Activins as regulators of branching morphogenesis Dev Biol 2001 238 1 12 11783989 10.1006/dbio.2001.0399
Zimmermann T Kunisch E Pfeiffer R Hirth A Stahl HD Sack U Laube A Liesaus E Roth A Palombo-Kinne E Isolation and characterization of rheumatoid arthritis synovial fibroblasts from primary culture – primary culture cells markedly differ from fourth-passage cells Arthritis Res 2001 3 72 76 11178129 10.1186/ar142
Le Blanc K Tammik C Rosendahl K Zetterberg E Ringden O HLA expression and immunologic properties of differentiated and undifferentiated mesenchymal stem cells Exp Hematol 2003 31 890 896 14550804 10.1016/S0301-472X(03)00110-3
Mourad W Mehindate K Schall TJ McColl SR Engagement of major histocompatibility complex class II molecules by superantigen induces inflammatory cytokine gene expression in human rheumatoid fibroblast-like synoviocytes J Exp Med 1992 175 613 616 1732419 10.1084/jem.175.2.613
Barnes E Salio M Cerundolo V Medlin J Murphy S Dusheiko G Klenerman P Impact of alpha interferon and ribavirin on the function of maturing dendritic cells Antimicrob Agents Chemother 2004 48 3382 3389 15328100 10.1128/AAC.48.9.3382-3389.2004
Keir ME Stoddart CA Linquist-Stepps V Moreno ME McCune JM IFN-alpha secretion by type 2 predendritic cells up-regulates MHC class I in the HIV-1-infected thymus J Immunol 2002 168 325 331 11751977
Deans RJ Moseley AB Mesenchymal stem cells: biology and potential clinical uses Exp Hematol 2000 28 875 884 10989188 10.1016/S0301-472X(00)00482-3
van Wissen M Snoek M Smids B Jansen HM Lutter R IFN-gamma amplifies IL-6 and IL-8 responses by airway epithelial-like cells via indoleamine 2,3-dioxygenase J Immunol 2002 169 7039 7044 12471139
Bertazzo A Punzi L Bertazzolo N Pianon M Pozzuoli A Costa CV Allegri G Tryptophan catabolism in synovial fluid of various arthropathies and its relationship with inflammatory cytokines Adv Exp Med Biol 1999 467 565 570 10721101
Schroecksnadel K Kaser S Ledochowski M Neurauter G Mur E Herold M Fuchs D Increased degradation of tryptophan in blood of patients with rheumatoid arthritis J Rheumatol 2003 30 1935 1939 12966593
Djouad F Fritz V Apparailly F Louis-Plence P Bony C Sany J Jorgensen C Noel D Reversal of the immunosuppressive properties of mesenchymal stem cells by tumor necrosis factor alpha in collagen-induced arthritis Arthritis Rheum 2005 52 1595 1603 15880818 10.1002/art.21012
Nakagawa S Toritsuka Y Wakitani S Denno K Tomita T Owaki H Kimura T Shino K Ochi T Bone marrow stromal cells contribute to synovial cell proliferation in rats with collagen induced arthritis J Rheumatol 1996 23 2098 2103 8970047
Kasperkovitz PV Timmer TC Smeets TJ Verbeet NL Tak PP van Baarsen LG Baltus B Huizinga TW Pieterman E Fero M Fibroblast-like synoviocytes derived from patients with rheumatoid arthritis show the imprint of synovial tissue heterogeneity: evidence of a link between an increased myofibroblast-like phenotype and high-inflammation synovitis Arthritis Rheum 2005 52 430 441 15692990 10.1002/art.20811
Ota F Maeshima A Yamashita S Ikeuchi H Kaneko Y Kuroiwa T Hiromura K Ueki K Kojima I Nojima Y Activin A induces cell proliferation of fibroblast-like synoviocytes in rheumatoid arthritis Arthritis Rheum 2003 48 2442 2449 13130463 10.1002/art.11249
Hermansson M Sawaji Y Bolton M Alexander S Wallace A Begum S Wait R Saklatvala J Proteomic analysis of articular cartilage shows increased type II collagen synthesis in osteoarthritis and expression of inhibin betaA (activin A), a regulatory molecule for chondrocytes J Biol Chem 2004 279 43514 43521 15292256 10.1074/jbc.M407041200
Kearns AE Demay MB BMP-2 induces the expression of activin betaA and follistatin in vitro J Cell Biochem 2000 79 80 88 10906757 10.1002/1097-4644(2000)79:1<80::AID-JCB80>3.0.CO;2-K
Hall BK Miyake T All for one and one for all: condensations and the initiation of skeletal development Bioessays 2000 22 138 147 10655033 10.1002/(SICI)1521-1878(200002)22:2<138::AID-BIES5>3.0.CO;2-4
Beattie GM Lopez AD Bucay N Hinton A Firpo MT King CC Hayek A Activin A maintains pluripotency of human embryonic stem cells in the absence of feeder layers Stem Cells 2005 23 489 495 15790770 10.1634/stemcells.2004-0279
Sato N Meijer L Skaltsounis L Greengard P Brivanlou AH Maintenance of pluripotency in human and mouse embryonic stem cells through activation of Wnt signaling by a pharmacological GSK-3-specific inhibitor Nat Med 2004 10 55 63 14702635 10.1038/nm979
|
16277684
|
PMC1297577
|
CC BY
|
2021-01-04 16:02:40
|
no
|
Arthritis Res Ther. 2005 Sep 20; 7(6):R1304-R1315
|
utf-8
|
Arthritis Res Ther
| 2,005 |
10.1186/ar1827
|
oa_comm
|
==== Front
Arthritis Res TherArthritis Research & Therapy1478-63541478-6362BioMed Central London ar18281627768210.1186/ar1828Research ArticleSegregation of a M404V mutation of the p62/sequestosome 1 (p62/SQSTM1) gene with polyostotic Paget's disease of bone in an Italian family Falchetti Alberto [email protected] Stefano Marco [email protected] Francesca [email protected] Monte Francesca [email protected] Alessia [email protected] Laura [email protected] Annalisa [email protected] Antonietta [email protected] Annamaria [email protected] Giancarlo [email protected] Maria Luisa [email protected] Department of Internal Medicine, University of Florence, Florence, Italy2 Department of Internal Medicine, University of Turin, Turin, Italy3 DeGene Spin-off, University of Florence, Florence, Italy2005 15 9 2005 7 6 R1289 R1295 30 3 2005 3 5 2005 1 8 2005 24 8 2005 Copyright © 2005 Falchetti 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.
Mutations of the p62/Sequestosome 1 gene (p62/SQSTM1) account for both sporadic and familial forms of Paget's disease of bone (PDB). We originally described a methionine→valine substitution at codon 404 (M404V) of exon 8, in the ubiquitin protein-binding domain of p62/SQSTM1 gene in an Italian PDB patient. The collection of data from the patient's pedigree provided evidence for a familial form of PDB. Extension of the genetic analysis to other relatives in this family demonstrated segregation of the M404V mutation with the polyostotic PDB phenotype and provided the identification of six asymptomatic gene carriers. DNA for mutational analysis of the exon 8 coding sequence was obtained from 22 subjects, 4 PDB patients and 18 clinically unaffected members. Of the five clinically ascertained affected members of the family, four possessed the M404V mutation and exhibited the polyostotic form of PDB, except one patient with a single X-ray-assessed skeletal localization and one with a polyostotic disease who had died several years before the DNA analysis. By both reconstitution and mutational analysis of the pedigree, six unaffected subjects were shown to bear the M404V mutation, representing potential asymptomatic gene carriers whose circulating levels of alkaline phosphatase were recently assessed as still within the normal range. Taken together, these results support a genotype–phenotype correlation between the M404V mutation in the p62/SQSTM1 gene and a polyostotic form of PDB in this family. The high penetrance of the PDB trait in this family together with the study of the asymptomatic gene carriers will allow us to confirm the proposed genotype–phenotype correlation and to evaluate the potential use of mutational analysis of the p62/SQSTM1 gene in the early detection of relatives at risk for PDB.
==== Body
Introduction
Paget's disease of bone (PDB; Online Mendelian Inheritance in Man (OMIM) entry no. 602080) is a metabolic bone disease characterized by accelerated bone resorption followed by the deposition of dense, chaotic bone matrix, affecting up to 3% of individuals of Caucasian ancestry above the age of 55 years [1]. Although PDB is genetically heterogeneous, in some familial cases of late onset PDB an autosomal dominant pattern of inheritance has been reported [2-4]. Mutations of the p62/sequestosome 1 (p62/SQSTM1) gene account for most of the sporadic and familial forms of PDB [1-5], and exons 7 and 8, encoding the ubiquitin-binding-associated domain (UBA), host a clustered mutational area [2-5]. p62 acts as a scaffold protein in signalling pathways downstream of the interleukin-1, tumour necrosis factor (TNF)-α and nerve growth factor receptors [6].
In a recent paper we described an M404V mutation in the UBA of the p62/SQSTM1 gene in an Italian population of patients affected by PDB [5]. This mutation has also been confirmed in other ethnic groups [7-9]. For the Italian patient carrying this A→G transition at exon 8 [5], collection of the family history demonstrated a clear inheritance for PDB. DNA analysis for the p62/SQSTM1 gene mutation was performed in all affected familial members and in several unaffected subjects, to evaluate the segregation of the M404V mutation with the PDB phenotype and to detect potentially asymptomatic gene carriers. Through this analysis we identified both a familial form of PDB, in which the M404V mutation segregates with a polyostotic phenotype of the disorder, and several asymptomatic gene carriers.
Materials and methods
Family recruitment and disease ascertainment
The Local Ethical Committee of the University of Florence approved this study. The PDB female proband (III-1) was clinically evaluated and genetically characterized as a carrier of a novel M404V mutation at exon 8 of the p62/SQSTM1 gene (Fig. 1) [5].
Through the family history a familial form of PDB (F01 pedigree, Fig. 1) was ascertained. The four-generation family, originating from central Italy, consists of 37 living subjects (22 females and 15 males; age range 33 to 92 years) and 18 deceased individuals (11 males and 7 females; Fig. 1). Members from generations I to III were farmers born and still living in a rural environment, whereas fourth-generation individuals, although born in the same environment as previous generations, moved to urban life after adolescence. Relevant clinical information on affected and gene-carrier members of the F01 pedigree was collected; they are summarized in Table 1.
All available family members were asked to undergo DNA mutational analysis and biochemical assessment after administration of an informed consent form.
No information was available on the first (I) generation (subjects I-1, I-1.0 and I-2; Fig. 1).
In the second (II) generation (Fig. 1) blood samples for genomic DNA evaluation were obtained from the only living subject (patient II-6), a 92-year-old male, suffering from a benign hyperplasia of the prostate (Table 1). Male subject II-3, the father of a III-6 affected individual (Fig. 1), was referred to as a carrier of multiple bone deformities and pain by living members of the family, strongly suggesting the presence of PDB disease in this individual. Figure 2a contributes to sustaining this hypothesis.
The proband III-1, in whom the M404V mutation was first detected [5], belongs to the third (III) generation (Fig. 1). Patient III-6 died because of an osteogenic sarcoma within or in a Pagetic bone. This patient was diagnosed as being affected by PDB at the age of 62 years because of the presence of bone pain, elevated serum alkaline phosphatase (AP) activity and X-rays indicating typical PDB. Bone scintigraphy showed signs of disease in the right pelvis, the right proximal femur and left ribs IV and VIII. Three years after the diagnosis of PDB, bone pain in the right pelvis increased markedly and a bone biopsy showed the presence of an osteogenic sarcoma on the Pagetic bone already metastasized to the lungs. The patient died at 65 years of age after surgical and chemotherapeutic interventions, some years before DNA analysis was performed on the F01 pedigree.
The ages of members from the fourth (IV) generation (IV-1, IV-2, IV-3, IV-5, IV-6, IV-13, IV-14, IV-15 and IV-16) ranged from 41 to 53 years. Neither clinical nor biochemical abnormalities suggestive of PDB are currently evident in this younger group (Fig. 1, Table 1).
Evaluation of AP, measured by an autoanalyzer, has been performed also in all the individuals undergoing mutational analysis. The upper limit of the reference range is 120 units/l.
DNA extraction, PCR and mutational analysis
After administration of an informed consent form, peripheral blood was obtained from 22 subjects: 4 PDB patients (III-1, III-3, III-12 and III-13) and 18 clinically unaffected members (II-6, III-7, III-8, III-9, III-10, III-14, III-18, III-19, III-20, IV-1, IV-2, IV-3, IV-5, IV-6, IV-13, IV-14, IV-15 and IV-16) (Fig. 1, Table 1). Genomic DNA was extracted from peripheral blood leukocytes with the use of a microvolume extraction method, QIAamp DNA Mini Kit (Qiagen GmbH, Hilden, Germany), in accordance with the manufacturer's instructions.
Exon 8 of the p62/SQSTM1 gene was amplified by PCR (I-Cycler; Bio-Rad Laboratories, Milan, Italy) using a couple of primers located in the flanking intron: 5'-CAGTGTGGCCTGTGAGGAC-3'/5'-CAGTGAGCCTTGGGTCTCG-3'. For each patient we used 0.1 μg of DNA, in a final buffer volume of 50 μl (67 mM Tris-HCl, 16.6 mM (NH4)SO4, 0.01% Tween 20, 1.5 mM MgCl2, 0.2 mM deoxyribonucleotides, each primer at 0.2 μM and 1 unit of Polytaq (Polymed, Florence, Italy)). Thirty PCR cycles were performed at 94°C for 30 s, 55°C for 30 s and 72°C for 1 min, after a first denaturing cycle at 94°C for 3 min. A final extension cycle of 5 min was performed at 72°C.
PCR products were tested by 2% ethidium bromide-stained agarose-gel electrophoresis, purified with a High Pure PCR Product Purification Kit (Roche, Indianapolis, IN, USA) and finally sequenced with a BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA). The sequencing reaction consisted of 25 repeated cycles of denaturation for 10 s at 96°C, annealing for 5 s at 55°C and extension for 2 min at 60°C. The sequencing products were purified with a DyeEx 2.0 Spin Kit (Qiagen GmbH, Hilden, Germany) to remove the excess dye terminator. A 5 μl sample of each purified sequence was then resuspended in 15 μl of formamide and denatured for 2 min at 95°C. Analysis of the forward and reverse sequences was performed on an ABI Prism 3100 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA).
Results
Clinical data
Suspicion or diagnosis of PDB was based on description by relatives, evidence of bone deformities and pain strongly suggestive of PDB (II-3 (Fig. 2a), II-8 and II-10) and, when possible, direct evidence of elevated total AP, X-ray scanning and bone scintigraphy (III-1, III-3 (Fig. 2b), III-6, III-12 and III-13; Fig. 1, Table 1).
After careful reconstitution of their clinical history, 12 subjects (6 males and 6 females), of which 11 were living, were reported to be the following: clinically ascertained as PDB patients (III-1, III-3 (Fig. 2b), III-6, III-12 and III-13) with increased circulating levels of AP (more than 120 units/l); presumably affected by PDB (II-3 (Fig. 2a), II-8 and II-10); and potentially mutant (II-1, II-2, II-7 and III-5; Table 1).
In accordance with previously described criteria, all affected members, clinically ascertained (III-1, III-3 (Fig. 2b), III-6, III-12 and III-13), exhibited polyostotic localization of PDB (III-1, III-3, III-6 and III-13), except patient III-12 (Table 1). In the last of these the monostotic left femur involvement was diagnosed only through standard X-ray examination, so the possibility of underestimation of skeletal involvement cannot be excluded (Table 1). Over all, considering only the clinically ascertained affected subjects (III-1, III-3, III-6, III-12 and III-13), the number of bones involved in this family is 3.8 ± 2.31 (mean ± SD). Three subjects in the family (II-3 (Fig. 2a), II-8 and II-10) were presumably affected by PDB on the basis of the history related by other members of the family and of the fact that progeny of II-3 and II-8 carried the M404V mutation. In these three cases the description of bone deformities was suggestive of multiple bone localization.
Until now none of the unaffected members has exhibited AP levels outside the normal range (that is, more than 120 units/l).
Mutational analysis
All the available ascertained affected PDB individuals from the third generation (III-1, III-3, III-12 and III-13) exhibited the M404V mutation of the p62/SQSTM1 gene (Table 1), confirming the pathogenetic nature of this p62/SQSTM1 gene mutation and suggesting segregation of the mutation with the polyostotic phenotype in this family. Although patient III-6 died as a result of an osteogenic sarcoma on a Pagetic bone several years before the genetic evaluation of F01 pedigree, he probably had the M404V mutation. Through mutational analysis the pedigree was carefully reconstructed, and this allowed us to propose that patients II-1, II-2, II-3, II-7, II-8 and III-5 were also carriers of the M404V mutation. In fact, their PDB-affected children (III-1, III-3, III-6, III-12 and III-13) exhibited the mutation (Table 1).
Of the unaffected subjects, II-6, III-7, III-8, III-9, III-10, III-14, III-18, III-19, III-20, IV-2, IV-3, IV-5 and IV-6 (age range from 35 to 92 years) were not carrying the M404V p62/SQSTM1 gene mutation, whereas subjects IV-1, IV-13, IV-14, IV-15 and IV-16 (age range from 41 to 53 years) were carrying the mutation (Table 1).
AP levels were still in the normal range (less than 120 units/l) in these gene carriers (Table 1).
Discussion
Several lines of evidence [2-7] support the role of the p62/SQSTM1 gene in the pathogenesis of PDB, even though the molecular mechanisms that underlie its functional activities are not fully understood. Similarly, little information has been collected about either the potential genotype–phenotype correlation between gene mutations and clinical manifestations of PDB [7] or the role of genetic testing in asymptomatic carriers within affected families. The findings described in this paper are of interest with regard to both issues.
The p62/SQSTM1 protein binds non-covalently to ubiquitin, co-localizing with ubiquitinated inclusions in several human diseases characterized by altered protein aggregation [10]. Moreover, the protein mediates several cellular functions including NFκB-dependent signalling and transcriptional activity, which are important for the recruitment and activation of osteoclastic cells [2].
The nuclear magnetic resonance structure of the p62-UBA domain has recently been determined, but its functional significance in the p62 protein is still unknown [11]. The study by Ciani and colleagues showed that the M404V mutation is able to modify the secondary structure of the domain and affects its ability to bind to Lys48-linked multiubiquitin chains in vitro [11].
Together with other p62/SQSTM1 gene mutations at the UBA domain, Cavey and colleagues [12] showed that M404V is able to cause the loss of monoubiquitin binding and impair in vitro Lys48-linked polyubiquitin binding, although these effects were reported only when the binding experiments were performed at the physiological temperature of 37°C. These findings suggest that PDB-related SQSTM1 mutations may confer a higher susceptibility to development of the disease by impairing the binding of the p62 protein to a ubiquitinated target. However, other molecular mechanisms, involving a key ubiquitinated substrate, could be invoked in the attempt to explain the acquisition of the PDB phenotype in individuals with mutations of the p62/SQSTM1 gene [11]. A structural analysis demonstrated in 4 of 70 PDB relatives with British ancestry that an M404V mutation involves residues on the hydrophobic surface patch implicated in ubiquitin binding [7]. Consequently, an M404V mutation affects the ability of a mutant UBA domain to bind polyubiquitin chains [7].
Using this structural information Hocking and colleagues reported that patients with truncating mutations of the p62/SQSTM1 gene exhibited a trend for more extensive PDB than those with mis-sense mutations such as M404V. They concluded that there is no correlation between the ubiquitin-binding properties of different mutant UBA domains and disease occurrence or extension of the same [7]. These findings therefore do not provide a speculative hypothesis for an explanation of the observed genotype–phenotype correlation in the Italian family with PDB described in this paper.
The heterozygous segregation of M404V mutation with the PDB phenotype in the F01 pedigree supports the pathogenetic role via a dominant-negative action [4]. Moreover, the evidence of a genotype–phenotype correlation in this family can also include epigenetic mechanisms that, through a common genetic background, can contribute, along with the M404V mutation, to the expression of a polyostotic PDB phenotype in the affected members. Interestingly, the commonly shared rural environment of all the members from generations I to III and, for a shorter period, generation IV of this family, together with the presence of past measles infection in all individuals analysed, could suggest a role for environmental factors in determining the polyostotic expression of the disease in M404V mutant subjects.
Clinical follow-up of asymptomatic carriers from generation IV might confirm or negate this observation. Studies on the penetrance of PDB have been performed by several authors [2-4,9], the PDB shows an incomplete clinical expression, meaning that some SQSTM1 gene carriers from affected families do not show clinical evidence of the disease. Moreover, some PDB-affected individuals, from affected families with a known SQSTM1 gene mutation, do not exhibit the mutation [2-4,9]. Conversely, individuals older than 55 years of age with a known SQSTM1 mutation from relatives affected with PDB, did not develop PDB [4]. A potential explanation for these findings is the existence of genetic heterogeneity, with possible modifier loci capable of controlling the clinical expression of PDB [4,9]. For individuals younger than 55 years of age with a known SQSTM1 mutation, originating from relatives affected with PDB, who had not yet developed PDB [4,8], the time needed for phenotypic expression of the disease could represent a limiting factor. In general, a lack of expression of the disease in recognized SQSTM1 gene carriers could be explained by a reduced exposure to environmental factors such as paramixovirus infections and/or by the progressive abandonment of the rural environment [4,7].
An important application of genetic analyses in families is the precocious identification of asymptomatic gene carriers. In this relative the p62/SQSTM1 disease-associated mutation was also present in individuals younger than 50 years of age [3,4]. So far the asymptomatic gene carriers have not exhibited any abnormality in the circulating levels of AP and have not shown any clinical signs suggestive of PDB. Although bone scanning is commonly recommended in patients with PDB older than 40 years of age, because of the ethical considerations observed in our country, bone scan tests cannot be performed unless AP levels are raised and consequently PDB bone localization cannot be excluded at this stage in asymptomatic mutant carriers. However, considering that a positive individual older than 40 years of age has an up to 80% likelihood of developing the disease by 70 years of age [13], the extremely high penetrance of PDB in this family clearly indicates the need for an accurate vertical follow-up of the six asymptomatic mutant carriers. This will allow us to confirm the suggested genotype–phenotype correlation in the currently asymptomatic carriers as well, and to assess the role of mutational analysis of the p62/SQSTM1 gene for early detection of the individuals at risk for developing PDB. At present, a positive test for the mutation of the p62/SQSTM1 gene in a patient with PDB does not have any impact on treatment [13].
Finally, one of the affected subjects (III-6) in this family developed an osteosarcoma that caused her death. Pagetoid osteosarcoma is a complication of PDB [14,15] and is most often observed in severe, long-standing PDB. Two previous reports described a direct lineage in which Pagetoid osteosarcoma developed in affected family members [16,17]. Although specific genetic mechanisms remain to be elucidated, some authors reported loss of heterozygosity for loci at chromosome 18q21-22 in Pagetoid osteosarcomas as well as in sporadic osteosarcomas [17,18]. The deleted region was shown to harbour the receptor activator of nuclear factor κB (RANK, TNFRSF11A) gene identified in a family affected by familial expansile osteolysis (OMIM entry no. 174810), a Paget-like syndrome [19]. Although the RANK gene has not been found to be mutated in PDB-affected individuals, a positive association between a polymorphic variant of this gene and PDB has been reported [20,21]. NFκB is also the potential molecular target of the mechanism underlying the altered osteoclastogenesis seen in PDB patients carrying mutated sequences in the UBA domain of the p62/SQSTM1 gene [2-12]. Inactivation of the p62/SQSTM1 gene could activate the RANK–NFκB signalling, as seen in the familial expansile osteolysis syndrome [19], with impairment of TNF-α-induced programmed cell death [22]. Such machinery is also crucial for immunity, lymphocyte development, tumorigenesis and cancer chemoresistance; NFκB functions are recognized as relevant to tumour promotion [23]. Even though the presence of the M404V mutation could not be assessed in patient III-6, these hypotheses strongly support the need for further investigation into the possible role of the p62 protein in the occurrence of osteosarcoma, both in PDB-affected patients and in individuals without PDB.
Conclusion
This paper describes a genotype–phenotype correlation in PDB cases with a mis-sense mutation in the p62/SQSTM1 gene. These results should be confirmed in other PDB patients of Italian and other ancestries. Moreover, the value of a pre-symptomatic gene test in PDB requires a vertical evaluation in well-characterized relatives, opening new possibilities for the practical application of genetic diagnosis in PDB family members and also in the general population. Finally, the knowledge of the function of p62/SQSTM1 gene mutations should enable us to uncover the pathogenesis of PDB and osteogenic osteosarcoma.
Abbreviations
AP = alkaline phosphatase; NFκB = nuclear factor κB; PCR = polymerase chain reaction; PDB = Paget's disease of bone; RANK = receptor activator of nuclear factor κB; TNF = tumour necrosis factor; UBA domain = ubiquitin-binding-associated domain.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
AF conceived of the study and participated in its design and coordination, acquisition of data, analysis and interpretation of data, and was fully involved in drafting the manuscript and revising it critically for important intellectual content. MDS made substantial contributions to the acquisition and interpretation of data. AF and MDS contributed equally to the work. FM performed the molecular genetic studies. FDM performed the molecular genetic studies together with FM. AG participated in the sequence alignment. LM performed the statistical analysis. AT supervised the performance statistical analysis. AA helped in the clinical activity. AC participated in the sequence alignment. GI participated in the design of the study and helped to draft the manuscript. MLB participated in the design and coordination and helped in drafting the manuscript and revising it critically for important intellectual content; she also gave final approval of the version to be published. All authors read and approved the final manuscript.
Acknowledgements
The authors thank Dr Tiziano Lusenti (Nephrology Unit, Reggio Emilia Santa Maria Hospital, Reggio Emilia, Italy) for his assistance in this study. This paper has been supported by the European Research Program, Fifth Framework Program 'Quality of Life and Management of Living Resources Research and Technological Development Program' on 'Genetic Markers for Osteoporosis', by the Cofin MIUR, PNR 2001–2003 (FIRB), and by the Fondazione Ente Cassa di Risparmio di Firenze (to MLB).
Figures and Tables
Figure 1 Family pedigree (F01). The proband, III-1, is indicated by an arrow. Members of family with ascertained clinical evidence of Paget's disease of bone (PDB; III-1, III-3, III-6, III-12 and III-13) are represented by black symbols. Subjects strongly suspected to be affected by PDB, as reported by personal history in relatives (II-3, II-8 and II-10), are indicated by horizontal bar symbols. Relatives potentially mutant on the basis of pedigree reconstruction (II-1, II-2, II-7 and III-5) are represented by vertical bar symbols. Grey symbols identify individuals known to have the M404V mutation but whose PDB disease was not expressed; open symbols indicate subjects not exhibiting either the mutation or clinical evidence of PDB. Underlined numbers indicate individuals in whom a genetic test was performed. Question marks identify individuals whose clinical phenotype is not verifiable.
Figure 2 Evidence of PDB in two affected subjects from F01 family. (a) Bone deformity of the right forearm of family member II-3. Relatives described him as having suffered from multiple bone deformities and pain. (b) X-ray scan of the right tibia of family member III-3, with a typical flame-shaped lytic wedge (arrow).
Table 1 Available clinical and mutational data on affected patients and gene carriers of PDB
Pedigree number (sex) Age at clinical diagnosis or DNA evaluation; present age (years) AP (U/l) PDB-related clinical finding Other relevant clinical data
II-3 (M)a Deceased Unknown Diffuse marked bone deformities (Fig. 2a) Died at age 76 years from Alzheimer's disease
II-8 (F)a Deceased Unknown Bone deformities at both lower extremities Died at age 52 years from colon-rectal cancer; also had breast cancer
II-10 (F)a Deceased Unknown Multiple marked diffuse skeletal deformities Died at age 92 years from unknown cause
III-1 (F)b 57; 65 357 Third lumbar vertebra, pelvis, right proximal femur Alive
III-3 (F)b 53; 79 560 Pelvis, both tibias Alive. Apparently healthy
III-5 (M)a Not assessed Unknown Diffuse bone pain Died at age 80 years from unknown cause
III-6 (F)a 62; deceased 2,259 Right pelvis and proximal femur, IV and VIII left ribs Died 12 years previously from osteogenic sarcoma on Pagetic bone (right pelvis)
III-12 (M)b 64; 82 380 Left hipc Alive. Benign prostate hyperplasia
III-13 (M)b 65; 83 610 T5, T10 and L4 vertebral bodies, sacrum, right tibia, right femur, right shoulder and collarbone Alive. Allergy to pollen, hypertensive cardiopathy
IV-1 (F) 41 <120 None Alive, age 42, healthy
IV-13 (F) 53 <120 None Alive, hypertension, age 54
IV-14 (F) 41 <120 None Alive, age 42, lumbar–sacral discal hernia, goitre
IV-15 (M) 47 <120 None Alive, age 48 allergy to pollen
IV-16 (F) 48 <120 None Alive, age 49, allergy to pollen
The M404V mutation was ascertained in individuals listed in bold. The highest observed levels of alkaline phosphatase (AP) are reported for each affected subject; the normal range is less than 120 units/l. PDB, Paget's disease of bone.
aIndividuals strongly suspected to be potential PDB patients after careful reconstruction of the familial clinical history. bThese subjects received two treatment courses with oral risedronate (30 mg/day) for 3 months followed by a 112-day follow-up period without treatment [24]; complete normalization of serum AP levels and bone pain remission were observed in all these treated subjects. cTotal bone scintigraphy was not performed on this subject; the skeletal extent of PDB is on the basis of X-ray evaluations.
==== Refs
Cooper C Schafheutle K Dennison E Kellingray S Guyer P Barker D The epidemiology of Paget's disease in Britain: is the prevalence decreasing? J Bone Miner Res 1999 14 192 197 9933472
Laurin N Brown JP Morissette J Raymond V Recurrent mutation of the gene encoding sequestosome 1 (SQSTM1/p62) in Paget disease of bone Am J Hum Genet 2002 70 1582 1588 11992264 10.1086/340731
Hocking LJ Lucas GJ Daroszewska A Mangion J Olavesen M Cundy T Nicholson GC Ward L Bennett ST Wuyts W Domain-specific mutations in sequestosome 1 (SQSTM1) cause familial and sporadic Paget's disease Hum Mol Genet 2002 11 735 739 10.1093/hmg/11.22.2735
Johnson-Pais TL Wisdom JH Weldon KS Cody JD Hansen MF Singer FR Leach RJ Three novel mutations in SQSTM1 identified in familial Paget's disease of bone J Bone Miner Res 2003 18 1748 1753 14584883
Falchetti A Di Stefano M Marini F Del Monte F Mavilia C Strigoli D De Feo ML Isaia G Masi L Amedei A Two novel mutations at exon 8 of Sequestosome 1 gene (SQSTM1) in an Italian series of patients affected by Paget's disease of bone (PDB) J Bone Miner Res 2004 19 1013 1017 15125799
Geetha T Wooten MW Structure and functional properties of the ubiquitin binding protein p62 FEBS Lett 2002 512 19 24 11852044 10.1016/S0014-5793(02)02286-X
Hocking LJ Lucas GJ Daroszewska A Cundy T Nicholson GC Donath J Walsh JP Finlayson C Cavey JR Ciani B Novel UBA domain mutations of SQSTM1 in Paget's disease of bone: genotype phenotype correlation, functional analysis, and structural consequences J Bone Miner Res 2004 19 1122 1127 15176995
Eekhoff EW Karperien M Houtsma D Zwinderman AH Dragoiescu C Kneppers AL Papapoulos SE Familial Paget's disease in The Netherlands: occurrence, identification of new mutations in the sequestosome 1 gene, and their clinical associations Arthritis Rheum 2004 50 1650 1654 15146436 10.1002/art.20224
Good DA Busfield F Fletcher BH Lovelock PK Duffy DL Kesting JB Andersen J Shaw JT Identification of SQSTM1 mutations in familial Paget's disease in Australian pedigrees Bone 2004 35 277 282 15207768 10.1016/j.bone.2004.01.010
Zatloukal K Stumptner C Fuchsbichler A Heid H Schnoelzer M Kenner L Kleinert R Prinz M Aguzzi A Denk H p62 is a common component of cytoplasmic inclusions in protein aggregation diseases Am J Pathol 2002 160 255 263 11786419
Ciani B Layfield R Cavey JR Sheppard PW Searle MS Structure of the ubiquitin-associated domain of p62 (SQSTM1) and implications for mutations that cause Paget's disease of bone J Biol Chem 2003 278 37409 37412 12857745 10.1074/jbc.M307416200
Cavey JR Ralston SH Hocking LY Sheppard PW Ciani B Searle MS Layfield R Loss of ubiquitin-binding associated with Paget's disease of bone p62 (SQSTM1) mutations J Bone Miner Res 2005 4 619 624
Cornelis F Genetics and clinical practice in rheumatology Joint Bone Spine 2003 70 458 464 14667553 10.1016/j.jbspin.2003.09.005
Barry HC Paget's Disease of Bone 1969 London: E & S Livingstone
Huvos AG Butler A Bretsky SS Osteogenic sarcoma associated with Paget's disease of bone: a clinicopathologic study of 65 patients Cancer 1983 52 1489 1495 6577936
Nassar VH Gravanis MB Familial osteogenic sarcoma occurring in Pagetoid bone Am J Clin Pathol 1981 76 235 239 6943928
McNaim JDK Damron TA Landas SK Ambrose JL Shrimpton AE Inheritance of osteosarcoma and Paget's disease of bone J Mol Diagnost 2001 3 171 177
Nelissery MJ Padalecki SS Brkanac Z Singer FR Roodman GD Unni KK Leach RJ Hansen MF Evidence for a novel osteosarcoma tumor-suppressor gene in the chromosome 18q region genetically linked with Paget's disease of bone Am J Hum Genet 1998 63 817 824 9718349 10.1086/302019
Hughes AE Ralston SH Marken J Bell C MacPherson H Wallace RG van Hul W Whyte MP Nakatsuka K Hovy L Anderson DM Mutations in TNFRSF11A, affecting the signal peptide of RANK, cause familial expansile osteolysis Nat Genet 2000 24 45 48 10615125 10.1038/71667
Wuyts W van Wesenbeeck L Morales-Piga A Ralston SH Hocking L Vanhoenacker F Westhovens R Verbruggen L Anderson D Van Hul W Evaluation of the role of RANK and OPG genes in Paget's disease of bone Bone 2001 28 104 107 11165949 10.1016/S8756-3282(00)00411-7
Sparks AB Peterson SN Bell C Loftus BJ Hocking L Cahill DP Frassica FJ Streeten EA Levine MA Fraser CM Mutation screening of the TNFRSF11A gene encoding receptor activator of NFκB (RANK) in familial and sporadic Paget's disease of bone and osteosarcoma Calcif Tissue Int 2001 68 151 155 11351498 10.1007/s002230001211
Bubici C Papa S Pham CG Zazzeroni F Franzoso G NF-κB and JNK: An intricate affair Cell Cycle 2004 3 1524 1529 15611622
Pikarsky E Porat RM Stein I Abramovitch R Amit S Kasem S Gutkovich-Pyest E Urieli-Shoval S Galun E Ben-Neriah Y NF-κB functions as a tumour promoter in inflammation-associated cancer Nature 2004 431 461 466 15329734 10.1038/nature02924
Siris ES Chines AA Altman RD Brown JP Johnston CC JrLang R McClung MR Mallette LE Miller PD Ryan WG Risedronate in the treatment of Paget's disease of bone: an open label, multicenter study J Bone Miner Res 1998 13 1032 1038 9626635
|
16277682
|
PMC1297578
|
CC BY
|
2021-01-04 16:02:40
|
no
|
Arthritis Res Ther. 2005 Sep 15; 7(6):R1289-R1295
|
utf-8
|
Arthritis Res Ther
| 2,005 |
10.1186/ar1828
|
oa_comm
|
==== Front
Arthritis Res TherArthritis Research & Therapy1478-63541478-6362BioMed Central London ar18291627768510.1186/ar1829Research ArticleInduction of a B-cell-dependent chronic arthritis with glucose-6-phosphate isomerase Bockermann Robert [email protected] David [email protected] Thomas [email protected] Rikard [email protected] Section for Medical Inflammation Research, University of Lund, Lund, Sweden2 Deutsches Rheumaforschungszentrum Berlin, Berlin, Germany3 Deutsches Rheumaforschungszentrum Berlin, and Institut für Immunologie, Klinikum der FSU, Jena, Germany4 Section for Medical Inflammation Research, University of Lund, Lund, Sweden2005 20 9 2005 7 6 R1316 R1324 8 6 2005 18 7 2005 16 8 2005 26 8 2005 Copyright © 2005 Bockermann 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.
Antibodies specific for glucose-6-phosphate isomerase (G6PI) from T-cell receptor transgenic K/BxN mice are known to induce arthritis in mice, and immunization of DBA/1 mice with G6PI led to acute arthritis without permanent deformation of their joints. Because rheumatoid arthritis is a chronic disease, we set out to identify the capacity of G6PI to induce chronic arthritis in mice. Immunization with recombinant human G6PI induced a chronically active arthritis in mice with a C3H genomic background, whereas the DBA/1 background allowed only acute arthritis and the C57BL/10 background permitted no or very mild arthritis. The disease was associated with the major histocompatibility region sharing an allelic association similar to that of collagen-induced arthritis (i.e. q > p > r). All strains developed a strong antibody response to G6PI that correlated only in the C3H.NB strain with arthritis severity. Similarly, a weak response to type II collagen in a few mice was observed, which was associated with arthritis in C3H.NB mice. Mice on the C3H background also developed ankylosing spondylitis in the vertebrae of the tail. Both C3H.Q and B10.Q mice deficient for B cells were resistant to arthritis. We conclude that G6PI has the ability to induce a chronic arthritis, which is MHC associated and B-cell dependent. Thus, there are striking similarities between this and the collagen-induced arthritis model.
==== Body
Introduction
Glucose-6-phosphate isomerase (G6PI) is a widely expressed protein with multiple functions. It is an essential cytosolic enzyme in the energy cycle and has glycolytic activity, but it also has additional functions as an extracellular signalling molecule. Thus, G6PI is also known as AMF (autocrine motility factor) and neuroleukine, and may play roles in both cancer and autoimmunity [1,2].
Coincidentally, it was found that G6PI plays an essential role in the development of arthritis in mice. This originally stemmed from the observation that a bovine pancreas ribonuclease specific T-cell receptor transgenic mouse crossed with NOD mice (the so-called K/BxN mouse) spontaneously developed arthritis. Through a series of elegant experiments it was demonstrated that this transgenic T-cell receptor recognized G6PI within the context of major histocompatibility complex (MHC) class II molecule Ag7 [3,4]. The transgenic autoreactive T cells triggered autoreactive B cells to produce arthritogenic antibodies specific to G6PI [4-6]. After transferring G6PI-reactive serum from arthritic K/BxN mice, these antibodies bound to peripheral joints and induced arthritis in a manner strikingly similar to that shown previously for antibodies to the cartilage-specific antigen collagen type II (CII) [7,8]. B-cell activation in response to G6PI appeared to occur primarily in lymph nodes draining the joints [9], indicating that recognition of G6PI is joint specific. The reason for this specificity is not apparent because the G6PI protein is a ubiquitous protein.
Even though there are inconsistent data regarding the role of G6PI in rheumatoid arthritis (RA) [10-13], it appears that antibodies to G6PI occur predominantly in patients with Felty's syndrome – a variant of RA [14]. It is still unclear whether this is a unique phenomenon of Felty's syndrome or whether it reflects a generalized higher autoreactivity in these patients. Support for the latter comes from a study using affinity purified sera from arthritis patients that compared different kinds of arthritides and suggested a role for G6PI antibodies [15]. G6PI is nevertheless an interesting autoantigen and may represent a unique pathway leading to aggressive subtypes of RA. It is thus interesting to investigate this model further and to compare it with other models, such as the classic collagen-induced arthritis (CIA) model [16]. Here we show that immunization with G6PI leads to a chronically active arthritis in mice with genes from the C3H background, and that susceptibility is both controlled by the MHC region and dependent on B cells.
Materials and methods
Mice
The mouse strains were bred and used in the animal facility of the Section for Medical Inflammation Research (University of Lund, Lund, Sweden). Mice of DBA/1J and C3H.NB origins were from Jackson Laboratories (Bar Harbor, Maine, USA), and those of B10.Q, B10.P and B10.RIII origins were from Dr Jan Klein (Max-Planck-Institut für Biologie, Abteilung Immungenetik, Tübingen, Germany). C3H.Q mice were established through backcrossing (4n) of the H2q fragment derived from an original C3H.Q mouse into C3H.NB [17]. The human DR4 transgenic and the backcrossed B-cell deficient mouse strains were described previously [18-20].
Experimental mice were matched for sex and age in all experiments. The founder μMT mouse was kindly provided by Dr Werner Müller (Institute of Genetics, Cologne, Germany), which we backcrossed to B10.Q for 13 generations before they were intercrossed for the experiment. C3H.Q μMT mice were backcrossed for 10 generations and finally intercrossed.
The experiments were conducted in accordance with guidelines from the Swedish Ethical Committee.
Antigens
Recombinant human G6PI was produced as previously described [21]. G6PI cDNA fragments were introduced into a modified pQE100 expression vector for expression of His-tagged proteins in Escherichia coli strain Bl21. Supernatants of bacterial lysates were subjected to purification over a Ni-NTA column (Qiagen, Hilden, Germany), in accordance with the manufacturer's instructions. Using the same strategy, human creatine kinase (hCK) and mouse G6PI (mG6PI) was produced. The purity of proteins was checked using standard SDS gels. Rat collagen type II (rCII) was prepared from SWARM chondrosarcoma by pepsin digestion [22] and further purified as described previously [23].
Immunization and arthritis scoring
An emulsion for immunization was made by sonication, using an aliquot of human G6PI and complete Freund's adjuvant (Difco, Detroit, MI, USA), resulting in an emulsion of 2 mg/ml human G6PI. A total of 100 μl of the emulsion (200 μg G6PI) was injected at the base of the tail in each mouse. For the titration experiments G6PI was diluted with phosphate-buffered saline to adjust to the required concentration. In some experiments mice were given an intradermal boost with 50 μg G6PI in incomplete Freund's adjuvant. Mice were visually scored for arthritis using an extended scoring protocol ranging from 1 to 15 for each paw, allowing a maximum score of 60 per mouse. Each arthritic (red and swollen) toe and knuckle was scored as 1, whereas an affected ankle was scored as 5 (total: 15/paw) [24].
Antibody analysis
Serum for analysis of antibody levels was taken at indicated time points and at the end of all experiments. Serum was diluted 1:1,000 for G6PI, mG6PI, hCK and 1:100 for rCII antibody analysis. ELISA Maxisorp plates (Nunc, Roskilde Denmark) were coated with 50 μl of 10 μg/ml of the recombinant proteins or rat CII. The amounts of total specific IgG was determined through quantitative ELISA using peroxidase-conjugated goat anti-mouse IgG (H+L; 115-035-062; Jackson ImmunoResearch, West Grove, PA, USA) secondary antibodies [25]. ABTS (2,2'-Azino-bis(3-Ethylbenzthiazoline-6-Sulfonic Acid), # 11204521001; Roche Diagnostics GmbH, Penzberg, Germany) was used as substrate. Values were measured at 405 nm and are expressed as optical density values.
Histology
At the end of the experiments paws, knees, and tails were fixated in 4% paraformaldehyde for 24 hours and decalcified with EDTA. The paraffin sections were stained with haematoxylin and erythrosine [26].
Statistical analysis
Frequency of arthritis was analyzed using the χ2 test, and antibody levels and arthritis severity were analyzed using the Mann-Whitney U-test. Disease score and antibody correlations were analyzed using the Spearman rho correlation test from the StatView software package (Version 5.0.1, SAS Institute Inc., Cary, NC, USA).
Results
Titration of the arthritogenic dose of G6PI
DBA/1 mice were used to confirm induction of arthritis using human recombinant G6PI in our animal house and to titrate the dose. Almost 100% of the mice developed arthritis upon immunization with all of the doses used, although the severity was dose dependent. With the lowest dose (100 μg) the arthritis started as early as day 9 and subsequently progressed to a severe arthritis peaking 2 weeks after immunization.
Thereafter the disease gradually resolved and no macroscopic signs of arthritis were apparent at day 40 (Fig. 1a), enabling mice to climb under the lids of their cages. The intermediate dose of 200 μg per mouse, which resulted in arthritis in 100%, was selected for further study.
Pronounced genetic control of chronic arthritis involving both MHC and non-MHC genes
G6PI immunization of different mouse strains resulted in marked differences in arthritis susceptibility and severity. The C3H.NB strain developed arthritis approximately 2 days later than did DBA/1 mice. However, the arthritis in C3H.NB mice was more severe and, most importantly, these mice developed a chronically active inflammation that lasted throughout the observation period of 90 days (Fig. 1b).
Histological analysis of joints at 90 days after immunization showed active erosive inflammation (Fig. 2e,j). It should be pointed out that only active inflammatory arthritis with redness and oedema was evaluated for clinical scoring. Even though the oedema declined over time in C3H mice, these animals still had tissue depositions that rendered the joints dysfunctional. There was also massive cell infiltration of the joint space causing erosion and destruction of bone and cartilage, as demonstrated by histology (Fig. 2e,j). The destructive character of this process was striking; for instance, it was even able to dissolve joint cartilage. Simultaneously, new bone formation could be observed, creating large osteophytes. In contrast, DBA/1 mice regained function of many finger joints without major bone remodelling after the inflammation went into remission (Fig. 2d,i), even though this strain is known to be prone to development of osteophytes [27].
The inflammatory process, in contrast to CIA, was not limited to synovial joints but also eroded vertebra and the annulus fibrosus, together with the nucleus pulposus in mice of the C3H background (Fig. 2g). Healthy vertebrae are shown in Fig. 2f. B10.P mice, which share the MHC region with the highly susceptible C3H.NB mice, were resistant to arthritis, showing the strong influence of non-MHC genes. The MHC congenic B10.Q strain, on the other hand, developed significant but mild acute arthritis, whereas another MHC congenic strain – B10.RIII – was also totally resistant to joint destruction. It should be noted that the B10.Q strain used in this study is different from the B10.Q mouse available from Jackson Laboratories, which has an arthritis-protective mutation of the Tyk2 gene [28], which explains the earlier reported resistance in these animals to G6PI-induced arthritis [21]. The (B10.Q × DBA/1)F1 mice developed arthritis almost as severe as that in DBA/1 mice, showing that part of the genetic contribution from DBA/1 dominates the suppressive B10.Q background genes. Mice expressing the human DR4 (0401) molecule on the B10 background were resistant to arthritis. These observations indicate that the most susceptible MHC haplotype is H2q, which is similar to earlier observations in the CIA model [29,30].
Because the highly susceptible C3H.NB strain harboured the less susceptible MHC haplotype (H2p), we tested it in comparison with the C3H.Q strain, which is a MHC congenic strain that carries the H2q haplotype. The C3H.Q strain developed slightly more chronic arthritis than the C3H.NB mice, although the difference between the strains in single experiments did not always achieve statistical significance because of the high severity of arthritis in both strains (Fig. 1c).
A booster immunization after resolution of arthritis induced a relapse in most strains (Fig. 1b), but this was milder than the first arthritic episode and started at the exact same time after immunization, suggesting that there is no memory effect from the primary immunization.
Development of G6PI arthritis is associated with a strong antibody response to G6PI and a weak response to type II collagen
All arthritis susceptible mouse strains developed a strong antibody response to human G6PI (hG6PI; Fig. 3a). However, the hG6PI specific antibody response did not always correlate with arthritis severity; strong correlation was found only in the C3H.NB strain (Table 1). The anti-G6PI antibody response made use of all IgG isotypes (data not shown). ELISA plates were also coated with recombinant mouse G6PI because human G6PI was used for immunization. The antibody responses were very similar using the two proteins (Fig. 3a,b). To exclude the His-tag as an allogeneic B-cell epitope, we also investigated the antibody response against a His-tag fusion protein of hCK (Fig. 3d) and His-tag labelled recombinant Aq as negative controls (data not shown). No significant response for these antigens could be detected. That the response was truly directed against conserved G6PI epitopes was also confirmed by using tissue purified commercial (Sigma-Aldrich Sweden AB, Stockholm, Sweden) rabbit G6PI (data not shown).
Nevertheless, many of the strains used, such as B10.RIII and B10.DR4, did not develop arthritis even though they exhibited strong antibody responses against G6PI, indicating the presence of other protective factors. Interestingly, arthritis susceptible strains developed significant titres to CII (Fig 3c), but again only the C3H.NB mice exhibited a positive correlation between anti-CII antibody response and arthritis severity (Table 1).
Development of G6PI-induced arthritis is B-cell dependent
To determine conclusively whether B cells play a critical role in the pathology of G6PI-induced arthritis we used mice with a disrupted IgM gene, which are therefore deficient in mature B cells. No arthritis developed in B-cell-deficient B10.Q mice (Fig. 4a). C3H.Q μMT mice developed only very mild oedema for no longer than 2 days (Fig. 4b), leaving no histological changes (data not shown).
Discussion
Immunization with G6PI induces arthritis of various degrees of severity and chronicity, depending on the mouse strain. Interestingly, the C3H genetic background permits a chronically active disease course that leads to loss of joint function. This is an important feature of an animal model of RA because the human disease is already chronic when it becomes diagnosed. RA is most likely often preceded by many years of subclinical inflammatory activity. This is not only reflected by raised C-reactive protein levels but also by the production of autoantibodies such as rheumatoid factors and antibodies to citrullinated proteins [31-33]. The chronic disease course of arthritis in C3H.Q and C3H.NB mice, induced with G6PI, will be useful in the analysis of mechanisms of chronicity and as a model to develop new therapeutic protocols.
Interestingly, the C3H background also allows a more severe CIA [16,34]. In addition, in both models, the DBA/1 mouse develops a severe but acute and self-limited type of arthritis. Another striking similarity between the G6PI model and CIA is the association with MHC. In both models the H2q haplotype confers a more severe form of arthritis than does H2p. In the CIA model this difference has been shown to be due to the Aq molecule, which binds the immunodominant CII260–270 glycopeptide with greater affinity than the corresponding Ap molecule [35]. It would therefore be of interest to identify the G6PI peptide that binds to Aq and to investigate its affinity to the different MHC molecules, in analogy to the CII peptide. However, a difference from CIA is that the H2r haplotype, despite a strong anti-G6PI antibody response, does not confer susceptibility to G6PI-induced arthritis, although in the CIA model the association with H2r is dependent on binding and recognition of peptides other than 260–270 [36]. Another apparent difference is that DR4 (DRB1*0401/DRA) expressing mice are susceptible to CIA [19,37] but not to G6PI-induced arthritis. This is possibly due to a threshold effect, in which the mice developed a strong autoimmune response to G6PI but which, combined with the relative nonpermissive B10 background, did not lead to arthritis. It may not be unexpected that G6PI-induced arthritis is critically dependent on functional B cells, as shown by our findings in B-cell deficient μMT mice on the B10 and the C3H backgrounds. However, it is interesting that a mild transient oedema was observed in B-cell-deficient mice on the highly arthritis susceptible C3H background.
It has been shown in the K/BxN transgenic model that anti-G6PI serum antibodies readily transfer arthritis [4,6]. In the protein-induced G6PI model, this has so far not been demonstrated [21], and the antibody titres in the different strains do not exhibit a convincing correlation with arthritis susceptibility. Thus, high levels of antibodies to G6PI do not always lead to arthritis, indicating that other pathogenic factors play a role. Interestingly, a few mice with severe arthritis developed detectable amounts of antibodies to CII. There is no evidence for a cross-reactivity between G6PI and CII, and the most likely explanation is an activation of an autoimmune response to cartilage-derived CII, as has been seen in pristine-induced arthritis and in various spontaneous arthritides in mice [38-41].
On the C3H background the G6PI model is also useful in investigating ankylosing spondylitis because it generates inflammation of vertebral joints followed by ankylosis after a single round of immunization. Careful serum analysis of patients with different forms of arthritides revealed that G6PI-specific antibodies may be identified not only in severe forms of RA but also in ankylosing spondylitis and Reiter's syndrome [15]. The G6PI-induced arthritis model on the C3H background demonstrates that an initial G6PI immune response is sufficient to induce destructive activity in the spine. Which C3H genetic factors actually contribute to this arthritis pathway remain to be determined, and this needs a careful and cautious analysis of the precise genetic background of any mice used.
Our work over many years with congenic and transgenic mice has made us aware of pitfalls relating to the purity of genetic backgrounds. One should be careful in extrapolating data without having full control over the genetic backgrounds. It is not too surprising that, for instance, the C3H.He mice used by Ji and coworkers [42] did not exhibit high sensitivity for their serum transfer model, as might be suggested by our results. Not only does the C3H.He mouse from Jackson, used by those investigators, has a defect in the Toll-like receptor 4 that renders it unresponsive to lipopolysaccharide stimulation, but also it carries another MHC haplotype (H2-K) compared with our congenic mice. Furthermore, it is likely that the different MHC congenic inbred strains have accumulated mutations over the years, as well as carrying several contaminating fragments due to incomplete backcrossing. Bearing these problems in mind, we backcrossed our C3H.Q mice for several generations to C3H.NB to be sure that we compared the MHC effect only. Therefore, it will be of great interest in future investigations to use a panel of highly controlled congenic mice to identify chronicity factors in C3H mice.
In an examination of the IgG isotypes active in human Reiter's syndrome, they appeared to be predominantly of the T-helper-2-like isotype IgG4, equivalent to IgG1 in K/BxN mice [15]. It will be interesting to investigate whether a T-helper-2 driven immune response is responsible for the chronic severity and spine involvement with the C3H background.
Taken together, there are several similarities but also differences between G6PI-induced arthritis and CIA. Most strikingly, the genetic control in the two models allows only acute arthritis in DBA/1 mice but a more chronic relapsing form in mice of the C3H background. In the CIA model the B10 background allows chronic development of arthritis [43]. Interestingly, both models appear to follow a central pathogenic pathway that involves B-cell autoreactivity and arthritogenic antibodies. Experiments using antibodies to CII and G6PI over the years have shown extensive similarities [6,8,42,44-49].
The most obvious difference between the two models is the tissue distribution of the autoantigen. G6PI is systemically distributed in the body because it is expressed intracellular in all cells as an enzyme of glycolysis and can furthermore be secreted. CII is also widely expressed during foetal development (for review see Holmdahl and coworkers [50]). In the adult CII expression is more restricted, mainly to cartilage (e.g. in diarthrodial joints, larynx, spine and sternum). It is also expressed in the vitreous body of the eye [50]. G6PI-induced arthritis developed much earlier after immunization than did CIA and with a stronger oedematous appearance, which could extend into the knees, although no prominent histological changes were observed in DBA/1 mice. Thus, in both CIA and G6PI-induced arthritis the tissue distribution of the autoantigen could not account for the specificity of the inflammatory disease. In fact, models with unknown autoantigens, like induction of arthritis with the alkane pristane, are also specific in that the resulting inflammation only affects joints [51,52].
One important question to address is where the immune system detects the autoantigen. The antigen might be differently expressed in various tissues and be processed differently depending on the kind of antigen-presenting cell. Another issue to address is the role played by synovial tissue in diarthrodial joints because these joints are predominantly affected. The cartilage surface may be of importance for triggering antibody-mediated inflammation, as shown for both anti-CII and anti-G6PI antibodies. Taken together, we believe that G6PI-induced arthritis is a very useful model for studies of RA and it may represent a unique pathway, in particular with respect to its autoantigen specificity and chronicity.
Conclusion
This study showed that G6PI-induced arthritis can be converted into a chronic inflammatory arthritis model by using the C3H genetic background. Mice of the C3H background also develop arthritis in their vertebra, supporting a role for G6PI reactivity in ankylosing spondylitis.
We conclude that G6PI has the ability to induce a chronic from of arthritis, which is MHC associated and B-cell dependent. Thus, there is a striking similarity between G6PI-induced arthritis and the CIA model. Genetic factors determining chronicity – a hallmark of RA – will be addressed using this model in future experiments.
Abbreviations
CIA = collagen-induced arthritis; CII = collagen type II; ELISA = enzyme-linked immunosorbent assay; G6PI = glucose-6-phosphate isomerase; hCK = human creatine kinase; MHC = major histocompatibility complex; RA = rheumatoid arthritis; rCII = rat collagen type II.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
RB performed the experiments, and was involved in designing the study and in writing the manuscript under the guidance of RH. DS and TK produced the recombinant proteins (hG6PI, mG6Pi and hCK) and were involved in designing the study and critically read the manuscript.
Acknowledgements
We thank Emma Mondoc for technical support for the histology and Carlos Palestro, Isabell Bohlin, Sandy Liedholm, Rebecka Ljungqvist and Alexandra Treschow-Bäcklund for husbandry of the animals.
This work was supported by grants from the Swedish Research Council, the Swedish Strategic Research Council, the Swedish Rheumatism Association, Kock and Österlund Foundations, Gustav V: s Foundation, and Crafoord Foundation.
Figures and Tables
Figure 1 Characterization of G6PI-induced arthritis. Shown are characteristics of glucose-6-phosphate isomerase (G6PI)-induced arthritis in different genetic backgrounds and major histocompatibility complex (MHC) congenics. (a) DBA/1 mice were immunized intradermally at the base of the tail with the indicated amounts of G6PI emulsified in complete Freund's adjuvant (CFA) to establish a dynamic immunization protocol allowing for an increase or decrease in disease severity. The course of disease was followed for 40 days after immunization. The graph shows the mean scores for all mice. Disease developed in 8/8 (400 μg/mouse), 9/9 (200 μg/mouse) and 8/9 (100 μg/mouse) mice. (b) After establishing the immunization protocol, mice with different MHC haplotypes and genetic backgrounds were immunized with 200 μg G6PI in CFA at the base of the tail. Active arthritis, characterized by redness and oedema, was scored over 90 days after immunization. Blood was drawn at day 40 for antibody analysis and the mice were boosted with human G6PI (50 μg/mouse in incomplete Freund's adjuvant) at day 48. The numbers of mice of each strain evaluated were as follows: (B10.Q × DBA/1)F1, n = 10; B10-Tg(DR4), n = 8; B10.P, n = 12; B10.Q, n = 26; B10.RIII, n = 6; C3H.NB, n = 9; and DBA/1, n = 10. (c) Because the MHC haplotype H-2p on the black background rendered mice resistant to G6PI-induced arthritis, the role of the beta chain of Ap was addressed on the highly susceptible C3H background. C3H.NB (H-2p; n = 21) and C3H.Q (H-2q; n = 19) mice were immunized intradermally with 200 μg G6PI in CFA and scored for 73 days. At no time point was a significant difference noticed between the two MHC congenic strains. Only a tendency toward more chronic progression could be seen in the C3H.Q mice during the late phase. In all experiments, error bars indicate the standard error of the mean.
Figure 2 Clinical and histological evaluation of arthritis. Clinical and histological evaluation demonstrates that arthritis induced with 200 μg glucose-6-phosphate isomerase (G6PI) in complete Freund's adjuvant (CFA) leads to chronic destructive arthritis in mice on the C3H background. (a) Healthy C3H.Q hind foot. (b) A C3H.Q hind foot 90 days after disease induction. The digits are still red and swollen. After day 90 paws were fixated and decalcified for paraffin sectioning. Histopathology demonstrates the destructive character of the GPI-induced arthritis in C3H in comparison with DBA/1 mice. Both mice achieved clinical scores in their hind feet of 15. The C3H mouse developed (e) an irreversible destruction of their joints through invasive pannus tissue accompanied by new bone formation, (j) destroying the whole architecture of the ankle, whereas DBA/1 mice have relatively intact joints, apart from (d) smaller erosions (arrows) and (i) hyperplasia. (c,h) Healthy control joints. The severity of the disease on the C3H background is also indicated by (g) the destruction of intervertebral structures such as the annulus fibrosus, nucleus pulposus and the vertebra themselves by inflammatory cells (arrow). (f) Healthy control tail. Staining with haematoxylin and erythrosine; original magnification 25× and 100×. af, annulus fibrosus; np, nucleus pulposus; o, osteophytes; pa, pannus; sy, synovial membrane; ta, talus; ti, tibia.
Figure 3 Antibody analysis. Indicated mouse strains were immunized with 200 μg human glucose-6-phosphate isomerase (hG6PI) in complete Freund's adjuvant and bled at day 40 for antibody analysis. ELISA plates were coated with (a) 10 μg/ml hG6PI, (b) mouse G6PI (mG6PI), (c) collagen type II (CII), or (d) human creatine kinase (hCK). Sera from nonimmunized mice (n = 5) of different genetic backgrounds were used as negative controls. The figures show the optical density (OD) value for total IgG responses at a serum dilution of 1:1,000 for hG6PI, mG6PI and hCK (panels a, b and d) and 1:100 for CII (panel c). The results are represented as box plots, indicating the median, the 25th and 75th centiles as boxes, and the 10th and 90th centiles as whiskers. Outliers are indicated as circles.
Figure 4 Role of B cells in the pathology of G6PI-induced arthritis. The role of B cells in the pathology of glucose-6-phosphate isomerase (G6PI)-induced arthritis was addressed using B-cell deficient μMT mice on B10.Q and C3H.Q backgrounds. Mice aged between 6 and 7 weeks were immunized with 200 μg G6PI in complete Freund's adjuvant and scored for arthritis. (a) Five out of seven B10.Q mice developed arthritis, whereas none of the six B10.Q μMT mice exhibited signs of inflammation. (b) Four out of 12 C3H.Q μMT mice exhibited mild oedema for no longer than 48 hours, whereas all nine C3H.Q littermate control mice demonstrated a destructive inflammatory disease course. Error bars indicate the standard error of the mean.
Table 1 Correlation between specific IgG-total and accumulative score
Strain Anti-hG6PI versus score Anti-CII versus score n
Rho P value Rho P value
C3H.NB 0.917* 0.009* 0.793* 0.026* 9
B10.Q 0.402* 0.046* -0.094 0.614 26
(B10.Q × DBA1)F1 0.018 0.956 0.139 0.67 10
DBA/1 0.317 0.370 0.033 0.924 10
To investigate correlations between arthritis severity and antibody production the Spearman correlation test was applied. Mice of indicated strains were immunized with 200 μg glucose-6-phosphate isomerase (G6PI) in complete Freund's adjuvant. Blood was drawn at day 40 and analyzed by ELISA for anti-hG6PI and anti-CII total IgG responses, as shown in Fig. 1. The accumulative arthritis score until day 40 was tested for correlation with antibody production using the Spearman rank correlation test. A rho value close to 1 indicates correlation of high ranks for IgG with high ranks for arthritis scores; 0 indicates that there is no correlation between values; and a number close to -1 indicates that high ranks for one variable correlate with low ranks for the other. *Significant positive correlations.
==== Refs
Sun YJ Chou CC Chen WS Wu RT Meng M Hsiao CD The crystal structure of a multifunctional protein: phosphoglucose isomerase/autocrine motility factor/neuroleukin Proc Natl Acad Sci USA 1999 96 5412 5417 10318897 10.1073/pnas.96.10.5412
Chou CC Sun YJ Meng M Hsiao CD The crystal structure of phosphoglucose isomerase/autocrine motility factor/neuroleukin complexed with its carbohydrate phosphate inhibitors suggests its substrate/receptor recognition J Biol Chem 2000 275 23154 23160 10770936 10.1074/jbc.M002017200
Korganow AS Ji H Mangialaio S Duchatelle V Pelanda R Martin T Degott C Kikutani H Rajewsky K Pasquali L From systemic T cell self reactivity to organ-specific autoimmune disease via immunoglobulins Immunity 1999 10 451 461 10229188 10.1016/S1074-7613(00)80045-X
Matsumoto I Staub A Benoist C Mathis D Arthritis provoked by linked T and B cell recognition of a glycolytic enzyme Science 1999 286 1732 1735 10576739 10.1126/science.286.5445.1732
Matsumoto I Maccioni M Lee DM Maurice M Simmons B Brenner M Mathis D Benoist C How antibodies to a ubiquitous cytoplasmic enzyme may provoke joint-specific autoimmune disease Nat Immunol 2002 3 360 365 11896391 10.1038/ni772
Maccioni M Zeder-Lutz G Huang H Ebel C Gerber P Hergueux J Marchal P Duchatelle V Degott C van egenmortel M Arthritogenic monoclonal antibodies from K/BxN mice J Exp Med 2002 195 1071 1077 11956298 10.1084/jem.20011941
Stuart JM Dixon FJ Serum transfer of collagen induced arthritis in mice J Exp Med 1983 158 378 392 6886622 10.1084/jem.158.2.378
Holmdahl R Rubin K Klareskog L Larsson E Wigzell H Characterization of the antibody response in mice with type II collagen-induced arthritis, using monoclonal anti-type II collagen antibodies Arthritis Rheum 1986 29 400 410 2421741
Mandik-Nayak L Wipke BT Shih FF Unanue ER Allen PM Despite ubiquitous autoantigen expression, arthritogenic autoantibody response initiates in the local lymph node Proc Natl Acad Sci USA 2002 99 14368 14373 12391319 10.1073/pnas.182549099
Schaller M Burton DR Ditzel HJ Autoantibodies to GPI in rheumatoid arthritis: linkage between an animal model and human disease Nat Immunol 2001 2 746 753 11477412 10.1038/90696
Schubert D Schmidt M Zaiss D Jungblut PR Kamradt T Autoantibodies to GPI and creatine kinase in RA Nat Immunol 2002 3 411 11976711 10.1038/ni0502-411a
Kassahn D Kolb C Solomon S Bochtler P Illges H Few human autoimmune sera detect GPI Nat Immunol 2002 3 411 412 11976712 10.1038/ni0502-411b
Matsumoto I Lee DM Goldbach-Mansky R Sumida T Hitchon CA Schur PH Anderson RJ Coblyn JS Weinblatt ME Brenner M Low prevalence of antibodies to glucose-6-phosphate isomerase in patients with rheumatoid arthritis and a spectrum of other chronic autoimmune disorders Arthritis Rheum 2003 48 944 954 12687536 10.1002/art.10898
van Gaalen FA Toes RE Ditzel HJ Schaller M Breedveld FC Verweij CL Huizinga TW Association of autoantibodies to glucose-6-phosphate isomerase with extraarticular complications in rheumatoid arthritis Arthritis Rheum 2004 50 395 399 14872481 10.1002/art.20028
Schaller M Stohl W Tan SM Benoit VM Hilbert DM Ditzel HJ Raised levels of anti-glucose-6-phosphate isomerase IgG in serum and synovial fluid from patients with inflammatory arthritis Ann Rheum Dis 2005 64 743 749 15458959 10.1136/ard.2004.025502
Holmdahl R Andersson M Goldschmidt TJ Gustafsson K Jansson L Mo JA Type II collagen autoimmunity in animals and provocations leading to arthritis Immunol Rev 1990 118 193 232 2079326
Kjellen P Jansson L Vestberg M Andersson AA Mattsson R Holmdahl R The H2-Ab gene influences the severity of experimental allergic encephalomyelitis induced by proteolipoprotein peptide 103–116 J Neuroimmunol 2001 120 25 33 11694316 10.1016/S0165-5728(01)00407-6
Svensson L Jirholt J Holmdahl R Jansson L B cell-deficient mice do not develop type II collagen-induced arthritis (CIA) Clin Exp Immunol 1998 111 521 526 9528892 10.1046/j.1365-2249.1998.00529.x
Andersson EC Hansen BE Jacobsen H Madsen LS Andersen CB Engberg J Rothbard JB Sönderstrup-McDevitt G Malmström V Holmdahl R Definition of MHC and T cell receptor contacts in the HLA-DR4 restricted immunodominant epitope in type II collagen and characterization of collagen-induced arthritis in HLA-DR4 and human CD4 transgenic mice Proc Natl Acad Sci USA 1998 95 7574 7569 9636191 10.1073/pnas.95.13.7574
Bäcklund J Carlsen S Höger T Holm B Fugger L Kihlberg J Burkhardt H Holmdahl R Predominant selection of T cells specific for glycosylated collagen type II peptide (263–270) in humanized transgenic mice and in rheumatoid arthritis Proc Natl Acad Sci USA 2002 99 9960 9965 12089323 10.1073/pnas.132254199
Schubert D Maier B Morawietz L Krenn V Kamradt T Immunization with glucose-6-phosphate isomerase induces T cell-dependent peripheral polyarthritis in genetically unaltered mice J Immunol 2004 172 4503 4509 15034067
Smith BD Martin GR Dorfman A Swarm R Nature of the collagen synthesized by a transplanted chondrosarcoma Arch Biochem Biophys 1975 166 181 186 1122133 10.1016/0003-9861(75)90378-1
Andersson M Holmdahl R Analysis of type II collagen reactive T cells in the mouse. I. Different regulation of autoreactive versus non-autoreactive anti-type II collagen T cells in the DBA/1 mouse Eur J Immunol 1990 20 1061 1066 1694131
Holmdahl R Animal models for rheumatoid arthritis useful for drug screening and evaluation Curr Res Rheum Arthr 1998 2 115 121
Holmdahl R Jansson L Larsson E Rubin K Klareskog L Homologous type II collagen induces chronic and progressive arthritis in mice Arthritis Rheum 1986 29 106 113 3947407
Holmdahl R Jonsson R Larsson P Klareskog L Early appearence of activated CD4 positive T lymphocytes and Ia-expressing cells in joints of DBA/1 mice immunized with type II collagen Lab Invest 1988 58 53 60 3257283
Corthay A Hansson AS Holmdahl R T lymphocytes are not required for the spontaneous development of entheseal ossification leading to marginal ankylosis in the DBA/1 mouse Arthritis Rheum 2000 43 844 851 10765929 10.1002/1529-0131(200004)43:4<844::AID-ANR15>3.0.CO;2-B
Shaw MH Boyartchuk V Wong S Karaghiosoff M Ragimbeau J Pellegrini S Muller M Dietrich WF Yap GS A natural mutation in the Tyk2 pseudokinase domain underlies altered susceptibility of B10.Q/J mice to infection and autoimmunity Proc Natl Acad Sci USA 2003 100 11594 11599 14500783 10.1073/pnas.1930781100
Wooley PH Luthra HS Stuart JM David CS Type II collagen induced arthritis in mice. I. Major histocompatibility complex (I-region) linkage and antibody correlates J Exp Med 1981 154 688 700 6792316 10.1084/jem.154.3.688
Brunsberg U Gustafsson K Jansson L Michaëlsson E Ährlund-Richter L Pettersson S Mattsson R Holmdahl R Expression of a transgenic class II Ab gene confers susceptibility to collagen-induced arthritis Eur J Immunol 1994 24 1698 1702 8026530
Aho K Palosuo T Raunio V Puska P Aromaa A Salonen JT When does rheumatoid disease start? Arthritis Rheum 1985 28 485 489 4004958
Rantapaa-Dahlqvist S de Jong BA Berglin E Hallmans G Wadell G Stenlund H Sundin U van Venrooij WJ Antibodies against cyclic citrullinated peptide and IgA rheumatoid factor predict the development of rheumatoid arthritis Arthritis Rheum 2003 48 2741 2749 14558078 10.1002/art.11223
van Gaalen FA Linn-Rasker SP van Venrooij WJ de Jong BA Breedveld FC Verweij CL Toes RE Huizinga TW Autoantibodies to cyclic citrullinated peptides predict progression to rheumatoid arthritis in patients with undifferentiated arthritis: a prospective cohort study Arthritis Rheum 2004 50 709 715 15022309 10.1002/art.20044
Bäcklund J Nandakumar KS Bockermann R Mori L Holmdahl R Genetic control of tolerance to type II collagen and development of arthritis in an autologous collagen-induced arthritis model J Immunol 2003 171 3493 3499 14500645
Kjellén P Brunsberg U Broddefalk J Hansen B Vestberg M Ivarsson I Engström Å Svejgaard A Kihlberg J Fugger L The structural basis of MHC control of collagen-induced arthritis; binding of the immunodominant type II collagen 256–270 glycopeptide to H-2Aq and H-2Ap molecules Eur J Immunol 1998 28 755 767 9521085 10.1002/(SICI)1521-4141(199802)28:02<755::AID-IMMU755>3.0.CO;2-2
Myers LK Miyahara H Terato K Seyer JM Stuart JM Kang AH Collagen-induced arthritis in B10.RIII mice (H-2r): identification of an arthritogenic T-cell determinant Immunology 1995 84 509 513 7540590
Rosloniec EF Brand DD Myers LK Esaki Y Whittington KB Zaller DM Woods A Stuart JM Kang AH Induction of autoimmune arthritis in HLA-DR4 (DRB1*0401) transgenic mice by immunization with human and bovine type II collagen J Immunol 1998 160 2573 2578 9510154
Wooley PH Seibold JR Whalen JD Chapdelaine JM Pristane-induced arthritis. The immunologic and genetic features of an experimental murine model of autoimmune disease Arthritis Rheum 1989 32 1022 1030 2765002
Hultqvist M Olofsson P Holmberg J Bäckström BT Tordsson J Holmdahl R Enhanced autoimmunity, arthritis, and encephalomyelitis in mice with a reduced oxidative burst due to a mutation in the Ncf1 gene Proc Natl Acad Sci USA 2004 101 12646 12651 15310853 10.1073/pnas.0403831101
Sakaguchi N Takahashi T Hata H Nomura T Tagami T Yamazaki S Sakihama T Matsutani T Negishi I Nakatsuru S Sakaguchi S Altered thymic T-cell selection due to a mutation of the ZAP-70 gene causes autoimmune arthritis in mice Nature 2003 426 454 460 14647385 10.1038/nature02119
Kotani M Tagawa Y Iwakura Y Involvement of autoimmunity against type II collagen in the development of arthritis in mice transgenic for the human T cell leukemia virus type I tax gene Eur J Immunol 1999 29 54 64 9933086 10.1002/(SICI)1521-4141(199901)29:01<54::AID-IMMU54>3.0.CO;2-M
Ji H Gauguier D Ohmura K Gonzalez A Duchatelle V Danoy P Garchon HJ Degott C Lathrop M Benoist C Genetic influences on the end-stage effector phase of arthritis J Exp Med 2001 194 321 330 11489951 10.1084/jem.194.3.321
Svensson L Nandakumar KS Johansson A Jansson L Holmdahl R IL-4-deficient mice develop less acute but more chronic relapsing collagen-induced arthritis Eur J Immunol 2002 32 2944 2953 12355448 10.1002/1521-4141(2002010)32:10<2944::AID-IMMU2944>3.0.CO;2-4
Ji H Ohmura K Mahmood U Lee DM Hofhuis FM Boackle SA Takahashi K Holers VM Walport M Gerard C Arthritis critically dependent on innate immune system players Immunity 2002 16 157 168 11869678 10.1016/S1074-7613(02)00275-3
Stuart JM Tomoda K Yoo TJ Townes AS Kang AH Serum transfer of collagen-induced arthritis. II. Identification and localization of autoantibody to type II collagen in donor and recipient rats Arthritis Rheum 1983 26 1237 1244 6626283
Terato K Hasty KA Reife RA Cremer MA Kang AH Stuart JM Induction of arthritis with monoclonal antibodies to collagen J Immunol 1992 148 2103 2108 1545120
Hietala MA Nandakumar KS Persson L Fahlen S Holmdahl R Pekna M Complement activation by both classical and alternative pathways is critical for the effector phase of arthritis Eur J Immunol 2004 34 1208 1216 15048732 10.1002/eji.200424895
Nandakumar KS Svensson L Holmdahl R Collagen type II specific monoclonal antibody induced arthritis (CAIA) in mice. Description of the disease and the influence of age, sex, and genes Am J Pathol 2003 163 1827 1837 14578183
Nandakumar KS Bäcklund J Vestberg M Holmdahl R Collagen type II (CII)-specific antibodies induce arthritis in the absence of T or B cells but the arthritis progression is enhanced by CII-reactive T cells Arthritis Res Ther 2004 6 R544 R550 15535832 10.1186/ar1217
Holmdahl R Malmström V Vuorio E Autoimmune recognition of cartilage collagens Ann Med 1993 25 251 264 8333924
Vingsbo C Sahlstrand P Brun JG Jonsson R Saxne T Holmdahl R Pristane-induced arthritis in rats: a new model for rheumatoid arthritis with a chronic disease course influenced by both major histocompatibility complex and non-major histocompatibility complex genes Am J Pathol 1996 149 1675 1683 8909256
Hansson AS Lu S Holmdahl R Extra-articular cartilage affected in collagen-induced, but not pristane-induced, arthritis models Clin Exp Immunol 2002 127 37 42 11882030 10.1046/j.1365-2249.2002.01712.x
|
16277685
|
PMC1297579
|
CC BY
|
2021-01-04 16:02:40
|
no
|
Arthritis Res Ther. 2005 Sep 20; 7(6):R1316-R1324
|
utf-8
|
Arthritis Res Ther
| 2,005 |
10.1186/ar1829
|
oa_comm
|
==== Front
Arthritis Res TherArthritis Research & Therapy1478-63541478-6362BioMed Central London ar18301627768610.1186/ar1830Research ArticleContrasting effects of peroxisome-proliferator-activated receptor (PPAR)γ agonists on membrane-associated prostaglandin E2 synthase-1 in IL-1β-stimulated rat chondrocytes: evidence for PPARγ-independent inhibition by 15-deoxy-Δ12,14prostaglandin J2 Bianchi Arnaud [email protected] David 1Sebillaud Sylvie 1Koufany Meriem 1Galteau Marie-Madeleine 1Netter Patrick 1Terlain Bernard 1Jouzeau Jean-Yves [email protected] Laboratoire de Physiopathologie et Pharmacologie Articulaires, UMR 7561 CNRS-UHP, 54505 Vandœuvre-lès-Nancy, France2005 22 9 2005 7 6 R1325 R1337 16 3 2005 22 4 2005 4 8 2005 29 8 2005 Copyright © 2005 Bianchi 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.
Microsomal prostaglandin E synthase (mPGES)-1 is a newly identified inducible enzyme of the arachidonic acid cascade with a key function in prostaglandin (PG)E2 synthesis. We investigated the kinetics of inducible cyclo-oxygenase (COX)-2 and mPGES-1 expression with respect to the production of 6-keto-PGF1α and PGE2 in rat chondrocytes stimulated with 10 ng/ml IL-1β, and compared their modulation by peroxisome-proliferator-activated receptor (PPAR)γ agonists. Real-time PCR analysis showed that IL-1β induced COX-2 expression maximally (37-fold) at 12 hours and mPGES-1 expression maximally (68-fold) at 24 hours. Levels of 6-keto-PGF1α and PGE2 peaked 24 hours after stimulation with IL-1β; the induction of PGE2 was greater (11-fold versus 70-fold, respectively). The cyclopentenone 15-deoxy-Δ12,14prostaglandin J2 (15d-PGJ2) decreased prostaglandin synthesis in a dose-dependent manner (0.1 to 10 μM), with more potency on PGE2 level than on 6-keto-PGF1α level (-90% versus -66% at 10 μM). A high dose of 15d-PGJ2 partly decreased COX-2 expression but decreased mPGES-1 expression almost completely at both the mRNA and protein levels. Rosiglitazone was poorly effective on these parameters even at 10 μM. Inhibitory effects of 10 μM 15d-PGJ2 were neither reduced by PPARγ blockade with GW-9662 nor enhanced by PPARγ overexpression, supporting a PPARγ-independent mechanism. EMSA and TransAM® analyses demonstrated that mutated IκBα almost completely suppressed the stimulating effect of IL-1β on mPGES-1 expression and PGE2 production, whereas 15d-PGJ2 inhibited NF-κB transactivation. These data demonstrate the following in IL-1-stimulated rat chondrocytes: first, mPGES-1 is rate limiting for PGE2 synthesis; second, activation of the prostaglandin cascade requires NF-κB activation; third, 15d-PGJ2 strongly inhibits the synthesis of prostaglandins, in contrast with rosiglitazone; fourth, inhibition by 15d-PGJ2 occurs independently of PPARγ through inhibition of the NF-κB pathway; fifth, mPGES-1 is the main target of 15d-PGJ2.
==== Body
Introduction
Prostaglandins (PGs) are well-known lipid mediators that reproduce the cardinal signs of inflammation [1] but also contribute to tumorigenesis, gastrointestinal protection or osteogenesis [2-5]. Decreasing their biosynthesis by the inhibition of cyclo-oxygenases (COXs) is thought to account for most of the therapeutical properties of non-steroidal anti-inflammatory drugs. During inflammation, the pathophysiological contribution of prostaglandins is supported by PGE2, the major mediator produced by monocytes in response to inflammatory stimulus, and prostacyclin (PGI2). However, since the discovery of at least two COX isoenzymes, the pathophysiological relevance of PG must be considered from a different point of view. First, inflammation can be ascribed to inducible COX-2-derived PG rather than to basal COX-1-derived PG [6]. Second, PGE2 and PGI2 are now recognized as end-point products of a coordinate enzymatic cascade comprising phospholipases A2, cyclooxygenases and terminal PG synthases whose activities are coupled preferentially between constitutive and inducible isoforms [7]. Third, PG produced by COX-2 switches from PGE2 to 15-deoxy-Δ12,14prostaglandin J2 (15d-PGJ2) in the course of acute inflammation [8]. Because 15d-PGJ2, a cyclopentenone by-product of PGD2, has shown anti-inflammatory properties in various experimental models [9,10], it has been proposed as an endogenous regulator of inflammation favouring the resolution of acute flares [11].
PGE synthase-1 (PGES-1), the enzyme converting the COX-derived PGH2 into PGE2, exists in multiple forms with distinct enzymatic properties, modes of expression, subcellular localizations and intracellular functions [12]. One of its isoforms, cPGES-1, is a cytosolic protein found as a complex with heat shock protein 90 (Hsp90) that is constitutively expressed in a wide variety of cells and tissues. Another isoform, microsomal PGE synthase-1 (mPGES-1), is a perinuclear membrane-associated protein belonging to the microsomal glutathione S-transferase family. In contrast with cPGES-1, its expression is induced by pro-inflammatory cytokines, growth factors, bacterial endotoxins and phorbol esters and is downregulated by anti-inflammatory corticosteroids [12]. As mentioned above, PGES-1 isoforms display distinct functional coupling with upstream COX in cells; cPGES-1 is predominantly coupled with constitutive COX-1, thereby contributing to basal PG synthesis, whereas mPGES-1 is preferentially linked with inducible COX-2 and contributes to stimulated PG synthesis [7]. Recently a novel PGES, mPGES-2 [13], was cloned and was shown to be highly expressed in heart and brain. Its role remains largely unknown, especially in inflammatory conditions.
Peroxisome-proliferator-activated receptor γ (PPARγ) is a ligand-activated nuclear transcription factor belonging to the nuclear hormone receptor superfamily. PPARγ binds, as a heterodimer with retinoid X receptor, to peroxisome-proliferator-response element (PPRE) located in the promoter of numerous target genes whose expression is regulated by PPARγ agonists. Agonists of PPARγ include synthetic ligands, as antidiabetic thiazolidinediones, and natural compounds, as fatty acids and 15d-PGJ2, which were shown initially to have a major function in adipocyte differentiation and glucose homeostasis [14-16]. However, PPARγ agonists were recently thought to contribute to the control of inflammation by inhibiting the transcriptional induction of pro-inflammatory cytokines (tumour necrosis factor-α, IL-1 and IL-6) or genes encoding inflammatory enzymes (inducible nitric oxide synthase and COX-2) in activated monocytic cells [17,18]. Similar pharmacological potencies were reported in chondrocytes [19] and synoviocytes [20] exposed to an inflammatory stimulus, giving a rationale to the anti-inflammatory effect of PPARγ agonists in experimental arthritis [10,21]. Because 15d-PGJ2 was thought to be a negative regulator of experimental inflammation [11], it is tempting to speculate that part of this effect could be supported by the regulation of PPARγ target genes, possibly through the control of transcription factors such as NF-κB or activator protein-1 [22,23].
Chondrocytes express both COX isoenzymes [24] and produce large amounts of eicosanoids under inflammatory conditions [25]. However, COX-2 represents only the first inducible step in the stimulated synthesis of PG [12] and its inhibition by PPARγ ligands remains moderate in articular cells [19,20]. We therefore investigated whether PPARγ agonists could reduce PG synthesis by inhibiting mPGES-1 in rat chondrocytes stimulated with IL-1β. Such a mechanism would be consistent with the ability of 15d-PGJ2 to inhibit PGE2 production and to downregulate mPGES-1 in microsomal fractions from CHO cells overexpressing mPGES [26].
The present study demonstrates an early induction of COX-2 and a delayed induction of mPGES-1 by IL-1β in rat chondrocytes, with the stimulated synthesis of prostaglandins fitting well the expression profile of mPGES-1 for PGE2 while remaining lower than the extent of COX-2 induction for 6-keto-PGF1α (the stable metabolite of PGI2). In our experimental system, 15d-PGJ2 lowered the 6-keto-PGF1α level and the expression of COX-2 but was much more potent towards the PGE2 level and the expression of mPGES-1, supporting the view that mPGES-1 is the rate-limiting step in PGE2 synthesis. The dose-dependent inhibitory potency of 15d-PGJ2 was not reproduced by the high-affinity PPARγ agonist rosiglitazone and was affected neither by blockade of PPARγ with the antagonist GW-9662 nor by PPARγ overexpression. Consistent with a PPARγ-independent mechanism was our final observation that 15d-PGJ2 decreased NF-κB transactivation, which is crucial for the induction of mPGES-1 and the stimulation of PGE2 synthesis by IL-1β in rat chondrocytes.
Materials and methods
Isolation and culture of rat chondrocytes
Chondrocytes were isolated from femoral heads of healthy Wistar male rats (130 to 150 g) (Charles River, Saint-Aubin-les-Elbeuf, France), killed under general anaesthesia (AErrane™; Baxter SA, Maurepas, France) in accordance with national animal care guidelines, after approval by our internal ethics committee. Cells were obtained by sequential digestion with pronase and collagenase [27], then washed twice in PBS and cultured to confluence in 75 cm2 flasks at 37°C in a humidified atmosphere containing 5% CO2. The medium used was DMEM/Ham's F-12 supplemented with L-glutamine (2 mM), penicillin (100 U/ml), streptomycin (100 μg/ml) and either 10% heat-inactivated FCS (Life Technologies) during subculturing or 1% FCS during experiments. Chondrocytes were used between passages 1 and 3 to prevent dedifferentiation.
Study design
Chondrocytes maintained in low (1%) FCS medium were stimulated with 10 ng/ml IL-1β (Sigma, St-Quentin-Fallavier, France) in the presence or absence (vehicle alone, 0.1% of final concentration in dimethylsulphoxide) of PPAR agonists added 4 hours before IL-1β. In a preliminary kinetic study, mRNA levels of COX-2 and mPGES-1 in cell layers were determined from 6 to 48 hours after challenge with IL-1β, whereas 6-keto-PGF1α and PGE2 levels were assayed from 6 to 36 hours in culture supernatants. Thereafter, COX-2 mRNA level was checked 12 hours after exposure to IL-1β, whereas the mPGES-1 mRNA level, the COX-2 and mPGES-1 protein levels, and the secreted 6-keto-PGF1α and PGE2 levels were evaluated at 24 hours. The PPARγ agonists rosiglitazone (Cayman, Ann Arbor, MI, USA) or 15d-PGJ2 (Calbiochem, Meudon, France) were used in the range 0.1 to 10 μM, whereas additional PPARγ agonist troglitazone (Cayman) and PPARγ antagonist GW-9662 (Cayman) were used at 10 μM.
Assay for chondrocyte viability
Cell viability was assessed by the mitochondrial-dependent reduction of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide (MTT; Sigma) into formazan [28]. In brief, cells were incubated for 24 hours at 37°C in the presence or absence of IL-1β and/or PPARγ agonists (added 4 hours before IL-1β) in low-FCS (1%) DMEM/Ham's F-12 medium. Chondrocytes were incubated further with MTT (1 mg/ml final concentration) for 4 hours at 37°C before the addition of lysing buffer (20% w/v SDS in a 50% aqueous solution of dimethylformamide, pH 4.7). After 24 hours of incubation at 37°C, solubilization of formazan crystals was quantified by measuring A580 on a Multiskan® microplate reader (Labsystems, Montigny-le-Bretonneux, France).
RNA extraction and real-time PCR analysis
Total RNA was isolated from chondrocyte layers using Trizol® (Invitrogen, Cergy-Pontoise, France). Two micrograms of total RNA were reverse-transcribed for 90 minutes at 37°C with 200 U of Moloney Murine Leukaemia Virus reverse transcriptase (Invitrogen) and hexamer random primers. Expression of COX-2, mPGES-1 and adiponectin (chosen as a specific PPARγ target gene [29]) mRNAs were quantified by real-time PCR with the Lightcycler® (Roche) technology and the SYBRgreen master mix system® (Qiagen, Courtabœuf, France). After amplification, a melting curve was constructed to determine the melting temperature of each PCR product; their sizes were checked on a 2% agarose gel stained with ethidium bromide (0.5 μg/ml). Each run included standard dilutions and positive and negative reaction controls. The mRNA levels of each gene of interest and of the ribosomal protein S29, chosen as a housekeeping gene, were determined in parallel for each sample. Results are expressed as the normalized ratio of mRNA level of each gene of interest over the S29 gene.
The gene-specific primer pairs used were as follows: mPGES-1, sense 5'-TCGCCTGGATACATTTCCTC-3', antisense 5'-GTCCCCCATTGTGGTATCTG-3'; COX-2, sense 5'-TACAAGCAGTGGCAAAGGCC-3', antisense 5'-CAGTATTGAGGAGAACAGATGGG-3'; adiponectin, sense 5'-AATCCTGCCCAGTCATGAAG-3', antisense 5'-TCTCCAGGAGTGCCATCTCT-3'; S29, sense 5'-AAGATGGGTCACCAGCAGCTCTACG-3', antisense 5'-AGACGCGGCAAGAGCGAGAA-3'.
Transient transfection experiments
Chondrocytes were seeded in six-well plates at 5 × 105 cells per well and grown to 80% confluence. Cells were transfected with either 500 ng of a PPARγ expression vector (pcDNA3.1 PPARγ, a gift from Dr H. Fahmi, Centre Hospitalier de l'Université de Montréal, Montréal, Canada), or 500 ng of a dominant-negative vector of NF-κB (IκBαΔN (Ala32, Ala36) from Clontech). Transfections were performed for 2 hours with 10 μl of polyethyleneimine reagent (Euromedex, Souffelweyersheim, France) in 1 ml of culture medium. At 24 hours after transfection, cells were stimulated with IL-1β for 24 hours in the presence or absence of PPARγ agonists.
Preparation of nuclear extracts and electrophoretic mobility-shift assay (EMSA)
Nuclear proteins were isolated as described elsewhere [30] with minor modifications. In brief, cells were scraped in a lysis buffer (10 mM HEPES, pH 7.9, 10 mM KCl, 1 mM dithiothreitol (DTT)) containing a protease-inhibitor cocktail and 0.5% Igepal®, then incubated for 15 min on ice. Nuclei were collected by centrifugation at 2,000 g for 5 min at 4°C and resuspended in 50 μl of HEPES buffer without Igepal® and KCl, but containing 420 mM NaCl. After a 30 min incubation on ice, nuclear debris were removed by centrifugation at 13,000 g for 10 min at 4°C; supernatants were collected and then stored at -80°C before use. The DNA sequences of the double-stranded oligonucleotides specific for NF-κB were 5'-GATCCAGTTGAGGGGACTTTCCCAGGCG-3' and 5'-GATCCGCCTGGGAAAGTCCCCTCAACTG-3'. Complementary strands were annealed and double-stranded oligonucleotides were labelled with [32P]dCTP by using the Klenow fragment of DNA polymerase (Invitrogen). Nuclear proteins (5 μg) were incubated for 10 min at 4°C in a binding buffer (20 mM Tris/HCl, pH 7.9, 5 mM MgCl2, 0.5 mM DTT, 0.5 mM EDTA and 20% glycerol) in the presence of 2 μg of poly(dIdC). The extracts were then incubated for 30 min at 4°C with 10,000 c.p.m. of 32P-labelled NF-κB probe. The samples were loaded on a 5% native polyacrylamide gel and run in 0.5 × Tris/borate/EDTA buffer. NF-κB-specific bands were confirmed by competition with a 100-fold excess of unlabelled probe, which resulted in no shifted band.
NF-κB transactivation analysis
Nuclear proteins were prepared with the TransAM® nuclear extract kit in accordance with the manufacturer's protocol (Active Motif Europe, Rixensart, Belgium). In brief, cells were scraped into PBS containing phosphatase and protease inhibitors, centrifuged, resuspended in 1 × hypotonic buffer and then kept on ice for 15 min. After the addition of detergent, lysates were centrifuged at 14,000 g for 30 s at 4°C. The pellets were resuspended in complete lysis buffer (20 mM HEPES, pH 7.5, 350 mM NaCl, 20% glycerol, 1% Igepal®, 1 mM MgCl2, 0.5 mM EDTA, 0.1 mM EGTA, 1 mM DTT, phosphatase and protease inhibitors) and shaken vigorously. After incubation on ice and centrifugation at 14,000 g for 10 min at 4°C, supernatants were collected and protein concentration was determined with a Bradford-based assay (Bio-Rad Laboratories, Marnes-la-Coquette, France).
NF-κB activation was determined with the TransAM® ELISA kit (Active Motif Europe). In brief, 5 μg of nuclear extract was added to each well of a 96-well plate into which an oligonucleotide with a NF-κB consensus binding site had been immobilized. After 1 hour of incubation with smooth agitation, wells were washed three times with washing buffer (100 mM PBS, pH 7.5, 500 mM NaCl and 1% Tween 20) and then incubated with p65 antibody (dilution 1:1,000 in washing buffer) for 1 hour at 20°C. After three successive washings with buffer, the wells were finally incubated for 1 hour with diluted horseradish peroxidase-conjugated antibody (dilution 1:1,000 in washing buffer) before the addition of 100 μl of developing solution (3,3',5,5'-tetramethylbenzidine substrate solution diluted in 1% dimethylsulphoxide). After 5 min of incubation, the reaction was stopped by the addition of 100 μl of 0.5 M H2SO4 and the final A450 was read on a Multiskan® microplate reader.
Assays for PGE2 and 6-keto-PGF1α
Levels of PGE2 and 6-keto-PGF1α were determined in culture supernatants with Assay Design® ELISA kits (Oxford Biomedical Research, Ann Arbor, MI, USA) in accordance with manufacturer's instructions. Assays are based on the combined use of a monoclonal antibody against PGE2 or PGF1α and an alkaline phosphatase-conjugated polyclonal antibody. After the addition of p-nitrophenyl phosphate substrate, A405 was read at on a micro Multiskan® plate reader. The limits of detection were 10 pg/ml and 1.4 pg/ml for PGE2 and 6-keto-PGF1α, respectively, with negligible cross-reactivity with PGE1 and PGF2α, respectively (manufacturer's data). Positive controls were used in each experiment.
Western blot analysis
Cells, seeded in six-well plates and grown to 90% confluence, were washed twice with ice-cold PBS and scraped off the wells in 1 × Laemmli blue for PPARγ or in TBS containing 0.1% SDS for other proteins. Cells were disrupted by sonication (five pulses) and centrifuged at 800 g for 10 min, before determination of protein concentration with a Bradford-based assay. Protein samples (5 μg) were analysed by SDS-PAGE (10% acrylamide for COX-2 and PPARγ, 12% for β-actin, and 15% for mPGES-1), and electroblotted on a poly(vinylidene difluoride) membrane. After 1 hour in blocking buffer (TBS-Tween with 5% nonfat dried milk), membranes (Immobilon; Waters, Saint-Quentin en Yvelines, France) were blotted overnight at 4°C with antibodies against β-actin (dilution 1:500; Sigma), mPGES-1 (dilution 1:200; Cayman), COX-2 (dilution 1:1,000; Cayman) or PPARγ (a gift from Professor Michel Dauça, Université Henri Poincaré, Vandœuvre-lès-Nancy, France; dilution 1:1,000), diluted in TBS-Tween with 5% bovine serum albumin. After three washings with TBS-Tween, the blot was incubated for 1 hour at room temperature with anti-rabbit IgG conjugated with horseradish peroxidase (Cell Signaling, Beverly, MA, USA) at 1:2,000 dilution in TBS-Tween containing 5% nonfat dried milk. After four washings with TBS-Tween, protein bands were detected by chemiluminescence with the Phototope Detection system in accordance with the manufacturer's instructions (Cell Signaling).
Statistical analysis
Results are expressed as means ± SD for at least three assays. Comparisons were made by ANOVA, followed by the Fisher protected least-squares difference post-hoc test with Statview™ 5.0 software (SAS Institute Inc). A P value of less than 0.05 was considered significant.
Results
Kinetics of COX-2/mPGES-1 expression and prostaglandin production in IL-1β-stimulated rat chondrocytes
Under basal conditions, PGE2 and 6-keto-PGF1α production was almost undetectable (Fig. 1a), whereas COX-2 and mPGES-1 mRNAs were expressed at a very low level (Fig. 1b). In response to IL-1β, PGE2 levels increased earlier (6 hours) than 6-keto-PGF1α levels (12 hours), although both peaked at 24 hours (Fig. 1a). At the time of maximal production, PGE2 levels were increased 70-fold and 6-keto-PGF1α levels 11-fold. Under these experimental conditions, COX-2 and mPGES-1 expression was induced from 6 hours, with maximal induction at 12 hours and 24 hours, respectively, after challenge with IL-1β (Fig. 1b). At these times, the extent of gene variation was higher for mPGES-1 (68-fold) than for COX-2 (37-fold).
Effect of PPARγ agonists on prostaglandin cascade in IL-1β-stimulated rat chondrocytes
As shown in Fig. 2a, IL-1β-induced PGE2 production was decreased by 92%, and 6-keto-PGF1α levels by 66%, by 10 μM 15d-PGJ2. The effect of 10 μM rosiglitazone on the stimulated levels of prostaglandins was less than the variation range of our biological system (-12% for PGE2 and +10% for 6-keto-PGF1α; Fig. 2a). Under IL-1-stimulated conditions, 10 μM 15d-PGJ2 decreased the expression of COX-2 and mPGES-1 by 40% and 92%, respectively, at the mRNA level (Fig. 2b) and by 52% and 73%, respectively, at the protein level (Fig. 2c). In contrast, 10 μM rosiglitazone increased COX-2 mRNAs by 37% and decreased mPGES-1 mRNAs by 10% (Fig. 2b), while leaving COX-2 protein unaffected and decreasing mPGES-1 protein by 36% (Fig. 2c). The inhibitory potency of 15d-PGJ2 on PGE2 levels was dose-related (-8% at 0.1 μM and -42% at 10 μM), whereas rosiglitazone was still ineffective at lower concentrations (-2% at 0.1 μM and -6% at 10 μM). As shown in Table 1, the proliferation of chondrocytes was increased by challenge with IL-1β but this effect was reduced neither by 15d-PGJ2 nor by rosiglitazone. Under IL-1-stimulated conditions, the PPARγ agonist troglitazone (10 μM) had a potency similar to that of rosiglitazone on mPGES-1 mRNAs (-12%), although its induction of COX-2 mRNAs was less (+25% versus +37%) and it was more inhibitory towards PGE2 levels (-25% versus -12%; data not shown). The basal levels of prostaglandins were unaffected by PPARγ agonists (Fig. 2a) despite a moderate inducing effect of 15d-PGJ2 on COX-2 mRNAs (Fig. 2b) and protein (Fig. 2c).
Effect of PPARγ blockade on inhibitory potency of 15d-PGJ2 on stimulated prostaglandin cascade
When 10 μM 15d-PGJ2 was tested in combination with the PPARγ antagonist GW-9662 at 10 μM, its inhibitory effect on IL-1-induced PGE2 (-94% versus -95%) and 6-keto-PGF1α (-64% versus -58%) levels remained unchanged (Fig. 3a). Similarly, the strong decrease in mPGES-1 mRNA (-93% versus -87%; Fig. 3b) and protein (-70% versus -65%; Fig. 3c) levels was unaffected. In all experiments, the inducing effect of IL-1β on prostaglandin release and gene expression was not modified by GW-9662. Because of the low efficacy of chondrocyte transfection with a PPRE-luciferase construct as a gene reporter assay, the functionality of PPARγ ligands was controlled by measuring changes in adiponectin expression. As shown in Fig. 3d, the adiponectin mRNA level was increased by 10 μM 15d-PGJ2 or rosiglitazone and returned to the basal level in the presence of GW-9662.
Effect of PPARγ overexpression on inhibitory potency of 15d-PGJ2 on stimulated prostaglandin cascade
Transfection of chondrocytes with a PPARγ expression vector did not change their response to IL-1β and provoked a limited increase in PGE2 level and mPGES-1 expression in resting cells (Fig. 4a, b). The inhibition of IL-1β-induced PGE2 release and mPGES-1 mRNA level by 10 μM 15d-PGJ2 was not impaired in cells overexpressing PPARγ (-88% versus -94% and -79% versus -82%, respectively; Fig. 4a, b). Control experiments showed that PPARγ protein was efficiently overexpressed (Fig. 4c), and that the level of adiponectin mRNA was enhanced by 15d-PGJ2 or rosiglitazone (Fig. 4d), in cells transfected with the PPARγ expression vector.
Contribution of NF-κB pathway to regulation of stimulated prostaglandin cascade by IL-1β and 15d-PGJ2 in rat chondrocytes
As shown in Fig. 5, transfection with a dominant-negative vector of NF-κB (IκBαΔN) almost completely eliminated the synthesis of PGE2 (Fig. 5a) and the expression of mPGES-1 (Fig. 5b) in IL-1β-stimulated chondrocytes. As with PPARγ, transient overexpression was associated with a negligible induction of PGE2 and mPGES-1 in resting cells (Fig. 5a, b). Gel-shift analysis (Fig. 5c) and TransAM® assay (Fig. 5d) confirmed that IL-1β induced NF-κB transactivation in rat chondrocytes and demonstrated that this activity was markedly decreased by 15d-PGJ2.
Discussion
Since the discovery of a preferential coupling between several inducible enzymes of the prostaglandin cascade [31], it has become necessary to re-evaluate which step is critical for the synthesis of mediators. COX and phospholipases A2 have long been considered the rate-limiting enzymes; this was confirmed indirectly by the successful launching of non-steroidal anti-inflammatory drugs for the treatment of inflammation, pain and fever. However, the discovery of inducible mPGES-1 opened new insights because it was expressed at a high level in joint tissues during experimental polyarthritis [32] as well as in periarticular soft tissues and brain during acute inflammation [33]. Moreover, PGE2 was shown to contribute to inflammation and hyperalgesia [34], and the pivotal role of mPGES-1 in its production was confirmed by the decrease in pain nociception and inflammatory reactions in mPGES-1-deficient mice [35]. Finally, in contrast with COX inhibition, blockade of mPGES-1 could theoretically favour the biotransformation of cyclic endoperoxide H2 into anti-inflammatory 15d-PGJ2 depending on the tissue expression of PGD synthase [36]. The pathophysiological role of mPGES-1 in inflammatory diseases is therefore worthy of study, and inhibitors of this enzyme might have potent therapeutical relevance [37].
In the present study we investigated first the respective time courses of prostaglandin production and induction of genes of the arachidonic acid cascade in chondrocytes activated with IL-1β, a pro-inflammatory cytokine with a central function in joint diseases [38]. We confirmed that normal rat chondrocytes were very sensitive to stimulation by IL-1β and produced large amounts of prostaglandins [39], with kinetics comparable to that of human osteoarthritic chondrocytes [19,40] or the immortalized T/C-28a2 cell line [41]. As expected, resting and activated chondrocytes produced several types of prostaglandin, although the extent of variation was much higher for PGE2 than for 6-keto-PGF1α [39,42]. Although IL-1β-induced PGE2 synthesis was associated with the induction of COX-2 expression in articular cells [19,39], it has been shown that COX-2 and mPGES-1 are coordinately upregulated, but with different time courses [37,39,43], and that their subcellular localizations overlap in the perinuclear region [40,43]. Our kinetics study confirmed an early induction of COX-2 and a delayed induction of mPGES-1 in IL-1β-stimulated chondrocytes [40], thereby mimicking the time course reported for inflamed rat tissues [33].
The increase in PGE2 level fitted well with the extent of mPGES-1 gene induction but not with that of COX-2, whereas changes in the 6-keto-PGF1α level were much smaller than the extent of COX-2 induction. Of course, each inducible enzyme of the arachidonic acid cascade is rate limiting in that it controls the bioavailability of substrate to downstream effectors [6,7]. However, our results strongly support the contention that mPGES-1 expression is the most limiting step in PGE2 synthesis, consistent with previous experiments with MK-886 [40], a five-lipoxygenase activating protein (FLAP) inhibitor with in vitro inhibitory potency towards mPGES-1 [44].
Because the stimulated synthesis of 6-keto-PGF1α requires successive metabolization by COX-2 and prostacyclin synthase (PGIS), the lower than expected increase could reflect a limited induction of PGIS in rat chondrocytes. Thus, induction of PGIS by IL-1β was less than double that in rat non-articular cells [45] despite its selective upregulation by COX-2 induction in human endothelial cells [46]. A decrease in PGIS expression, contrasting with an increase in mPGES-1 expression, was also reported in inflamed tissues of rat with adjuvant polyarthritis [32]. Alternatively, other metabolic pathways might have been favoured such as the conversion of cyclic endoperoxides into other prostaglandins [42], depending on the substrate concentration dependences of the terminal synthases [6,46]. Arachidonic acid could also have been transformed into hydroxylated non-prostaglandin metabolites, which can be synthesized in IL-1β-stimulated chondrocytes [25], depending on the balance between the COX and lipoxygenase pathways [47]. In all instances, IL-1β stimulated all inducible steps of the arachidonic acid cascade to produce PGE2 maximally in rat chondrocytes.
The study of the expression of COX-2 or mPGES-1 and the release of prostaglandins in activated chondrocytes showed that 15d-PGJ2 was strongly inhibitory, whereas the high-affinity PPARγ agonist rosiglitazone was marginally potent in the same concentration range. Although 15d-PGJ2 and rosiglitazone were able to induce adiponectin expression, thereby demonstrating their potency to activate PPARγ, these results, irrespective of the binding affinity of agonists to PPARγ [48], supported the idea that this isotype was not primarily involved. It is interesting to note that the inducing effect of rosiglitazone on COX-2 mRNA was not confirmed at the protein level and that it was slightly inhibitory on mPGES-1, resulting in an unchanged PGE2 level. When we tried to decrease the inhibitory potency of 15d-PGJ2 by antagonizing its binding to PPARγ with GW-9662, we failed to observe any changes in gene mRNAs and PGE2 levels. As a corollary, the efficient overexpression of PPARγ did not enhance the potency of 15d-PGJ2 in our experimental system. Finally, despite the existence of a PPRE consensus site in the promoter of human COX-2 [49] and evidence that 15d-PGJ2 stimulates COX-2 gene expression in rat chondrocytes as in human synovial fibroblasts [50], we failed to observe any change in the basal production of PGE2, as reported previously in human osteoarthritic chondrocytes [51].
Taken together, our data strongly support the contention that 15d-PGJ2 was acting independently of PPARγ. Very few data are available in the rat species, but a PPARγ-dependent inhibition of inducible arachidonic acid cascade was reported in cardiac myocytes stimulated with IL-1β [52]. Because the inhibitory potency of 15d-PGJ2 on the COX-2, mPGES-1 and PGE2 levels was closely similar in both studies, we suggest that this discrepancy might be supported by cell type specificities. Indeed, the decrease in the levels of prostacyclin metabolites was different between cardiac myocytes and chondrocytes (no inhibition versus -66%) for a comparable extent of COX-2 inhibition (-40 to -50%), whereas the synthetic PPARγ agonist troglitazone was much more inhibitory towards PGE2 levels in the former cell type. In human chondrocytes, the inhibitory potency of 15d-PGJ2 was similar to our results on PGE2 levels [51], although supported by a stronger inhibition of COX-2 and a PPARγ-dependent inhibition of mPGES-1 [53]. In this cell type, the dose-dependent effect of 15d-PGJ2 was also thought to be mainly supported by the activation of PPARγ for the control of other inflammatory mediators [54] and apoptosis [55]. The biological responses to PPAR agonists are well known to differ between species [56], but our data support the notion that the potency of PPARγ agonists on joint cells might be influenced by differences in both cell type and species. Consistently, 15d-PGJ2 and troglitazone were shown to inhibit PGE2 production and mPGES-1 expression in IL-1β-stimulated human synovial fibroblasts [57], whereas troglitazone was totally ineffective on LPS-induced COX-2 expression in rat cells [20]. Finally, one could underline that the contribution of PPARγ might also depend on 15d-PGJ2 concentration, because the inhibition of PGE2 production was reported to be PPARγ-dependent in the nanomolar range while becoming PPARγ-independent in the micromolar range [58]. Despite a variable contribution of the PPARγ isotype depending on the biological system used, the present study confirms that 15d-PGJ2 downregulates inducible steps of the arachidonic acid cascade in joint cells, thereby probably contributing to its anti-arthritic properties [10].
The inhibitory potency of 15d-PGJ2 was PPARγ-independent but dose-related, which does not favour non-specific activity. This led us to investigate whether 15d-PGJ2 could interact with the NF-κB pathway, which is known to be one of its major targets in many cell types [59,60]. A previous study of the mouse mPGES-1 promoter indicated that it lacked binding sites for NF-κB, the cAMP-response element, and E-box, which have been implicated in COX-2 induction, implying that the mechanisms for inducible expression of COX-2 and mPGES-1 were distinct in this species [61]. In human synovial fibroblasts, transcriptional regulation of the mPGES-1 gene by IL-1β was shown to be closely dependent on the transcription factor early growth response factor-1 (Egr-1) [57], although activator protein-1 and specificity protein-1 binding sites were also found [62]. In human chondrocytes, IL-1β was demonstrated to use overlapping, but distinct, signalling pathways to induce COX-2 and mPGES-1, with a major role for ERK1/2 and p38β MAPK in controlling the latter [41]. However, in a non-articular human cell type, a substantial role for NF-κB was demonstrated recently in the coordinate induction of COX-2 and mPGES-1 by IL-1β [63]. As indicated previously, some of these signalling pathways can be inhibited in a PPARγ-dependent manner, possibly secondary to the squelching of transcription cofactors such as CBP/p300 by protein-protein interaction with PPARγ [64]. Consequently, such a mechanism is unlikely to explain the PPARγ-independent inhibitory potency of 15d-PGJ2 in our system.
Although the promoter of rat mPGES-1 has not so far been explored, our data with mutated IκBα are consistent with a major role of NF-κB in the control of its transcriptional activity. We showed further that 15d-PGJ2 inhibited IL-1β-induced NF-κB nuclear binding (with the use of EMSA) and transactivation (with a TransAM® assay). This inhibitory effect was consistent with the ability of 15d-PGJ2 to decrease IκB kinase (IKK) activity, by limiting the phosphorylation of its catalytic subunit IKKβ, and to prevent IκBα degradation by the proteasome [65]. Because of the high chemical reactivity of its cyclopentenone ring with substances containing nucleophilic groups, such as the cysteinyl thiol group of proteins [66], possible mechanisms may include covalent binding of 15d-PGJ2 to IKK [67] or alkylation of a conserved cysteine residue located in the p65 subunit DNA-binding domain of NF-κB [68]. A possible chemical interaction with NF-κB components is further sustained by the ability of 15d-PGJ2 to suppress the induction of COX-2 in PPARγ-deficient macrophages [14]. However, we did not investigate whether NF-κB binds directly to mPGES-1 rat promoter, and the delayed induction of mPGES-1 by IL-1β supports indirect regulation. NF-κB was consistently shown to regulate the early expression of Egr-1 [69], which has been implicated in the regulation of murine and human mPGES-1 [57,61]. Alternatively, we cannot exclude the possibility that inhibition of COX-2 by 15d-PGJ2 might participate partly in its inhibitory potency towards mPGES-1, because PGE2 production associated with COX-2 is involved in the induction of mPGES-1 by IL-1β in rheumatoid synovial fibroblasts [43].
Conclusion
The data reported here demonstrate that IL-1β activates COX-2 and mPGES-1 sequentially in rat chondrocytes and that the production of large amounts of PGE2 depends mainly on the expression of mPGES-1. In our cell type, 15d-PGJ2 displayed a strong inhibitory effect on prostaglandin levels and gene expression, whereas rosiglitazone was poorly active in the same concentration range. Despite its efficient activation of PPARγ, the effect of 15d-PGJ2 occurred through a PPARγ-independent mechanism. The activation of the NF-κB pathway was critical for mediating the inducing effect of IL-1β on PGE2 levels and mPGES-1 expression in rat chondrocytes, and was abolished by 15d-PGJ2. On the basis of the pathophysiological role of PGE2 in rheumatic diseases, our data support the general meaning that 15d-PGJ2 could behave as an endogenous regulator of inflammation if it was synthesized in sufficient amounts within joint tissues.
Abbreviations
15d-PGJ2 = 15-deoxy-Δ12,14prostaglandin J2; COX-2 = cyclo-oxygenase-2; DMEM = Dulbecco's modified Eagle's medium; DTT = dithiothreitol; EMSA = electrophoretic mobility-shift assay; FCS = fetal calf serum; IKK = IκB kinase; IL = interleukin; mPGES-1 = microsomal prostaglandin E synthase-1; NF-κB = nuclear factor-κB; PBS = phosphate-buffered saline; PCR = polymerase chain reaction; PG = prostaglandin; PGI2, prostacyclin; PGIS = prostacyclin synthase; PPAR = peroxisome-proliferator-activated receptor; PPRE = peroxisome-proliferator-response element.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
AB and DM performed the molecular studies and drafted the manuscript. SS and MK performed the immunoassays and the statistical analysis. MMG and PN supervised the study design and the manuscript. BT and JYJ conceived the study and participated in its design and final presentation. All authors read and approved the final manuscript.
Acknowledgements
This work was supported by grants from the Association de la Recherche contre la Polyarthite and the Communauté Urbaine du Grand Nancy.
Figures and Tables
Figure 1 Time course of prostaglandins production, COX-2 and mPGES-1 mRNA expression, in IL-1β-stimulated chondrocytes. Rat cells were exposed to 10 ng/ml IL-1β for 6, 12, 24, 36 or 48 hours before total RNA extraction and collection of culture supernatant. (a) Prostaglandin levels (PGE2, 6-keto-PGF1α) assayed by ELISA in culture supernatant; (b) relative abundances of cyclo-oxygenase-2 (COX-2) and microsomal prostaglandin E synthase-1 (mPGES-1) mRNAs, analysed by real-time PCR and normalized to S29 mRNA. Prostaglandin levels and PCR COX-2/S29 or mPGES-1/S29 mRNA ratios presented in histograms are expressed as means ± SD for at least three independent experiments. Statistically significant differences (P < 0.05) from controls: * for PGE2 or COX-2; † for 6-keto-PGF1α or mPGES-1
Figure 2 Effect of PPARγ agonists on IL-1β-induced prostaglandins levels, COX-2 and mPGES-1 mRNAs. After 4 hours of pretreatment with 10 μM 15-deoxy-Δ12,14prostaglandin J2 (15d-PGJ2) or rosiglitazone, chondrocytes were incubated with 10 ng/ml IL-1β for 12 or 24 hours. (a) PGE2 and 6-keto-PGF1α levels assayed by ELISA in culture supernatant; (b) relative abundances of cyclo-oxygenase-2 (COX-2) and microsomal prostaglandin E synthase-1 (mPGES-1) mRNAs, analysed by real-time PCR and normalized to S29 mRNA (c) COX-2 and mPGES-1 protein levels assessed by western blotting and normalized to β-actin level. Results are expressed as means ± SD for at least three independent experiments. Statistically significant differences (P < 0.05): *, comparison with non-stimulated controls; #, comparison with IL-1β-stimulated cells.
Figure 3 Effect of PPARγ blockade on the inhibition of IL-1β-induced responses by 15d-PGJ2. Chondrocytes were pretreated for 4 hours with 10 μM 15-deoxy-Δ12,14prostaglandin J2 (15d-PGJ2) in the presence or absence of 10 μM GW9662 (a specific antagonist of peroxisome-proliferator-activated receptor γ (PPARγ)), then stimulated with 10 ng/ml IL-1β for 24 hours before analysis of prostaglandin production and mPGES-1 expression. (a) PGE2 and 6-keto-PGF1α levels assayed by ELISA in culture supernatant; (b) relative abundance of microsomal prostaglandin E synthase-1 (mPGES-1) mRNA analysed by real-time PCR and normalized to S29 mRNA; (c) mPGES-1 protein level assessed by western blotting and normalized to β-actin level; (d) modulation of adiponectin (a PPARγ target gene) mRNAs by PPARγ ligands, analysed by real-time PCR and normalized to S29 mRNA. Results are expressed as means ± SD for at least three independent experiments. Statistically significant differences (P < 0.05): *, comparison with non stimulated controls; #comparison with IL-1β-stimulated cells; †, comparison with PPARγ agonists alone or in combination with PPARγ antagonist.
Figure 4 Effect of PPARγ overexpression on the inhibition of IL-1β-induced responses by 15d-PGJ2. Chondrocytes in six-well plates were transfected with pcDNA3.1 peroxisome-proliferator-activated receptor γ (PPARγ) construct (500 ng) for 36 hours. Thereafter, cells were pretreated for 4 hours with 10 μM 15-deoxy-Δ12,14prostaglandin J2 (15d-PGJ2), then stimulated with 10 ng/ml IL-1β for 24 hours before extraction of total RNA and collection of culture supernatant. (a) PGE2 levels assayed by ELISA in culture supernatant; (b) relative abundance of microsomal prostaglandin E synthase-1 (mPGES-1) mRNAs analysed by real-time PCR and normalized to S29 mRNA; (c) western blot control experiment of PPARγ and β-actin expression; (d) modulation of adiponectin (a PPARγ target gene) mRNAs by PPARγ agonists and pcDNA3.1 PPARγ construct, analysed by real-time PCR and normalized to S29 mRNA. Results are expressed as means ± SD for at least three independent experiments. Statistically significant differences (P < 0.05): *, comparison with non-stimulated controls; #, comparison with IL-1β-stimulated cells; †, comparison with PPARγ agonists alone or in combination with PPARγ plasmid.
Figure 5 Contribution of NF-κB pathway to IL-1β-induced responses and 15d-PGJ2 inhibitory effects. In one set of experiments (a, b), chondrocytes cultured in six-well plates were transfected with 500 ng of IKBα dominant-negative (IκBαΔN) vector for 24 hours, then stimulated for 24 hours with 10 ng/ml IL-1β. (a) PGE2 levels in culture supernatant assayed by ELISA; (b) Relative abundance of microsomal prostaglandin E synthase-1 (mPGES-1) mRNAs analysed by real-time PCR and normalized to S29 mRNA. Results are expressed as means ± SD for at least three independent experiments. In another set of experiments (c, d), chondrocytes cultured in six-well plates were exposed to 10 ng/ml IL-1β for 15 min in the presence or absence of 10 μM 15-deoxy-Δ12,14prostaglandin J2 (15d-PGJ2) before extraction of nuclear proteins. Activation of NF-κB was determined by EMSA (c) and by ELISA with the TransAm® technology (d). Results in (d) are expressed as relative arbitrary units with IL-1β treatment set at 100, and are representative of three different experiments. Statistically significant differences (P < 0.05): *, comparison with non-stimulated controls; #, comparison with IL-1β-stimulated cells.
Table 1 Effects of peroxisome-proliferator-activated receptor γ agonists on viability of IL-1β-stimulated chondrocytes
Agonist added A580 Percentage of control
None (control) 0.81 ± 0.05 100
IL-1β (10 ng/ml) 1.12 ± 0.04* 138
15d-PGJ2 (10 μM) 0.91 ± 0.07 112
IL-1β + 15d-PGJ2 1.10 ± 0.05 135
Rosiglitazone (10 μM) 0.95 ± 0.09 117
IL-1β + rosiglitazone 1.12 ± 0.07 138
15d-PGJ2, 15-deoxy-Δ12,14prostaglandin J2.
*, P < 0.05, comparison with non-stimulated controls
==== Refs
Willoughby DA Colville-Nash PR Seed MP Inflammation, prostaglandins, and loss of function J Lipid Mediat 1993 6 287 293 8357986
Norrdin RW Jee WS High WB The role of prostaglandins in bone in vivo Prostaglandins Leukot Essent Fatty Acids 1990 41 139 149 2281118 10.1016/0952-3278(90)90081-U
Matsumoto H Naraba H Murakami M Kudo I Yamaki K Ueno A Oh-ishi S Concordant induction of prostaglandin E2 synthase with cyclooxygenase-2 leads to preferred production of prostaglandin E2 over thromboxane and prostaglandin D2 in lipopolysaccharide-stimulated rat peritoneal macrophages Biochem Biophys Res Commun 1997 230 110 114 9020023 10.1006/bbrc.1996.5894
Shinomiya S Naraba H Ueno A Utsunomiya I Maruyama T Ohuchida S Ushikubi F Yuki K Narumiya S Sugimoto Y Regulation of TNFalpha and interleukin-10 production by prostaglandins I2 and E2: studies with prostaglandin receptor-deficient mice and prostaglandin E-receptor subtype-selective synthetic agonists Biochem Pharmacol 2001 61 1153 1160 11301049 10.1016/S0006-2952(01)00586-X
Harris SG Padilla J Koumas L Ray D Phipps RP Prostaglandins as modulators of immunity Trends Immunol 2002 23 144 150 11864843 10.1016/S1471-4906(01)02154-8
Dubois RN Abramson SB Crofford L Gupta RA Simon LS Van De Putte LB Lipsky PE Cyclooxygenase in biology and disease FASEB J 1998 12 1063 1073 9737710
Tanioka T Nakatani Y Semmyo N Murakami M Kudo I Molecular identification of cytosolic prostaglandin E2 synthase that is functionally coupled with cyclooxygenase-1 in immediate prostaglandin E2 biosynthesis J Biol Chem 2000 275 32775 32782 10922363 10.1074/jbc.M003504200
Gilroy DW Colville-Nash PR McMaster S Sawatzky DA Willoughby DA Lawrence T Inducible cyclooxygenase-derived 15-deoxyΔ12–14PGJ2 brings about acute inflammatory resolution in rat pleurisy by inducing neutrophil and macrophage apoptosis FASEB J 2003 17 2269 2271 14563690
Cuzzocrea S Wayman NS Mazzon E Dugo L Di Paola R Serraino I Britti D Chatterjee PK Caputi AP Thiemermann C The cyclopentenone prostaglandin 15-deoxy-Δ12,14-prostaglandin J2 attenuates the development of acute and chronic inflammation Mol Pharmacol 2002 61 997 1007 11961117 10.1124/mol.61.5.997
Kawahito Y Kondo M Tsubouchi Y Hashiramoto A Bishop-Bailey D Inoue K Kohno M Yamada R Hla T Sano H 15-deoxy-Δ12,14-PGJ2 induces synoviocyte apoptosis and suppresses adjuvant-induced arthritis in rats J Clin Invest 2000 106 189 197 10903334
Willoughby DA Moore AR Colville-Nash PR Gilroy D Resolution of inflammation Int J Immunopharmacol 2000 22 1131 1135 11137620 10.1016/S0192-0561(00)00064-3
Murakami M Nakatani Y Tanioka T Kudo I Prostaglandin E synthase Prostaglandins Other Lipid Mediat 2002 68-69 383 399 12432931 10.1016/S0090-6980(02)00043-6
Tanikawa N Ohmiya Y Ohkubo H Hashimoto K Kangawa K Kojima M Ito S Watanabe K Identification and characterization of a novel type of membrane-associated prostaglandin E synthase Biochem Biophys Res Commun 2002 291 884 889 11866447 10.1006/bbrc.2002.6531
Chawla A Barak Y Nagy L Liao D Tontonoz P Evans RM PPAR-gamma dependent and independent effects on macrophage-gene expression in lipid metabolism and inflammation Nat Med 2001 7 48 52 11135615 10.1038/83336
Chawla A Schwarz EJ Dimaculangan DD Lazar MA Peroxisome proliferator-activated receptor (PPAR) gamma: adipose-predominant expression and induction early in adipocyte differentiation Endocrinology 1994 135 798 800 8033830 10.1210/en.135.2.798
Tontonoz P Hu E Spiegelman BM Stimulation of adipogenesis in fibroblasts by PPAR gamma 2, a lipid-activated transcription factor Cell 1994 79 1147 1156 8001151 10.1016/0092-8674(94)90006-X
Jiang C Ting AT Seed B PPAR-gamma agonists inhibit production of monocyte inflammatory cytokines Nature 1998 391 82 86 9422509 10.1038/35154
Colville-Nash PR Qureshi SS Willis D Willoughby DA Inhibition of inducible nitric oxide synthase by peroxisome proliferator-activated receptor agonists: correlation with induction of heme oxygenase 1 J Immunol 1998 161 978 984 9670978
Boyault S Simonin MA Bianchi A Compe E Liagre B Mainard D Becuwe P Dauca M Netter P Terlain B Bordji K 15-Deoxy-Δ12,14-PGJ2, but not troglitazone, modulates IL-1β effects in human chondrocytes by inhibiting NF-κB and AP-1 activation pathways FEBS Lett 2001 501 24 30 11457450 10.1016/S0014-5793(01)02614-X
Simonin MA Bordji K Boyault S Bianchi A Gouze E Becuwe P Dauca M Netter P Terlain B PPAR-gamma ligands modulate effects of LPS in stimulated rat synovial fibroblasts Am J Physiol Cell Physiol 2002 282 C125 C133 11742805
Cuzzocrea S Mazzon E Dugo L Patel NS Serraino I Di Paola R Genovese T Britti D De Maio M Caputi AP Thiemermann C Reduction in the evolution of murine type II collagen-induced arthritis by treatment with rosiglitazone, a ligand of the peroxisome proliferator-activated receptor gamma Arthritis Rheum 2003 48 3544 3556 14674008 10.1002/art.11351
Ricote M Li AC Willson TM Kelly CJ Glass CK The peroxisome proliferator-activated receptor-γ is a negative regulator of macrophage activation Nature 1998 391 79 82 9422508 10.1038/34178
Spiegelman BM PPARgamma in monocytes: less pain, any gain? Cell 1998 93 153 155 9568708 10.1016/S0092-8674(00)81567-6
Blanco FJ Guitian R Moreno J de Toro FJ Galdo F Effect of antiinflammatory drugs on COX-1 and COX-2 activity in human articular chondrocytes J Rheumatol 1999 26 1366 1373 10381057
Kerr JS Stevens TM Davis GL McLaughlin JA Harris RR Effects of recombinant interleukin-1 beta on phospholipase A2 activity, phospholipase A2 mRNA levels, and eicosanoid formation in rabbit chondrocytes Biochem Biophys Res Commun 1989 165 1079 1084 2558648 10.1016/0006-291X(89)92712-5
Quraishi O Mancini JA Riendeau D Inhibition of inducible prostaglandin E(2) synthase by 15-deoxy-Δ12,14-prostaglandin J2 and polyunsaturated fatty acids Biochem Pharmacol 2002 63 1183 1189 11931852 10.1016/S0006-2952(02)00844-4
Kuettner KE Pauli BU Gall G Memoli VA Schenk RK Synthesis of cartilage matrix by mammalian chondrocytes in vitro. I. Isolation, culture characteristics, and morphology J Cell Biol 1982 93 743 750 6288734 10.1083/jcb.93.3.743
Hansen MB Nielsen SE Berg K Re-examination and further development of a precise and rapid dye method for measuring cell growth/cell kill J Immunol Methods 1989 119 203 210 2470825 10.1016/0022-1759(89)90397-9
Yang B Brown KK Chen L Carrick KM Clifton LG McNulty JA Winegar DA Strum JC Stimpson SA Pahel GL Serum adiponectin as a biomarker for in vivo PPARgamma activation and PPARgamma agonist-induced efficacy on insulin sensitization/lipid lowering in rats BMC Pharmacol 2004 4 23 15491498 10.1186/1471-2210-4-23
Dignam JD Lebovitz RM Roeder RG Accurate transcription initiation by RNA polymerase II in a soluble extract from isolated mammalian nuclei Nucleic Acids Res 1983 11 1475 1489 6828386
Kudo I Murakami M Diverse functional coupling of prostanoid biosynthetic enzymes in various cell types Adv Exp Med Biol 1999 469 29 35 10667306
Claveau D Sirinyan M Guay J Gordon R Chan CC Bureau Y Riendeau D Mancini JA Microsomal prostaglandin E synthase-1 is a major terminal synthase that is selectively up-regulated during cyclooxygenase-2-dependent prostaglandin E2 production in the rat adjuvant-induced arthritis model J Immunol 2003 170 4738 4744 12707354
Guay J Bateman K Gordon R Mancini J Riendeau D Carrageenan-induced paw edema in rat elicits a predominant prostaglandin E2 (PGE2) response in the central nervous system associated with the induction of microsomal PGE2 synthase-1 J Biol Chem 2004 279 24866 24872 15044444 10.1074/jbc.M403106200
Portanova JP Zhang Y Anderson GD Hauser SD Masferrer JL Seibert K Gregory SA Isakson PC Selective neutralization of prostaglandin E2 blocks inflammation, hyperalgesia, and interleukin 6 production in vivo J Exp Med 1996 184 883 891 9064348 10.1084/jem.184.3.883
Kamei D Yamakawa K Takegoshi Y Mikami-Nakanishi M Nakatani Y Oh-Ishi S Yasui H Azuma Y Hirasawa N Ohuchi K Reduced pain hypersensitivity and inflammation in mice lacking microsomal prostaglandin E synthase-1 J Biol Chem 2004 279 33684 33695 15140897 10.1074/jbc.M400199200
Murakami Y Akahoshi T Hayashi I Endo H Hashimoto A Kono S Kondo H Kawai S Inoue M Kitasato H Inhibition of monosodium urate monohydrate crystal-induced acute inflammation by retrovirally transfected prostaglandin D synthase Arthritis Rheum 2003 48 2931 2941 14558100 10.1002/art.11271
Stichtenoth DO Thoren S Bian H Peters-Golden M Jakobsson PJ Crofford LJ Microsomal prostaglandin E synthase is regulated by proinflammatory cytokines and glucocorticoids in primary rheumatoid synovial cells J Immunol 2001 167 469 474 11418684
Dayer JM The pivotal role of interleukin-1 in the clinical manifestations of rheumatoid arthritis Rheumatology (Oxford) 2003 42 Suppl 2 ii3 ii10 12817089
Nedelec E Abid A Cipolletta C Presle N Terlain B Netter P Jouzeau J Stimulation of cyclooxygenase-2-activity by nitric oxide-derived species in rat chondrocyte: lack of contribution to loss of cartilage anabolism Biochem Pharmacol 2001 61 965 978 11286988 10.1016/S0006-2952(01)00559-7
Kojima F Naraba H Miyamoto S Beppu M Aoki H Kawai S Membrane-associated prostaglandin E synthase-1 is upregulated by proinflammatory cytokines in chondrocytes from patients with osteoarthritis Arthritis Res Ther 2004 6 R355 R365 15225371 10.1186/ar1195
Masuko-Hongo K Berenbaum F Humbert L Salvat C Goldring MB Thirion S Up-regulation of microsomal prostaglandin E synthase 1 in osteoarthritic human cartilage: critical roles of the ERK-1/2 and p38 signaling pathways Arthritis Rheum 2004 50 2829 2838 15457451 10.1002/art.20437
Knott I Dieu M Burton M Houbion A Remacle J Raes M Induction of cyclooxygenase by interleukin 1: comparative study between human synovial cells and chondrocytes J Rheumatol 1994 21 462 466 8006889
Kojima F Naraba H Sasaki Y Beppu M Aoki H Kawai S Prostaglandin E2 is an enhancer of interleukin-1beta-induced expression of membrane-associated prostaglandin E synthase in rheumatoid synovial fibroblasts Arthritis Rheum 2003 48 2819 2828 14558087 10.1002/art.11261
Mancini JA Blood K Guay J Gordon R Claveau D Chan CC Riendeau D Cloning, expression, and up-regulation of inducible rat prostaglandin E synthase during lipopolysaccharide-induced pyresis and adjuvant-induced arthritis J Biol Chem 2001 276 4469 4475 11067848 10.1074/jbc.M006865200
Itoh A Nishihira J Makita H Miyamoto K Yamaguchi E Nishimura M Effects of IL-1beta, TNF-alpha, and macrophage migration inhibitory factor on prostacyclin synthesis in rat pulmonary artery smooth muscle cells Respirology 2003 8 467 472 14629650 10.1046/j.1440-1843.2003.00491.x
Caughey GE Cleland LG Penglis PS Gamble JR James MJ Roles of cyclooxygenase (COX)-1 and COX-2 in prostanoid production by human endothelial cells: selective up-regulation of prostacyclin synthesis by COX-2 J Immunol 2001 167 2831 2838 11509629
Martel-Pelletier J Mineau F Fahmi H Laufer S Reboul P Boileau C Lavigne M Pelletier JP Regulation of the expression of 5-lipoxygenase-activating protein/5-lipoxygenase and the synthesis of leukotriene B4 in osteoarthritic chondrocytes: role of transforming growth factor beta and eicosanoids Arthritis Rheum 2004 50 3925 3933 15593193 10.1002/art.20632
Willson TM Cobb JE Cowan DJ Wiethe RW Correa ID Prakash SR Beck KD Moore LB Kliewer SA Lehmann JM The structure-activity relationship between peroxisome proliferator-activated receptor gamma agonism and the antihyperglycemic activity of thiazolidinediones J Med Chem 1996 39 665 668 8576907 10.1021/jm950395a
Meade EA McIntyre TM Zimmerman GA Prescott SM Peroxisome proliferators enhance cyclooxygenase-2 expression in epithelial cells J Biol Chem 1999 274 8328 8334 10075740 10.1074/jbc.274.12.8328
Kalajdzic T Faour WH He QW Fahmi H Martel-Pelletier J Pelletier JP Di Battista JA Nimesulide, a preferential cyclooxygenase 2 inhibitor, suppresses peroxisome proliferator-activated receptor induction of cyclooxygenase 2 gene expression in human synovial fibroblasts: evidence for receptor antagonism Arthritis Rheum 2002 46 494 506 11840453 10.1002/art.10055
Fahmi H Pelletier JP Mineau F Martel-Pelletier J 15d-PGJ2 is acting as a 'dual agent' on the regulation of COX-2 expression in human osteoarthritic chondrocytes Osteoarthritis Cartilage 2002 10 845 848 12435328 10.1053/joca.2002.0835
Mendez M LaPointe MC PPARgamma inhibition of cyclooxygenase-2, PGE2 synthase, and inducible nitric oxide synthase in cardiac myocytes Hypertension 2003 42 844 850 12885795 10.1161/01.HYP.0000085332.69777.D1
Li X Afif H Cheng S Martel-Pelletier J Pelletier JP Ranger P Fahmi H Expression and regulation of microsomal prostaglandin E synthase-1 in human osteoarthritic cartilage and chondrocytes J Rheumatol 2005 32 887 895 15868626
Fahmi H Di Battista JA Pelletier JP Mineau F Ranger P Martel-Pelletier J Peroxisome proliferator-activated receptor gamma activators inhibit interleukin-1beta-induced nitric oxide and matrix metalloproteinase 13 production in human chondrocytes Arthritis Rheum 2001 44 595 607 11263774 10.1002/1529-0131(200103)44:3<595::AID-ANR108>3.0.CO;2-8
Shan ZZ Masuko-Hongo K Dai SM Nakamura H Kato T Nishioka K A potential role of 15-deoxy-Δ12,14-prostaglandin J2 for induction of human articular chondrocyte apoptosis in arthritis J Biol Chem 2004 279 37939 37950 15213234 10.1074/jbc.M402424200
Blanquart C Barbier O Fruchart JC Staels B Glineur C Peroxisome proliferator-activated receptors: regulation of transcriptional activities and roles in inflammation J Steroid Biochem Mol Biol 2003 85 267 273 12943712 10.1016/S0960-0760(03)00214-0
Cheng S Afif H Martel-Pelletier J Pelletier JP Li X Farrajota K Lavigne M Fahmi H Activation of peroxisome proliferator-activated receptor gamma inhibits interleukin-1β-induced membrane-associated prostaglandin E2 synthase-1 expression in human synovial fibroblasts by interfering with Egr-1 J Biol Chem 2004 279 22057 22065 15023995 10.1074/jbc.M402828200
Berry EB Keelan JA Helliwell RJ Gilmour RS Mitchell MD Nanomolar and micromolar effects of 15-deoxy-delta 12,14-prostaglandin J2 on amnion-derived WISH epithelial cells: differential roles of peroxisome proliferator-activated receptors gamma and delta and nuclear factor kappa B Mol Pharmacol 2005 68 169 178 15821150
Nosjean O Boutin JA Natural ligands of PPARγ: are prostaglandin J2 derivatives really playing the part? Cell Signal 2002 14 573 583 11955950 10.1016/S0898-6568(01)00281-9
Scher JU Pillinger MH 15d-PGJ2: the anti-inflammatory prostaglandin? Clin Immunol 2005 114 100 109 15639643 10.1016/j.clim.2004.09.008
Naraba H Yokoyama C Tago N Murakami M Kudo I Fueki M Oh-Ishi S Tanabe T Transcriptional regulation of the membrane-associated prostaglandin E2 synthase gene. Essential role of the transcription factor Egr-1 J Biol Chem 2002 277 28601 28608 12034740 10.1074/jbc.M203618200
Ekstrom L Lyrenas L Jakobsson PJ Morgenstern R Kelner MJ Basal expression of the human MAPEG members microsomal glutathione transferase 1 and prostaglandin E synthase genes is mediated by Sp1 and Sp3 Biochim Biophys Acta 2003 1627 79 84 12818425
Catley MC Chivers JE Cambridge LM Holden N Slater DM Staples KJ Bergmann MW Loser P Barnes PJ Newton R IL-1β-dependent activation of NF-κB mediates PGE2 release via the expression of cyclooxygenase-2 and microsomal prostaglandin E synthase FEBS Lett 2003 547 75 79 12860389 10.1016/S0014-5793(03)00672-0
Subbaramaiah K Lin DT Hart JC Dannenberg AJ Peroxisome proliferator-activated receptor gamma ligands suppress the transcriptional activation of cyclooxygenase-2. Evidence for involvement of activator protein-1 and CREB-binding protein/p300 J Biol Chem 2001 276 12440 12448 11278336 10.1074/jbc.M007237200
Boyault S Bianchi A Moulin D Morin S Francois M Netter P Terlain B Bordji K 15-Deoxy-Δ12,14-prostaglandin J2 inhibits IL-1β-induced IKK enzymatic activity and IκBα degradation in rat chondrocytes through a PPARγ-independent pathway FEBS Lett 2004 572 33 40 15304320 10.1016/j.febslet.2004.06.090
Fukushima M Biological activities and mechanisms of action of PGJ2 and related compounds: an update Prostaglandins Leukot Essent Fatty Acids 1992 47 1 12 1438462 10.1016/0952-3278(92)90178-L
Rossi A Kapahi P Natoli G Takahashi T Chen Y Karin M Santoro MG Anti-inflammatory cyclopentenone prostaglandins are direct inhibitors of IκB kinase Nature 2000 403 103 108 10638762 10.1038/47520
Straus DS Pascual G Li M Welch JS Ricote M Hsiang CH Sengchanthalangsy LL Ghosh G Glass CK 15-deoxy-Δ12,14-prostaglandin J2 inhibits multiple steps in the NF-κB signaling pathway Proc Natl Acad Sci USA 2000 97 4844 4849 10781090 10.1073/pnas.97.9.4844
Thyss R Virolle V Imbert V Peyron JF Aberdam D Virolle T NF-kappaB/Egr-1/Gadd45 are sequentially activated upon UVB irradiation to mediate epidermal cell death EMBO J 2005 24 128 137 15616591 10.1038/sj.emboj.7600501
|
16277686
|
PMC1297580
|
CC BY
|
2021-01-04 16:02:40
|
no
|
Arthritis Res Ther. 2005 Sep 22; 7(6):R1325-R1337
|
utf-8
|
Arthritis Res Ther
| 2,005 |
10.1186/ar1830
|
oa_comm
|
==== Front
Arthritis Res TherArthritis Research & Therapy1478-63541478-6362BioMed Central London ar18311627769010.1186/ar1831Research ArticleReproducibility and sensitivity to change of various methods to measure joint space width in osteoarthritis of the hip: a double reading of three different radiographic views taken with a three-year interval Maheu Emmanuel [email protected] Christian [email protected] Marc [email protected] Maxime [email protected] Salah [email protected] Isabelle 6Isabelle_KERLOCH/EXPANSCIENCE/[email protected]ères Bernard [email protected] Tim D [email protected] Eric [email protected] Michel G [email protected] Service de Rhumatologie, Hôpital Saint Antoine, Paris, France2 4 Place Martin Nadaud, 75020 Paris, France3 Medical Director, Clinica et Statistica, Issy les Moulineaux, France4 Université René Descartes, Paris V, Rheumatology Department, Hôpital Cochin, Paris, France5 Statistician, Clinica et Statistica, Issy les Moulineaux, France6 Project Manager, Expanscience Labs, Courbevoie, France7 Service de Rhumatologie, Hôpital Rangueil, Toulouse, France8 St Thomas's Hospital, London, UK9 Service de Rhumatologie, Hôpital, Lyon, France10 Service de Rhumatologie, Hôpital Léopold Bellan, Paris, France2005 5 10 2005 7 6 R1375 R1385 4 7 2005 18 8 2005 25 8 2005 30 8 2005 Copyright © 2005 Maheu 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.
Joint space width (JSW) and narrowing (JSN) measurements on radiographs are currently the best way to assess disease severity or progression in hip osteoarthritis, yet we lack data regarding the most accurate and sensitive measurement technique. This study was conducted to determine the optimal radiograph and number of readers for measuring JSW and JSN. Fifty pairs of radiographs taken three years apart were obtained from patients included in a structure modification trial in hip osteoarthritis. Three radiographs were taken with the patient standing: pelvis, target hip anteroposterior (AP) and oblique views. Two trained readers, blinded to each other's findings, time sequence and treatment, each read the six radiographs gathered for each patient twice (time interval ≥15 days), using a 0.1 mm graduated magnifying glass. Radiographs were randomly coded for each reading. The interobserver and intraobserver cross-sectional (M0 and M36) and longitudinal (M0–M36) reproducibilities were assessed using the intraclass coefficient (ICC) and Bland–Altman method for readers 1 and 2 and their mean. Sensitivity to change was estimated using the standardized response mean (SRM = change/standard deviation of change) for M0–M36 changes. For interobserver reliability on M0–M36 changes, the ICCs (95% confidence interval [CI]) were 0.79 (0.65–0.88) for pelvic view, 0.87 (0.78–0.93) for hip AP view and 0.86 (0.76–0.92) for oblique view. Intraobserver reliability ICCs were 0.81 (0.69–0.89) for observer 1 and 0.97 (0.95–0.98) for observer 2 for the pelvic view; 0.87 (0.78–0.92) and 0.97 (0.96–0.99) for the hip AP view; and 0.73 (0.57–0.84) and 0.93 (0.88–0.96) for the oblique view. SRMs were 0.61 (observer 1) and 0.82 (observer 2) for pelvic view; 0.64 and 0.75 for hip AP view; and 0.77 and 0.70 for oblique view. All three views yielded accurate JSW and JSN. According to the best reader, the pelvic view performed slightly better. Both readers exhibited high precision, with SRMs of 0.6 or greater for assessing JSN over three years. Selecting a single reader was the most accurate method, with 0.3 mm precision. Using this cutoff, 50% of patients were classified as 'progressors'.
==== Body
Introduction
Osteoarthritis (OA) is the most common rheumatic disease, and is becoming a major public health problem with the ageing of the population and the growing incidence of obesity in developed countries [1]. Treatment aims both to reduce symptom severity and to prevent or slow down disease progression and activity. Many symptom-modifying therapies have been proposed with various levels of evidence (for a recent review, see Zhang and coworkers [2]). However, we still lack a disease-modifying therapy because there is no treatment with proven efficacy in preventing, stopping, or retarding the disease process [2]. The structural process in OA affects cartilage, which is decreased in quality and thickness. Other structures may be involved in the damage observed in OA, including subchondral bone, articular capsule, synovium, meniscus and soft periarticular tissues. Hip OA is very common. It affects about 10% of the general population aged 65–74 years [3]. The prevalence of symptomatic hip OA increases dramatically with age.
Several trials have been conducted to identify structure-modifying drugs in hip OA, but as yet no such agent has exhibited convincing efficacy in this regard. The structural progression of OA is currently assessed on plain radiographic views by measuring the joint space width (JSW) and joint space narrowing (JSN) over a period of time [4]. This assessment is at present based on chondrometry, as described by Lequesne [5-7]. Other methods have been proposed, such as digitalized chondrometry (i.e. measurement of JSW or joint space surface with computer assistance [8]). Good reliability and sensitivity have been demonstrated for both methods [9,10]. At present, manual chondrometry – measurement of JSW at the narrowest point using a 1/10 mm graduated magnifying glass – performed by trained readers is the most commonly used technique. It has been shown to be sensitive to change and able to detect minor changes such as 0.5 mm over a one or two year period [11,12].
Recently published expert consensus recommendations [13,14] advocate the use of manual or digitalized measurement of joint space at the narrowest point on plain radiographic views of the pelvis in trials of structure-modifying treatment. However, there remains uncertainty concerning the optimal view for performing the measurement (anteroposterior [AP] pelvic view, feet in internal rotation of 15°, target hip AP view, or oblique view, which was proposed by Lequesne and Laredo [15] to be the 'false profile') and the number of readers that should perform the measurements in such trials. In 1987 Altman and coworkers [16] recommended three readers, but no evidence has yet been reported to support whether one, two, or even three readers should perform the measurements. It has been documented that radiography should be carried out in the standing rather than in the supine position [17,18]. The oblique view and plain pelvic view were compared in a pilot study conducted in 50 patients [19]. The combination of both views allowed identification of JSN in an additional one-third of patients, but the study did not attempt to identify the most sensitive view for performing chondrometry in a structure-modification trial.
The present study aimed to answer the following questions. Which radiographic view of the hip provides the most accurate measurement of JSW and JSN progression in hip OA? Should future trials of the structure-modifying effect of a treatment employ one or two trained readers for optimal assessment of disease progression and reliability of JSW measurement in hip OA?
Materials and methods
Patients
Hip radiographs were obtained from patients included in the ERADIAS study – an ongoing randomized, three-year, prospective, multicentre, double-blind, placebo-controlled trial of avocado/soybean unsaponifiables in hip OA. The study was approved by the ethics review board of the Pitié-Salpétrière Hospital. Included were outpatient with symptomatic hip OA (according to the American College of Rheumatology criteria [20]), who were 45–75 years old and who had a manually measured JSW on plain AP pelvic radiograph of 1–4 mm at baseline. All patients gave written informed consent to participate in the trial. Radiographs were verified by an independent assessor before study entry to ensure that patients were affected by OA; to ensure that the JSW was between 1 and 4 mm and assign the patient to one of the two strata (see below); and to exclude patients with isolated posteroinferior JSN, identified on the oblique view.
Selection of radiographs
Radiographs from 50 patients were selected at random from radiographs of patients who had completed the three-year duration of the trial on 13 July 2004. Patients in the trial were stratified at entry into two strata: those with baseline JSW below 2.5 mm and those with baseline JSW 2.5 mm or greater, in order to ensure that the whole spectrum of disease was represented. For each patient, the protocol was to obtain three different radiographic views each year: plain radiograph of the pelvis, and target hip AP view and oblique view (Lequesne's false profile). Radiographs performed at baseline and at month 36 (M36 ± 3 months) were selected. The number of sets of radiographs required in each stratum was 25.
Radiographic techniques
All radiographs were obtained at a standard size of 1/1 with the patient in a weight-bearing position. The X-ray beam was orientated AP, horizontal, and perpendicular to the table. The distance between X-ray source and film was 100 cm. Pelvis radiographs were performed with 15 ± 5° internal rotation of the feet and with the X-ray beam directed at the upper edge of the pubis symphysis. For hip AP views, 15 ± 5° internal rotation of the foot was also required but the the X-ray beam was directed at the joint space (with fluoroscopy). Oblique views were obtained using the technique described by Lequesne [13]. Patients were positioned with the foot axis (second metatarsus) parallel to the inferior edge of the radiography table and with the X-ray beam directed at the joint space (fluoroscopy). A sketch of the feet on the ground was drawn on heavy-weight paper during initial radiography and was used to position the patient at each subsequent examination.
Radiation exposure for each patient was 0.7 mSv (milliSieverts) for the pelvic view, 0.3 mSv for the hip AP view and 0.3 mSv for the oblique view. According to current private ambulatory practice in France, the cost of each of these views is 24.30 Euro (rated Z15 each, a Z costing 1.62 Euro).
Blinding process for radiographs
Two lists of randomization (one per stratum) were used to code radiographs (using an alphanumeric code). Different alphanumeric codes were assigned to radiographs for each reading in order to avoid any identification of a set of radiographs that had already been read (reading one: list numbers 1–50; reading two: list numbers 51–100). Readers were blinded to the time sequence; letters A or B were randomly assigned to code the time sequence (M0 [baseline] or M36) on radiographs. Therefore, each radiograph was identified both by a letter and a number. All coded films (three views at M0 and three at M36, yielding a total of six films) for a single patient were gathered in an envelope.
Reading procedures
Two trained readers (CC and EM) measured JSW using a 0.1 mm graduated magnifying glass. For each radiograph they were unaware of patient's identity, drug assignment, time sequence of the radiographs and each other's findings. Each set of six radiographs was read twice with a minimum time interval of 15 days between the two readings. Each radiograph was read on a horizontally positioned light box in order to identify the location and take an accurate measurement of the narrowest JSW area. All six views for each patient were read at the same time. About 10 sets of radiographs were read during each reading session (60 radiographs). A break was planned during each session so as not to exceed more than 2 consecutive hours of reading. Altogether, 300 radiographs were read twice, giving a total of 600 radiographs read. For the pelvic view, the target hip (i.e. the hip responsible for the patient's inclusion in the trial) to be read was indicated by a mark made by those in charge of randomization and labelling of radiographs. Readings were done between 24 August 2004 and 5 October 2004 by the two readers.
Measurement of joint space width
The JSW of the hip joint was measured at the narrowest point for each view, in accordance with a previously described method [6]. Briefly, the site of measurement was marked by the reader using a special pencil that produces removable marks. The interbone distance was measured at this site with the help of a 0.1 mm graduated magnifying glass directly applied to the radiographic film and reported on a specifically designed case report form. The mark was then removed by the reader. For the oblique view, measurement had to be performed in the anterior and upper part of the circumference between the femoral head and the acetabulum, because no significant articular cartilage thickness could be measured at the posteroinferior segment of the view, especially after patients with posteroinferior hip OA had been excluded from the trial.
Data management
Data were checked and queries sent to each observer when appropriate. For the Western and Ontario MacMaster University (WOMAC) score calculation, rules provided by the author were used [21]. Double key data entry was performed between 2 September and 5 October 2004.
Statistical analysis
Descriptive data were recorded at baseline for the 50 patients selected: age, gender, BMI, mean disease duration, WOMAC score and Lequesne's index [22]. The data from radiographic readings were presented for each view and for each observer (reader 1, reader 2 and mean of the two readers) for M0 and M36, and their difference (M36–M0) using descriptive statistics (number, mean, standard deviation [SD], minimum, and maximum). The number of hips exhibiting a joint space change of 0.5 mm or more and those with a change of 0.3 mm or more between M0 and M36 were calculated for each view. The metrologic measurements taken for each view and each reader are shown in Table 1.
Accuracy of JSW measurement evaluated by intraobserver and interobserver reproducibility was assessed using the intraclass coefficient of correlation (ICC) [23] and using the Bland–Altman plotting method [24], which indicates the smallest detectable difference (SDD; i.e. the amount of detectable change above the random measurement error). Estimates of ICC were derived in the framework of a two-way fixed effect model. The 95% confidence interval (CI) was estimated using the method described by Fleiss and Shrout [25]. Mean difference, SD of the difference, 95% CI approximation of bias, limits, and 95% CI of upper and lower limits of agreements between measures were calculated. Using the SDD, the proportions of patients who could be considered to be 'progressors' were calculated.
Sensitivity to change of radiographic measures was estimated based on differences in JSW between M36 and M0 (from reading 1) using the standardized response mean (SRM; mean change/SD of change). The 95% CI of SRM estimates were calculated using the Jackknife technique [26] using the software S-PLUS professional (S-PLUS 6 for Windows; Insightful Corp., Seattle, WA, USA).
Paired tests and limits of agreements were used for comparisons between views and observers. When the null hypothesis (i.e. normal distribution) was rejected, the paired Wilcoxon test was used.
Results
One hundred and forty-eight patients were included in the clinical trial between 7 February 2000 and 31 July 2001. The dropout rate in this sample was 45.9% (68/148), leaving 80 patients who completed the three years of follow-up. Radiographs of 29 patients were rejected for the following reasons: radiographs not received (five patients); one view missing or not available (11); radiographs sent for duplication and meanwhile not available (4); M0 or M36 radiograph not performed at the right time (i.e. more than 1 month delay; 2); M36 radiograph not obtained within the predefined time limit (i.e. 36 ± 3 months; 3); and poor radiograph quality (4). Among radiographs for the remaining 51 patients (26 in the low stratum and 25 in the high stratum), one patient was excluded by a random process to keep 25 radiographs in each stratum. Descriptive clinical data for the 50 patients whose radiographs were selected are shown in Table 2.
General results of radiographic measurements for each view and each observer (the mean of observers 1 and 2 is considered a third observer) are summarized in Table 3.
Interobserver reproducibility
Data (mean of differences at baseline [± SD] and M0–M36 changes, ICC values) are provided for each view in Table 4. ICC values were 0.80 for the pelvic view, 0.88 for the target hip AP view and 0.72 for the target hip oblique view, indicating a good interobserver reproducibility. There was a systematic bias between the two readers; specifically, JSW measurements for reader 2 were slightly but systematically higher than those of the reader 1.
Intraobserver reproducibility
Cross-sectional intraobserver reproducibility of radiographic measurements at baseline
The mean differences between repeated measurements of baseline radiographs are given in Table 5 for each view. ICC values were very high for both readers on all three views.
Longitudinal intraobserver reproducibility of measurements of joint space changes between baseline and M36
The mean differences in repeated measurements of changes in JSW between baseline (M0) and M36 are given in Table 6 for each reader and each view. The Bland–Altman plotting method results for intraobserver reproducibility of measurements of changes between baseline and M36 are summarized in Fig. 1 for both readers and the three different views. ICC values were also very high for each observer for all three views, as shown in Table 6.
Both readers exhibited very good precision, as assessed using the ICC. Reader 2 was more accurate for all measures, as assessed both by ICC and Bland–Altman graphics (Fig. 1b). Adding a second reader or calculating the mean of the two readers did not confer any additional precision.
Sensitivity to change over time
The SRM values were high, ranging from 0.61 (pelvic view, reader 1) to 0.82 (pelvic view, reader 2; Table 6). The estimate of the precision of the SRM calculated was performed using the Jackknife technique; 95% CI Jackknife SRMs are given in Table 6. According to values calculated in this study, radiographic measurement of JSW on the three views was sensitive. Reader 2 was more sensitive to change than was reader 1. All radiographic views appeared to provide similar levels of responsiveness. However, the pelvic view seemed to be the most sensitive in measuring changes in JSW – a basic property in trials of structure-modifying treatment.
JSW measurement is a continuous variable, and therefore it does not permit one to classify patients as disease progressors or nonprogressors. To translate this continuous variable into a dichotomous progression variable, we calculated the SDD, which can be derived using the Bland and Altman graphical approach. Its value is obtained by 2 SDs of the mean of differences between the two measurements. As may be calculated from data shown in Tables 5 and 6, the SDD for reader 2 was 0.32 mm for measurements of JSW and M0–M36 JSW changes on pelvic view and 0.30 mm and 0.28 mm, respectively, for measurements of JSW and M0–M36 JSW changes on the hip AP view.
The proportions of patients who could be classified as 'progressors' using the 0.3 mm cutoff or using the 0.5 mm cutoff previously described [11] are given in Table 7. Based on the reading precision offered by reader 2, the cutoff value of 0.3 mm was selected. Reader 1 identified 52%, 52% and 56% of progressors on pelvic, hip AP and hip oblique views, respectively. Reader 2 identified 48%, 54% and 52%, respectively. Using the 0.5 mm cutoff value, the respective proportions of progressors were 34%, 34% and 46% for reader 1, and 40%, 40% and 38% for reader 2.
Combining the results of measurements taken from the pelvic view and those taken from the oblique view led to a higher rate of identified progressors. Using the 0.3 mm cutoff, reader 1 identified 64% of progressors versus 52% on the pelvic view; using the 0.5 mm cutoff, 52% of progressors versus 34% were identified. The corresponding figures for reader 2 were 48% versus 48% and 46% versus 40%.
Comparisons between views
The mean difference between the JSW measurements on pelvic and hip AP views was 0.01 ± 0.18 mm for reader 2 (at first reading), which was not statistically significant (P = 0.91 by Wilcoxon test). The mean difference for the same reader between the JSW measurements on pelvic and oblique views was 0.01 ± 0.64 mm, which was also not statistically significant (P = 0.89 by Student's t test). The study of correlations between measurements of M0–M36 JSW changes by reader 2 on pelvic and hip AP views exhibited very high correlation (Pearson correlation coefficient = 0.94; P < 0.0001).
Discussion
Several radiographic views allow assessment of JSW and joint space changes in hip OA. To our knowledge, this is the first study to compare directly the metrologic measurement properties of JSW assessed using different radiographic views in hip OA obtained in the same sample of patients. Because the evaluation of a structure-modifying effect of a treatment is currently based on JSW measurement on radiographs, it is critical to optimize the technique used in order to maximize the precision of the measure. It must be noted that, in the present study, radiographs of poor quality or not performed within the predefined time limits from seven patients (9%) were excluded, which is not the procedure usually employed in clinical trial; instead, all radiographs are kept in such trial for use in an intent-to-treat analysis.
Our findings did not reveal significant differences between the ability of the different views to measure JSW reliably. With regard to intraobserver precision (either transversal at M0 or longitudinal between M0 and M36), and only considering the results for the better of the two readers, any of the three views could be used in a structural evaluation in hip OA because they yielded almost the same precision in assessment of JSW and joint space change. The limits of agreement at baseline ranged from -0.3 mm to +0.3 mm both for pelvic view and hip AP view, and for M0–M36 JSN measurement they ranged from -0.37 mm to +0.27 mm for the pelvic view and from -0.28 mm to +0.28 mm for the hip AP view. Cross-sectional and transversal intraobserver reproducibilities were consistent; the same values for dispersion (SD) were registered from the data for reader 2 from his readings of each of the three views at M0 and M0–M36 measurements.
In the present study interobserver reproducibility was less accurate than intraobserver (as shown in Tables 4, 5, 6). However, in the case of centralized reading performed by a single selected reader, it is clear that intraobserver precision is far more important than interobserver precision when examining the metrologic properties of an assessment tool aimed at measuring changes over time or with a given treatment.
The measurement of M0–M36 change in JSW provides an opportunity to assess the real measurement error. Indeed, it includes the error in measurement on a single radiograph (M0) along with the ability to detect change over time, and also includes the variability in measurement related to differences in patient repositioning at the second radiograph. When the aim is to select a tool to evaluate changes over time and/or to compare changes between groups, one must consider longitudinal intraobserver reliability and sensitivity to change, as given by the SRM. In the present study SRM values were good in all cases and for both readers. A SRM above 0.6 is considered good to excellent, whereas SRM values between 0.3 and 0.6 correspond to slight to moderate responsiveness. Unsurprisingly, measurements by the best reader provided the highest SRM values (ranging from 0.70 for the oblique view to 0.82 for the pelvic view). These values are consistent with SRMs calculated in previous studies comparing manual and digitalized assessment of joint space in hip OA [9,10].
Assessment of intraobserver precision provides an opportunity to calculate the SDD (i.e. the minimal amount of change that can be considered a change superior to the measurement error). The SDD allows determination of a cutoff value that segregates patients into those who had 'progressed' (i.e. lost cartilage thickness) and those who had not. This is of considerable importance in a trial in which the aim is to assess significant changes. The high precision in measurements by the second reader allowed us to select a 0.3 mm cutoff value, which is much lower than the 0.5 mm cutoff value usually recommended from previous studies [9,11]. Such a cutoff used in future clinical trials of structure-modifying treatment could result in increased statistical power and a reduction of the number of patients required. It would certainly permit a shorter duration of the trial (e.g. two years instead of three).
The present study shows that the precision of the measure is more dependent on the precision of the readers than on the radiographic view selected. Although the three views examined in the present study offered comparable precision in the assessment of JSW, either pelvic or hip AP view seems to be a good choice, offering a good reliability in measuring either JSW on a single view or joint space changes over time in pairs of radiographs taken 36 months apart. Although the best precision was obtained using the hip AP view, based on the Bland–Altman results and the SRM calculation, it may be more practical to choose the pelvic view (only slightly inferior to the hip AP view) because it also provides information on the contralateral hip.
The oblique view gives information that cannot be obtained when examining an AP view of the hip, even following exclusion of patients with isolated posteroinferior JSN, as was done in the present study. In a sample of hip OA patients with JSN in various locations, Conrozier and coworkers [19] showed that assessing the oblique in addition to the pelvic view resulted in identification of an additional 30% of patients with JSN. Our findings support the hypothesis that the combination of views could be superior to the use of a single view in identifying those patients whose joint space has changed. According to reader 2, 62% of patients could be classified as 'progressors' (i.e. patients exhibiting a decrease in JSW ≥ 0.3 mm) based on the combination of pelvic or oblique views, as compared with 48% of patients identified as progressors based on the pelvic view alone.
Using the pelvic or hip AP view, or combining one of them and the oblique view to assess structural modification in hip OA remains an option, depending on the trial aims and design. One could recommend that primary measurement of JS change be done using a single front view (either pelvic or hip AP) and that changes on both pelvic or hip face and oblique view be studied as secondary outcomes.
In France, in accordance with current ambulatory practice, the costs of each view were the same (costs are, of course, country dependent). The patients' radiation exposure was not very different between pelvic view, and hip AP or oblique views. Selection of the radiograph should not depend on such characteristics.
With regard to the number of readers that should be employed, our results conflict with previous recommendations that several readers be used [16]. A single reader was superior to the combination of two. Based on the results of this study, we recommend that the best reader be selected from among several trained readers before starting 'blinded' reading. This assumes that the reader has undergone preliminary training and that the reader will be selected to assess the primary outcome before the start of the trial. In our study we should like to identify two factors from among many possible explanations for the differences observed between the two readers, which could be taken into account in future trials: reader 2 was the most experienced of the two readers, having performed JSW assessment in several trials over the past 20 years; and furthermore, there were optical differences between the readers (reader 2 is a well corrected myopic and reader 1 a presbyopic). The latter factor leads reader 2 to remove his glasses when reading, with his myopia helping to magnify the image he reads. Optical impairments could be taken into account in the selection of readers; a myopic is preferable to a presbyopic.
Conclusion
Our results show that the three radiographs usually performed in the radiographic examination of the hip offer good precision for assessment of JSW. However, pelvic or hip AP view allow more accurate measurement. The selection of one trained reader is preferable to using several readers in a trial. Furthermore, the better the precision of the reader, the fewer the patients required for the trial. A precision of 0.3 mm joint space change over time is attainable, using such procedures. When choosing this cutoff, 50% of the patients could be identified as 'progressors' in the sample selected in the present study, which would enhance statistical power greatly. Further investigations are required to compare digitized with manual chondrometry on these three views and joint space measurement on a single AP view versus the combination of AP and oblique views.
Abbreviations
AP = anteroposterior; CI = confidence interval; ICC = intraclass coefficient of correlation; JSN = joint space narrowing; JSW = joint space width; OA = osteoarthritis; SD = standard deviation; SDD = smallest detectable difference; SRM = standardized response mean; WOMAC = Western and Ontario MacMaster University.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
EM devised the protocol of the trial and that of the present study with MM, performed the readings with CC, contributed to data analysis and wrote the manuscript. CC performed the readings and significantly contributed to the protocol development, data analysis and manuscript revision. MM devised the statistical section of the protocol, supervised the blinding process of radiographs, performed the randomization of radiographs, and verified the statistical analysis. MD significantly participated in devising the protocol of the study and in the analysis of data; he contributed to writing the manuscript and its revision. SG wrote with MM the statistical section of the protocol, checked the data and performed the statistical analysis. IK performed the follow-up of the trial and the radiographic study, and data management. BM, TS and EV significantly contributed to devising the protocol, analysis of data and revision of the manuscript. MGL was principal investigator of the trial; he significantly contributed to developing the design of the study, helped to define methods of measurement, and participated in manuscript development and revision.
Acknowledgements
We gratefully acknowledge Dr Philippe Coste and Expanscience Laboratories for providing data, material assistance and financial support to carry out this project.
Figures and Tables
Figure 1 Intraobserver precision. Shown is the intraobserver precision, summarized using the Bland and Altman plotting method, for the assessment of changes in joint space width (JSW) between baseline (M0) and 36 months (M36) for the two readers and the three different radiographic views. A total of 50 sets of three radiographs taken at M0 and M36 were read twice by 2 readers with a 15 day interval. (a) Intraobserver reproducibility for reader 1 (M0–M36): 1, pelvic view; 2, hip anteroposterior (AP) view; 3, oblique view. (b) Intraobserver reproducibility for reader 2 (M0–M36): 1, pelvic view; 2, hip AP view; 3, oblique view.
Table 1 Scheduled tested properties for each view and for each observer
Observer Intraobserver reproducibility (M0) Intraobserver reproducibility (M0–M36) Interobserver reproducibility (M0) Interobserver reproducibility (M0–M36) Sensitivity to change (M0–M36)
Value from observer 1 x x x x x
Value from observer 2 x x x
Mean of values from observers 1 and 2 x x x
Data Reading 1 and reading 2 of M0 for each observer Difference between M0 and M36 for readings 1 and 2 for each observer Reading 1 at M0 for two observers Reading 1 for the difference between M0 and M36 for two observers Difference between M0 and M36 in reading 1 for each observer
Table 2 Baseline characteristics of the 50 patients for whom radiographs were available
Baseline characteristic Low stratum: JSW <2.5 mm (n = 25) High stratum: JSW ≥ 2.5 mm (n = 25) Total (n = 50)
Age (years; mean ± SD) 60.6 ± 8.2 59.9 ± 10.9 60.2 ± 9.6
Sex (n [%] female) 15/25 (60%) 14/25 (56%) 29/50 (58%)
BMI (kg/m2; mean ± SD) 27.5 ± 5.5 25.8 ± 3.0 26.6 ± 4.5
Disease duration (years; mean ± SD) 5.2 ± 4.5 3.6 ± 3.2 4.4 ± 4.0
Lequesne index (0–24; mean ± SD) 7.2 ± 2.0 6.9 ± 1.9 7.1 ± 2
WOMAC (total score: 0–100; mean ± SD) 35 ± 23.3 27.10 ± 15.75 31.12 ± 20.15
Global pain on VAS (mm; mean ± SD) 39.4 ± 25.8 35. 7 ± 15.1 37.6 ± 21.0
BMI, body mass index; JSW, joint space width; SD, standard deviation; VAS, visual analogue scale; WOMAC, Western and Ontario MacMaster University.
Table 3 General results of radiographic measurements
View Observer Reading n M0 M36 ΔM36–M0
Mean (mm) SD Min Max Mean (mm) SD Min Max Mean (mm) SD Min Max
Pelvic 1 1 50 2.33 0.81 0.70 4.10 1.93 1.06 0.0 4.40 -0.40 0.65 -2.10 +0.80
2 50 2.34 0.81 0.70 4.20 1.85 0.97 0.0 4.10 -0.49 0.54 -1.90 +0.40
2 1 50 2.67 1.07 0.90 5.50 2.13 1.35 0.0 5.50 -0.54 0.66 -2.30 +0.40
2 50 2.66 1.02 0.70 5.10 2.16 1.31 0.0 5.20 -0.49 0.67 -2.20 +0.20
Mean (1 + 2) 1 50 2.50 0.91 0.70 4.65 2.03 1.17 0.0 4.95 -0.47 0.62 -2.05 +0.45
2 50 2.50 0.89 0.70 4.50 2.01 1.11 0.0 4.55 -0.49 0.57 -2.05 +0.30
Hip AP 1 1 50 2.41 0.86 0.70 4.30 2.02 1.13 0.0 4.40 -0.39 0.61 -2.00 +0.60
2 50 2.40 0.86 0.80 4.50 1.94 1.07 0.0 4.30 -0.46 0.59 -1.90 +0.70
2 1 50 2.69 1.05 0.60 5.50 2.19 1.34 0.0 5.50 -0.49 0.66 -2.10 +0.50
2 50 2.68 1.01 0.70 5.50 2.19 1.30 0.0 5.00 -0.49 0.66 -2.20 +0.60
Mean (1 + 2) 1 50 2.55 0.94 0.65 4.90 2.11 1.22 0.0 4.95 -0.44 0.62 -2.05 +0.55
2 50 2.54 0.92 0.75 5.00 2.06 1.17 0.0 4.55 -0.48 0.60 -1.95 +0.65
Oblique 1 1 50 2.37 0.79 0.70 3.70 1.92 1.05 0.0 3.80 -0.46 0.59 -2.20 +0.50
2 50 2.29 0.78 0.40 3.60 1.92 1.02 0.0 4.10 -0.37 0.50 -1.70 +0.70
2 1 50 2.69 1.00 0.50 5.00 2.21 1.28 0.0 5.00 -0.47 0.67 -2.40 +1.20
2 50 2.70 1.01 0.60 5.00 2.18 1.27 0.0 5.00 -0.52 0.64 -2.30 +0.20
Mean (1 + 2) 1 50 2.53 0.85 0.75 4.15 2.07 1.13 0.0 3.95 -0.46 0.61 -2.15 +0.65
2 50 2.50 0.87 0.50 4.20 2.05 1.13 0.0 4.95 -0.45 0.53 -1.95 +0.35
Shown are the general results of radiographic measurements of joint space width (JSW) at baseline (M0) and 36 months (M36), and the joint space change for the three views and two observers (and their mean and SD). AP, anteroposterior; Δ, difference; SD, standard deviation.
Table 4 Interobserver reproducibility of joint space width measurements
View Observer Baseline JSW reading 1 (mm; mean ± SD) Mean of difference (mm; mean ± SD) ICC (95% CI) Joint space change from M0 to M36 in reading 1 (mm) ICC (95% CI)
Pelvic 1 2.33 ± 0.81 -0.34 ± 0.52 0.80 (0.52–0.91) -0.40 ± 0.65 0.79 (0.65–0.88)
2 2.67 ± 1.07 -0.54 ± 0.66
Hip AP 1 2.41 ± 0.86 -0.27 ± 0.39 0.88 (0.66–0.95) -0.39 ± 0.61 0.87 (0.78–0.93)
2 2.69 ± 1.05 -0.49 ± 0.66
Oblique 1 2.37 ± 0.79 -0.31 ± 0.61 0.72 (0.50–0.85) -0.46 ± 0.59 0.86 (0.76–0.92)
2 2.69 ± 1.00 -0.47 ± 0.67
Shown is the interobserver reproducibility of joint space width (JSW) measurements between the two readers for the three radiographic views. AP, anteroposterior; CI, confidence interval; ICC, intraclass coefficient of correlation; SD, standard deviation.
Table 5 Cross-sectional intraobserver reproducibility of joint space width measurements at baseline
View Observer JSW (mm; mean ± SD) Mean difference (mm; mean ± SD) Limits of agreements ICC values (95% CI)
Reading 1 Reading 2
Pelvic view 1 2.33 ± 0.81 2.34 ± 1.07 -0.01 ± -0.24 -0.49 to +0.47 0.96 (0.93–0.98), 0.99
2 2.67 ± 0.81 2.66 ± 1.02 -0.02 ± 0.16 -0.30 to +0.34 (0.98–0.99) 0.98 (0.97–0.99)
Mean (1 + 2) 2.50 ± 0.91 2.50 ± 0.89 0.004 ± 0.14 -0.28 to +0.28
Hip AP 1 2.41 ± 0.86 2.40 ± 0.86 0.01 ± 0.25 -0.49 to +0.51 0.96 (0.93–0.98), 0.99
2 2.69 ± 1.05 2.68 ± 1.01 0.01 ± 0.15 -0.29 to +0.31 (0.98–0.99) 0.98 (0.97–0.99)
Mean (1 + 2) 2.55 ± 0.94 2.54 ± 0.92 0.01 ± 0.16 -0.30 to +0.32
Oblique 1 2.37 ± 0.79 2.29 ± 0.78 0.08 ± 0.37 -0.89 to +0.71 0.88 (0.80–0.93), 0.98
2 2.69 ± 0.78 2.70 ± 1.01 -0.02 ± 0.21 -0.41 to +0.51 (0.96–0.99) 0.96 (0.94–0.98)
Mean (1 + 2) 2.53 ± 0.85 2.50 ± 0.87 0.03 ± 0.21 -0.29 to +0.35
Shown is the cross-sectional intraobserver reproducibility of radiographic measurements of joint space width (JSW) performed on baseline radiographs. CI, confidence interval; ICC, intraclass coefficient of correlation; SD, standard deviation.
Table 6 Longitudinal (M0–M36) intraobserver reproducibility and sensitivity to change of joint space width measurements
View Observer JSW change (mm; mean ± SD) Difference (mm; mean ± SD) 95% CI approximation of bias Limits of agreements ICC value (95% CI) SRM calculated on reading 1 (95% CI)
Reading 1 Reading 2
Pelvic 1 -0.40 ± 0.65 -0.49 ± 0.54 0.09 ± 0.36 -0.01 to +0.19 -0.63 to +0.81 0.81 (0.69–0.89) -0.61 (-0.65 to -0.57)
2 -0.54 ± 0.66 -0.49 ± 0.67 -0.05 ± 0.16 -0.09 to -0.01 -0.37 to +0.27 0.97 (0.95–0.98) -0.82 (-0.84 to -0.80)
Mean (1 + 2) -0.47 ± 0.62 -0.49 ± 0.57 0.02 ± 0.18 -0.2 to +0.06 -0.34 to +0.38 0.95 (0.92–0.97) -0.76 (-0.77 to -0.73)
Hip AP 1 -0.39 ± 0.61 -0.46 ± 0.59 0.07 ± 0.30 -0.01 to +0.15 -0.53 to +0.67 0.87 (0.78–0.92) -0.64 (-0.67 to -0.61)
2 -0.49 ± 0.66 -0.49 ± 0.66 -0.00 ± 0.14 -0.04 to +0.04 -0.28 to +0.28 0.97 (0.96–0.99) -0.75 (-0.77 to -0.73)
Mean (1 + 2) -0.44 ± 0.62 -0.48 ± 0.60 0.03 ± 0.16 -0.01 to +0.07 -0.29 to +0.35 0.96 (0.94–0.98) -0.72 (-0.74 to -0.70)
Oblique 1 -0.46 ± 0.59 -0.37 ± 0.50 -0.09 ± 0.40 -0.21 to +0.03 -0.89 to +0.71 0.73 (0.57–0.84) -0.77 (-0.80 to -0.74)
2 -0.47 ± 0.67 -0.52 ± 0.64 0.05 ± 0.23 -0.10 to +0.11 -0.41 to +0.51 0.93 (0.88–0.96) -0.70 (-0.74 to -0.67)
Mean (1 + 2) -0.46 ± 0.61 -0.45 ± 0.53 0.03 ± 0.16 -0.10 to +0.07 -0.29 to +0.35 0.91 (0.85–0.95) -0.76 (-0.79 to -0.73)
Shown is the longitudinal intraobserver reproducibility and sensitivity to change (standardized response mean [SRM]) of radiographic measurements of joint space width (JSW; change from baseline [M0] to M36) on 3 different views read by two observers (and their mean). AP, anteroposterior; CI, confidence interval; ICC, intraclass coefficient of correlation; SD, standard deviation.
Table 7 Proportion of patients considered to be progressors
View Observer Patients with ≥ 0.3 mm change between M0 and M36 Patients with ≥ 0.5 mm change between M0 and M36
Pelvic 1 52% 34%
2 48% 40%
Both 40% 28%
Hip AP 1 52% 34%
2 54% 40%
Both 46% 32%
Oblique 1 56% 46%
2 52% 38%
Both 46% 34%
Shown are the proportions of patients considered to be progressors, using two different smallest detectable differences (SDDs) as cutoff values for defining progression (0.3 mm and 0.5 mm) for the two readers and three views. AP, anteroposterior.
==== Refs
Felson DT Zhang Y An update on the epidemiology of knee and hip osteoarthritis with a view to prevention Arthritis Rheum 1998 41 1343 1355 9704632 10.1002/1529-0131(199808)41:8<1343::AID-ART3>3.0.CO;2-9
Zhang W Doherty M Arden N Bannwarth B Bijlsma J Gunther KP Hauselmann HJ Herrero-Beaumont G Jordan K Kaklamanis P EULAR evidence based recommendations for the management of hip osteoarthritis: report of a task force of the EULAR Standing Committee for International Clinical Studies Including Therapeutic (ESCISIT) Ann Rheum Dis 2005 64 669 681 15471891 10.1136/ard.2004.028886
van Sasse JL van Romunde LK Cats A Vandenbroucke JP Valkenburg HA Epidemiology of osteoarthritis: Zoetermeer survey. Comparison of radiologic osteoarthritis in a Dutch population with that in 10 other populations Ann Rheum Dis 1989 48 271 280 2712610
Altman R Brandt K Hochberg M Moskowitz R Bellamy N Bloch DA Buckwalter J Dougados M Ehrlich G Lequesne M Design and conduct of clinical trials in patients with osteoarthritis: recommendations from a task force of the Osteoarthritis Research Society. Results from a Workshop Osteoarthritis Cartilage 1996 4 217 243 11048620
Lequesne M Chondrometry. Quantitative evaluation of joint space width and rate of joint space loss in osteoarthritis of the hip Rev Rhum Engl Ed 1995 62 155 158 7788330
Lequesne M Kuettner K, Goldberg V Quantitative measurements of joint space during progression of osteoarthritis: 'chondrometry' Osteoarthritis Disorders 1995 Rosemont: American Academy of Orthopedic Surgeons 427 444
Ravaud P Dougados M Radiographic assessment in osteoarthritis J Rheumatol 1997 24 786 791 9101519
Conrozier T Tron AM Balblanc JC Mathieu P Piperno M Fitoussi G Bochu M Vignon E [Measurement of the hip joint space using automatic digital image analysis] Rev Rhum Ed Fr 1993 60 137 143 8136806
Conrozier T Lequesne M Favre H Taccoen A Mazières B Dougados M Vignon M Vignon E Measurement of the radiological hip joint space width. An evaluation of various methods of measurement Osteoarthritis Cartilage 2001 9 281 286 11300752 10.1053/joca.2000.0386
Maillefert JF Sharp JT Aho LS Dougados M Comparison of a computer-based method and the classical manual method for radiographic joint space width assessment in hip osteoarthritis J Rheumatol 2002 29 2592 2596 12465158
Dougados M Gueguen A Nguyen M Berdah L Lequesne M Mazières B Vignon E Radiological progression of hip osteoarthritis: definition, risk factors and correlations with clinical status Ann Rheum Dis 1996 55 356 362 8694574
Lequesne M Maheu E Cadet C Dreiser RL Structural effect of avocado/ soybean unsaponifiables on joint space loss in osteoarthritis of the hip. Arthritis Rheum 2002 47 50 58 11932878 10.1002/art1.10239
Abadie E Ethgen D Avouac B Bouvenot G Branco J Bruyère O Calvo G Devogelaer JP Dreiser RL Herrero-Beaumont G Recommendations for the use of new methods to assess the efficacy of disease-modifying drugs in the treatment of osteoarthritis Osteoarthritis Cartilage 2004 12 263 268 15023377 10.1016/j.joca.2004.01.006
Altman RD Bloch DA Dougados M Hochberg M Lohmander S Pavelka K Spector T Vignon E Measurement of structural progression in osteoarthritis of the hip: the Barcelona consensus group Osteoarthritis Cartilage 2004 12 515 524 15219566 10.1016/j.joca.2004.04.004
Lequesne MG Laredo JD The faux profil (oblique view) of the hip in the standing position. Contribution to the evaluation of osteoarthritis of the adult hip Ann Rheum Dis 1998 57 676 681 9924210
Altman RD Fries JF Bloch DA Carstens J Cooke TD Genent H Gofton P Groth H McShane DJ Murphy WA Radiographic assessment of progression in osteoarthritis Arthritis Rheum 1987 30 1214 1225 3689459
Pessis E Chevrot A Drape JL Leveque C Sarazin L Minoui A Le Blevec G Chemla N Dupont AM Godefroy D Study of the joint space of the hip on supine and weight-bearing digital radiographs Clin Radiol 1999 54 528 532 10484220 10.1016/S0009-9260(99)90850-4
Conrozier T Lequesne MG Tron AM Mathieu P Berdah L Vignon E The effects of position on the radiographic joint space in osteoarthritis of the hip Osteoarthritis Cartilage 1997 5 17 22 9010875
Conrozier T Bochu M Gratacos J Piperno M Mathieu P Vignon E Evaluation of the "Lequesne false profile" of the hip in patients with hip osteoarthritis Osteoarthritis Cartilage 1999 7 295 300 10329304 10.1053/joca.1998.0203
Altman RD Alarcon G Appelrouth D Bloch D Borenstein D Brandt K Brown C Cooke TD Daniel W Feldman D The American College of Rheumatology criteria for the classfication and reporting of osteoarthritis of the hip Arthritis Rheum 1991 34 505 514 2025304
Bellamy N Musculoskeletal Metrology 1993 Dordrecht, The Netherlands: Kluwer Academic Publishers
Lequesne M Méry C Samson M Gérard P Indexes of severity for osteoarthritis of the hip and knee. Validation-value in comparison with other assessment tests Scand J Rheumatol Suppl 1987 65 85 89 3479839
McGraw KO Wong SP Forming inferences about some intraclass correlation coefficients Psychol Methods 1996 1 30 46 10.1037/1082-989X.1.1.30
Bland JM Altman DG Statistical methods for assessing agreement between two methods for clinical measurement Lancet 1986 1 307 310 2868172
Shrout PE Fleiss JL Intraclass coefficients: uses in assessing rater reliability Psychol Bull 1979 86 420 428 10.1037//0033-2909.86.2.420
Shao J Tu D The Jackknife and the Bootstrap 1996 New York: Springer-Verlag
|
16277690
|
PMC1297581
|
CC BY
|
2021-01-04 16:02:39
|
no
|
Arthritis Res Ther. 2005 Oct 5; 7(6):R1375-R1385
|
utf-8
|
Arthritis Res Ther
| 2,005 |
10.1186/ar1831
|
oa_comm
|
==== Front
Arthritis Res TherArthritis Research & Therapy1478-63541478-6362BioMed Central London ar18321627768910.1186/ar1832Research ArticleMost nuclear systemic autoantigens are extremely disordered proteins: implications for the etiology of systemic autoimmunity Carl Philip L [email protected] Brenda RS [email protected] Philip L [email protected] Department of Pharmacology, University of North Carolina, Chapel Hill, NC 27599, USA2 R. L. Juliano Structural Bioinformatics Core Facility, University of North Carolina, Chapel Hill, NC 27599, USA3 Division of Rheumatology, University of Pennsylvania School of Medicine and Philadelphia VA Medical Center, Philadelphia, PA 19104, USA2005 6 10 2005 7 6 R1360 R1374 25 4 2005 2 6 2005 4 8 2005 31 8 2005 Copyright © 2005 Carl 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.
Patients with systemic autoimmune diseases usually produce high levels of antibodies to self-antigens (autoantigens). The repertoire of common autoantigens is remarkably limited, yet no readily understandable shared thread links these apparently diverse proteins. Using computer prediction algorithms, we have found that most nuclear systemic autoantigens are predicted to contain long regions of extreme structural disorder. Such disordered regions would generally make poor B cell epitopes and are predicted to be under-represented as potential T cell epitopes. Consideration of the potential role of protein disorder may give novel insights into the possible role of molecular mimicry in the pathogenesis of autoimmunity. The recognition of extreme autoantigen protein disorder has led us to an explicit model of epitope spreading that explains many of the paradoxical aspects of autoimmunity – in particular, the difficulty in identifying autoantigen-specific helper T cells that might collaborate with the B cells activated in systemic autoimmunity. The model also explains the experimentally observed breakdown of major histocompatibility complex (MHC) class specificity in peptides associated with the MHC II proteins of activated autoimmune B cells, and sheds light on the selection of particular T cell epitopes in autoimmunity. Finally, the model helps to rationalize the relative rarity of clinically significant autoimmunity despite the prevalence of low specificity/low avidity autoantibodies in normal individuals.
==== Body
Introduction
Why some proteins become autoantigens is one of the mysteries of immunology. Indeed, as Paul Plotz put it in a recent review, "The repertoire of target autoantigens is a Wunderkammer – a collection of curiosities – of molecules with no obvious linking principle" [1]. Most immunologists believe, probably with good reason, that making real progress in understanding and treating autoimmune diseases depends on solving this mystery.
While a single property might explain why these few proteins become autoantigens, it seems more likely that a combination of factors unites these proteins. Plotz divides such factors into four groups: structural properties, catabolism and fate after cell death, concentration and the microenvironment, and immunological and inflammatory properties. This paper will primarily deal with the first of Plotz's factors, the structural properties of autoantigens. Among the structural properties he lists are, citing the work of Dohlman and colleagues [2,3]: a highly charged surface, repetitive surface elements, bound nucleic acid, and the presence of a coiled coil. In this paper, we provide computational evidence that the first three of these properties can be understood as arising from the fact that most nuclear systemic autoantigens are extremely disordered proteins, and suggest that the fourth property, the presence of a coiled coil, occurs far less frequently than does disorder. We also show that several of the other factors mentioned by Plotz that may influence the selection of autoantigens also fit nicely into the picture of nuclear systemic autoantigens as extremely disordered proteins. We will argue that disordered proteins are apt to be poor activators of B cells for multiple reasons, and hence that B cells targeted to extremely disordered proteins are apt to escape immune deletion. Furthermore, because extremely disordered proteins tend to be highly sensitive to proteolysis and are predicted to have poor affinity for major histocompatibility complex (MHC) II, these proteins are also predicted to be under-represented as T cell epitopes. In the Discussion we propose a model of how the pool of potentially autoreactive B cells might subsequently become activated and lead to pathological consequences. This model explicitly incorporates the fact that, in addition to being disordered, the majority of nuclear systemic antigens are large complexes of highly expressed structural macromolecules. The model predicts that it should normally be difficult to identify T cell populations that activate autoimmune B cells, and that such activation might not require cell-to-cell contact between B and T cells. Considerable evidence supports both of these predictions. At the same time the model explains why, paradoxically, some type of T cell-B cell contact is required in the development of autoimmunity. Finally, the model provides insights into why a specific T cell epitope is most commonly associated with the SmB autoantigen in systemic lupus erythematosus (SLE).
Defining protein disorder
The dominant picture of protein structure is that proteins fold to a unique native state of lowest energy. There is now an increased appreciation that the native state may not be a single structure after all, but rather an ensemble of closely related structures [4,5]. More recently has come an appreciation that large regions of some proteins never fold at all, at least in the absence of a binding partner. Regions that lack a fixed tertiary structure as determined by weak or missing electron density in a solved X-ray structure are identified as intrinsically disordered. In what follows we shall use the terms 'disordered protein' and 'disordered region' somewhat interchangeably, while recognizing that a 'disordered protein' can have regions of extensive order. It is important to distinguish between a disordered region that has a multiplicity of structures and a region such as a loop that lacks alpha-helical or beta-sheet secondary structure but may exist in a single structure.
While some aspects of protein disorder were appreciated more than 50 years ago, we can thank Dunker and Obradovic and their colleagues [6] for the current renaissance of interest in the concept. A more rigorous discussion of the concept of protein disorder is provided by Dunker et al. [6,7]. Excellent recent reviews of protein disorder are provided by Uversky, Gillespie and Fink [8], Fink [9], and Dyson and Wright [10], who call such proteins 'natively unfolded' or 'intrinsically unstructured'.
To develop software capable of predicting disordered regions, Dunker, Obradovic and their colleagues analyzed experimentally determined structures with disordered regions. They developed a neural network model to predict disorder, trained on regions of missing electron density in X-ray structures and disordered regions in NMR structures. The current default PONDR® predictor at the PONDR® web site [11] is VL-XT [12-14]. It is a hybrid of three earlier predictors: VL1 used for internal regions starting and ending 11 residues from the protein terminus; XN, an amino terminus predictor; and XC, a carboxyl terminus predictor. These predictors use a variety of input attributes including coordination number, net charge, hydropathy, and the presence of particular combinations of amino acids. The false positive error rate, that is, the prediction of disorder when a region is known to be ordered, of the VL-XT predictor is estimated at 22% on a per residue basis. However, the predictor is far better at predicting long regions of disorder, so that the false positive rate per residue drops to 1.7% per residue for consecutive regions of predicted disorder ≥40 residues. Further details on the training and accuracy of the various PONDR® predictors are available on the PONDR® web site.
Some additional PONDR® predictors are available at DisProt [15], but these have not been used in this study.
PONDR® scores are characterized by a disorder index q, which can range from 0 to 1, and are averaged over a window of nine amino acids. The boundary between order and disorder is conventionally set at q = 0.5. There is no clear criterion for extreme disorder. In this paper we call a protein extremely disordered if it contains at least one long disordered region (LDR) of 39 or more consecutive residues as predicted by PONDR®.
One should note that there are now several other web-based predictors of protein disorder available based on different algorithms and training sets. Examples are the DISOPRED [16] and DISEMBL™ [17] predictors. DISEMBL™ also has a complementary program GlobPlot™ [18] that focuses on predicting order. For the 19 LDRs presented in the figures, we have also determined the degree of disorder using the two DISEMBL™ and the DISOPRED disorder predictors. For all the predictors, on average 57% to 70% of the residues in the LDR predicted by PONDR® were confirmed to be disordered. This agreement suggests that our conclusions about LDRs are not strongly dependent on the particular disorder predictor used.
Materials and methods
A database of 51 nuclear systemic autoantigens (hNuSysAAG) was generated by SWISS-PROT text searches using SRS [19] combined with literature searches for autoantigens not yet annotated in SWISS-PROT. Keywords used in searching SWISS-PROT included 'human (organism) and nuclear and (autoantigen or autoimmune or antigen)' or 'human (organism) and nuclear and (scleroderma or sclerosis or lupus or sjogren)'. In a few cases, for example, the histones, we added widely recognized systemic nuclear autoantigens that were not annotated as autoantigens in SWISS-PROT. Proteins were removed from the initial search results for the following reasons: non-nuclear subcellular location (although it is not always clear how to classify the cellular location of a protein that is largely located in the cytoplasm, such as Ro 52K, but that shuttles to the nucleus – we generally assigned a nuclear location to such proteins despite the degree of ambiguity involved); not related to a systemic autoimmune disease; origin in a complex that was autoantigenic, but the protein was not autoantigenic itself. Three additional control databases were generated from SWISS-PROT: 10,962 human proteins (hSP); 2,335 human nuclear proteins (hNuSP); and 8,627 human non-nuclear proteins (hNNuSP).
All the predictions of order/disorder presented in this paper were made with the VL-XT predictor available at the PONDR® web site [11]. The predictions of class II dependent T cell epitopes were made with the ProPred predictor [20].
Results
Most nuclear systemic autoantigens are predicted to contain extremely disordered regions
PONDR® predictions for proteins vary from highly ordered to almost completely disordered. In Fig. 1 we show typical patterns for several human proteins, none of which are known autoantigens, and all of which are in the Protein Data Bank (PDB) [21], a structural database that is known to contain largely ordered proteins. In contrast, the PONDR® plots of several nuclear systemic autoantigens are shown in Fig. 2. It is clear that the autoantigens shown in Fig. 2 are predicted to be far more disordered than the non-autoantigenic proteins shown in Fig. 1. To gain insight into the significance of the relationship between disorder and autoantigenicity, we performed analyses of the various databases described earlier.
Of the 51 autoantigens in our hNuSysAAG database, 76% of the proteins met our criterion for extreme disorder, which was comparable with 75% of the proteins in hNuSP. In contrast, only 49% of hSP and 42% of hNNuSP met our criterion for extreme disorder. Thus, while nuclear autoantigens are no more disordered than nuclear proteins as a whole, nuclear proteins in general are significantly more likely to be disordered than non-nuclear proteins. It is interesting to note that 50% of the proteins annotated in SWISS-PROT as autoantigens are nuclear proteins but only 21% of human proteins are nuclear, implying disorder may play a role in this enrichment of nuclear proteins as autoantigens.
Our results can be compared to a recent paper by Iakoucheva et al. [22] that demonstrated that proteins associated with cancer (79% of proteins) and proteins associated with signal transduction (66% of proteins) are more highly disordered than the typical eukaryotic protein in the SWISS-PROT database (47% of proteins) or the PDB (13% of proteins). Note that these authors have defined a long disordered region as 30 or more residues compared with our criterion of 39 or more residues. Using Iakoucheva et al.'s criterion, we found that 83% of the proteins in hNuSysAAG met the requirement for the long disordered region. Thus, the proteins in hNuSysAAG are at least as disordered as the cancer-associated and signaling proteins studied by Iakoucheva et al. [22].
Some additional evidence also suggests that disorder and autoantigenicity are linked. In particular, the most common autoantigens in the Sm particle are Sm B/B', Sm D1 and Sm D3. All three proteins contain a long disordered region ≥39 consecutive residues. In contrast, a PONDR® analysis of Sm E, Sm F, and Sm G, proteins in the Sm particle that are rarely if ever autoantigens, lack long disordered regions (data not shown).
Experimental evidence that nuclear systemic autoantigens are extremely disordered proteins
Certain experimental evidence suggests that most nuclear systemic autoantigens are indeed, as predicted, disordered. For example, the La autoantigen is known to be especially sensitive to proteolysis consistent with a disordered structure [23,24]. The amino terminus of DNA topoisomerase I has been shown to be disordered by limited proteolysis [25], circular dichroism and gel filtration [26]. Furthermore, the positively charged tails of the histones are proteolytically sensitive and are not observed to contribute electron density [27].
In general, it is difficult to crystallize extremely disordered proteins. Thus X-ray studies of extremely disordered proteins tend either to focus on the ordered domains of the proteins that can be readily crystallized, or are studies of protein complexes where some disordered domains become ordered on binding. While NMR studies are not restricted to proteins that can crystallize, only small proteins are readily amenable to NMR methods so that often only domains of larger proteins are studied. Despite these limitations, direct evidence illustrated in Fig. 3 indicates that PONDR® predictions of disordered regions correlate well with structural determinations for several nuclear systemic autoantigens.
The fact that the structural studies in each of these cases stop close to the predicted boundary between order and disorder strongly suggests that the indicated regions have been correctly identified as disordered by PONDR®. Some of the disparity between prediction and experiment may be explained by complex formation. For example, in topoisomerase I, PONDR® predicts disorder from 365–404 and 437–475 whereas structures of topoisomerase I in complex with DNA show these regions are ordered. These residues possibly act as linkers connecting domains of topoisomerase I that interact with opposite sides of the DNA; they may be unstructured in the apoprotein and become ordered upon binding DNA.
Properties of disordered proteins of relevance to the nature of autoantigens
The amino acid composition of disordered regions is distinct from that of ordered regions [6]. Typically disordered regions are deficient in Trp, Cys, Phe, Ile, Tyr, Val, Leu, and Asn. They are enriched in Ala, Arg, Gly, Gln, Ser, Pro, Glu, and Lys. This bias in amino acid composition is reflected in the fact that disordered regions typically have a strong net charge, which is the first attribute of autoantigens mentioned by Plotz [1]. One consequence of this skewed amino acid composition of disordered regions is that many strongly disordered regions have very low sequence complexity as measured by Shannon's entropy [13], which can in turn lead to a preference for repetitive surface elements, the second of Plotz's factors thought to influence autoantigen structure. (However, not all regions of low sequence complexity are disordered.) The low sequence complexity of autoantigens is readily observed using a Web-based tool such as the GlobPlot™ server [18]. Although statistics on the fraction of all proteins that contain segments of low complexity are not readily available, we note that of 24 low complexity regions found in 13 of the most common nuclear systemic autoantigens, all but two occur in regions of disorder as determined by PONDR® (data not shown).
Many functions have been ascribed to disordered proteins [7], but one of the most prominent is binding to nucleic acid [7,10]. This is also a factor mentioned by Plotz as a third characteristic of the structure of autoantigens. In addition, recent work [28] shows that sites of phosphorylation are correlated with sites of protein disorder. Because phosphorylation/dephosphorylation are factors mentioned by Plotz as likely to be important in the selection of autoantigens [1], this is one more piece of evidence, albeit indirect, that disorder is apt to play a role in this process. The fourth structural criterion characteristic of autoantigens noted by Plotz (citing Dohlman et al. [2]), is the predicted presence of a coiled coil. The mechanism by which coiled coils may promote antigenicity is unclear, but Howard et al. [29] showed that a region at the amino terminus of the autoantigen histidyl-tRNA synthetase (which Coils [30] predicts to be a strong coiled coil (data not shown)) may promote autoimmunity by activation of dendritic cells. When we examined our database of nuclear systemic autoantigens using the Coils predictor, we found that coiled coils were present in 29% of our proteins whereas long disordered regions were present in 76% of our proteins. (Dohlman et al. [2] report a value of 36.7% coiled coils in their database of systemic autoantigens compared to 8.7% in the SwissProt and 1.1% in the PDB.) Thus, in agreement with Dohlman et al. [2] coiled coils appear to be over-represented in our collection of nuclear systemic autoantigens. Coiled coils are predicted roughly as frequently in our autoantigens that have long disordered regions as in the minority that do not. However, it is interesting to note that the most frequently encountered nuclear systemic autoantigens, such as the histones, the Sm proteins, and the U1 and centromere binding proteins, are all completely devoid of predicted coiled coils and are extremely disordered. (It should be noted that Dohlman et al. [2] stated that U1 snRNP70K and CENB possessed coiled coils. However, using an updated version of the Coils predictor that was unavailable to Dohlman et al., we found that these two predictions were in error. When the predictions were run using additional weighting of the amino acids appearing in positions 1 and 4 of the heptad repeat, which helps to rule out false positives, we were unable to confirm the putative coiled coils.)
In some cases, a region predicted by PONDR® to be disordered overlaps with a region predicted by Coils to be a coiled coil. An example is Ro 52K. Here the two disordered regions are predicted to be 124–174 and 183–261; the predicted coiled coils cover 128–165 and 189–234. Ottosson et al. [31] present experimental evidence showing the peptide 200–239 'had a partly α-helical secondary structure with major contribution of random coil,' that is, both the Coils and the PONDR® predictor seemed to be partially correct. In summary, we have confirmed the results of Dohlman et al. [2] that coiled coils seem to be common in autoantigens, but there is currently no evidence that this conclusion conflicts with the prediction that nuclear systemic autoantigens are disordered.
Disordered regions are predicted to make poor T cell antigens
B cells generally require T cell help to become activated and secrete their antibody product. Although T cells are required for the production of antinuclear autoantibodies in multiple animal models and probably also in humans, it has been notoriously difficult to isolate nuclear antigen-reactive T cells and to explore their specificity and function. We examined the predicted ability of several nuclear systemic autoantigens to function as T cell epitopes (when presented by MHC class II molecules) and asked if these sequences resided in areas of disorder; we used the web server ProPred [20,32]. This site implements the computer program TEPITOPE, which predicts peptide sequences that offer promise as promiscuous T cell epitopes [33]. The available evidence, though limited, suggests that TEPITOPE predicts many sequences that are experimentally verified T cell epitopes, although it also predicts many sequences to be T cell epitopes that cannot be verified as such [34-36]. This latter point is hardly surprising as TEPITOPE's predictions are based solely on binding to MHC II and do not attempt to model cellular compartmentalization of the antigen and specific proteolysis of the protein. The most extensive analysis [37] suggests that at least 50% of TEPITOPES predictions are verifiable, although the data also suggest that predictions for certain MHC alleles may be more accurate than others. We wondered if disordered regions might be particularly poor candidates for strong binding to MHC II proteins and, therefore, unlikely to be T cell epitopes.
Representative results for several HLA-DR alleles are shown in Fig. 4. If one compares the overall pattern of PONDR® predictions from Fig. 2 with the T cell antigen prediction from Fig. 4, one can see that the strongly disordered regions of the PONDR® plots correspond to regions of the T cell epitope plot in which only a very few even potential epitopes are located. By a potential epitope we mean epitope represented by a peak in the ProPred output without necessarily considering whether that peak is above the threshold. In fact, the vast majority of the potential epitopes illustrated in Fig. 4 are below threshold and, therefore, would not be predicted to be epitopes. For reasons of space we only show the results for four alleles and the four autoantigens whose PONDR® plot was displayed in Fig. 2. For example, for Histone H1b in Fig. 2a the PONDR® plot shows strong disorder in the region from residues 1–51 and from 112–218. The former region in Fig. 4a is somewhat depleted of potential T cell epitopes and the latter nearly devoid of potential epitopes. For U1 RNP70K the PONDR® plot in Fig. 2b shows strong regions of disorder at residues 52–91, 162–209, and 224–418. Although there still appear to be some possible epitope candidates in the former two regions in Fig. 4b, the latter region is again nearly devoid of potential epitopes. In the PONDR® plot of Fig. 2c, the disordered regions of Ro 52K from 124–174 and 183–261 can readily be seen to correspond to a slight diminution in the frequency of prospective epitopes in Fig. 4c. While the effect here is far less dramatic than in the case of the three other autoantigens pictured, the degree of disorder seen in Fig. 2 for Ro 52K is considerably less than for the other autoepitopes. Finally, the strongly disordered region in Sm B/B' from residues 51–240 in Fig. 2d corresponds to a marked deficit of potential T cell candidates in the same region in Fig. 4d compared to the number of potential epitopes in the first 50 residues. An even more dramatic demonstration of the correspondence of regions of extreme disorder and a lack of potential T cell epitopes will be discussed in Fig. 5. Taken together, these data suggest that disordered regions, probably because of their conformational flexibility, masking by nucleic acids and other proteins and their proteolytic lability, make poor antigens. Thus, both intuitions about what makes a good antigen and the computational analysis of predicted MHC II T cell epitopes support the notion that there will be few T cells targeted to extremely disordered regions. Proteins with extensive regions of disorder are thus likely to elicit poor T cell responses. B cells reactive against these nuclear antigens are unlikely to receive cognate help, and would be neither activated nor deleted. These clones thus represent a potential source of autoreactive antibodies.
Autoantibodies recognize both ordered and disordered regions
Given that clones targeted to extremely disordered proteins are a potential source of autoimmune antibodies, it is natural to wonder if in fact one can subsequently detect autoantibodies directed against the disordered regions. The obvious way to explore this question is to compare epitope maps for some common autoantigens with the maps of disordered regions provided by PONDR®. This exercise is, however, more difficult than it might seem. For example, Moutsopoulos et al. [38] have reviewed the epitope mapping data for Ro 60 kD, Ro 52 kD, and La 48 kD. It is apparent from their paper that different groups using different techniques on different patient samples have identified different linear epitopes and that, for many of the autoantigens, most of the protein sequence has been identified as an autoepitope by one group or another. Nonetheless, one can ask if disordered regions ever appear as autoepitopes. The answer is a clear yes. For example, in Ro 52K multiple authors have located an autoepitope at residues 216–292. Much of this epitope overlaps with the predicted strongly disordered region in Ro 52K from residues 183–261 (see Fig. 2c). Similarly, autoantigen La shows a predicted strongly disordered region from residues 369–408, which is another region targeted by autoantibodies. Many other B cell epitopes to Sm B have been located largely at the carboxyl terminus of the protein [39]. As is readily seen in Fig. 2d, this region of the protein is predicted to be largely disordered. Furthermore, linear epitope mapping may not be finding the most relevant conformational epitopes. So while it is clear that many epitopes on autoantigens are located in disordered regions of the antigen, it is also true that large regions of autoantigens are often autoepitopes, rendering any correspondence between disordered regions and autoepitopes less than convincing.
Protein disorder and epitope spreading
Spreading describes the extension of immune reactivity from an initial region of strong antigenicity towards a polypeptide into other epitopes of the autoantigen, or even from an epitope in one polypeptide to another polypeptide in a macromolecular complex such as the nucleosome or the Sm particle [40,41]. Spreading can lead to a more rapid and intense secondary response, longer lasting immune memory and multiple other advantages [40]. In a disease such as SLE, the reactivity can even extend into a different type of macromolecule such as DNA or RNA. Judith James and her colleagues have carried out several elegant experimental demonstrations of spreading. In a key study [42] they showed that immunization of rabbits with the peptide PPPGMRPP, a repeated sequence within the carboxyl terminus of Sm B/B', led to a spreading of the B cell response to many different structures on the SmB/B' autoantigen. A salient observation was that the antibodies reactive against these secondary determinants were in general not cross-reactive with the initiating peptide. In subsequent work [43], these authors showed that the closely related peptide PPPGRPP found in the nuclear antigen 1 (EBNA1) of the Epstein-Barr virus (EBV) was also capable of eliciting a lupus-like disease in rabbits. This result is of great interest given the evidence that the authors cite that EBV may be an etiological agent of autoimmune disease. A reasonable hypothesis is thus that EBV might attempt to circumvent immune surveillance by utilizing molecular mimicry. The subsequent attempt to deal with an EBV infection might lead to an autoimmune attack, initially on similar sequences in the B/B' polypeptide followed by spreading to the rest of the Sm particle.
To further explore the relevance of disorder to the idea of spreading we carried out a PONDR® analysis of the EBNA1 protein. The results are shown in Fig. 5. The results shown in Fig. 5a extend the notion of molecular mimicry [44] by suggesting that the EBNA1 protein has evolved to present, as nearly as possible, a disordered face to the immune system. The PPPGRPP epitope is one of the few regions of the protein that is relatively ordered, and because it mimics a self-antigen of Sm B/B' the immune system has a difficult job in defending against EBV infection. An antibody response against the ordered epitope risks subsequent development of autoimmune disease because the same spreading, which presumably allows defense against the disordered regions of EBNA1, carries the risk of a similar spreading to other epitopes in the Sm particle.
This view of the battle between the virus and the immune system is further amplified by the results of the analysis of MHC II T cell epitopes using the ProPred server shown in Fig. 5b. Here we can see that the extremely disordered regions of the virus contain essentially no predicted T cell epitopes in the context of MHC II. This is further strong evidence that a suspected pathogen implicated in autoimmune disease has escaped immune surveillance by using disorder to 'fly below' the level of sensitivity of the T cell receptor. Thus the virus seems to use both disorder and molecular mimicry as part of the infectious process. There have been earlier suggestions that protein disorder may allow viruses or presumably other pathogens to evade immune detection [45,46]. While the above example supports the notion of molecular mimicry as an important process in the development of autoimmune disease, we do not wish to suggest that other mechanisms that might lead to autoimmunity have been ruled out. Indeed, it seems that defects in apoptosis allowing exposure of cryptic disordered antigens to the immune system might be an important mechanism in many cases [12,47,48].
As another example of how a consideration of protein disorder can shed light on the phenomenon of spreading we consider further work from James' group [49]. They examined the immunogenicity and antigenicity in rabbits of two strong epitopes of the lupus autoantigen small nuclear ribonucleoprotein particle U1 snRNPA protein (also known as the U1A protein). One peptide, A3, was a strong immunogen, and in the months following initial immunization antibodies against this peptide exhibited spreading to other common epitopes of U1 snRNPA. In contrast, the A6 peptide was a weaker immunogen, and antibodies to this epitope do not show spreading. Not only was spreading associated solely with the A3 epitope, but also this epitope, unlike the A6 epitope, was able to induce clinical signs of autoimmune disease such as leukopenia and renal insufficiency. The authors asked why these two epitopes, located fairly close together in the same polypeptide, exhibit such different immunological and pathological properties. They point out that the two peptides have similar high isoelectric points, which are fair indicators of antigenicity in the snRNP system, and that A6, like some other autoimmune epitopes, is relatively non-immunogenic. It may be significant that, as shown in the PONDR® plot in Fig. 6, the A3 epitope that is capable of inducing spreading and autoimmune disease like the EBNA1 epitope shown in Fig. 5, is in a strongly ordered region located adjacent to regions of strong disorder of the PONDR® plot. In contrast, the A6 epitope is in a region of strong disorder. Once again in support of these notions, we have carried out an analysis of the predicted T cell epitopes in these regions. The results shown in Fig. 6b confirm a paucity of T cell MHC II epitopes in the extremely disordered region 96–226. In particular, there are few even potential T cell epitopes predicted in the region from 103–115 where the A6 peptide is located.
Recent work on the mechanism of spreading from Gordon, McCluskey and colleagues [50] extending their earlier studies of the Ro/La system [51,52] suggest that one can obtain an antibody response to several regions of the La autoantigen following immunization with recombinant La. In contrast, when they immunized with Ro 52K or Ro 60K, the only region of La in which spreading was seen to occur was the carboxy-terminal region which, as shown in Fig. 3a, is the only region of La that is strongly disordered. These results are again consistent with the pattern of spreading moving from ordered to disordered regions.
Discussion
Any theory of autoimmunity needs to account for at least two observations. The first is of the existence of large numbers of self-reactive immune cells, normally deleted or inactivated during tolerization, with specificity for a limited number of autoantigens. The second is that having escaped destruction, these immune cells can somehow subsequently become activated. The appreciation that many nuclear autoantigens are disordered can shed light on possible mechanisms by which both of these events can occur.
A priori one might expect the disordered regions of proteins to be poor antigens. By definition they exist in multiple conformations, which would suggest that it would be difficult to develop a conformation-specific antibody against such a region. In addition, disordered regions are very sensitive to proteolysis [7]. Furthermore, because disordered regions are often bound to other proteins or to nucleic acids, they may be masked and physically unavailable to the immune system [49]. Finally, as shown by the ProPred analysis, disordered regions are only rarely apt to be T cell epitopes. In summary, the recognition that most nuclear systemic autoantigens contain long disordered regions goes a long way towards explaining why a pool of potentially autoreactive B cells, of very low affinity that are targeted largely towards disordered regions, persists even in healthy individuals.
However, the very success of the concept of autoantigen disorder in explaining the persistence of B cells directed to self-epitopes only intensifies the difficulty of understanding how disordered regions could ever become the targets of autoimmune attack. Having argued that disordered regions are largely invisible to both T and B cells, how can we explain why in a few percent of individuals this invisibility is breached and autoimmune disease ensues? We agree with earlier authors that the key event is likely to be spreading. Although the data presented support the notion that spreading initiates at ordered epitopes and can spread through disordered regions to elicit autoimmune disease, we have said little about how this might occur. What exactly is the role of the ordered epitope in initiating spreading, and how might it contribute to the activation of the pool of self-reactive progenitor B cells potentially targeted to disordered regions? We suggest that a key to this process lies in the large size, high level of expression, and polyvalent nature of most of the nuclear systemic autoantigens and in particular the fact that frequently these autoantigens are part of structural macromolecular complexes. The model is diagrammed in Fig. 7.
In Fig. 7 the 'primary' progenitor B cell displaying the autoreactive surface Ig (sIg) binds to and processes the determinant and displays the resulting epitopes to T cells in the context of MHC II and accessory proteins in a typical immune synapse. We designate this sort of T cell cytokine activation of progenitor B cells 'cis' activation to indicate that the activation is to a progenitor B cell displaying sIg directly in close proximity to a T cell bearing a homologous T cell receptor via a conventional cognate cell to cell immune synapse.
In our view, the 'secondary' B progenitor cell in Fig. 7 becomes activated via a rather different mechanism. B cell progenitors capable of efficient, high-affinity binding to disordered determinants are few. Instead, there are many B cells that bind with low affinity to these determinants, a binding which is insufficient for their deletion or inactivation. The result is the persistence of large numbers of B cells reactive to disordered determinants on proteins. Furthermore, due to the proteolytic lability of strongly disordered peptides, peptides derived from disordered regions cannot be efficiently presented in the context of MHC II, as suggested by the gaps in the T cell epitope profile for disordered regions shown in Figs 4, 5, 6. Thus, it is difficult to present peptides derived from disordered regions in a conventional immune synapse. However, if a second snRNP should bind to a cross reactive B cell progenitor via its own sIg, a 'secondary' B cell progenitor with sIg that binds (weakly) to the circular epitope can be brought into close proximity of the activated T cell, such that cytokines from this T cell, which can act only over short distances, can act in 'trans' to activate the secondary B progenitor cell. This secondary B cell progenitor is thus activated by 'eavesdropping' on the signals being sent from the T cell to the primary B progenitor cell. In this model, the snRNP acts as a molecular scaffold to bring the two B progenitor cells into close proximity of the T cell to allow cytokine eavesdropping.
The model outlined in Fig. 7 bears some resemblance to other published models of spreading [53,54] but it differs in several notable aspects. Fatenejad and Craft [51] propose two models for spreading. In both models, a T cell is able to activate B cells with different specificities, but all the T/B interactions proceed via a conventional immune synapse. This is also the case for the model of Deshmukh et al. [52]. The weakness of all these models is that they presuppose secondary B cells binding to and processing secondary antigens in the absence of T cell help. Yet if this process occurred with any frequency, one would expect that such B cells would be activated and subject to immune deletion. The key difference between our model and those proposed is that we do not require such processing by potentially autoimmune B cell precursors because the secondary B cell precursors use 'eavesdropping' to obviate the need for conventional T/B immune synapses. Our model makes several predictions about the nature of the T and B cells that participate in autoimmune disease. Some of these predictions are characteristic of a wide range of models of autoimmunity and are, therefore, not terribly informative in deciding for or against the model. Still, it is important to note that the model is consistent with a great deal of information that is available about autoimmunity, for example, that autoimmune B cell populations are clonal, that the diseases are antigen driven, that the presence of an autoreactive B cell receptor is insufficient in itself to drive that cell into an autoimmune response, that there is no global defect in B cell elimination in autoimmunity and that helper T cells are required for autoimmune disease [55].
More telling are some less obvious predictions. The first of these is that only T cells with a very limited range of specificities are needed to activate secondary B cells carrying a wide range of specificities. The study of T-cell clones specific for several autoantigens of snRNPs strongly supports this prediction [56]. The model also predicts that soluble factors alone are insufficient to drive autoantibody production, and that some of the interactions in systemic autoimmunity are MHC II restricted [57].
Another prediction of the model is that one might find associated with the MHC II protein of secondary B cells peptides that would normally not have access to the MHC II pathway. This breakdown of pathway specificity might occur because there is no T cell synapse to ensure that only class II peptides are presented by the secondary B cells. Such a breakdown in pathway specificity has been experimentally observed [58]. Another prediction is that autoimmunity should be a relatively rare phenomenon on a per cell basis [59]. The model presupposes a syzygy of three immune cells linked via an autoantigen scaffolding. It seems likely that this is a relatively rare event compared to a normal T/B interaction of two cells, but made more likely for highly expressed proteins. The recent observation by Greidinger et al. [60] that direct T cell contact with B cells is not needed for T cell help in autoimmunity is also in striking agreement with our model.
Perhaps the most dramatic prediction of the model is that the T cell epitopes associated with autoimmunity should most frequently be derived from ordered regions of autoantigens because such regions can engage in conventional immune synapses. Talken et al. [61] have presented evidence supporting this prediction. They identified a series of peptides derived from SmB capable of stimulating T cell clones isolated from patients with SLE. They identified only three T cell epitopes from a total of seven patients. Of these peptides, a single peptide denoted as SmB-E1 comprising residues 16–33 of the SmB sequence was by far the most frequently encountered. This peptide was able to promote the growth of 23 out of 54 total T cell clones. Clones responsive to this peptide were present in five out of the seven patients. Considering that only a single clone was derived from two patients, it is apparent that this epitope is by far the most frequently encountered T cell epitope directed against SmB. A longitudinal analysis showed that response to this epitope remained stable for the two years of the study. Strikingly, this peptide (residues 16–33) is derived from the few residues (17%) in SmB that are predicted to be ordered (Fig. 2d) as predicted by the scaffolding model. We have assembled a considerable amount of additional evidence supporting this prediction (data not shown).
Conclusion
Nuclear autoantigens exhibit a remarkable degree of disorder. This property may explain the singular skewing of autoantibodies toward these nuclear proteins. We present a framework for considering how protein disorder might lead to autoreactivity. Our scheme unites the notions of tolerance, molecular mimicry, spreading and nucleic acid or protein binding by autoantigens into a coherent whole, but is conservative in that except for introducing the notion of disorder it does not posit any novel attributes of pathogens, the immune system, protein structure or autoantigens that have not been suggested in the past. There are preliminary suggestions that disorder may contribute to the development of autoantigens in the cytoplasm, such as the 60S acidic ribosomal proteins and golgins, and in some types of organ specific disease, such as multiple sclerosis (myelin basic protein), and celiac disease (tissue transglutaminase). Whatever the exact details that emerge from further analysis, we suggest that there is reason to suppose that protein order/disorder has a part to play in explaining the Wunderkammer of autoantigens.
Abbreviations
EBV = Epstein-Barr virus; hNNuSP = human non-nuclear protein database; hNuSP = human nuclear protein database; hNuSysAAG = human nuclear systemic autoantigen database; hSP = human protein database; LDR = long disordered region; MHC = major histocompatibility complex; sIg = surface Ig; SLE = systemic lupus erythematosus; snRNP = small nuclear ribonucleoprotein particle.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
DL Carl and DL Cohen are responsible for most of the analysis, interpretation, and writing of the manuscript. BRST contributed invaluable informatics and statistical input as well as suggesting key aspects of the scaffolding model.
Supplementary Material
Additional file 1
An Excel file containing a table listing the human nuclear systemic autoantigens in our study, along with their Swiss-Prot accession number, associated disease and length of all disordered regions of more than 20 residues.
Click here for file
Acknowledgements
Access to PONDR® was provided by Molecular Kinetics (6201 La Pas Trail – Ste 160, Indianapolis, IN 46268, USA; 317-280-8737; E-mail: [email protected]). VL-XT is copyright©1999 by the WSU Research Foundation, all rights reserved. PONDR® is copyright©2004 by Molecular Kinetics, all rights reserved. We also acknowledge the services provided by the Swiss Institute of Bioinformatics via the Expasy server [62,63] and access to the ProPred server [20]. We are grateful to Marshall Edgell and Paul Plotz for many useful discussions and comments on the manuscript and to the Department of Pharmacology, UNC-CH, for support of the RL Juliano Structural Bioinformatics Core Facility. DL Carl wishes to acknowledge the role of Doug Davies in introducing him to the field of protein disorder and Keith Dunker for useful comments on the manuscript. DL Cohen was supported by grants from the South Jersey Lupus Foundation, the US Department of Veterans Affairs, the Alliance for Lupus Research, and the National Institute of Arthritis and Musculoskeletal Diseases.
Figures and Tables
Figure 1 PONDR® predictions of disorder for four familiar human proteins. The SwissProt Accession Numbers [63] are given in parentheses. (a) Alpha-1-antitrypsin (P01009); (b) hemoglobin B (P02023); (c) calmodulin (P62158); (d) transthyretin precursor human (P02766). The line at PONDR® score 0.5 defines the disorder threshold and is an arbitrary measure used to distinguish order from disorder. The PONDR® predictor used here and in all other diagrams in this paper is VL-XT, which is the default predictor on the PONDR® web site.
Figure 2 The PONDR® plot of several autoantigens selected from Table 1 (Additional file 1). The proteins shown are: (a) histone H1b (P10412); (b) U1 RNP70K (P08621); (c) Ro 52K (P19474); (d) SmB/B (P14678). The heavy horizontal black bars indicate regions of 39 or more successive disordered residues with a PONDR® score greater than the threshold of 0.5.
Figure 3 PONDR® predictions compared to experimental structural determinations for various autoantigens. (a) La autoantigen (Swiss-Prot: P05455). The shaded box above the plot (residues 231–325) is the region that Jacks et al. [64] determined to be ordered via NMR. The empty boxes (residues 214–230 and residues 326–408) are regions determined to be unstructured or disordered. The inset (PDB: 1OWX; La222-334) shows the conformational flexibility of disordered regions at the amino and carboxyl terminii of the La fragment. (b) DNA topoisomerase I (Swiss-Prot: P11387). The structure was determined by X-ray methods for a protein-DNA complex (PDB: 1EJ9) encompassing residues 203–765 of DNA topoisomerase I. Residues 634–713 (empty box) are missing and, therefore, disordered in the structure [65]. The lightly shaded box at the amino terminus is the region that was determined to be disordered in the references cited above. (c) Histone H3 (Swiss-Prot: P68431). The structure of chicken H3 in a histone octamer complex (PDB: 2HIO) was determined by X-ray methods for residues 1–135. Residues 1–42 are missing, presumably due to disorder [66]. (d) Sm D1 (Swiss-Prot: P62314). The structure of a protein complex between Sm D1 (residues 2–119) and Sm D2 was studied by X-ray methods (PDB: 1B34) [67]. Residues 82–119 from Sm D1 are missing from the structure.
Figure 4 T cell epitopes for several autoantigens predicted by the ProPred server. (a) histone H1b (Swiss-Prot: P10412). (b) U1 RNP70K (Swiss-Prot: P08621). (c) Ro 52K (Swiss-Prot: P19474). (d) Sm B/B' (Swiss-Prot: P14678). Only four alleles are shown for each protein for the HLA antigens (from the top down): DRB1_0101; DRB1_0102; DRB_0301; and DRB1_0305. The patterns for the remaining MHC II alleles follow the same general trends. The black bars highlight the long disordered regions of the sequence as pictured in Fig. 2. The horizontal dotted red line is the threshold score-here set at the default value of 3%, which is used to differentiate between binders and non-binders. A threshold of 3% means that the protein sequence belongs to the 3% best scoring natural peptides. The lower the threshold percentage the fewer false positive peptides will be predicted to be T cell epitopes.
Figure 5 Disorder and T cell epitope prediction for EBV Nuclear Antigen 1. (a) PONDR® plot of the Epstein Barr Nuclear Antigen 1 protein (Swiss-Prot: P03211). The PPPGRPP epitope that induces cross-reactivity to an epitope on Sm B/B' is found in residues 398–412, almost exactly at the sharp minimum of the PONDR® plot. This is the only known cross-reacting epitope in the virus. (b) T cell epitopes of EBNA1 predicted by the ProPred server. Only the results for alleles HLA-DRB_01, HLA-DRB_0102, HLA-DRB1_0301, and HLA-DRB_0305 are shown. The remaining 47 alleles show a very similar picture. The threshold is set at 3%. The black bars delimit the strongly disordered regions of the PONDR® plot shown in (a). It is apparent that the highly disordered region of the first approximately 400 amino acids is predicted to be nearly devoid of potential T cell epitopes. The epitope from residues 398–412 that cross-reacts with the SmB protein is predicted to be most reactive with alleles HLA DRB5_0101 and DRB5_0105, although just slightly below a 3% threshold (data not shown).
Figure 6 Disorder and T cell epitope for U1 snRNPA. (a) PONDR® plot of the U1 snRNPA protein (Swiss-Prot: P09012). The location of the strongly immunogenic peptide A3 (residues 44–56), which induces spreading and systemic autoimmune disease, is indicated by XXX. The weakly immunogenic peptide A6 (residues 103–115), which does not induce spreading or autoimmune disease [49], is indicated by xxx. (b) ProPred analysis of the U1 snRNPA protein in the context of MHC II. Only the results for alleles HLA-DRB_01, HLA-DRB_0102, HLA-DRB1_0301, and HLA-DRB_0305 are shown. The remaining 47 alleles show a very similar picture. The threshold is set at 3%. The black bar delimits the long disordered region of (a).
Figure 7 A scaffolding model for antigen spreading. Shown is the target of an autoimmune response; here a snRNP particle that expresses at least two antigenic determinants. The determinant represented by the rectangle, which might be for example the PPPGRRP sequence on EBNA1, is assumed to cross react with the determinant PPPGMRPP on the snRNP via a conventional immune synapse. We denote progenitor B cells participating in these cognate interactions with T cells as primary progenitor B cells. Also shown is a second determinant on the small nuclear ribonucleoprotein particle (snRNP; represented by a circle) that is assumed to be more strongly disordered than the rectangular determinant. Progenitor B cells reacting with this determinant, termed secondary progenitor B cells, are capable of spreading the immune response via an eavesdropping mechanism as discussed in the text. EBV, Epstein-Barr virus.
==== Refs
Plotz PH The autoantibody repertoire: searching for order Nat Rev Immunol 2003 3 73 78 12511877 10.1038/nri976
Dohlman JG Lupas A Carson M Long charge-rich alpha-helices in systemic autoantigens Biochem Biophys Res Commun 1993 195 686 696 8373407 10.1006/bbrc.1993.2100
Brendel V Dohlman J Blaisdell BE Karlin S Very long charge runs in systemic lupus erythematosus-associated autoantigens Proc Natl Acad Sci U S A 1991 88 1536 1540 1996354
Wrabl JO Larson SA Hilser VJ Thermodynamic propensities of amino acids in the native state ensemble: implications for fold recognition Protein Sci 2001 10 1032 1045 11316884 10.1110/ps.01601
Arrington CB Teesch LM Robertson AD Defining protein ensembles with native-state NH exchange: kinetics of interconversion and cooperative units from combined NMR and MS analysis J Mol Biol 1999 285 1265 1275 9887275 10.1006/jmbi.1998.2338
Dunker AK Lawson JD Brown CJ Williams RM Romero P Oh JS Oldfield CJ Campen AM Ratliff CM Hipps KW Intrinsically disordered protein J Mol Graph Model 2001 19 26 59 11381529 10.1016/S1093-3263(00)00138-8
Dunker AK Brown CJ Lawson JD Iakoucheva LM Obradovic Z Intrinsic disorder and protein function Biochemistry 2002 41 6573 6582 12022860 10.1021/bi012159+
Uversky VN Gillespie JR Fink AL Why are "natively unfolded" proteins unstructured under physiologic conditions? Proteins 2000 41 415 427 11025552 10.1002/1097-0134(20001115)41:3<415::AID-PROT130>3.0.CO;2-7
Fink AL Natively unfolded proteins Curr Opin Struct Biol 2005 15 35 41 15718131 10.1016/j.sbi.2005.01.002
Dyson HJ Wright PE Intrinsically unstructured proteins and their functions Nat Rev Mol Cell Biol 2005 6 197 208 15738986 10.1038/nrm1589
PONDR® Predictors of Natural Disordered Regions
Li X Romero P Rani M Dunker AK Obradovic Z Predicting Protein Disorder for N-, C-, and Internal Regions Genome Inform Ser Workshop Genome Inform 1999 10 30 40 11072340
Romero P Obradovic Z Li X Garner EC Brown CJ Dunker AK Sequence complexity of disordered protein Proteins 2001 42 38 48 11093259 10.1002/1097-0134(20010101)42:1<38::AID-PROT50>3.0.CO;2-3
Romero P Obradovic Z Dunker K Sequence Data Analysis for Long Disordered Regions Prediction in the Calcineurin Family Genome Inform Ser Workshop Genome Inform 1997 8 110 124 11072311
DisProt – Database of Protein Disorder
The DISOPRED2 Disorder Prediction Server
DISEMBL™ – Intrinsic Protein Disorder Prediction
GlobPlot – Intrinsic Protein Disorder, Domain & Globularity Prediction
SRS Sequence Retrieval System at ExPASy
ProPred – MHC Class-II Binding Peptide Prediction Server
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
Iakoucheva LM Brown CJ Lawson JD Obradovic Z Dunker AK Intrinsic disorder in cell-signaling and cancer-associated proteins J Mol Biol 2002 323 573 584 12381310 10.1016/S0022-2836(02)00969-5
Habets WJ den Brok JH Boerbooms AM van de Putte LB van Venrooij WJ Characterization of the SS-B (La) antigen in adenovirus-infected and uninfected HeLa cells Embo J 1983 2 1625 1631 6357777
Casiano CA Tan EM Antinuclear autoantibodies: probes for defining proteolytic events associated with apoptosis Mol Biol Rep 1996 23 211 216 9112231 10.1007/BF00351171
Stewart L Ireton GC Champoux JJ The domain organization of human topoisomerase I J Biol Chem 1996 271 7602 7608 8631794 10.1074/jbc.271.13.7602
Stewart L Ireton GC Parker LH Madden KR Champoux JJ Biochemical and biophysical analyses of recombinant forms of human topoisomerase I J Biol Chem 1996 271 7593 7601 8631793 10.1074/jbc.271.13.7593
Luger K Richmond TJ The histone tails of the nucleosome Curr Opin Genet Dev 1998 8 140 146 9610403 10.1016/S0959-437X(98)80134-2
Iakoucheva LM Radivojac P Brown CJ O'Connor TR Sikes JG Obradovic Z Dunker AK The importance of intrinsic disorder for protein phosphorylation Nucleic Acids Res 2004 32 1037 1049 14960716 10.1093/nar/gkh253
Howard OM Dong HF Yang D Raben N Nagaraju K Rosen A Casciola-Rosen L Hartlein M Kron M Yiadom K Histidyl-tRNA synthetase and asparaginyl-tRNA synthetase, autoantigens in myositis, activate chemokine receptors on T lymphocytes and immature dendritic cells J Exp Med 2002 196 781 791 12235211 10.1084/jem.20020186
COILS – Prediction of Coiled Coil Regions in Proteins
Ottosson L Salomonsson S Hennig J Sonesson SE Dorner T Raats J Kuchroo VK Sunnerhagen M Wahren-Herlenius M Structurally derived mutations define congenital heart block-related epitopes within the 200–239 amino acid stretch of the Ro52 protein Scand J Immunol 2005 61 109 118 15683447 10.1111/j.0300-9475.2005.01542.x
Singh H Raghava GP ProPred: prediction of HLA-DR binding sites Bioinformatics 2001 17 1236 1237 11751237 10.1093/bioinformatics/17.12.1236
Sturniolo T Bono E Ding J Raddrizzani L Tuereci O Sahin U Braxenthaler M Gallazzi F Protti MP Sinigaglia F Hammer J Generation of tissue-specific and promiscuous HLA ligand databases using DNA microarrays and virtual HLA class II matrices Nat Biotechnol 1999 17 555 561 10385319 10.1038/9858
Manici S Sturniolo T Imro MA Hammer J Sinigaglia F Noppen C Spagnoli G Mazzi B Bellone M Dellabona P Protti MP Melanoma cells present a MAGE-3 epitope to CD4(+) cytotoxic T cells in association with histocompatibility leukocyte antigen DR11 J Exp Med 1999 189 871 876 10049951 10.1084/jem.189.5.871
Cochlovius B Stassar M Christ O Raddrizzani L Hammer J Mytilineos I Zoller M In vitro and in vivo induction of a Th cell response toward peptides of the melanoma-associated glycoprotein 100 protein selected by the TEPITOPE program J Immunol 2000 165 4731 4741 11035118
de Lalla C Sturniolo T Abbruzzese L Hammer J Sidoli A Sinigaglia F Panina-Bordignon P Cutting edge: identification of novel T cell epitopes in Lol p5a by computational prediction J Immunol 1999 163 1725 1729 10438899
Kwok WW Gebe JA Liu A Agar S Ptacek N Hammer J Koelle DM Nepom GT Rapid epitope identification from complex class-II-restricted T-cell antigens Trends Immunol 2001 22 583 588 11698198 10.1016/S1471-4906(01)02038-5
Moutsopoulos NM Routsias JG Vlachoyiannopoulos PG Tzioufas AG Moutsopoulos HM B-cell epitopes of intracellular autoantigens: myth and reality Mol Med 2000 6 141 151 10965491
Zieve GW Khusial PR The anti-Sm immune response in autoimmunity and cell biology Autoimmun Rev 2003 2 235 240 12965173 10.1016/S1568-9972(03)00018-1
James JA Harley JB B-cell epitope spreading in autoimmunity Immunol Rev 1998 164 185 200 9795776
Powell AM Black MM Epitope spreading: protection from pathogens, but propagation of autoimmunity? Clin Exp Dermatol 2001 26 427 433 11488833 10.1046/j.1365-2230.2001.00852.x
James JA Gross T Scofield RH Harley JB Immunoglobulin epitope spreading and autoimmune disease after peptide immunization: Sm B/B'-derived PPPGMRPP and PPPGIRGP induce spliceosome autoimmunity J Exp Med 1995 181 453 461 7530756 10.1084/jem.181.2.453
James JA Scofield RH Harley JB Lupus humoral autoimmunity after short peptide immunization Ann N Y Acad Sci 1997 815 124 127 9186649
Wucherpfennig KW Mechanisms for the induction of autoimmunity by infectious agents J Clin Invest 2001 108 1097 1104 11602615 10.1172/JCI200114235
House-Pompeo K Xu Y Joh D Speziale P Hook M Conformational changes in the fibronectin binding MSCRAMMs are induced by ligand binding J Biol Chem 1996 271 1379 1384 8576127 10.1074/jbc.271.3.1379
Kwong PD Doyle ML Casper DJ Cicala C Leavitt SA Majeed S Steenbeke TD Venturi M Chaiken I Fung M HIV-1 evades antibody-mediated neutralization through conformational masking of receptor-binding sites Nature 2002 420 678 682 12478295 10.1038/nature01188
Levine JS Koh JS The role of apoptosis in autoimmunity: immunogen, antigen, and accelerant Semin Nephrol 1999 19 34 47 9952279
Lorenz HM Herrmann M Winkler T Gaipl U Kalden JR Role of apoptosis in autoimmunity Apoptosis 2000 5 443 449 11256887 10.1023/A:1009692902805
McClain MT Lutz CS Kaufman KM Faig OZ Gross TF James JA Structural availability influences the capacity of autoantigenic epitopes to induce a widespread lupus-like autoimmune response Proc Natl Acad Sci U S A 2004 101 3551 3556 14988508 10.1073/pnas.0306267101
Gordon TP Kinoshita G Cavill D Keech C Farris A Kaufman K McCluskey J Purcell A Restricted specificity of intermolecular spreading to endogenous La (SS-B) and 60 kDa Ro (SS-A) in experimental autoimmunity Scand J Immunol 2002 56 168 173 12121436 10.1046/j.1365-3083.2002.01115.x
McCluskey J Farris AD Keech CL Purcell AW Rischmueller M Kinoshita G Reynolds P Gordon TP Determinant spreading: lessons from animal models and human disease Immunol Rev 1998 164 209 229 9795778
Topfer F Gordon T McCluskey J Intra- and intermolecular spreading of autoimmunity involving the nuclear self-antigens La (SS-B) and Ro (SS-A) Proc Natl Acad Sci U S A 1995 92 875 879 7846070
Fatenejad S Craft J Intrastructural help in diversification of humoral autoimmune responses Clin Exp Immunol 1996 106 1 4 8870690 10.1046/j.1365-2249.1996.d01-821.x
Deshmukh US Gaskin F Lewis JE Kannapell CC Fu SM Mechanisms of autoantibody diversification to SLE-related autoantigens Ann N Y Acad Sci 2003 987 91 98 12727627
Cohen P Paul WE Systemic Autoimmunity Fundamental Immunology 2003 Philadelphia: Lippincott Williams & Wilkins 1371 1400
Talken BL Bailey CW Reardon SL Caldwell CW Hoffman RW Structural analysis of TCRalpha and beta chains from human T-Cell clones specific for small nuclear ribonucleoprotein polypeptides Sm-D, Sm-B and U1–70 kDa: TCR complementarity determining region 3 usage appears highly conserved Scand J Immunol 2001 54 204 210 11439168 10.1046/j.1365-3083.2001.00930.x
Sobel ES Kakkanaiah VN Kakkanaiah M Cheek RL Cohen PL Eisenberg RA T-B collaboration for autoantibody production in lpr mice is cognate and MHC-restricted J Immunol 1994 152 6011 6016 8207226
Freed JH Marrs A VanderWall J Cohen PL Eisenberg RA MHC class II-bound self peptides from autoimmune MRL/lpr mice reveal potential T cell epitopes for autoantibody production in murine systemic lupus erythematosus J Immunol 2000 164 4697 4705 10779775
Zhang X Smith DS Guth A Wysocki LJ A receptor presentation hypothesis for T cell help that recruits autoreactive B cells J Immunol 2001 166 1562 1571 11160197
Greidinger EL Gazitt T Jaimes KF Hoffman RW Human T cell clones specific for heterogeneous nuclear ribonucleoprotein A2 autoantigen from connective tissue disease patients assist in autoantibody production Arthritis Rheum 2004 50 2216 2222 15248220 10.1002/art.20287
Talken BL Schafermeyer KR Bailey CW Lee DR Hoffman RW T cell epitope mapping of the Smith antigen reveals that highly conserved Smith antigen motifs are the dominant target of T cell immunity in systemic lupus erythematosus J Immunol 2001 167 562 568 11418695
ExPASy Proteomics Server
Gasteiger E Gattiker A Hoogland C Ivanyi I Appel RD Bairoch A ExPASy: The proteomics server for in-depth protein knowledge and analysis Nucleic Acids Res 2003 31 3784 3788 12824418 10.1093/nar/gkg563
Jacks A Babon J Kelly G Manolaridis I Cary PD Curry S Conte MR Structure of the C-terminal domain of human La protein reveals a novel RNA recognition motif coupled to a helical nuclear retention element Structure (Camb) 2003 11 833 843 12842046 10.1016/S0969-2126(03)00121-7
Redinbo MR Champoux JJ Hol WG Novel insights into catalytic mechanism from a crystal structure of human topoisomerase I in complex with DNA Biochemistry 2000 39 6832 6840 10841763 10.1021/bi992690t
Arents G Burlingame RW Wang BC Love WE Moudrianakis EN The nucleosomal core histone octamer at 3.1 A resolution: a tripartite protein assembly and a left-handed superhelix Proc Natl Acad Sci U S A 1991 88 10148 10152 1946434
Kambach C Walke S Young R Avis JM de la Fortelle E Raker VA Luhrmann R Li J Nagai K Crystal structures of two Sm protein complexes and their implications for the assembly of the spliceosomal snRNPs Cell 1999 96 375 387 10025403 10.1016/S0092-8674(00)80550-4
|
16277689
|
PMC1297582
|
CC BY
|
2021-01-04 16:02:40
|
no
|
Arthritis Res Ther. 2005 Oct 6; 7(6):R1360-R1374
|
utf-8
|
Arthritis Res Ther
| 2,005 |
10.1186/ar1832
|
oa_comm
|
==== Front
Arthritis Res TherArthritis Research & Therapy1478-63541478-6362BioMed Central London ar18331627768710.1186/ar1833Research ArticleReduced transforming growth factor-beta signaling in cartilage of old mice: role in impaired repair capacity Blaney Davidson EN [email protected] A [email protected] EL [email protected] der Kraan PM [email protected] den Berg WB [email protected] Experimental Rheumatology and Advanced Therapeutics, St Radboud University Medical Centre Nijmegen, Geert Grooteplein 26, 6525 GA Nijmegen, The Netherlands2005 30 9 2005 7 6 R1338 R1347 4 7 2005 26 7 2005 18 8 2005 1 9 2005 Copyright © 2005 Blaney Davidson 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.
Osteoarthritis (OA) is a common joint disease, mainly effecting the elderly population. The cause of OA seems to be an imbalance in catabolic and anabolic factors that develops with age. IL-1 is a catabolic factor known to induce cartilage damage, and transforming growth factor (TGF)-beta is an anabolic factor that can counteract many IL-1-induced effects. In old mice, we observed reduced responsiveness to TGF-beta-induced IL-1 counteraction. We investigated whether expression of TGF-beta and its signaling molecules altered with age. To mimic the TGF-beta deprived conditions in aged mice, we assessed the functional consequence of TGF-beta blocking. We isolated knee joints of mice aged 5 months or 2 years, half of which were exposed to IL-1 by intra-articular injection 24 h prior to knee joint isolation. Immunohistochemistry was performed, staining for TGF-beta1, -2 or -3, TGF-betaRI or -RII, Smad2, -3, -4, -6 and -7 and Smad-2P. The percentage of cells staining positive was determined in tibial cartilage. To mimic the lack of TGF-beta signaling in old mice, young mice were injected with IL-1 and after 2 days Ad-LAP (TGF-beta inhibitor) or a control virus were injected. Proteoglycan (PG) synthesis (35S-sulfate incorporation) and PG content of the cartilage were determined. Our experiments revealed that TGF-beta2 and -3 expression decreased with age, as did the TGF-beta receptors. Although the number of cells positive for the Smad proteins was not altered, the number of cells expressing Smad2P strongly dropped in old mice. IL-1 did not alter the expression patterns. We mimicked the lack of TGF-beta signaling in old mice by TGF-beta inhibition with LAP. This resulted in a reduced level of PG synthesis and aggravation of PG depletion. The limited response of old mice to TGF-beta induced-IL-1 counteraction is not due to a diminished level of intracellular signaling molecules or an upregulation of intracellular inhibitors, but is likely due to an intrinsic absence of sufficient TGF-beta receptor expression. Blocking TGF-beta distorted the natural repair response after IL-1 injection. In conclusion, TGF-beta appears to play an important role in repair of cartilage and a lack of TGF-beta responsiveness in old mice might be at the root of OA development.
==== Body
Introduction
Osteoarthritis (OA) is a common joint disease characterized by cartilage damage, osteophyte formation and thickening of the joint capsule. The etiology of OA is unknown, but OA is strongly correlated with age. OA may be a result of an age-related alteration in responsiveness of cells to anabolic and catabolic stimuli.
IL-1 is a cytokine that plays an important catabolic role in OA. IL-1 is highly expressed by chondrocytes of joints that are affected by OA, both in mice and humans [1,2]. Patients with OA have high levels of IL-1 in their synovial fluids as well [3]. IL-1 itself can induce cartilage damage [4] by reducing proteoglycan (PG) synthesis, increasing matrix metalloproteinase expression [5], and stimulating nitric oxide production [6].
Transforming growth factor (TGF)-beta is an important anabolic factor in OA. It is very beneficial for cartilage as it stimulates PG and collagen type II synthesis and can downregulate cartilage-degrading enzymes [7-13]. In addition, TGF-beta is able to counteract IL-1 induced suppression of PG synthesis [9,14-16]. Through this action TGF-beta is able to protect cartilage from damage by IL-1 [9,17,18]. In humans, expression of an asporin variant with a high TGF-beta inhibitory effect is significantly correlated with an increased incidence of OA [19].
Old animals show more prolonged suppression of PG synthesis after IL-1 exposure than young mice [4] and display a reduced response to counteraction of IL-1 by TGF-beta [20]. This indicates a shift in response to catabolic and anabolic stimuli, eventually leading to loss of cartilage homeostasis and OA.
TGF-beta signals predominantly through two receptors, TGF-beta-RI (ALK5) and TGF-beta-RII. TGF-beta binds to the type II receptor, recruits and phosphorylates the type I receptor and subsequently activates its receptor Smad, Smad2 or Smad3, by phosphorylation [21]. Thereafter, the phosphorylated Smad2 or Smad3 forms a complex with the common-Smad, Smad4. The complex is subsequently translocated to the nucleus where TGF-beta responsive genes are transcribed [22]. Inside the cell there are also inhibitory Smads (Smad6 and Smad7) that can prevent TGF-beta signaling [23,24].
We postulate that the lack of responsiveness to TGF-beta counteraction of IL-1 in old mice is due to an overall lack of responsiveness to TGF-beta caused by a down regulation of receptors and/or Smad expression or and increase in inhibitory Smads. Therefore, we investigated the expression of the various TGF-betas (1, 2 and 3) as well as their signaling molecules (TGF-beta-RI and TGF-beta-RII, Smad2, Smad-2P, Smad3, Smad4, Smad6 and Smad7) immunohistochemically in the cartilage of knee joints of young and old mice. In addition, we assessed whether these expression levels were altered differently in young and old mice by intra-articular injection of IL-1α.
We show that old mice have a profoundly lower expression of TGF-beta receptors (I and II) than young mice, which correlates with less Smad-2 phosphorylation. IL-1 itself had little effect on the expression of TGF-beta signaling molecules in cartilage.
To investigate whether reduced TGF-beta response could cause the reduced repair capacity in old mice, we mimicked the lack of TGF-beta by blocking TGF-beta activity with latency associated peptide (LAP) after IL-1 insult. This demonstrated that endogenous TGF-beta was required for a normal repair response and that lack thereof aggravates cartilage damage.
Materials and methods
Animals
Male C57Bl/6 mice aged 5 months or 2 years were used. Animals were kept in filtertop cages with woodchip bedding under standard pathogen free conditions. They were fed a standard diet with tap water ad libitum. The local animal committee approved this study.
Experimental design
TGF-beta counteraction of IL-1 effects is most likely mediated by TGF-betaRI, TGF-betaRII and the intercellular Smad proteins. We investigated if young (n = 14) and old (n = 14) mice differ in expression of these TGF-beta signaling mediators. Therefore, knee joints were isolated and prepared for immunohistochemistry. Half of the joints were prepared for paraffin sections, half were prepared for frozen sections. The number of cells staining positive for the various proteins were measured with a computerized imaging system. In addition to the comparison between young and old mice, we checked whether IL-1α injection 24 h prior to knee joint isolation (10 ng) (R&D Systems, Wiesbaden, Germany) influenced the expression patterns. Thus, the right knee joint of every mouse was injected with IL-1α and the left knee served as the non-injected group.
To assess whether lack of TGF-beta could indeed cause reduced repair capacity in old mice, young mice (n = 14) were injected intra-articularly with IL-1. Two days later we injected an adenovirus over-expressing the TGF-beta inhibitor LAP [25]. This inhibitor scavenges endogenous TGF-beta in the synovial fluid, preventing it from binding to its receptor. After 4 days, patellae were isolated for measurement of PG synthesis or whole knee joints were isolated for histology to measure PG content in the cartilage.
Histology
For the different classes of Smads, knee joints were decalcified for 14 days in EDTA/PVP and subsequently cryosections of total knee joints (7 μM) were prepared and stored at -20°C. Before use, sections were air-dried for 30 minutes and freshly prepared paraformaldehyde (4%, 5 minutes) was used to fix the sections.
Immunohistochemistry for TGF-beta1, TGF-beta2, TGF-beta3, TGF-betaRI, TGF-betaRII and Smad-2P, as well as Safranin O/Fast Green staining, were performed on paraffin sections from total knee joints. Knee joints were fixed in phosphate buffered formalin for 7 days. They were dehydrated using an automated tissue-processing apparatus (Tissue Tek VIP, Sakura, Ramsey, MN, USA) and embedded in paraffin. Tissue sections of 7 μM were prepared.
Immunohistochemistry
Sections were deparaffinized and washed with PBS. For antigen unmasking, sections were incubated in citrate buffer (0.1 M sodiumcitrate, 0.1 M citric acid) for 2 h. Endogenous peroxidase was blocked with 1% hydrogen peroxidase in methanol for 30 minutes. Thereafter, sections were blocked with 5% normal serum of the species in which the secondary antibody was produced. Specific primary antibodies against TGF-beta1, TGF-beta2, and TGF-beta3 (1.0 μg/ml), TGF-betaRI and TGF-betaRII, Smad2, Smad3 and Smad4 (0.5 μg/ml), Smad6 (1.0 μg/ml), Smad7 (3.3 μg/ml) and Smad-2P (1:100) were incubated overnight at 4°C. (Smad6 antibody was purchased from Invitrogen (Breda, The Netherlands), Smad-2P from Cell Signaling Technology (Beverly, MA, USA), and all other primary antibodies were purchased from Santa Cruz Biotechnology Inc (Santa Cruz, CA, USA)). As a negative control, the primary antibody was replaced with goat or rabbit IgGs. After extensive washing in PBS, the appropriate biotin labeled secondary antibody was used at a concentration of 2 μg/ml in 1% bovine serum albumin/PBS for 2 h (Vector Laboratories Inc, Burlingame, CA, USA), followed by a biotin-streptavidine detection system according to the manufacturer's protocol (Vector Laboratories Inc.). Bound complexes were visualized via reaction with 3',3'-diaminobenzidine (Sigma Chemicals Co, Zwijndrecht, The Netherlands) and H2O2 resulting in a brown precipitate. Sections were briefly counterstained with hematoxylin and mounted with Permount.
Image analysis: quantification of positively stained articular chondrocytes
For the different antigens, the number of positive articular chondrocytes in the tibia was determined by a blinded observer. The microscopic image was displayed on a computer monitor using the Qwin image analysis system (Leica Imaging Systems, Rijswijk, The Netherlands) and a Leica DC 300F digital camera. The area representing the non-calcified articular cartilage was selected by hand. For each antigen, a threshold was set in such a manner that only chondrocytes that were found to be positive (brown stained cell) as judged by the observer were selected. The computer program determined the number of positive cells in the cartilage for the different antigens. For each knee joint, the expression of the different antigens was measured in at least three tissue sections. The intensity of the staining was not taken into account as no obvious differences were observed in staining intensities in the different experimental groups: young/old or -IL-1/+IL-1. The obtained values were averaged and the average per treatment group was determined. To correct for differences in cell number between young and old mice, the average number of chondrocytes in the non-calcified cartilage was determined in sections stained with hematoxylin only for both paraffin and frozen sections. This was based on a similar selection procedure to that described above with the exception that selection of chondrocytes was based on the blue staining from heamatoxylin instead of brown staining. The average number of chondrocytes per sections was calculated for every joint.
Image analysis: proteoglycan content
PG content of articular cartilage was measured in sections stained with Safranin O and Fast Green using a computerized imaging system as previously described [25]. Briefly, Safranin O stains PGs in the cartilage red. A blinded observer captured an image on screen and selected the cartilage. The computer then measured the amount of blue light passing through the selected area. The higher the amount of light passing through, the lower the amount of PGs in cartilage. The average of three sections per knee joint was calculated.
Proteoglycan synthesis
PG synthesis was assessed by measurement of 35S-sulfate incorporation. Isolated patellae were immediately placed in Dulbecco's modified Eagle's medium with gentamicin (50 mg/ml) and pyruvate. After half an hour, this medium was replaced by medium containing 35S-sulfate 20 μCi/ml in which patellae were incubated for 3 h at 37°C and 5% CO2. Thereafter, patellae were further prepared for determining the amount of 35S-sulfate incorporation in the articular cartilage as previously described [22].
Statistical analysis
Results were analyzed with the Student's t-test and considered significant if the p-value was smaller than 0.05.
Results
Chondrocyte cell number is reduced with age
The percentage of cells expressing the different TGF-beta signaling proteins in murine cartilage was calculated by correction for the total number of cells present in the articular cartilage of the tibia. Therefore, the total number of cells in both medial and lateral tibial cartilage was quantified for all experimental groups by computerized quantification of cell number in hematoxyline and eosin (H&E) stained sections. Old mice had a significantly lower number of chondrocytes in the tibial cartilage: the reduction was more pronounced in the medial tibial cartilage, with a reduction in cell number of 34%; the number of cells in lateral tibial cartilage had reduced 17%. (Fig. 1). Treatment with IL-1 had no effect on the total number of cells (data not shown).
Reduction of various TGF-beta signaling molecules with age
To assess whether the reduced TGF-beta responsiveness in old mice was due to a lower amount of TGF-beta expression we compared the number of TGF-beta positive cells in the tibial cartilage of young (5 months old) and old mice (2 years old) in immunohistochemically stained sections. The number of positive cells was quantified with a computerized imaging system and corrected for the total amount of chondrocytes. In both medial and lateral tibial cartilage, age had no effect on the number of TGF-beta1 expressing cells. However, the number of cells expressing TGF-beta2 in old mice had reduced from 30% to almost no positive cells left (average of 0.2%) on the medial side of the joint and from 32% to 2% on the lateral tibial cartilage (Fig. 2f, h). TGF-beta3 showed a similar pattern in medial tibial cartilage, where the number of positive cells was 31% in young mice compared to 1% in old mice. On the lateral side of the joint, ageing also resulted in a lower number of TGF-beta3 positive cells, but this was not significant (Fig. 3).
We also examined the effect of aging on the number of cells staining positive for the TGF-betaRs. TGF-betaRI was expressed by a significantly lower number of cells in the medial tibial cartilage in old mice compared to young mice, 2% compared to 21%, respectively. On the lateral side, the number of TGF-betaRI positive cells was also lower in old mice, but this was not significant. The amount of cells expressing TGF-betaRII was significantly lower in old mice, both on the medial and on the lateral side of the joint. On the medial side, the number of immunopositive cells was reduced with age from 27% in young mice to 4% in old mice; in the lateral tibial cartilage the reduction was from 26% to 6% (Figs 4 and 2e, g).
In contrast to the receptors, the number of cells positive for the several Smad molecules had hardly changed with age. The percentage of cells positive for receptor-Smad Smad2 was equal in young and old mice (Fig. 2a, c). The expression of receptor-Smad Smad3 had increased in old mice as well as the percentage of cells positive for the common-Smad, Smad4, but only in the medial tibial cartilage. The inhibitory Smad, Smad6, was not altered with age in the medial tibial cartilage, but was higher in old mice on the lateral side. Smad 7 was significantly higher in old mice, but this was limited to the medial tibial cartilage (Fig. 5).
Despite the lack of difference in Smad2 expression between young and old mice, the phosphorylated variant of this Smad, Smad-2P, was significantly reduced in old mice in both medial and tibial cartilage. In the medial tibial cartilage, the percentage of cells staining positive for Smad-2P was 53% in young mice compared to 5% in old mice. On the lateral side, aging had lowered the amount of immunopositive cells from 85% to 30% (Figs 6 and 2b, d). This indicates a decrease in active TGF-beta signaling in old mice, possibly related to the decreased number of TGF-betaRs in old mice.
To assess whether IL-1 itself altered TGF-beta signaling in old mice, thereby reducing the counteractive abilities of TGF-beta to IL-1, we examined the effect of IL-1 injection on the expression of TGF-beta signaling components in the articular cartilage. Injection of IL-1 24 h prior to knee joint isolation resulted in an increased expression of TGF-beta1 and TGF-beta2 in lateral tibial cartilage in old mice and a higher number of Smad2 positive cells in the medial tibial cartilage. IL-1 treatment did not influence TGF-beta receptor, Smad or Smad2P expression in old mice (Fig. 7). IL-1 did not alter the expression of the TGF-beta signaling components in young mice.
Effect of blocking TGF-beta on proteoglycan synthesis and proteoglycan content
To assess the functional consequence of depressed TGF-beta signaling, we blocked TGF-beta by adenoviral overexpression of the TGF-beta inhibitor LAP two days after IL-1 insult. Four days after primary insult, knee joints were isolated for assessment of PG synthesis and PG content. PG synthesis was measured by 35S-sulfate incorporation into cartilage ex vivo. A normal response to IL-1 insult is an initial drop in PG synthesis the first 2 days after IL-1 injection, followed by a rapid increase in synthesis within the next 2 days [4]. The increased synthesis levels are above normal turnover levels. LAP over-expression after IL-1 injection was able to completely block this intrinsic increase in PG synthesis as shown by the 35S-sulfate incorporation, which was lower than after IL-1 insult alone (Fig. 8a).
In addition, PG content was measured by quantification of Safranin O staining intensity of the cartilage. The block of endogenous TGF-beta resulted in an aggravation of cartilage damage as the PG content of the cartilage was significantly reduced beyond IL-1 induced PG depletion (Fig. 8b). These data show that deprivation of TGF-beta resulted in a reduced repair capacity of the cartilage.
Discussion
OA is characterized by cartilage damage with an increasing incidence with age. The etiology of OA is unknown, but an imbalance between catabolic and anabolic factors appears to be involved. Whereas chondrocytes of young mice respond well to TGF-beta counteraction of IL-1, those of old mice show less efficient counteracting of IL-1 by TGF-beta [20]. In addition, they display prolonged suppression of PG synthesis. This might be due to a decreased response to TGF-beta in cartilage of old mice. We compared, therefore, the expression of TGF-beta and the TGF-beta signaling components in cartilage of young and old mice. The cartilage of old mice contained a lower number of cells than young mice. We thus corrected our findings for the total number of cells in the examined cartilage. In this study, only the tibial cartilage is discussed, but similar changes occurred in the femoral cartilage. The reduced cell number we found in old mice corresponds to the decreased number of cells that was found in cartilage of human donors older than 40 [26]. A decrease in chondrocyte cell number could be due to an age-related decline in (TGF-beta-induced) chondrocyte proliferation rate [27,28].
Our results show that old mice have significantly lower numbers of cells expressing TGF-beta2 and TGF-beta3 than young mice. In addition, old animals had a significantly lower number of chondrocytes expressing TGF-betaRs. The lack of responsiveness to TGF-beta counteraction in old mice is not likely a result of alterations in Smad expression, as they are unaffected or even elevated by aging. Smad3 was elevated in tibial cartilage, and in the medial tibial cartilage we found an elevation of Smad4 with age. The basal material for signaling inside the cell is present, only the action is lacking. This lack of action might be due to the reduced receptor expression combined with a drop in TGF-beta2 and TGF-beta3 in old mice. This could also explain the lower Smad2 phosphorylation in old mice. Smad2 itself is not a problem as it is present in equal numbers in both young and old mice, but if there are less receptors and less ligands, Smads are unlikely to be phosphorylated in high amounts.
In lateral tibial cartilage we found an elevation of Smad6 expression with age, while in medial tibial cartilage Smad7 was elevated with age; these changes were restricted to one cartilage surface only instead of both. Although it might contribute, it is unlikely that this elevation is the cause of the overall unresponsiveness to TGF-beta.
We wanted to make sure that IL-1 itself did not alter TGF-beta signaling and cause the reduced counteraction. Therefore, mice were exposed to IL-1 prior to knee joint isolation. IL-1 treatment had only little effect on TGF-beta signaling. In old mice, we found an upregulation of TGF-beta1 and TGF-beta2 in lateral tibial cartilage. In the medial tibial cartilage, we observed an IL-1-induced increase in Smad2. Although there was elevation of these factors, it had no effect on Smad-2P, indicating that IL-1 treatment did not alter the outcome of TGF-beta signaling.
Iqbal et al. [29] found a decrease in the expression of mRNA for TGF-beta1, TGF-beta2 and TGF-beta3 with age in equine cartilage, supporting our findings. It is not clear why the TGF-beta isoforms show a different pattern but it is known that all three isoforms are differentially regulated and have a different promotor region. Also, during embryogenis all three isoforms show a different, developmental stage related expression pattern [30]. Gómez-Camarillo et al. [31] also showed a progressive decrease of TGF-betaRI with age. Matsunaga et al. [32] found similar expression patterns in cervical intervertebral discs in mice. They showed a decrease in expression of TGF-beta1, TGF-beta2 and TGF-beta3 as well as TGF-betaRI and TGF-betaRII with age. In myogenic progenitor cells in mice, Beggs et al. [33] described similar observations. They found that TGF-betaRI and TGF-betaRII were downregulated and Smad2, Smad3, Smad4 and Smad7 remained unchanged [33]. These data indicate that our findings are similar to those found in other species and cell types and that the phenomenon of reduced TGF-betaRs and reduced TGF-beta expression it is not limited to cartilage of murine knee joints.
IL-1 treatment increased the expression of TGF-beta1 and TGF-beta2 in tibial cartilage. Andriamanalijaona et al. [34] have also shown the ability of IL-1 to increase TGF-beta1 of articular chondrocytes. Kaiser et al. [35] showed that IL-1 treatment resulted in elevated expression of Smad7 mRNA in vitro after 3 days. In our in vivo experiment, however, no significant alterations in inhibitory Smad expression were found. In contrast to Kaiser et al. [35], we measured the percentage of cells expressing Smad7 one day after IL-1 injection in vivo. The discrepancies in time, measurement and system probably explain why different results were found.
We previously examined TGF-beta expression in OA. In severe OA in STR/ort mice, we did not find any TGF-beta expression or Smad-2P at all, whereas younger STR/ort mice with only mild damage still expressed both factors (data not shown). In addition, others have also found discrepancies between OA cartilage and healthy cartilage with respect to TGF-beta expression. Gomez-Camarillo and Kouri [31] showed that TGF-beta1 receptors were very scarce in experimental OA. The drop in expression levels of TGF-beta and their signaling molecules that we found in old mice might precede OA.
The expression patterns in the cartilage suggest that a lack of TGF-beta signaling plays a potential role in the reduced repair capacity in old mice and possibly in OA. To further investigate whether the disturbed TGF-beta signaling could cause a reduction in repair, we inhibited endogenous TGF-beta after IL-1 insult. This resulted in a total block of the increased PG synthesis, thereby reducing the intrinsic repair capacity of the cartilage. The reduced PG synthesis resulted in an aggravation of the IL-1-induced PG loss in cartilage. These results show that not only do old mice have a reduced TGF-beta signaling capacity, but also that disrupted TGF-beta signaling can indeed induce a distorted repair capacity of cartilage.
It has been hypothesized that TGF-beta treatment can be used as a factor for cartilage repair. However, old mice respond poorly to TGF-beta, so the use of TGF-beta for repair might be more difficult than expected. It has already been shown that human articular chondrocytes stimulated with TGF-beta1, fibroblast growth factor-2 and platelet derived growth factor-BB, contained more glycosaminoglycans than non-stimulated controls, but only if donors were younger than 40 [26]. In addition, stimulation of equine articular cartilage with TGF-beta resulted in lower [35S]Na2SO4 incorporation in horses of 20 years old compared to 9 month old horses [26,29]. Although the response to TGF-beta is reduced with age, it does not mean that the cartilage does not respond at all. There was still an increase in incorporation of [35S]Na2SO4 after TGF-beta stimulation found by Livne et al. [36] in mice, but it has to be considered that this response in old animals cannot be compared to the massive stimulation in young animals. However, finding ways to stimulate cartilage repair bypassing the TGF-beta receptor pathway appears to be an attractive option to boost repair of aged cartilage.
Conclusion
Our data show that there are less chondrocytes expressing TGF-betaRs in cartilage in old mice. Smad expression is unchanged, but Smad2 phosphorylation is reduced with age. These data suggest that the reduced TGF-beta counteraction of IL-1 induced cartilage damage of old mice is due to an overall lack in TGF-beta signaling capacity. Blocking endogenous TGF-beta in young mice induced a distorted repair capacity in cartilage. The reduced ability of chondrocytes to respond to anabolic factors during aging might play a role in the development of the age-related disease OA.
Abbreviations
IL-1 = interleukin-1; LAP = latency associated peptide; OA = osteoarthritis; PBS = phosphate buffered saline; PG = proteoglycan; TGF-beta = transforming growth factor-beta; TGF-betaR = transforming growth factor-beta receptor.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
ENBD participated in the animal experiments and immunohistochemistry, carried out histological measurements, analyzed the data and drafted the manuscript. AS participated in the animal experiments, immunohistochemistry and analysis of the young versus old mice comparison. ELV participated in the animal experiments, carried out histological processing of the knee joints, participated in immunohistochemistry and performed 35S-sulfate measurements. PMK conceived of the study, participated in the design and coordination and helped to draft the manuscript. WBB participated in study design and revision of the final manuscript.
Acknowledgements
This study was supported by the Dutch Rheumatism Association "Het Nationaal Reumafonds".
Figures and Tables
Figure 1 Number of cells in medial and lateral tibial cartilage of 5 month and 2 year old C57Bl/6 mice. Paraffin sections of knee joints of young (5 months old) and old (2 years old) mice were stained with hematoxylin and eosin after which a blinded observer used a computerized imaging system to count the number of chondrocytes in the tibial cartilage. Old mice have a lower number of cells in their cartilage than young mice. The reduction in cell number is more pronounced on the medial side of the joint. Error bars display the standard error. For statistical analysis, a Student's t-test was used. * = p < 0.05; ** = p < 0.005; *** = p < 0.0005.
Figure 2 Staining of various transforming growth factor (TGF)-beta signaling molecules in cartilage. Paraffin sections of knee joints of young and old mice were stained with antibodies against (a,c) Smad2, (b,d) Smad-2P, (e,g) TGF-beta receptor II (TGF-betaRII) and (f,h) TGF-beta2. The medial tibia of the young mice clearly show a high number of cells staining positive for (b) Smad-2P, (e) TGF-betaRII and (f) TGF-beta2, whereas the (d,g,h) old mice had only a very low number of cells staining positive for these factors. (a,c) Smad2 staining remained unchanged with age. F, femur; T, tibia.
Figure 3 Percentage of cells expressing transforming growth factor (TGF)-beta in medial and lateral tibial cartilage. Paraffin sections of knee joints of young (5 months old) and old (2 years old) mice were stained immunohistochemically with antibodies against TGF-beta1, TGF-beta2 or TGF-beta3. Subsequently, the number of cells staining positive in the cartilage were scored with a computerized imaging system and corrected for the total number of cells. (a) In medial cartilage, TGF-beta2 and TGF-beta3 expression were significantly reduced with age. (b) In lateral cartilage, TGF-beta2 was significantly reduced. Error bars display the standard error. For statistical analysis, a Student's t-test was used. * = p < 0.05.
Figure 4 Percentage of cells expressing transforming growth factor (TGF)-beta receptors (TGF-betaRs). Paraffin sections of knee joints of young (5 months old) and old mice (2 years old) were stained immunohistochemically with antibodies against TGF-betaRI or TGF-betaRII. Subsequently, the number of cells staining positive in the cartilage were scored with a computerized imaging system and corrected for the total number of cells. The expression of both receptors was reduced with age in both (a) medial and (b) lateral tibial cartilage, but the reduced TGF-betaRII was significant only in lateral tibial cartilage. Error bars display the standard error. For statistical analysis, a Student's t-test was used. * = p < 0.05; ** = p < 0.005; *** = p < 0.0005.
Figure 5 Percentage of cells expressing Smad in medial and lateral tibial cartilage. Frozen sections of knee joints of young (5 months old) and old (2 years old) mice were stained immunohistochemically with antibodies against Smad2, Smad3, Smad4, Smad6 or Smad7. Subsequently, the number of cells staining positive were scored with a computerized imaging system and corrected for the total number of cells. (a) In medial tibial cartilage, expression of Smad3, Smad4 and Smad7 increased with age. (b) In lateral tibial cartilage Smad3 and Smad6 expression increased with age. Error bars display the standard error. For statistical analysis, a Student's t-test was used. * = p < 0.05; *** = p < 0.0005.
Figure 6 Percentage of cells expressing Smad-2P in medial and lateral tibial cartilage. Paraffin sections of knee joints of young (5 months old) and old (2 years old) mice were stained immunohistochemically with antibodies against Smad-2P. Subsequently, the number of cells staining positive were scored with a computerized imaging system and corrected for the total number of cells. The Smad-2P expression was reduced with age in both medial and lateral tibial cartilage. Error bars display the standard error. For statistical analysis, a Student's t-test was used. *** = p < 0.0005.
Figure 7 Effect of IL-1 on expression of TGF-beta signaling proteins in cartilage. Knee joints of (a) young (5 months old) and (b) old (2 years old) mice were injected with IL-1 24 h prior to isolation of the knee joints. Paraffin sections of knee joints were stained immunohistochemically for TGF-beta1, TGF-beta2, TGF-beta3, TGF-beta receptor I (TGF-betaRI), TGF-betaRII, Smad2, Smad3, Smad4, Smad6, Smad7 and Smad-2P. Subsequently, the number of cells staining positive were scored with a computerized imaging system and corrected for the total number of cells. After IL-1 injection, Smad2 expression increased only in the medial tibial cartilage and TGF-beta1 and TGF-beta2 expression increased only in the lateral tibial cartilage. Error bars display the standard error. For statistical analysis, a Student's t-test was used. * = p < 0.05.
Figure 8 Effect of transforming growth factor (TGF)-beta deprivation on intrinsic cartilage repair capacity. Murine knee joints of young mice were injected with IL-1. After two days an adenovirus expressing the TGF-beta inhibitor latency associated peptide (LAP) was injected intra-articularly. Four days after the initial injections with IL-1, patellae were isolated for 35S-sulfate incorporation and whole knee joints were isolated for histology. (a) 35S-sulfate incorporation into isolated patellar cartilage after treatment with IL-1 and Ad-LAP. IL-1 treatment induces an initial decrease in 35S-sulfate incorporation, but by day 4 the 35S-sulfate incorporation increased above normal levels, indicating an overshoot in proteoglycan synthesis. By blocking endogenous TGF-beta with LAP, this overshoot is completely abolished. (b) Proteoglycan content of patellar cartilage after treatment with IL-1 and Ad-LAP. IL-1 injection results in depletion of proteoglycans in cartilage. Blocking endogenous TGF-beta with LAP results in an aggravation of this depletion beyond IL-1 induced damage alone.
==== Refs
Moos V Fickert S Muller B Weber U Sieper J Immunohistological analysis of cytokine expression in human osteoarthritic and healthy cartilage J Rheumatol 1999 26 870 879 10229409
Towle CA Hung HH Bonassar LJ Treadwell BV Mangham DC Detection of interleukin-1 in the cartilage of patients with osteoarthritis: a possible autocrine/paracrine role in pathogenesis Osteoarthritis Cartilage 1997 5 293 300 9497936
Webb GR Westacott CI Elson CJ Osteoarthritic synovial fluid and synovium supernatants up-regulate tumor necrosis factor receptors on human articular chondrocytes Osteoarthritis Cartilage 1998 6 167 176 9682783 10.1053/joca.1998.0109
van Beuningen HM Arntz OJ van den Berg WB In vivo effects of interleukin-1 on articular cartilage. Prolongation of proteoglycan metabolic disturbances in old mice Arthritis Rheum 1991 34 606 615 2025312
Ganu V Goldberg R Peppard J Rediske J Melton R Hu SI Wang W Dunander C Heingard D Inhibition of interleukin-1alpha-induced cartilage oligomeric matrix protein degradation in bovine articular cartilage by matrix metalloproteinase inhibitors: potential role for matrix metalloproteinases in the generation of cartilage oligomeric matrix protein fragments in arthritic synovial fluid Arthritis Rheum 1998 41 2143 2151 9870871 10.1002/1529-0131(199812)41:12<2143::AID-ART9>3.0.CO;2-P
Khatib AM Siegfried G Messai H Quintero M Barbara A Mitrovic RD Basal and induced nitric oxide and cGMP productions are decreased in senescent cultured rat articular chondrocytes Mech Ageing Dev 1998 101 21 32 9593310 10.1016/S0047-6374(97)00150-4
Ballock RT Heydemann A Izumi T Reddi AH Regulation of the expression of the type-II collagen gene in periosteum-derived cells by three members of the transforming growth factor-beta superfamily J Orthop Res 1997 15 463 467 9246095 10.1002/jor.1100150321
Izumi T Scully SP Heydemann A Bolander ME Transforming growth factor beta 1 stimulates type II collagen expression in cultured periosteum-derived cells J Bone Miner Res 1992 7 115 121 1549955
van Beuningen HM van der Kraan PM Arntz OJ van den Berg WB Protection from interleukin 1 induced destruction of articular cartilage by transforming growth factor beta: studies in anatomically intact cartilage in vitro and in vivo Ann Rheum Dis 1993 52 185 191 8484670
van Beuningen HM van der Kraan PM Arntz OJ van den Berg WB In vivo protection against interleukin-1-induced articular cartilage damage by transforming growth factor-beta 1: age-related differences Ann Rheum Dis 1994 53 593 600 7979598
van Beuningen HM van der Kraan PM Arntz OJ van den Berg WB Transforming growth factor-beta 1 stimulates articular chondrocyte proteoglycan synthesis and induces osteophyte formation in the murine knee joint Lab Invest 1994 71 279 290 8078307
Edwards DR Murphy G Reynolds JJ Whitham SE Docherty AJ Angel P Heath JK Transforming growth factor beta modulates the expression of collagenase and metalloproteinase inhibitor EMBO J 1987 6 1899 1904 2820711
Edwards DR Leco KJ Beaudry PP Atadja PW Veillette C Riabowol KT Differential effects of transforming growth factor-beta 1 on the expression of matrix metalloproteinases and tissue inhibitors of metalloproteinases in young and old human fibroblasts Exp Gerontol 1996 31 207 223 8706790 10.1016/0531-5565(95)02010-1
Smith P Shuler FD Georgescu HI Ghivizzani SC Johnstone B Niyibizi C Robbins PD Evans CH Genetic enhancement of matrix synthesis by articular chondrocytes: comparison of different growth factor genes in the presence and absence of interleukin-1 Arthritis Rheum 2000 43 1156 1164 10817570 10.1002/1529-0131(200005)43:5<1156::AID-ANR26>3.0.CO;2-M
Redini F Mauviel A Pronost S Loyau G Pujol JP Transforming growth factor beta exerts opposite effects from interleukin-1 beta on cultured rabbit articular chondrocytes through reduction of interleukin-1 receptor expression Arthritis Rheum 1993 36 44 50 8424835
Takahashi N Rieneck K van der Kraan PM van Beuningen HM Vitters EL Bendtzen K van den Berg WB Elucidation of IL-1/TGF-beta interactions in mouse chondrocyte cell line by genome-wide gene expression Osteoarthritis Cartilage 2005 13 426 438 15882566 10.1016/j.joca.2004.12.010
Chandrasekhar S Harvey AK Transforming growth factor-beta is a potent inhibitor of IL-1 induced protease activity and cartilage proteoglycan degradation Biochem Biophys Res Commun 1988 157 1352 1359 3144973 10.1016/S0006-291X(88)81024-6
Lum ZP Hakala BE Mort JS Recklies AD Modulation of the catabolic effects of interleukin-1 beta on human articular chondrocytes by transforming growth factor-beta J Cell Physiol 1996 166 351 359 8591995 10.1002/(SICI)1097-4652(199602)166:2<351::AID-JCP13>3.0.CO;2-E
Kizawa H Kou I Iida A Sudo A Miyamoto Y Fukuda A Mabuchi A Kotani A Kawakami A Yamamoto S An aspartic acid repeat polymorphism in asporin inhibits chondrogenesis and increases susceptibility to osteoarthritis Nat Genet 2005 37 138 144 15640800 10.1038/ng1496
Scharstuhl A van Beuningen HM Vitters EL van der Kraan PM van den Berg WB Loss of transforming growth factor counteraction on IL-1-mediated effects in cartilage of old mice Ann Rheum Dis 2002 61 1095 1098 12429542 10.1136/ard.61.12.1095
Itoh S Itoh F Goumans MJ ten Dijke P Signaling of transforming growth factor-beta family members through Smad proteins Eur J Biochem 2000 267 6954 6967 11106403 10.1046/j.1432-1327.2000.01828.x
Shioda T Lechleider RJ Dunwoodie SL Li HC Yahata T de Caestecker MP Fenner MH Roberts AB Isselbacher KJ Transcriptional activating activity of Smad4: Roles of SMAD hetero-oligomerization and enhancement by an associating transactivator Proc Natl Acad Sci USA 1998 95 9785 9790 AU: please provide the names of the first 10 authors of ref.22 9707553 10.1073/pnas.95.17.9785
Imamura T Takase M Nishihara A Oeda E Hanai J Kawabata M Miyazono K Smad6 inhibits signalling by the TGF-beta superfamily Nature 1997 389 622 626 9335505 10.1038/39355
Nakao A Afrakhte M Moren A Nakayama T Christian JL Heuchel R Itoh S Kawabata M Heldin NE Heldin CH ten Dijke P Identification of Smad7, a TGFbeta-inducible antagonist of TGF- beta signalling Nature 1997 389 631 635 9335507 10.1038/39369
Scharstuhl A Vitters EL van der Kraan PM van den Berg WB Reduction of osteophyte formation and synovial thickening by adenoviral overexpression of transforming growth factor beta/bone morphogenetic protein inhibitors during experimental osteoarthritis Arthritis Rheum 2003 48 3442 3451 14673995 10.1002/art.11328
Barbero A Grogan S Schafer D Heberer M Mainil-Varlet P Martin I Age related changes in human articular chondrocyte yield, proliferation and post-expansion chondrogenic capacity Osteoarthritis Cartilage 2004 12 476 484 15135144 10.1016/j.joca.2004.02.010
Rosen F McCabe G Quach J Solan J Terkeltaub R Seegmiller JE Lotz M Differential effects of aging on human chondrocyte responses to transforming growth factor beta: increased pyrophosphate production and decreased cell proliferation Arthritis Rheum 1997 40 1275 1281 9214428
Guerne PA Blanco F Kaelin A Desgeorges A Lotz M Growth factor responsiveness of human articular chondrocytes in aging and development Arthritis Rheum 1995 38 960 968 7612045
Iqbal J Dudhia J Bird JL Bayliss MT Age-related effects of TGF-beta on proteoglycan synthesis in equine articular cartilage Biochem Biophys Res Commun 2000 274 467 471 10913361 10.1006/bbrc.2000.3167
Goumans MJ Mummery C Functional analysis of the TGFbeta receptor/Smad pathway through gene ablation in mice Int J Dev Biol 2000 44 253 265 10853822
Gomez-Camarillo MA Kouri JB Ontogeny of rat chondrocyte proliferation: studies in embryo, adult and osteoarthritic (OA) cartilage Cell Res 2005 15 99 104 15740638
Matsunaga S Nagano S Onishi T Morimoto N Suzuki S Komiya S Age-related changes in expression of transforming growth factor-beta and receptors in cells of intervertebral discs J Neurosurg 2003 98 1 Suppl 63 67 12546390
Beggs ML Nagarajan R Taylor-Jones JM Nolen G Macnicol M Peterson CA Alterations in the TGFbeta signaling pathway in myogenic progenitors with age Aging Cell 2004 3 353 361 15569352 10.1111/j.1474-9728.2004.00135.x
Andriamanalijaona R Felisaz N Kim SJ King-Jones K Lehmann M Pujol JP Boumediene K Mediation of interleukin-1beta-induced transforming growth factor beta1 expression by activator protein 4 transcription factor in primary cultures of bovine articular chondrocytes: possible cooperation with activator protein 1 Arthritis Rheum 2003 48 1569 1581 12794825 10.1002/art.11020
Kaiser M Haag J Soder S Bau B Aigner T Bone morphogenetic protein and transforming growth factor beta inhibitory Smads 6 and 7 are expressed in human adult normal and osteoarthritic cartilage in vivo and are differentially regulated in vitro by interleukin-1beta Arthritis Rheum 2004 50 3535 3540 15529348 10.1002/art.20750
Livne E Laufer D Blumenfeld I Differential response of articular cartilage from young growing and mature old mice to IL-1 and TGF-beta Arch Gerontol Geriatr 1997 24 211 221 15374127 10.1016/S0167-4943(96)00753-4
|
16277687
|
PMC1297583
|
CC BY
|
2021-01-04 16:02:39
|
no
|
Arthritis Res Ther. 2005 Sep 30; 7(6):R1338-R1347
|
utf-8
|
Arthritis Res Ther
| 2,005 |
10.1186/ar1833
|
oa_comm
|
==== Front
Arthritis Res TherArthritis Research & Therapy1478-63541478-6362BioMed Central London ar18341627768810.1186/ar1834Research ArticleEffect of a small molecule inhibitor of nuclear factor-κB nuclear translocation in a murine model of arthritis and cultured human synovial cells Wakamatsu Kyoko [email protected] Toshihiro [email protected] Nobuyuki [email protected] Kazuo [email protected] Tetsuo [email protected] Department of Microbiology and Immunology, Tokyo Medical and Dental University Graduate School of Health Sciences, Tokyo, Japan2 Department of Medicine and Rheumatology, Tokyo Medical and Dental University Graduate School, Tokyo, Japan3 Department of Applied Chemistry, Keio University, Kanagawa, Japan2005 30 9 2005 7 6 R1348 R1359 1 6 2005 30 6 2005 30 8 2005 2 9 2005 Copyright © 2005 Wakamatsu 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 small cell-permeable compound, dehydroxymethylepoxyquinomicin (DHMEQ), does not inhibit phosphorylation and degradation of IκB (inhibitor of nuclear factor-κB [NF-κB]) but selectively inhibits nuclear translocation of activated NF-κB. This study aimed to demonstrate the antiarthritic effect of this novel inhibitor of the NF-κB pathway in vivo in a murine arthritis model and in vitro in human synovial cells. Collagen-induced arthritis was induced in mice, and after onset of arthritis the mice were treated with DHMEQ (5 mg/kg body weight per day). Using fibroblast-like synoviocyte (FLS) cell lines established from patients with rheumatoid arthritis (RA), NF-κB activity was examined by electrophoretic mobility shift assays. The expression of molecules involved in RA pathogenesis was determined by RT-PCR, ELISA, and flow cytometry. The proliferative activity of the cells was estimated with tritiated thymidine incorporation. After 14 days of treatment with DHMEQ, mice with collagen-induced arthritis exhibited decreased severity of arthritis, based on the degree of paw swelling, the number of swollen joints, and radiographic and histopathologic scores, compared with the control mice treated with vehicle alone. In RA FLS stimulated with tumor necrosis factor-α, activities of NF-κB components p65 and p50 were inhibited by DHMEQ, leading to suppressed expression of the key inflammatory cytokine IL-6, CC chemokine ligand-2 and -5, matrix metalloproteinase-3, intercellular adhesion molecule-1, and vascular cell adhesion molecule-1. The proliferative activity of the cells was also suppressed. This is the first demonstration of an inhibitor of NF-κB nuclear translocation exhibiting a therapeutic effect on established murine arthritis, and suppression of inflammatory mediators in FLS was thought to be among the mechanisms underlying such an effect.
==== Body
Introduction
Rheumatoid arthritis (RA) is a chronic inflammatory disease that affects nearly 1% of the population worldwide and can lead to significantly impaired quality of life. Mortality rates are also significantly increased in patients with RA, and currently available therapies are often unable to change the course of the disease; therefore, further improvements in therapy are required. In this regard the recent application of biologic agents such as monoclonal antibodies to tumor necrosis factor (TNF)-α and IL-6 receptor, and recombinant soluble TNF-α receptor have been of great interest. Many cytokines, chemokines, adhesion molecules and matrix degrading enzymes have been demonstrated to play a role in synovial proliferation and joint destruction, which are the main pathologic features of RA. Notably, the efficacy of these biologic agents has indicated that intervention in a single cytokine pathway can achieve significant suppression of the complex inflammatory network and ameliorate disease. However, there are negative aspects to therapy with biologic agents, such as opportunistic infections, infusion reactions, high cost, and the fact that there are some patients in whom RA remains active regardless of the use of biologics. Therefore, further development of small molecular agents that specifically interrupt the critical intracellular pathways that are activated in RA synovium could prove beneficial.
The transcription factor nuclear factor-κB (NF-κB) forms a heterodimer or a homodimer of the subunit members, and in the cytoplasm of unstimulated cells it binds to natural inhibitors of NF-κB (IκB), which prevent it from entering the nucleus. The most common activated form of NF-κB in inflammatory cells consists of a p65 subunit and a p50 or p52 subunit [1-3]. In synovial tissue from patients with RA, p65 and p50 have been shown to be present in the nuclei of macrophage-like synoviocytes, fibroblast-like synoviocytes (FLS), and vascular endothelial cells, and probably play a pivotal role in the pathogenesis of RA [4-7]. The cytokines IL-1 and TNF-α activate and can be activated by NF-κB, and this positive regulatory loop amplifies the expression of other cytokines, chemokines, adhesion molecules, and enzymes in inflamed tissue [2]. Therefore, NF-κB should be considered a primary target for new types of anti-inflammatory treatments. Indeed, several recent studies have already shown significant effectiveness of this strategy. For example, in vivo experiments using murine arthritic models that employed intra-articular adenoviral gene transfer of dominant negative IκB kinase β [8] or super repressor IκBα [9], or alternatively intra-articular injection of NF-κB decoy oligonucleotides [9,10] demonstrated decreased severity of joint swelling. Moreover, ex vivo adenoviral gene transfer of IκBα into human synovial tissue inhibited the expression of inflammatory mediators [11]. Apart from gene transfer techniques, intravenous injection of a chimeric protein comprising the super-repressor IκBα fused to the membrane-transducing domain of the HIV Tat protein was shown to be effective in a rat model of acute pleuritis, although arthritis was not addressed in that study [12].
Only a limited number of studies testing the in vivo effects of small molecular weight compounds on arthritis have been reported [13]. These include a proteasome inhibitor PS-341 [14], and IκB kinase inhibitors BMS-345541 [15] and SPC 839 [16], which improved clinical and pathologic findings in murine arthritis. Another NF-κB inhibitor designated SP100030 was also shown to suppress collagen-induced arthritis (CIA) [17], but it appeared to be less efficient, possibly because it selectively affects T cells and not fibroblasts or endothelial cells. Recently, a peptide inhibitor of NF-κB that blocks association of NEMO (NF-κB essential modulator) with IκB kinases has been shown to ameliorate carrageenan-induced mouse paw inflammation, CIA, and RANKL (receptor activator of NF-κB ligand)-induced osteoclastogenesis [18,19].
Because RA is a chronic systemic disease, low molecular weight, cell-permeable agents that can block the NF-κB pathway with high specificity – if they become available – may have an advantage over gene transfer methods. In our search for such an inhibitor, we designed a compound named dehydroxymethylepoxyquinomicin (DHMEQ) using the parent structure of the antibiotic epoxyquinomicin C. We demonstrated that DHMEQ inhibits TNF-α-induced nuclear translocation of NF-κB, and does not inhibit phosphorylation and degradation of IκB, or a c-Jun N-terminal kinase (JNK) and a caspase-activating pathway in Jurkat T cells [20,21]. Here, we extended our study to test the therapeutic effect of DHMEQ on CIA, and to test the efficacy on the inhibition of the inflammatory pathway in human RA FLS. The results showed that this unique inhibitor of NF-κB nuclear translocation may hold promise for treatment of RA.
Materials and methods
Inhibitor of nuclear factor-κB
DHMEQ was synthesized as described previously [22]. It was dissolved in 100% dimethyl sulfoxide (DMSO) at 20 mg/ml and kept in aliquots at -30°C. Before use in cell culture, it was diluted with the medium described below to a final DMSO concentration of 0.05% or less, at which no effect of DMSO per se on NF-κB activity was observed.
Induction of collagen-induced arthritis
Animal experiments were approved by the Institutional Animal Care and Use Committee of Tokyo Medical and Dental University. Male 8-week-old DBA/1J mice were purchased from Oriental Yeast (Tokyo, Japan). Bovine collagen type II (Collagen Research Center, Tokyo, Japan) was dissolved in 50 mmol/l acetic acid at 4 mg/ml and emulsified in an equal volume of Freund's complete adjuvant (Difco Laboratories, Detroit, MI, USA). Mice were immunized intradermally with 100 μl of the emulsion at the base of the tail. After 21 days (day 0) the same amount of the antigen emulsified in the same adjuvant was intradermally injected at the base of the tail as a booster immunization. By day 5, 20 out of the 25 mice developed signs of arthritis and were randomly allocated to two groups of 10 mice each: an experimental group and a control group. From days 5 to 18, 100 μg DHMEQ (5 mg/kg body weight) dissolved in 50 μl of 100% DMSO was injected subcutaneously every day into the inguinal region of the mice in the experimental group. Mice in the control group received 50 μl DMSO, injected similarly.
Assessment of arthritis
The thickness of each hind paw was measured using a pair of digital slide calipers by an investigator who was blinded to the treatment groups. Out of 16 joints in each hind paw (i.e. the ankle, midfoot, first to fifth metatarsophalangeal [MTP] joints, interphalangeal joint of the first toe, and second to fifth proximal and distal interphalangeal joints), the swollen joints were identified using magnified pictures taken with a digital camera and counted. Radiographic assessment of bilateral second to fourth MTP joints was carried out using the following scoring systems: for soft tissue swelling 0 = not obvious, 1 = mild, 2 = marked; and for bone erosion 0 = not obvious, 1 = erosion < 0.3 mm in diameter, 2 = erosion > 0.3 mm in diameter. Use of this system yields a possible score between 0 and 12 per animal for each item. The left hind paw of each mouse was dissected, formalin-fixed, decalcified, embedded in paraffin, and stained with hematoxylin and eosin. Synovitis and bone destruction around the MTP joints of the sections were scored as follows: synovitis 0 = not obvious, 1 = synovitis < 0.2 mm at maximum thickness, 2 = synovitis > 0.2 mm at maximum thickness; bone destruction 0 = not obvious, 1 = obvious bone erosion, 2 = marked bone erosion associated with penetration of pannus into the marrow space. Radiographic and histopathologic assessments were performed by five investigators who were blinded to the assignment of mouse groups.
Cell cultures
RA FLS lines were established, as described previously [23], from the synovial tissues of RA patients obtained at surgery. RA patients fulfilled the American College of Rheumatology criteria. All procedures involving human tissues were approved by the Ethics Committee of Tokyo Medical and Dental University, and consent forms were obtained from the patients involved in the study. The cells were cultured in 100 mm dishes with Dulbecco's modified Eagle's medium (high glucose) containing 10% heat-inactivated fetal calf serum (FCS; Givco, Rockville, MD, USA) and antibiotics. Cells were passaged between four and eight times and were used when the cultures had reached about 80% cell layer confluence.
RT-PCR
RA FLS in 100 mm dishes were starved for 16 hours in medium without FCS, and then 10 μg/ml DHMEQ or vehicle was added. Twenty minutes later they were stimulated with 5 ng/ml TNF-α (PeproTech, London, UK) for 30 minutes, washed with phosphate-buffered saline (PBS), and detached from the dishes by treatment with trypsin-EDTA. Total RNA was isolated using RNeasy Mini Kit (Qiagen, Valencia, CA, USA) and treated with DNase I (Takara, Ohtsu, Japan), and RT-PCR was carried out using OneStep RT-PCR Kit (Qiagen) and the following primers: β-actin (5'-GTCCTCTCCCAAGTCCACACA, 3'-CTGGTCTCAAGTCAGTGTACAGGTAA), CC chemokine ligand (CCL)5 (formerly called RANTES; 5'-CGCTGTCATCCTCATTGCTA, 3'-GCTGTCTCGAACTCCTGACC), CCL2 (formerly called MCP-1; 5'-GCCTCCAGCATGAAAGTCTC, 3'-TAAAACAGGGTGTCTGGGGA), IL-1β (5'-TGCACGATGCACCTGTACGA, 3'-AGGCCCAAGGCCACAGGTAT), IL-6 (5'-GTTCCTGCAGAAAAAGGCAAAG, 3'-CTGAGGTGCCCATGCTACATTT), matrix metalloproteinase (MMP)-3 (5'-ATGGAGCTGCAAGGGGTGAG, 3'-CCCGTCACCTCCAATCCAAG), and vascular endothelial cell growth factor (VEGF) (5'-ATTGGAGCCTTGCCTTGCTG, 3'-CCAGGGTCTCGATTGGATGG). The cycling program was 94°C for 1 minute, 62°C for 1 minute, and 72°C for 1 minute for 25 or 30 cycles, followed by a final extension for 1 minute. The PCR products were electrophoresed on 1.5% agarose gel and stained with ethidium bromide. The relative intensities of the bands were quantified using image analysis software (NIH Image version 1.63; National Institute of Health, Bethesda, MD, USA).
ELISA
RA-FLS were cultured and starved similarly as described above except 24-well culture plates were used. Twenty minutes after addition of DHMEQ (10 μg/ml) or vehicle, TNF-α (5 ng/ml) was added and 24 hours later the supernatant was harvested and kept at -20°C until use. ELISA kits for CCL2, CCL5, IL-1β and IL-6 were purchased from BioSource (Camacillo, CA, USA), and that for MMP-3 was from MBL (Nagoya, Japan).
Flow cytometry
RA-FLS were cultured and starved as described above except 60 mm culture dishes were used. Twenty minutes after addition of DHMEQ (10 μg/ml) or vehicle, TNF-α (5 ng/ml) was added and 14 hours later the cells were washed with PBS, detached by trypsin-EDTA, and suspended in PBS. The prepared cells were incubated with monoclonal antibody to intercellular adhesion molecule (ICAM)-1 (BD Sciences, San Jose, CA, USA), vascular cell adhesion molecule (VCAM)-1 (BD Sciences), or isotype-matched mouse IgG for 20 minutes, followed by a detection with phycoerythrin-conjugated goat anti-mouse IgG (Southern Biotechnology Associates, Birmingham, AL, USA), and analyzed with an Epics XL flow cytometer (Beckman Coulter, Miami, FL, USA).
For detection of apoptotic cells, cells were prepared as above, DHMEQ (0, 5, or 10 μg/ml) or 1 μmol/l staurosporin (Wako, Osaka, Japan) was added, the cells were stimulated with TNF-α (5 ng/ml) 20 minutes later, further incubated for 14 hours, and stained with Cy3-labeled annexin V (MBL).
Proliferation assay
RA FLS were cultured at a density of 104/well in 96-well culture plates for 18 hours and then starved for 24 hours in Dulbecco's modified Eagle's medium with 2 μmol/l 2-mercaptoethanol, without FCS. After serum starvation, 0–10 μg/ml DHMEQ was added followed 20 minutes later by addition of 5 ng/ml TNF-α and 5 μCi/ml [3H]thymidine. The cells were further cultured for 48 hours and incorporation of 3H during the last 24 hours was measured in a scintillation counter.
Electrophoretic mobility shift assay
RA FLS were cultured, starved, and stimulated as described above for RT-PCR, and nuclear extracts were prepared using NucBaster Protein Extraction Kit (Novagen, Darmstadt, Germany). The protein concentrations of the extracts were estimated by BCA Protein Assay Kit (Pierce, Rockford, IL, USA), and the extracts were kept at -80°C until use. 32P-labeled oligonucleotide containing the NF-κB binding sequence (5'-AGTTGAGGGGACTTTCCCAGGC-3') was used as a probe. Ten micrograms of the nuclear extract was incubated with 2 μg poly(dI-dC) for 30 minutes at room temperature in 20 μl reaction buffer containing 20 mmol/l HEPES, 20% glycerol, 100 mmol/l KCl, and 0.2 mmol/l EDTA, at pH 7.9. Following incubation, 32P-labeled probe was added to the mixture with or without 100-fold excess unlabeled oligonucleotide as a competitor and incubated for a further 30 minutes at room temperature. The protein-DNA complexes were separated from the free probe by 4% PAGE. For supershift assays, 1 μg of antibody to p50, p52 (Santa Cruz Biotech, Santa Cruz, CA, USA), or p65 (Chemicon, Temecula, CA, USA) was added to the sample and incubated for 30 minutes at 4°C before electrophoresis.
Statistical analysis
Results were compared using two-sided, unpaired Student's t-tests.
Results
Therapeutic effect of DHMEQ on collagen-induced arthritis
The in vivo anti-inflammatory effect of DHMEQ was first demonstrated in a type of CIA model described by Matsumoto and coworkers [21], in which they showed a prophylactic effect of this compound when administered at 2–4 mg/kg, three times a week, from the day of booster immunization, although the CIA protocol was different from that in the present study and they did not use adjuvant. To test the therapeutic effect on a standard CIA model, we included only those mice that had apparently begun to develop arthritis by day 5 after the booster immunization with collagen and complete adjuvant, and started therapy with DHMEQ at 100 μg (5 mg/kg) daily.
After 14 days of therapy (from days 5 to 18), the thickness of the hind paws in the DHMEQ treated group (mean ± SD; 5.97 ± 0.66 mm) was significantly lower than that in the control group (6.95 ± 0.88 mm) that received vehicle alone (Fig. 1a). The number of swollen joints was also significantly lower in the DHMEQ-treated group (9.20 ± 4.64 versus 14.20 ± 3.68; Fig. 1b). During the first week of treatment, these young animals exhibited growth retardation, as estimated from their body weights, likely resulting from severe inflammation. However, this slow-down was significantly less in the DHMEQ group (weight gain from days 5 to 12; 1.29 ± 1.25 g) than in the control group (0.14 ± 1.01 g; Fig. 1c), suggesting that DHMEQ alleviated inflammation and was tolerable at the dose tested. Radiographs showed various degrees of soft tissue swelling and destructive changes in bone (Fig. 2a, b). The scores of these findings were both significantly lower in the group treated with DHMEQ (Fig. 2c, d). At the histologic level, various degrees of synovitis and bone destruction were observed (Fig. 3a, b). In severe cases, marked infiltration of mononuclear cells were present within the synovium, and pannus was frequently observed to penetrate into the bone marrow space. Again, DHMEQ reduced these findings significantly (Fig. 3c, d).
Inhibition of nuclear factor-κB in rheumatoid arthritis fibroblast-like synoviocytes by DHMEQ
To examine the mechanisms underlying the antiarthritic effect of DHMEQ, as well as its effect on human cells, RA FLS lines were established from several patients with RA and used for the present experiments. We should like to note that it was previously confirmed that these cells expressed neither CD14 nor HLA class II [23], which means that they did not contain either macrophages or dendritic cells.
The effect of DHMEQ on NF-κB activation in RA FLS was examined using electrophoretic mobility shift assay (Fig. 4). Unstimulated RA FLS in serum-free medium exhibited only a faint band corresponding to NF-κB, but stimulation with 5 ng/ml TNF-α increased the intensity of the band dramatically. In supershift assays, anti-p50 antibody virtually abrogated the band of NF-κB, and anti-p65 antibody also remarkably diminished the intensity of the band. Anti-p52 antibody was much less effective, suggesting that the major components of the activated NF-κB in TNF-α-stimulated RA FLS were p65 and p50. An excess amount of unlabeled NF-κB probe abolished the band, confirming the specificity of this assay. When DHMEQ was added 20 minutes before the stimulation with TNF-α, the band representing NF-κB was abrogated almost completely at 10 μg/ml but not significantly at 1 μg/ml. Based on these findings, the following experiments were carried out using 10 μg/ml DHMEQ unless otherwise indicated.
Suppression of inflammatory mediators by DHMEQ
Among the many molecules that are involved in inflammatory responses, a set of representative chemokines, ILs, MMP-3, and VEGF were selected to test the effect of DHMEQ. IL-6 is known to be an NF-κB-dependent cytokine [24,25] and is one of the key cytokines in RA pathogenesis, as evidenced by the fact that an anti-IL-6 receptor monoclonal antibody has been shown to reduce significantly RA disease activity in clinical trials [26]. We previously showed that chemokines CCL2 and CCL5 play a role not only in inflammatory cell migration but also in activation of RA FLS in an autocrine or paracrine manner [23]. Other investigators have shown that an antagonist to CCL2 suppressed arthritis in a murine model [27]. MMP-3 is among the cartilage-degrading enzymes and is known to be regulated by the NF-κB pathway in RA synovium [28]. VEGF is one of the angiogenic factors that are involved in the neovascularization in RA joints [29].
The effect of DHMEQ on mRNA expression of these key molecules was first examined by RT-PCR. As shown in Fig. 5, there was some heterogeneity in the mRNA expression levels depending on the cell line used. However, mRNA levels of CCL2, IL-6, and MMP-3 tended to be consistently increased by TNF-α stimulation and suppressed by DHMEQ. CCL5 mRNA was not detected in one of the cell lines (#5), but in other cell lines it was enhanced by TNF-α and suppressed by DHMEQ. IL-1β mRNA was barely detectable in some of the cell lines tested, at least under these assay conditions. However, after TNF-α stimulation IL-1β mRNA was clearly expressed in cell line #2 and faintly in lines #1 and #5; in all cases the expression was diminished by treatment with DHMEQ. In contrast, the level of VEGF mRNA was neither significantly increased by TNF-α nor suppressed by DHMEQ. We applied a constant amount of RNA to each tube, and observed virtually constant intensity of β-actin mRNA, irrespective of treatment with DHMEQ; this suggested that this compound did not affect the housekeeping activity of the cells.
The suppressive effect of DHMEQ on the TNF-α-induced expression of CCL2, CCL5, IL-1β, IL-6 and MMP-3 was further tested at the protein level by ELISA (Fig. 6). CCL2, CCL5 and IL-6 were barely detected in the serum-free culture supernatant of the cells unless the cells were stimulated, but were produced abundantly after TNF-α stimulation. This production was significantly suppressed by DHMEQ. Similarly, production of MMP-3 tended to be suppressed by DHMEQ, although this was not statistically significant. Because the level of IL-1β was lower than the detection limit (10 pg/ml), even after stimulation with TNF-α, we could not confirm the effect of DHMEQ on IL-1β expression by ELISA.
Suppression of adhesion molecule expression by DHMEQ
Adhesion molecules ICAM-1 (CD54) and VCAM-1 (CD106) are expressed at higher levels in the synovial tissue of RA than in osteoarthritis [30,31] and are implicated in the interaction between leukocytes and RA FLS that contributes to the synovitis. Flow cytometric analysis showed that 2.7 ± 0.5% (mean ± standard deviation [SD] of four independent experiments) of RA FLS expressed ICAM-1 in serum-free medium, and the ratio of cells expressing ICAM-1 markedly increased up to 44.5 ± 11.7% after stimulation with TNF-α (Fig. 7). In the presence of DHMEQ, however, this ratio significantly decreased to 5.3 ± 3.4%. On the other hand, VCAM-1 was expressed on only 0.9 ± 1.0% (mean ± SD of three independent experiments) of the unstimulated cells, but 29.0 ± 11.2% of the cells expressed VCAM-1 after TNF-α stimulation (Fig. 8); this ratio significantly decreased to 3.1 ± 2.5% by the effect of DHMEQ.
Suppression of proliferative activity of rheumatoid arthritis fibroblast-like synoviocytes by DHMEQ
One of the prominent characteristics of RA FLS is their proliferative activity, which leads to pannus formation as well as swelling of the joints, and pathways controlling the proliferation of RA FLS include an NF-κB-dependent pathway [32]. Representative data shown in Fig. 9 indicate that RA FLS incorporate a certain amount of [3H]thymidine even without stimulation (mean ± SD; 207 ± 36 counts/minute), which was significantly higher than the background level (29 ± 1.2 counts/minute; P < 0.01). This moderate activity increased to 582 ± 53 counts/minute with stimulation with TNF-α, and DHMEQ significantly suppressed this proliferative activity in a dose-dependent manner. At 5.0 μg/ml or higher concentration of DHMEQ, proliferative activity of the cells was lower than that of unstimulated cells, suggesting that DHMEQ suppressed spontaneous proliferation as well as TNF-α-induced proliferation.
Cytotoxic effect of DHMEQ on rheumatoid arthritis fibroblast-like synoviocytes
To test whether DHMEQ exhibits cytotoxicity at the concentration that suppressed inflammatory mediators, serum-starved RA FLS were further incubated for 14 hours after stimulation with TNF-α in serum-free medium with DHMEQ or the apoptosis inducer staurosporin. Thereafter, expression of annexin V binding phospholipid on the cell surface – an indicator of early phase of apoptosis – was measured using Cy3-labeled annexin V. With staurosporin, 10.1% of the cells were annexin V positive and the mean fluorescence intensity was 1.1 (Fig, 10). In contrast, in the presence of 5 and 10 μg/ml DHMEQ, the ratio and mean fluorescence intensity of annexin V positive cells remained less than 1% and 0.5%, respectively. Trypan blue dye exclusion test also showed virtually 100% viability of the cells incubated with DHMEQ (not shown).
Discussion
Many stimuli are known to activate NF-κB, including TNF-α, IL-1β, anti-CD3 antibody (in T cells), oxidative stresses, viral products and lipopolysaccharides, and these act by means of protein kinases that phosphorylate IκB, leading to degradation of IκB by the proteasome and passage of NF-κB into the nucleus [2]. NF-κB regulates the expression of many genes that are involved in inflammatory responses, including TNF-α, IL-1β, IL-6, CCL2, CCL5, MMP-3, ICAM-1, VCAM-1, inducible nitric oxide synthase, and cyclo-oxygenase-2, all of which are known to participate in the pathogenesis of RA. Products of these genes coordinately enhance inflammatory reactions resulting in further activation of NF-κB. In fact, NF-κB components p50 and p65 were demonstrated to be activated in both macrophage-like and fibroblast-like synoviocytes as well as vascular endothelial cells in RA-derived synovial tissue but not in normal synovium [4-7]. In RA synovium p50 and p65 expression increases, especially at sites adjacent to the cartilage-pannus junction, and is thought to be implicated in cartilage destruction [33]. It is clear, therefore, that NF-κB is an important target molecule for RA therapy. Aspirin, sodium salicylate, corticosteroids, sulfasalazine, and gold salts were demonstrated, at least in part, to exhibit their activity by way of NF-κB suppression [34-37], and novel agents that are more specific to the NF-κB pathway than these classical agents – and are less costly than the recently marketed biologics – would be of great value. Indeed, inhibition of the NF-κB pathway by gene therapy [8-11] or by small molecular weight compounds [13-18] has recently been tested in experimental models of arthritis, demonstrating the efficacy of this strategy.
We previously showed that DHMEQ does not inhibit phosphorylation and degradation of IκB, but inhibits nuclear transport of p65 in TNF-α-stimulated COS-1 cells transfected with the DNA that encodes p65 combined with green fluorescent protein [20]. It does not affect TNF-α-induced activation of JNK, or nuclear transport of Smad2 or the large T antigen. This molecule should therefore be considered a unique inhibitor of NF-κB that acts at the level of nuclear translocation. This specificity is an advantage of DHMEQ over the inhibitors of the upstream molecules of the NF-κB pathway, because kinase inhibitors or proteasome inhibitors may suffer from disadvantages relating to specificity and undesired effects that are unrelated to the NF-κB pathway. Another advantage of DHMEQ over gene transfer methods is its simplicity of administration. In our model of mouse arthritis, subcutaneous daily injections of 5 mg/kg DHMEQ resulted in a significant therapeutic effect on arthritis. Although the efficacy of other administration protocols and the pharmacokinetics of DHMEQ remain to be studied in detail, small, cell-permeable compounds appear to have fewer obstacles to be overcome in comparison with gene transfer strategies because RA is a chronic systemic inflammatory disorder.
As a first step toward application in human cells, we tested the effect of DHMEQ on the function of RA synovial cells in culture. Thus far, studies that showed effects of NF-κB inhibitors using human synovial cells have been limited. However, suppression of expression of the key inflammatory cytokine IL-6, chemokines CCL2 and CCL5, matrix-degrading enzyme MMP-3, and adhesion molecules ICAM-1 and VCAM-1, as well as proliferative activity of the cells, suggested that DHMEQ may be efficacious in the treatment of RA synovitis. In the electrophoretic mobility shift assay, 10 μg/ml DHMEQ nearly completely inhibited NF-κB activity of the TNF-α-stimulated RA FLS, but its effect was not significant at 1 μg/ml (Fig. 4). Nevertheless, in the proliferation assay 1.3 μg/ml DHMEQ significantly suppressed TNF-α-stimulated thymidine uptake by RA FLS (Fig. 9). Variation is to a certain extent inevitable in experiments using RA FLS, but this discrepancy was reproducible. It seems possible, in assays that require longer incubation of cells, that secondary effects of NF-κB inhibition resulting from suppressed expression of cytokines and other regulatory molecules of cellular activity may merge with the direct effect of DHMEQ.
VEGF is thought to promote angiogenesis and enhance vascular permeability in inflamed tissue, but we did not observe suppression of VEGF mRNA expression by DHMEQ. A recent investigation showed suppression of IL-6-induced VEGF production by fibroblasts using a JNK inhibitor, which suggests that expression of VEGF is predominantly regulated not by the NF-κB pathway but by the activator protein-1 pathway [38]. In this regard, it was reported that cyclosporin A, which is a widely used immunosuppressant and is effective to some degree in RA, suppressed expression of VEGF by RA-FLS by way of suppressing activator protein-1 binding activity [29].
The most important goal in RA therapy is the prevention of bone destruction in order to maintain normal function of the joints. It was recently demonstrated that DHMEQ suppresses osteoclastogenesis in a culture system of mouse bone marrow derived macrophage precursor cells stimulated with RANKL (receptor activator of NF-κB ligand) and macrophage colony-stimulating factor, and suppresses the bone-resorbing activity of mature osteoclasts [39]. Therefore, it is of interest to study further the in vivo effect of this inhibitor on bone-resorbing activity in severe arthritis.
Along with efficacy, safety issues should be addressed. In the present study we observed no abnormality in the behavior of the mice treated with DHMEQ; rather, they exhibited less weight loss than did the control mice during the active period of inflammation. In in vitro experiments, annexin V staining and trypan blue staining confirmed that cell viability did not decrease. The mRNA expression of VEGF and β-actin was not affected by DHMEQ. However, NF-κB is known to play a role in preventing cell apoptosis [40]. Massive hepatocyte apoptosis in p65-deficient mice is an extreme example of the antiapoptotic role of NF-κB [41]. We also observed significant apoptosis in some tumor cells transplanted into nude mice treated with DHMEQ (8 mg/kg per day, by intraperitoneal injection), but the mice did not exhibit adverse effects [42]. The relationship between the anti-inflammatory effect of DHMEQ and apoptosis of cells in inflamed as well as normal tissue remains to be further examined.
Conclusion
We showed herein that an NF-κB nuclear translocation inhibitor DHMEQ had a therapeutic effect on CIA in mice, and suppressed the expression of inflammatory molecules and the proliferative activity of TNF-α-stimulated RA FLS. Although its effect on other human cell types, especially T cells, vascular endothelial cells and osteoclasts, are currently under investigation, these findings suggest that FLS are among the important targets on which DHMEQ exerts its antiarthritic effect. This agent may be a promising candidate for further clinical development.
Abbreviations
CCL = CC chemokine ligand; CIA = collagen-induced arthritis; DHMEQ = dehydroxymethylepoxyquinomicin; DMSO = dimethyl sulfoxide; ELISA = enzyme-linked immunosorbent assay; FCS = fetal calf serum; FLS = fibroblast-like synoviocyte; ICAM = intercellular adhesion molecule; IκB = inhibitor of NF-κB; IL = interleukin; JNK = c-Jun N-terminal kinase; MMP = matrix metalloproteinase; MTP = metatarsophalangeal; NF-κB = nuclear factor-κB; PBS = phosphate-buffered saline; RA = rheumatoid arthritis; RT-PCR = reverse transcriptase-polymerase chain reaction; SD = standard deviation; TNF = tumor necrosis factor; VCAM = vascular cell adhesion molecule; VEGF = vascular endothelial cell growth factor.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
KW carried out in vitro experiments. TN and TK designed the study, carried out in vivo experiments, and drafted the manuscript. NM collected the clinical materials and revised the manuscript. KU synthesized a critical chemical. All authors read and approved the final manuscript.
Acknowledgements
We thank Fumiko Inoue for her expert technical assistance. This study was supported in part by the Ministry of Health, Labour and Welfare of Japan.
Figures and Tables
Figure 1 Clinical effect of NF-κB inhibitor DHMEQ on collagen-induced arthritis in DBA/1J mice. After onset of arthritis, animals were treated with 100 μg/day dehydroxymethylepoxyquinomicin (DHMEQ; ■; n = 10) or vehicle (□; n = 10). (a) Sum of the thickness of the right and the left hind paws of each mouse after 2 weeks of treatment. Each paw was measured twice and the average was plotted. (b) Sum of the number of swollen joints (described in the Materials and methods section) in the right and the left hind paws of each mouse after 2 weeks of treatment. Each paw was counted twice and the average was plotted. Maximum possible number is 32 per mouse. (c) Change in body weight of each mouse during the first week of treatment. Horizontal bars represent the mean. NF-kB, nuclear factor-κB.
Figure 2 Effect of DHMEQ on radiographic findings in collagen-induced arthritis in mice. (a) A representative radiograph of the left metatarsophalangeal (MTP) joints of a mouse treated with dehydroxymethylepoxyquinomicin (DHMEQ), which shows small bone erosions, and (b) that of a control mouse, which shows remarkable soft tissue swelling and large bone erosions. (c) Soft tissue swelling and (d) bone erosions of bilateral second, third and fourth MTP joints observed in the radiographs were scored as described in the Materials and methods section. Values are expressed as the mean ± standard deviation of the total scores of 10 mice in each group, determined by five independent observers.
Figure 3 Effect of DHMEQ on histopathologic findings in collagen-induced arthritis in mice. (a) A representative specimen of the metatarsophalangeal (MTP) joint of a dehydroxymethylepoxyquinomicin (DHMEQ)-treated mouse, showing almost normal findings, and (b) that of a control mouse showing remarkable cell infiltration in the synovium and bone destruction accompanied by pannus invasion into the marrow space. The severity of (c) synovitis and (d) bone destruction in the specimens were scored as described in the Materials and methods section. Values are expressed as the mean ± standard deviation of the total scores of 10 mice in each group, determined by five independent observers.
Figure 4 Inhibition by DHMEQ of NF-κB in rheumatoid arthritis fibroblast-like synoviocytes. Nuclear extracts were obtained from unstimulated and tumor necrosis factor (TNF)-α-stimulated rheumatoid arthritis (RA) fibroblast-like synoviocytes (FLS) and nuclear factor-κB (NF-κB) DNA-binding activity was examined by electrophoretic mobility shift assays. For supershift assays, the DNA-protein mixture was incubated with antibodies to p65, p50, or p52 before electrophoresis. To confirm the specificity of the assay, 100-fold excess of unlabeled NF-κB probe was included as a competitor. To assess whether dehydroxymethylepoxyquinomicin (DHMEQ) inhibits NF-κB activation in RA FLS, the cells were incubated with DHMEQ for 20 minutes before the stimulation with TNF-α. Data shown are representative of three independent experiments.
Figure 5 Effect of DHMEQ on inflammatory mediator mRNA expression by RA FLS stimulated with TNF-α. (a) Five rheumatoid arthritis (RA) fibroblast-like synoviocyte (FLS) cell lines (#1–#5) obtained from different patients were stimulated with tumor necrosis factor (TNF)-α in the presence or absence of 10 μg/ml dehydroxymethylepoxyquinomicin (DHMEQ) and mRNA expression of CC chemokine ligand (CCL)2, CCL5, IL-6, IL-1β, matrix metalloproteinase (MMP)-3, and vascular endothelial cell growth factor (VEGF) was examined by RT-PCR. (b) Densitometric analysis of these results. Intensity of each band was normalized relative to that of β-actin in the same lane, and the mean ± standard deviation of the five cell lines are shown. *P < 0.05 versus T. D, DHMEQ; T, TNF-α.
Figure 6 Suppressive effect of DHMEQ on inflammatory mediator production by RA-FLS at the protein level. Rheumatoid arthritis (RA) fibroblast-like synoviocytes (FLS) were stimulated with tumour necrosis factor (TNF)-α in the presence or absence of 10 μg/ml dehydroxymethylepoxyquinomicin (DHMEQ), and levels of secreted CC chemokine ligand (CCL)2, CCL5, IL-6, and matrix metalloproteinase (MMP)-3 in the culture supernatants were measured using ELISA. Values are expressed as the mean ± standard deviation of three independent experiments.
Figure 7 Suppression of ICAM-1 expression by DHMEQ. Shown, using flow cytometry, is suppression of intercellular adhesion molecule (ICAM)-1 expressed on tumor necrosis factor (TNF)-α-stimulated rheumatoid arthritis (RA) fibroblast-like synoviocytes (FLS) by dehydroxymethylepoxyquinomicin (DHMEQ). Cells were preincubated for 20 minutes with 10 μg/ml DHMEQ or vehicle. TNF-α stimulated or unstimulated RA FLS were incubated with isotype-matched control IgG or anti-ICAM-1 antibody, followed by phycoerythrin-labeled second antibody. (a) Representative data are shown, along with (b) the means ± standard deviation of ICAM-1-positive cells in four independent experiments.
Figure 8 Suppression of VCAM-1 expression by DHMEQ. Suppression of vascular cell adhesion molecule (VCAM)-1 expressed on tumor necrosis factor (TNF)-α-stimulated rheumatoid arthritis (RA) fibroblast-like synoviocytes (FLS) by dehydroxymethylepoxyquinomicin (DHMEQ). Flow cytometric analysis was carried out (as in Fig. 7) except that anti-VCAM-1 antibody was used. (a) Representative data are shown, along with (b) the means ± standard deviation of VCAM-1-positive cells in three independent experiments.
Figure 9 Suppression of proliferative activity of RA FLS by DHMEQ. Rheumatoid arthritis (RA) fibroblast-like synoviocytes (FLS) were stimulated with tumor necrosis factor (TNF)-α or unstimulated in the presence or absence of dehydroxymethylepoxyquinomicin (DHMEQ), cultured for 48 hours, and incorporation of [3H]thymidine during the last 24 hours was measured. Values are expressed as mean ± standard deviation of triplicate measurements. Data shown are representative of three independent experiments. *P < 0.01, **P < 0.001. cpm, counts/minute.
Figure 10 Cytotoxicity of DHMEQ. Significant cytotoxicity was not observed in rheumatoid arthritis (RA) fibroblast-like synoviocytes (FLS) treated with dehydroxymethylepoxyquinomicin (DHMEQ). Cells were stimulated with 5 ng/ml tumor necrosis factor (TNF)-α and incubated in serum-free medium for 14 hours with 0–10 μg/ml DHMEQ or with the apoptosis inducer staurosporin (1 μmol/l), and Cy3-labeled annexin V binding cells were measured by flow cytometry. Data shown are representative of three independent experiments. MFI, mean fluorescence intensity.
==== Refs
Roshak AK Jackson JR McGough K Chabot-Fletcher M Mochan E Marshall LA Manipulation of distinct NFκB proteins alters interleukin-1β-induced human rheumatoid synovial fibroblast prostaglandin E2 formation J Biol Chem 1996 271 31496 31501 8940164 10.1074/jbc.271.49.31496
Barnes PJ Karin M Nuclear factor-κB: a pivotal transcription factor in chronic inflammatory diseases N Engl J Med 1997 336 1066 1071 9091804 10.1056/NEJM199704103361506
Firestein GS NF-κB: Holy Grail for rheumatoid arthritis? Arthritis Rheum 2004 50 2381 2386 15334448 10.1002/art.20468
Handel ML McMorrow LB Gravallese EM Nuclear factor-κB in rheumatoid synovium. Localization of p50 and p65 Arthritis Rheum 1995 38 1762 1770 8849348
Marok R Winyard PG Coumbe A Kus ML Gaffney K Blades S Mapp PI Morris CJ Blake DR Kaltschmidt C Baeuerle PA Activation of the transcription factor nuclear factor-κB in human inflamed synovial tissue Arthritis Rheum 1996 39 583 591 8630106
Fujisawa K Aono H Hasunuma T Yamamoto K Mita S Nishioka K Activation of transcription factor NF-κB in human synovial cells in response to tumor necrosis factor α Arthritis Rheum 1996 39 197 203 8849369
Han Z Boyle DL Manning AM Firestein GS AP-1 and NF-κB regulation in rheumatoid arthritis and murine collagen-induced arthritis Autoimmunity 1998 28 197 208 9892501
Tak PP Gerlag DM Aupperle KR van de Geest DA Overbeek M Bennett BL Boyle DL Manning AM Firestein GS Inhibitor of nuclear factor κB kinase β is a key regulator of synovial inflammation Arthritis Rheum 2001 44 1897 1907 11508443 10.1002/1529-0131(200108)44:8<1897::AID-ART328>3.0.CO;2-4
Miagkov AV Kovalenko DV Brown CE Didsbury JR Cogswell JP Stimpson SA Baldwin AS Makarov SS NF-κB activation provides the potential link between inflammation and hyperplasia in the arthritic joint Proc Natl Acad Sci USA 1998 95 13859 13864 9811891 10.1073/pnas.95.23.13859
Tomita T Takeuchi E Tomita N Morishita R Kaneko M Yamamoto K Nakase T Seki H Kato K Kaneda Y Ochi T Suppressed severity of collagen-induced arthritis by in vivo transfection of nuclear factor κB decoy oligodeoxynucleotides as a gene therapy Arthritis Rheum 1999 42 2532 2542 10615998 10.1002/1529-0131(199912)42:12<2532::AID-ANR5>3.0.CO;2-2
Bondeson J Foxwell B Brennan F Feldmann M Defining therapeutic targets by using adenovirus: blocking NF-κB inhibits both inflammatory and destructive mechanisms in rheumatoid synovium but spares anti-inflammatory mediators Proc Natl Acad Sci USA 1999 96 5668 5673 10318942 10.1073/pnas.96.10.5668
Blackwell NM Sembi P Newson JS Lawrence T Gilroy DW Kabouridis PS Reduced infiltration and increased apoptosis on leukocytes at sites of inflammation by systemic administration of a membrane-permeable IκBα repressor Arthritis Rheum 2004 50 2675 2684 15334484 10.1002/art.20467
Roshak AK Callahan JF Blake SM Small-molecule inhibitors of NF-κB for the treatment of inflammatory joint disease Curr Opin Pharmacol 2002 2 316 321 12020477 10.1016/S1471-4892(02)00165-0
Palombella VJ Conner EM Fuseler JW Destree A Davis JM Laroux FS Wolf RE Huang J Brand S Elliott PJ Role of the proteasome and NF-κB in streptococcal cell wall-induced polyarthritis Proc Natl Acad Sci USA 1998 95 15671 15676 9861028 10.1073/pnas.95.26.15671
McIntyre KW Shuster DJ Gillooly KM Dambach DM Pattoli MA Lu P Zhou X Qiu Y Zusi FC Burke JR A highly selective inhibitor of IκB kinase, BMS-345541, blocks both joint inflammation and destruction in collagen-induced arthritis in mice Arthritis Rheum 2003 48 2652 2659 13130486 10.1002/art.11131
Hammaker D Sweeney S Firestein GS Signal transduction networks in rheumatoid arthritis Ann Rheum Dis 2003 62 Suppl 2 ii86 ii89 14532158
Gerlag DM Ransone L Tak PP Han Z Palanki M Barbosa MS Boyle D Manning AM Firestein GS The effect of a T cell-specific NF-κB inhibitor on in vitro cytokine production and collagen-induced arthritis J Immunol 2000 165 1652 1658 10903776
Jimi E Aoki K Saito H D'Acquisto F May MJ Nakamura I Sudo T Kojima T Okamoto F Fukushima H Selective inhibition of NF-κB blocks osteoclastogenesis and prevents inflammatory bone destruction in vivo Nat Med 2004 10 617 624 15156202 10.1038/nm1054
di Meglio P Ianaro A Ghosh S Amelioration of acute inflammation by systemic administration of a cell-permeable peptide inhibitor of NF-κB activation Arthritis Rheum 2005 52 951 958 15751079 10.1002/art.20960
Ariga A Namekawa J Matsumoto N Inoue J Umezawa K Inhibition of tumor necrosis factor-α-induced nuclear translocation and activation of NF-κB by dehydroxymethylepoxyquinomicin J Biol Chem 2002 277 24625 24630 11983688 10.1074/jbc.M112063200
Matsumoto N Ariga A To-e S Nakamura H Agata N Hirano S Inoue J Umezawa K Synthesis of NF-κB activation inhibitors derived from epoxyquinomicin C Bioorg Med Chem Lett 2000 10 865 869 10853648 10.1016/S0960-894X(00)00114-1
Suzuki Y Sugiyama C Ohno O Umezawa K Preparation and biological activities of optically active dehydroxymethylepoxyquinomicin, a novel NF-κB inhibitor Tetrahedron 2004 60 7061 7066 10.1016/j.tet.2004.01.103
Nanki T Nagasaka K Hayashida K Saita Y Miyasaka N Chemokines regulate IL-6 and IL-8 production by fibroblast-like synoviocytes from patients with rheumatoid arthritis J Immunol 2001 167 5381 5385 11673556
Miyazawa K Mori A Yamamoto K Okudaira H Transcriptional roles of CCAAT/enhancer binding protein-β, nuclear factor-κB, and C-promoter binding factor 1 in interleukin (IL)-1β-induced IL-6 synthesis by human rheumatoid fibroblast-like synoviocytes J Biol Chem 1998 273 7620 7627 9516466 10.1074/jbc.273.13.7620
Georganas C Liu H Perlman H Hoffmann A Thimmapaya B Pope RM Regulation of IL-6 and IL-8 expression in rheumatoid arthritis synovial fibroblasts: the dominant role for NF-κB but not C/EBPβ or c-Jun J Immunol 2000 165 7199 7206 11120852
Nishimoto N Yoshizaki K Miyasaka N Yamamoto K Kawai S Takeuchi T Hashimoto J Azuma J Kishimoto T Treatment of rheumatoid arthritis with humanized anti-interleukin-6 receptor antibody: a multicenter, double-blind, placebo-controlled trial Arthritis Rheum 2004 50 1761 1769 15188351 10.1002/art.20303
Gong JH Ratkay LG Waterfield JD Clark-Lewis I An antagonist of monocyte chemoattractant protein 1 (MCP-1) inhibits arthritis in the MRL-lpr mouse model J Exp Med 1997 186 131 137 9207007 10.1084/jem.186.1.131
Andreakos E Smith C Kiriakidis S Monaco C de Martin R Brennan FM Paleolog E Feldmann M Foxwell BM Heterogeneous requirement of IκB kinase 2 for inflammatory cytokine and matrix metalloproteinase production in rheumatoid arthritis: implications for therapy Arthritis Rheum 2003 48 1901 1912 12847684 10.1002/art.11044
Cho ML Cho CS Min SY Kim SH Lee SS Kim WU Min DJ Min JK Youn J Hwang SY Cyclosporine inhibition of vascular endothelial growth factor production in rheumatoid synovial fibroblasts Arthritis Rheum 2002 46 1202 1209 12115224 10.1002/art.10215
Furuzawa-Carballeda J Alcocer-Varela J Interleukin-8, interleukin-10, intercellular adhesion molecule-1 and vascular cell adhesion molecule-1 expression levels are higher in synovial tissue from patients with rheumatoid arthritis than in osteoarthritis Scand J Immunol 1999 50 215 222 10447928 10.1046/j.1365-3083.1999.00573.x
Li P Sanz I O'Keefe RJ Schwarz EM NF-κB regulates VCAM-1 expression on fibroblast-like synoviocytes J Immunol 2000 164 5990 5997 10820282
Firestein GS Invasive fibroblast-like synoviocytes in rheumatoid arthritis. Passive responders or transformed aggressors? Arthritis Rheum 1996 39 1781 1790 8912499
Benito MJ Murphy E Murphy EP van den Berg WB FitzGerald O Bresnihan B Increased synovial tissue NF-κB1 expression at sites adjacent to the cartilage-pannus junction in rheumatoid arthritis Arthritis Rheum 2004 50 1781 1787 15188354 10.1002/art.20260
Kopp E Ghosh S Inhibition of NF-κB by sodium salicylate and aspirin Science 1994 265 956 959 8052854
Barnes PJ Adcock I Anti-inflammatory actions of steroids: molecular mechanisms Trends Pharmacol Sci 1993 14 436 441 7510080 10.1016/0165-6147(93)90184-L
Wahl C Liptay S Adler G Schmid RM Sulfasalazine: a potent and specific inhibitor of nuclear factor kappa B J Clin Invest 1998 101 1163 1174 9486988
Yang JP Merin JP Nakano T Kato T Kitade Y Okamoto T Inhibition of the DNA-binding activity of NF-κB by gold compounds in vitro FEBS Lett 1995 361 89 96 7890047 10.1016/0014-5793(95)00157-5
Naruishi K Nishimura F Yamada-Naruishi H Omori K Yamaguchi M Takashiba S C-Jun N-terminal kinase (JNK) inhibitor, SP600125, blocks interleukin (IL)-6-induced vascular endothelial growth factor (VEGF) production: cyclosporine A partially mimics this inhibitory effect Transplantation 2003 76 1380 1382 14627919 10.1097/01.TP.0000085661.52980.95
Takatsuna H Asagiri M Kubota T Oka K Osada T Sugiyama C Saito H Aoki K Ohya K Takayanagi H Umezawa K Inhibition of RANKL-induced osteoclastogenesis by (-)-DHMEQ, a novel NF-kappaB inhibitor, through downregulation of NFATc1 J Bone Miner Res 2005 20 653 662 15765185
Chen F Castranova V Shi X New insights into the role of nuclear factor-κB in cell growth regulation Am J Pathol 2001 159 387 397 11485895
Beg AA Sha WC Bronson RT Ghosh S Baltimore D Embryonic lethality and liver degeneration in mice lacking the RelA component of NF-kappa B Nature 1995 376 167 170 7603567 10.1038/376167a0
Kikuchi E Horiguchi Y Nakashima J Kuroda K Oya M Ohigashi T Takahashi N Shima Y Umezawa K Murai M Suppression of hormone-refractory prostate cancer by a novel nuclear factor κB inhibitor in nude mice Cancer Res 2003 63 107 110 12517785
|
16277688
|
PMC1297584
|
CC BY
|
2021-01-04 16:02:40
|
no
|
Arthritis Res Ther. 2005 Sep 30; 7(6):R1348-R1359
|
utf-8
|
Arthritis Res Ther
| 2,005 |
10.1186/ar1834
|
oa_comm
|
==== Front
Arthritis Res TherArthritis Research & Therapy1478-63541478-6362BioMed Central London ar18371627769110.1186/ar1837Research ArticleThe impact of HLA-DRB1 genes on extra-articular disease manifestations in rheumatoid arthritis Turesson Carl [email protected] Daniel J [email protected] Cornelia M [email protected] Lennart TH [email protected] Jörg J [email protected] Ingemar F [email protected] Gunnar [email protected]äll-Wåhlin Britt-Marie [email protected] Lennart [email protected] Sonja A [email protected] Eric L [email protected] Department of Rheumatology, Malmö University Hospital, Södra Förstadsgatan 101, 205 02 Malmö, Sweden2 Division of Rheumatology, Mayo Clinic College of Medicine, 200 First Street SW, Rochester, Minnesota 55905, USA3 Department of Health Sciences Research, Mayo Clinic College of Medicine, 200 First Street SW, Rochester, Minnesota 55905, USA4 Lowance Center for Human Immunology, Emory University School of Medicine, 101 Woodruff Circle, Atlanta, Georgia 30322, USA5 Spenshult Hospital for Rheumatic Diseases, 313 92 Oskarström, Sweden6 Department of Rheumatology, Lund University Hospital, Kioskgatan 3, 221 85 Lund, Sweden7 Department of Clinical Microbiology and Immunology, Lund University Hospital, Sölvegatan 23, 223 62 Lund, Sweden2005 11 10 2005 7 6 R1386 R1393 27 7 2005 31 8 2005 6 9 2005 8 9 2005 Copyright © 2005 Turesson 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 objective of this study was to examine HLA-DRB1 and HLA-DQB1 genotypes in patients with severe extra-articular rheumatoid arthritis (ExRA) and to compare them with the genotypes of rheumatoid arthritis (RA) patients without extra-articular manifestations. Patients with severe ExRA were recruited from a large research database of patients with RA, from two cohorts of prevalent RA cases, and from a regional multicenter early RA cohort. Cases with ExRA manifestations (n = 159) were classified according to predefined criteria. Controls (n = 178) with RA but no ExRA were selected from the same sources. Cases and controls were matched for duration of RA and for clinical center. PCR based HLA-DRB1 and HLA-DQB1 genotyping was performed using the Biotest SSP kit, with additional sequencing in order to distinguish DRB1*04 subtypes. Associations between alleles and disease phenotypes were tested using multiple simulations of random distributions of alleles. There was no difference in global distribution of HLA-DRB1 and HLA-DQB1 alleles between patients with ExRA and controls. DRB1*0401 (P = 0.003) and 0401/0401 homozygosity (P = 0.002) were more frequent in Felty's syndrome than in controls. The presence of two HLA-DRB1*04 alleles encoding the shared epitope (SE) was associated with ExRA (overall odds ratio 1.79, 95% confidence interval 1.04–3.08) and with rheumatoid vasculitis (odds ratio 2.44, 95% confidence interval 1.22–4.89). In this large sample of patients with ExRA, Felty's syndrome was the only manifestation that was clearly associated with HLA-DRB1*0401. Other ExRA manifestations were not associated with individual alleles but with DRB1*04 SE double dose genotypes. This confirms that SE genes contribute to RA disease severity and ExRA. Other genetic and environmental factors may have a more specific impact on individual ExRA manifestations.
==== Body
Introduction
Rheumatoid arthritis (RA) is a systemic inflammatory disease that, in a substantial proportion of patients, is associated with the development of extra-articular manifestations. These extra-articular RA (ExRA) manifestations can have a defining impact on disease outcome, including increased premature mortality compared with RA in general [1-4]. Severe ExRA occurs both in patients recently diagnosed with RA and in those with long-standing disease [2]. Suggested predictors of ExRA include clinical, serologic, and genetic factors [5].
There is strong evidence of a role for genetic factors in the etiology of RA [6-8], and genetic polymorphisms are probably involved in the wide variation in disease expression. As for most diseases classified according to a list of criteria, rather than specific diagnostic tests, the disease phenotype in RA is heterogeneous. The presence of disease susceptibility alleles may define subsets of patients with different disease courses, including patients with mild, nonerosive disease and those with a true RA phenotype and progressive disease, with extensive joint damage and ExRA manifestations. On the other hand, genetic markers not related to disease susceptibility may influence disease progression and risk for developing ExRA.
HLA (human leukocyte antigen) alleles have been implicated in a number of chronic inflammatory diseases. RA has been associated with the 'shared epitope' (SE) of HLA-DRB1, which includes DRB1*04 and DRB1*01 alleles [9]. Recent genome-wide scanning studies using microsatellite loci have confirmed that there is strong linkage between this region and RA [10,11]. RA-associated HLA-DRB1*04 alleles have been reported mainly in patients with severe disease [12-16]. A meta-analysis of studies of disease progression in RA [17] revealed an association between HLA-DRB1*04 and erosive disease, and in a recently reported survey of an extensively investigated cohort of patients with early RA [18] homozygosity for HLA-DRB1*04 was a major predictor of development of erosions. DRB1*04 alleles have also been specifically associated with ExRA [19-21], and a specific impact of DRB1*04 homozygosity has been suggested. Some authors have reported an association with the 0401/0401 genotype [21,22] whereas others have found the 0401/0404 genotype to be more frequent among patients with ExRA [23]. These discrepancies may reflect variability in the relative frequencies of HLA-DRB1*0401 in different populations. For example, in East Asian populations, in which DRB1*0401 is rare and DRB1*0405 is the most frequent RA associated HLA-DRB1 genotype [24], the latter allele has also been reported to be associated with an increased risk for ExRA manifestations [25].
All previous studies of major histocompatibility class (MHC) class II genes and ExRA have been based on small patient samples, limiting the generalizability of the results. Most studies were not sufficiently powered to examine the effect of linkage disequilibrium within the MHC, including HLA-DQB1 alleles. Previous investigations did not use consistent and well characterized definitions of ExRA, which is a matter of vital importance to the study of disease phenotypes in RA [26].
The purpose of this study was to investigate associations between HLA-DRB1 and HLA-DQB1 alleles and severe ExRA manifestations in a multicenter case-control study of patients with well characterized disease. To our knowledge, this is the largest sample of patients with severe ExRA ever reported. We report that patients with ExRA manifestations are more likely to carry a double dose of DRB1*04 SE alleles, and we demonstrate that the impact of individual DRB1 alleles is limited.
Materials and methods
Patients
Patients with severe ExRA according to predefined criteria [2,3] were recruited from the rheumatology laboratory database of the Mayo Clinic (Rochester, MN, USA), from two clinic-based cohorts of patients with ExRA from Malmö University Hospital and Lund University Hospital (Sweden), and from a Swedish multicenter early RA cohort (the Better AntiRheumatic PharmacOTherapy [BARFOT] cohort). ExRA manifestations studied included pericarditis, pleuritis, Felty's syndrome, scleritis, episcleritis, glomerulonephritis, vasculitis-related neuropathy, major cutaneous vasculitis, and vasculitis involving other organs. Felty's syndrome was defined as RA-associated neutropenia and splenomegaly, with other potential causes excluded or unlikely. In addition, the criteria for severe ExRA were modified to include RA-associated interstitial lung disease, as previously described [5]. Controls, defined as patients with RA without current or previous signs of extra-articular disease manifestations, in accordance with the same criteria [2,3,5], were selected from the corresponding centers. One patient with RA (control) was matched to each patient with ExRA (case) according to duration of RA and clinical center. All cases and controls fulfilled 1987 American College of Rheumatology criteria for classification of RA [27].
Eighty-eight patients fulfilling the predefined criteria for ExRA (see above) were identified from the Mayo Clinic rheumatology laboratory database, and their medical case records were subjected to a structured review, as previously described [22]. A random sample of 184 patients with RA but without ExRA were identified from this database after careful medical record review. Controls from this sample were matched with cases for duration of RA ± 5 years. DNA samples were available from 86 ExRA cases and 85 controls for HLA typing.
Another cohort of patients was recruited from a prospective study of extra-articular disease manifestations and vascular comorbidities in RA from the rheumatology outpatient clinic of Malmö University Hospital. Consecutive patients with recently diagnosed severe extra-articular disease manifestations were invited to participate. Patients with non-extra-articular RA, matched to extra-articular patients for age, sex and disease duration (± 1 year), were selected from a community-based register of RA patients in the city of Malmö [28] or from a community-based early RA inception cohort from the same area. Samples from 28 patients with ExRA (cases) and 28 matched patients with RA but without ExRA (controls) were available for analysis. Thirty-five patients with ExRA (cases) and 42 patients with RA but without extra-articular disease (controls), matched for disease duration, from a case-control study of predictors of ExRA at the University Hospitals in Malmö and Lund [22] were also included in the analysis. Results of HLA-DR and HLA-DQ genotyping for some of these patients were reported previously [22].
In addition, patients were recruited from the BARFOT registry [29], which includes patients participating in a structured program for follow up of newly diagnosed RA in southern Sweden. This registry includes virtually all adult patients with new onset of inflammatory polyarthritis within the catchment area of the six participating rheumatology centers of the BARFOT program (total population is approximately 1.5 million), including patients fulfilling the 1987 American College of Rheumatology classification criteria for RA [27]. From 1992 to 2001, a total of 1,589 consecutive patients were recruited to the registry. Referring rheumatologists are encouraged to report ExRA manifestations occurring in these patients to the register. All reported ExRA cases (n = 35) were reviewed and classified according to the study criteria [2,3]. Of these, 26 patients fulfilled the criteria for ExRA. Controls without ExRA were matched to the cases by sex, age at inclusion, disease duration and, when possible, geographic region. All potential controls were reviewed in order to ensure that they did not have a history of ExRA. Samples for genotyping were available from ten ExRA cases and 24 non-ExRA controls in this subset.
Data on serologic tests for rheumatoid factor (RF) and antinuclear antibodies (ANAs), and information on smoking status are prospectively collected as part of a structured follow up of patients in the BARFOT study. Data on these parameters for patients from the other centers (Malmö, Lund and the Mayo Clinic) were obtained by thorough review of all available clinical records.
All patients gave informed consent to participate in the study. The study was approved by the Research Ethics Committee at Lund University and by the Institutional Review Board at the Mayo Clinic.
Genotyping
DNA for HLA-DRB1/DQB1 typing of patients recruited from the Mayo Clinic was isolated from peripheral blood mononuclear cells using the DNA Isolation Kit for Mammalian Blood (Roche Applied Sciences, Indianapolis, IN, USA). For patients from the Swedish RA cohorts, DNA was extracted from whole blood using the QIAamp minikit (Qiagen, Hilden, Germany) at the DNA/RNA Genotyping Laboratory, SWEGENE Resource Center for Profiling Polygenic Diseases (Lund University and Malmö University Hospital, Sweden). The purified DNA was used for HLA-DRB1 and HLA-DQB1 determination with the PCR-based Micro-SSP DRB and DQB generic typing trays (Biotest AG, Dreiech, Germany). Because the DRB kit does not detect HLA-DRB1*04 allelic variations, all samples that were positive for HLA-DRB1*04 were re-amplified by PCR using a primer set that amplified all HLA-DRB1*04 alleles: 5'-GTTTCTTGGAGCAGGTTAAACA-3' (HLA-DRB1*04) and 5'-GCCGCTGCACTGTGAAGCTCTC-3' (HLA-DRB1 generic). Samples were then purified using the High Pure PCR Product Purification Kit (Roche Applied Sciences) and sequenced in the Mayo Clinic molecular biology core facility on a PRISM 37 DNA Sequencer (Applied Biosystems, Foster City, CA, USA) with the HLA-DRB1 primer as the initiating primer. The specific HLA-DRB1*04 allele was then assigned on the basis of the sequencing results. For the statistical analysis, the SE encoding rare DRB1*0401-like alleles *0409, *0413, *0416 and *0421 were classified as *0401; alleles *0408, *0410 and *0419 were classified as *0404. The DRB1*0405 alleles were analyzed as a separate entity. All other DRB1*04 alleles were classified as DRB1*04 non-SE alleles.
Statistical analysis
The age at RA diagnosis and the duration of RA at inclusion in ExRA cases and non-ExRA controls with RA were compared using the t test. The sex distribution, the number of smokers and the number of patients with a positive RF test or ANA test at any time were compared between the cases and controls using Pearson's χ2 statistic.
To compare the distribution of alleles between cases and controls, we used Armitage's trend in proportions, which does not treat the two alleles within a person as independent (i.e. it does not assume Hardy-Weinberg equilibrium). This approach reduces to the usual Pearson χ2 statistic for comparing allele frequencies when genotype proportions match Hardy-Weinberg proportions [30], and is the preferred way to compare allele frequencies [31]. However, the usual Armitage test for trend is for only two alleles. A multivariate extension for more than two alleles, which compares allele counts between cases and controls, is based on the score statistic for logistic regression. For this score statistic, each subject receives a vector of scores, where each element of the vector counts alleles of each type. From this score statistic, we computed a global test of association between case/control status and all alleles of HLA-DRB1 and HLA-DQB1 separately. Because the distribution of this statistic is not known, we performed simulations to compute P values. The case/control status was randomly permuted, and the simulated statistic computed and compared with the observed statistic. This simulation process was repeated 1,000 times to compute P values, both for the maximum statistic and allele-specific Armitage trend tests. The distribution of combinations of HLA-DRB1 and DQ alleles (i.e. the distribution of HLA-DRB1-DQ haplotypes) was similarly compared in cases and controls.
To evaluate the association of single or double dose of HLA-DRB1*04 SE subtypes with case/control status, or with a particular manifestation of ExRA, we used logistic regression to calculate odds ratios (ORs) and 95% confidence interval (CI).
Results
A total of 159 patients with severe ExRA according to predefined criteria [5,23] were identified. Forty-three patients had vasculitis, defined as biopsy proven vasculitis or major cutaneous vasculitis diagnosed by a dermatologist. Additional subgroups analyzed were neuropathy (mononeuropathy or polyneuropathy; n = 40), interstitial lung disease (n = 25), Felty's syndrome (n = 21) and pericarditis (n = 27). These were compared with 178 controls with non-extra-articular RA. Disease duration and age at RA onset was similar in cases and controls (mean 11.3 years versus 12.5 years for duration, and mean 50.1 years versus 50.4 years for age at RA onset; Table 1). There was a trend toward a relative predominance of male patients in the ExRA group (P = 0.06). However, this comparison is skewed because of the matching of cases and controls for sex in two of the subsamples. A positive test for RF (P < 0.0001) or ANAs (P < 0.0001) at any time were both significantly associated with ExRA.
Some of the individual severe ExRA manifestations occurred together more frequently than expected. Among the 21 patients with Felty's syndrome, three (14%) had evidence of vasculitis. In the subset with vasculitis, 15 out of 43 (35%) had neuropathy and seven (16%) had interstitial lung disease.
Overall effects of HLA-DRB1 alleles
The distribution of HLA-DRB1 was not significantly different between ExRA cases and non-ExRA controls (global P = 0.19; Table 2). The most frequent HLA-DRB1 allele in both groups was HLA-DRB1*0401, and this allele tended to be more common among patients with ExRA (allele frequency 0.326 versus 0.263; P = 0.09). HLA-DRB1*0401 was significantly associated with Felty's syndrome (allele frequency 0.475; P = 0.003) but not with other individual manifestations when compared with non-extra-articular RA (Fig. 1). The rare allele HLA-DRB1*12 was more common in the ExRA subgroup (allele frequency 0.023 versus 0.003; P = 0.02). The DRB1*0405 (allele frequency 0.019 versus 0.003; P = 0.01) and DRB1*0404 (allele frequency 0.119 versus 0.085; P = 0.14) alleles were also more frequent in patients with ExRA than in non-ExRA controls. One of the HLA-DRB1*04 alleles encoding the SE (DRB1*0401, *0404, or *0405) was present in 105 out of 151 ExRA patients as compared with 96 out of 178 non-ExRA patients (OR 1.77, 95% CI 1.13–2.77). The impact of the presence of DRB1*04 SE alleles on risk for ExRA was variable for the different manifestations (Fig. 2). Patients with RA and vasculitis were more likely to carry DRB1*04 SE alleles than patients with RA and no vasculitis (OR 2.07, 95% CI 1.00–4.25). Similar trends were found for Felty's syndrome and neuropathy, but the associations were not significant (Fig. 2).
Effects of homozygosity for the shared epitope
The homozygous genotype HLA-DRB1*0401/0401 was significantly more frequent in patients with Felty's syndrome (genotype frequency 0.286; P = 0.002) and patients with pericarditis (genotype frequency 0.185; P = 0.043) than in non-ExRA controls (frequency 0.068). Other ExRA manifestations were not associated with any specific homozygous genotype. The presence of two HLA-DRB1*04 SE alleles was significantly associated with ExRA overall (OR 1.79, 95% CI 1.04–3.08), Felty's syndrome (OR 2.63, 95% CI 1.04–6.63), and vasculitis (OR 2.44, 95% CI 1.22–4.89) compared with patients with RA who lacked these manifestations. By contrast, pericarditis, neuropathy, and interstitial lung disease were not associated with double dose of HLA-DRB1*04 SE alleles (Table 3).
Effects of HLA-DQB alleles
The distribution of HLA-DQB alleles was not significantly different between ExRA cases and non-ExRA controls (P = 0.11; Table 4). The relatively rare allele HLA-DQ4 tended to occur more frequently in ExRA cases (allele frequency 0.046 versus 0.014; P = 0.037). Other than that, there was no significant difference in the occurrence of DQB alleles between patients with ExRA overall or individual ExRA manifestations and non-ExRA controls. There was no significant global difference in the frequency of homozygous HLA-DQB genotypes between cases and controls except for patients with ExRA and pericarditis (P = 0.04). HLA-DQ8/DQ8 homozygosity was more common in patients with pericarditis than in non-ExRA patients with RA (genotype frequency 0.120 versus 0.029; P = 0.021).
Analyses of linkage disequilibrium
Haplotype analysis indicated that the association between ExRA and HLA-DRB1*04 SE homozygosity was due to the importance of the DRB1*04 genotype, rather than being secondary to associations with HLA-DRB1-DQB haplotypes (data not shown). Similarly, the associations between Felty's syndrome and DRB1*0401, and between pericarditis and DQ8/DQ8 were not explained by DRB1-DQB haplotype associations.
Discussion
In this large sample of patients with severe ExRA, we found Felty's syndrome to be associated with HLA-DRB1*0401. There was no significant difference in the global distribution of HLA-DRB1 or HLA-DQB alleles compared with non-extra-articular RA controls for any other manifestation or for ExRA overall. Patients with severe ExRA were more likely to carry HLA-DRB1*04 SE alleles, and genotypes featuring a double dose of DRB1*04 SE alleles were associated with rheumatoid vasculitis, Felty's syndrome, and all ExRA combined.
A number of studies have indicated a role for HLA-DR4 related genes in ExRA [26]. In Caucasians of Northern European origin, severe ExRA has been associated with DRB1*0401/0401 homozygosity in particular [21,22]. In a recent meta-analysis of HLA-DRB1 genotyping studies of patients with RA-associated vasculitis conducted by Gorman and coworkers [32], vasculitis was found to be associated with the genotypes 0401/0401, 0401/0404, and 0401/0101. In other meta-analyses by the same group, double dose of DRB1*04 SE alleles was associated with radiographic signs of progressive joint damage in northern European Caucasians [17], but there was no significant association between SE and the presence of rheumatoid nodules [33]. Taken together, these findings indicate that DRB1*04 SE double gene dose is associated with disease severity in RA, and that such genotypes may contribute specifically to risk for severe ExRA manifestations.
On the other hand, there was considerable heterogeneity across individual ExRA manifestations. The association between Felty's syndrome and DRB1*0401 is well established [34,35]. In contrast, we did not observe any significant association with single or double DRB1*04 gene dose for patients with pericarditis, neuropathy, or interstitial lung disease. This indicates that the importance of HLA-DRB1 alleles may be variable for different manifestations, although our failure to detect an effect could be due to sample size or selection.
Severe ExRA manifestations tend to cluster in individual patients with RA [36]. The high prevalence of vasculitis in patients with Felty's syndrome observed in the present study is consistent with the literature [37], and may in part be due to shared genetic factors such as HLA-DRB1*04 alleles. In a survey of the community-based Olmsted county RA cohort [36] we found clustering of a number of different ExRA features, including a frequent co-occurrence of vasculitis with neuropathy and rheumatoid lung disease. We made similar observations in the present study. Such clustering may be explained by both genetic and environmental factors.
The association between HLA-DRB1 genotypes and RA disease severity, including ExRA, has been interpreted as reflecting the importance of T cells in the pathogenesis of RA [26]. HLA-DR and other MHC molecules are involved in presentation of antigens to T cells, and in positive and negative selection of T cells in the thymus. Because there appears to be a stoichiometric relationship between MHC molecules on the cell surface and positive selection mechanisms in thymic maturation of T cells, it has been suggested that the explanation for the gene dose effect seen in ExRA is its effect on T-cell diversity [21,38]. The T-cell repertoire in patients with RA is markedly contracted, with less diversity and emergence of dominant T-cell clonotypes [39]. T-cell abnormalities in patients with ExRA include expansion of CD8+ large granular lymphocytes [40] and of immunosenescent CD4+CD28- cells [41,42], and extensive CD4+ infiltrates in RA-associated interstitial pneumonitis [43]. The importance of HLA-DRB1 genes and other genes with a role in T-cell selection and T-cell function for these phenomena require further study.
In accordance with previous studies, we found patients with ExRA to be more likely to be RF positive and ANA positive [22,44]. This suggests a role for both B cells and T cells, possibly including dysregulated B cell-T cell interaction, in ExRA.
New genetic associations that were not postulated and have not been reproduced should be interpreted with caution. Given the nonsignificant results of the global distribution tests, the associations between ExRA and some rare DRB1 and DQB1 alleles (i.e. DRB1*12 and DQ4) are probably due to chance. The negative global test for HLA-DRB1 alleles in ExRA overall also suggests that the impact of DRB1*04 SE on the risk for severe ExRA manifestations is not strong, although it is reproducible in separate patient samples.
The lack of association between ExRA and HLA-DQB1 alleles, and the lack of association with HLA-DRB1-DQB1 haplotypes favors a specific role for HLA-DRB1 genes in ExRA, rather than secondary associations due to linked genes. Nevertheless, we cannot exclude the possibility that linkage disequilibrium with other genes in MHC explain our results.
The patients included in this study were recruited from four different centers, and the background RA population from which they were sampled is not fully characterized, at least not for the patients seen at Lund University Hospital and at the Mayo Clinic. On the other hand, these patients were recruited during a period when there was particular interest in patients with severe ExRA at each of the centers, suggesting that they should reflect the majority of patients with ExRA seen and be representative of the ExRA population as a whole.
In multicenter studies of genetic markers, ethnic heterogeneity of the studied patient samples must be considered. However, the majority of the patients included at the Mayo Clinic were Caucasians of northern European origin, similar to the patients from southern Sweden. Thus, our result could be generalized to RA patients with this ethnic background but not to other populations.
Conclusion
In a study of a large sample of patients with ExRA, we have confirmed an association between HLA-DRB1*0401 and Felty's syndrome, but we found no association between ExRA overall or other individual manifestations and specific HLA-DRB1 alleles. Double dose HLA-DRB1*04 SE genotypes are associated with a modestly increased risk for vasculitis and other ExRA manifestations. Other genetic and environmental factors are likely to be more important for the systemic features of RA.
Abbreviations
ANA = antinuclear antibody; CI = confidence interval; ExRA = extra-articular rheumatoid arthritis; HLA = human leukocyte antigen; MHC = major histocompatibility complex; OR = odds ratio; PCR = polymerase chain reaction; RF = rheumatoid factor; SE = shared epitope.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
CT conceived the study, was responsible for the recruitment and classification of patients, and drafted the manuscript. DS performed the statistical analysis and helped to draft the manuscript. CW participated in the design and coordination of the study, and recruited a subset of the patients. LJ participated in the recruitment of a subset of the patients and the interpretation of the statistical data, and helped to draft the manuscript. JG recruited a subset of the patients and participated in the design and coordination of the study. GS participated in the recruitment of patients and the molecular genetics analysis. IP participated in the recruitment of patients, the extraction of clinical data, and the interpretation of the statistical analysis. BMNW participated in the recruitment and classification of patients and the extraction of clinical data. LT performed part of the molecular genetics analysis and helped to draft the manuscript. SD performed part of the molecular genetics analysis and participated in the coordination of the study. EM conceived the study together with CT, performed part of the molecular genetics analysis, participated in the design and coordination of the study, and in the interpretation of the statistical data, and helped to draft the manuscript. All authors read and approved the final manuscript.
Acknowledgements
The authors would like to thank Angelina Lippert for excellent work with the genotyping. We also thank the BARFOT study group for their support and for contributing patients to the study, and all colleagues at the Mayo Clinic, Malmö University Hospital, Lund University Hospital, Spenshult Hospital for Rheumatic Diseases, Karolinska University Hospital, Huddinge, and at the general hospitals in Helsingborg, Kalmar, Kristianstad and Mölndal for their help in patient recruitment. This study was supported by NIH grant K24 AR 47578-01A1, the Swedish Rheumatism Association, the Swedish Society for Medicine, Lund University, and a grant from the Mayo Clinic.
Figures and Tables
Figure 1 Variation in frequency of HLA-DRB1*0401 by disease phenotype in RA. HLA-DRB1*0401 was significantly more frequent in patients with Felty's syndrome (P = 0.003) than in non-extra-articular rheumatoid arthritis (RA) controls, but patients with other manifestations did not differ significantly from controls. ExRA, extra-articular RA; ILD, interstitial lung disease; Neuro, vasculitis related neuropathy.
Figure 2 ExRA manifestations among those carrying carrying HLA-DRB1*04 shared epitope alleles. Shown are odds ratios (ORs) with 95% confidence intervals (CIs) for different extra-articular rheumatoid arthritis (ExRA) manifestations for patients carrying HLA-DRB1*04 shared epitope alleles. ILD, interstitial lung disease.
Table 1 Demographic data and clinical predictors of ExRA
ExRA Non-ExRA P
Number 159 178
Age at RA diagnosis (years; mean ± SD) 50.1 ± 14.4 50.4 ± 14.8 0.87
Disease duration (years; mean ± SD) 11.3 ± 11.2 12.5 ± 11.3 0.34
Male/female (n) 75/84 66/112 0.06
RF positivea (%) 87.2 58.3 <0.0001
ANA positiveb (%) 60.8 33.8 <0.0001
aInformation available for 149 extra-articular rheumatoid arthritis (ExRA) and 163 non-ExRA patients.
bInformation available from 120 ExRA and 151 non-ExRA patients. ANA, antinuclear antibody; RA, rheumatoid arthritis; RF, rheumatoid factor; SD, standard deviation.
Table 2 Frequencies of HLA-DRB1 alleles in patients with ExRA compared with patients with non-extra-articular RA
HLA-DRB1 allele Allele frequency P
ExRA Non-ExRA
DRB1*01 0.119 0.130 0.74
DRB1*03 0.071 0.113 0.12
DRB1*0401 0.326 0.263 0.09
DRB1*0404 0.119 0.085 0.14
DRB1*0405 0.019 0.003 0.01
Non-SE DRB1*04 0.019 0.045 0.09
DRB1*07 0.052 0.065 0.54
DRB1*08 0.013 0.014 1.00
DRB1*09 0.023 0.011 0.23
DRB1*10 0.016 0.020 0.73
DRB1*11 0.036 0.054 0.28
DRB1*12 0.023 0.003 0.02
DRB1*13 0.042 0.045 0.83
DRB1*15 0.097 0.110 0.65
Global: P = 0.19. ExRA, extra-articular rheumatoid arthritis; SE, shared epitope.
Table 3 The effect of homozygosity for HLA-DRB1*04 shared epitope alleles on risk for severe ExRA manifestations
Manifestations Homozygotes (yes/no; %) OR (95% CI) P
Cases with manifestation Controls without manifestation
All ExRA 39/116 (25) 28/149 (16) 1.79 (1.04–3.08) 0.04
Pericarditis 8/19 (29) 59/246 (19) 1.76 (0.73–4.19) 0.21
Felty's syndrome 8/13 (38) 59/252 (19) 2.63 (1.04–6.63) 0.04
Neuropathy 8/32 (20) 59/233 (20) 0.99 (0.43–2.25) 0.98
Interstitial lung disease 3/22 (12) 64/243 (21) 0.52 (0.15–1.78) 0.30
Vasculitis 15/28 (35) 52/237 (18) 2.44 (1.22–4.89) 0.01
CI, confidence interval; ExRA, extra-articular rheumatoid arthritis; OR, odds ratio.
Table 4 Frequencies of HLA-DQ alleles in patients with ExRA compared with patients with non-extra-articular RA
HLA-DQ allele Allele frequency P
ExRA Non-ExRA
DQ2 0.122 0.121 0.91
DQ3 0.017 0.029 0.50
DQ4 0.046 0.014 0.04
DQ5 0.129 0.188 0.06
DQ6 0.139 0.165 0.37
DQ7 0.252 0.249 1.00
DQ8 0.222 0.170 0.13
DQ9 0.066 0.043 0.22
Global: P = 0.11. ExRA, extra-articular RA.
==== Refs
Gordon DA Stein JL Broder I The extra-articular features of rheumatoid arthritis. A systematic analysis of 127 cases Am J Med 1973 54 445 452 4348718 10.1016/0002-9343(73)90040-5
Turesson C Jacobsson L Bergstrom U Extra-articular rheumatoid arthritis: prevalence and mortality Rheumatology (Oxford) 1999 38 668 674 10461483 10.1093/rheumatology/38.7.668
Turesson C O'Fallon WM Crowson CS Gabriel SE Matteson EL Occurrence of extraarticular disease manifestations is associated with excess mortality in a community based cohort of patients with rheumatoid arthritis J Rheumatol 2002 29 62 67 11824973
Gabriel SE Crowson CS Kremers HM Doran MF Turesson C O'Fallon WM Matteson EL Survival in rheumatoid arthritis: a population-based analysis of trends over 40 years Arthritis Rheum 2003 48 54 58 12528103 10.1002/art.10705
Turesson C Jacobsson LT Epidemiology of extra-articular manifestations in rheumatoid arthritis Scand J Rheumatol 2004 33 65 72 15163106 10.1080/03009740310004621
Aho K Koskenvuo M Tuominen J Kaprio J Occurrence of rheumatoid arthritis in a nationwide series of twins J Rheumatol 1986 13 899 902 3820198
Silman AJ MacGregor AJ Thomson W Holligan S Carthy D Farhan A Ollier WE Twin concordance rates for rheumatoid arthritis: results from a nationwide study Br J Rheumatol 1993 32 903 907 8402000
Grant SF Thorleifsson G Frigge ML Thorsteinsson J Gunnlaugsdottir B Geirsson AJ Gudmundsson M Vikingsson A Erlendsson K Valsson J The inheritance of rheumatoid arthritis in Iceland Arthritis Rheum 2001 44 2247 2254 11665965 10.1002/1529-0131(200110)44:10<2247::AID-ART387>3.0.CO;2-Y
Gregersen PK Silver J Winchester RJ The shared epitope hypothesis. An approach to understanding the molecular genetics of susceptibility to rheumatoid arthritis Arthritis Rheum 1987 30 1205 1213 2446635
Jawaheer D Seldin MF Amos CI Chen WV Shigeta R Monteiro J Kern M Criswell LA Albani S Nelson IL A genomewide screen in multiplex rheumatoid arthritis families suggests genetic overlap with other autoimmune diseases Am J Hum Genet 2001 68 927 936 11254450 10.1086/319518
Jawaheer D Seldin MF Amos CI Chen WV Shigeta R Etzel C Damle A Xiao X Chen D Lum RF Screening the genome for rheumatoid arthritis susceptibility genes: a replication study and combined analysis of 512 multicase families Arthritis Rheum 2003 48 906 916 12687532 10.1002/art.10989
Weyand CM Hicok KC Conn DL Goronzy JJ The influence of HLA-DRB1 genes on disease severity in rheumatoid arthritis Ann Intern Med 1992 117 801 806 1416553
Gough A Faint J Salmon M Hassell A Wordsworth P Pilling D Birley A Emery P Genetic typing of patients with inflammatory arthritis at presentation can be used to predict outcome Arthritis Rheum 1994 37 1166 1170 8053954
Weyand CM McCarthy TG Goronzy JJ Correlation between disease phenotype and genetic heterogeneity in rheumatoid arthritis J Clin Invest 1995 95 2120 2126 7738179
Salvarani C Macchioni PL Mantovani W Bragliani M Collina E Cremonesi T Battistel B Boiardi L HLA-DRB1 alleles associated with rheumatoid arthritis in Northern Italy: correlation with disease severity Br J Rheumatol 1998 37 165 169 9569071 10.1093/rheumatology/37.2.165
Combe B Dougados M Goupille P Cantagrel A Eliaou JF Sibilia J Meyer O Sany J Daures JP Dubois A Prognostic factors for radiographic damage in early rheumatoid arthritis: a multiparameter prospective study Arthritis Rheum 2001 44 1736 1743 11508423 10.1002/1529-0131(200108)44:8<1736::AID-ART308>3.0.CO;2-I
Gorman JD Lum RF Chen JJ Suarez-Almazor ME Thomson G Criswell LA Impact of shared epitope genotype and ethnicity on erosive disease: a meta-analysis of 3,240 rheumatoid arthritis patients Arthritis Rheum 2004 50 400 412 14872482 10.1002/art.20006
Goronzy JJ Matteson EL Fulbright JW Warrington KJ Chang-Miller A Hunder GG Mason TG Nelson AM Valente RM Crowson CS Prognostic markers of radiographic progression in early rheumatoid arthritis Arthritis Rheum 2004 50 43 54 14730598 10.1002/art.11445
Ollier W Venables PJ Mumford PA Maini RN Awad J Jaraquemada D D'Amaro J Festenstein H HLA antigen associations with extra-articular rheumatoid arthritis Tissue Antigens 1984 24 279 291 6335792
Hillarby MC Clarkson R Grennan DM Bate AS Ollier W Sanders PA Chattophadhyay C Davis M O'Sullivan MM Williams B Immunogenetic heterogeneity in rheumatoid disease as illustrated by different MHC associations (DQ, Dw and C4) in articular and extra-articular subsets Br J Rheumatol 1991 30 5 9 1991218
Weyand CM Xie C Goronzy JJ Homozygosity for the HLA-DRB1 allele selects for extraarticular manifestations in rheumatoid arthritis J Clin Invest 1992 89 2033 2039 1602009
Turesson C Jacobsson L Bergstrom U Truedsson L Sturfelt G Predictors of extra-articular manifestations in rheumatoid arthritis Scand J Rheumatol 2000 29 358 364 11132204 10.1080/030097400447552
Perdriger A Chales G Semana G Guggenbuhl P Meyer O Quillivic F Pawlotsky Y Role of HLA-DR-DR and DR-DQ associations in the expression of extraarticular manifestations and rheumatoid factor in rheumatoid arthritis J Rheumatol 1997 24 1272 1276 9228123
Kim HY Kim TG Park SH Lee SH Cho CS Han H Predominance of HLA-DRB1*0405 in Korean patients with rheumatoid arthritis Ann Rheum Dis 1995 54 988 990 8546532
Koh WH Chan SH Lin YN Boey ML Association of HLA-DRB1*0405 with extraarticular manifestations and erosions in Singaporean Chinese with rheumatoid arthritis J Rheumatol 1997 24 629 632 9101492
Turesson C Weyand CM Matteson EL Genetics of rheumatoid arthritis: Is there a pattern predicting extraarticular manifestations? Arthritis Rheum 2004 51 853 863 15478157 10.1002/art.20693
Arnett FC Edworthy SM Bloch DA McShane DJ Fries JF Cooper NS Healey LA Kaplan SR Liang MH Luthra HS The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis Arthritis Rheum 1988 31 315 324 3358796
Jacobsson L Lindroth Y Marsal L Tejler L [The Malmo model for private and public rheumatological outpatient care. Cooperation makes it possible to introduce disease modifying treatment quickly] Lakartidningen 2001 98 4710 4716 11715248
Svensson B Schaufelberger C Teleman A Theander J Remission and response to early treatment of RA assessed by the Disease Activity Score. BARFOT study group. Better Anti-rheumatic Farmacotherapy Rheumatology (Oxford) 2000 39 1031 1036 10986311 10.1093/rheumatology/39.9.1031
Sasieni PD From genotypes to genes: doubling the sample size Biometrics 1997 53 1253 1261 9423247
Devlin B Roeder K Genomic control for association studies Biometrics 1999 55 997 1004 11315092 10.1111/j.0006-341X.1999.00997.x
Gorman JD David-Vaudey E Pai M Lum RF Criswell LA Particular HLA-DRB1 shared epitope genotypes are strongly associated with rheumatoid vasculitis Arthritis Rheum 2004 50 3476 3484 15529352 10.1002/art.20588
Gorman JD David-Vaudey E Pai M Lum RF Criswell LA Lack of association of the HLA-DRB1 shared epitope with rheumatoid nodules: an individual patient data meta-analysis of 3,272 Caucasian patients with rheumatoid arthritis Arthritis Rheum 2004 50 753 762 15022316 10.1002/art.20119
Lanchbury JS Jaeger EE Sansom DM Hall MA Wordsworth P Stedeford J Bell JI Panayi GS Strong primary selection for the Dw4 subtype of DR4 accounts for the HLA-DQw7 association with Felty's syndrome Hum Immunol 1991 32 56 64 1685490 10.1016/0198-8859(91)90117-R
Coakley G Brooks D Iqbal M Kondeatis E Vaughan R Loughran TP JrPanayi GS Lanchbury JS Major histocompatility complex haplotypic associations in Felty's syndrome and large granular lymphocyte syndrome are secondary to allelic association with HLA-DRB1 *0401 Rheumatology (Oxford) 2000 39 393 398 10817772 10.1093/rheumatology/39.4.393
Turesson C Christianson TJH McClelland RL Matteson EL Clustering of extra-articular manifestations in rheumatoid arthritis Arthritis Rheum 2004 51 S243 10.1002/art.20693
Schneider HA Yonker RA Katz P Longley S Panush RS Rheumatoid vasculitis: experience with 13 patients and review of the literature Semin Arthritis Rheum 1985 14 280 286 3909408 10.1016/0049-0172(85)90047-2
Weyand CM Goronzy JJ Functional domains on HLA-DR molecules: implications for the linkage of HLA-DR genes to different autoimmune diseases Clin Immunol Immunopathol 1994 70 91 98 8299234 10.1006/clin.1994.1015
Wagner UG Koetz K Weyand CM Goronzy JJ Perturbation of the T cell repertoire in rheumatoid arthritis Proc Natl Acad Sci USA 1998 95 14447 14452 9826720 10.1073/pnas.95.24.14447
Loughran TP JrStarkebaum G Large granular lymphocyte leukemia. Report of 38 cases and review of the literature Medicine (Baltimore) 1987 66 397 405 3626848
Martens PB Goronzy JJ Schaid D Weyand CM Expansion of unusual CD4+ T cells in severe rheumatoid arthritis Arthritis Rheum 1997 40 1106 1114 9182921
Goronzy JJ Weyand CM Aging, autoimmunity and arthritis: T-cell senescence and contraction of T-cell repertoire diversity: catalysts of autoimmunity and chronic inflammation Arthritis Res Ther 2003 5 225 234 12932282 10.1186/ar974
Turesson C Matteson EL Colby TV Vuk-Pavlovic Z Vassallo R Weyand CM Tazelaar HD Limper AH Increased CD4+ T cell infiltrates in rheumatoid arthritis-associated interstitial pneumonitis compared with idiopathic interstitial pneumonitis Arthritis Rheum 2005 52 73 79 15641082 10.1002/art.20765
Voskuyl AE Zwinderman AH Westedt ML Vandenbroucke JP Breedveld FC Hazes JM Factors associated with the development of vasculitis in rheumatoid arthritis: results of a case-control study Ann Rheum Dis 1996 55 190 192 8712883
|
16277691
|
PMC1297586
|
CC BY
|
2021-01-04 16:02:40
|
no
|
Arthritis Res Ther. 2005 Oct 11; 7(6):R1386-R1393
|
utf-8
|
Arthritis Res Ther
| 2,005 |
10.1186/ar1837
|
oa_comm
|
==== Front
Arthritis Res TherArthritis Research & Therapy1478-63541478-6362BioMed Central London ar18381627769210.1186/ar1838Research ArticleAssociation between anti-nucleophosmin and anti-cardiolipin antibodies in (NZW × BXSB)F1 mice and human systemic lupus erythematosus Lartigue Aurelia 1Drouot Laurent 1Jouen Fabienne 2Charlionet Roland 1Tron François [email protected] Danièle 121 INSERM U519 and Institut Fédératif de Recherche Multidisciplinaire sur les Peptides, Faculté de Médecine et Pharmacie, 22 boulevard Gambetta, 76183 Rouen Cedex, France2 Laboratoire d'Immunopathologie Clinique et Expérimentale, CHU de Rouen, 1 rue de Germont, 76000, Rouen cedex, France2005 13 10 2005 7 6 R1394 R1403 8 6 2005 18 7 2005 11 8 2005 9 9 2005 Copyright © 2005 Lartigue 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.
We showed previously that nucleophosmin (NPM), a nucleolar phosphoprotein, is recognized by sera from (NZW × BXSB)F1 (WB) mice, a model of systemic lupus erythematosus (SLE) and anti-phospholipid syndrome. In the present study we analysed the prevalence and kinetics of anti-NPM autoantibodies in WB mice by a solid-phase ELISA with recombinant human (rh) NPM as the antigen and showed that most male WB mouse sera had anti-NPM antibodies that were responsible for their indirect immunofluorescence staining pattern on Hep-2 cells. Anti-NPM antibodies were significantly associated with anti-cardiolipin (aCL) antibodies. This antibody profile mirrored that observed in certain human SLE sera because anti-NPM antibodies were detected in 28% of the sera from patients with SLE and were similarly associated with aCL antibodies. The demonstration that rhNPM bound to cardiolipin (CL) in vitro and increased the CL-binding activity of a WB-derived aCL monoclonal antibody indicates that NPM can interact with CL to form SLE-related immunogenic particles that might be responsible for the concomitant production of anti-NPM and aCL antibodies.
==== Body
Introduction
(NZW × BXSB)F1 (WB) mice develop an autoimmune disease whose histological and immunological manifestations resemble those of human systemic lupus erythematosus (SLE) [1]. Male WB produce anti-nuclear antibodies (ANA), including anti-deoxyribonucleic acid (DNA) autoantibodies and high levels of anti-cardiolipin (aCL) antibodies that are thought to contribute to the pathogenesis of myocardial infarction and thrombocytopenia observed in these animals [2]. aCL antibodies present in male WB mice require a plasma cofactor such as β2-glycoprotein I (β2GPI), to bind to cardiolipin (CL) and thus possess binding properties similar to those of aCL antibodies observed in the serum of patients with SLE [3,4]. Male WB mice are therefore considered an appropriate model for the secondary anti-phospholipid syndrome associated with SLE.
The precise nature of epitopes recognized by β2GPI-dependent aCL antibodies remains a matter of debate. Some groups consider that aCL antibodies do not recognize CL or β2GPI alone but bind to either a CL-β2GPI complex [5] or cryptic epitopes generated by their association [6]. Others think that aCL antibodies bind to β2GPI in the absence of CL [7]. The complexity of the interaction of anti-phospholipid antibodies with their respective antigens is further illustrated by the demonstration that β2GPI is not the unique cofactor involved in their binding activity. Indeed, other phospholipid-binding proteins have been described, such as prothrombin, protein C, protein S or annexin V, most of which participate in the coagulation cascade [8,9].
We previously observed that WB mouse-derived monoclonal antibodies (mAbs) selected for their capacities to react with CL in the presence of FCS also reacted with nuclear antigens, as shown by their nucleolar immunofluorescence-labelling pattern on HEp-2 cells [10]. One of these mAbs, 4B7, reacted with nucleophosmin (NPM; also known as B23), a nucleolar protein involved in the assembly and transport of ribosomes [11]. Subsequently, we showed by immunoscreening of a two-dimensional (2D) PAGE-separated HL-60 cell protein map with WB mouse sera and mass spectrometry (MS) that the males of this strain mount an ordered autoimmune B-cell response directed against various antigens, consistently including NPM [12].
These observations led us to analyse further the prevalence and kinetics of anti-NPM antibodies, their relationships with aCL antibodies in WB mouse and human SLE sera and the mechanisms that might account for the association between these two antibody populations.
Materials and methods
Mice and sera
Female NZW and male BXSB mice were purchased, respectively, from Bomholtgard Breeding and Research Center (Ry, Denmark) and The Jackson Laboratory (Bar Harbor, ME, USA), maintained in our animal facilities and crossbred to obtain WB offspring. Control male CD1 mice were purchased from Charles River Laboratories (Saint-Germain-sur-l'Arbresle, France). All mice were housed in the same room and fed on the same diet. Mice were bled every two months throughout their lifetime. Sera were stored at -20°C until used. Sera from (NZB × NZW)F1 and MRLlpr/lpr mice were also tested. Mouse studies were approved by the animal Ethical Committee of Normandy (ceean number 1004-27).
Patients and sera
Serum samples were collected from 82 patients with SLE that met the criteria of the American College of Rheumatology. The serological profile of these patients was analysed for ANA by indirect immunofluorescence using HEp-2 cells, anti-DNA antibodies by the Farr assay and anti-β2GPI by ELISA (Varelisa β2-glycoprotein I IgG antibody EIA kit; Pharmacia Diagnostics, Freiburg, Germany). Serum samples were obtained from 103 healthy blood donors (agreement 31/10/2003). The study was approved by the Ethical Committee of Haute Normandie (number 99135HP).
Indirect immunofluorescence assay
The immunofluorescence pattern of sera from WB mice and patients with SLE was determined on Hep-2 cells as described previously [10]. Inhibition experiments were performed by preincubating an anti-NPM mouse mAb (Invitrogen immunodetection, San Francisco, CA, USA), mouse and patient sera for 1 hour with 20 μg of recombinant NPM.
Generation of mAb 4B7
Splenocytes from a four-month-old male WB mouse were fused with P3 × 63Ag8.653 myeloma cell line and the resulting hybridoma secreting mAb 4B7 was selected on the basis of the capacity of its supernatant to react with CL in ELISA. mAb 4B7 was purified as described elsewhere [10].
Production and purification of recombinant human NPM in Escherichia coli
Total RNA was extracted from HL-60 cells by using TRIzol reagent (Invitrogen, Eragny, France). The NPM cDNA was obtained by transcription of oligo(dT) primer RNA with Moloney murine leukaemia virus reverse transcriptase (Invitrogen). An 879-base-pair DNA fragment was amplified from the cDNA by PCR with the primers NPM-NdeI (5' -GGAATTCCATATGGAAGATTCGATGGACATGGAC-3', sense) and NPM-BamHI (5' -CCGGATCCTTACTTGTCATCGTCGTCCTTGTAGTCCGTACGAAGAGACTTCCTCCACTGCC-3', anti-sense) designed to create a FLAG tag followed by a stop codon and a BamHI site. The purified PCR product was cloned into Nde-BamHI-digested pET-15b (Novagen, Madison, WI, USA) downstream from the histidine tag. The resulting plasmid was transfected into E. coli BL21(DE3) pLysE (Novagen, Darmstadt, Germany). Expression was induced by incubation with isopropyl β-D-thiogalactoside (final concentration 1 mM) for 2.5 hours. E. coli pellets were suspended and lysed in lysis buffer (50 mM NaH2PO4, 300 mM NaCl, 10 mM imidazole, 1 mg/ml lysozyme and 1 mg/ml protease inhibitors, pH 8). The suspension was sonicated at 40% intensity for 2.5 minutes (Vibra Cell; Bioblock Scientific, Illkirch, France) and centrifuged at 7,000 g for 15 minutes at 4°C. The supernatant was subjected to repeated aspiration and expulsion through a fine needle to mechanically break DNA and then passed through 0.22 μm filter. Batch purification was performed with Ni2+-nitrilotriacetate (NTA)-Sepharose (Qiagen, Hilden, Germany). The resin was washed with buffer (50 mM NaH2PO4, 1 M NaCl, 20 mM imidazole, 20% (v/v) ethanol, pH 8) until UV absorbance at 280 nm became negative. The recombinant human nucleophosmin (rhNPM) was eluted with 50 mM NaH2PO4, 300 mM NaCl, 150 mM imidazole, pH 8. The rhNPM-enriched fractions were dialysed against water and then freeze-dried. This purification was performed for a second time with the same protocol. The protein concentration was determined with the Bradford protein assay kit (Bio-Rad Laboratories Inc., Marnes-la-Coquette, France) and the yield was about 2.5 mg per 150 ml of E. coli culture. Purity was controlled by SDS-PAGE analysis with a 10% polyacrylamide gel followed by western blotting with an anti-histidine-tag mAb (Sigma-Aldrich Corp., St Louis, MO, USA). rhNPM was used to detect anti-NPM antibodies, including those present in mouse sera, because they were previously shown to bind human NPM expressed by a human cell line [12] and there is 95% identity between human NPM and NO38, the murine equivalent of human NPM.
ELISA for anti-NPM autoantibodies
High-binding plates (96-well, Microlon; Dutscher, Issy-les-Moulineaux, France) were coated with 10 μg/ml purified rhNPM in 0.05 M carbonate-bicarbonate buffer (pH 9.5) and incubated overnight at 4°C. After being washed in PBS containing 0.05% Tween, wells were blocked with PBS containing 5% (w/v) BSA for 2 hours at 23°C. Plates were washed three times with 0.05% PBS-Tween, followed by the addition to duplicate wells of mouse or human of sera, diluted 1:50 or 1:100, respectively, in diluting buffer (1% (w/v) BSA-PBS), and then incubated for 2 hours at room temperature. After being washed, biotin-conjugated goat anti-mouse IgG (Caltag Laboratories, Hamburg, Germany) diluted 1:10,000 was added, and incubated for 1 hour at room temperature. After three washes, alkaline phosphatase-conjugated streptavidin (1:10,000 dilution; Caltag Laboratories) was added and incubated for 10 minutes at room temperature. After three washes, plates were revealed with p-nitrophenyl phosphate (Sigma). The absorbance at 405 nm (A405) was read. Horseradish peroxidase-conjugated goat anti-human IgG (Sigma) was added and revealed with 3,3',5,5'-tetramethylbenzidine (Sigma). A405 was read. Positive mouse sera were defined as those giving an A405 reading greater than the mean value plus 2 SD of sera from 60 normal control CD1 male mice. Positive human sera were defined as those giving an A405 reading greater than the mean value plus 3 SD of sera from 103 healthy controls.
ELISA for anti-CL antibodies
Each well of polystyrene microtitre plates (96-well; Luxlon, Nemours, France) was coated with 50 μl of bovine heart CL (10 μg/ml) in absolute alcohol. The plates were incubated overnight at 4°C to allow the ethanol to evaporate. After blocking of non-specific binding sites by incubation with 10% FCS, the plates were washed with PBS and incubated with sera diluted in FCS. The plates were washed three times followed by incubation with alkaline phosphatase-conjugated goat anti-mouse IgG (Rockland, Gilbertsville, PA, USA) or with horseradish peroxidase-conjugated goat anti-human IgG. Reactivity was determined as described above. Positive mouse sera were defined as above, whereas positive human sera were defined as those giving an A405 reading of more than 20 immunoglobulin G phospholipid international units (GPLU).
NPM binding to CL
To determine whether NPM reacts with CL, CL-coated plates were incubated with different rhNPM concentrations for 1 hour at room temperature. After blocking of non-specific binding sites with 0.5% gelatin, plates were washed three times with PBS. Anti-histidine or anti-FLAG (Sigma-Aldrich) mAb was added and incubated for 1 hour. After three washes, alkaline phosphatase-conjugated goat anti-mouse IgG was added and revealed with p-nitrophenyl phosphate.
Biacore analysis
The Biacore Biosensor system (Biacore, Uppsala, Sweden) was used to study the interaction between rhNPM and CL. Vesicles were prepared by drying 5 mg of CL under vacuum and hydrating the lipid in 1 ml of PBS as described previously [13]. The vesicles (400 μg/ml) were captured on the surface of sensor chip L1, consisting of dextran modified with lipophilic compounds. We successively injected into the system 10 μl of 20 mM CHAPS, 150 μl of vesicles and 10 μl of 10 mM NaOH to stabilize the baseline, yielding about 4,000 to 5,000 resonance units (RU) of bound antigen. Then different rhNPM concentrations (100 to 500 μg/ml) were injected and the sensor chip was regenerated with 20 mM CHAPS. Haemoglobin (Sigma) and human recombinant envoplakin produced in our laboratory were used as controls. In a second series of experiments, rhNPM was immobilized on the surface of sensor chip NTA, allowing the binding of histidine-tagged proteins. In brief, we injected 20 μl of 500 μM NiCl2 and then 40 μl of 200 μg/ml NPM, which yielded about 1,500 RU of antigen. The CL vesicles (400 μg/ml) were injected and the sensor chip was regenerated with 350 mM EDTA.
Effect of rhNPM on aCL-binding activity
To examine the effect of rhNPM on aCL-binding activity, purified 4B7 mAb (0.8 mg/ml) was incubated on CL-coated plates in the presence or absence of NPM, in an ELISA as described above.
Preparation of HL-60 cell-protein extract for 2D PAGE
Human promyelocytic leukaemia cell line (American Type Culture Collection, Manassas, VA, USA) was grown at 37°C in a humidified atmosphere (95% air, 5% CO2) in 50 ml of RPMI 1640 (Invitrogen, Eragny, France), supplemented with 10% FCS Sigma), antibiotics (Invitrogen) and 1 mM sodium pyruvate (Sigma). Cells were washed three times with PBS and isolated by centrifugation at 15,000 g for 5 minutes at room temperature. HL-60 cells (2 × 108) were suspended in 10% (v/v) trichloroacetic acid, 0.12% (w/v) dithiothreitol (DTT) and stored overnight at -20°C. Cells were centrifuged at 4°C for 30 minutes at 15,000 g. Cell pellets were suspended again in 0.12% (w/v) DTT and kept at -20°C for 1 hour before being centrifuged. Dry pellets were suspended in lysis buffer (9 M urea, 2% (w/v) CHAPS, 1% (w/v) DTT, 2% (v/v) protease inhibitors (Sigma)) and then centrifuged at 4°C for 20 minutes at 1,500 g. The protein content was determined with the PlusOne 2D Quant kit (Amersham, Buckinghamshire, UK).
This lysate was subjected to 2D PAGE as previously described [12]. The immunoreactive spots were detected with human sera, diluted 1:100. After being washed, membranes were incubated with alkaline phosphatase-conjugated goat anti-human IgG (Amersham) and revealed with Nitro Blue Tetrazolium salt and 5-bromo-4-chloroindol-3-yl phosphate substrate (Roche, Meylan, France).
Protein identification
The immunoreactive spots were excised from polyacrylamide gels with Ettan Spot Picker (Amersham) and digested by trypsin (proteomics grade; Sigma) with Ettan Digester (Amersham). Samples were analysed by matrix-assisted laser desorption/ionization-time-of-flight MS to obtain peptide mass information. Spectra obtained were compared with those registered in protein databases (SWISS-PROT and NCBInr). Data were matched against the databases with the use of the MS-Fit program (accessible through ProteinProspector).
Statistical analysis
Absorbances of sera obtained from male and female WB mice of the same age were compared with the Mann-Whitney U-test. The percentages of different antibodies in the sera of patients with SLE were compared by using a χ2 test. The relationship between the titres of different antibodies in mouse sera was evaluated with Spearman's correlation test.
Results
Male WB mouse sera frequently react with NPM
To determine the prevalence of anti-NPM antibodies, sera collected from normal CD1 (n = 54) and WB lupus-prone mice at different times of life were analysed by solid-phase ELISA using rhNPM. As shown in Fig. 1a, anti-NPM antibodies were present early in life in male WB mice and were detected in more than 75% of sera from animals more than three months old. Anti-NPM antibodies appeared later in female WB mice (25% at 3 months), were less frequent (40% at 4 to 6 months) and gave low A405 values. The A405 of male WB sera were significantly higher than those of control mouse sera at each period (P = 0.01) and tested dilution (Fig. 1b), whereas female WB A405 values differed significantly from normal mouse sera only at four months (P = 0.03). All anti-NPM antibody-positive mouse sera, tested by immunoblotting on 2D PAGE-separated HL-60 cell-protein map, bound to the native human NPM, thereby confirming their previously reported reactivity with this nucleolar autoantigen [12]. Anti-NPM positive mouse sera gave a nucleolar staining pattern on Hep-2 cells by indirect immunofluorescence analysis similar to that observed with an anti-NPM mouse mAb (Fig. 2a,c). Preincubation of the mAb and mouse sera with recombinant NPM abrogated the nucleolar staining (Fig. 2b,d).
NPM-binding activity in other mouse strains
We then tested WB parental strains, female NZW and male BXSB mice, but no reactivity against NPM was observed. Similarly, offspring of NZB and NZW mice also did not develop anti-NPM antibodies. In contrast, anti-NPM antibodies were present in 60% of 20 sera from MRLlpr/lpr mice more than three months old.
Anti-NPM and aCL antibodies are associated in male WB mice
Because WB lupus-prone mice are also characterized by aCL antibody production, we looked for an association between anti-NPM and aCL antibodies in their sera. Indeed, anti-NPM and aCL antibodies were found to be associated in sera from three-month-old male WB mice (χ2 = 18.14; P < 0.0001), and their A405 values were positively correlated (r = 0.750; P < 0.0001; Fig. 3a). Similarly, A405 values of anti-NPM and aCL antibodies in sera from MRLlpr/lpr mice more than three months old were positively correlated (r = 0.585; P < 0.01; Fig. 3b).
Anti-NPM antibodies are present in patients with SLE and are associated with aCL antibodies
The demonstration that anti-NPM antibodies are frequently produced in male WB mice prompted us to search for anti-NPM antibodies in the sera of 82 patients with SLE. Indeed, 23 (28%) of these sera were positive for anti-NPM antibody (Fig. 4) and the antibodies seemed to be more frequent (although not significantly) in males than in females (5/9 versus 18/73; Table 1). These anti-NPM antibody-positive SLE sera were analysed by immunoblotting on HL-60 cell protein separated by 2D PAGE (Fig. 5a); most of them (80%) reacted consistently with a spot of molecular mass (36 kDa) and pI (4.5) that had the same coordinates as that recognized by male WB mice and was characteristic of NPM (Fig. 5c). The immunoreactive spots bound by human and mouse sera were excised and analysed by matrix-assisted laser desorption/ionization-time-of-flight MS and corresponded to NPM (Fig. 5b, d).
Anti-NPM antibody-positive SLE sera yielded homogeneous nuclear and nucleolar staining on Hep-2 cells by indirect immunofluorescence analysis (Fig. 2e). The nucleolar staining disappeared when sera were preincubated with recombinant NPM (Fig. 2f).
aCL antibodies were detected in 32 (39%) of the sera from patients with SLE. Pertinently, as in WB mice, the analysis of the distribution of anti-NPM and aCL antibodies in these SLE sera (Table 1) indicates that these autoantibodies were associated (χ2 = 9.2; P = 0.002). In contrast, ANA, anti-DNA and anti-β2GPI antibodies rates did not differ significantly between anti-NPM-positive and anti-NPM-negative patients. Interestingly, of the 23 anti-NPM-positive sera, 18 (78%) did not react with β2GPI; conversely, of the 12 anti-β2GPI-aCL-positive sera, only three were positive for anti-NPM, suggesting that the presence of these two autoantibody populations is mutually exclusive.
NPM interacts in vitro with CL
The demonstration that anti-NPM and aCL antibodies were associated in WB mice and certain patients with SLE was reminiscent of the previously described association between anti-β2GPI and aCL antibodies; it led us to ask whether NPM could bind to CL. CL-coated plates were incubated with increasing concentrations of rhNPM, which was revealed with an anti-histidine or anti-FLAG mAb. Figure 6a shows that NPM bound to CL-coated wells in a dose-dependent manner. This binding was confirmed by Biacore analysis. CL vesicles were captured on the surface of sensor chip L1 and NPM was injected at a concentration of 400 μg/ml. Figure 7 shows that CL vesicles bound to NPM (Fig. 7a, with a difference of 3,550 RU) but did not bind to irrelevant proteins (haemoglobin and envoplakin; Fig. 7b). Similarly, NPM captured on an NTA sensor chip bound CL vesicles, giving a response of 2,885 RU (Fig. 7c).
NPM increases the CL binding of a murine mAb
The demonstration that NPM could bind to CL prompted us to study the effect of rhNPM on the CL-binding activity of purified 4B7, an aCL mAb derived from a WB mouse whose serum contained both aCL and anti-NPM antibodies [10]. Indeed, in ELISA, 4B7 reactivity with CL increased markedly in the presence of NPM, which therefore acted as a cofactor (Fig. 6b). Similarly, with the use of sensor chip L1 coated with CL vesicles or CL vesicles plus NPM, the binding of 4B7 to the chip was enhanced 1.5-fold (2,578 versus 3,989 RU; Fig. 8).
Discussion
NPM (B23) is an abundant nucleolar phosphoprotein with multiple functions: the assembly and/or transport of ribosomes [11], chaperone activities [14] and a regulatory role in cell proliferation [15,16]. NPM was previously shown to be targeted by antibodies produced in patients with either non-organ-specific autoimmune diseases [17-19] or cancer, namely hepatocellular and breast carcinoma [20,21].
By using a sensitive ELISA with rhNPM as the antigen, we showed that anti-NPM antibodies are present in most male WB lupus-prone mice and are therefore a constant feature of the antibody response in these animals: anti-NPM positivity appeared early in life, increased with age, and peaked at three to four months, before death. In contrast to our previous observations obtained by immunoblot analysis of 2D gel-separated NPM [12], this sensitive ELISA enabled us to detect these antibodies in female WB mice too, although later, less frequently and initially at lower A405 values. Pertinently, anti-NPM antibodies in male WB mice were consistently associated with aCL, and both antibody populations appeared concomitantly, as shown by sequential analyses of WB sera. Anti-NPM antibodies could also be detected in MRLlpr/lpr mice and again were significantly associated with aCL antibodies, which are frequently produced by this lupus mouse strain. This antibody pattern mirrored that observed in certain human SLE sera. Indeed, anti-NPM antibodies were present in 28% of our 82 patients with SLE; they were more frequent in males and were significantly associated with aCL antibodies. The association of anti-NPM and aCL antibodies was first suggested by Li and colleagues [17]. Indeed, in their analysis of 164 sera obtained from patients with various autoimmune diseases selected by the presence of anti-nucleolar antibodies, 6 had anti-NPM and aCL antibodies and SLE. However, our results, showing that anti-NPM antibodies are constantly detected in WB mice and frequently observed in patients with SLE, demonstrate that anti-NPM antibodies constitute a frequent and new marker in mouse and human lupus, establish a clear relationship between anti-NPM and aCL antibodies and finally define a subset of patients with aCL antibodies.
The association of aCL and anti-NPM antibodies in WB lupus-prone mice and patients with SLE might be explained by two different mechanisms: first, cross-reactivity due to the expression of a shared epitope by CL and NPM, or second, the ability of NPM to interact with CL to form an immunogenic complex able to induce the two antibody populations and/or a unique antibody population able to react with both NPM and CL (dual reactivity), as reported for lupus-related antigen particles [22-24]. The former hypothesis is not supported by the presence of anti-NPM-positive/aCL-negative sera and, conversely, aCL-positive/anti-NPM-negative sera and by the demonstration that anti-NPM antibodies induced by immunization of normal mice do not react with CL (data not shown); this conclusion was also reached by Li and colleagues [17], who showed that affinity-purified anti-NPM antibodies from SLE sera did not bind to CL. We therefore tested the second hypothesis, which proposed that lupus-related antigens are made of physically linked epitopes, such as DNA-histones [22] or Sm-DNA [23,24] and implies that NPM is able to bind to CL and behave as an aCL antibody cofactor. Plasmon resonance analysis of rhNPM-CL interaction clearly showed that NPM binds to CL in vitro to form complexes. This binding might be attributed to the functional N-terminal domain, which contains a high density of hydrophobic residues involved in chaperone activity [25]. The same technology enabled us to show that the reactivity of 4B7 mAb to NPM-CL complexes was markedly enhanced in comparison with its reactivity to either NPM or CL alone, suggesting that 4B7 is representative of mAbs exhibiting a dual specificity similar to that previously reported for certain anti-histone murine mAbs, whose binding activity is increased by DNA [22,26]. In human SLE sera, such autoantibodies probably exist but their identification remains elusive because of the polyclonal nature of sera, which may simultaneously contain monoreactive anti-NPM and anti-CL antibodies.
These results could also lead us to consider that NPM acts as an aCL cofactor, at least in this murine model of lupus. So far, several cofactors of anti-phospholipid antibodies have been described; most of them are soluble proteins involved in coagulation [27,28]. We have showed here that a nuclear autoantigen can behave as an aCL cofactor. This observation raises important questions: is NPM the unique nuclear autoantigen acting as an aCL cofactor? When and where does NPM interact with CL to form an immunogenic complex able to initiate antibody responses to both proteins? NPM is an abundant nuclear protein [29] that can translocate to the cytoplasm [30]. Recently, it was found to be localized in the cell membrane and to be a component of the fructose lysine-specific receptor expressed by monocyte-like cell lines [31]. Thus, NPM could well interact with anionic phospholipids at the cell membrane to form a typical lupus-related immunogenic complex, which could be released. Experiments are under way to determine the cellular localization of NPM in different categories of cells and during various cellular processes, such as apoptosis, which is thought to have a major role in the breakage of B cell tolerance in SLE [32]. The answer to these questions will help to explain the high frequency of anti-phospholipid antibodies in patients with SLE and to identify other nuclear autoantigens able to behave as an aCL cofactor. Another important objective is to precisely define the sensitivity and specificity of anti-NPM antibodies for lupus and to determine whether their presence is correlated with disease activity or clinical manifestations in patients with SLE. Such an analysis is under way in a large series of patients with SLE and various autoimmune diseases and will help to clarify the disease significance of this autoantibody population.
Conclusion
In this study we show that anti-NPM antibodies constitute a frequent marker in WB and MRLlpr/lpr lupus-prone mice and in human SLE, establish a clear relationship between anti-NPM and aCL antibodies and define a subset of patients with aCL antibodies. The demonstration that NPM binds to CL in vitro and increases the CL-binding activity of a WB-derived aCL mAb indicates that NPM can interact with CL to form SLE-related immunogenic particles that might be responsible for the concomitant production of anti-NPM and aCL antibodies.
Abbreviations
2D = two-dimensional; aCL = anti-cardiolipin; ANA = anti-nuclear antibodies; β2GPI = β2-glycoprotein I; BSA = bovine serum albumin; CL = cardiolipin; DTT = dithiothreitol; ELISA = enzyme-linked immunosorbent assay; FCS = fetal calf serum; mAb = monoclonal antibodies; MS = mass spectrometry; NPM = nucleophosmin; NTA = nitrilotriacetate; PBS = phosphate-buffered saline; PCR = polymerase chain reaction; rhNPM = recombinant human nucleophosmin; RU = resonance units; SLE = systemic lupus erythematosus; WB = (NZW × BXSB)F1.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
AL participated in the design of the study, performed the immunoanalysis and wrote the manuscript. LD participated in recombinant protein synthesis. FB provided sera from patients. RC performed mass spectrometry analysis. FT participated in the design of the study and wrote the manuscript. DG performed Biacore analysis, participated in the design of the study and wrote the manuscript. All authors read and approved the final manuscript.
Acknowledgements
This study was supported by INSERM.
Figures and Tables
Figure 1 Nucleophosmin-binding activity of WB lupus-prone and control mouse sera. (a) Sera from male (◆) and female (■) WB mice, and from male CD1 (▴) mice, collected during months 1 to 2, 2 to 3, 3 to 4 and 4 to 6, were tested in solid-phase ELISA. The horizontal dotted line represents the cutoff value (mean A405 of controls + 2 SD). Sera with A405 < 0.522 were considered negative. (* P = 0.03, ** P = 0.01, *** P = 0.0001; Mann-Whitney test). (b) Different dilutions of male (◆) and female (■) WB and male (▴) CD1 mouse sera collected at 3.5 months were tested in ELISA.
Figure 2 Immunofluorescence staining pattern of anti-nucleophosmin-positive WB mouse and SLE sera on HEp-2 cells. An anti-nucleophosmin (anti-NPM) mouse mAb and anti-NPM antibody-positive mouse sera gave a nucleolar pattern (a,c). Anti-NPM-antibody-positive systemic lupus erythematosus (SLE) sera yielded homogeneous nuclear and nucleolar staining (e). The nucleolar staining was abrogated when the mAb and sera from mouse (b,d) or patient (f) were previously incubated with 10 μg of recombinant human nucleophosmin.
Figure 3 Correlation between anti-nucleophosmin and anti-cardiolipin antibodies in lupus mice. A positive correlation between anti-nucleophosmin and anti-cardiolipin antibody reactivities in 38 sera from male WB mice (a) and 20 sera from MRLlpr/lpr mice (b) more than 3 months old.
Figure 4 Nucleophosmin-binding activity of sera from patients with SLE. (a) Sera from patients with systemic lupus erythematosus (SLE) (◆) or from healthy controls (-) were analysed by ELISA using recombinant human nucleophosmin. The dotted line represents the cutoff value (mean A405 of controls + 3 SD). Sera giving an A405 < 0.286 were considered negative. (b) Different dilutions of serum from an SLE patient (◆) and a control (■) were tested in ELISA.
Figure 5 Immunoblot analyses of sera positive for anti-nucleophosmin antibody. (a,c) Anti-nucleophosmin antibody obtained respectively from a patient with systemic lupus erythematosus (SLE) (a) and a male WB mouse (c) using a 2D PAGE-separated HL-60 cell protein map as the substrate. (b,d) Mass spectra of the proteins bound by the SLE (b) and mouse (d) serum.
Figure 6 Binding of rhNPM to cardiolipin and effects of nucleophosmin on cardiolipin-binding activity of 4B7 mAb. (a) Cardiolipin was coated on plastic plates and incubated with various concentrations of recombinant human nucleophosmin (rhNPM). (b) mAb 4B7 was incubated on cardiolipin-coated plates in the absence or presence of rhNPM. Bars represent the mean A405 of duplicate experiments and error bars represent SD.
Figure 7 Biacore analysis of recombinant human nucleophosmin (rhNPM)-cardiolipin interaction. (a,b) Cardiolipin vesicles were captured on the surface of the sensor chip L1 and rhNPM (a), haemoglobin or envoplakin (b) were injected. (c) rhNPM was immobilized on the surface of the Ni2+-nitrilotriacetate sensor chip and cardiolipin vesicles were injected.
Figure 8 Biacore analysis of binding of 4B7 mAb to cardiolipin-nucleophosmin complexes. Cardiolipin vesicles were captured on the surface of sensor chip L1. Binding of purified 4B7 mAb to cardiolipin vesicles alone (a) or after the injection of recombinant human nucleophosmin (b).
Table 1 Demographic and serological profiles of patients with SLE
Characteristic SLE patients Anti-NPM antibodies
(n = 82) Positive (n = 23) Negative (n = 59)
Age, years (mean ± SD) 38.3 ± 12.3 39.8 ± 14.9 37.8 ± 11.5
Females 73 (89.0) 18* 55
Males 9 (10.8) 5* 4
aCL 32 (39) 15 (65.2)** 17 (28.8)**
ANA-positive 66 (80.5) 23 (100) 43 (72.8)
Anti-DNA 46 (56) 13 (56.5) 33 (55.9)
Anti-β2GPI 15 (18.3) 5 (21.7) 10 (16.9)
Anti-β2GPI-CL+ 12 (14.6) 3 (13) 9 (15.2)
aCL, anti-cardiolipin; ANA, anti-nuclear antibodies; β2GPI, β2-glycoprotein I; CL, cardiolipin; NPM, nucleophosmin; SLE, systemic lupus erythematosus. Values are n (%) unless indicated otherwise.
* χ2 = 2.4; P = 0.12. ** χ2 = 9.2; P = 0.002.
==== Refs
Hang LM Izui S Dixon FJ (NZW × BXSB)F1 hybrid. A model of acute lupus and coronary vascular disease with myocardial infarction J Exp Med 1981 154 216 221 7252427 10.1084/jem.154.1.216
Koike T Anticardiolipin antibodies in NZW × BXSB F1 mice Lupus 1994 3 241 246 7804309
Hashimoto Y Kawamura M Ichikawa K Suzuki T Sumida T Yoshida S Matsuura E Ikehara S Koike T Anticardiolipin antibodies in NZW × BXSB F1 mice. A model of antiphospholipid syndrome J Immunol 1992 149 1063 1068 1634762
Gharavi AE Cucurull E Tang H Silver RM Branch DN Effect of antiphospholipid antibodies on beta(2) glycoprotein I-phospholipid Interaction Am J Reprod Immunol 1998 39 310 315 9602248
McNeil HP Simpson RJ Chesterman CN Krilis SA Anti-phospholipid antibodies are directed against a complex antigen that includes a lipid-binding inhibitor of coagulation: beta 2-glycoprotein I (apolipoprotein H) Proc Natl Acad Sci USA 1990 87 4120 4124 2349221
Pengo Y Biasiolo A Fior MG Autoimmune antiphospholipid antibodies are directed against a cryptic epitope expressed when beta 2-glycoprotein I is bound to a suitable surface Thromb Haemost 1995 73 29 34 7537919
Roubey RA Eisenberg RA Harper MF Winfield JB 'Anticardiolipin' autoantibodies recognize beta 2-glycoprotein I in the absence of phospholipid. Importance of Ag density and bivalent binding J Immunol 1995 154 954 960 7814895
Roubey RA Immunology of the antiphospholipid antibody syndrome Arthritis Rheum 1996 39 1444 1454 8814055
Alarcon-Segovia D Cabral AR The concept and classification of anti-phospholipid/cofactor syndromes Lupus 1996 5 364 367 8902761
Thébault S Gilbert D Machour N Marvin L Lange C Tron F Charlionet R Two-dimensional electrophoresis and mass spectrometry identification of proteins bound by a murine monoclonal anti-cardiolipin antibody: a powerful technique to characterize the cross-reactivity of a single autoantibody Electrophoresis 2000 21 2531 2539 10939468 10.1002/1522-2683(20000701)21:12<2531::AID-ELPS2531>3.0.CO;2-E
Savkur RS Olson MO Preferential cleavage in pre-ribosomal RNA by protein B23 endoribonuclease Nucleic Acids Res 1998 26 4508 4515 9742256 10.1093/nar/26.19.4508
Thébault S Gilbert D Hubert M Drouot L Machour N Lange C Tron F Orderly pattern of development of the autoantibody response in (New Zealand White × BXSB)F1 lupus mice: characterization of target antigens and antigen spreading by two-dimensional gel electrophoresis and mass spectrometry J Immunol 2002 169 4046 4053 12244208
Haruta K Kobayashi S Hirose S Horiai A Ohyanagi M Tanaka M Kawano T Shirai T Takasaki Y Hashimoto H Monoclonal anti-cardiolipin antibodies from New Zealand Black x New Zealand White F1 mice react to thrombomodulin J Immunol 1998 160 253 258 9551978
Okuwaki M Matsumoto K Tsujimoto M Nagata K Function of nucleophosmin/B23, a nucleolar acidic protein, as a histone chaperone FEBS Lett 2001 506 272 276 11602260 10.1016/S0014-5793(01)02939-8
Feuerstein N Chan PK Mond JJ Identification of numatrin, the nuclear matrix protein associated with induction of mitogenesis, as the nucleolar protein B23: implication for the role of the nucleolus in early transduction of mitogenic signals J Biol Chem 1988 263 10608 10612 3392030
Jiang PS Yung BY Down-regulation of nucleophosmin/B23 mRNA delays the entry of cells into mitosis Biochem Biophys Res Commun 1999 257 865 870 10208875 10.1006/bbrc.1999.0551
Li XZ McNeilage LJ Whittingham S Autoantibodies to the major nucleolar phosphoprotein B23 define a novel subset of patients with anticardiolipin antibodies Arthritis Rheum 1989 32 1165 1169 2789047
Pfeifle J Anderer FA Franke M Characterisation of nucleolar proteins as autoantigens using human autoimmune sera Ann Rheum Dis 1986 45 978 986 3545105
Ulanet DB Wigley FM Gelber AC Rosen A Autoantibodies against B23, a nucleolar phosphoprotein, occur in scleroderma and are associated with pulmonary hypertension Arthritis Rheum 2003 49 85 92 12579598 10.1002/art.10914
Imai H Ochs R Kiyosawa K Furuta S Nakamura R Tan EM Nucleolar antigens and autoantibodies in hepatocellular carcinoma and other malignancies Am J Pathol 1992 140 859 870 1314027
Brankin B Skaar TC Brotzman M Trock B Clarke R Autoantibodies to the nuclear phosphoprotein nucleophosmin in breast cancer patients Cancer Epidemiol Biomarkers Prev 1998 7 1109 1115 9865429
Losman MJ Fasy TM Novick KE Monestier M Monoclonal autoantibodies to subnucleosomes from a MRL/Mp-+/+ mouse. Oligoclonality of the antibody response and recognition of a determinant composed of histones H2A, H2B and DNA J Immunol 1992 148 1561 1569 1371530
Retter MW Eisenberg RA Cohen PL Clarke SH Sm and DNA binding by dual reactive B cells requires distinct VH, V kappa, and VH CDR3 structures J Immunol 1995 155 2248 2257 7636271
Retter MW Cohen PL Eisenberg RA Clarke SH Both Sm and DNA are selecting antigens in the anti-Sm B cell response in autoimmune MRL/lpr mice J Immunol 1996 156 1296 1306 8558010
Hingorani K Szebeni A Olson MOJ Mapping the functional domains of nucleolar protein B23 J Biol Chem 2000 275 24451 24457 10829026 10.1074/jbc.M003278200
Losman MJ Fasy TM Novick KE Monestier M Relationships among antinuclear antibodies from autoimmune MRL mice reacting with histone H2A-H2B dimers and DNA Int Immunol 1993 5 513 523 8318454
Galli M Beretta G Daldossi M Bevers EM Barbui T Different anticoagulant and immunological properties of anti-prothrombin antibodies in patients with antiphospholipid antibodies Thromb Haemost 1997 77 486 491 9065999
Galli M Comfurius P Barbui T Zwaal RF Bevers EM Anticoagulant activity of beta 2-glycoprotein I is potentiated by a distinct subgroup of anticardiolipin antibodies Thromb Haemost 1992 68 297 300 1440495
Pasquali JL Azerad G Martin T Muller S The double reactivity of a human monoclonal rheumatoid factor to IgG and histones is related to distinct binding sites Eur J Immunol 1988 18 1127 1130 3402531
Borer RA Lehner CF Eppenberger HM Nigg EA Major nucleolar proteins shuttle between nucleus and cytoplasm Cell 1989 56 379 390 2914325 10.1016/0092-8674(89)90241-9
Brandt R Nawka M Kellermann J Salazar R Becher D Krantz S Nucleophosmin is a component of the fructoselysine-specific receptor in cell membranes of Mono Mac 6 and U937 monocyte-like cells Biochim Biophys Acta 2004 1670 132 136 14738996
Casciola-Rosen L Rosen A Petri M Schlissel M Surface blebs on apoptotic cells are sites of enhanced procoagulant activity: implications for coagulation events and antigenic spread in systemic lupus erythematosus Proc Natl Acad Sci USA 1996 93 1624 1629 8643681 10.1073/pnas.93.4.1624
|
16277692
|
PMC1297587
|
CC BY
|
2021-01-04 16:02:40
|
no
|
Arthritis Res Ther. 2005 Oct 13; 7(6):R1394-R1403
|
utf-8
|
Arthritis Res Ther
| 2,005 |
10.1186/ar1838
|
oa_comm
|
==== Front
Arthritis Res TherArthritis Research & Therapy1478-63541478-6362BioMed Central London ar18391627769310.1186/ar1839Research ArticlePyridoxine supplementation corrects vitamin B6 deficiency but does not improve inflammation in patients with rheumatoid arthritis Chiang En-Pei I [email protected] Jacob [email protected] Pamela J 2Dallal Gerard [email protected] Ronenn [email protected] Department of Food Science and Biotechnology, National Chung-Hsing University, 250 Kuo-Kuang Road, Taichung, Taiwan 402, Republic of China2 Vitamin Metabolism Laboratory, Jean Mayer USDA Human Nutrition Research Center on Aging, Tufts University, 711 Washington Street, Boston, MA 02111, USA3 Biostatistics Unit, Jean Mayer USDA Human Nutrition Research Center on Aging, Tufts University, 711 Washington Street, Boston, MA 02111, USA4 Nutrition, Exercise Physiology, and Sarcopenia Laboratory Jean Mayer USDA Human Nutrition Research Center on Aging, Tufts University, 711 Washington Street, Boston, MA 02111, USA5 Tufts-New England Medical Center, 136 Harrison Avenue, Boston, MA 02111, USA2005 14 10 2005 7 6 R1404 R1411 16 6 2005 9 8 2005 6 9 2005 14 9 2005 Copyright © 2005 Chiang 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.
Patients with rheumatoid arthritis have subnormal vitamin B6 status, both quantitatively and functionally. Abnormal vitamin B6 status in rheumatoid arthritis has been associated with spontaneous tumor necrosis factor (TNF)-α production and markers of inflammation, including C-reactive protein and erythrocyte sedimentation rate. Impaired vitamin B6 status could be a result of inflammation, and these patients may have higher demand for vitamin B6. The aim of this study was to determine if daily supplementation with 50 mg of pyridoxine for 30 days can correct the static and/or the functional abnormalities of vitamin B6 status seen in patients with rheumatoid arthritis, and further investigate if pyridoxine supplementation has any effects on the pro-inflammatory cytokine TNF-α or IL-6 production of arthritis. This was a double-blinded, placebo-controlled study involving patients with rheumatoid arthritis with plasma pyridoxal 5'-phosphate below the 25th percentile of the Framingham Heart Cohort Study. Vitamin B6 status was assessed via plasma and erythrocyte pyridoxal 5'-phosphate concentrations, the erythrocyte aspartate aminotransferase activity coefficient (αEAST), net homocysteine increase in response to a methionine load test (ΔtHcy), and 24 h urinary xanthurenic acid (XA) excretion in response to a tryptophan load test. Urinary 4-pyridoxic acid (4-PA) was measured to examine the impact of pyridoxine treatment on vitamin B6 excretion in these patients. Pro-inflammatory cytokine (TNF-α and IL-6) production, C-reactive protein levels and the erythrocyte sedimentation rate before and after supplementation were also examined. Pyridoxine supplementation significantly improved plasma and erythrocyte pyridoxal 5'-phosphate concentrations, erythrocyte αEAST, urinary 4-PA, and XA excretion. These improvements were apparent regardless of baseline B6 levels. Pyridoxine supplementation also showed a trend (p < 0.09) towards a reduction in post-methionine load ΔtHcy. Supplementation did not affect pro-inflammatory cytokine production. Although pyridoxine supplementation did not suppress pro-inflammatory cytokine production in patients with rheumatoid arthritis, the suboptimal vitamin B6 status seen in rheumatoid arthritis can be corrected by 50 mg pyridoxine supplementation for 30 days. Data from the present study suggest that patients with rheumatoid arthritis may have higher requirements for vitamin B6 than those in a normal healthy population.
==== Body
Introduction
Patients with rheumatoid arthritis have reduced circulating levels of vitamin B6 compared to healthy subjects [1-3]. We have demonstrated that low plasma pyridoxal 5'-phosphate levels reflect the impaired functional vitamin B6 status in these patients. Plasma pyridoxal 5'-phosphate levels correlated with both the net homocysteine increase in response to a methionine load test (ΔtHcy) and 24 h urinary xanthurenic acid excretion in response to a tryptophan load test (XA) [4]. We also demonstrated that the inadequate vitamin B6 status seen in patients with rheumatoid arthritis was not due to insufficient dietary intake or excessive excretion, but was related to the inflammatory status of their underlying disease [4,5]. Abnormal vitamin B6 status in rheumatoid arthritis has been associated with spontaneous tumor necrosis factor (TNF)-α production [1] and markers of inflammation, including C-reactive protein (CRP) and erythrocyte sedimentation rate [5]. We recently showed that adjuvant arthritis caused tissue-specific depletion of vitamin B6 in rats [6], suggesting that the impaired vitamin B6 metabolism in patients with rheumatoid arthritis result from inflammation, and these patients may have higher requirements for vitamin B6 than those in a normal healthy population.
Vitamin B6 supplementation for patients with rheumatoid arthritis has been considered. Earlier studies reported that short-term pyridoxine treatment normalized tryptophan metabolism in patients with rheumatoid arthritis, but did not improve arthritis symptoms [7-9]. These studies were limited by small sample size, absence of placebo controls or blinding, and limited assessment of B6 metabolism, relying instead on pyridoxal 5'-phosphate levels, which are altered by inflammation itself. Furthermore, the cause of subnormal vitamin B6 status in rheumatoid arthritis remains to be determined and it is not known whether vitamin B6 supplementation improves functional vitamin B6 indices in these patients. The present study is the first one to systematically investigate the efficacy of vitamin B6 supplementation on static and functional vitamin B6 indices in patients with rheumatoid arthritis.
Although vitamin B6 supplementation appeared ineffective for symptom relief in rheumatoid arthritis, it should still be considered in these patients because of the potential adverse consequences of vitamin B6 insufficiency. Vitamin B6 deficiency in animals has been related to atherosclerotic lesions [10]. More recently, researchers demonstrated a relationship between vitamin B6 deficiency and atherosclerosis in human population-based studies, and they reported that this relationship was independent of plasma total homocysteine (tHcy) levels both before and after methionine loading [11,12]. Furthermore, vitamin B6 deficiency is associated with post-methionine load hyperhomocysteinemia, another known independent risk factor for cardiovascular disease [13-15]. We previously reported that patients with rheumatoid arthritis have mild but significantly elevated ΔtHcy in response to methionine load compared to age- and gender-matched healthy controls [2,16]. This led us to evaluate the efficacy of giving vitamin B6 supplements to rheumatoid arthritis patients with respect to decreasing the elevated ΔtHcy and improve functional vitamin B6 status. The goal of the present study was to investigate whether treatment with 50 mg pyridoxine for 30 days improves static and functional indices of vitamin B6 status in patients with rheumatoid arthritis.
Materials and methods
Study population
Thirty six adults with rheumatoid arthritis were recruited through the Tufts New England Medical Center (NEMC) Rheumatology Clinic as previously described [5]. Written informed consent was obtained from all subjects in accordance with the regulations of the NEMC/Tufts University Human Investigation Review Committee. Briefly, men and women over 18 years old fulfilling the American College of Rheumatology criteria for rheumatoid arthritis were eligible [17]. Patients with pregnancy, oral contraceptive use, anemia (hemoglobin ≤ 10 mg/dl), thrombocytopenia (platelet count ≤ 50,000/ul), abnormal liver transaminase (serum aspartate aminotransferase or alanine aminotransferase ≥ 50 IU/l), renal insufficiency (serum creatinine ≥ 1.5 mg/dl), diabetes, or cancer were excluded. Patients taking supplements containing vitamin B6 were asked to stop for ≥ 1 month before their participation in the study.
Study protocol
This double-blinded, randomized and placebo controlled trial was conducted in the General Clinical Research Center (GCRC) at Tufts-NEMC. Prior to enrollment, blood screening and urinalysis were performed to ensure qualification and to identify individuals with low circulating vitamin B6 for the study. To test the efficacy of vitamin B6 supplementation on those patients with reduced plasma pyridoxal 5'-phosphate, baseline (phase 1) vitamin B6 status was determined using a two day test procedure as follows. Patients taking methotrexate were asked to come at least 24 h after their weekly dose of this drug. On the first day of the evaluation (day 1), each subject arrived in the GCRC at 8 a.m. after having eaten breakfast. Each subject received a standard oral tryptophan load test (5 g powdered L-tryptophan dissolved in chocolate milk; Ajinomoto, Teaneck, NJ, USA) and collected urine for the next 24 h. The urine was kept refrigerated without additives during the collection period. Separate 24 h urine collection was done in the week prior to day 1 for the measurement of baseline XA and 4-pyridoxic acid (4-PA) excretion.
Subjects were asked to fast overnight starting at 8 p.m. on day 1 for the methionine load test next morning. After completion of the 24 h urine collection in the morning of day 2, each subject received a standard methionine load test [18]. Baseline fasting blood was drawn in a tube containing ethylenediaminetetraacetic acid (EDTA) (Becton Dickinson, Franklin Lakes, NJ, USA) for determination of plasma pyridoxal 5'-phosphate, fasting tHcy level, erythrocyte pyridoxal 5'-phosphate concentration, erythrocyte aspartate aminotransferase activity (EAST), and CRP concentrations. Aliquots were also collected for routine hematology and chemistry analyses. Peripheral blood mononuclear cells (PBMC) were collected from heparinized blood and isolated by Ficoll-Hypaque centrifugation, then washed and cultured for 24 h in 96-well flat-bottom plates with ultrafiltered, pyrogen-free RPMI 1640 medium (Sigma, St. Louis, MO, USA) that was supplemented with 100 μg/ml streptomycin and 100 U/ml penicillin, with 1% autologous heat-inactivated pooled serum and 1% L-glutamine. After incubation, plates were then frozen at -80°C until assay.
After collection of fasting blood on day 2, each patient was then given a standard oral methionine load test (100 mg/kg body weight powdered methionine dissolved in orange juice; Ajinomoto, Teaneck, NJ, USA). Blood was drawn 4 h after the methionine load for determination of the post-load tHcy level. Fasting plasma pyridoxal 5'-phosphate levels were determined within 1 week and the level was compared to the 25th percentile of the Framingham Offspring Heart Cohort [19]. Patients with a plasma pyridoxal 5'-phosphate level within the lowest quartile of the appropriate age and gender Framingham population (cycle 6, offspring group) were recruited for the supplementation phase of the study (phase 2). The 25th percentile cutoff for plasma pyridoxal 5'-phosphate in women below 55 years and women at or above 55 years were 33.7 and 37.5 nmol/l, respectively. For men below 55 years and for men at or above 55 years it was 49.7 and 35.6 nmol/l, respectively [19].
Study interventions
Qualified subjects started taking the study treatment within one week of plasma pyridoxal 5'-phosphate analysis. These subjects were randomly assigned through the NEMC pharmacy to receive either active vitamin B6 (B6 group) or placebo (placebo group) tablets in double-blinded fashion for 30 days. To minimize the potential confounding effect of methotrexate and prednisone treatment on the functional tests, subjects were stratified by prednisone and methotrexate treatment, and then the subjects in each group were randomized to receive either active or placebo treatment. The randomization procedure was under guidance of a statistician and performed by registered pharmacists not directly involved in the present study.
The placebo tablet, made specifically for the study, was identical in appearance as well as ingredients to the active tablet, except that the active tablet (Nutro Laboratories, South Plainfield, NJ, USA) contained 50 mg of pyridoxine hydrochloride and the placebo did not (Tishcon Corp., Westbury, NY, USA). Both tablets contained microcrystalline cellulose, croscarmellose sodium, calcium phosphate, stearic acid, and magnesium stearate, ingredients commonly found in over-the-counter vitamin B6 supplements. Each phase 2 participant was asked to take one assigned tablet daily throughout the 30 day period. To assure compliance with the treatment regimen, each subject was given a personal study calendar with the 30 supplement days highlighted. The subject was asked to record the time of ingestion of each tablet on the calendar. In addition, the study coordinator made phone calls to remind each subject to take the tablets during the 30 day supplement period. The subjects were asked to return the bottle for a tablet count at the end of the 30 day treatment. To test the efficacy of the vitamin B6 supplementation, each subject went through the same 2 day testing procedure described above at the end of the 30 day supplementation period.
Laboratory analyses
Blood hematology and chemistry analyses and urinalysis were performed at the Clinical Laboratory of NEMC, Boston, MA. CRP concentrations were determined by enzyme immunoassay kit (Virgo CRP150 kit, Hemagen, Waltham, MA, USA). Pyridoxal 5'-phosphate concentration was assayed by the tyrosine decarboxylase enzymatic procedure of Camp et al. [20] with a modification of the extraction procedure for plasma and erythrocytes. The modification is described as follows: a 20 μl plasma aliquot was precipitated with 4 volumes of 5% trichloroacetic acid for deproteinization. Erythrocytes were washed with 0.9% saline 3 times and the freshly washed erythrocytes were extracted with an equal volume of 10% (w/v) perchloroacetic acid. After centrifugation, the supernatants were stored at -70°C until the analysis. The erythrocyte pyridoxal 5'-phosphate results were expressed as nmol/l of packed erythrocyte at a hemotocrit of 100%. Fasting and post-methionine load plasma tHcy concentrations [21] and 4-PA [22] were determined by high performance liquid chromatography (HPLC) using a Hitachi L-7100 intelligent pump connected to an L-7400 UV detector (Hitachi, Tokyo, Japan). Baseline and post-tryptophan load urinary XA were measured by a colorimetric method [23]. EAST activity was measured using the Cobas Fara II Centrifugal Analyzer (Roche Dianostic system Inc., Nutley, NJ, USA) [24]. The ratio of pyridoxal 5'-phosphate saturated and unsaturated enzyme is expressed as the activity coefficient αEAST. Plasma TNF-α concentrations and PBMC TNF-α and IL-6 production was assayed with the commercially available quantitative enzyme immunoassays (Quantikine, R&D Systems, Minneapolis, MN, USA). Total PBMC cytokine production was measured in unstimulated cells (spontaneous production).
Statistical analysis
Differences in means between the baseline indices of the active group versus the placebo group were evaluated by Student's t-tests to examine if the randomization was successful. Differences were considered significant if the two-tailed p-value was <0.05. Plasma pyridoxal 5'-phosphate, tHcy, urinary XA, and 4-PA levels were log-transformed to achieve normality. Analysis of covariance (ANCOVA) was used to test the treatment effect of pyridoxine. The model was adjusted for the baseline (phase 1) value. A Pearson's correlation coefficient was calculated to examine the relationship between plasma pyridoxal 5'-phosphate levels and the inflammatory marker CRP before and after the treatment period. All statistical analyses were performed using Systat 10.0 for Windows ™ (SPSS, Chicago, IL, USA).
Results
Thirty-six patients with rheumatoid arthritis who met the eligibility requirements for the study were recruited for phase 1 of the study. Three patients dropped out because of scheduling problems or due to the concern over ingestion of the methionine and/or tryptophan. Of the 33 patients who completed the phase 1 procedure, 28 patients (85%) were found to have plasma pyridoxal 5'-phosphate levels within the lowest quartile of the age- and gender-matched population of the Framingham Offspring Study and thus qualified for the supplementation phase (Table 1). The number of pills consumed by each participant during the treatment period was divided by the total number of pills supplied to each subject (n = 30). The average percentage of pill consumption and the standard deviation in each group was calculated. Based on the tablet counts after the completion of the study, the compliance of treatment regimen was 97.8 ± 6.3 (%) for the B6 group and 98.3 ± 5.2 (%) for the placebo group. Baseline characteristics in the B6 and the placebo groups were comparable, indicating that randomization was appropriate (Table 1).
Indicators of vitamin B6 status before and after treatment are shown in Table 2. All markers of vitamin B6 status improved significantly in the B6 group after supplementation, except for net increase in total homocysteine concentration, which only showed a modest trend towards improvement. None of the vitamin B6 status parameters showed significant improvements after treatment in the placebo group. Analysis of co-variance further demonstrated that initial levels of plasma pyridoxal 5'-phosphate, ΔtHcy, post-load urinary XA, αEAST (Table 2) and CRP and erythrocyte sedimentation rate (ESR) (Table 3) in phase 1 were strong predictors of those indicators after treatment in phase 2. After adjusting for the initial levels in (before treatment), the vitamin B6 supplementation significantly improved plasma and erythrocyte pyridoxal 5'-phosphate concentrations, αEAST, post-load XA and 24 h 4-PA excretion. We found a trend for normalization of plasma ΔtHcy in the vitamin B6 treatment versus placebo group (p = 0.086, ANCOVA). In patients who had abnormal ΔtHcy (above 15 μmol/l) before treatment (n = 22/28), the effect of vitamin B6 treatment was significant (p < 0.02). Plasma pyridoxal 5'-phosphate and ΔtHcy levels were related to CRP in patients with rheumatoid arthritis [5], thus CRP could be a potential targets for vitamin B6 supplementation. The correlations between CRP and plasma pyridoxal 5'-phosphate and ΔtHcy disappeared in the B6 group after supplementation, whereas the relationships remained in the placebo group after the 30 day treatment (CRP versus plasma pyridoxal 5'-phosphate, r = -0.61, p = 0.02; CRP versus ΔtHcy, r = 0.48, p = 0.098) (n = 14). We found that vitamin B6 supplementation had no effect on inflammatory cytokines, plasma CRP, ESR, or rheumatoid factor levels in these patients (Table 3).
Discussion
Abnormal vitamin B6 metabolism has been reported in rheumatoid arthritis for decades [3,7,8,25,26]. Considering the close associations between vitamin B6 indices and the clinical and biochemical inflammatory markers [5], it is likely that inflammation causes vitamin B6 deficiency, yet it is also possible that impaired vitamin B6 status contributes to more severe inflammation in these patients. The present study demonstrates that 50 mg of pyridoxine hydrochloride supplementation for 1 month can significantly improve vitamin B6 status in patients with rheumatoid arthritis regardless of the etiology of such inadequacy. In contrast, vitamin B6 supplementation was ineffective in suppressing inflammatory cytokine production, or reducing ESR, plasma CRP or rheumatoid factor levels in these patients. As improving vitamin B6 status did not alleviate inflammation, it is unlikely that vitamin B6 inadequacy directly causes or worsens the inflammatory condition. We suggest that the impaired vitamin B6 metabolism in patients with rheumatoid arthritis results from inflammation, and these patients may be in higher demand of vitamin B6 to cope with the ongoing inflammatory condition. In the present study, 85% (28/33) of our participants had a plasma pyridoxal 5'-phosphate concentration below the 25th percentile of the Framingham population data at baseline. We previously reported the presence of functional vitamin B6 inadequacy in rheumatoid arthritis patients: ΔtHcy after a methionine load test was significantly higher in patients with rheumatoid arthritis than healthy matched controls, indicating that impaired trans-sulfuration accompanied the low plasma vitamin B6 levels [2,16].
Both clinical studies and animal experiments suggest that inflammation causes tissue specific depletion of vitamin B6 [4,5,16]. It is not clear how different tissues may respond to vitamin B6 supplementation during inflammation. While patients with rheumatoid arthritis have abnormal systemic functional status of vitamin B6 (as measured by ΔtHcy level in response to a methionine load test), they appear to have normal functional vitamin B6 status specifically in the erythrocytes (as measured by αEAST) [16]. Previously, we demonstrated that vitamin B6 status in erythrocytes is more sensitive to dietary vitamin B6 intake compared to plasma pyridoxal 5'-phosphate concentration or functional indices, including ΔtHcy and XA excretion in patients with rheumatoid arthritis [4]. Based on this observation, we expected αEAST to be more responsive to vitamin B6 supplementation compared to the methionine load test. The results from the present study support our speculation. All subjects in the B6 group had improvements in αEAST, including those patients who had a normal initial αEAST before supplementation. After supplementation, the mean reduction in αEAST was 32% of the original αEAST, and all individuals after the supplementation had an αEAST in the desirable range (αEAST ≤ 1.5) suggested by Leklem [27]. Individuals in the placebo group had no significant change in αEAST. In conclusion, erythrocyte αEAST reflects vitamin B6 intake rather than systemic B6 functional status, and is more sensitive to vitamin B6 supplementation in these patients.
Post-tryptophan load XA excretion above 146.2 μmol/day (30 mg/day) was considered as the cutoff for inadequacy in healthy volunteers after ingestion of 5 g of L-tryptophan [28]. In our phase 1 screening, 19 of the 28 patients had post-tryptophan XA excretion levels above this threshold. After 30 days of vitamin B6 treatment, 13 of the 14 patients in the B6 treated group had normal levels of post-tryptophan load XA excretion, whereas only 2 of the patients with abnormal XA in the placebo group fell in the 'adequate range' after treatment. Our results suggest that pyridoxine treatment can normalize tryptophan metabolism in those patients with abnormal tryptophan metabolism.
With respect to different indicators for functional vitamin B6 status, we found that the effect of the pyridoxine treatment on the response to a methionine load test was not as strong as αEAST or post-load XA. There was a mild vitamin B6 treatment effect after adjusting for initial ΔtHcy in phase 1 (ANCOVA, p = 0.086). Twenty-five percent (7/28) of these patients had a 'normal' ΔtHcy (below 15 μmol/l) before treatment, which might account for the overall modest treatment effect of pyridoxine in our study.
Subnormal vitamin B6 status has also been shown in some asthma patients. It was reported that vitamin B6 supplementation (20 mg/day) for 6 weeks significantly reduced post-methionine load ΔtHcy in asthma patients with low vitamin B6 status, but it had no significant effect in controls with normal ΔtHcy response [29]. The initial ΔtHcy level in our present study is comparable with those in the above study (pre-supplementation mean ± SD = 23.9 ± 11.3 μmol/l), and we also found a rather modest effect of vitamin B6 supplementation on ΔtHcy in our participants with rheumatoid arthritis. In the present study, there was a significant treatment effect of vitamin B6 supplementation (p = 0.022) in subjects with an initial ΔtHcy level above 15 μmol/l, suggesting that there may be a threshold effect of pyridoxine on ΔtHcy levels. Treatment with vitamin B6 may only lower ΔtHcy in individuals who start with elevated ΔtHcy levels. Conversely, the disrupted homocysteine metabolism may not be simply due to vitamin B6 inadequacy in these patients as 2 of the 14 subjects in the B6 group with abnormal initial ΔtHcy still had a similarly abnormal response to methionine load after the 30 day vitamin B6 supplementation (ΔtHcy > 30 nmol/l). As elevated ΔtHcy was found to be associated with enhanced disease activity in these patients [5], we suggest that alleviating the disease activity of rheumatoid arthritis with medication may help correct the abnormal methionine load outcomes. Further studies are warranted to study whether suppressing inflammation improves vitamin B6 status. It is also possible that factors other than vitamin B6 status, such as heterozygosity or deficiency of cystathionine β-synthase, may be responsible for the elevated ΔtHcy in these individuals. In this case, vitamin B6 supplementation alone may or may not be sufficient to correct the abnormal outcomes in response to a methionine load test.
We previously demonstrated the potential interfering effect of methotrexate on the methionine load test [2]. In addition, Bruckner et al. [25] demonstrated that drug therapy such as corticosteroids may have effects on tryptophan metabolism. We therefore performed block randomization to improve comparability between the B6 and the placebo group and minimize potential confounding effects of medication use. There was no difference in weight, height, age, methotrexate or prednisone dose, duration of disease, number of painful/swollen joints, The Health Assessment Questionnaire (HAQ) disability score, erythrocyte sedimentation rate, or rheumatoid factor between the placebo and B6 group at baseline, indicating that our strategy of randomization of treatment was effective.
Conclusion
Vitamin B6 supplementation is effective in improving static and functional vitamin B6 status regardless of the etiology of the vitamin B6 deficiency in patients with rheumatoid arthritis with plasma pyridoxal 5'-phosphate below the 25th percentile of the Framingham Heart Cohort Study. All static measurements of vitamin B6 status, including plasma and erythrocyte pyridoxal 5'-phosphate, αEAST, urinary XA excretion in response to a tryptophan load test, and 24 h 4-PA excretion, were significantly improved by the 30 day vitamin B6 treatment. We suggest that vitamin B6 supplementation should be considered in rheumatoid arthritis patients to improve vitamin B6 status, and to reduce the potential adverse consequences of B6 vitamin deficiency. In light of the potential benefits of improving B6 status in patients with rheumatoid arthritis, further studies should be conducted to determine the optimal dose that maximizes the biochemical as well as functional indices reflecting vitamin B6 therapy.
Abbreviations
4-PA = 4-pyridoxic acid; αEAST = erythrocyte aspartate aminotransferase activity coefficient; CRP = C-reactive protein; ΔtHcy = net homocysteine increase in response to a methionine load test; EAST = erythrocyte aspartate aminotransferase; ESR = erythrocyte sedimentation rate; GCRC = General Clinical Research Center; NEMC = New England Medical Center; PBMC = peripheral blood mononuclear cells; tHcy = plasma total homocysteine; TNF = tumor necrosis factor; XA = 24 h urinary xanthurenic acid excretion in response to a tryptophan load test.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
All authors made substantive intellectual contributions to the present study. E-PC conceived of the study, acquired partial funding, and carried out the human experiments, including study designs, coordination, biochemical analyses, data acquisition, analysis and interpretation, and drafted the manuscript. JS participated in the design of the study, acquisition of funding, and was involved in revising the manuscript critically for important intellectual content. GED participated in the design of the study and performed the statistical analysis. RR conceived of the study, acquired funding, and performed all clinical assessments in study subjects, and revised the manuscript critically for important intellectual content.
Acknowledgements
The authors thank Bernadette Muldoon RN, Karin Kohin, and Sarah Olson for their assistance in recruiting, the staff in Nutrition Evaluation Laboratory and the Tufts NEMC Clinical Laboratory for various analyses, the Tufts NEMC research pharmacy for randomization of the treatments, and the GCRC nurse staff for assistance with the study procedure. This study would not have been completed without their generous assistance. This project has been supported in part by a grant from the National Science Council of Taiwan (Grant # NSC 94-2320-B005-009; to E-PC). E-PC was also a recipient of a Dissertation Award from the Arthritis Foundation in the US. This project was also supported by the US Department of Agriculture under cooperative agreement no. 58-1950-9-001. Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the authors and do not necessarily reflect the view of the US Department of Agriculture. This study was also supported in part by grant RR-00054 from the National Center for Research Resources, for the General Clinical Research Center, New England Medical Center and Tufts University School of Medicine (RR).
Figures and Tables
Table 1 Description of subjects
Placebo group (n = 14) Vitamin B6 group (n = 14)
Age 57.5 (11.0) 53.9 (12.6)
Sex (F:M) 9:5 12:2
Height (cm) 168.4 (10.3) 164.7 (9.1)
Methotrexate (yes/all) 7/14 8/14
Methotrexate dose (mg/week) 7.5 (10.1) 10.2 (11.7)
Prednisone (yes/all) 9/14 11/14
Prednisone dose (mg/week) 3.1 (3.5) 4.3 (4.0)
NSAIDs use (yes/all) 10/14 11/14
Duration of disease (years) 11.6 (8.2) 8.5 (5.6)
Number of painful joints 5.1 (5.1) 7.9 (8.9)
Number of swollen joints 8.9 (8.7) 8.0 (9.5)
The Health Assessment Questionnaire disability score 1–3 scale 1.45 (1.18) 1.17 (0.94)
Erythrocyte sedimentation rate 30.2 (21.4) 36.0 (29.9)
Rheumatoid factor (IU/ml) 87.2 (69.2) 88.8 (82.9)
Albumin (g/dl) 3.8 (0.5) 3.4 (0.4)
Alkaline phosphatase (IU/l) 76.6 (14.5) 73.6 (23.6)
24 h creatinine (mg/dl) 1.02 (0.38) 1.01 (0.42)
C-reactive protein (mg/l) 16.7 (16.2) 8.6 (12.7)
Values represent mean (SD). NSAIDs, non steroidal anti-inflammatory drugs.
Table 2 Measurements of vitamin B6 status before and after 30 day treatment
Placebo group (n = 14) B6 group (n = 14) p value (baseline)a p value (treat)b
Before After Before After
Plasma PLP (nmol/l)c 22.8 (15.4–31.5) 23.6 (15.2–43.0) 27.0 (20.4–30.9) 144.5 (84.5–236.7) <0.0001 <0.0001
Erythrocyte PLP (nmol/l) 26.0 (20.8–39.4) 41.6 (28.5–53.7) 44.6 (37.5–54.0) 116.4 (65.3–424.7) 0.623 0.002
αEAST 1.88 (1.67–1.99) 1.85 (1.64–1.96) 1.80 (1.68–1.93) 1.33 (1.29–1.40) 0.001 <0.0001
ΔtHcy (μmol/l)c 19.2 (15.0–27.5) 17.9 (13.0–25.8) 24.9 (16.4–35.9) 19.0 (15.5–28.7) <0.0001 0.086
Post-load XA (μmol/24 h)c 173 (132–243) 137 (103–354) 183 (30–653) 102 (39–371) 0.001 0.042
4-PA (μg/24 h) 0.7 (0.5–1.2) 0.8 (0.5–170) 0.8 (0.5–2.0) 4.2 (0.8–12.8) 0.338 <0.0001
Data are presented as median (95% CI). aEffects of each baseline (before treatment) value on its post-treatment outcome. bTreatment effects of placebo and vitamin B6 were examined by analysis of covariance after adjusting for baseline value. cPlasma pyridoxal 5'-phosphate (PLP), urinary xanthurenic acid excretion in response to a tryptophan load test (post-load XA), and plasma total homocysteine (tHcy) concentrations were log-transformed to reach normal distribution for statistical analyses. αEAST, erythrocyte aspartate aminotransferase activity coefficient; 4-PA, 24 h 4-pyridoxic acid excretion; ΔtHcy, net homocysteine increase in response to a methionine load test.
Table 3 Inflammatory cytokines, C-reactive protein, erythrocyte sedimentation rate, and rheumatoid factor before and after 30 day treatment
Placebo group (n = 14) B6 group (n = 14) p value (baseline)a p value (treat)b
Before After Before After
PBMC IL-6 (pg/ml)c 490 (289–832) 1,369 (202–1,665) 1,112 (437–1,352) 1,476 (918–1,602) 0.698 0.315
PBMC TNF-α (ng/ml)d 224.6 (118.4–361.8) 341.5 (242.6–654.1) 114.1 (319.1–89.2) 178.7 (59.6–391.0) 0.320 0.963
Serum TNF-α (pg/ml) 1.7 (0.7–3.8) 2.1(0.3–5.5) 1.5 (0.9–2.7) 2.0 (0.9–3.6) 0.134 0.166
Serum CRP (mg/l) 13.0 (5.90–27.6) 7.0 (4.4–27.5) 2.0 (0.1–17.2) 3.0 (0.6–14.8) 0.387 <0.0001
ESR 31.0 (19.4–52.6) 32.0 (24.0–49.7) 27.5 (18.8–41.6) 31.0 (22.4–38.9) 0.425 <0.0001
RF (IU/ml) 72.0 (43.3–131.2) 93.8 (37.1–132.5) 76.4 (47.5–130.0) 73.8 (47.3–122.8) 0.697 <0.0001
Data are presented as median (95% CI). aEffects of each baseline (before treatment) value on its post-treatment outcome. bTreatment effects (placebo versus vitamin B6) were examined by analysis of covariance, adjusting for baseline (before) value. cSpontaneous production of IL-6 by peripheral blood mononuclear cells (PBMCs). dSpontaneous production of tumor necrosis factor (TNF)-α by PBMCs. CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; RF, rheumatoid factor.
==== Refs
Roubenoff R Roubenoff RA Selhub J Nadeau MR Cannon JG Freeman LM Dinarello CA Rosenberg IH Abnormal vitamin B6 status in rheumatoid cachexia. Association with spontaneous tumor necrosis factor alpha production and markers of inflammation Arthritis Rheum 1995 38 105 109 7818558
Roubenoff R Dellaripa P Nadeau MR Abad LW Muldoon BA Selhub J Rosenberg IH Abnormal homocysteine metabolism in rheumatoid arthritis Arthritis Rheum 1997 40 718 722 9125255
Sanderson CR Davis RE Bayliss CE Serum pyridoxal in patients with rheumatoid arthritis Ann Rheum Dis 1976 35 177 180 942275
Chiang EP Bagley PJ Roubenoff R Nadeau M Selhub J Plasma pyridoxal 5'-phosphate concentration is correlated with functional vitamin B-6 indices in patients with rheumatoid arthritis and marginal vitamin B-6 status J Nutr 2003 133 1056 1059 12672918
Chiang EP Bagley PJ Selhub J Nadeau M Roubenoff R Abnormal vitamin B(6) status is associated with severity of symptoms in patients with rheumatoid arthritis Am J Med 2003 114 283 287 12681455 10.1016/S0002-9343(02)01528-0
Chiang EP Smith DE Selhub J Dallal G Wang YC Roubenoff R Inflammation causes tissue-specific depletion of vitamin B6 Arthritis Res Ther 2005 7 R1254 R1262 16277678 10.1186/ar1821
Flinn JH Price JM Yess N Brown RR Excretion of tryptophan metabolites by patients with rheumatoid arthritis Arthritis Rheum 1964 7 201 210 14168449
Mckusick AB Sherwin RW Jones LG Hsu JM Urinary excretion of pyridoxine and 4-pyridoxic acid in rheumatoid arthritis Arthritis Rheum 1964 7 636 653 14240622
Jaffe IA Altman K The effect of pyridoxine on abnormal tryptophan metabolism in rheumatoid arthritis Arthritis Rheum 1964 7 319
Rinehart JF Greenberg LD Vitamin B6 deficiency in the rhesus monkey; with particular reference to the occurrence of atherosclerosis, dental caries, and hepatic cirrhosis Am J Clin Nutr 1956 4 318 325 13339702
Stampfer MJ Malinow MR Willett WC Newcomer LM Upson B Ullmann D Tishler PV Hennekens CH A prospective study of plasma homocyst(e)ine and risk of myocardial infarction in US physicians JAMA 1992 268 877 881 1640615 10.1001/jama.268.7.877
Verhoef P Meleady R Daly LE Graham IM Robinson K Boers GH Homocysteine, vitamin status and risk of vascular disease; effects of gender and menopausal status. European COMAC Group Eur Heart J 1999 20 1234 1244 10454975 10.1053/euhj.1999.1522
Clarke R Daly L Robinson K Naughten E Cahalane S Fowler B Graham I Hyperhomocysteinemia: an independent risk factor for vascular disease N Engl J Med 1991 324 1149 1155 2011158
McCully KS Vascular pathology of homocysteinemia: implications for the pathogenesis of arteriosclerosis Am J Pathol 1969 56 111 128 5792556
Malinow MR Plasma homocyst(e)ine: a risk factor for arterial occlusive diseases J Nutr 1996 126 4 Suppl 1238S 1243S 8642463
Chiang EP Vitamin B6 status and homocysteine metabolism in patients with rheumatoid arthritis PhD thesis 2001 Tufts University, School of Nutrition Science and Policy
Arnett FC Edworthy SM Bloch DA McShane DJ Fries JF Cooper NS Healey LA Kaplan SR Liang MH Luthra HS The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis Arthritis Rheum 1988 31 315 324 3358796
Bostom AG Roubenoff R Dellaripa P Nadeau MR Sutherland P Wilson PW Jacques PF Selhub J Rosenberg IH Validation of abbreviated oral methionine-loading test Clin Chem 1995 41 948 949 7768022
Selhub J Jacques PF Bostom AG Wilson PW Rosenberg IH Relationship between plasma homocysteine and vitamin status in the Framingham study population. Impact of folic acid fortification Public Health Rev 2000 28 117 145 11411265
Camp VM Chipponi J Faraj BA Radioenzymatic assay for direct measurement of plasma pyridoxal 5'-phosphate Clin Chem 1983 29 642 644 6831691
Araki A Sako Y Determination of free and total homocysteine in human plasma by high-performance liquid chromatography with fluorescence detection J Chromatogr 1987 422 43 52 3437026
Gregory JF 3rdKirk JR Determination of urinary 4-pyridoxic acid using high performance liquid chromatography Am J Clin Nutr 1979 32 879 883 433815
Hoes MJ Kreutzer EK Sijben N Xanthurenic acid excretion in urine after oral intake of 5 grams L-tryptophan by healthy volunteers: standardisation of the reference values J Clin Chem Clin Biochem 1981 19 259 264 7264566
Williams DG Methods for the estimation of three vitamin dependent red cell enzymes Clin Biochem 1976 9 252 255 1000784
Bruckner FE Smith HG Lakatos C Chamberlain MA Tryptophan metabolism in rheumatoid neuropathies Ann Rheum Dis 1972 31 311 315 5045910
Bett IM Metabolism of tryptophan in rheumatoid arthritis Ann Rheum Dis 1962 21 63 69 13868642
Leklem JE Vitamin B-6: a status report J Nutr 1990 120 Suppl 11 1503 1507 2243296
Tillostoon JA Sauberlish HE Baker EM Canham JE Use of carbon 14 labeled vitamins in human nutrition studies: pyridoxine Proceedings of the 7th International Congress Nutrition 1966 5 554
Ubbink JB van der Merwe A Delport R Allen RH Stabler SP Riezler R Vermaak WJ The effect of a subnormal vitamin B-6 status on homocysteine metabolism J Clin Invest 1996 98 177 184 8690790
|
16277693
|
PMC1297588
|
CC BY
|
2021-01-04 16:02:40
|
no
|
Arthritis Res Ther. 2005 Oct 14; 7(6):R1404-R1411
|
utf-8
|
Arthritis Res Ther
| 2,005 |
10.1186/ar1839
|
oa_comm
|
==== Front
Arthritis Res TherArthritis Research & Therapy1478-63541478-6362BioMed Central London ar18401627769410.1186/ar1840Research ArticleBetween adaptive and innate immunity: TLR4-mediated perforin production by CD28null T-helper cells in ankylosing spondylitis Raffeiner Bernd [email protected] Christian [email protected] Christina [email protected] Werner [email protected] Christian [email protected] Sandra C [email protected] Michael [email protected] Beatrix [email protected] Michael [email protected] Department of Internal Medicine, Innsbruck Medical University, Austria2 Ludwig Boltzmann Institute for Rehabilitation of Internal Diseases, Saalfelden, Austria3 Institute for Biomedical Aging Research, Austrian Academy of Science, Austria2005 18 10 2005 7 6 R1412 R1420 21 4 2005 14 6 2005 26 8 2005 27 9 2005 Copyright © 2005 Raffeiner 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.
CD3+CD4+CD28null and CD3+CD8+CD28null T cells are enriched in patients with immune-mediated diseases compared with healthy controls. This study shows that CD4+CD28null T cells express Toll-like receptors recognizing bacterial lipopolysaccharides in ankylosing spondylitis, psoriatic arthritis and rheumatoid arthritis. In ankylosing spondylitis, TLR4 (23.1 ± 21.9%) and, to a smaller extent, TLR2 (4.1 ± 5.8%) were expressed on CD4+CD28null T cells, whereas expression was negligible on CD4+CD28+ and CD8+ T cells. CD4+CD28null T cells produced perforin upon stimulation with lipopolysaccharide, and this effect was enhanced by autologous serum or recombinant soluble CD14. Perforin production could be prevented with blocking antibodies directed against CD14 or TLR4. Incubation of peripheral blood mononuclear cells with tumour necrosis factor alpha led to an upregulation of TLR4 and TLR2 on CD4+CD28null T cells in vitro, and treatment of patients with antibodies specifically directed against tumour necrosis factor alpha resulted in decreased expression of TLR4 and TLR2 on CD4+CD28null T cells in vivo. We describe here a new pathway for direct activation of cytotoxic CD4+ T cells by components of infectious pathogens. This finding supports the hypothesis that CD4+CD28null T cells represent an immunological link between the innate immune system and the adaptive immune system.
==== Body
Introduction
Pattern recognition receptors (PRRs) are a family of receptors of the innate immune system binding to conserved pathogen-associated molecular patterns [1]. The most important PRRs are the Toll-like receptors (TLRs), which allow monocytes, neutrophils, dendritic cells, natural killer (NK) cells and B cells to recognize bacterial components, viruses, fungi and host material such as heat shock proteins [2-5]. Receptor engagement leads to the translocation of NF-κB and to gene transcription of proinflammatory cytokines. Lipopolysaccharide (LPS) from Gram-negative bacteria is the main ligand for TLR4. LPS binding to TLR4 is promoted by CD14, which can be present either in a membrane-bound form (mCD14) or a soluble form (sCD14) [6,7]. Whether LPS is a low-affinity ligand for TLR2 is still controversial [8].
The interactions between the innate and adaptive immune systems are crucial to promote proinflammatory reactions against pathogens and to ensure maintenance of vital self-tolerance. TLRs are expressed on both innate and adaptive immune cells and are critically involved in this interplay. TLR-stimulated dendritic cells induce specific T cells to differentiate into memory cells [9,10], and microbial induction of the TLR pathway on dendritic cells also blocks the suppressive effects of regulatory T cells [11,12]. In addition, TLRs are themselves expressed on T cells. In murine T cells, LPS signalling induces production of INF-γ and T-helper-1 accentuated immune responses [13,14]. After in vitro activation, mouse CD4+ T cells express TLR3 and TLR9, and treatment of these cells with synthetic ligands for TLR3 and TLR9, viral dsRNA and bacterial unmethylated DNA enhances their survival [15]. LPS can directly activate regulatory T cells, and can thereby increase their suppressive function [16]. Activated human T cells express high levels of cell-surface TLR2 and produce elevated levels of cytokines in response to bacterial lipopeptide, a TLR2 ligand [17].
TLRs have also been shown to be important for the pathogenesis of immune-mediated diseases. In rheumatoid arthritis (RA), for example, TLR2 expression occurs in inflamed synovial tissue predominantly at sites of attachment and invasion into the cartilage and bone [18]. The TLR2-mediated stimulation of synovial fibroblasts with bacterial components promotes the release of proinflammatory cytokines and leads to a higher expression of TLR2.
In ankylosing spondylitis (AS), as in other immune-mediated diseases, an unusual proinflammatory and cytotoxic CD4+ T-cell subgroup has been described, which lacks the co-stimulatory molecule CD28. These CD4+CD28null T cells depend on alternative pathways for coactivation and can indeed obtain such signals by NK receptors recognizing ubiquitous major histocompatibility complex class I molecules [19-23]. Anomalous expression of NKG2D on these T cells together with upregulated MIC ligands in the inflamed synovial tissue of RA have also been shown to provide co-stimulatory signals [24]. CD4+CD28null T cells have therefore been considered an immunological link between the adaptive and the innate defence system [25].
As CD4+CD28null T cells have NK cell features, and as NK cells express TLRs on their surface [4], we investigated the expression of TLRs as an alternative stimulatory pathway on CD4+CD28null T cells.
Patients and methods
A total of 90 consecutive patients with spondyloarthropathies, 72 patients with RA and 64 age-matched healthy controls were enrolled into the study. Spondyloarthropathy was defined according to the European Spondyloarthropathy Study Group criteria [26]. Out of the spondyloarthropathy-defined patients, 65 patients had a diagnosis of AS according to the modified New York criteria [27] and 25 patients had psoriatic arthritis (PsA) as defined by the diagnostic criteria of Moll and Wright [28]. RA was diagnosed according to the criteria of the American College of Rheumatology [29]. Probands did not show any history, clinical or laboratory sign for infections nor malignant diseases. Healthy controls had no history of an immune-mediated disease. Heparinized blood samples were drawn from peripheral veins after informed and written consent according to the local ethics committee.
Patients' characteristics (including age, sex, the presence of rheumatoid factor and anti-cyclic citrullinated peptide antibodies, HLA-B27 status, axial involvement, erythrocyte sedimentation rate and C-reactive protein) are summarized in Table 1. Nine of the AS patients were treated with antibodies directed against tumour necrosis factor alpha (TNF-α) (infliximab = Remicade®; Aesca, Vienna, Austria) at a dosage of 3 mg/kg body weight.
Cell preparation and surface staining
Peripheral blood mononuclear cells (PBMCs) were isolated by Ficoll density gradient centrifugation. For surface staining, PBMCs were incubated, as appropriate, with fluorescein isothiocyanate-conjugated anti-CD4, anti-CD8 or anti-CD28, with phycoerythrin-conjugated anti-CD28 and with peridinin chlorophyll protein-conjugated anti-CD3, anti-CD4 and anti-CD8 monoclonal antibodies (Becton Dickinson, San Diego, CA, USA). Specific mAbs directed against CD14 (fluorescein isothiocyanate), TLR4 and TLR2 (phycoerythrin; eBioscience, San Diego, CA, USA) were used to analyse expression of PRRs. Cells were incubated for 30 minutes at 4°C with the antibodies. After washing with PBS, cells were fixed in 4% paraformaldehyde (cellfix; Becton Dickinson). Stained cells were analysed on a FACS-Calibur analyser (Becton Dickinson). At least 100,000 events were counted for each acquisition. Data were analysed using WinMDI software (version 2.5, Joseph Trotter; Scripps Research Institute, La Jolla, CA, USA). Cells were considered positively stained when fluorescence levels were higher than those of the corresponding isotype controls.
For the isolation of CD4+CD28+ and CD4+CD28null T cells, the MACS® CD4+ T-cell multisort kit and magnetic bead labelled anti-phycoerythrin antibodies were applied according to the manufacturer's instructions (Miltenyi Biotech, Amsterdam, The Netherlands). To further increase the purity of T cells, sorted fractions were incubated in 24-well plates for 2 hours at 37°C to allow adherence of contaminating monocytes. Purity of isolated fractions was determined by flow cytometry as already described.
Reverse transcription-polymerase chain reaction
Total RNA was extracted from isolated cell fractions using 1 ml Tri-Reagent (Sigma-Aldrich, St Louis, MO, USA), 200 μl chloroform (Sigma-Aldrich), 0.5 ml isopropanol (Sigma-Aldrich) and 1 μl glycogen (Roche, Basel, Switzerland) as previously described [30]. RNA was reverse transcribed to cDNA applying the Reverse Transcription System (Promega, Madison, WI, USA). PCR amplification was performed on 1 μg total RNA and 10 pmol primers (TLR2: forward, 5'-GGCCAGCAAATTACCTGTGTG-3'; reverse, 5'-CTGAGCCTCGTCCATGGGCCACTCC-3'; TLR4: forward, 5'-TGCAATGGATCAAGGACCAGAGGC-3'; reverse, 5'-GTGCTGGGACACCACAACAATCACC-3'; and β2-microglobulin: forward, 5'-CTCCGTGGCCTTAGCTGTG-3'; reverse, 5'-TTTGGAGTACGCTGGATAGCC-3') using the HotStar Master Mix Kit (Qiagen Valencia, CA, USA). The PCR reaction was performed according to a modified protocol [31] on the PTC-100 Thermal Cycler (MJ Research, Waltham, MA, USA) at 95°C for 3 minutes, at 95°C for 1 minute, 58°C for 2 minute and 72°C for 1 minute (35 cycles), and at 72°C for 2 minutes. Negative control samples were prepared by amplification reactions in the absence of cDNA. PCR products were separated on 1% agarose gel containing ethidium bromide and were visualized by UV illumination.
LPS studies and intracellular cytokine staining
Isolated PBMCs were resuspended in RPMI 1640 containing 2 mM L-glutamine and were distributed on 24-well plates at a density of 1 × 106 cells per well. LPS of Escherichia coli serotype 026:B6 (Sigma) was added at a concentration of 10 μg/ml [16,17], sCD14 (R&D systems, Minneapolis, MN, USA) at 25 μg/ml and autologous serum at a concentration of 5% as indicated.
To test inhibitory effects of blocking anti-CD14 or anti-TLR4 antibodies, cells were resuspended in RPMI 1640 with 5% autologous serum and were incubated with 10 μg/ml blocking anti-CD14 antibody (R&D Systems), 10 μg/ml anti-TLR4 antibody (Torrey Pines Biolabs, TX, USA) and 10 μg/ml isotype control antibody (Becton Dickinson) for 1 hour before adding LPS at a concentration of 10 μg/ml. Brefeldin A (Sigma) was added to functional experiments at a concentration of 10 μg/ml to avoid release of produced cytokines. After incubation for 16 hours at 37°C the cells were washed, stained with CD4-peridinin chlorophyll protein mAbs and CD28-phycoerythrin mAbs and were permeabilized with 0.05% Tween 20 to stain intracellularly with fluorescein isothiocyanate-conjugated anti-perforin mAbs or control immunoglobulin (Becton Dickinson).
Stimulation of short-term cell lines
Short-term cell lines were established after incubation of fresh PBMCs on stimulating immobilized anti-CD3 mAbs (OKT3; eBioscience) for 18 hours. Cells were then cultured in densities of 0.5 × 106 – 2 × 106 in RPMI 1640 containing 10% FCS, 2 mM L-glutamine and 20 U/ml recombinant human IL-2 (Sigma). The medium was changed every 2–3 days. Cell lines were used after 7 days for stimulation assays over 24 hours with 20 ng/ml recombinant human TNF-α (Prepotech, London, UK).
Enzyme-linked immunosorbent assay
Soluble levels of CD14 in blinded sera were tested in duplicates with an ELISA kit (R&D systems) according to the manufacturer's instructions.
Statistical analysis
Statistical analysis was performed using the SPSS program (version 11.5; SPSS Inc., Chicago, IL, USA). The Kolmogorov–Smirnov test was used to test for normal distribution, and the Mann–Whitney U test and the Wilcoxon test were used as appropriate. At least six assays were performed for each experiment. P < 0.05 was considered significant and P < 0.01 was highly significant. Data are shown as box plots with the lines within the boxes representing the median, the boxes representing the 25th–75th percentiles and the lines outside the boxes including all values except mavericks.
Results
Comparison between the prevalences of circulating CD4+CD28null and CD8+CD28null T cells in AS, PsA, RA and healthy controls
Peripheral blood from patients with AS, PsA and RA showed an increased prevalence of CD3+CD4+CD28null T cells and CD3+CD8+CD28null T cells compared with age-matched healthy controls. One representative example of a three-colour flow cytometry analysis showing the prevalence of CD3+CD4+CD28null and CD3+CD8+CD28null T cells is shown in Figure 1a. The mean percentages of CD3+CD4+CD28null T cells out of CD3+CD4+ T cells were 5.1 ± 9.8%, 5.1 ± 6.8%, 4.6 ± 5.2% and 1.5 ± 4.5% for AS, PsA, RA and healthy controls (each with P < 0.001), respectively. The mean percentages of CD3+CD8+CD28null T cells out of CD3+CD8+ T cells were 38.4 ± 21.9% in AS, 44.4 ± 21.7% in PsA and 46 ± 26.2% in RA, compared with 22.3 ± 12.4% in the control group (each with P < 0.001, Figure 1b).
Increased expression of pattern recognition receptors on CD4+CD28null T cells
To test T cells for the expression of TLR2 and TLR4 mRNA, monocyte-depleted CD4+CD28null and CD4+CD28+ T cells were isolated from patients as well as from healthy controls (CD4+CD28+ T cells). As shown in Figure 2a, purity was high for both with the CD3+CD4+CD28null and CD3+CD4+CD28+ fractions ranging between 94.2% and 99.7%. A representative example out of three independent RT-PCR experiments is depicted in Figure 2b: CD4+CD28null T cells express TLR2 and TLR4 mRNA, whereas TLR4 transcripts were not detected in CD4+CD28+ T cells from patients and healthy controls. In contrast, variable but significant levels of TLR2 mRNA were present in the CD3+CD4+CD28+ population even from healthy individuals.
To address PRR surface expression on T cells, PBMCs from patients as well as from healthy controls were incubated with mAbs against CD4, CD14, CD28, TLR4 and TLR2, and were analysed by flow cytometry. As shown in Figure 3a, lymphocytes were gated on the forward and side scatter and specific gates were set to focus CD4+CD28null and CD4+CD28+ T cells for the analysis of TLR expression.
Overall, all analysed PRRs (CD14, TLR4 and TLR2) were detected on the surface of CD4+CD28null T cells but not on CD28+ T cells irrespective of the underlying diseases tested. CD14 was expressed on 13.3 ± 20.4% of CD4+CD28null T cells versus 0.7 ± 1% of CD4+CD28+ T cells in AS, on 8.8 ± 15.7% of CD4+CD28null T cells versus 1.0 ± 2.3% of CD4+CD28+ T cells in PsA, and on 11.3 ± 17.2% of CD4+CD28null T cells versus 0.6 ± 0.6% of CD4+CD28+ T cells in RA (each with P < 0.001). TLR-4, the main receptor for LPS recognition, was significantly expressed on CD4+CD28null T cells in AS (23.1 ± 21.9% versus 0.9 ± 1.2%), in PsA (12.4 ± 18.1% versus 0.4 ± 0.5%) and in RA (23.1 ± 24.7% versus 0.6 ± 0.9%; each with P < 0.001). TLR2 was more frequently expressed on CD4+CD28null T cells than on CD28+ cells (4.1 ± 5.8% versus 1.0 ± 1.1%, P < 0.001) in AS, but not in PsA or in RA (Figure 3b). Negligible surface expression of PRRs without any difference between CD28null and CD28+ T cells were seen on CD8+ T cells from patients (data not shown). CD4+CD28+, CD8+CD28+ and CD8+CD28null T cells from healthy controls showed no significant expression of PRRs. CD4+CD28null T cells from healthy controls expressed PRR to some extent but, as the prevalence of this subpopulation is low, no significant data could be acquired (data not shown).
LPS-mediated perforin production of CD4+CD28null T cells depends on CD14 and TLR4
For functional testing of TLR-mediated lymphocytic stimulation, fresh PBMCs were incubated for 16 hours with LPS and the T cells were analysed for their intracellular production of perforin. As shown in Figure 4a,b, perforin was produced upon LPS stimulation by CD4+CD28null T cells (13 ± 10.7% perforin+ cells), but perforin expression was negligible in CD4+CD28+ T cells (0.4% ± 0.3% perforin+ cells, P = 0.009). Combining recombinant sCD14 with LPS doubled the percentage of perforin+CD4+CD28null T cells (24 ± 15.3% perforin+ cells) compared with LPS stimulation alone (P = 0.0001). A comparable additional effect was seen when cultures stimulated with LPS were supplemented with 5% autologous serum (26.9 ± 16.6% versus 13 ± 10.7% perforin+ cells, P = 0.001) but not with FCS (data not shown). CD4+CD28+ T cells did not produce perforin after co-incubation with LPS, even after addition of sCD14 or autologous serum (Figure 4a and data not shown).
These findings implicated the possible occurrence of natural sCD14 in sera from AS patients. Blinded samples from patients with AS and from healthy controls were therefore analysed using enzyme-linked immunoassays. As shown in Figure 5a, the levels of sCD14 were higher in AS patients than in healthy controls (1,653.6 ± 463 pg/ml versus 1,170 ± 259 pg/ml, P = 0.008). To investigate whether sCD14 from autologous serum was crucial for a stronger response of CD4+CD28null T cells to LPS, cells were incubated with a blocking antibody directed against CD14 prior to the addition of LPS. This antibody is capable of binding both sCD14 and mCD14. As expected, LPS-induced perforin production of CD4+CD28null T cells was reversed by blocking sCD14 and mCD14 (24.0 ± 16.3% versus 3.1 ± 2.5% perforin+ cells, P = 0.006) but not by isotype control antibody (Figure 5b).
Blocking assays with antibodies directed against TLR4 were then performed to specifically address the role of TLR4 in LPS-mediated perforin production of CD4+CD28null T cells. As shown in Figure 5c, preincubation with anti-TLR4 antibody inhibited activation of CD4+CD28null T cells by LPS (3.4 ± 2.3% with anti-TLR4 antibody versus 23.8 ± 14.7% perforin+ cells with isotype control antibody, P = 0.002).
Effects of TNF-α in vitro and therapeutic blockade of TNF-α in vivo on PRR expression of CD4+CD28null T cells
In vitro assays were performed to test the effect of TNF-α on the expression of PRRs. Incubation of PBMCs with TNF-α for 24 hours increased the expression of TLR4 and TLR2, but not of CD14 on CD4+CD28null T cells (from 9.2 ± 25.8% to 26.6 ± 27.7% for TLR4, P < 0.001 and from 1.1 ± 3.1% to 2.4 ± 4.1% for TLR2, P = 0.008; Figure 6a). Expression of CD14, TLR4 and TLR2 were neither induced on CD4+CD28+ T cells nor on CD8+ T cells after incubation with TNF-α (data not shown).
To examine the effects of TNF-α blocking treatment on the expression of PRRs on CD4+CD28null T cells in vivo, peripheral CD4+CD28null T cells from AS patients were tested before and after successful treatment with TNF-α-specific chimeric antibodies. As shown in Figure 6b, CD4+CD28null T cells from patients during active AS disease showed higher levels of PRRs than after successful treatment with the TNF-α blocking agent. CD14 was reduced from 10.6 ± 16.6% before treatment to 2.5 ± 2% after treatment (P = 0.011), TLR4 was reduced from 46.9 ± 32.7% to 11.7 ± 12.5% (P = 0.008) and TLR2 was reduced from 11.7 ± 19.4% to 1.8 ± 2.9% (P = 0.012).
Discussion
The present study shows an increased expression of PRRs on human circulating CD4+ T cells lacking the CD28 co-stimulatory molecule. TLR4 can be considered an alternative signalling pathway for cytotoxic CD4+CD28null T cells, but neither for their CD28+ counterparts nor for CD8+ T cells. The concomitant expression of T-cell receptor (TCR) and PRRs on the cell surface further supports the role of CD4+CD28null T cells as an immunological link between the adaptive and the innate defence system, and is in accordance with earlier descriptions of co-existing NK receptors and TCR on these cells [25]. In all chronic immune diseases tested (AS, PsA and RA) more CD4+CD28null T cells expressed TLR4 than TLR2, thus stressing the superior role of TLR4 over TLR2. Indeed, a significant surface expression of TLR2 on CD4+CD28null T cells has only been found in patients with AS, but not in patients with PsA and RA (Figure 1b), which is consistent with an earlier histological study in RA that did not find TLR2 on CD3+ T cells in the synovial tissue [32].
To assure a high purity of T-cell populations was a critical part in this study. T cells were therefore not only purified by MACS® technology for the RT-PCR assays, but were also monocyte depleted. A high purity of CD3+CD4+CD28null cells ranging from 94.2% up to 99.1% was thus obtained, as shown in Figure 2a. In a separate approach, surface expression of TLRs was studied by fluorescence-activated cell sorting analysis with gates carefully set to focus on the lymphocyte population on the forward scatter and the side scatter. An additional gate was then set on the population expressing high levels of CD4, ensuring monocytes that express lower levels of CD4 were excluded [33]. Detection of mRNA and surface expression of TLRs were therefore used as two independent techniques to ensure the presence of TLRs in the examined cell populations.
From the functional perspective, TLR4 is the key receptor for Gram-negative bacteria. In our in vitro model the effect of bacterial exposure on the cytotoxic function of CD4+CD28null T cells was simulated by addition of LPS from an E. coli strain. TLR4 binds LPS and thus provides activating signals to the CD4+CD28null T cells, which can be reversed with TLR4 blocking antibodies (Figure 5c). This mechanism clearly depends on CD14, which allows signal transmission [34]. CD14 in AS may either occur on cell membranes of CD4+CD28null T cells (Figure 3b) or as a soluble molecule in the serum (Figure 5a). As the percentages of CD4+CD28null T cells expressing mCD14 were much lower than the percentages of cells expressing TLR4, we added either recombinant CD14 or autologous sera for stimulation assays (Figure 4a,b). Indeed, addition of CD14 nearly doubled the percentage of perforin+CD4+CD28null T cells upon LPS stimulation, whereas the addition of anti-CD14 antibody completely abolished the effect of LPS (Figure 5b). LPS binding protein, another TLR-related molecule, is also known to support binding of LPS to TLR4. Serum levels of LPS binding protein correlate with inflammation in RA and reactive arthritis [35], but have not so far been studied in AS. In our experiments both the addition of recombinant sCD14 and autologous serum had a comparable additional effect on LPS-induced perforin production of CD4+CD28null T cells, which indicates serum LPS binding protein not to be indispensable in AS. Taking these facts together, activated CD4+CD28null T cells produce perforin upon LPS-mediated activation in a CD14-dependent and TLR4-dependent manner.
As we used PBMCs for functional assays, we cannot exclude that LPS also activated antigen-presenting cells within the PBMCs. However, direct LPS-mediated effects on TLR4+ T cells appear more relevant: TLR- T cells were not activated in the presence of LPS, and the percentage of perforin+CD4+CD28null T cells correlated well with the prevalence of TLR4-expressing CD4+CD28null T cells (data not shown). Antigen-presenting cells would not need addition of CD14 for activation by LPS anyway, as CD14 is widely expressed on antigen-presenting cells.
Direct TLR-mediated activation of human T cells has been previously shown for activated CD8+ T cells and CD4+CD45RO+ memory T cells from healthy individuals with high surface levels of TLR2, but not TLR4 [17]. Although a number of CD4+CD28+ T cells express the memory marker CD45RO (data not shown), we did not detect TLR2 on these cells. A possible explanation for the discrepancy with our results may be that the mAbs used recognize different epitopes or variants of TLRs. We showed that expression of mRNA for TLR2 was present in both CD4+CD28+ and CD4+CD28null T cells to a varying extent (Figure 2b). In contrast, we found TLR4 exclusively in CD4+CD28null T cells on both the mRNA and the protein level. The activation of TLR4 on CD4+CD28null T cells was independent of TCR-mediated stimulation for perforin production, and TLR4 signalling did not lead to an additive effect on concomitant cross-linking of TCR (data not shown). The high affinity of TLR4 to LPS without the obligatory need of the TCR signal may therefore have an influence on the susceptibility of CD4+CD28null T cells from AS patients to Gram-negative bacterial components.
As TNF-α directly influences CD28 gene transcription and may facilitate the emergence of CD4+CD28null T cells in chronic inflammatory syndromes [36], we also studied the effects of TNF-α on PRRs in vitro. In line with its effects on monocytic TLRs on the mRNA level [37], TNF-α also resulted in an increased protein expression of TLR4 and TLR2 on CD4+CD28null T cells (Figure 6a). Accordingly, expression of TLRs on CD4+CD28null T cells from patients with active AS disease before treatment (Figure 6b) were higher than those of unselected AS patients (examined for Figure 3b). The expression of CD14, TLR4 and TLR2 was then reduced on fresh CD4+CD28null T cells from AS patients treated with TNF-α blocking agents, further indicating the important role of TNF-α for the upregulation of surface expression of these PRRs also on CD4+CD28null T cells (Figure 6b).
Conclusion
The finding of PRRs on cytotoxic CD4+CD28null T cells of patients with AS, PsA or RA represents a new pathophysiological link between the innate and the adaptive immune system. In vitro activation of CD4+CD28null T cells by LPS is mediated by TLR4 and depends on CD14. Additional work has to be carried out to explain the downstream mechanisms of action and the clinical implications of these findings.
Abbreviations
AS = ankylosing spondylitis; ELISA = enzyme-linked immunosorbent assay; FCS = foetal calf serum; IFN-γ = interferon gamma; IL = interleukin; LPS = lipopolysaccharide; mAb = monoclonal antibody; mCD14 = membrane-bound CD14; NF = nuclear factor; NK = natural killer; PBMC = peripheral blood mononuclear cell; PBS = phosphate-buffered saline; PCR = polymerase chain reaction; PRR = pattern recognition receptor; PsA = psoriatic arthritis; RA = rheumatoid arthritis; RT = reverse transcriptase; sCD14 = soluble CD14; TCR = T-cell receptor; TLR = Toll-like receptor; TNF-α = tumour necrosis factor alpha.
Competing interests
The 'Verein zur Förderung der Hämatologie, Onkologie und Immunologie' (Innsbruck, Austria) which sponsors the laboratory, had been supported to a minor extent by Aesca, Austria. The authors declare that they have no competing interests.
Authors' contributions
BR, C Dejaco, C Duftner and CG carried out the cell culture work, WK carried out the ELISAs. CG also helped to coordinate the study. SCV and MK performed RT-PCR. BR, C Dejaco, C Duftner and MS designed the study, performed the statistical analysis and drafted the manuscript. BGL critically provided important discussion on the data. All authors read and approved the final manuscript.
Acknowledgements
This work was supported by the Innsbruck Medical University, the 'Verein zur Förderung der Hämatologie, Onkologie und Immunologie' (Innsbruck, Austria) and by the 'Verein zur Förderung der Ausbildung und wissenschaftlichen Tätigkeit von Südtirolern an der Universität Innsbruck' (Innsbruck, Austria) (to C Dejaco).
Figures and Tables
Figure 1 Prevalence of circulating CD3+CD4+CD28null and CD3+CD8+CD28null cells in ankylosing spondylitis, psoriatic arthritis and rheumatoid arthritis. (a) Representative example showing the prevalence of CD3+CD4+CD28null T cells (percentage out of C3+CD4+) and CD3+CD8+CD28null T cells (percentage out of C3+CD8+). The histogram shows staining for CD3 (filled grey curve) and isotype control antibody (black line). Dot plots are then gated on CD3+ cells. (b) Box plots summarize the prevalence of CD3+CD4+CD28null and CD3+CD8+CD28null T cells in ankylosing spondylitis (AS), psoriatic arthritis (PsA) and rheumatoid arthritis (RA) patients. The Mann-Whitney U test was used to determine the statistical differences between patients and the age-matched healthy control group (CO). ***P < 0.001.
Figure 2 Messenger RNA expression of TLR2 and TLR4 in CD3+CD4+CD28null and CD3+CD4+CD28+ T cells. (a) Fluorescence-activated cell sorting analysis shows the purity of CD3+CD4+CD28null and CD3+CD4+CD28+ T cells. (b) mRNA expression of TLR2, TLR4 and β2-microglobulin (β2m, housekeeping gene) in CD3+CD4+CD28null T cells (CD28-) and in CD3+CD4+CD28+ T cells (CD28+). Peripheral blood mononuclear cells were used as positive control (pos co), and a negative control (neg co) was performed in the absence of cDNA. A representative example out of three independent experiments is given.
Figure 3 Surface expression of CD14, TLR4 and TLR2 on CD4+CD28+ and CD28null cells in ankylosing spondylitis, psoriatic arthritis and rheumatoid arthritis. (a) Representative dot plots and histograms show TLR4 expression (filled red curve, black line represents isotype control) on CD4+CD28+ and CD4+CD28null T cells. Gates were set on lymphocytes (forward scatter and sideward scatter) as well as on CD28+ and CD28null cells expressing high levels of CD4. (b) Box plots summarize the expression of CD14, TLR4 and TLR2 on CD4+CD28+ and CD28null T cells in patients as indicated. The Wilcoxon test was used to determine the statistical differences between the groups. ***P < 0.001. SSC, side scatter; FSC, forward scatter; AS, ankylosing spondylitis; PsA, psoriatic arthritis; RA, rheumatoid arthritis.
Figure 4 Effects of LPS, sCD14 and autologous serum on perforin production by CD4+ T cells. Fresh peripheral blood mononuclear cells of patients with ankylosing spondylitis were incubated with medium (as a negative control), soluble CD14 (sCD14) or 10 μg/ml lipopolysaccharide (LPS) alone or with LPS in combination with 25 μg/ml sCD14 and 5% autologous serum for 16 hours. After staining with fluorescence-marked monoclonal antibodies directed against perforin, CD4 and CD28, cells were counted by flow cytometry. (a) Histograms show perforin production by CD4+CD28+ (upper row) and CD4+CD28null T cells (lower row) in response to medium, sCD14, LPS, LPS + sCD14 and LPS + serum as indicated (red curves). Black lines represent isotype control staining. Values indicate the mean fluorescence intensity. Gates for CD4+CD28null and CD4+CD28+ T cells were set as shown in Figure 3a. (b) Box blots show percentages of perforin-producing CD4+CD28null T cells from seven independent experiments. Differences were tested for significance using the Wilcoxon test. ***P ≤ 0.001.
Figure 5 CD14 and TLR4-mediated effects. (a) ELISA assays were performed to analyse levels of soluble CD14 (sCD14) in sera from patients with ankylosing spondylitis (AS) (n = 50) and healthy controls (CO) (n = 23). The Mann-Whitney test was used to determine the statistical differences between the group of patients and the control group. **P < 0.01. A blocking antibody (Ab) directed against (b) CD14 and (c) TLR4 or an isotype control were added to peripheral blood mononuclear cells from patients with AS maintained in 5% autologous serum. After 1 hour, lipopolysaccharide (LPS) stimulation at a concentration of 10 μg/ml for 16 hours was started. Box blots show percentages of perforin-producing CD4+CD28null T cells from seven independent experiments. Differences were tested for significance using the Wilcoxon test. **P < 0.01.
Figure 6 Effects of TNF-α on expression of pattern recognition receptors in vitro and in vivo. (a) Peripheral blood mononuclear cells were stimulated with 20 ng/ml tumour necrosis factor-α (TNF-α) for 24 hours, and CD4+CD28null T cells were analysed for expression of CD14, TLR4 and TLR2. Box plots summarize data from seven independent experiments. Medians were compared using the Wilcoxon test. ***P < 0.001, **P < 0.01. (b) CD4+CD28null T cells in patients with active ankylosing spondylitis treated with infliximab at a dosage of 3 mg/kg body weight were tested for the expression of pattern recognition receptors (PRRs) before and 3 weeks after injection (n = 8). The expression of CD14, TLR4 and TLR2 was detected by flow cytometry. The Wilcoxon test was used to determine differences in expression of PRRs before (pre) and under successful TNF-α blocking treatment (post). **P < 0.01, *P < 0.05.
Table 1 Patients' characteristics
Controls (n = 64) Ankylosing spondylitis (n = 65) Psoriatic arthritis (n = 25) Rheumatoid arthritis (n = 72)
Age (years) 49.2 ± 12.4 43.1 ± 10.9 52.2 ± 12 57.9 ± 11
Gender (% female) 50.8 28.4 45.8 75.6
Rheumatoid factor positivity (%) 0.0 n.d. n.d. 77.8
CCP-Ab positivity (%) 0.0 n.d. n.d. 80.5
HLA-B27 positivity (%) n.d. 78.6 31.3 n.d.
Axial involvement (% positive) 0.0 100.0 48.0 0.0
Erythrocyte sedimentation rate >15 mm/hour (% positive) 0.0 62.7 88.2 86.2
C-reactive protein >0.7 μg/dl (% positive) 0.0 26.1 41.7 46.8
CCP-Ab, antibodies directed against cyclic citrullinated peptides; n.d., not determined.
==== Refs
Janeway CA Jr Approaching the asymptope? Evolution and revolution in immunology Cold Spring Harb Symp Quant Biol 1989 54 1 13 2700931
Iwasaki A Medzhitov R Toll-like receptor control of the adaptive immune responses Nat Immunol 2004 5 987 995 15454922 10.1038/ni1112
Faure E Equils O Sieling PA Thomas L Zhang FX Kirschning CJ Polentarutti N Muzio M Arditi M Bacterial lipopolysaccharide activates NF-κB through Toll-like receptor 4 (TLR-4) in cultured human dermal endothelial cells. Differential expression of TLR-4 and TLR-2 in endothelial cells J Biol Chem 2000 275 11058 11063 10753909 10.1074/jbc.275.15.11058
Chalifour A Jeannin P Gauchat JF Blaecke A Malissard M N'Guyen T Thieblemont N Delneste Y Direct bacterial protein PAMP recognition by human NK cells involves TLRs and triggers α-defensin production Blood 2004 104 1778 1783 15166032 10.1182/blood-2003-08-2820
Ohashi K Burkart V Flohé S Kolb H Cutting edge: heat shock protein 60 is a putative endogenous ligand of the Toll-like receptor-4 complex J Immunol 2000 164 558 561 10623794
Wright SD Ramos RA Tobias PS Ulevitch RJ Mathison JC CD14, a receptor for complexes of lipopolysaccharide (LPS) and LPS binding protein Science 1990 249 1431 1433 1698311
Poltorak A HE X Smirnova I Liu MY van Huffel C Du X Birdwell D Alejos E Silva M Galanos C Defective LPS signalling in C3H/HeJ and C57BL/10ScCr mice: mutations in Tlr4 gene Science 1998 282 2085 2088 9851930 10.1126/science.282.5396.2085
Lien E Sellati TJ Yoshimura A Flo TH Rawadi G Finberg RW Carroll JD Espevik T Ingalls RR Radolf JD Golenbock DT Toll-like receptor 2 functions as a pattern recognition receptor for diverse bacterial products J Biol Chem 1999 274 33419 33425 10559223 10.1074/jbc.274.47.33419
Maxwell JR Rossi RJ McSorley SJ Vella AT T cell clonal conditioning: a phase occurring early after antigen presentation but before clonal expansion is impacted by Toll-like receptor stimulation J Immunol 2004 172 248 259 14688332
Medzhitov R Janeway CA Jnr Innate immunity N Engl J Med 2000 343 338 344 10922424 10.1056/NEJM200008033430506
Pasare C Medzhitov R Toll-dependent control mechanisms of CD4 T cell activation Immunity 2004 21 733 741 15539158 10.1016/j.immuni.2004.10.006
Pasare C Medzhitov R Toll-pathway-dependent blockade of CD4+CD25+ T cell-mediated suppression by dendritic cells Science 2003 299 1033 1036 12532024 10.1126/science.1078231
Matsuguchi T Takagi K Musikacharoen T Yoshikai Y Gene expressions of lipopolysaccharide receptors, toll-like receptors 2 and 4, are differently regulated in mouse T lymphocytes Blood 2000 95 1378 1385 10666214
Sobek V Birkner N Falk I Wurch A Kirschning CJ Wagner H Wallich R Lamers MC Simon MM Direct Toll-like receptor 2 mediated co-stimulation of T cells in the mouse system as a basis for chronic inflammatory joint disease Arthritis Res Ther 2004 6 R433 R446 15380043 10.1186/ar1212
Gelman AE Zhang J Choi Y Turka LA Toll-like receptor ligands directly promote activated CD4+ T cell survival J Immunol 2004 172 6065 6073 15128790
Caramalho I Lopes-Carvalho T Ostler D Zelenay S Haury M Demengeot J Regulatory T cells selectively express Toll-like receptors and are activated by lipopolysaccharide J Exp Med 2003 197 403 411 12591899 10.1084/jem.20021633
Komai-Koma M Jones L Ogg GS Xu D Liew FY TLR2 is expressed on activated T cells as a costimulatory receptor Proc Natl Acad Sci USA 2004 101 3029 3034 14981245 10.1073/pnas.0400171101
Seibl R Birchler T Loeliger S Hossle JP Gay RE Saurenmann T Michel BA Seger RA Gay S Lauener RP Expression and regulation of Toll-like receptor 2 in rheumatoid arthritis synovium Am J Pathol 2003 162 1221 1227 12651614
Schmidt D Goronzy JJ Weyand CM CD4+ CD7- CD28- T cells are expanded in rheumatoid arthritis and are characterized by autoreactivity J Clin Invest 1996 97 2027 2037 8621791
Lamprecht P Moosig F Csernok E Seitzer U Schnabel A Mueller A Gross WL CD28 negative T cells are enriched in granulomatous lesions of the respiratory tract in Wegener's granulomatosis Thorax 2001 56 751 757 11562512 10.1136/thorax.56.10.751
Markovic-Plese S Cortese I Wandinger KP McFarland HF Martin R CD4+CD28- costimulation-independent T cells in multiple sclerosis J Clin Invest 2001 108 1185 1194 11602626 10.1172/JCI200112516
Duftner C Goldberger C Falkenbach A Würzner R Falkensammer B Pfeiffer KP Maerker-Hermann E Schirmer M Prevalence, clinical relevance and characterization of circulating cytotoxic CD4+CD28- T cell in ankylosing spondylitis Arthritis Res Ther 2003 5 R292 R300 12932293 10.1186/ar793
Namekawa T Snyder MR Yen JH Goehring BE Leibson PJ Weyand CM Goronzy JJ Killer cell activating receptors function as costimulatory molecules on CD4+CD28null T cells clonally expanded in rheumatoid arthritis J Immunol 2000 165 1138 1145 10878393
Groh V Bruhl A El-Gabalawy H Nelson JL Spies T Stimulation of T cell autoreactivity by anomalous expression of NKG2D and its MIC ligands in rheumatoid arthritis Proc Natl Acad Sci USA 2003 100 9452 9457 12878725 10.1073/pnas.1632807100
Warrington KJ Takemura S Goronzy JJ Weyand CM CD4+, CD28- T cells in rheumatoid arthritis patients combine features of the innate and adaptive immune systems Arthritis Rheum 2001 44 13 20 11212151 10.1002/1529-0131(200101)44:1<13::AID-ANR3>3.0.CO;2-6
Dougados M van der Linden S Juhlin R Huitfeldt B Amor B Calin A Cats A Dijkmans B Olivieri I Pasero G The European Spondylarthropathy Study Group preliminary criteria for the classification of spondylarthropathy Arthritis Rheum 1991 34 1218 27 1930310
Goie The HS Steven MM van der Linden SM Cats A Evaluation of diagnostic criteria for ankylosing spondylitis: a comparison of the Rome, New York and modified New York criteria in patients with a positive clinical history screening test for ankylosing spondylitis Br J Rheumatol 1985 24 242 249 3160423
Wright V Moll JMH Wright V, Moll JMH Psoriatic arthritis Seronegative Polyarthritis 1976 Amsterdam: North Holland Publishing Company 169 223
Arnett FC Edworthy SM Bloch DA McShane DJ Fries JF Cooper NS Healey LA Kaplan SR Liang MH Luthra HS The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis Arthritis Rheum 1988 31 315 324 3358796
Herndler-Brandstetter D Schwaiger S Veel E Fehrer C Cioca DP Almanzar G Keller M Pfister G Parson W Wurzner R CD25-expressing CD8+ T cells are potent memory cells in old age J Immunol 2005 175 1566 1574 16034095
Schaefer TM Desouza K Fahey JV Beagley KW Wira CR Toll-like receptor (TLR) expression and TLR-mediated cytokine/chemokine production by human uterine epithelial cells Immunology 2004 112 428 436 15196211 10.1111/j.1365-2567.2004.01898.x
Iwahashi M Yamamura M Aita T Okamoto A Ueno A Ogawa N Akashi S Miyake K Godowski PJ Makino H Expression of Toll-like receptor 2 on CD16+ blood monocytes and synovial tissue macrophages in rheumatoid arthritis Arthritis Rheum 2004 50 1457 1467 15146415 10.1002/art.20219
Filion LG Izaguirre CA Garber GE Huebsh L Aye MT Detection of surface and cytoplasmic CD4 on blood monocytes from normal and HIV-1 infected individuals J Immunol Methods 1990 135 59 69 1703191 10.1016/0022-1759(90)90256-U
Pugin J Schürer-Maly C-C Leturcq D Moriarty A Ulevitch RJ Tobias PS Lipopolysaccharide activation of human endothelial and epithelial cells is mediated by lipopolysaccharide-binding protein and soluble CD14 Proc Natl Acad Sci USA 1993 90 2744 2748 7681988
Heumann D Bas S Gallay P Le Roy D Barras C Mensi N Glauser MP Vischer T Lipopolysaccharide binding protein as a marker of inflammation in synovial fluid of patients with arthritis: correlation with interleukin 6 and C-reactive protein J Rheumatol 1995 22 1224 1229 7562749
Bryl E Vallejo AN Weyand CM Goronzy JJ Down-regulation of CD28 expression by TNF-α J Immunol 2001 167 3231 3238 11544310
Muzio M Bosisio D Polentarutti N D'amico G Stoppacciaro A Mancinelli R van't Veer C Penton-Rol G Ruco LP Allavena P Mantovani A 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
|
16277694
|
PMC1297589
|
CC BY
|
2021-01-04 16:02:40
|
no
|
Arthritis Res Ther. 2005 Oct 18; 7(6):R1412-R1420
|
utf-8
|
Arthritis Res Ther
| 2,005 |
10.1186/ar1840
|
oa_comm
|
==== Front
Arthritis Res TherArthritis Research & Therapy1478-63541478-6362BioMed Central London ar18451627769510.1186/ar1845Research ArticleIdentification of citrullinated α-enolase as a candidate autoantigen in rheumatoid arthritis Kinloch Andrew 1Tatzer Verena 1Wait Robin 1Peston David 1Lundberg Karin 1Donatien Phillipe 1Moyes David 1Taylor Peter C 1Venables Patrick J [email protected] Kennedy Institute of Rheumatology, Imperial College London, Charing Cross Hospital Campus, 1 Aspenlea Road, London W6 8LH, UK2005 19 10 2005 7 6 R1421 R1429 5 5 2005 1 6 2005 8 9 2005 29 9 2005 Copyright © 2005 Kinloch 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.
Antibodies against citrullinated proteins are highly specific for rheumatoid arthritis (RA), but little is understood about their citrullinated target antigens. We have detected a candidate citrullinated protein by immunoblotting lysates of monocytic and granulocytic HL-60 cells treated with peptidylarginine deiminase. In an initial screen of serum samples from four patients with RA and one control, a protein of molecular mass 47 kDa from monocytic HL-60s reacted with sera from the patients, but not with the serum from the control. Only the citrullinated form of the protein was recognised. The antigen was identified by tandem mass spectrometry as α-enolase, and the positions of nine citrulline residues in the sequence were determined. Serum samples from 52 patients with RA and 40 healthy controls were tested for presence of antibodies against citrullinated and non-citrullinated α-enolase by immunoblotting of the purified antigens. Twenty-four sera from patients with RA (46%) reacted with citrullinated α-enolase, of which seven (13%) also recognised the non-citrullinated protein. Six samples from the controls (15%) reacted with both forms. α-Enolase was detected in the RA joint, where it co-localised with citrullinated proteins. The presence of antibody together with expression of antigen within the joint implicates citrullinated α-enolase as a candidate autoantigen that could drive the chronic inflammatory response in RA.
==== Body
Introduction
Rheumatoid arthritis (RA) is a common and disabling disease affecting about 1% of the population [1]. Unlike most other autoimmune rheumatic diseases, the dominant autoantigens are unknown. Because rheumatoid factors are present in up to 75% of patients with RA, it has been suggested that immunoglobulin G is the antigen. However, rheumatoid factors are also present in patients with other diseases and in up to 5% of healthy individuals [2]. Other antibodies are also present in sera from patients with RA, including antiperinuclear factor [3] and antikeratin antibody [4]. Because both antiperinuclear factor and antikeratin antibody react with human filaggrin and related proteins [5] they were collectively designated 'anti-filaggrin antibodies'. It was subsequently reported that binding of anti-filaggrin antibody epitopes is dependent on the presence of citrulline, an amino acid derived from arginine as a result of a post-translational modification catalysed by the enzyme peptidylarginine deiminase (PAD) [6,7].
These findings have been exploited in anti-cyclic citrullinated peptide (anti-CCP) assays, which are more sensitive (80%) and specific (97%) for RA than rheumatoid factors are [8]. Anti-CCPs may occur early in disease [9], or even before clinical manifestations [10]. Anti-CCP positivity also predicts a more aggressive form of RA [11,12]. Anti-filaggrin antibodies have been found at higher concentrations in synovial membrane than in synovial fluid and peripheral blood [13] from patients with RA. However, filaggrin is notably absent from the RA joint [8]. This suggested that there might be other citrullinated proteins in the joint driving the immune response. Citrullinated fibrin is a candidate because it is present in interstitial deposits in the synovial membrane [13] and is recognised by anti-citrullinated-filaggrin antibodies. Endogenous citrullination of fibrin has also been demonstrated in murine models of arthritis [14]. However, immunisation of mice with citrullinated fibrinogen did not induce arthritis [15,16]. Another candidate is citrullinated vimentin, now known to be identical to the Sa antigen [17,18], the presence of which has been demonstrated in synovial membrane [19].
It is not known whether citrullinated vimentin and fibrin are just two of multiple citrullinated autoantigens in RA, or whether there is a dominant autoantigen that has yet to be described. The premise of the current study is that, if there were such a candidate, it is likely to be present in myeloid cells, the dominant cell type in the rheumatoid joint. We therefore studied the promyelocytic HL-60 cell line, which can readily be differentiated into cells with a monocytic or granulocytic phenotype that also express PAD [20]. Untreated and citrullinated lysates of HL-60s were probed with an initial screening panel of serum from patients with RA, to identify reactive polypeptides. These were then partly purified and identified by tandem mass spectrometry. This approach has enabled us to propose citrullinated α-enolase as a novel candidate autoantigen for RA.
Materials and methods
Patient samples
Serum was obtained with informed consent from 52 patients with RA attending the Rheumatology Clinic, Charing Cross Hospital, London. All met the classification criteria for RA [21]. Control serum samples were obtained from healthy volunteers. Multiple synovial biopsies were taken under direct vision from each of three predetermined sites within the knee joint during arthroscopic examination in eight patients with RA and four with osteoarthritis. Informed consent was obtained from each patient before arthroscopy. All biopsies taken during a single examination were fixed for 24 hours in 10% neutral buffered formalin and then processed into paraffin wax. Ethical approval was granted by the Riverside Research Ethics Committee and the Hammersmith NHS Trust Research Ethics Committee.
Isolation of RA synovial cells
Synovial cells were isolated from synovium that had been surgically removed from three patients, undergoing total knee, hip or elbow replacement. After the removal of fat, synovium was cut into small pieces in complete medium (RPMI 1640, 10% fetal calf serum, 1% penicillin and streptomycin) in a plastic tissue culture dish. The tissue was drained in a sieve and scraped into a beaker containing 20 ml of complete medium, 100 μg of collagenase A and 3 μg of DNase A, mixed thoroughly and incubated for up to 90 minutes at 37°C until 'stringy'. The mixture was shaken vigorously and diluted with complete medium to a final volume of 50 ml. Synovial cells were pelleted by centrifugation (200 g for 10 minutes at 24°C).
Culture and differentiation of HL-60 cells
HL60 cells were cultured in complete medium and passaged every third day. For differentiation to PAD-expressing monocytes or granulocytes, 3 × 105 cells/ml were incubated for 3 days with either 100 nM 1α,25-dihydroxyvitamin D3 (Wako Chemicals, Neuss, Germany) or 1 μM trans-retinoic acid (Sigma, Poole, UK) [20].
Preparation of whole-cell lysates and subcellular fractionation
Cells were lysed at 2.5 × 106 cells per 150 μl of lysis buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.1% SDS, 1% Nonidet P40, 100 mg/ml aprotinin). Protein concentrations were measured by the Bio-Rad DC Protein assay (Bio-Rad, Hercules, CA, USA) and diluted with PBS. Subcellular fractionation was performed by resuspending PBS-washed cells in lysis buffer (10 mM Tris-HCl, pH 7.5, 1 mM potassium acetate, 1.5 mM magnesium acetate, 2 mM dithiothreitol, 1 mM phenylmethylsulphonyl fluoride, 10 μg/ml aprotinin, 1 μg/ml leupeptin, 10 μg/ml pepstatin), incubating on ice for 30 minutes and disrupting with a Dounce homogeniser. Homogenates were centrifuged (500 g for 10 minutes) to pellet the nuclear fraction, which was washed, disrupted by sonication and solubilised in 0.5% Nonidet P40. The supernatant from the nuclear fractionation was centrifuged at 100,000 g, giving a membrane-rich pellet (P100) and a cytosolic supernatant (S100).
Deimination of proteins in vitro
Deimination was performed as described previously [7]. In brief, whole-cell lysates and subcellular fractions were diluted to a concentration of 0.86 mg protein/ml in PAD buffer (0.1 M Tris-HCl, pH 7.4, 10 mM CaCl2, 5 mM dithiothreitol, 1 mM phenylmethylsulphonyl fluoride, 10 μg/ml aprotinin, 1 μg/ml leupeptin, 10 μg/ml pepstatin) and were deiminated in vitro with rabbit muscle PAD (7 U/mg of protein; Sigma) for 2 hours at 50°C. The reaction was stopped by boiling in Laemmli buffer for 10 minutes. Non-neuronal α-enolase (Hytest, Turku, Finland) was deiminated in the same buffer at a concentration of 0.365 mg/ml. All samples were stored at -20°C until use.
Immunoblotting
Whole-cell lysates were separated on 10% NuPAGE Bis-TrisGels (Invitrogen, Paisley, Renfrewshire, UK), transferred to nitrocellulose membranes, blocked with 5% non-fat milk in PBS/0.1% Tween, and probed with human serum diluted 100-fold with the blocking solution. Goat anti-α-enolase antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) was used at a dilution of 1:100. Membranes were washed three times for 15 minutes with PBS/0.1% Tween and incubated with peroxidase-conjugated secondary antibody (Jackson Immuno Research, West Grove, PA, USA), anti-human (recognising immunoglobulin G, immunoglobulin M and immunoglobulin A) and rabbit anti-goat respectively. After a further wash, membranes were developed with the use of enhanced chemiluminescence (Amersham Biosciences, Little Chalfont, Buckinghamshire, UK) in accordance with the manufacturer's instructions. Deiminated proteins were identified with an anti-citrulline (modified) detection kit (catalogue no. 07–-390; Upstate, Lake Placid, NY, USA). The presence of antibodies against citrullinated and non-citrullinated antigens (1.92 μg per well) was established by blotting with serum at a dilution of 1:40.
Two-dimensional gel electrophoresis
In vitro deiminated S100 fractions of 1α,25-dihdroxyvitamin D3-differentiated HL-60 cells were desalted with spin desalting columns (Pierce, Northumberland, UK) and were dissolved in 2D lysis buffer (9.5 M urea, 1% (w/v) dithiothreitol, 2% CHAPS and 0.5% carrier ampholyte (Amersham Biosciences) supplemented with proteinase inhibitors. The samples were loaded by in-gel rehydration into linear pH 3 to 10 immobilised pH gradient dry strips 13 cm long (Amersham Biosciences). Isoelectric focusing was performed with a Multiphor II flatbed electrophoresis system (Amersham Biosciences) at 300 V for 1 minute, then ramped to 3,500 V for 1.5 hours and maintained at 3,500 V for 3.5 hours. Before separation in the second dimension, disulphide bonds were reduced by incubation with 65 mM dithiothreitol (15 minutes in 2% SDS, 6 M Urea, 30% v/v glycerol and 150 mM Tris-HCl, pH 8.8). Free thiol groups were alkylated by treatment with 260 mM iodoacetamide for 15 minutes. The strips were transferred to a 10% polyacrylamide gel and run at 8 mA. Gels were fixed and silver stained with a protocol compatible with mass spectrometry [22].
Mass spectrometry
In-gel digestion with trypsin was performed with an Investigator Progest robotic digestion system (Genomic Solutions, Huntington, UK) as described previously [23]. Tandem electrospray mass spectra were recorded with a Q-Tof hybrid quadrupole/orthogonal acceleration time-of-flight spectrometer (Micromass, Manchester, UK) interfaced to a Micromass CapLC chromatograph. Samples were dissolved in 0.1% aqueous formic acid, and introduced into the spectrometer by means of a Pepmap C18 column (300 μm × 0.5 cm; LC Packings, Amsterdam, The Netherlands), and were eluted with an acetonitrile/0.1% formic acid gradient (5% to 70% acetonitrile over 20 minutes).
The capillary voltage was set to 3,500 V, and data-dependent tandem mass spectrometry acquisitions were performed on precursors with charge states of 2, 3 or 4 over a survey mass range of 400 to 1,300. Proteins were identified by correlation of uninterpreted tandem mass spectra to entries in SwissProt/TrEMBL, using ProteinLynx Global Server (Version 1.1; Micromass) [24]. The database was created by merging the FASTA format files of SwissProt, TrEMBL and their associated splice variants. No taxonomic, mass or pI constraints were applied. One missed cleavage per peptide was allowed, and the fragment ion mass tolerance window was set to 100 p.p.m. All matching spectra were reviewed by an expert, and citrullinated residues were localised by manual interpretation of sequence-specific fragment ions with the MassLynx program PepSeq (Micromass).
Slide preparation and immunohistochemistry
Synovial tissue biopsies were processed into paraffin wax by fixation in 10% neutral buffered formalin for 24 hours. The tissue was then progressively dehydrated by passage through a series of graded alcohols and xylene. The samples were mounted on silane-treated slides, which were incubated for 10 minutes with 2% hydrogen peroxide/98% methanol, blocked for 10 minutes in horse serum and then incubated for 60 minutes with anti-α-enolase antibody diluted 1:400. Citrullinated proteins were detected with the anti-modified citrullinated protein kit (Upstate). Unmodified sections were used as controls. The slides were washed in TBS and incubated for 30 minutes with either biotinylated horse anti-goat (for enolase) or biotinylated pig anti-rabbit (for citrulline) antibodies, at a concentration of 1:400 and washed again before incubation for 30 minutes with avidin-biotin-HRP (PK6100; Vector Biolabs) at a 1:100 concentration and staining for 5 minutes with diaminobenzidine (SK4100; Vector Biolabs). Finally the slides were counterstained for 1 minute with haematoxylin, washed in tap water, dehydrated, cleared and mounted.
Results
Identification of a 47 kDa citrullinated protein as a target for antibodies in sera from patients with RA
Each of four serum samples from patients with RA, but not the control serum, reacted strongly with a band with an apparent mass of 47 kDa (boxed in Figure 1) in the PAD-treated lysates of HL-60 cells. No reaction at 47 kDa was observed with non-deiminated lysates. Reactivity at 47 kDa was strongest with HL-60 cells that had been differentiated to monocytes, although a similar polypeptide was also seen in lysates from cells with the granulocyte phenotype (data not shown). Endogenous citrullination in the HL-60 cells was undetectable with the antibody against modified citrulline, but after treatment with PAD in vitro, abundant citrullinated polypeptides were observed. The RA sera, particularly RA1 and RA4, seemed to be relatively selective for the 47 kDa polypeptide, with only four to six additional bands identifiable in each blot. Serum from RA2 and RA3 showed more diffuse reactivity, although a 47 kDa polypeptide predominated. This suggested that, among the numerous potential antigens generated by citrullination of proteins in cells of monocytic phenotype, there was apparent selectivity among our four screening sera for one citrullinated polypeptide migrating at 47 kDa. We therefore performed further experiments to identify this protein.
Nuclear, cytosolic and membrane fractions were prepared from HL-60 cells by differential centrifugation, and were deiminated with PAD as before. Immunoblotting with one of the RA sera showed that the 47 kDa antigen was enriched in the cytosolic (S100) fraction (Figure 2).
Identification of the 47 kDa autoantigen as citrullinated α-enolase
The deiminated S100 fraction was separated by one-dimensional SDS-PAGE and stained with Coomassie blue; the putative band recognised by sera from patients with RA was excised, digested with trypsin and analysed by tandem mass spectrometry. Fourteen peptides were sequenced (Table 1), all of which mapped onto α-enolase (SwissProt accession number P06733). In total 242 residues of non-redundant amino acid sequence were obtained, corresponding to 56% coverage. To confirm that the stained band co-localised with the protein recognised by sera from patients with RA, the deiminated S100 fraction was separated by two-dimensional electrophoresis, blotted onto nitrocellulose and probed with serum samples RA1 and RA4. Both recognised a doublet of spots that matched a feature on the silver-stained gel of apparent molecular mass 47 kDa, and with a pI of 5 (Figure 3). These spots were excised and identified by mass spectrometry as α-enolase. When the membranes were re-probed with an antibody specific for the carboxy-terminal region of α-enolase, we observed a similar pattern to that obtained with serum from patients with RA, confirming the identity of the autoantigen.
Conversion of arginine to citrulline results in a mass increase of 0.984 Da and a loss of the positive charge from the side chain, resulting in a significant acidic shift in the two-dimensional electrophoretic migration. Moreover, the pattern of peptides obtained on tryptic digestion will be altered, because the modified residues are refractory to trypsinolysis and yield peptides containing internal citrulline, rather than carboxy-terminal arginine (Tables 1 and 2). Six peptides containing internal citrulline residues were sequenced, which enabled the localisation of nine sites of modification (Table 1). None of these peptides were present in tryptic digests of unmodified α-enolase (Table 2). The pI determined by two-dimensional electrophoresis was also consistent with this extensive citrullination, being about 5.0.
Other antigens, recognised more sporadically by sera from patients with RA (Figure 3), were also characterised by mass spectrometry. They included elongation factor 1α (SwissProt accession number P68104) and adenyl cyclase-associated protein 1 (SwissProt accession number Q01518), both of which were shown to be citrullinated.
Higher prevalence of antibodies against citrullinated α-enolase than against native α-enolase in serum from patients with RA
Twenty-four of the RA serum samples (46%) reacted with the citrullinated α-enolase, seven of which (13%) also reacted with the non-citrullinated form of the protein. Six of the controls (15%) reacted with both (Figure 4). All of the 17 RA samples that reacted only with the citrullinated form of α-enolase were positive for anti-CCP2 (data not shown).
α-Enolase is abundant in synovium from patients with RA
Immunohistochemical analysis of inflamed synovial sections showed that all eight RA and four osteoarthritis samples examined expressed α-enolase. Expression in RA sections was greatest in the more hyperplastic subsynovial layer (Figure 5b). In the osteoarthritis sections, α-enolase staining was predominantly localised in vascular endothelial cells (Figure 5a). The antibody against modified citrulline (Figure 6a) indicated that citrullinated proteins were present in the region that stained positively for enolase, although the intensity of the staining indicated that levels of citrullination were relatively low. Immunoblotting of synovial cell lysates from three patients with RA demonstrated that a band co-migrating with purified α-enolase reacted with the anti-α-enolase antibody (Figure 7).
Discussion
In this study we characterised citrullinated α-enolase as a dominant antigen in citrullinated lysates of differentiated HL-60 cells targeted by a screening panel of serum from patients with RA. The identity of the antigen was established by mass spectrometry, and the sites of nine citrulline residues within the protein were determined by tandem mass spectrometry. Further confirmation was obtained by two-dimensional electrophoresis and Western blotting with a specific anti-enolase antibody. With the use of purified protein, 46% of a larger panel of sera from patients with RA reacted with citrullinated α-enolase by immunoblotting. This suggests that citrullinated α-enolase is at least as immunodominant as citrullinated filaggrin or citrullinated vimentin, because, by immunoblotting, the frequency of antibodies against citrullinated filaggrin has been reported as 41 to 58% [25-28] and against citrullinated vimentin as 22 to 40% [19,29,30]. Improved sensitivity and specificity of RA diagnosis may well be obtained by testing RA sera with peptides derived from citrullinated epitopes of α-enolase, as has been demonstrated for citrullinated filaggrin in the first-generation anti-CCP test, in which the sensitivity increased to more than 70%.
α-Enolase, unlike filaggrin, is abundantly expressed in the synovial membrane. Several lines of evidence indicate that it is citrullinated in the joint. First, it was detected in the myeloid-like HL-60s cell line, which expresses PAD and has a similar phenotype to that of cells abundant in the joint. Second, it was detected as a synovial antigen that co-localised with staining for citrullinated proteins. The staining shown in Figure 6a suggests that only a small proportion of the antigen is citrullinated in vivo, which might explain why we were unable to demonstrate citrullination of α-enolase by Western blotting of immunoprecipitates from synovial cells.
Although this is the first report that citrullinated α-enolase is a common target antigen in RA, native α-enolase has previously been observed as an infrequent target antigen for several autoimmune diseases [31-34]. For example, Saulot and colleagues [34] observed that antibodies against (placental) α-enolase occurred in 25% of patients with RA and were predictive of radiological progression. They found that only 8 of the 36 patients reacting with placental α-enolase also reacted with the recombinant protein. This contrasted with 19% of patients with systemic lupus erythematosus and 15% of patients with systemic sclerosis whose serum samples reacted with both forms of α-enolase. They hypothesised that RA sera reacting with placental α-enolase, but not recombinant antigen, were recognising a post-translationally modified form of α-enolase. Although it is tempting to speculate that the modification they predicted is citrullination, the most abundant of the triplet of spots identified as α-enolase in their study migrated in two-dimensional electrophoresis at a pI of 7.0, consistent with native α-enolase. However, it is possible that the two more acidic α-enolase spots, which they attributed to phosphorylation, might in fact be citrullinated. The expression of PAD2 protein in the placenta would account for a degree of deimination either in vivo or during extraction. It is also consistent with the identification of the Sa antigen, also of placental origin, as citrullinated vimentin [17]. The higher frequency of anti-citrullinated α-enolase in our study than that of Saulot and colleagues might be due to the fact that our cell lysates were extensively deiminated in vitro. This is demonstrated by the uniform migration of deiminated enolase at a pI of 5 and by the replacement of arginine by citrulline in all the peptides listed in Table 1.
In our study, 15% of the control sera reacted with native α-enolase and also with citrullinated α-enolase, whereas reactivity with the citrullinated form alone was restricted to the patients with RA. This is, again, consistent with the results of Saulot and colleagues, assuming that the placental form of the protein was partly deiminated. In turn, this suggests that RA-specific antibodies might be driven by peptides containing one or more of the 17 potential citrulline residues in the sequence of α-enolase. Binding to non-arginine containing regions might account for the 'background', and hence the apparent loss of disease specificity seen when immunoblotting with the normal sera in our study, and the non-RA sera in that of Saulot and colleagues. One way to test this would be to examine reactivity to peptides derived from citrullinated epitopes from α-enolase. Such studies are currently in progress. If both sensitivity and specificity increases for RA it might provide another assay for diagnosis of the disease. More importantly it would provide further data to support the concept that α-enolase is a driving autoantigen in RA.
The properties of citrullinated α-enolase make it an attractive synovial antigen for driving the immune response. α-Enolase is a highly conserved, multifunctional protein that, in addition to its role in glycolysis, binds plasminogen. It is known to be upregulated by hypoxia [35] and by proinflammatory stimuli [36], both of which are features of the synovial membrane microenvironment in RA. α-Enolase is expressed during cell differentiation and is used as marker of differentiation in the grading of tumours [37]. In myeloid cells, the dominant cell type in the inflamed synovium, it is co-expressed with PAD2 and PAD4. In the present study we have shown that its distribution in the subsynovium is similar to that of citrullinated proteins and PAD [38], but we have not yet demonstrated its citrullination in vivo.
There is substantial similarity between human and prokaryotic α-enolases (47% identity with that from Streptococcus pyogenes, for example), and antibodies raised against streptococcal surface α-enolase also recognise the human enzyme [36]. Thus the presence of antibodies against uncitrullinated α-enolase, in serum of individuals without RA, might be attributable to cross-reaction with bacterial epitopes. Expression of an enzyme able to citrullinate peptidylarginine has been demonstrated in the oral organism Porphyromonas gingivalis [39], which provides a mechanism by which antibacterial antibodies cross-react with endogenous citrullinated proteins and initiate loss of tolerance.
Conclusion
We have demonstrated that antibodies against citrullinated α-enolase are found in 46% of serum samples from patients with RA, and that native α-enolase is abundantly expressed in rheumatoid synovium. It is upregulated by factors such as hypoxia, which are characteristic of the rheumatoid joint, and its amino-acid sequence is highly conserved between prokaryotes and higher eukaryotes, making citrullinated α-enolase a candidate target antigen in RA; this merits further investigation.
Abbreviations
CCP = cyclic citrullinated peptide; PAD = peptidylarginine deiminase; PBS = phosphate-buffered saline; RA = rheumatoid arthritis.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
AK performed immunoblotting, screened sera for antibodies, detected α-enolase in synoviocytes, and participated in study design and drafting of the manuscript. VT performed the initial Western blotting, cellular fractionation and two-dimensional electrophoresis experiments. DP, KL and PD performed the immunohistochemistry. DM participated in the study design and established the cell culture methodology. PCT performed the synovial biopsies and participated in study design and drafting of the manuscript. RW was responsible for mass spectrometric characterisation of citrullinated α-enolase, participated in the study design and helped to draft and edit the manuscript. PJV conceived of the study, participated in its design and edited the manuscript. All authors read and approved the final version.
Acknowledgements
We are grateful to Shajna Begum for assistance with preparation of samples for mass spectrometry and to Lauren Schewitz for performing the synovial cell isolation. We thank the Arthritis Research Campaign, the Medical Research Council and the Wellcome Trust for their support.
Figures and Tables
Figure 1 Screen to identify undeiminated and deiminated proteins reacting with RA and non-RA serum samples. Proteins from HL60 lysates incubated (+) or without (-) peptidylarginine deiminase (PAD) blotted with an antibody specific for modified citrulline residues (anti-citrulline) and a screening panel of rheumatoid arthritis (RA1 to RA4) and non-RA (control) serum. The rectangular box indicates a citrullinated protein reacting strongly with each of the RA serum samples but not the control.
Figure 2 Intracellular expression of immunogenic citrullinated 47 kDa protein. Presence of citrullinated 47 kDa protein reactive with rheumatoid arthritis serum 1 in different subcellular fractions of undifferentiated HL60s (U) and HL60 monocytes (M) (S100, cytosolic; P100, membrane; Nuc, nuclear) showing enrichment in the S100 (cytosolic) fraction.
Figure 3 Characterisation of the 47 kDa protein by two-dimensional electrophoresis. Proteins in the 47 kDa rich monocytic S100 fraction were separated by two-dimensional electrophoresis according to charge (x-axis) and molecular mass (y-axis). (a) The full complement of proteins was observed by silver staining. (c,d) Proteins reacting with rheumatoid arthritis serum samples 1 (c) and 4 (d) were highlighted by immunoblotting. (b) The highly reactive 47 kDa protein was confirmed as α-enolase by immunoblotting with the goat anti-α-enolase antibody. CAP1, adenyl cyclase-associated protein 1; EF1α, elongation factor 1α.
Figure 4 Prevalence of antibodies against deiminated and undeiminated α-enolase in RA patients and healthy controls. Immunoblotting of in vitro citrullinated and untreated α-enolase with serum samples from patients with rheumatoid arthritis (RA) and normal controls, showing that about half of the serum samples from the RA group contain antibodies with selectivity for citrullinated α-enolase.
Figure 5 Localisation of α-enolase in synovial membranes. Immunohistochemistry of biopsy sections from patients with osteoarthritis (a) and rheumatoid arthritis (RA) (b) with the goat anti-α-enolase antibody showing expression of α-enolase (stained brown) in cells in the subsynovium of the patient with RA and in endothelial cells in the patient with osteoarthritis. Cell nuclei are counterstained blue.
Figure 6 Localisation of citrullinated proteins in synovial membranes. (a) Immunostaining of citrullinated proteins by the anti-modified citrulline kit was mainly confined to the subsynovium. (b) No staining was visible on the control. (c) Staining produced by the anti-α-enolase antibody on an adjacent section was much stronger, and included the subsynovial cells which also stained for citrullinated antigens. Original magnification × 20 in all cases.
Figure 7 Presence of α-enolase in synovial cells from patients with rheumatoid arthritis (RA). Immunoblotting of lysates from RA synovial cells with anti-α-enolase (+PAD, deiminated α-enolase; – PAD, undeiminated α-enolase) showing a 47 kDa protein in synoviocytes, from three patients with RA, reacting with the goat anti-α-enolase antibody, which co-migrates with purified α-enolase.
Table 1 Peptides from deiminated α-enolase sequenced by tandem mass spectrometry
m/z (charge) Location Matched sequence
401.24 (2+) 221–227 EGLELLK
480.77 (2+) 81–88 LNVTEQEK
674.34 (2+) 394–405 TGAPC(Cit)SE(Cit)LAK
696.87 (2+) 422–433 FAG(Cit)NF(Cit)NPLAK
817.41 (2+) 343–357 VNQIGSVTESLQACK
837.38 (3+) 285–385 DYPVVSIEDPFDQDDWGAWQK
846.95 (2+) 406–419 YNQLL(Cit)IEEELGSK
980.98 (2+) 202–220 DATNVGDEGGFAPNILENK
878.45 (3+) 5–27 IHA(Cit)EEIFDS(Cit)GNPTVEVDLFTSK
1,177.10 (2+) 372–393 SGETEDTFIADLVVGLCTGQIK
915.14 (3+) 202–227 DATNVGDEGGFAPNILENKEGLELLK
988.15 (3+) 256–280 YDLDFKSPDDPS(Cit)YISPDQLADLYK
1,017.04 (2+) 306–325 FTASAGIQVVGDDLTVTNPK
925.24 (4+) 126–161 GVPLY(Cit)HIADLAGNSEVILPVPAFNVINGGSHAGNK
Cit, citrulline.
Table 2 Peptides from non-citrullinated α-enolase sequenced by tandem mass spectrometry
m/z (charge) Location Matched sequence
401.24 (2+) 221–227 EGLELLK
403.73 (2+) 406–411 YNQLLR
452.75 (2+) 412–419 IEEELGSK
480.77 (2+) 81–88 LNVTEQEK
572.30 (2+) 183–192 IGAEVYHNLK
703.86 (2+) 15–27 GNPTVEVDLFTSK
713.34 (2+) 269–280 YISPDQADLYK
482.29 (3+) 80–91 KLNVTEQEKIDK
817.41 (2+) 343–357 VNQIGSVTESLQACK
785.09 (3+) 372–393 SGETEDTFIADLVVGLCTGQIK
915.14 (3+) 202–227 DATNVGDEGGFAPNILENKEGLELLK
753.66 (4+) 132–161 HIADLAGNSEVILPVPAFNVINGGSHAGNK
==== Refs
Lee DM Weinblatt ME Rheumatoid arthritis Lancet 2001 358 903 911 11567728 10.1016/S0140-6736(01)06075-5
Mageed RA van Venrooij WJ, Maini RN The RF antigen Manual of Biological Markers of Disease 1996 Dordrecht: Kluwer Academic Publishing 1 27
Nienhuis RL Mandema E A new serum factor in patients with rheumatoid arthritis; the antiperinuclear factor Ann Rheum Dis 1964 23 302 305 14178016
Young BJ Mallya RK Leslie RD Clark CJ Hamblin TJ Anti-keratin antibodies in rheumatoid arthritis Br Med J 1979 2 97 99 111762
Sebbag M Simon M Vincent C Masson-Bessiere C Girbal E Durieux JJ Serre G The antiperinuclear factor and the so-called antikeratin antibodies are the same rheumatoid arthritis-specific autoantibodies J Clin Invest 1995 95 2672 2679 7539459
Schellekens GA de Jong BA van den Hoogen FH van de Putte LB van Venrooij WJ Citrulline is an essential constituent of antigenic determinants recognized by rheumatoid arthritis-specific autoantibodies J Clin Invest 1998 101 273 281 9421490
Girbal-Neuhauser E Durieux JJ Arnaud M Dalbon P Sebbag M Vincent C Simon M Senshu T Masson-Bessiere C Jolivet-Reynaud C The epitopes targeted by the rheumatoid arthritis-associated antifilaggrin autoantibodies are posttranslationally generated on various sites of (pro)filaggrin by deimination of arginine residues J Immunol 1999 162 585 594 9886436
Nijenhuis S Zendman AJ Vossenaar ER Pruijn GJ van Venrooij WJ Autoantibodies to citrullinated proteins in rheumatoid arthritis: clinical performance and biochemical aspects of an RA-specific marker Clin Chim Acta 2004 350 17 34 15530456 10.1016/j.cccn.2004.07.016
van Gaalen FA Linn-Rasker SP van Venrooij WJ de Jong BA Breedveld FC Verweij CL Toes RE Huizinga TW Autoantibodies to cyclic citrullinated peptides predict progression to rheumatoid arthritis in patients with undifferentiated arthritis: a prospective cohort study Arthritis Rheum 2004 50 709 715 15022309 10.1002/art.20044
Rantapaa-Dahlqvist S de Jong BA Berglin E Hallmans G Wadell G Stenlund H Sundin U van Venrooij WJ Antibodies against cyclic citrullinated peptide and IgA rheumatoid factor predict the development of rheumatoid arthritis Arthritis Rheum 2003 48 2741 2749 14558078 10.1002/art.11223
Vencovsky J Machacek S Sedova L Kafkova J Gatterova J Pesakova V Ruzickova S Autoantibodies can be prognostic markers of an erosive disease in early rheumatoid arthritis Ann Rheum Dis 2003 62 427 430 12695154 10.1136/ard.62.5.427
Forslind K Ahlmen M Eberhardt K Hafstrom I Svensson B BARFOT Study Group Prediction of radiological outcome in early rheumatoid arthritis in clinical practice: role of antibodies to citrullinated peptides (anti-CCP) Ann Rheum Dis 2004 63 1090 1095 15308518 10.1136/ard.2003.014233
Masson-Bessiere C Sebbag M Durieux JJ Nogueira L Vincent C Girbal-Neuhauser E Durroux R Cantagrel A Serre G In the rheumatoid pannus, anti-filaggrin autoantibodies are produced by local plasma cells and constitute a higher proportion of IgG than in synovial fluid and serum Clin Exp Immunol 2000 119 544 552 10691929 10.1046/j.1365-2249.2000.01171.x
Vossenaar ER Nijenhuis S Helsen MM van der Heijden A Senshu T van den Berg WB van Venrooij WJ Joosten LA Citrullination of synovial proteins in murine models of rheumatoid arthritis Arthritis Rheum 2003 48 2489 2500 13130468 10.1002/art.11229
Hida S Miura NN Adachi Y Ohno N Influence of arginine deimination on antigenicity of fibrinogen J Autoimmun 2004 23 141 150 15324932 10.1016/j.jaut.2004.06.002
Rubin B Sonderstrup G Citrullination of self-proteins and autoimmunity Scand J Immunol 2004 60 112 120 15238080 10.1111/j.0300-9475.2004.01457.x
Vossenaar ER Despres N Lapointe E van der Heijden A Lora M Senshu T van Venrooij WJ Menard HA Rheumatoid arthritis specific anti-Sa antibodies target citrullinated vimentin Arthritis Res Ther 2004 6 R142 R150 15059278 10.1186/ar1149
Vossenaar ER Radstake TR van der Heijden A van Mansum MA Dieteren C de Rooij DJ Barrera P Zendman AJ van Venrooij WJ Expression and activity of citrullinating peptidylarginine deiminase enzymes in monocytes and macrophages Ann Rheum Dis 2004 63 373 381 15020330 10.1136/ard.2003.012211
Despres N Boire G Lopez-Longo FJ Menard HA The Sa system: a novel antigen-antibody system specific for rheumatoid arthritis J Rheumatol 1994 21 1027 1033 7932409
Nakashima K Hagiwara T Ishigami A Nagata S Asaga H Kuramoto M Senshu T Yamada M Molecular characterization of peptidylarginine deiminase in HL-60 cells induced by retinoic acid and 1α,25-dihydroxyvitamin D3 J Biol Chem 1999 274 27786 27792 10488123 10.1074/jbc.274.39.27786
Arnett FC Edworthy SM Bloch DA McShane DJ Fries JF Cooper NS Healey LA Kaplan SR Liang MH Luthra HS The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis Arthritis Rheum 1988 31 315 324 3358796
Shevchenko A Wilm M Vorm O Mann M Mass spectrometric sequencing of proteins silver-stained polyacrylamide gels Anal Chem 1996 68 850 858 8779443 10.1021/ac950914h
Wait R Gianazza E Eberini I Sironi L Dunn MJ Gemeiner M Miller I Proteins of rat serum, urine, and cerebrospinal fluid: VI. Further protein identifications and interstrain comparison Electrophoresis 2001 22 3043 3052 11565799 10.1002/1522-2683(200108)22:14<3043::AID-ELPS3043>3.0.CO;2-M
Wait R Miller I Eberini I Cairoli F Veronesi C Battocchio M Gemeiner M Gianazza E Strategies for proteomics with incompletely characterized genomes: the proteome of Bos taurus serum Electrophoresis 2002 23 3418 3427 12373772 10.1002/1522-2683(200210)23:19<3418::AID-ELPS3418>3.0.CO;2-7
Vittecoq O Incaurgarat B Jouen-Beades F Legoedec J Letourneur O Rolland D Gervasi G Menard JF Gayet A Fardellone P Autoantibodies recognizing citrullinated rat filaggrin in an ELISA using citrullinated and non-citrullinated recombinant proteins as antigens are highly diagnostic for rheumatoid arthritis Clin Exp Immunol 2004 135 173 180 14678280 10.1111/j.1365-2249.2004.02341.x
Vincent C Simon M Sebbag M Girbal-Neuhauser E Durieux JJ Cantagrel A Fournie B Mazieres B Serre G Immunoblotting detection of autoantibodies to human epidermis filaggrin: a new diagnostic test for rheumatoid arthritis J Rheumatol 1998 25 838 846 9598877
Palosuo T Lukka M Alenius H Kalkkinen N Aho K Kurki P Heikkila R Nykanen M von Essen R Purification of filaggrin from human epidermis and measurement of antifilaggrin autoantibodies in sera from patients with rheumatoid arthritis by an enzyme-linked immunosorbent assay Int Arch Allergy Immunol 1998 115 294 302 9566352 10.1159/000069460
Paimela L Palosuo T Aho K Lukka M Kurki P Leirisalo-Repo M von Essen R Association of autoantibodies to filaggrin with an active disease in early rheumatoid arthritis Ann Rheum Dis 2001 60 32 35 11114279 10.1136/ard.60.1.32
Goldbach-Mansky R Lee J McCoy A Hoxworth J Yarboro C Smolen JS Steiner G Rosen A Zhang C Menard HA Rheumatoid arthritis associated autoantibodies in patients with synovitis of recent onset Arthritis Res 2000 2 236 243 11056669 10.1186/ar93
Hayem G Chazerain P Combe B Elias A Haim T Nicaise P Benali K Eliaou J Kahn MF Sany J Meyer O Anti-Sa antibody is an accurate diagnostic and prognostic marker in adult rheumatoid arthritis J Rheumatol 1999 26 7 13 9918234
Moodie FD Leaker B Cambridge G Totty NF Segal AW Alpha-enolase: a novel cytosolic autoantigen in ANCA positive vasculitis Kidney Int 1993 43 675 681 8455367
Gitlits VM Sentry JW Matthew ML Smith AI Toh BH Autoantibodies to evolutionarily conserved epitopes of enolase in a patient with discoid lupus erythematosus Immunology 1997 92 362 368 9486109 10.1046/j.1365-2567.1997.00355.x
Pratesi F Moscato S Sabbatini A Chimenti D Bombardieri S Migliorini P Autoantibodies specific for alpha-enolase in systemic autoimmune disorders J Rheumatol 2000 27 109 115 10648026
Saulot V Vittecoq O Charlionet R Fardellone P Lange C Marvin L Machour N Le Loet X Gilbert D Tron F Presence of autoantibodies to the glycolytic enzyme alpha-enolase in sera from patients with early rheumatoid arthritis Arthritis Rheum 2002 46 1196 1201 12115223 10.1002/art.10252
Aaronson RM Graven KK Tucci M McDonald RJ Farber HW Non-neuronal enolase is an endothelial hypoxic stress protein J Biol Chem 1995 270 27752 27757 7499243 10.1074/jbc.270.46.27752
Fontan PA Pancholi V Nociari MM Fischetti VA Antibodies to streptococcal surface enolase react with human alpha-enolase: implications in poststreptococcal sequelae J Infect Dis 2000 182 1712 1721 11069244 10.1086/317604
Sugio K Sugaya M Hanagiri T Yasumoto K [Tumor marker in primary lung cancer] J UOEH 2004 26 473 479 15624359
Suzuki A Yamada R Chang X Tokuhiro S Sawada T Suzuki M Nagasaki M Nakayama-Hamada M Kawaida R Ono M Functional haplotypes of PADI4, encoding citrullinating enzyme peptidylarginine deiminase 4, are associated with rheumatoid arthritis Nat Genet 2003 34 395 402 12833157 10.1038/ng1206
McGraw WT Potempa J Farley D Travis J Purification, characterization, and sequence analysis of a potential virulence factor from Porphyromonas gingivalis, peptidylarginine deiminase Infect Immun 1999 67 3248 3256 10377098
|
16277695
|
PMC1297593
|
CC BY
|
2021-01-04 16:02:40
|
no
|
Arthritis Res Ther. 2005 Oct 19; 7(6):R1421-R1429
|
utf-8
|
Arthritis Res Ther
| 2,005 |
10.1186/ar1845
|
oa_comm
|
==== Front
Crit CareCritical Care1364-85351466-609XBioMed Central London cc37671627771510.1186/cc3767ResearchEfficiency of 7.2% hypertonic saline hydroxyethyl starch 200/0.5 versus mannitol 15% in the treatment of increased intracranial pressure in neurosurgical patients – a randomized clinical trial [ISRCTN62699180] Harutjunyan Lilit [email protected] Carsten [email protected] Andreas [email protected] Matthias [email protected] Stefan [email protected] Jens [email protected] Anaesthesiologist, Department of Anesthesia and Critical Care, Martin-Luther-University Halle-Wittenberg, Halle, Germany2 Neurosurgeon, Department of Neurosurgery, Martin-Luther-University Halle-Wittenberg, Halle, Germany3 Head, Department of Anesthesia and Critical Care, Klinikum Wolfsburg, Wolfsburg, Germany4 Professor of Anesthesiology and Pain Therapy, Department of Anesthesia and Critical Care, Martin-Luther-University Halle-Wittenberg, Halle, Germany5 Anaesthesiologist and Intensivist, Department of Anesthesia and Critical Care, Martin-Luther-University Halle-Wittenberg, Halle, Germany2005 9 8 2005 9 5 R530 R540 6 5 2005 6 6 2005 14 6 2005 17 6 2005 Copyright © 2005 Harutjunya 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
This prospective randomized clinical study investigated the efficacy and safety of 7.2% hypertonic saline hydroxyethyl starch 200/0.5 (7.2% NaCl/HES 200/0.5) in comparison with 15% mannitol in the treatment of increased intracranial pressure (ICP).
Methods
Forty neurosurgical patients at risk of increased ICP were randomized to receive either 7.2% NaCl/HES 200/0.5 or 15% mannitol at a defined infusion rate, which was stopped when ICP was < 15 mmHg.
Results
Of the 40 patients, 17 patients received 7.2% NaCl/HES 200/0.5 and 15 received mannitol 15%. In eight patients, ICP did not exceed 20 mmHg so treatment was not necessary. Both drugs decreased ICP below 15 mmHg (p < 0.0001); 7.2% NaCl/HES 200/0.5 within 6.0 (1.2–15.0) min (all results are presented as median (minimum-maximum range)) and mannitol within 8.7 (4.2–19.9) min (p < 0.0002). 7.2% NaCl/HES 200/0.5 caused a greater decrease in ICP than mannitol (57% vs 48%; p < 0.01). The cerebral perfusion pressure was increased from 60 (39–78) mmHg to 72 (54–85) mmHg by infusion with 7.2% NaCl/HES 200/0.5 (p < 0.0001) and from 61 (47–71) mmHg to 70 (50–79) mmHg with mannitol (p < 0.0001). The mean arterial pressure was increased by 3.7% during the infusion of 7.2% NaCl/HES 200/0.5 but was not altered by mannitol. There were no clinically relevant effects on electrolyte concentrations and osmolarity in the blood. The mean effective dose to achieve an ICP below 15 mmHg was 1.4 (0.3–3.1) ml/kg for 7.2% NaCl/HES 200/0.5 and 1.8 (0.45–6.5) ml/kg for mannitol (p < 0.05).
Conclusion
7.2% NaCl/HES 200/0.5 is more effective than mannitol 15% in the treatment of increased ICP. A dose of 1.4 ml/kg of 7.2% NaCl/HES 200/0.5 can be recommended as effective and safe. The advantage of 7.2% NaCl/HES 200/0.5 might be explained by local osmotic effects, because there were no clinically relevant differences in hemodynamic clinical chemistry parameters.
==== Body
Introduction
The development or presence of secondary brain injury in patients with intracranial pathology has been associated with increased morbidity and mortality. An increase in intracranial pressure (ICP) accompanied by a low cerebral perfusion pressure (CPP) should therefore be avoided in these patients. Several clinical studies have demonstrated that outcome is improved by adequate pharmacological or neurosurgical treatment optimizing ICP [1-3]. According to established treatment guidelines, an ICP >20 mmHg and a CPP <60 mmHg are considered critical [4-8]. Early recognition of such critical episodes by multimodal neuromonitoring, and selection of an effective and safe drug for treatment are essential for neuroprotection.
Osmotherapy has been used since the early 20th century to treat increased ICP. The physiological basis and concept of osmotherapy was first published in 1919 [9]. Intravenous infusion of mannitol is considered to be the 'gold standard' for the treatment of increased ICP. Barbiturates and TRIS buffer are still used as alternative treatments, although their use in clinical practice is limited by cardiovascular and metabolic side effects [10-13]. In addition, experimental and clinical evidence has shown that 'small volume resuscitation' has a positive effect in the treatment of increased ICP in trauma patients [14-16].
Experimentally, intravenous application of hypertonic saline increases global cerebral perfusion as well as the right-shifted oxygen dissociation curve, both with consecutive improvement of oxygen delivery. At the same time, an increase of cerebral compliance and decrease in ICP occur by decrease of the brain edema [17].
Although several experimental and clinical studies have investigated the effects of hypertonic saline or mannitol on ICP, only a few studies comparing these drugs in neurosurgical patients have been published [18-22]. Furthermore, there are no clinical data available for recommendation of an 'effective dose' of hypertonic saline in clinical practice.
The purpose of this study was to compare the efficacy and safety of 7.2% NaCl/HES 200/0.5 and mannitol 15% in neurosurgical patients with increased ICP. This study focuses on the effects of both drugs on ICP, CPP, mean arterial pressure (MAP), hematocrit, serum sodium and osmolarity. Furthermore, we attempted to recommend an effective dose for the application of hypertonic saline.
Methods
After approval by the local ethics committee and written informed consent being obtained from the patients' legal relatives, neurosurgical patients with severe neuronal damage (e.g. cerebral trauma, spontaneous intracerebral bleeding or subarachnoidal bleeding) were enrolled in this prospective randomized study. The patients were randomized to receive either 7.2% NaCl/HES 200/0.5 (HyperHAES®, Fresenius Kabi Deutschland GmbH, Bad Homburg) or mannitol (Osmofundin® 15%-N, B. Braun Melsungen AG, Melsungen, Germany), to treat increased ICP.
Inclusion criteria were: age >18 years, severe brain damage (Glasgow Coma Score <8) with cerebral edema – visualized by CT scan and continuous monitoring of ICP. Exclusion criteria were: elevated ICP due to space-occupying lesions with indication for neurosurgical intervention (e.g. bleeding, hydrocephalus), severe renal failure, metabolic disorders, initial serum sodium >150 mmol/l and initial serum osmolarity >320 mosm/kg.
Standard treatment protocol
All patients were intubated and received pressure-controlled mechanical ventilation (Bilevel Positive Airway Pressure (BiPAP), etCO2 4.2–4.8 kPa, FiO2 0.3–1.0). Care was taken to keep the arterial partial oxygen pressure above 15 kPa, the hemoglobin concentration above 5.5 mmol/l and the CPP above 70 mmHg. If necessary, blood pressure was supported with vasopressor therapy. Blood glucose was adjusted to values between 6–8 mmol/l by continuous application of human insulin. Patients' core temperature was measured via the bladder, with a target temperature of 36.0–37.0°C. If the core temperature exceeded 37.0°C, external cooling blankets were used to cool the patient, otherwise patients were covered either with an additional blanket or with an active heating blanket (Bair Hugger; Augustine Medical, Eden Prairie, MN, USA).
Analgosedation and continuous patient monitoring were managed according to the standards of the Department of Anesthesiology and Critical Care at the Martin-Luther-University Halle-Wittenberg, Germany. Analgosedation at days 1–4 was performed using propofol and sufentanil or remifentanil. Thereafter, midazolam and sufentanil were administered. The standard monitoring included electrocardiogram, invasive arterial blood pressure, central venous pressure, peripheral oxygen saturation (SpO2) and intraparenchymal ICP measurement (Codman Microsensor ICP Monitoring System; Codman & Shurtleff Inc, Raynham, MA, USA).
An increase in ICP was treated first by deepening the sedation and analgesia by titrating the medication and adjusting to adequate ventilator settings. If ICP exceeded the 20 mmHg threshold for more than 5 min, the study medication (mannitol or 7.2% NaCl/HES 200/0.5 (herein referred to as '7.2% hypertonic saline' or 'hypertonic saline') was infused via the central venous line using an automated infusion system at a defined infusion rate. The infusion was stopped when ICP was reduced to <15 mmHg, defined as the treatment goal. However, in the case of sustained ICP problems (ICP >15 mmHg or CPP <70 mmHg) after these measures, bolus applications of thiopentone (maximum single bolus: 5 mg/kg) were allowed. In these patients, the possibility of a space-occupying lesion was excluded by CT scan.
Data acquisition and statistical analysis
Mean arterial blood pressure, heart rate, SpO2, ICP and calculated CPP were continuously measured. Analysis of these parameters was performed at the following time points: initiation of infusion; after termination of infusion (ICP <15 mmHg achieved); 10 min after terminating infusion; 30 min after terminating infusion; and 60 min after terminating infusion. Serum sodium level and hematocrit were measured every 4 h and the serum osmolarity every 12 h. The values taken before the therapy, as well as the maximum values subsequently achieved, were analyzed. Individual outcomes were assessed at the end of stay in the intensive care unit (ICU) using the differentiation between survivors and non-survivors.
The random code for group assignment was generated by computer. The software package Stat View 4.0 (Abacus Concepts Inc, Berkeley, CA, USA) was used for all statistical calculations. All demographic data are presented as mean ± SD. The clinical values in both groups were not normally distributed. Results are presented as median (minimum-maximum range). Groups were compared using the non-parametric Mann-Whitney U-Test and the Wilcoxon Signed Rank was employed to analyze the effect of the medication used within each group; p < 0.05 was regarded as statistically significant and computed significance levels are given.
Results
A total of 40 neurosurgical patients were recruited according to the inclusion criteria and randomized to receive either 7.2% NaCl/HES 200/0.5 (n = 17) or mannitol 15% (n = 15) to treat increased ICP. Only 32 patients were evaluated since in eight patients, ICP did not exceed 20 mmHg, therefore no study medication was administered.
Demographic data of all analyzed patients are summarized in Table 1. There were no significant differences between the two groups. No relevant clinical characteristics were revealed in the eight patients not undergoing osmotic therapy.
Analgosedation was started in all patients using our standard protocol. In four patients in the 7.2% hypertonic saline group and five patients in the mannitol group, propofol was substituted by thiopental because of sustained ICP problems.
Heart rate and blood pressure
The average baseline heart rate was 78 (58–95) bpm in the mannitol and 76 (52–92) bpm in the hypertonic saline group (p = NS). The infusion of study medication produced no clinically relevant changes in heart rate and no arrhythmias.
The initial MAP was 84 (68–92) mmHg in the mannitol group and 82 (64–98) mmHg in the hypertonic saline group (p = NS). Maximal changes could be analyzed in the mannitol group after 10 min (83 (69–105) mmHg) and in patients receiving hypertonic saline after 30 min (85 (74–98) mmHg) (Fig. 1, Table 2).
The individual maximum increase of MAP during the observation time after infusion of mannitol was 5.8% to 88 (72–106) mmHg and after infusion of hypertonic saline was 7.6% to 85 (74–98) mmHg. The time of the maximal increase was individual for each patient as well.
ICP and CPP
Prior to administration of the study medication, the mean ICP was 23 (19–30) mmHg in the mannitol group and 22 (19–31) mmHg in the hypertonic saline group (p = NS). After infusion with mannitol, the ICP decreased to 14 (7–20) mmHg and after infusion with hypertonic saline it decreased to 15 (8–18) mmHg (p < 0.0001). This effect was achieved within 8.7 (4.2–19.9) min by mannitol and 6.0 (1.2–15.0) min by hypertonic saline (p < 0.0002) and maintained over the 1 h observation period. The lowest ICP was 12 (6–19) mmHg in the mannitol and 10 (6–14) mmHg in the hypertonic saline group (p < 0.05), observed 30 min after the end of infusion. Thus, the maximum decrease in ICP produced by hypertonic saline was 57% and that of mannitol 48%. Sixty minutes after the end of infusion, the ICP in the hypertonic saline group was still lower than that of the mannitol group (11 (5–18) mmHg; vs 14 (7–20) mmHg; p < 0.005) (Fig. 2, Table 2).
Prior to administration of study medication, the mean CPP was 61 (47–71) mmHg in the mannitol and 60 (39–78) mmHg in the hypertonic saline group (p = NS; Fig. 3). At the end of infusion, a significant increase of CPP to 70 (50–79) mmHg after mannitol infusion (p < 0.0001) and 72 (54–85) mmHg after hypertonic saline infusion (p < 0.0001) occurred. This improvement was maintained during the whole study period. The maximal increase in CPP occurred in both groups after 30 min (mannitol +18%; hypertonic saline +27%; p < 0.05). CPP was significantly higher in the hypertonic saline group (p < 0.01, Fig. 3, Table 2) 30 and 60 min after the end of infusion.
The 15 patients in the mannitol group had a total of 53 episodes of increased ICP exceeding 20 mmHg requiring infusion of study medication (3.5 treatments/patient). For 49 of these episodes (92.5%), infusion of mannitol was effective and reduced ICP to <15 mmHg within 8.7 (4.2–19.9) min. For one episode, mannitol produced a delayed effect, appearing 20 min after application of a total of 235 ml mannitol (2.6 ml/kg). In three episodes, however, ICP could not be reduced below 15 mmHg by an infusion of up to 2.1 ml/kg of mannitol. In two of these patients, thiopental was given intravenously at up to 3 mg/kg and in one patient a unilateral decompressive craniectomy was performed.
In the 17 patients in the hypertonic saline group, 57 periods of increased ICP occurred (3.3 treatments/patient). 7.2% NaCl/HES 200/0.5 was effective in 55 episodes (96.5%), reducing ICP to <15 mmHg within 6.0 (1.2–15.0) min. In one episode, hypertonic saline (3 ml/kg) was only effective after an additional bolus of thiopental 3 mg/kg was given and, in another episode, ICP could not be reduced below 15 mmHg by an infusion of up to 3.1 ml/kg of hypertonic saline. Finally, mild hyperventilation (etCO2 ~28–30 mmHg) achieved the target ICP value <15 mmHg.
The median dose of mannitol (145 (70–332) ml/application; 1.8 (0.45–6.5) ml/kg) required to reduce ICP below 15 mmHg was significantly higher than that of hypertonic saline (100 (35–250) ml/application; 1.4 (0.3–3.1) ml/kg). Repeated administration of mannitol caused an increase of the required single dose in six out of 15 patients (40%) and a decrease in two patients (13%). Repeated administration of hypertonic saline caused an increase of the required single dose in two patients (12%) and a decrease in seven patients (41%).
Clinical chemistry
Hematocrit was not significantly changed by infusion of mannitol (0.3 (0.27–0.42) vs 0.29 (0.26–0.40)) and hypertonic saline (0.29 (0.24–0.37) vs 0.29 (0.24–0.36)). A temporary, but statistically significant increase of serum sodium occurred after infusion of the hypertonic saline from 143 (136–148) mmol/l to 148 (144–153) mmol/l (p < 0.001). Serum osmolarity increased significantly after infusion of hypertonic saline: 284 (273–300) mosm/kg to 300 (284–319) mosm/kg (p < 0.001), as well as after infusion of mannitol: 286 (270–315) mosm/kg to 295 (278–327) mosm/kg (p < 0.001).
Outcome
Ten patients (58.8%) assigned to the group receiving hypertonic saline survived, the remaining seven patients died (41.2%). In the group with the mannitol treatment, six patients survived (40.0%) and nine patients died (60.0%). The chi-square test revealed no statistical significance.
In patients who survived, a lower dose of the osmotic agent had been administered. Survivors in the hypertonic saline group received a significant lower dose of 1.4 (0.32–2.8) ml/kg hypertonic saline. In non-survivors, the dosage given was 1.7 (0.9–3.1) ml/kg (p < 0.05). In the mannitol group, patients who survived received 1.7 (0.5–3.4) ml/kg mannitol versus 1.9 (1.0–6.5) ml/kg mannitol in patients who died (p = NS). Therefore, a statistical significance regarding the influence of the specific osmolarity, either of hypertonic saline or mannitol, given with each treatment, on changes of the cerebral hemodynamics (ICP, CPP) or patients' individual outcomes could not be analyzed.
Discussion
The strong relationship between incidence of increased ICP and outcome in patients with neuronal damage emphasizes the vulnerability of the injured brain and the need for adequate treatment. The management of severely injured neurosurgical patients has changed over recent decades, especially regarding the introduction and acceptance of clinical guidelines among neurosurgeons and intensivists [4,10,23,24]. It has become a generally accepted treatment goal to keep the CPP above 70 mmHg, because episodes of CPP <60 mmHg or ICP >20 mmHg are associated with a worse outcome [6-8]. These goals are incorporated into current treatment protocols, which are constantly analyzed with regards to their efficacy and feasibility, and updated accordingly. Osmotic agents are important components of all treatment protocols, especially mannitol as it is a well-established treatment for increased ICP following brain injury. Surveys of the critical care management of head-injured patients show that 83% of the centers in the United States and 100% of the centers in the United Kingdom used mannitol to control ICP [25-27]. The clinical use of mannitol is, however, limited by renal complications and the fast increase of the osmotic gradient followed by its reversal due to disruption of the blood-brain barrier (BBB) [28-31]. Furthermore, mannitol (at concentrations which may be reached in clinical conditions) and the hyperosmotic stress itself can activate the process of apoptotic cell death [32].
Recent data have demonstrated different osmotic effects of mannitol. Videen and co-workers [33] observed that after administration of 1.5 g/kg bolus of mannitol in six patients with acute complete middle cerebral artery infarctions, the brain in the non-infarcted hemisphere shrank more than in the infarcted hemisphere. This may increase the inter-hemispheric pressure difference and worsen tissue shift [33].
Hypertonic saline is an interesting alternative to mannitol, because there is experimental and clinical evidence that it can reduce ICP and improve CPP [34-39]. Experimental studies in animals suffering from a combination of hemorrhagic shock and head trauma demonstrated a significant reduction of ICP, an improvement of CPP and/or a reduction of brain edema [34-36,40,41].
The efficacy of hypertonic saline after isolated brain injury, however, has rarely been investigated. Qureshi et al. [22] examined different concentrations of hypertonic saline (23.4%, 3.0%) versus mannitol after isolated experimental intracerebral hemorrhage in a canine model. The acute effects on ICP and CPP were most prominent after infusion of hypertonic saline 23.4%, but were better sustained after infusion of hypertonic saline 3%. The water content was highest after mannitol infusion in most regions of the brain, especially in the white matter ipsilateral to the hematoma. The authors speculated that these results were due to a certain permeability of the BBB. The most positive effect on water content was seen after hypertonic saline 3% [22].
Berger et al. [42] compared the efficacy of hypertonic saline and mannitol to reduce ICP after a combination of two different neuronal injuries. Initially, a cold-induced focal lesion was used to induce a vasogenic brain edema in rabbits, then intracranial hypertension was induced by a further inflation of an epidural balloon. The authors demonstrated that hypertonic solution as well as mannitol can reduce the ICP efficiently. After the first application, the effect of mannitol was enhanced compared with the hypertonic solution (98 ± 14 min vs 189 ± 27 min; p < 0.054), but became the same after repeated applications. It is remarkable that mannitol was more effective in decreasing the water content in brain tissue in the traumatized hemisphere, whereas hypertonic solution lowered the water content in the contralateral brain tissue. An accumulation of mannitol could occur, followed by a possible reversal of the local osmotic gradient. These different effects on brain tissue could be an explanation for the failed therapeutic efficiency after mannitol and emphasized the advantages of hypertonic solutions [42]. Furthermore, Prough et al. observed a higher regional cerebral blood flow in dogs with induced intracerebral hemorrhage after hypertonic saline without any increase of the CPP [43].
The positive effect of 7.2% hypertonic saline on ICP has also been demonstrated in several clinical studies investigating patients with therapy-refractory ICP increase due to isolated brain injury but without hemorrhagic shock [21,44-46]. Hypertonic saline had no effects on MAP in these euvolaemic patients [46].
Schwarz et al. [47] evaluated the efficacy of hypertonic saline hydroxyethyl starch 7.55% in comparison with mannitol 20% in stroke patients with increased ICP. Hypertonic saline hydroxyethyl starch was effective in all, mannitol in only 70% of patients. The maximum ICP decrease was seen 25 min after the start of hypertonic saline infusion and 45 min after the start of mannitol infusion. There was no constant effect on CPP in the hypertonic saline group, whereas CPP rose significantly in the mannitol-treated group. The authors concluded that hypertonic saline hydroxyethyl starch seems to lower ICP more effectively but does not increase CPP as much as mannitol [47].
Hypertonic saline has also been used to reduce ICP in patients with brain tumors or subarachnoid hemorrhage. Suarez et al. [48] reported a significant decrease of ICP and increase of CPP in these patients, when application of mannitol had been previously unsuccessful. Similar results were observed by Horn et al. in patients with traumatic brain injury and subarachnoidal hemorrhage, where hypertonic saline 7.5% adequately reduced ICP after mannitol therapy had failed [44].
Based on these findings, patients with isolated head trauma can also be expected to benefit from hypertonic saline. This patient population covers some specific patho-physiological conditions, characterized by diffuse axonal injuries, hemorrhages, and necrotic and edematous tissue, which can lead to different therapeutic strategies and a failed positive effect of hypertonic saline compared with patients with other intracranial mass lesions [49,50]. Munar et al. [51] investigated the acute effects of 7.2% hypertonic saline on ICP, cerebral blood flow and systemic hemodynamics in patients with moderate and severe traumatic brain injury during the first 72 h after injury. Hypertonic saline significantly reduces ICP without changes in MAP and relative global cerebral blood flow, expressed as 1/AVDO2. These results suggest that hypertonic saline decreases ICP by means of an osmotic mechanism [51].
Not all studies, however, reported positive effects of hypertonic saline on ICP, especially if hypertonic saline was infused continuously. Qureshi et al. analyzed the effect of continuous administration of hypertonic saline 2% or 3% in patients with head trauma. They reported a higher in-hospital mortality rate in patients receiving hypertonic solutions and described no favorable impact on the rate of necessary medical interventions during the patient's treatment in the ICU. The influence of hypertonic saline on the supposedly disrupted BBB after head injury was mainly used to explain the failed effect. A disrupted BBB can lead to an accumulation of sodium resulting in an reversal of the osmotic gradient with concomitant increase of ICP [52]. However, Hartl et al. demonstrated a reduced water content in areas with a disturbed BBB in a model with or without a focal cryogenic brain lesion and hemorrhagic shock [53].
Our results showed that bolus application of either study medication, mannitol 15% or hypertonic saline 7.2%, significantly decreases ICP and increases CPP (Table 2). The effect of hypertonic saline on ICP was significantly better than that of mannitol. Clinically important effects of both drugs on MAP could not be determined, although some statistically significant differences were observed at a few measurement points. Therefore, it can be concluded that local cerebral dehydration is the main mechanism of both substances in decreasing ICP and optimizing CPP. The higher potency of hypertonic saline suggests that its local effect is more clearly pronounced.
However, the mechanisms whereby hypertonic solutions reduce ICP are multifactorial and are still discussed with some controversy. The main principle seems to be the 'local dehydration' of brain tissue drawing water from parenchyma to the intravascular space following an osmotic gradient [54]. Comparing this with the osmotic effect of mannitol, a second mechanism to explain the effect of the ICP-reduction must exist. This hypothesis is supported by the results of Berger et al. [42]. He found, in rats with induced head injury, a similar positive effect on ICP with regards to the amount and duration of the decrease, but a higher CPP in the rats receiving mannitol. Contrary to our results, the MAP increased after hypertonic saline, whereas the MAP temporarily decreased after mannitol. The authors hypothesized that the different effects of the two solutions are the result of a selective permeability of the BBB and/or the different reflection coefficients. A disrupted BBB would have to be the result of an accumulation of both solutions in the brain tissue. Therefore different mechanisms of local cerebral dehydration must exist [42]. These hypotheses are supported by the results of Worthley et al. and Kaufmann et al. Both working groups demonstrated that the ICP-decreasing effect is limited after repeated bolus applications of mannitol, but a further application of hypertonic saline lead to a further ICP reduction [55,56]. However, a direct vasodilatation of pial vessels [57-59], the reduction of blood viscosity due to enhancement of the intravascular volume, the rapid absorption of cerebrospinal fluid and restoration of the normal membrane potentials are other effects to positively affect the ICP [60,61]. Our results only support the hypothesis about the local dehydration of brain tissue. Systemic hemodynamic effects for the given dosage couldn't be demonstrated, but the decreased ICP leads to the improved CPP. All homeostatic side effects after hypertonic saline, for instance hypernatriemia and increased serum osmolarity, are temporary and without systemic hemodynamic side effects. Such complications as described in the literature, emphasize cardiac failure with lung edema, metabolic acidosis, coagulopathia subdural hematoma and central pontine myolysis as the most important [22,40,48]. With the intention of limiting the side effects of changes in electrolytes and osmolarity, a standardized laboratory measurement procedure is needed.
The substantial difference in the design of the present and a comparable study is the fact that we did not administer a fixed total dose, but infused the study medication at a defined infusion rate until ICP decreased to <15 mmHg, the primary goal of our treatment. No clinical study has so far identified an exact dose-effect relationship for hypertonic saline. Only one comparable clinical study confirms the superiority of 2.0 ml/kg of hypertonic saline 7.5% over mannitol 20% in head-injured patients [21]. This study concluded that 2 ml/kg of 7.2% NaCl/HES 200/0.5 can be recommended as an effective dose to reduce increased ICP [21]. In our study, an average dose of 1.5 ± 0.6 ml/kg of hypertonic saline adequately reduced ICP below 15 mmHg. Furthermore, because of our application mode with an defined application rate and a target ICP of <15 mmHg we could demonstrate a failed influence of the osmotic load given with each treatment.
Regardless of all positive effects in our study, there are some limitations that need to be discussed, most of all, the small patient population of each group and the heterogeneity in the underlying neurological illness. The primary intention of our study was pragmatic and adjusted on the typical clinical routine. However, we included neurosurgical patients with severe neuronal damage independent from the individual pathogenesis. To compensate for this to a certain degree, we used a randomized study design. Furthermore, until now there have been only limited data available for comparison of these two osmotic agents in a clinical setting. A small amount of evidence is available that hypertonic saline has some advantages compared with mannitol in the treatment of patients with intracranial hypertension after trauma, subarachnoid bleeding or stroke [21,47,62,63].
Conclusion
7.2% NaCl/HES 200/0.5 and mannitol 15% are effective and safe drugs in the treatment of increased ICP, although 7.2% NaCl/HES 200/0.5 is more effective than mannitol. A dose of 1.4 ml/kg can be recommended as an initial dose. The advantage of hypertonic saline can be explained by individual local osmotic effects, because no relevant systemic changes occur. The observed effects on electrolytes and plasma osmolarity are not significantly different between the two osmotic drugs and have no clinical relevance here. Further experimental and clinical research is required to evaluate the optimal administration regime, the best treatment strategies adapted to the individual patient's needs and the impact on patients' morbidity and mortality.
Key messages
• 7.2% NaCl/HES 200/0.5 is more effective than mannitol in the treatment of increased ICP
• A dose of 1.4 ml/kg 7.2% NaCl/HES 200/0.5 can be recommended as an initial dose
• The local dehydration of brain tissue after application of 7.2% NaCl/HES 200/0.5 seems to be the primary mechanism for the improved CPP
Abbreviations
BBB = blood-brain barrier; CPP = cerebral perfusion pressure; GCS = Glasgow Coma Score; ICH = intracerebral hemorrhage; ICU = intensive care unit; SAH = subarachnoid hemorrhage; SAPS = simplified acute physiology score; SHT = severe head trauma; SpO2 = peripheral oxygen saturation
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
All of the authors were involved in designing the study and collecting data. JS and LH were involved in the statistical analysis. SG revised the article and was responsible for translation into English. All authors read and approved the final manuscript.
Acknowledgements
The authors are grateful to the intensive care nursing staff who cared for the patients and followed the study protocol.
Figures and Tables
Figure 1 Box-and-whisker plots of the MAP. Data are plotted for the first hour after administration of 7.2% NaCl/HES 200/0.5 (HS) or mannitol 15% (M). In patients receiving 7.2% NaCl/HES 200/0.5, the MAP change was statistically significant compared with the value at the start of treatment († p < 0.05). The changes with mannitol were not statistically significant within the group, but significant after 30 min to HS (*p < 0.05). MAP, mean arterial pressure.
Figure 2 Box-and-whisker plots of the ICP. Data are plotted for the first hour after intravenous administration of 7.2% NaCl/HES 200/0.5 (HS) or mannitol (M). The ICP decreases after injection of the respective test substance significantly in comparison with the baseline value at the start of treatment († p < 0.0001). After 30 min and 60 min, a statistically significant difference was seen between the two treatment regimes (p < 0.05) ICP, intracranial pressure.
Figure 3 Box-and-whisker plots of the mean CPP. Data are plotted within the first hour after administration of 7.2% NaCl/HES 200/0.5 (HS) or mannitol (M). The CPP increases significantly compared with the start of treatment († p < 0.0001). After 30 min and 60 min, a statistically significant difference was seen between the two treatment regimes (p < 0.01). CPP, cerebral perfusion pressure.
Table 1 Demographic data of analyzed patients
Mannitol 15% (n = 15) 7.2% NaCl/HES 200/0.5 (n = 17)
Age 47 ± 16 47 ± 16
Weight 89 ± 27 87 ± 24
Gender, M/F 8/7 9/8
Initial GCS 5.8 ± 1.4 6 ± 1.3
SAPS score 42.5 ± 13 39.6 ± 9.6
Days on ICU 23.3 ± 14.8 22.8 ± 15.5
Basic illness
SAH 5 4
Brain infarct 4 3
Isolated SHT III° 4 6
ICH 1 3
Other 1 1
Surgical intervention 13 13
7.2% NaCl/HES 200/0.5, 7.2% hypertonic saline hydroxyethyl starch 200/0.5; GCS, Glasgow Coma Score; ICH, intracerebral hemorrhage; ICU, intensive care unit; SAPS, simplified acute physiology score; SHT, severe head trauma.
Table 2 Time course of heart rate, MAP, ICP and the CPP for the two different treatment groups
Start infusion Terminating infusion +10 min +30 min +60 min
Heart rate, l/min
7.2% NaCl/HES 200/0.5 76 [52–92] 78 [60–104] 77 [62–107] 78 [62–101] 79 [61–99]
Mannitol 15% 78 [58–95] 80 [58–96] 80 [60–95] 81 [58–93] 79 [56–96]
MAP, mmHg
7.2% NaCl/HES 200/0.5 84 [64–98] 84* [68–96] 84* [67–97] 85* [74–100] 84 [63–94]
Mannitol 15% 84 [68–92] 85 [65–98] 83 [69–105] 81 [69–106] 82 [68–108]
ICP, mmHg
7.2% NaCl/HES 200/0.5 22 [19–31] 15** [8–18] 12** [2–16] 10**,++ [6–14] 11**,+ [5–18]
Mannitol 15% 23 [19–30] 14** [7–20] 13** [4–19] 12** [6–19] 14** [7–20]
CPP, mmHg
7.2% NaCl/HES 200/0.5 60 [39–78] 72** [54–85] 72** [55–89] 75**, #[62–86] 73**, #[58–88]
Mannitol 15% 61 [47–71] 70** [50–79] 70** [56–92] 72** [60–93] 69** [56–89]
*p < 0.05, **p < 0.0001 compared with start infusion. +p < 0.0001, ++p < 0.01, #p < 0.05 between treatment regimes. HR, heart rate; CPP, cerebral perfusion pressure; ICP, intracranial pressure; MAP, mean arterial pressure.
==== Refs
Lang EW Chesnut RM Intracranial pressure and cerebral perfusion pressure in severe head injury New Horiz 1995 3 400 409 7496748
Chesnut RM Medical management of severe head injury: present and future New Horiz 1995 3 581 593 7496770
Caroli M Locatelli M Campanella R Balbi S Martinelli F Arienta C Multiple intracranial lesions in head injury: clinical considerations, prognostic factors, management, and results in 95 patients Surg Neurol 2001 56 82 88 11580939 10.1016/S0090-3019(01)00540-7
Procaccio F Stocchetti N Citerio G Berardino M Beretta L Della Corte F D'Avella D Brambilla GL Delfini R Servadei F Tomei G Guidelines for the treatment of adults with severe head trauma (part II). Criteria for medical treatment J Neurosurg Sci 2000 44 11 18 10961491
Vukic M Negovetic L Kovac D Ghajar J Glavic Z Gopcevic A The effect of implementation of guidelines for the management of severe head injury on patient treatment and outcome Acta Neurochir (Wien) 1999 141 1203 1208 10592121 10.1007/s007010050419
Rosner MJ Rosner SD Johnson AH Cerebral perfusion pressure: management protocol and clinical results J Neurosurg 1995 83 949 962 7490638
Rosner MJ Introduction to cerebral perfusion pressure management Neurosurg Clin N Am 1995 6 761 773 8527917
Rosner MJ Daughton S Cerebral perfusion pressure management in head injury J Trauma 1990 30 933 940 discussion 940-931 2117669
Weed LH McKibben PS Experimental alteration of brain bulk Am J Physiol 1919 48 531 558
Bullock R Chesnut RM Clifton G Ghajar J Marion DW Narayan RK Newell DW Pitts LH Rosner MJ Wilberger JW Guidelines for the management of severe head injury. Brain Trauma Foundation Eur J Emerg Med 1996 3 109 127 9028756
McMullen JE Aoki H Brown EB JrKittle CF Hemodynamic effects of tris (hydroxymethyl) aminomethane Surg Forum 1965 16 188 191 5835142
Traeger SM Henning RJ Dobkin W Giannotta S Weil MH Weiss M Hemodynamic effects of pentobarbital therapy for intracranial hypertension Crit Care Med 1983 11 697 701 6884049
The Brain Trauma Foundation. The American Association of Neurological Surgeons. The Joint Section on Neurotrauma and Critical Care. Use of barbiturates in the control of intracranial hypertension J Neurotrauma 2000 17 527 530 10937896
Nakayama S Sibley L Gunther RA Holcroft JW Kramer GC Small-volume resuscitation with hypertonic saline (2,400 mOsm/liter) during hemorrhagic shock Circ Shock 1984 13 149 159 6744520
Rocha-e-Silva M Negraes GA Soares AM Pontieri V Loppnow L Hypertonic resuscitation from severe hemorrhagic shock: patterns of regional circulation Circ Shock 1986 19 165 175 3719918
Maningas PA Resuscitation with 7.5% NaCl in 6% dextran-70 during hemorrhagic shock in swine: effects on organ blood flow Crit Care Med 1987 15 1121 1126 2445530
Kempski O Obert C Mainka T Heimann A Strecker U "Small volume resuscitation" as treatment of cerebral blood flow disturbances and increased ICP in trauma and ischemia Acta Neurochir Suppl 1996 66 114 117 8780808
Berger S Schurer L Hartl R Deisbock T Dautermann C Murr R Messmer K Baethmann A 7.2% NaCl/10% dextran 60 versus 20% mannitol for treatment of intracranial hypertension Acta Neurochir Suppl (Wien) 1994 60 494 498 7526628
De Vivo P Del Gaudio A Ciritella P Puopolo M Chiarotti F Mastronardi E Hypertonic saline solution: a safe alternative to mannitol 18% in neurosurgery Minerva Anestesiol 2001 67 603 611 11731749
Erard AC Walder B Ravussin P [Effects of equiosmolar load of 20% mannitol, 7.5% saline and 0.9% saline on plasma osmolarity, haemodynamics and plasma concentrations of electrolytes] Ann Fr Anesth Reanim 2003 22 18 24 12738015
Vialet R Albanese J Thomachot L Antonini F Bourgouin A Alliez B Martin C Isovolume hypertonic solutes (sodium chloride or mannitol) in the treatment of refractory posttraumatic intracranial hypertension: 2 mL/kg 7.5% saline is more effective than 2 mL/kg 20% mannitol Crit Care Med 2003 31 1683 1687 12794404 10.1097/01.CCM.0000063268.91710.DF
Qureshi AI Wilson DA Traystman RJ Treatment of elevated intracranial pressure in experimental intracerebral hemorrhage: comparison between mannitol and hypertonic saline Neurosurgery 1999 44 1055 1063 discussion 1063-1054 10232539 10.1097/00006123-199905000-00064
Procaccio F Stocchetti N Citerio G Berardino M Beretta L Della Corte F D'Avella D Brambilla GL Delfini R Servadei F Tomei G Guidelines for the treatment of adults with severe head trauma (part I). Initial assessment; evaluation and pre-hospital treatment; current criteria for hospital admission; systemic and cerebral monitoring J Neurosurg Sci 2000 44 1 10 10961490
Bulger EM Nathens AB Rivara FP Moore M MacKenzie EJ Jurkovich GJ Management of severe head injury: institutional variations in care and effect on outcome Crit Care Med 2002 30 1870 1876 12163808 10.1097/00003246-200208000-00033
Ghajar J Hariri RJ Narayan RK Iacono LA Firlik K Patterson RH Survey of critical care management of comatose, head-injured patients in the United States Crit Care Med 1995 23 560 567 7874910 10.1097/00003246-199503000-00023
Jeevaratnam DR Menon DK Survey of intensive care of severely head injured patients in the United Kingdom BMJ 1996 312 944 947 8616307
Matta B Menon D Severe head injury in the United Kingdom and Ireland: a survey of practice and implications for management Crit Care Med 1996 24 1743 1748 8874315 10.1097/00003246-199610000-00023
Garcia-Sola R Pulido P Capilla P The immediate and long-term effects of mannitol and glycerol. A comparative experimental study Acta Neurochir (Wien) 1991 109 114 121 1907077 10.1007/BF01403005
Kofke WA Tempelhoff R Increased intracranial pressure in head trauma patients given fentanyl or sufentanil Anesthesiology 1993 78 620 621 8457073
Node Y Yajima K Nakazawa S [Rebound phenomenon of mannitol and glycerol: clinical studies] No To Shinkei 1983 35 1241 1246 6422967
Oken DE Renal and extrarenal considerations in high-dose mannitol therapy Ren Fail 1994 16 147 159 8184141
Famularo G The puzzle of neuronal death and life: is mannitol the right drug for the treatment of brain oedema associated with ischaemic stroke? Eur J Emerg Med 1999 6 363 368 10646927
Videen TO Zazulia AR Manno EM Derdeyn CP Adams RE Diringer MN Powers WJ Mannitol bolus preferentially shrinks non-infarcted brain in patients with ischemic stroke Neurology 2001 57 2120 2122 11739839
Gunnar W Jonasson O Merlotti G Stone J Barrett J Head injury and hemorrhagic shock: studies of the blood brain barrier and intracranial pressure after resuscitation with normal saline solution, 3% saline solution, and dextran-40 Surgery 1988 103 398 407 2451302
Gunnar W Kane J Barrett J Cerebral blood flow following hypertonic saline resuscitation in an experimental model of hemorrhagic shock and head injury Braz J Med Biol Res 1989 22 287 289 2790301
Gunnar WP Merlotti GJ Barrett J Jonasson O Resuscitation from hemorrhagic shock. Alterations of the intracranial pressure after normal saline, 3% saline and dextran-40 Ann Surg 1986 204 686 692 2431664
Walsh JC Zhuang J Shackford SR A comparison of hypertonic to isotonic fluid in the resuscitation of brain injury and hemorrhagic shock J Surg Res 1991 50 284 292 1999918 10.1016/0022-4804(91)90192-O
Prough DS Johnson JC Poole GV JrStullken EH Johnston WE JrRoyster R Effects on intracranial pressure of resuscitation from hemorrhagic shock with hypertonic saline versus lactated Ringer's solution Crit Care Med 1985 13 407 411 3987319
Prough DS Johnson JC Stump DA Stullken EH Poole GV JrHoward G Effects of hypertonic saline versus lactated Ringer's solution on cerebral oxygen transport during resuscitation from hemorrhagic shock J Neurosurg 1986 64 627 632 3950745
Sheikh AA Matsuoka T Wisner DH Cerebral effects of resuscitation with hypertonic saline and a new low-sodium hypertonic fluid in hemorrhagic shock and head injury Crit Care Med 1996 24 1226 1232 8674340 10.1097/00003246-199607000-00027
Battistella FD Wisner DH Combined hemorrhagic shock and head injury: effects of hypertonic saline (7.5%) resuscitation J Trauma 1991 31 182 188 1994077
Berger S Schurer L Hartl R Messmer K Baethmann A Reduction of post-traumatic intracranial hypertension by hypertonic/hyperoncotic saline/dextran and hypertonic mannitol Neurosurgery 1995 37 98 107 8587698
Prough DS Whitley JM Taylor CL Deal DD DeWitt DS Rebound intracranial hypertension in dogs after resuscitation with hypertonic solutions from hemorrhagic shock accompanied by an intracranial mass lesion J Neurosurg Anesthesiol 1999 11 102 111 10213437
Horn P Munch E Vajkoczy P Herrmann P Quintel M Schilling L Schmiedek P Schurer L Hypertonic saline solution for control of elevated intracranial pressure in patients with exhausted response to mannitol and barbiturates Neurol Res 1999 21 758 764 10596385
Hartl R Ghajar J Hochleuthner H Mauritz W Hypertonic/hyperoncotic saline reliably reduces ICP in severely head-injured patients with intracranial hypertension Acta Neurochir Suppl 1997 70 126 129 9416299
Khanna S Davis D Peterson B Fisher B Tung H O'Quigley J Deutsch R Use of hypertonic saline in the treatment of severe refractory posttraumatic intracranial hypertension in pediatric traumatic brain injury Crit Care Med 2000 28 1144 1151 10809296 10.1097/00003246-200004000-00038
Schwarz S Schwab S Bertram M Aschoff A Hacke W Effects of hypertonic saline hydroxyethyl starch solution and mannitol in patients with increased intracranial pressure after stroke Stroke 1998 29 1550 1555 9707191
Suarez JI Qureshi AI Bhardwaj A Williams MA Schnitzer MS Mirski M Hanley DF Ulatowski JA Treatment of refractory intracranial hypertension with 23.4% saline Crit Care Med 1998 26 1118 1122 9635664 10.1097/00003246-199806000-00038
Mayer SA Sacco RL Shi T Mohr JP Neurologic deterioration in noncomatose patients with supratentorial intracerebral hemorrhage Neurology 1994 44 1379 1384 8058133
Qureshi AI Safdar K Weil J Barch C Bliwise DL Colohan AR Mackay B Frankel MR Predictors of early deterioration and mortality in black Americans with spontaneous intracerebral hemorrhage Stroke 1995 26 1764 1767 7570722
Munar F Ferrer AM de Nadal M Poca MA Pedraza S Sahuquillo J Garnacho A Cerebral hemodynamic effects of 7.2% hypertonic saline in patients with head injury and raised intracranial pressure J Neurotrauma 2000 17 41 51 10674757
Qureshi AI Suarez JI Castro A Bhardwaj A Use of hypertonic saline/acetate infusion in treatment of cerebral edema in patients with head trauma: experience at a single center J Trauma 1999 47 659 665 10528599
Hartl R Schurer L Goetz C Berger S Rohrich F Baethmann A The effect of hypertonic fluid resuscitation on brain edema in rabbits subjected to brain injury and hemorrhagic shock Shock 1995 3 274 279 7541301
Shackford SR Bourguignon PR Wald SL Rogers FB Osler TM Clark DE Hypertonic saline resuscitation of patients with head injury: a prospective, randomized clinical trial J Trauma 1998 44 50 58 9464749
Worthley LI Cooper DJ Jones N Treatment of resistant intracranial hypertension with hypertonic saline. Report of two cases J Neurosurg 1988 68 478 481 3343621
Kaufmann AM Cardoso ER Aggravation of vasogenic cerebral edema by multiple-dose mannitol J Neurosurg 1992 77 584 589 1527619
Wahl M [Resistance regulation of cerebral vessels] Fortschr Med 1973 91 1279 4751357
Bosse O Kuschinsky W Wahl M [Effects of osmolarity changes on the resistance regulation of the pia mater vessels in the cat] Pflugers Arch 1972 332 Suppl 332 R59 5065828
Kuschinsky W Wahl M Bosse O Thurau K Perivascular potassium and pH as determinants of local pial arterial diameter in cats. A microapplication study Circ Res 1972 31 240 247 5049739
Muizelaar JP Wei EP Kontos HA Becker DP Cerebral blood flow is regulated by changes in blood pressure and in blood viscosity alike Stroke 1986 17 44 48 3080824
Qureshi AI Suarez JI Use of hypertonic saline solutions in treatment of cerebral edema and intracranial hypertension Crit Care Med 2000 28 3301 3313 11008996 10.1097/00003246-200009000-00032
Schwarz S Georgiadis D Aschoff A Schwab S Effects of hypertonic (10%) saline in patients with raised intracranial pressure after stroke Stroke 2002 33 136 140 11779902 10.1161/hs0102.100877
Tseng MY Al-Rawi PG Pickard JD Rasulo FA Kirkpatrick PJ Effect of hypertonic saline on cerebral blood flow in poor-grade patients with subarachnoid hemorrhage Stroke 2003 34 1389 1396 12730557 10.1161/01.STR.0000071526.45277.44
|
16277715
|
PMC1297608
|
CC BY
|
2021-01-04 16:04:56
|
no
|
Crit Care. 2005 Aug 9; 9(5):R530-R540
|
utf-8
|
Crit Care
| 2,005 |
10.1186/cc3767
|
oa_comm
|
==== Front
Crit CareCritical Care1364-85351466-609XBioMed Central London cc37681627770910.1186/cc3768ResearchMannan-binding lectin and procalcitonin measurement for prediction of postoperative infection Siassi Michael [email protected] Jutta [email protected] Rudi [email protected] Michael [email protected] Steffen [email protected] Werner [email protected] Joachim [email protected] Department of Surgery, University Hospital Erlangen, Erlangen, Germany2 Regional Centre for Blood Transfusion and Clinical Immunology, Aalborg Hospital, Aalborg, Denmark3 Department of Anaesthesiology, University Hospital Jena, Jena, Germany4 Department of Medical Microbiology and Immunology, University of Aarhus, Aarhus, Denmark5 Professor, Department of Surgery, University Hospital Erlangen, Erlangen, Germany6 Department of Anaesthesiology, University Hospital Erlangen, Erlangen, Germany2005 19 7 2005 9 5 R483 R489 2 5 2005 27 5 2005 7 6 2005 20 6 2005 Copyright © 2005 Siassi 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
Postoperative infection is a major cause of morbidity and mortality. We investigated two serum markers for their ability to identify patients at risk for postoperative infection. Mannan-binding lectin (MBL) is a central molecule of the innate immune system and MBL deficiency is known to predispose to infection. Procalcitonin (PCT) is a sensitive marker for bacterial infection.
Methods
We investigated 162 patients undergoing elective surgery for cancer of the gastrointestinal tract. Patients were classified as having no complications (group A), having infection for unknown reason (group B) or having sepsis after events like aspiration or anastomotic leakage (group C). Analysis was done pre- and postoperatively for serum levels of MBL, PCT and C-reactive-protein. DNA was preoperatively sampled and stored and later analysed for genetic polymorphisms of MBL.
Results
The preoperative serum levels of MBL were significantly lower in group B patients than in group A patients (1332 ± 466 ng/ml versus 2523 ± 181 ng/ml). PCT measured on day one post-surgery was significantly higher in group B patients than in group A (3.33 ± 1.08 ng/ml versus 1.38 ± 0.17 ng/ml). Patients with an aberrant MBL genotype had a significantly higher risk of postoperative infections than wild-type carriers (p < 0.05).
Conclusion
Preoperative MBL and early postoperative PCT measurement may help identify patients at risk for postoperative infection.
==== Body
Introduction
Infection is a major reason for postoperative morbidity and mortality. Despite the use of new treatment modalities, improvements in technology and increased experience, morbidity rates are high and sepsis is the most common reason for mortality in surgical intensive care units [1]. Infection and sepsis in surgical patients occurs for various reasons. Some infections can be attributed to distinct events leading to an overwhelming bacterial load that would cause sepsis even in healthy persons, such as anastomotic leakage and aspiration (group C in this study). In other patients, however, an initial source of infection is not apparent, but it still occurs (group B in this study). Events of this type include bactaeremia of unknown reason, hospital acquired pneumonia, infection of in-dwelling catheters and bacterial translocation through the enteral mucosa. In contrast to the first group, these patients are exposed to a bacterial load that can normally be counteracted by the immune system even in the state of acute-phase metabolism. A compromised immune response of the host may predispose to clinically serious courses of infection. Various markers, including C-reactive protein (CRP), tumor necrosis factor α, IL-1, IL-6 and IL-8, have been studied for their ability to predict, diagnose and to differentiate infection, systemic inflammatory response syndrome and sepsis [2-4]. These markers have in common that they indicate activation of the immune system after infection has occurred. It is general surgical knowledge that postoperative infections usually occur after day 5 postoperatively, although there are no statistical data on this issue. Therefore, not only preoperative markers but also indicators in the early postoperative phase would be of predictive value.
Mannan-binding lectin (MBL) is a central part of the innate immune system. It belongs to a group of proteins called collectins. Its structure enables multiple binding to repeating oligosaccharide structures typical of bacterial surfaces [5]. After binding to micro-organisms, MBL activates the MBL-associated serine protease-2 and thus the lectin pathway of complement activation [6].
Previous studies have shown an increased susceptibility to bacterial, viral or fungal infections in patients with decreased MBL-serum levels [7,8]. A recent study identifed serum MBL level as an independent risk factor for survival in non-surgical intensive care unit patients [9]. The MBL concentration in serum is, in part, determined genetically. Some haplotypes confer low MBL concentrations or the secretion of non-functional protein. The main variants in exon 1 of the gene encoding MBL 2 are termed B, C and D variants, with A indicating the wild type. There are also polymorphisms in the 5' regulatory region at position -550 (H/L), -221 (X/Y variants) and in a 5' untranslated region at position +4 (P/Q variants). Due to linkage between polymorphisms, only seven common haplotypes exist with some leading to low MBL serum levels. Mutations in exon 1 (A/0 and 0/0 types) and the AX/AX type especially lead to low MBL serum levels. Altogether, depending on the disease studied, up to 25% of a Caucasian population may have insufficient MBL serum levels [10].
Procalcitonin (PCT), the prohormone of calcitonin, is normally produced in the C-cells of the thyroid gland and its concentration in the plasma of healthy subjects is very low (10–50 pg/ml) [11]. It is induced by bacterial endotoxin or inflammatory cytokines and both has a chemoattractant role and affects nitric oxide production.
PCT is preferentially induced during severe generalised bacterial, parasitic or fungal infections with systemic manifestations rather than in viral infections or inflammatory reactions of non-infectious origin [12].
In order to characterise patients with an increased susceptibility to postoperative sepsis, we studied the levels of MBL and PCT. To assess the influence of MBL polymorphisms, a genotype-analysis was performed. As a reference, a widely used marker of inflammation, CRP, was measured.
Materials and methods
Patients
We investigated patients undergoing major elective surgery for malignant disease of the gastrointestinal tract at the Department of Surgery in the University Hospital Erlangen (Erlangen, Germany) from January 1, 2000, until December 31, 2002. Demographic data for these patients is given in Table 1. Exclusion criteria were age below 18 years, pre-existing infection and emergency surgery. Patients were followed clinically until hospital discharge and postoperative complications were recorded according to the criteria of the American Council of Chest Physicians/Society of Critical Care Medicine [13]. Complications were termed 'postoperative infection' when signs of sepsis or systemic inflammatory response syndrome occurred with no obvious bacterial contamination or specific surgical problem (i.e. anastomotic leakage) (group B). Patients with other septic events were grouped separately (group C).
Blood samples were taken preoperatively on day 3 for MBL analysis and on day 1 and 3 postoperatively for PCT and CRP analysis; serum was stored at -76°C. The study was approved by the institutional ethics committee of the University of Erlangen.
Measurement of MBL, PCT and CRP
MBL was measured by ELISA (Statens Serum Institut, Copenhagen, Denmark). MBL genotyping was performed using a real-time PCR assay on a LightCycler™ instrument (Roche Diagnostics, Mannheim, Germany). In this approach, PCR and melting temperature (Tm) curve analysis are combined based on the principle of mutation detection by melting point analysis with a fluorescence resonance energy transfer hybridisation probe. The three mutations in exon 1 were detected in one capillary using a sensor probe covering the three mutations. Amplification of the variants located upstream of the coding sequence was performed by single colour detection for the H/L polymorphism and multiplexing by dual colour probes was used for simultaneous genotyping of X/Y and P/Q. The details of sample preparation and primer and probe design have been described elsewhere [14].
For statistical analysis, two groups were made. Group 1 included the genotypes leading to normal MBL levels; these are the homozygous wild-type carriers with the exception of the AX/AX type. Group 2 included all carriers of heterozygous or homozygous variations in exon 1 (A/O and O/O type) and the AX/AX type.
Serum PCT levels were determined by a specific and ultrasensitive immunoluminometric assay (Lumitest ProCa-S®, BRAHMS-Diagnostica, Berlin, Germany), which allowed measurement of the concentration of procalcitonin in human serum and plasma in the picogram range (5–5770 pg/ml) for diagnosis of locally restricted bacterial infections. Two monoclonal antibodies that bind PCT (the antigen) at two different binding sites (the calcitonin and katacalcin segments) were used. One of these antibodies (polyclonal, sheep) was luminescence labelled (the tracer), and the other (monoclonal, mouse) was fixed to the inner walls of the tube (coated tube system). During the course of incubation, both antibodies react with PCT molecules in the sample to form a sandwich. The luminescence signal is measured using a suitable luminometer and the LUMltest® Basiskit reagents.
CRP-analysis was done by turbidimetry (Olympus, Hamburg, Germany).
Statistical analysis
All serum levels are displayed as mean ± standard error of the mean (SEM). The statistical analysis was done using the t-test after logarithmic transformation of the raw data. CRP values were compared using the Mann-Whitney test. Correlation analysis was done using Spearman's rank correlation. All p-values are considered two-tailed. All tests were done using the SPSS 11.0 statistics software (SPSS, Munich, Germany).
Results
Of the 172 patients included in the study, complete data for analysis were available for 162. Of these, 137 had no septic events (group A), 10 patients suffered from postoperative infections as defined above (group B) (characteristics are given in Table 2) and 15 patients had septic complications based on a defined postoperative event (group C) (Table 3).
The mean preoperative and postoperative MBL serum concentrations of all patients were 2462 ± 175 and 2375 ± 160 ng/ml, respectively (p = 0.6). The serum level of PCT rose from 0.24 ± 0.1 preoperative to 1.5 ± 0.17 ng/ml postoperative (p < 0.05).
The mean preoperative MBL serum level in patients with postoperative infections (group B) was 1332 ± 466 ng/ml compared to 2523 ± 181 ng/ml in group A patients with no complications (p < 0.05). In patients who developed sepsis after a defined event (group C), preoperative MBL was 2047 ± 254 ng/ml, which did not differ significantly from group A. Postoperative MBL levels in group B and group A patients differed significantly at 1156 ± 393 ng/ml and 2442 ± 166 ng/ml, respectively (p < 0.05) (Fig. 1; complete data are given in Table 4).
The mean preoperative PCT level was 1.05 ± 1.0 ng/ml in patients with (group B) and 0.19 ± 0.1 ng/ml in patients without (group A) postoperative infection (p > 0.05). Postoperatively, there was a significant difference in PCT values between group B (3.33 ± 1.08 ng/ml) and group A (1.38 ± 0.17 ng/ml) (Fig. 2; complete data are given in Table 5).
Mean preoperative CRP was 16.0 ± 2.8 ng/l, 10.4 ± 3.0 ng/l and 11.1 ± 5.0 ng/ml in groups A, B and C, respectively (p > 0.05). On day 3 post surgery, the CRP values were 149.0 ± 5.9 ng/l, 209.4 ± 35.8 ng/l (p > 0.05) and 240.7 ± 22.3 ng/l (p < 0.05) in groups A, B and C, respectively. The measurement on day 1 did not show significant differences between the three groups.
It was possible to perform a MBL genotype analysis in 59 patients. Patients carrying the A/A type but not the XA/XA type (group 1, n = 35) had a mean preoperative MBL level of 3097.1 ± 475.1 ng/ml, whereas the mean serum MBL in patients who were heterozygotic or homozygotic for any mutation in exon 1 (group 2; A/0 (n = 21) and XA/XA type (n = 3)) was 1794.0 ± 374.6 ng/ml (p = 0.04). The Spearman rank correlation coefficient between genotype group and preoperative serum MBL was -0.315 (p = 0.02) (Fig. 3). Of the group 1 and group 2 patients, 2/35 (6%) and 6/24 (25%) developed postoperative infections (group B), respectively (p = 0.035)
Discussion
The search for a preoperative molecular marker defining patients at risk for postoperative infections is of great clinical interest because these patients may benefit from intensified monitoring. In this study, we show that low MBL serum levels and aberrant genotype are associated with a higher rate of postoperative infections. This correlates with earlier studies reporting a higher risk for infections in patients with MBL deficiency [7-9]. In contrast, a study in patients with fever of unknown cause showed no association between MBL deficiency and the course of infection [15]. In comparison to our study, however, the patient collective was not homogenous, with only fever as the primary entry criterion; the severity of sepsis differed substantially between patients, whereas the patient collective in our study was more homogenous. Also, this study only dealt with patients already having an infection and did not provide a 'control group' of patients not suffering from infection. The differences in the results between the two studies may, therefore, be due to different study designs and patient collectives. The risk of postoperative infection correlates with the type of surgery [16], which could cause bias. In our study, only patients undergoing elective surgery for gastrointestinal cancer were included. All patients underwent a resection of the gastrointestinal tract, causing some spillage of bacteria. The group was thus homogenous for the surgery-associated risk of infection.
An important issue in the design of our study was the distinction of patients who suffered complications leading to a bacterial challenge that would overwhelm even a normal immune system (group C) from patients with infection for unknown reason (group B). Mixing these cases in one group would lead to bias because immunological parameters may not play a great role in massive infection as it is encountered in group C patients.
In our study, serum MBL levels did not show a significant increase postoperatively. Postoperative MBL levels were also lower in patients with infections compared to those without. We could, therefore, not show an 'acute phase' like behaviour, as proposed in other studies. This may be due to the short postoperative phase investigated in our study. The previously described significant increase in postoperative MBL levels occurred on day 9 after surgery [17]. This late increase was not covered by our study design. Nevertheless, the comparison of procalcitonin and CRP as classic acute phase proteins and MBL showed a clear difference in their postoperative behaviour. We thus would not encourage the use of the term 'acute-phase protein' for MBL in the postoperative situation.
The analysis of the different genotypes of MBL showed a correlation between mutant genotypes and lower MBL serum levels as described before. The different genotypes were also strongly associated with postoperative infections; in our study, MBL serum levels and MBL genotyping showed similar correlation to infections in the samples that were tested for both. There are conflicting data on the clinical relevance of MBL mutations. A study on patients with pneumococcal disease showed an increased risk only in patients homozygous or functionally homozygous for MBL deficiency [7]. In contrast, a study on febrile neutropenia in children undergoing chemotherapy showed a clinical effect in patients with low MBL serum levels that was not limited to patients with exon 1 mutations [8]. Because the influence of serum levels on clinical outcome was the primary end-point in this study, we decided to group the serotypes according to their influence on MBL serum concentrations.
It has formerly been described that MBL genotyping is superior to the measurement of serum levels [18]. In our study, we show that, in the preoperative situation, measurement of MBL serum levels is as good a clinical marker as genotyping. Whereas the measurement of serum levels can easily be done by ELISA, MBL genotyping requires complex procedures that are not readily available in the clinical setting. This may facilitate future clinical use of MBL measurement.
Because MBL is now available both in a plasma-derived and a recombinant form, the question arises of whether supplementation in MBL-deficient individuals could minimise the risk of infections. The size of our study sample was too small to allow for multivariate analysis. We could not, therefore, identify MBL deficiency as an independent risk factor. The therapeutic use of high dose MBL in subjects with normal MBL levels must still be considered experimental and this approach should be addressed by larger studies.
In contrast to MBL, PCT showed no significant association between its preoperative serum level and the risk of postoperative infection. The trend towards higher preoperative PCT levels in group B patients may indicate pre-existing infection or systemic inflammatory response prior to surgery and needs further investigation.
Nevertheless, patients who developed infection had significantly higher PCT levels in the early postoperative phase. The measurement was made on day 1, whereas most infections occur later in the postoperative period. In contrast, CRP, which is widely used as a marker for infection, only showed a significant increase in group C patients on day 3. Because PCT is a sensitive marker of bacterial infection and systemic inflammation, this indicates that the actual bacterial load does not alone predispose to infection. It may instead show that the individual immune response plays a greater role. Despite the fact that preoperative PCT levels failed to predict infection, its early postoperative measurement (day 1) may help identify patients at risk for infection later on.
In our view, an immunologic factor that predisposes to infection can only play a role in an infection that occurs when the bacterial load is that of the average patient. In overwhelming infections caused by a massive bacterial load, those factors will not play a clinically significant role. Our results add new aspects to other studies that have shown increased susceptibility to infection in MBL-deficient individuals in non-surgical cases.
Conclusion
Low preoperative MBL serum levels, as well as high PCT levels in the early postoperative phase, correlate with the occurence of postoperative infections. These markers may thus be useful for distinguishing patients at risk for infection. Prospective studies are needed to determine whether such patients benefit from intensified monitoring or prophylactic therapy.
Key messages
○ Postoperative infection is a major cause for morbidity and mortality in gastrointestinal surgery.
○ Decreased serum MBL concentrations are associated with an increased risk of infection.
○ Preoperative MBL and early postoperative PCT measurement may help identify patients at risk for postoperative infections.
Abbreviations
CRP = C-reactive protein; ELISA = enzyme-linked immunosorbent assay; IL = interleukin; MBL = mannan-binding lectin; PCR = polymerase chain reaction; PCT = procalcitonin; SEM = standard error of the mean.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
MS and JS conceived of the study, developed the study design, were responsible for patient recruitment and sample collection and carried out the statistical analysis. JR carried out the MBL serum measurements. RS performed the MBL genotyping. MM participated in the PCT analysis and ST participated in the design of the study and helped draft the manuscript. ST participated in the study design and data analysis. WH participated in the study design and drafting of the manuscript. All authors read and approved the final manuscript.
Figures and Tables
Figure 1 Comparison of pre- and postoperative serum-MBL in group A, B and C patients. Preoperative (pre-OP) and postoperative (post-OP) mannan-binding lectin (MBL) serum levels in patients with no postoperative infections (group A), with postoperative infections (group B) and with postoperative infections after a defined event (group C). Error bars indicate the standard error of the mean.
Figure 2 Comparison of pre- and postoperative serum-PCT in group A, B and C patients. Preoperative (pre-OP) and postoperative (post-OP) procalcitonin (PCT) serum levels in patients with no postoperative infections (group A), with postoperative infections (group B) and with postoperative infections after a defined event (group C). Error bars indicate the standard error of the mean.
Figure 3 Correlation between MBL genotype group and MBL serum levels. Preoperative (pre-OP) mannan-binding lectin (MBL) levels in patients homozygous for the wild-type except the AX/AX type (group 1), and carriers of variations in exon 1 (A/O and O/O type) and the AX/AX type (group 2). Spearman rank correlation coefficient between genotype and preoperative serum MBL = -0.32, p = 0.02.
Table 1 Patient demographic data
Age (years; mean and range) 62.3 (25–83)
Gender
Female 56
Male 116
Type of surgery
Resection of pancreas 28 (16.3%)
Gastrectomy 29 (16.9)
Esophageal resection 17 (9.9%)
Colorectal resection 98 (57.0%)
Table 2 Postoperative infections (group B)
Infection Number of patients (n = 10)
Catheter sepsis 4
Pneumonia 2
Abdominal abscess 3
Urinary tract infection 1
Table 3 Other septic complications (group C)
Complication Number of patients (n = 15)
Anatomotic leak 13
Aspiration 2
Table 4 Preoperative and postoperative serum levels of mannan-binding lectin
Group Preoperative MBL (ng/ml) Postoperative MBL (ng/ml)
All patients 2462 ± 175 2375 ± 160
Group A 2523 ± 181 2442 ± 166
Group B 1332 ± 466 1156 ± 393
Group C 2047 ± 254 2266 ± 389
MBL, mannan-binding lectin.
Table 5 Preoperative and postoperative serum levels of procalcitonin
Group Preoperative PCT (ng/ml) Postoperative PCT (ng/ml)
All patients 0.24 ± 0.1 1.5 ± 0.17
Group A 0.19 ± 0.1 1.4 ± 0.17
Group B 1.05 ± 1.0 3.3 ± 1.08
Group C 0.07 ± 0.04 1.1 ± 0.07
PCT, procalcitonin.
==== Refs
Sands KE Bates DW Lanken PN Graman PS Hibberd PL Kahn KL Parsonnet J Panzer R Orav EJ Snydman DR Epidemiology of sepsis syndrome in 8 academic medical centers. Academic Medical Center Consortium Sepsis Project Working Group JAMA 1997 278 234 240 9218672 10.1001/jama.278.3.234
Oberhoffer M Russwurm S Bredle D Chatzinicolau K Reinhart K Discriminative power of inflammatory markers for prediction of tumor necrosis factor-f and interleukin-6 in ICU patients with systemic inflammatory response syndrome (SIRS) or sepsis at arbitrary time points Intensive Care Med 2000 26 S170 S174 10.1007/s001340051138
Harbarth S Holeckova K Froidevaux C Pittet D Ricou B Grau GE Vadas L Pugin J Geneva Sepsis Network Diagnostic value of procalcitonin, interleukin-6, and interleukin-8 in critically ill patients admitted with suspected sepsis Am J Respir Crit Care Med 2001 164 396 402 11500339
Selberg O Hecker H Martin M Klos A Bautsch W Kohl J Discrimination of sepsis and systemic inflammatory response syndrome by determination of circulating plasma concentrations of procalcitonin, protein complement 3a, and interleukin-6 Crit Care Med 2000 28 2793 2798 10966252 10.1097/00003246-200008000-00019
Turner MW Mannose-binding lectin: the pluripotent molecule of the innate immune system Immunol Today 1996 17 532 540 8961631 10.1016/0167-5699(96)10062-1
Petersen S Thiel S Jensenius JC The mannan-binding lectin pathway of complement activation: biology and disease association Mol Immunol 2001 38 133 149 11532276 10.1016/S0161-5890(01)00038-4
Roy S Knox K Segal S Griffiths D Moore CE Welsh KI Smarason A Day NP McPheat WL Crook DW Hill AV MBL genotype and risk of invasive pneumococcal disease: a case-control study Lancet 2002 359 1569 1573 12047967 10.1016/S0140-6736(02)08516-1
Neth O Hann I Turner MW Klein NJ Deficiency of mannose-binding lectin and burden of infection in children with malignancy: a prospective study Lancet 2001 358 614 618 11530147 10.1016/S0140-6736(01)05776-2
Hansen TK Thiel S Wouters PJ Christiansen JS Van den Berghe G Intensive insulin therapy exerts antiinflammatory effects in critically ill patients and counteracts the adverse effect of low mannose-binding lectin levels J Clin Endocrinol Metab 2003 88 1082 1088 12629088 10.1210/jc.2002-021478
Jack DL Bidwell J Turner MW Wood N Simultaneous genotyping for all three known stuctural mutations in the human mannose-binding lectin gene Hum Mutat 1997 9 41 46 8990007 10.1002/(SICI)1098-1004(1997)9:1<41::AID-HUMU7>3.0.CO;2-S
Meisner M Procalcitonin A new, innovative infection parameter 2000 3 Thieme: New York
Meisner M Lohs T Huettemann E Schmidt J Hueller M Reinhart K The plasma elimination rate and urinary secretion of procalcitonin in patients with normal and impaired renal function Eur J Anaesthesiol 2001 18 79 87 11270029 10.1046/j.0265-0215.2000.00783.x
Bone RC Balk RA Cerra FB Dellinger RP Fein AM Knaus WA Schein RM Sibbald WJ Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. The ACCP/SCCM Consensus Conference Committee. American College of Chest Physicians/Society of Critical Care Medicine Chest 1992 101 1644 1655 1303622
Steffensen R Hoffmann K Varming K Rapid genotyping of MBL2 gene mutations using real-time PCR with fluorescent hybridisation probes J Immunol Methods 2003 278 191 199 12957407 10.1016/S0022-1759(03)00190-X
Tacx AN Groeneveld AB Hart MH Aarden LA Hack CE Mannan binding lectin in febrile adults: no correlation with microbial infection and complement activation J Clin Pathol 2003 56 956 959 14645358 10.1136/jcp.56.12.956
Fleischmann KE Goldman L Young B Lee TH Association between cardiac and noncardiac complications in patients undergoing noncardiac surgery: outcomes and effects on length of stay Am J Med 2003 115 515 520 14599629 10.1016/S0002-9343(03)00474-1
Thiel S Holmskov U Hviid L Laursen SB Jensenius JC The concentration of the C-type lectin, mannan-binding protein, in human plasma increases during an acute phase response Clin Exp Immunol 1992 90 31 35 1395098
Garred P Larsen F Madsen HO Koch C Mannose-binding lectin deficiency-revisited Mol Immunol 2003 40 73 84 12914814 10.1016/S0161-5890(03)00104-4
|
16277709
|
PMC1297609
|
CC BY
|
2021-01-04 16:04:55
|
no
|
Crit Care. 2005 Jul 19; 9(5):R483-R489
|
utf-8
|
Crit Care
| 2,005 |
10.1186/cc3768
|
oa_comm
|
==== Front
Crit CareCritical Care1364-85351466-609XBioMed Central London cc37711628503410.1186/cc3771ResearchCase report: severe heat stroke with multiple organ dysfunction – a novel intravascular treatment approach Broessner Gregor [email protected] Ronny [email protected] Gerhard [email protected] Peter [email protected] Klaus [email protected] Christian [email protected] Bettina [email protected] Erich [email protected] Resident, Department of Neurology, Neurological Intensive Care Unit, Medical University, Innsbruck, Austria2 Assistant Professor, Department of Neurology, Neurological Intensive Care Unit, Medical University, Innsbruck, Austria3 Assistant Professor, Department of Neurology, Neurological Intensive Care Unit, Medical University, Innsbruck, Austria4 Professor and Chairman, Department of Neurology, Neurological Intensive Care Unit, Medical University, Innsbruck, Austria2005 20 7 2005 9 5 R498 R501 7 5 2005 23 6 2005 Copyright © 2005 Broessner 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
We report the case of a patient who developed a severe post-exertional heat stroke with consecutive multiple organ dysfunction resistant to conventional antipyretic treatment, necessitating the use of a novel endovascular device to combat hyperthermia and maintain normothermia.
Methods
A 38-year-old male suffering from severe heat stroke with predominant signs and symptoms of encephalopathy requiring acute admission to an intensive care unit, was admitted to a ten-bed neurological intensive care unit of a tertiary care hospital. The patient developed consecutive multiple organ dysfunction with rhabdomyolysis, and hepatic and respiratory failure. Temperature elevation was resistant to conventional treatment measures. Aggressive intensive care treatment included forced diuresis and endovascular cooling to combat hyperthermia and maintain normothermia.
Results
Analyses of serum revealed elevation of proinflammatory cytokines (TNF alpha, IL-6), cytokines (IL-2R), anti-inflammatory cytokines (IL-4) and chemokines (IL-8) as well as signs of rhabdomyolysis and hepatic failure. Aggressive intensive care treatment as forced diuresis and endovascular cooling (CoolGard® and CoolLine®) to combat hyperthermia and maintain normothermia were used successfully to treat this severe heat stroke.
Conclusion
In this case of severe heat stroke, presenting with multiple organ dysfunction and elevation of cytokines and chemokines, which was resistant to conventional cooling therapies, endovascular cooling may have contributed significantly to the reduction of body temperature and, possibly, avoided a fatal result.
==== Body
Introduction
Heat stroke is a life-threatening disease characterized by hyperpyrexia (elevated core body temperature exceeding 40°C) and predominant central nervous system dysfunction resulting in delirium, convulsion or coma [1]. In many clinical and pathogenetic aspects, heat stroke resembles sepsis, requiring aggressive intensive care treatments, and there is growing evidence that endotoxemia and cytokines may be implicated in its pathogenesis [1-3]. We report a case of severe heat stroke with secondary multiple organ dysfunction being successfully treated with an intravascular cooling device.
Case report
A 38-year-old male underwent a hiking tour on a hot, humid day in late July 2003. At the end of this exhausting trip he complained of dizziness, finally falling into an 'apathic' state. On the arrival of the emergency physician, the patient suffered from a generalized epileptic seizure. Subsequently, the comatose patient (Glasgow Coma Scale 6 (E 1, V 1, M 4)) developed respiratory insufficiency and cardiovascular failure (blood pressure 60/20 mmHg, heart rate 166/min). He was immediately intubated (using fentanyl (0.3 mg), etomidate (40 mg) and midazolam (20 mg)) and transported to our neurological intensive care unit (NICU).
On admission, the patient was deeply sedated and under analgesia, but still suffering from hypotension requiring immediate use of catecholamines (norepinephrine). The patient had normal weight (body mass index = 24) and no significant previous medical history. The initial cerebral computed tomography (CT) scan in combination with CT angiography did not reveal any pathologies and, to exclude an infectious origin for the central nervous system dysfunction, a lumbar puncture was carried out yielding normal cerebrospinal fluid. An initial extensive laboratory work up revealed impaired liver function (glutamic-oxaloacetic transaminase 312U/l (normal range: 10 to 50 U/l), glutamic-pyruvic transaminase 244 U/l (normal range: 10 to 50 U/l), gamma-glutamylcyclotransferase 94 U/l (normal range: 10 to 66 U/l) and a prothrombin time of 60% (normal range: 70 to 130%). Serum creatinine levels as well as blood urea nitrogen (BUN) were elevated (creatinine 2.6 mg/100 ml (normal range: 0.8 to 1.3 mg/100 ml) and BUN 30 mg/100 ml (normal range: 5 to 25 mg/100 ml)) indicating the beginning of renal failure. This situation was further complicated by rhabdomyolysis with elevation of myoglobin and creatine kinase (CK) (myoglobin peak level 33.124 μg/l (day 2), normal range: 0 to 116 μg/l) and CK peak level 102.4 U/l (day 4), normal range: 38 to 174 U/l.
At the time of admission, core body temperature measured by urinary bladder probe (Foley catheter; Kendall Curity, Mansfield, MA, USA), was 40.8°C. During the first 20 h of treatment, conventional temperature control methods including high-dose non-steroidal anti-inflammatory drugs (NSAIDs) (acetylsalicylic acid 1000 mg and paracetamol 2000 mg) and opioids (pethidine 100 mg), as well as external cooling devices such as cooling blankets (Blanketrol II®, Cincinnati Sub-Zero, Cincinnati, OH, USA) and Bair Hugger® (Arizant Healthcare Inc, Eden Prairie, MN, USA), which were applied for 8 h, did not lead to any significant decrease in core body temperature (Figure 1). Because of subsequent deterioration of the patient's condition and insufficient conventional temperature control, an aggressive treatment approach with a novel intravascular cooling system (CoolGard 3000® and Cool Line™, Alsius, Irvine, CA, USA) was begun. The heat-exchange catheter (Cool Line™) was placed into the left superior vena cava and cooled saline was infused through a closed loop system into two heat-exchange balloons located near the distal end of the catheter. The temperature of the saline solution was adjusted automatically by the CoolGard 3000®, which is an external temperature control unit, according to feedback to the external pump/refrigerant device from a microthermister attached to a Foley bladder catheter. Target temperature was set at 37°C for the first 25 h of intravascular treatment and subsequently at 37.5°C. Target temperature was reached within 7 h at a maximum cooling rate of 0.6°C/h and 'cooling' was prolonged at this level.
Multiple organ dysfunction and secondary rhabdomyolysis led to increased levels of myoglobin and CK (myoglobin peak level 33.124 μg/l (day 2), CK peak level 102.4 U/l (day 4)). To prevent imminent renal failure, forced diuresis was initiated and continued for 40 h using high-dose furosemide and fluids, resulting in an urinary excretion rate of more than 400 ml/h, leading to a fluid turnover of up to 24,000 ml/24 h. With this aggressive measure, we suceeded in avoiding the use of extracorporal hemofiltration and the renal parameters returned to normal values within 3 days.
Core body temperature was maintained at about 37°C and subsequently maintained at 37.5°C (± 0.2°C) with the use of the intravascular catheter over the next 5 days (Figure 1). Several attempts to stop the active cooling within this period (Figure 1) led to an immediate steep increase of core body temperature, which forced us to prolong this very efficacious endovascular treatment. Finally, after 111.5 h, CoolGard® treatment was stopped, since most of the laboratory parameters had stabilized; the patient did not suffer from hyperthermia thereafter.
To confirm the diagnosis of severe heat stroke and to measure the systemic inflammatory response [2], we analyzed levels of plasma cytokines and serum chemokines 60 h after admission. The results are shown in Table 1: soluble interleukin (IL) receptor (sIL-2R) 1500 pg/ml, IL-4 3 pg/ml, IL-6 204 pg/ml, IL-8 40 pg/ml and tumor necrosis factor alpha (TNF alpha) 38 pg/ml (IL-4, IL-6, IL-8, TNF alpha analyzed by immunoenzymometric assay, Biosource, Nivelles, Belgium; IL-2R analyzed by immunoenzymometric assay, Immunotec, Marseille, France). On days 5 and 7 after admission, the values of IL-6 had decreased to 96 pg/ml (day 5) and 34 pg/ml (day 7), respectively.
Initially diagnosed aspiration pneumonia as well as sinusitis maxillaris diagnosed on the initial cerebral CT scan, were treated with tazobactam/piperacilline and clindamycin. On day 3, somatosensoric potentials did not show any pathologic results. The patient was extubated on day 8 and transferred to a regular neurological ward on day 12 with neither signs of any focal neurological nor overt cognitive deficits. At the time of discharge from the NICU, laboratory parameters had returned to normal values.
Results and discussion
Immediate cooling and support of organ-system function are the two major therapeutic objectives in patients with heat stroke [1,3,4]. Using conventional temperature control measures such as NSAIDs or external cooling devices (cooling blankets and Bair Hugger®), even applied for several hours, was ineffective in combating hyperthermia in this case. So far, only one case has been reported in which a heat exchange balloon was inserted in the femoral vein [5] leading to reduction of core body temperature to 37-39°C. We succeeded in maintaining the preset normothermia (37 to 37.5°C) for more than 5 days, thus both preventing neurological sequelae and rescuing failing organ functions, in a patient with an expected mortality rate of up to 50% [1,6]. For active cooling, we used the Cool Line™ catheter placed into the upper vena cava in combination with CoolGard 3000®. Studies could show that this system is an efficacious tool for combating hyperthermia in patients with severe primary intracranial diseases [7-9] but has not been validated so far as a therapeutic tool in heat stroke.
The laboratory work up of chemokines (IL-8), proinflammatory cytokines (TNF alpha, IL-6), cytokines (IL-2R) and anti-inflammatory cytokines (IL-4) revealed an elevation of all parameters, which is of particular interest as it has been postulated that these cytokines and chemokines may play an important role in the pathogenesis of heat stroke [2]. In particular, the excessive elevation of IL-6 and IL-2R found in our patient is remarkable as these two markers may predict disease severity [1,2]. Considering these facts, the favorable neurologic outcome of our patient after having suffered from this 'sepsis-like syndrome' including multiple organ dysfunction, may be an indicator that intravascular cooling and maintenance of normothermia influences the inflammatory response and may lead to improved outcome in patients with heat stroke.
Conclusion
Heat stroke is a life-threatening disease requiring immediate admission to an ICU. The progression to multiple organ dysfunction can be fatal as many organ systems may be affected. The primary therapeutic goal must be to lower the core body temperature, which may be impossible with conventional measures. In our patient, intravascular treatment was efficacious and feasible. Prospective and controlled studies comparing the efficacy of various cooling techniques in NICU patients have proven the feasibility and efficacy of this endovascular cooling device (CoolGard 3000® and Cool Line™), thus it should be considered as a possible alternative to conventional treatment in heat stroke patients. In our patient, maintenance of normothermia (37 to 37.5°C) led to a favorable outcome with no neurologic impairment after the 'sepsis-like' heat stroke. Thus, further randomized and controlled studies are warranted to evaluate intravascular cooling as a possible tool in combating severe heat stroke.
Key messages
• Heat stroke is a life-threatening disease requiring immediate admission to an intensive care unit.
• Lowering the core body temperature must be the primary goal but conventional temperature control measures, as in our case, might be insufficient in decreasing core body temperature.
• Intravascular cooling was efficacious and feasible in maintaining "normothermia" (37°C – 37.5°C) in our patient, leading to a favorable outcome.
• Intravascular cooling could be considered as a possible alternative to conventional treatment in heat stroke patients.
Abbreviations
BUN = blood urea nitrogen; CK = creatine kinase; IL = interleukin; NICU = neurological intensive care unit; NSAIDs = non-steroidal anti-inflammatory drugs; R = receptor; TNF = tumor necrosis factor.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
GB and ES coordinated the data analysis and drafted the manuscript. RB, GF and BP participated in analysis of clinical data. KE and CB participated in analysis of CoolGard 3000® data. PL helped to draft the manuscript. All authors read and approved the final manuscript.
Figures and Tables
Figure 1 Course of core body temperature in a patient with heat stroke. The red line denotes the core body temperature while using 'conventional' temperature control methods. The blue line denotes the core body temperature while using an endovascular (CoolGard®) cooling treatment. Blue arrows denote the start of CoolGard® treatment. Red arrows denote attempts to terminate the active cooling procedure.
Table 1 Course of TNF alpha and IL levels in a patient with heat stroke
Parameters Normal range, pg/ml Measured values on day 3, pg/ml Measured values on day 5, pg/ml Measured values on day 7, pg/ml
sIL-2R 0–4.8 1500
IL-4 NA 3
IL-6 0–3 204 96 34
IL-8 NA 40
TNF alpha 0–20 38
IL, interleukin; NA, not available; TNF, tumor necrosis factor.
==== Refs
Bouchama A Knochel JP Heat stroke N Engl J Med 2002 346 1978 1988 12075060 10.1056/NEJMra011089
Lu KC Wang JY Lin SH Chu P Lin YF Role of circulating cytokines and chemokines in exertional heatstroke Crit Care Med 2004 32 399 403 14758154 10.1097/01.CCM.0000108884.74110.D9
Grogan H Hopkins PM Heat stroke: implications for critical care and anaesthesia Br J Anaesth 2002 88 700 707 12067009 10.1093/bja/88.5.700
White JD Riccobene E Nucci R Johnson C Butterfield AB Kamath R Evaporation versus iced gastric lavage treatment of heatstroke: comparative efficacy in a canine model Crit Care Med 1987 15 748 750 3608531
Megarbane B Resiere D Delahaye A Baud FJ Endovascular hypothermia for heat stroke: a case report Intensive Care Med 2004 30 170 14600811 10.1007/s00134-003-2053-z
Bouchama A Heatstroke: a new look at an ancient disease Intensive Care Med 1995 21 623 625 8522663 10.1007/BF01711537
Schmutzhard E Engelhardt K Beer R Brossner G Pfausler B Spiss H Unterberger I Kampfl A Safety and efficacy of a novel intravascular cooling device to control body temperature in neurologic intensive care patients: a prospective pilot study Crit Care Med 2002 30 2481 2488 12441758 10.1097/00003246-200211000-00013
Marion DW Controlled normothermia in neurologic intensive care Crit Care Med 2004 32 Suppl 2 S43 45 15043227 10.1097/01.CCM.0000110731.69637.16
Diringer MN Neurocritical Care Fever Reduction Trial Group Treatment of fever in the neurologic intensive care unit with a catheter-based heat exchange system Crit Care Med 2004 32 559 564 14758179 10.1097/01.CCM.0000108868.97433.3F
|
16285034
|
PMC1297610
|
CC BY
|
2021-01-04 16:04:56
|
no
|
Crit Care. 2005 Jul 20; 9(5):R498-R501
|
utf-8
|
Crit Care
| 2,005 |
10.1186/cc3771
|
oa_comm
|
==== Front
Crit CareCritical Care1364-85351466-609XBioMed Central London cc37721627770810.1186/cc3772ResearchRespiratory compliance but not gas exchange correlates with changes in lung aeration after a recruitment maneuver: an experimental study in pigs with saline lavage lung injury Henzler Dietrich [email protected] Paolo [email protected] Rolf [email protected] Annette [email protected] Andreas H [email protected] Rolf [email protected] Ralf [email protected] Senior Anesthesiologist, Anesthesiology Department, University Hospital RWTH Aachen, Germany2 Professor of Anesthesiology, Environment, Health and Safety Department, University of Insubria, Varese, Italy3 Intensivist, Surgical Intensive Care Department, University Hospital RWTH Aachen, Germany4 Resident, Anesthesiology Department, University Hospital RWTH Aachen, Germany5 Department of Clinical Radiology, University Hospital RWTH Aachen, Germany6 Professor of Anesthesiology, Anesthesiology Department, University Hospital RWTH Aachen, Germany7 Head, Surgical Intensive Care Department, University Hospital RWTH Aachen, Germany2005 13 7 2005 9 5 R471 R482 8 5 2005 27 5 2005 10 6 2005 24 6 2005 Copyright © 2005 Henzler 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 cited.
Introduction
Atelectasis is a common finding in acute lung injury, leading to increased shunt and hypoxemia. Current treatment strategies aim to recruit alveoli for gas exchange. Improvement in oxygenation is commonly used to detect recruitment, although the assumption that gas exchange parameters adequately represent the mechanical process of alveolar opening has not been proven so far. The aim of this study was to investigate whether commonly used measures of lung mechanics better detect lung tissue collapse and changes in lung aeration after a recruitment maneuver as compared to measures of gas exchange
Methods
In eight anesthetized and mechanically ventilated pigs, acute lung injury was induced by saline lavage and a recruitment maneuver was performed by inflating the lungs three times with a pressure of 45 cmH2O for 40 s with a constant positive end-expiratory pressure of 10 cmH2O. The association of gas exchange and lung mechanics parameters with the amount and the changes in aerated and nonaerated lung volumes induced by this specific recruitment maneuver was investigated by multi slice CT scan analysis of the whole lung.
Results
Nonaerated lung correlated with shunt fraction (r = 0.68) and respiratory system compliance (r = 0.59). The arterial partial oxygen pressure (PaO2) and the respiratory system compliance correlated with poorly aerated lung volume (r = 0.57 and 0.72, respectively). The recruitment maneuver caused a decrease in nonaerated lung volume, an increase in normally and poorly aerated lung, but no change in the distribution of a tidal breath to differently aerated lung volumes. The fractional changes in PaO2, arterial partial carbon dioxide pressure (PaCO2) and venous admixture after the recruitment maneuver did not correlate with the changes in lung volumes. Alveolar recruitment correlated only with changes in the plateau pressure (r = 0.89), respiratory system compliance (r = 0.82) and parameters obtained from the pressure-volume curve.
Conclusion
A recruitment maneuver by repeatedly hyperinflating the lungs led to an increase of poorly aerated and a decrease of nonaerated lung mainly. Changes in aerated and nonaerated lung volumes were adequately represented by respiratory compliance but not by changes in oxygenation or shunt.
See related commentary
==== Body
Introduction
Severe impairment of oxygenation in acute lung injury and in the acute respiratory distress syndrome (ARDS) is caused by an inhomogenous ventilation-perfusion distribution () and an increase in shunt fraction. The amount of aerated lung is markedly reduced due to alveolar collapse and flooding [1,2]. Mechanical ventilation has been shown to further aggravate the mismatch [3]. Even though it is unclear if the optimal treatment should aim to improve gas exchange, to prevent additional lung damage or to resolve the existing damage, one of the commonly used treatment concepts is the open-lung approach [4], aiming at recruitment and maintenance of ventilated lung volume. In general, recruitment means to transform nonaerated into aerated lung. These regions can open and close or can be kept opened if sufficient positive endexpiratory pressure (PEEP) is applied. Significant controversy exists over the optimal method to achieve alveolar recruitment and to the definition of recruitment, whether it means re-opening of collapsed alveoli or edema clearance [2]. Improvement in oxygenation is commonly used to detect recruitment, although gas exchange is also influenced by many other factors, like ventilation-perfusion distribution, pulmonary blood flow and regional vascular regulation [5,6]. The assumption that the gas exchange parameters adequately represent the mechanical process of alveolar opening has not been proven so far. The best available technique to detect recruitment is computed lung tomography [7] where the decrease of atelectatic lung can be visualized [8]. Since computer tomographic (CT) scanning cannot be performed repeatedly under clinical conditions, different parameters must be obtained at the bedside in order to indicate successful recruitment. The aim of this study was to investigate whether commonly used measures of lung mechanics better detect lung tissue collapse and changes in lung aeration after a recruitment maneuver as compared to measures of gas exchange.
Materials and methods
After governmental approval, eight anesthetized female pigs (31.3 ± 1.9 kg) were orotracheally intubated and ventilated in constant flow mode with a fraction of inspired oxygen (FiO2) of 1.0, a tidal volume of 8 ml/kg with an inspiratory-expiratory (I:E) ratio of 1:1 and PEEP of 10 cmH2O throughout the study. Deep anesthesia was maintained with a continuous infusion of propofol (7.7 ± 1.7 mgkg-1h-1) and fentanyl (8.0 ± 2.2 μgkg-1h-1) and animals were additionally paralyzed with pancuronium (0.3 ± 0.1 mgkg-1h-1) for the actual experimental phase. Handling of animals conferred to the guidelines laid out in the Guide for the Care and Use of Laboratory Animals [9].
Arterial and pulmonary artery catheters (Becten Dickinson, Heidelberg, Germany) were placed and cardiac output was determined through thermodilution with equipment from Datex-Ohmeda (Duisburg, Germany). The extravascular lung water index was determined by transcardiopulmonary thermodilution with equipment from Pulsion (Munich, Germany). Gas flow and airway pressures were measured at the proximal end of the tracheal tube. The esophageal pressure was measured using a balloon catheter (International Medical, c/o Allegiance, Kleve, Germany). Expiratory volumes were corrected as described previously [10]. A more detailed description can be found in Additional file 1.
Experimental protocol
Acute lung injury was induced through repeated lung lavage as described previously [11] and allowed to stabilize until the arterial blood partial oxygen pressure (PaO2) had been below 100 mmHg for 60 minutes. The following measurements were obtained before and 10 minutes after a recruitment maneuver was performed.
Lung volumes
Contiguous multi-slice CT scans of the whole lung (Siemens Sensation 16, Forchheim, Germany) were taken at end-expiratory and end-inspiratory occlusion [1,12]. From the reconstructed slices (2 mm) the lung was delineated by hand from the inner pleura. The calculations for hyperinflated parenchyma (HYP; -1000 to -900 Hounsfield units (HU)), normally aerated (NORM; -900 to -500 HU), poorly aerated (POOR; -500 to -100 HU) and non-aereated parenchyma (NON; -100 to +100 HU) were done by the CT software with a pixel size of 0.59 mm. The resulting areas were multiplied with the slice thickness and then added together for lung volumes (VTOT, VHYP, VNORM, VPOOR, VNON). The intrathoracic gas volume was calculated as VGAS = VTOT × HUMEAN/-1000 and the intrathoracic tissue volume was calculated as VTISS = VTOT - VGAS. The lung volumes consisted of VGAS + VTISS, for example, a mean HU of -500 representing 50% gas and 50% tissue. Recruitment was defined as a decrease in the nonaerated lung volume after the recruitment maneuver [13].
Venous admixture and dead space
Arterial and mixed venous blood samples were collected simultaneously and analyzed immediately using equipment by Radiometer, Copenhagen, Denmark. Venous admixture (QVA/QT) was calculated using the shunt equation [14] and dead space (VD/VT) according to the modified Bohr equation.
Compliance of the respiratory system
The static compliance of the respiratory system (CRS) was computed using the occlusion technique [15].
Inflation compliance and recruitable volume
An inflation-deflation pulmonary pressure-volume curve (PV-curve) starting from zero end-expiratory pressure (ZEEP) was performed using a new tool that was built into the ventilator (Galileo Gold, Hamilton, Rhäzüns, Switzerland). Objective analysis of inflation and deflation curves was performed by fitting it to the Venegas-Harris equation [16]. The corner points stating the point of maximum compliance increase and decrease, being the mathematical equivalents of lower and upper inflection points, were calculated. The maximum inflation compliance (CINF) was calculated through numerical differentiation of the true inflection point. The recruitable volume (VREC) was defined as the end-expiratory volume difference between the inflation and deflation pressure obtained at PEEP level (10 cmH2O).
The actual recruitment maneuver was performed by inflating the lungs three times with a pressure of 45 cmH2O for 40 s [8,17-19], with 10 normal tidal breaths between inflations. A detailed description of animal preparation and measurements can be found in Additional file 1. After the experiment, the animals were killed with a barbiturate overdose.
Statistical analysis
All data are reported as mean ± SD. To correlate the parameters under investigation with the CT measurements, the Pearson's coefficient (r) was calculated. Where appropriate, multiple linear regression was used. The validity of the model was verified by a Durbin-Watson statistic. Because correlations of parameters with end-inspiratory or end-expiratory CT measurements exhibited equal results, only the end-expiratory data are presented. To determine the parameter with the strongest influence, the dimensionless standardized beta coefficient (betaS) was calculated. Pre- and post-recruitment maneuver (RM) values were compared using Wilcoxon's signed ranks test. In the case of parameters exhibiting a significant difference, the dimensionless fractional change for any parameter 'X' was then calculated as fractional change (X) = XpostRM/XpreRM - 1 and correlation analysis performed as explained above. Fractional change values are expressed as percentages. Statistical significance was accepted at p < 0.05 (SPSS 11.0, SPSS, Chicago, USA).
Results
Correlation of the CT data with gas exchange and respiratory mechanics parameters before and after a recruitment maneuver
Parameters correlating with aerated lung
No significant correlations were found between the gas exchange or respiratory mechanics parameters and normally aerated lung volume. Instead, a significant correlation was observed between poorly aerated lung volume and the PaO2 (r = 0.569, p = 0.022) (Fig. 1c) and also between VPOOR and respiratory system compliance (r = 0.719, p = 0.006) (Fig. 1a) and the inflation pressure maximum compliance increase (r = 0.655, p = 0.008).
Parameters correlating with nonaerated lung
Venous admixture correlated directly with nonaerated lung volume (r = 0.678, p = 0.004) (Fig. 1d), but the PaO2 did not (p = 0.098). Similarly, nonaerated lung volume correlated with physiologic dead space (r = 0.534, p = 0.04), but not with the arterial blood partial carbon dioxide pressure (PaCO2; p = 0.154). Of the respiratory mechanics parameters, the respiratory system compliance (r = -0.587, p = 0.035) and the inflation point of maximum compliance decrease (r = -0.77, p = 0.001) correlated with the nonaerated lung volume (Fig. 1b). Multiple regression analysis revealed that the best prediction of nonaerated volume was achieved by a combination of inflation point of maximum compliance decrease (betaS = -0.563) and venous admixture (betaS = 0.45).
Effects of the recruitment maneuver
CT lung volume measurements
Atelectasis and consolidation were found predominately in the dependent two-thirds of the lung (Fig. 2). The recruitment maneuver caused a significant decrease in nonaerated lung volume by approximately 22% (Table 1). It is important to note that the recruitment was associated with an increase in poorly aerated and normally aerated lung volume. The individual changes in CT lung volumes are shown in Fig. 3. The increase of VPOOR (21.7%, betaS = 0.668) contributed more to recruitment than the increase of VNORM (11%, betaS = 0.641).
The 13% increase in VGAS represents an increase in the functional residual capacity, because the inspiratory-expiratory volume difference did not change (211 ± 33 ml pre-RM versus 221 ± 45 ml post-RM, p = 0.46). No differences in tidal volumes were found between the measurement with CT and spirometry. Importantly, the inspiratory-expiratory volume change in nonaereated regions (62 ± 18 ml), representing opening and collapse of alveoli, was not significantly reduced after the recruitment maneuver (43 ± 26 ml, p = 0.114). The fractional change (VGAS), however, was not correlated with any parameter of gas exchange or respiratory mechanics; it only correlated with fractional change (VNORM), which could be expected from recruitment.
Effects on gas exchange
The distributions of the fractional changes of the parameters under investigation can be seen in Fig. 4. Overall, a significant improvement in oxygenation (fractional change (PaO2), +33%) and a shunt reduction (fractional change (QVA/QT), -20.8%) were observed (Table 2). The fractional change (PaO2) did not correlate well with the increase of normally or poorly aerated lung (r = 0.51, p = 0.18), however, nor did the fractional change (QVA/QT) correlate with the decrease of nonaerated lung (r = 0.50, p = 0.21) (Fig. 5a,b). No significant changes in PaCO2 nor dead space were observed. From these data it seems that the changes in gas exchange parameters do not correlate with the changes in aerated or nonaerated volumes caused by a recruitment maneuver.
Effects on respiratory mechanics
In accordance with the CT-measurements, there were no changes in tidal volume, but peak and plateau pressures did decrease (Table 3), which correlated with the fractional change (VNON) (Fig. 5c). There was a significant increase in compliance and recruitable volume. The increase in CRS correlated positively with the increase in poorly aerated lung (r = 0,822, p = 0.012) and inversely with the decrease in nonaerated lung volumes (r = -0.721, p = 0.043). The decrease of nonaerated lung volume could be predicted from the equation fractional change (VNON) = -0.69 × fractional change (CRS). This means the decrease of atelectasis can be estimated to be roughly two-thirds of the increase in CRS. Interestingly, we found no significant correlations with normally aerated lung volume.
After the recruitment maneuver, the PV-curve was expanded vertically (see Additional file 1; Fig. 4). The resultant increase in the inflational point of maximum compliance increase correlated with the increase in the sum of VNORM and VPOOR (r = 0.914) (Fig. 5d). The fractional changes of VREC correlated positively with an increase in VPOOR (r = 0.863, p = 0.034) and also inversely with a decrease in VNON (r = -0.775 (p = 0.041).
Effects on hemodynamics
With no changes in sedation and fluid management, only heart rate and cardiac output decreased after the recruitment maneuver. However, no changes in systemic or pulmonary pressures nor vascular resistance could be observed. The extravascular lung water index indicated massive pulmonary edema, but did not change after the recruitment maneuver either (see Additional file 1; Table 2).
In summary, changes in compliance of the respiratory system but not in gas exchange parameters correlated with changes in nonaerated and aerated lung before and after a recruitment maneuver at the same PEEP level of 10 cmH2O.
Discussion
Experimental considerations
We investigated parameters used to indicate the amount and the change of aerated and nonaerated lung in acute lung injury. We chose the lavage model in pigs for this because it is known to be easily recruitable. This model has been shown to cause lung inflammation [20], ventilation-perfusion mismatch equal to other models [21] and an increase in extravascular lung water and excess tissue [22]. Furthermore, the preferential distribution of atelectasis to the dependent lung could also be demonstrated in patients with ARDS by use of CT scanning [12]. The number of experiments is in line with recent studies investigating respiratory mechanics in acute lung injury [23,24]. Increasing the power may have resulted in more subtle correlations, although we have found some correlations to be significant (certain effect) and others not (possible effect).
Our definition of recruitment may be questioned, because what we measured really is a density scale proportional to gas-tissue distributions. Thus, the decrease in a portion of HU labeled 'atelectasis' does not necessarily mean opening of alveoli. Instead, edema fluid could be squeezed out of the lung and pushed into poorly aerated lung; however, we did not find changes in extravascular lung water [22] or lung tissue after the recruitment maneuver. Therefore, the observed changes in differently aerated lung volumes could have been caused by transformation of completely collapsed lung into partly opened lung or by an increased homogeneity in the distribution of alveolar fluid [25]. Importantly, the observed changes in aerated lung volume were relatively small 10 minutes after the recruitment maneuver and do not support the usefulness of such a maneuver, which has also been demonstrated in clinical studies [26]. Possibly higher levels of PEEP could have enhanced recruitment, but to avoid possible influences of PEEP on the physiological parameters studied we maintained the same level of PEEP (10 cmH2O).
Evaluation of gas exchange parameters
Although impaired oxygenation is the main symptom in acute lung injury [27] correlated with atelectasis [28,29], our study suggests that PaO2 is less related to the amount of atelectatic lung than to the aerated lung that remains for ventilation. These studies suggested that there was a linear correlation between PaO2 or shunt and atelectasis formation, especially if atelectasis was below 5% of total lung [28]. Lung healthy subjects were studied, however, and only one slice of the lung close to the diaphragm was analyzed, representing the area where most atelectases occur. So atelectasis as a fraction of the whole lung was probably much lower. Furthermore, there seems to be a difference in the characteristic of atelectasis formation between otherwise healthy lungs and injured lungs with high proportions of instable alveolar units that are poorly ventilated. Poorly aerated lung has been considered as low regions. Because we found a correlation between the PaO2 and poorly aerated lung, it is possible that the regional blood flow through these regions was considerably high. Therefore, intrapulmonary shunt does not only happen in totally collapsed, but also in low , units. What the clinician wants to know is whether a certain improvement in oxygenation can predict the amount of recruitment. Improvements in gas exchange after recruitment are attributed mainly to two basic mechanisms: first, by redirection of blood flow from nonaerated to aerated lung regions and reduction of venous admixture, which we observed; and second, which we did not observe, through an increase in alveolar ventilation, leading to a reduction in PaCO2. In several clinical studies that have failed to demonstrate a benefit for active recruitment [26,30,31], oxygenation parameters, but not mechanical parameters, were used for decision making. Because we could not find the PaO2 changes representative of recruitment, even in a very recruitable model, this could have important implications on the interpretation of these studies. It seems that the amount of oxygenation improvement is not so much determined by the reduction of nonaerated lung, but by the blood flow through these regions.
Evaluation of respiratory mechanics parameters
The plateau pressure and static lung compliance correlated equally with nonaerated and poorly aerated lung volumes. It appears that in lung injury, VPOOR and VNON are the main determinants in overall lung compliance. Following the argument of Barnas et al. [32] that the elastance (E) of the rib cage compartment is parallel to the elastance of the diaphragm-abdomen compartment, the elastances of the differently aerated lung compartments could behave similarly and thus be described by the equation 1/ELUNG = k1/EHYP + k2/ENORM + k3/EPOOR + k4/ENON, where the constants k1–4 depend on their fraction of total lung volume. Thus in healthy lungs, EL is mainly dependent on ENORM, because it has the highest fraction of lung volume. But with increasing fractions of EPOOR and ENON (with much higher values than ENORM) they will become increasingly determinant for lung compliance. This hypothesis is supported by multiple regression analysis, showing that the fractional change of CINF was most dependent on VPOOR (betaS 0.550) and VNON (betaS -0.331).
The PV-curve has been used to obtain information about diseased lungs [33-36]. Although the calculated curve may not equally fit all data [37], the mathematical analysis of the PV-curve is objective and the best available algorithm so far [38]. Because the PV-curve characteristics reflect a dynamic investigation of the lung, they have been used to set the parameters of ventilation [39]. We did not investigate whether the point of maximum compliance increase really reflects the lower inflection point (LIP). We were surprised that the inflation point of maximum compliance increase actually increased after recruitment in a nonlinear way (Fig. 5d), with a sharp increase beyond an increase in aerated lung >20%. If the point of maximum compliance increase truly represented the commencement of alveolar recruitment, it should be lower in conditions with less atelectasis. An explanation for this phenomenon could be that recruitment happens throughout the inflation curve [36], making the existence of a singular threshold opening pressure unlikely. Also, inflation LIP has been shown to only poorly represent the pressure at which recruited lung stays open [33,40]. But since we did observe an increase in the LIP with recruitment, the logical consequence would be to increase PEEP after the recruitment maneuver.
Another parameter of the PV-curve, VREC has been used as an indicator of recruited volume in several investigations [36,41,42], but it had never been validated with actual CT measurements. Especially in ventilation with FiO2 1.0, the VREC represents unstable lung units prone to collapse. In our results, there was a significant increase in VREC after the recruitment maneuver, which correlated with the observed changes in VPOOR and VNON. This means that a significant portion of the recruited lung still collapsed endexpiratory, probably because we did not increase PEEP after the recruitment. Therefore, VREC could not only serve as a measurement for recruited lung, but also for the lung in danger of being de-recruited.
Conclusion
The findings of this study suggest that an improvement in oxygenation does not necessarily mean recruitment of nonaerated lung and that measures to recruit collapsed lung will have unpredictable results on gas exchange. The effects were diverse in magnitude and predicted changes in oxygenation and shunt did not correlate with alveolar recruitment. Poorly aerated lung regions were the main determinant for the observed changes in plateau pressure, respiratory system compliance and recruitable volume.
Lung recruitment might be grossly overestimated when simply looking at the PaO2. Also, the effects of a standard open-lung maneuver or currently advocated PEEP strategies on recruitment are relatively small [43]. Because we did not focus on optimal recruitment but on the relationship of certain parameters with changes in lung aeration, however, we used a recruitment procedure as proposed previously. Obviously, this specific recruitment maneuver was not sufficient to homogenize lung ventilation. Common treatment strategies in ARDS aim to improve oxygenation, and the mechanical properties of ventilator settings are adjusted according to gas exchange parameters (e.g. PEEP/FiO2 tables). The poor correlation we have found between oxygenation and recruitment might be a reason that several of these approaches have failed to show a benefit for the patients treated this way. We speculate that parameters other than gas exchange should be investigated as targets in treating these patients.
Key messages
• The respiratory mechanics parameters correlated with the amount of aerated lung better than gas exchange parameters, with the venous admixture being the only oxygenation parameter that correlated with nonaerated lung volume.
• A recruitment maneuver without PEEP adjustment led to a decrease of nonaerated lung, presumably towards poorly aerated lung mainly. This did not significantly alter the distribution of a tidal breath to the differently aerated lung regions, however, implying that there was no reduction in the opening and collapse of alveoli.
• Changes in aerated and nonaerated lung volumes after the recruitment maneuver were adequately represented by changes in plateau pressure, respiratory system compliance and recruitable volume.
• An improvement in oxygenation does not necessarily mean recruitment of nonaerated lung and measures to recruit collapsed lung will have unpredictable results on gas exchange.
• In the clinical context, or even worse in clinical studies, using PaO2 changes as a surrogate for lung recruitment should be done with caution, as it lacks a clear physiological basis.
Abbreviations
ARDS = acute respiratory distress syndrome; CINF = maximum inflation compliance; CRS = compliance of the respiratory system; CT = computer tomography; E = elastance; FiO2 = fraction of inspired oxygen; HU = Hounsfield unit; LIP = lower inflection point; PaO2 = arterial partial oxygen pressure; PEEP = positive end-expiratory pressure; PV-curve = (respiratory system) pressure volume curve; QVA/QT = venous admixture (according to Berggren's formula); RM = recruitment maneuver 45 cmH2O/40 s; = ventilation-perfusion distribution; VD/VT = physiological dead space (according to Bohr/Enghoff's formula); VGAS = intrathoracic gas volume; VHYP = volume of hyperinflated lung parenchyma; VNON = volume of nonaerated lung parenchyma; VNORM = volume of normally aerated lung parenchyma; VPOOR = volume of poorly aerated lung parenchyma; VREC = recruitable volume at end-expiration; VTISS = intrathoracic tissue volume.
Competing interests
DH has received an unrestricted research grant in 2003 from Hamilton Medical Deutschland GmbH, by which the study was partially funded. All other authors declare that they have no competing interests.
Authors' contributions
DH conceived the study, participated in the design and execution of the study, the analysis of data and finalized the manuscript. PP participated in analysis and interpretation of the data and revised the manuscript. RD participated in the animal experiments and the analysis of data. AU participated in the animal experiments and the analysis of multi-slice CT data. AM did the radiology studies and participated in the analysis of multi-slice CT data. RR participated in the study design and coordination and helped to draft the manuscript. RK participated in the study design, interpretation of results and writing of the manuscript.
Supplementary Material
Additional File 1
Additional information on materials and methods.
Click here for file
Acknowledgements
We are thankful to Ingo Weber, MD, Anesthesiology Department of the University Hospital RWTH Aachen, for English editing of the manuscript. We would also like to thank Thaddeus Stopinski and Kira Scherer, Institute for Animal Research at the University Hospital RWTH Aachen, for their invaluable help and assistance.
Figures and Tables
Figure 1 Correlation of expiratory multi-slice CT lung volumes with respiratory mechanics and gas exchange parameters. CRS, static compliance of respiratory system; PaO2, arterial partial oxygen pressure; Pmcd, pressure of maximum compliance decrease on inflation curve; QVA/QT, venous admixture; VNON, nonaerated lung volume; VPOOR, poorly aerated lung volume.
Figure 2 Representative CT scan of one animal at three different levels (apical, middle, basal). (a) Expiratory occlusion (10 cmH2O) before and after the recruitment maneuver. Lung volumes in this animal changed as follows: VHYP +1%, VNORM +15%, VPOOR +17%, VNON -30%, VGAS +11%. (b) Inspiratory occlusion at plateau pressure before and after the recruitment maneuver. Lung volumes in this animal changed as follows: VHYP +6%, VNORM +17%, VPOOR +26%, VNON -29%, VGAS +17%. VGAS, intrathoracic gas volume; VHYP, volume of hyperinflated lung parenchyma; VNON, volume of nonaerated lung parenchyma; VNORM, volume of normally aerated lung parenchyma; VPOOR, volume of poorly aerated lung parenchyma.
Figure 3 Distribution of differently aerated lung volumes. Individual curves for eight animals before (solid line) and after (dashed line) a recruitment maneuver. Multi-slice CT of the whole lung with characterization of lung parenchyma according to Hounsfield units at end-expiration. VHYP, volume of hyperinflated lung parenchyma; VNON, volume of nonaerated lung parenchyma; VNORM, volume of normally aerated lung parenchyma; VPOOR, volume of poorly aerated lung parenchyma.
Figure 4 Fractional changes in investigated parameters (means with confidence intervals). Cinf, maximum inflation compliance; Crs, static compliance of respiratory system; PaO2, arterial partial oxygen pressure; Pplat, plateau pressure; QVA/QT, venous admixture; VNON, nonaerated lung volume; VNORM, normally aerated lung volume; VPOOR, poorly aerated lung volume; Vrec, recruitable volume at PEEP.
Figure 5 Correlation of the fractional changes (FC; %) of parameters with multi-slice CT lung volumes. Regression lines with 95% individual confidence intervals.(a) Insignificant correlation of arterial partial oxygen pressure (PaO2) with nonaerated lung. Note the large confidence intervals. (b) Insignificant correlation of venous admixture (QVA/QT) with nonaerated lung. (c) Close relation between changes in plateau pressure (PPLAT) and poorly aerated lung. (d) Pressure of maximum compliance increase on inflation curve (Pmci) correlates non-linearly with aerated volume (volume of normally aerated lung parenchyma (VNORM) + volume of poorly aerated lung parenchyma (VPOOR)). Note the sharp increase of Pmci beyond 20% increase in aerated lung volume.
Table 1 Lung volumes measured by multi-slice computer tomography
Pre-recruitment maneuver Post-recruitment maneuver P-value fractional change (%)
Expiration
VHYP (ml) 60 ± 21 67 ± 28 0.025 11.2 ± 10
VNORM (ml) 577 ± 142 649 ± 206 0.036 11.0 ± 12
VPOOR (ml) 406 ± 83 493 ± 112 0.017 21.7 ± 18
VNON (ml) 357 ± 53 275 ± 72 0.012 -23.3 ± 15
VTOT (ml) 1401 ± 136 1483 ± 175 0.025 5.8 ± 5
VGAS (ml) 629 ± 83 711 ± 133 0.012 13.1 ± 10
VTISS (ml) 838 ± 62 832 ± 60 0.263 -
Inspiration
VHYP (ml) 109 ± 38 115 ± 42 0.093 -
VNORM (ml) 789 ± 140 889 ± 197 0.012 12.4 ± 12
VPOOR (ml) 397 ± 94 478 ± 124 0.017 20.9 ± 18
VNON (ml) 295 ± 54 232 ± 75 0.012 -22.3 ± 16
VTOT (ml) 1589 ± 139 1713 ± 150 0.012 7.9 ± 5
VGAS (ml) 838 ± 84 939 ± 128 0.012 12.5 ± 8
VTISS (ml) 819 ± 56 838 ± 64 0.263 -
Data are reported as mean ± SD. VGAS, total lung gas volume; VHYP, hyperinflated lung volume; VNON, non-aereated lung volume; VNORM, normally aereated lung volume; VPOOR, poorly aerated lung volume; VTISS, total lung tissue volume; VTOT, total lung volume.
Table 2 Gas exchange and hemodynamics parameters
Pre-recruitment maneuver Post-recruitment maneuver P-value fractional change (%)
PaO2 (mmHg) 71 ± 21 94 ± 28 0.017 33.0 ± 23
PaCO2 (mmHg) 81 ± 20 81 ± 19 0.575 -
PvO2 (mmHg) 45 ± 10 49 ± 10 0.093 -
QVA/QT (%) 50.2 ± 9.9 39.3 ± 8.6 0.036 -20.8 ± 16
VD/VT (%) 84 ± 2.9 83.7 ± 3.4 0.31 -
HR (min-1) 85 ± 84 77 ± 21 0.025 -11.3 ± 9
MAP (mmHg) 80 ± 15 83 ± 24 0.498 -
QT (l min-1) 3.7 ± 0.2 3.4 ± 0.2 0.018 -9.6 ± 6
VO2 (ml min-1) 138 ± 39 141 ± 35 0.889 -
DO2 (ml min-1) 401 ± 118 412 ± 101 0.575 -
EVLWI (ml kg-1) 20.6 ± 7.9 21.1 ± 9.6 0.499 -
Data are reported as mean ± SD. DO2, oxygen delivery; EVLWI, extravascular lung water index; HR, heart rate; MAP, mean arterial pressure; PaCO2, arterial carbon dioxide partial pressure; PaO2, arterial partial oxygen pressure; , mixed venous partial oxygen pressure; QT, cardiac output; QVA/QT, venous admixture; VD/VT, dead space fraction; VO2, oxygen consumption.
Table 3 Respiratory mechanics parameters
Pre-recruitment maneuver Post-recruitment maneuver P-value fractional change(%)
PIP (cmH2O) 36.6 ± 4 31.1 ± 3.7 0.012 -12.5 ± 6
PPLAT (cmH2O) 30.7 ± 3.1 27.2 ± 2.8 0.028 -13.8 ± 7
CRS (ml cmH2O-1) 13.5 ± 2.2 17.9 ± 2.6 0.028 34.5 ± 17
Pmci,INF (cmH2O) 22.4 ± 11.9 32.3 ± 5.4 0.046 113 ± 192
Pmcd,INF (cmH2O) 43.3 ± 9.5 56.6 ± 15.5 0.075 -
CINF (ml cmH2O-1) 24.4 ± 14.7 42.0 ± 14.5 0.028 101.8 ± 92
Pmci,DEF (cmH2O) 9.4 ± 2.2 9.9 ± 1.1 0.463 -
Pmcd,DEF (cmH2O) 19.9 ± 2.0 21.4 ± 1.9 0.046 7.0 ± 0.7
VREC (ml) 183 ± 135 256 ± 145 0.028 66.5 ± 47
Data are reported as mean ± SD. CINF, maximum inflation compliance; PIP, peak inspiratory pressure; PPLAT, plateau pressure; CRS, static respiratory system compliance; Pmci,DEF, point of maximum compliance increase of deflation curve; Pmcd,DEF, point of maximum compliance decrease of deflation curve; Pmcd,INF, point of maximum compliance decrease of inflation curve; Pmci,INF, point of maximum compliance increase of inflation curve; VREC, recruitable volume at 10 cmH2O.
==== Refs
Gattinoni L Caironi P Pelosi P Goodman LR What has computed tomography taught us about the acute respiratory distress syndrome? Am J Respir Crit Care Med 2001 164 1701 1711 11719313
Hubmayr RD Perspective on lung injury and recruitment: a skeptical look at the opening and collapse story Am J Respir Crit Care Med 2002 165 1647 1653 12070067 10.1164/rccm.2001080-01CP
Tokics L Hedenstierna G Strandberg A Brismar B Lundquist H Lung collapse and gas exchange during general anesthesia: effects of spontaneous breathing, muscle paralysis, and positive end-expiratory pressure Anesthesiology 1987 66 157 167 3813078
Lachmann B Open up the lung and keep the lung open Intensive Care Med 1992 18 319 321 1469157 10.1007/BF01694358
Rossaint R Hahn SM Pappert D Falke KJ Radermacher P Influence of mixed venous PO2 and inspired O2 fraction on intrapulmonary shunt in patients with severe ARDS J Appl Physiol 1995 78 1531 1536 7615466
Brimioulle S Julien V Gust R Kozlowski JK Naeije R Schuster DP Importance of hypoxic vasoconstriction in maintaining oxygenation during acute lung injury Crit Care Med 2002 30 874 880 11940762 10.1097/00003246-200204000-00027
Gattinoni L Pelosi P Vitale G Pesenti A D'Andrea L Mascheroni D Body position changes redistribute lung Computed-Tomographic density in patients with acute respiratory failure Anesthesiology 1991 74 15 23 1986640
Crotti S Mascheroni D Caironi P Pelosi P Ronzoni G Mondino M Marini JJ Gattinoni L Recruitment and derecruitment during acute respiratory failure: a clinical study Am J Respir Crit Care Med 2001 164 131 140 11435251
Institute of Laboratory Animal Resources, National Research Council Guide for the Care and Use of Laboratory Animals 1996 National Academy Press Washington, D.C. Ref Type: Internet Communication
Jonson B Beydon L Brauer K Mansson C Valind S Grytzell H Mechanics of respiratory system in healthy anesthetized humans with emphasis on viscoelastic properties J Appl Physiol 1993 75 132 140 8376259
Dembinski R Max M Bensberg R Rossaint R Kuhlen R Pressure Support Compared with Controlled Mechanical Ventilation in Experimental Lung Injury Anesth Analg 2002 94 1570 1576 12032029 10.1097/00000539-200206000-00037
Puybasset L Cluzel P Gusman P Grenier P Preteux F Rouby JJ Regional distribution of gas and tissue in acute respiratory distress syndrome. I. Consequences for lung morphology. CT Scan ARDS Study Group Intensive Care Med 2000 26 857 869 10990099 10.1007/s001340051274
Puybasset L Gusman P Muller JC Cluzel P Coriat P Rouby JJ Regional distribution of gas and tissue in acute respiratory distress syndrome. III. Consequences for the effects of positive end-expiratory pressure. CT Scan ARDS Study Group. Adult Respiratory Distress Syndrome Intensive Care Med 2000 26 1215 1227 11089745 10.1007/s001340051340
Berggren SM The oxygen deficit of arterial blood caused by non-ventilated parts of the lung Acta Physiol Scand Suppl 1942 4 4 92
Polese G Rossi A Appendini L Brandi G Bates JH Brandolese R Partitioning of respiratory mechanics in mechanically ventilated patients J Appl Physiol 1991 71 2425 2433 1778942
Venegas JG Harris RS Simon BA A comprehensive equation for the pulmonary pressure-volume curve J Appl Physiol 1998 84 389 395 9451661
Pelosi P Goldner M McKibben A Adams A Eccher G Caironi P Losappio S Gattinoni L Marini JJ Recruitment and derecruitment during acute respiratory failure: an experimental study Am J Respir Crit Care Med 2001 164 122 130 11435250
Marini JJ Recruitment maneuvers to achieve an "open lung" – whether and how? Crit Care Med 2001 29 1647 1648 11505154 10.1097/00003246-200108000-00032
Grasso S Mascia L del Turco M Malacarne P Giunta F Brochard L Slutsky AS Marco RV Effects of recruiting maneuvers in patients with acute respiratory distress syndrome ventilated with protective ventilatory strategy Anesthesiology 2002 96 795 802 11964585 10.1097/00000542-200204000-00005
Kaisers U Max M Walter J Kuhlen R Pappert D Falke KJ Rossaint R Partial liquid ventilation with small volumes of FC 3280 increases survival time in experimental ARDS Eur Respir J 1997 10 1955 1961 9311485 10.1183/09031936.97.10091955
Neumann P Hedenstierna G Ventilation-perfusion distributions in different porcine lung injury models Acta Anaesthesiol Scand 2001 45 78 86 11152038 10.1034/j.1399-6576.2001.450113.x
Luecke T Roth H Herrmann P Joachim A Weisser G Pelosi P Quintel M PEEP decreases atelectasis and extravascular lung water but not lung tissue volume in surfactant-washout lung injury Intensive Care Med 2003 29 2026 2033 12897993 10.1007/s00134-003-1906-9
Grasso S Terragni P Mascia L Fanelli V Quintel M Herrmann P Hedenstierna G Slutsky AS Ranieri VM Airway pressure-time curve profile (stress index) detects tidal recruitment/hyperinflation in experimental acute lung injury Crit Care Med 2004 32 1018 1027 15071395 10.1097/01.CCM.0000120059.94009.AD
Downie JM Nam AJ Simon BA Pressure-volume curve does not predict steady-state lung volume in canine lavage lung injury Am J Respir Crit Care Med 2004 169 957 962 14764430 10.1164/rccm.200305-614OC
Rouby JJ Puybasset L Nieszkowska A Lu Q Acute respiratory distress syndrome: lessons from computed tomography of the whole lung Crit Care Med 2003 31 S285 S295 12682454 10.1097/01.CCM.0000057905.74813.BC
Brower RG Morris A MacIntyre N Matthay MA Hayden D Thompson T Clemmer T Lanken PN Schoenfeld D Effects of recruitment maneuvers in patients with acute lung injury and acute respiratory distress syndrome ventilated with high positive end-expiratory pressure Crit Care Med 2003 31 2592 2597 14605529
Ashbaugh DG Bigelow DB Petty TL Levine BE Acute respiratory distress in adults Lancet 1967 2 319 323 4143721 10.1016/S0140-6736(67)90168-7
Neumann P Rothen HU Berglund JE Valtysson J Magnusson A Hedenstierna G Positive end-expiratory pressure prevents atelectasis during general anaesthesia even in the presence of a high inspired oxygen concentration Acta Anaesthesiol Scand 1999 43 295 301 10081535 10.1034/j.1399-6576.1999.430309.x
Hedenstierna G Tokics L Strandberg A Lundquist H Brismar B Correlation of gas exchange impairment to development of atelectasis during anaesthesia and muscle paralysis Acta Anaesthesiol Scand 1986 30 183 191 3085429
Brower RG Lanken PN MacIntyre N Matthay MA Morris A Ancukiewicz M Schoenfeld D Thompson BT Higher versus lower positive end-expiratory pressures in patients with the acute respiratory distress syndrome N Engl J Med 2004 351 327 336 15269312 10.1056/NEJMoa032193
Gattinoni L Tognoni G Pesenti A Taccone P Mascheroni D Labarta V Malacrida R Di Giulio P Fumagalli R Pelosi P Brazzi L Latini R Effect of prone positioning on the survival of patients with acute respiratory failure N Engl J Med 2001 345 568 573 11529210 10.1056/NEJMoa010043
Barnas GM Green MD Mackenzie CF Fletcher SJ Campbell DN Runcie C Broderick GE Effect of posture on lung and regional chest wall mechanics Anesthesiology 1993 78 251 259 8439019
Maggiore SM Jonson B Richard JC Jaber S Lemaire F Brochard L Alveolar derecruitment at decremental positive end-expiratory pressure levels in acute lung injury: comparison with the lower inflection point, oxygenation, and compliance Am J Respir Crit Care Med 2001 164 795 801 11549535
Vieillard-Baron A Prin S Schmitt JM Augarde R Page B Beauchet A Jardin F Pressure-volume curves in acute respiratory distress syndrome: clinical demonstration of the influence of expiratory flow limitation on the initial slope Am J Respir Crit Care Med 2002 165 1107 1112 11956053
Jonson B Svantesson C Elastic pressure-volume curves: what information do they convey? Thorax 1999 54 82 87 10343639
Jonson B Richard JC Straus C Mancebo J Lemaire F Brochard L Pressure-volume curves and compliance in acute lung injury: evidence of recruitment above the lower inflection point Am J Respir Crit Care Med 1999 159 1172 1178 10194162
Henzler D Orfao S Rossaint R Kuhlen R Modification of a sigmoidal equation for the pulmonary pressure-volume curve for asymmetric data J Appl Physiol 2003 95 2183 2184 14555679
Harris RS Hess DR Venegas JG An objective analysis of the pressure-volume curve in the acute respiratory distress syndrome Am J Respir Crit Care Med 2000 161 432 439 10673182
Amato MB Barbas CS Medeiros DM Magaldi RB Schettino GP Lorenzi FG Kairalla RA Deheinzelin D Munoz C Oliveira R Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome N Engl J Med 1998 338 347 354 9449727 10.1056/NEJM199802053380602
Lichtwarck-Aschoff M Hedlund AJ Nordgren KA Wegenius GA Markstrom AM Guttmann J Sjostrand UH Variables used to set PEEP in the lung lavage model are poorly related Br J Anaesth 1999 83 890 897 10700789
Richard JC Brochard L Vandelet P Breton L Maggiore SM Jonson B Clabault K Leroy J Bonmarchand G Respective effects of end-expiratory and end-inspiratory pressures on alveolar recruitment in acute lung injury Crit Care Med 2003 31 89 92 12544999 10.1097/00003246-200301000-00014
Ranieri VM Eissa NT Corbeil C Chasse M Braidy J Matar N Milic-Emili J Effects of positive end-expiratory pressure on alveolar recruitment and gas exchange in patients with the adult respiratory distress syndrome Am Rev Respir Dis 1991 144 544 551 1892293
Grasso S Fanelli V Cafarelli A Anaclerio R Amabile M Ancona G Fiore T Effects of high versus low positive end-expiratory pressures in acute respiratory distress syndrome Am J Respir Crit Care Med 2005 171 1002 1008 15665322 10.1164/rccm.200407-940OC
|
16277708
|
PMC1297611
|
CC BY
|
2021-01-04 16:04:56
|
no
|
Crit Care. 2005 Jul 13; 9(5):R471-R482
|
utf-8
|
Crit Care
| 2,005 |
10.1186/cc3772
|
oa_comm
|
==== Front
Crit CareCritical Care1364-85351466-609XBioMed Central London cc37741627771010.1186/cc3774ResearchRecombinant human activated protein C resets thrombin generation in patients with severe sepsis – a case control study de Pont Anne-Cornélie JM [email protected] Kamran [email protected] Barbara A [email protected] Jonge Evert [email protected] Margreeth B [email protected] Joost CM [email protected]üller Harry R [email protected] Marcel [email protected] Intensivist, Department of Intensive Care, Academic Medical Center, University of Amsterdam, The Netherlands2 Laboratory Researcher, Department of Vascular Medicine, Academic Medical Center, University of Amsterdam, The Netherlands3 Clinical Epidemiologist, Department of Epidemiology and Biostatistics, Academic Medical Center, University of Amsterdam, The Netherlands4 Professor of Intensive Care Medicine, Department of Intensive Care, Academic Medical Center, University of Amsterdam, The Netherlands5 Head of the Laboratory of Vascular Medicine, Department of Vascular Medicine, Academic Medical Center, University of Amsterdam, The Netherlands6 Professor of Vascular Medicine, Department of Vascular Medicine, Academic Medical Center, University of Amsterdam, The Netherlands7 Professor of Internal Medicine, Department of Internal Medicine, Academic Medical Center, University of Amsterdam, The Netherlands2005 21 7 2005 9 5 R490 R497 24 4 2005 3 6 2005 24 6 2005 28 6 2005 Copyright © 2005 de Pont 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
Recombinant human activated protein C (rhAPC) is the first drug for which a reduction of mortality in severe sepsis has been demonstrated. However, the mechanism by which this reduction in mortality is achieved is still not clearly defined. The aim of the present study was to evaluate the dynamics of the anticoagulant, anti-inflammatory and pro-fibrinolytic action of rhAPC in patients with severe sepsis, by comparing rhAPC-treated patients with case controls.
Methods
In this prospectively designed multicenter case control study, 12 patients who were participating in the ENHANCE study, an open-label study of rhAPC in severe sepsis, were treated intravenously with rhAPC at a constant rate of 24 μg/kg/h for a total of 96 h. Twelve controls with severe sepsis matching the inclusion criteria received standard therapy. The treatment was started within 48 h after the onset of organ failure. Blood samples were taken before the start of the infusion and at 4, 8, 24, 48, 96 and 168 h, for determination of parameters of coagulation and inflammation.
Results
Sepsis-induced thrombin generation as measured by thrombin-antithrombin complexes and prothrombin fragment F1+2, was reset by rhAPC within the first 8 h of infusion. The administration of rhAPC did not influence parameters of fibrinolysis and inflammation. There was no difference in outcome or occurrence of serious adverse events between the treatment group and the control group.
Conclusion
Sepsis-induced thrombin generation in severely septic patients is reset by rhAPC within the first 8 h of infusion without influencing parameters of fibrinolysis and inflammation.
==== Body
Introduction
During severe sepsis, activation of the inflammatory cascade leads to cell damage and organ failure. In recent years, the importance of the cross-talk between coagulation and inflammation in severe sepsis has been well defined. This has led to the hypothesis that inhibitors of coagulation might have a dual effect, that is, interruption of the cascades of both coagulation and inflammation. Recombinant human activated protein C (rhAPC, drotrecogin alfa (activated), Xigris®) is the first drug for which a reduction of mortality in severe sepsis has been demonstrated [1]. Indeed, rhAPC is an effective anticoagulant and also has distinct anti-inflammatory effects, at least in vitro. However, the mechanism by which the reduction in mortality by rhAPC is achieved is still not clearly defined. Several mechanisms have been proposed. Firstly, rhAPC may inhibit the formation of thrombin by proteolytically degrading coagulation factors Va and VIIIa. Thrombin has a central role in coagulation due to its ability to convert fibrinogen to fibrin, but also as the most potent agonist of platelet activation [2]. Thrombin may also affect the production of inflammatory cytokines by binding to protease-activated receptors (PARs) in mononuclear cells [3]. Secondly, rhAPC may inhibit the action of plasminogen activator inhibitor type I (PAI-I), thereby restoring the suppressed fibrinolytic state during sepsis [4]. Thirdly, binding of rhAPC to the endothelial protein C receptor (EPCR) may influence gene expression profiles of cells by blocking nuclear factor kappa B nuclear translocation, which is required for increases in proinflammatory cytokines and adhesion molecules [5]. Direct activation of PAR-1 by the APC-EPCR complex is another mechanism by which APC may affect inflammation [6]. Fourthly, it is hypothesized that rhAPC inhibits the adherence of activated leukocytes to activated endothelium [5,7]. However, the relative importance of each of these mechanisms for the mortality reduction in severe sepsis by rhAPC is still unclear.
The aim of the current study was to evaluate the dynamics of the anticoagulant, anti-inflammatory and pro-fibrinolytic action of rhAPC in patients with severe sepsis, by comparing rhAPC-treated patients with case controls. For this purpose, we employed sensitive assays for the assessment of activation and inhibition of the coagulant, inflammatory and fibrinolytic system.
Methods
Study design
The ENHANCE study, an open-label study of rhAPC in severe sepsis, was conducted worldwide and more than 2000 patients were included. In the Netherlands, four sites participated in the current ENHANCE substudy, three academic hospitals and one large teaching hospital. After completion of the ENHANCE study, an equal number of patients with severe sepsis meeting identical inclusion criteria were prospectively enrolled in this substudy to serve as case controls. At the time of performance of the study, rhAPC was not yet licensed and available for routine treatment of patients with severe sepsis.
Patients
The study was approved by the institutional review board and written informed consent was obtained from all participants or their authorized representatives. Patients were eligible for the trial if they had a known or suspected infection, three or more signs of systemic inflammation and a sepsis-induced dysfunction of at least one organ system that lasted no longer than 48 h. In patients enrolled in the ENHANCE study, treatment with rhAPC was started within 48 h after they met the inclusion criteria. The time of starting the rhAPC infusion was considered as t = 0. In the controls, the time at which rhAPC would have been started if the patient had been in the treatment group, was considered as t = 0. Blood samples were taken at the same time points in both the treatment group and the control group.
Treatment
In the treatment group, rhAPC was administered intravenously at a constant rate of 24 μg/kg body weight per hour for a total duration of 96 h. The infusion was interrupted for 1 h before any percutaneous procedure and was resumed 1 h later. During the infusion of rhAPC, no other anticoagulant was administered except nadroparin or dalteparin in a prophylactic dose. Except for the administration of rhAPC, the treatment of patients in both groups was identical.
Blood collection
Platelet counts were routinely determined daily. Blood for the analysis of parameters of coagulation and inflammation was collected from an arterial line just before t = 0 and at 4, 8, 24, 48, 96 and 168 h. Blood for platelet counts and cytokine assays was collected in K3-EDTA-containing tubes. All other blood samples were collected in citrated vacutainer tubes. Plasma was prepared by centrifugation of blood at 2500 × g twice for 20 mins at 16°C, followed by storage at -80°C until assays were performed.
Laboratory assays
The plasma concentrations of thrombin-antithrombin complexes (TAT), prothrombin fragment F1+2 (F1+2), and plasmin-antiplasmin complexes (PAP) were measured by ELISA (Dade Behring, Marburg, Germany). Activated partial thromboplastin time (APTT) and prothrombin time (PT) were performed on an automated coagulation analyzer (Behring Coagulation System, BCS) with reagents and protocols from the manufacturer (Dade Behring). Protein C was determined using the Coamatic protein C activity kit from Chromogenix (Milan, Italy). Total protein S antigen was assayed by ELISA using antibodies from DAKO (Glostrup, Denmark). Protein C and S deficiencies were defined as activity levels of <65% of the level measured in normal pool plasma. Free protein S was measured by precipitating the C4b-binding protein-bound fraction with polyethylene glycol 8000 and measuring the concentration of free protein S in the supernatant. Free protein S deficiency was defined as an activity of <26% of the total protein S level. Protein C inhibitor (PCI) was determined by ELISA as described by Elisen et al. [8]. Normal PCI levels vary from 55 to 142% of the PCI level measured in normal pool plasma. Thrombin-activatable fibrinolysis inhibitor (TAFI) antigen levels were assayed by ELISA as described by Mosnier et al. [9]. Normal TAFI levels vary from 50 to 150% of the TAFI level measured in normal pool plasma. D-dimers were assayed with an ELISA (Asserachrom D-Dimer, Roche, Almere, the Netherlands). Platelet counts were assessed by a Cell-dyn 4000 analyzer (Abbott Laboratories, Abbott Park, IL, USA). Tumor necrosis factor alpha (TNF-alpha), interleukin (IL)-6, and IL-10 were measured using commercial ELISA kits (Central Laboratory of the Netherlands Red Cross Blood Transfusion Service, Amsterdam, the Netherlands). The assay detection limits were 3 pg/ml for TNF-alpha, 0.6 pg/ml for IL-6 and 1.2 pg/ml for IL-10. All assays were performed by the Laboratory for Clinical Chemistry, Hematology and Transfusion and the Laboratory of Experimental Vascular Medicine at the Academic Medical Center Amsterdam, the Netherlands.
Evaluation of patients
Patients were followed for 28 days after enrollment or until death. Baseline characteristics were assessed within 24 h prior to enrollment. Microbiologic culture results were assessed daily from enrollment through day 28.
Statistical analysis
Data were analyzed using SPSS for Windows, v11.0 (SPSS Inc, Chicago, IL, USA). Differences between the treatment group and the case control group were tested by analysis of repeated measures using mixed linear models. Changes in time within the same group were analyzed by 1-way analysis of variance. Values are given as means ± SD. Significance was defined as p < 0.05.
Results
Patient characteristics
During the ENHANCE study, 12 patients were enrolled in the substudy at the four participating sites. Another 12 patients with severe sepsis were enrolled prospectively as case controls at two of the four participating sites. The baseline characteristics of the two patient groups are shown in Table 1. There were more patients with malignancy in the control group, but all other baseline characteristics were similar. The lung was the most common site of infection in both groups and Gram-negative infections were most common. The time elapsed between meeting the inclusion criteria and t = 0 was 12.3 ± 13.2 h in the rhAPC group versus 26.7 ± 12.5 h in the control group (p = 0.01).
Thrombin generation
Administration of rhAPC resulted in a reduction of sepsis-induced thrombin generation, as reflected by a decrease in the levels of TAT and F1+2 to 45 and 30% below baseline, respectively, within 8 h, without a significant change for the remaining 7 days (Fig. 1). In the control group, TAT and F1+2 levels increased to 2 and 1.4 times baseline, respectively, within 4 days. The difference in F1+2 levels between the two groups reached significance after 8 h (p = 0.03) and remained significant for the remaining 7 days (p = 0.03). In the rhAPC group, the APTT rose to a maximum of 1.4 times baseline within 4 h after the start of the infusion (p = 0.004), whereas the APTT remained stable in the control group. In both treatment groups, the PT decreased over time. However, this decrease only reached statistical significance in the control group on day 4 and day 7 (data not shown).
Protein C pathway
Ninety-two percent of all patients were protein C deficient at baseline, with mean protein C levels of 44 ± 20% in the rhAPC group and 47 ± 12% in the control group (NS). As shown in Fig. 2, protein C levels normalized in the course of 2 days in both treatment groups. All patients were deficient in protein C inhibitor with mean levels of 16 ± 13% in the rhAPC group and 21 ± 11% in the control group (NS). The levels of protein C inhibitor more than doubled over time in both groups (p = 0.004). The increase was more pronounced in the control group, but the difference between groups did not reach statistical significance (Fig. 2).
At baseline, deficiency in total and free protein S was present in 63 and 79% of all patients, respectively. Mean total and free protein S levels in the rhAPC group were 53 ± 17% and 19 ± 7%, respectively, and in the control group 60 ± 20% and 19 ± 10%, respectively (NS). The levels of total and free protein S normalized in the course of 2 and 4 days, respectively (Fig. 3).
Platelet counts
Baseline platelet counts were 173 ± 121 × 109/l in the rhAPC group and 156 ± 85 × 109/l in the control group (NS). In the rhAPC group, there was a trend toward an increase in platelet count: platelets increased from 173 ± 121 to 270 ± 190 × 109/l on day 6 (p = 0.06). In the control group, the platelet count increased from 156 ± 85 to 202 ± 189 × 109/l on day 6 (p = 0.44). The difference between the two groups was too small to reach statistical significance.
Fibrinolysis
Parameters of fibrinolysis are shown in Fig. 4. Baseline D-dimer levels did not differ significantly between groups and did not change significantly over time. PAP levels tended to increase in the rhAPC group, whereas they remained stable in the control group. However, the difference between groups was too small to reach statistical significance. In both groups, TAFI levels were depressed at baseline (56 ± 26% in the rhAPC group and 64 ± 16% in the control group), returning to normal in the course of 4 days, without a significant difference between groups.
Cytokines
The time course of cytokine levels is depicted in Fig. 5. Levels of TNF-alpha remained stable in both the rhAPC group and the control group. Levels of IL-6 and IL-10 gradually decreased over time in both groups, without a significant difference between groups.
Outcome
The outcome of patients in both groups is summarized in Table 2. In total, five patients died within 28 days, two in the rhAPC group and three in the control group, which comes down to a 28-day mortality rate of 17 and 25%, respectively (p = 1.0). In this small sample of patients, there was no statistically significant difference between the rhAPC group and the control group with respect to length of ICU stay, length of hospital stay and percentage of patients discharged home.
Complications
The occurrence of adverse events is summarized in Table 2. A serious bleeding event occurred in only one patient in the rhAPC group. In this patient, a central venous line was inserted erroneously without stopping the rhAPC infusion. The subsequent ongoing bleeding from the puncture site ultimately required a red blood cell (RBC) transfusion. Blood transfusion requirements were similar in the rhAPC group and the control group (0.44 ± 0.53 versus 0.23 ± 0.35 RBC units per day, respectively, p = 0.27).
Discussion
In the present clinical study, we studied the dynamics of the anticoagulant, pro-fibrinolytic and anti-inflammatory action of rhAPC when used in severe sepsis, by comparing rhAPC treated patients with case controls. We demonstrated that sepsis-induced thrombin generation was reset by rhAPC, as reflected by a decrease in TAT and F1+2 levels within 8 h of infusion. We did not find any influence of rhAPC on parameters of fibrinolysis and inflammation. Although the delay between meeting the inclusion criteria and t = 0 was longer in the control group, we do not think that this difference has influenced our results. Indeed, shifting the control group curves in Figs 1, 2, 3, 4, 5 to the right for 12 h did not change the results of the comparison between the two treatment groups.
The inhibition of thrombin generation by rhAPC might be the main mechanism by which mortality reduction in patients with severe sepsis was achieved in the PROWESS study [1]. Mortality in severe sepsis is usually due to multiple organ failure, which is believed to be caused by microvascular thrombosis, impairing the blood supply to various organs [10,11]. Under physiological circumstances, thrombin generation is regulated by the protein C system in order to prevent microvascular thrombosis. During sepsis, however, the expression of thrombomodulin and EPCR on the endothelial cell surface is downregulated, leading to inadequate activation of protein C and thus to inadequate inhibition of thrombin generation.
Our findings confirm the results of Dhainaut et al., who demonstrated that treatment with rhAPC attenuates thrombin generation, as reflected by a significant inhibition of TAT and F1+2 [12]. In our study, the inhibition was even more pronounced: treatment with rhAPC prevented the increase in thrombin generation that occurred in the control group. Interestingly, TAT and F1+2 levels did not change from 8 h until 7 days after starting the treatment, even after stopping the rhAPC infusion. These results are in contrast with those of Dhainaut et al., who found an increase in levels of TAT and F1+2 on day 5. There are several possible explanations for this difference. Firstly, we did not take measurements on days 5 and 6 and might have missed a transient rise in thrombin generation. Secondly, the rhAPC group in the PROWESS study might have been more severely ill at inclusion, as the mean APACHE II score was higher than in our rhAPC group (24.6 ± 7.6 versus 21 ± 6). It is conceivable that in more severely ill patients, normalization of thrombin generation takes more time. Thirdly, the time from inclusion to drug infusion was 17.5 ± 12.8 h in the PROWESS study, as compared with 12.3 ± 13.2 h in our study. It is also conceivable that the shorter delay to treatment might have influenced the speed of recovery. If rhAPC is indeed able to reset thrombin generation within 8 h in less severely ill patients when treated within 12 h of admission, one could argue that, under these circumstances, a shorter duration of rhAPC infusion might be sufficient to achieve the same extent of inhibition of thrombin generation. This could have important consequences for the recommended duration of treatment. However, based on the results of the present study, one cannot conclude that limitation of the duration of rhAPC treatment would yield the same results. Additional studies are needed to determine under which circumstances the duration of rhAPC infusion can be limited without influencing efficacy.
At baseline, 92% of our septic patients were protein C deficient with a mean protein C level of 45.8%. This finding is consistent with the results of earlier studies. Boldt et al. found a baseline protein C level of 47.8% in septic patients [13] and in the PROWESS study, Bernard et al. found median baseline protein C levels of 47 and 50% in the rhAPC group and the control group, respectively [1]. The depletion of protein C during sepsis is caused by a combination of degradation of protein C by neutrophil elastase and inadequate biosynthesis in the liver [11,14]. In our study, the protein C levels returned to normal in the course of 2 days in both treatment groups, whereas in the study by Dhainaut et al., normalization of protein C levels took 3.5 days in the rhAPC group and 5 days in the control group [12]. The increased time needed for the normalization of protein C levels might reflect the greater severity of illness of patients in this study.
In the present study, we did not find a convincing effect of the administration of rhAPC on fibrinolysis. The levels of D-dimers remained unchanged over time in both the rhAPC group and the control group. This is in contrast with the findings of Bernard et al., who found a significant decrease in D-dimer levels in the rhAPC group as compared with the control group [1]. The fact that we did not find such an effect may be due to the small number of patients and the great interpatient variability in D-dimer levels. PAP levels showed a tendency to increase in the rhAPC group, but this increase was too small to reach statistical significance. In agreement with our findings, Dhainaut et al. did not find an effect of rhAPC on PAI-1, a marker of fibrinolysis, when they used the method of repeated measurements [12]. They concluded that their results do not provide a strong basis for a pro-fibrinolytic effect of rhAPC, and our results support this conclusion.
In the present study, we did not find an effect of rhAPC on cytokine levels. Levels of IL-6 and IL-10 gradually declined to normal in the course of 2 days and the level of TNF-alpha remained unchanged over time in both treatment groups. In the PROWESS study, the decrease in IL-6 levels was significantly greater in the rhAPC group as compared with the control group [1]. However, in the post-hoc analysis of the PROWESS data by Dhainaut et al., there were no significant differences in IL-6 levels between the rhAPC group and the control group [12]. Moreover, Dhainaut et al. did not find any difference in levels of TNF-alpha and IL-10 between the two treatment groups. Our findings confirm these results. Dhainaut et al. conclude that their results do not provide a strong basis for a systemic anti-inflammatory effect of rhAPC in vivo at the therapeutic dose used. Our results support this conclusion. Indeed, the anti-inflammatory effect of rhAPC has only been demonstrated in vitro to date [15], using rhAPC concentrations 100- to 1000-fold the concentration achieved in therapeutic circumstances [16,17].
In the present study, no difference in outcome was found between the rhAPC group and the control group, which is probably due to the small number of patients. The numbers of serious adverse events did not differ between groups.
Conclusion
This study demonstrates that rhAPC resets sepsis-induced thrombin generation within the first 8 h of infusion, without influencing parameters of fibrinolysis and inflammation.
Key messages
• Recombinant human activated protein C resets thrombin generation within the first 8 h of infusion.
• The administration of recombinant activated protein C does not influence parameters of fibrinolysis and inflammation.
Abbreviations
APC = activated protein C; APTT = activated partial thromboplastin time; COPD = chronic obstructive pulmonary disease; EDTA = ethylene diamine tetraacetic acid; ELISA = enzyme-linked immunosorbent assay; ENHANCE = extended evaluation of recombinant activated protein C; EPCR = endothelial protein C receptor; F1+2 = prothrombin fragment F1+2; ICU = intensive care unit; IL = interleukin; NS = not significant; PAI-1 = plasminogen activator inhibitor type 1; PAP = plasmin-antiplasmin complexes; PAR-1 = protease activated receptor type 1; PCI = protein C inhibitor; PROWESS = protein C worldwide evaluation in severe sepsis; PT = prothrombin time; rhAPC = recombinant human activated protein C; SD = standard deviation; SOFA = sepsis-related organ failure assessment ; TAFI = thrombin-activatable fibrinolysis inhibitor; TAT = thrombin-antithrombin complexes; TNF = tumor necrosis factor.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
ACJMdP carried out the data collection and drafted the manuscript. KB and JCMM were responsible for the laboratory analysis. BAH performed the statistical analysis. EdJ and MBV participated in the coordination of the study. HB participated in the study design and helped to draft the manuscript. ML conceived the study, created the design and helped to draft the manuscript. All authors read and approved the final manuscript.
Acknowledgements
This study was financially supported by Eli Lilly and Company, Indianapolis, IN, USA. In addition to the authors, the following institutions and investigators participated in the study: Groningen: University Medical Center Groningen, Department of Intensive Care: H Delwig; Rotterdam, University Medical Center Rotterdam, Surgical Intensive Care Unit: B van den Hoven; and Zwolle, Isala Klinieken: F Snellen.
Figures and Tables
Figure 1 Levels of TAT and F1+2. Plasma levels of (a) TAT and (b) F1+2 in the rhAPC group (▴) and the control group (○). Data represent mean ± SD. F1+2, prothrombin fragment F1+2; rhAPC, recombinant human activated protein C; TAT, thrombin-antithrombin complexes.
Figure 2 Levels of protein C and protein C inhibitor. Plasma levels of (a) protein C and (b) protein C inhibitor in the rhAPC group (▴) and the control group (○). The levels of protein C and protein C inhibitor are expressed as the percentage of the level measured in normal pool plasma. Data represent mean ± SD. rhAPC, recombinant human activated protein C.
Figure 3 Levels of total and free protein S. Plasma levels of (a) total protein S and (b) free protein S in the rhAPC group (▴) and the control group (○). The levels of total protein S are expressed as the percentage of the level measured in normal pool plasma. The levels of free protein S are expressed as a percentage of total protein S. Data represent mean ± SD. rhAPC, recombinant human activated protein C.
Figure 4 Fibrinolysis. Plasma levels of (a) D-dimer, (b) PAP and (c)TAFI-Ag in the rhAPC group (▴) and the control group (○). TAFI levels are expressed as the percentage of the level measured in normal pool plasma. Data represent mean ± SD. PAP, plasmin-antiplasmin complexes; TAFI-Ag, thrombin-activatable fibrinolysis inhibitor antigen; rhAPC, recombinant human activated protein C.
Figure 5 Cytokines. Plasma levels of (a) TNF-alpha, (b) IL-6 and (c) IL-10 in the rhAPC group (▴) and the control group (○). Assay detection limits were 3.0 pg/ml for TNF-alpha, 0.6 pg/ml for IL-6 and 1.2 pg/ml for IL-10. Data represent mean ± SD. IL, interleukin; rhAPC, recombinant human activated protein C; TNF, tumor necrosis factor.
Table 1 Baseline characteristics of the patients
Characteristic rhAPC group (n = 12) Control group (n = 12)
Age, years 57 ± 16 66 ± 14
Male sex (%) 7 (58) 7 (58)
Pre-existing conditions, n 7 16
Malignancy 2 7
Diabetes 1 3
Hypertension 2 2
COPD 1 2
Myocardial infarction 0 1
Pancreatitis 1 1
Recent surgery (%) 3 (25) 3 (25)
APACHE II score 21 ± 6 22 ± 6
Mechanical ventilation 12 11
Shock 12 12
Vasopressor use 11 11
Renal replacement therapy 4 4
Number of dysfunctional organ systems 4 ± 1 3 ± 1
SOFA score at inclusion 10 ± 3 9 ± 2
Site of infection, n
Lung 8 8
Urinary tract 2 0
Other 2 4
Causes of infection, n
Gram positive 6 4
Gram negative 7 5
Anaerobes 0 1
Fungi 1 2
Data represent mean ± SD. APACHE, Acute Physiology and Chronic Health Evaluation; COPD, chronic obstructive pulmonary disease; SOFA, sepsis-related organ failure assessment.
Table 2 Outcome and adverse events
Characteristic rhAPC group (n = 12) Control group (n = 12) p value
28-day survivors, n 10 9 1.00
Length of ICU stay, days 15 ± 14 20 ± 15 0.41
Length of hospital stay, days, 35 ± 20 46 ± 41 0.38
Patients discharged home (%) 5 (41.7) 5 (41.7) 1.00
At least one adverse event, n 9 7 0.41
Serious bleeding event, n 1 0 0.33
Data represent mean ± SD.
==== Refs
Bernard GR Vincent JL Laterre PF LaRosa SP Dhainaut JF Lopez-Rodriguez A Steingrub JS Garber GE Helterbrand JD Ely EW Fisher CJ Jrfor the PROWESS study group Efficacy and safety of recombinant human activated protein C for severe sepsis N Engl J Med 2001 344 699 709 11236773 10.1056/NEJM200103083441001
Brass LF Thrombin and platelet activation Chest 2003 124 3 Suppl 18S 25S 12970120 10.1378/chest.124.3_suppl.18S
Naldini A Pucci A Carney DH Fanetti G Carraro F Thrombin enhancement of interleukin-1 expression in mononuclear cells: involvement of proteinase-activated receptor-1 Cytokine 2002 20 191 199 12550103 10.1006/cyto.2002.2001
Morris PE Hite RD Ohl C Relationship between inflammation and coagulation pathway in patients with severe sepsis: implications for therapy with actived protein C BioDrugs 2002 16 403 417 12463764
Esmon CT Protein C anticoagulant pathway and its role in controlling microvascular thrombosis and inflammation Crit Care Med 2001 29 7 Suppl S48 S51 11445734 10.1097/00003246-200107001-00018
Riewald M Petrovan RJ Donner A Ruf W Activated protein C signals through the thrombin receptor PAR1 in endothelial cells J Endotoxin Res 2003 9 317 321 14577849 10.1179/096805103225002584
Hoffmann JN Vollmar B Laschke MW Inthorn D Fertmann J Schildberg FW Menger MD Microhemodynamic and cellular mechanisms of activated protein C action during endotoxemia Crit Care Med 2004 32 1011 1017 15071394 10.1097/01.CCM.0000120058.88975.42
Elisen MG Maseland MH Church FC Bouma BN Meijers JC Role of the A+ helix in heparin binding to protein C inhibitor Thromb Haemost 1996 75 760 766 8725720
Mosnier LO von dem Borne PAK Meijers JCM Bouma BN Plasma TAFI levels influence the clot lysis time in healthy individuals in the presence of an intact intrinsic pathway of coagulation Thromb Haemost 1998 80 829 835 9843179
Esmon C The protein C pathway Crit Care Med 2000 28 9 Suppl S44 S48 11007197 10.1097/00003246-200009001-00010
Levi M Choi G Schoots I Schultz MJ van der Poll T Beyond sepsis: Activated protein C and ischemia-reperfusion injury Crit Care Med 2004 32 5 Suppl S309 S312 15118536 10.1097/01.CCM.0000126362.38567.52
Dhainaut JF Yan SB Margolis BD Lorente JA Russell JA Freebairn RC Spapen HD Riess H Basson B Johnson G 3rdKinasewitz GT for the PROWESS sepsis study group Drotrecogin alfa (activated) (recombinant human activated protein C) reduces host coagulopathy response in patients with severe sepsis Thromb Haemost 2003 90 642 653 14515185
Boldt J Papsdorf M Rothe A Kumle B Piper S Changes of the hemostatic network in critically ill patients – is there a difference between sepsis, trauma, and neurosurgery patients? Crit Care Med 2000 28 445 450 10708181 10.1097/00003246-200002000-00026
Dhainaut JF Yan SB Cariou A Mira JP Soluble thrombomodulin, plasma-derived unactivated protein C, and recombinant human activated protein C in sepsis Crit Care Med 2002 30 5 Suppl S318 S324 12004254 10.1097/00003246-200205001-00023
Esmon CT Crosstalk between inflammation and thrombosis Maturitas 2004 47 305 314 15063484 10.1016/j.maturitas.2003.10.015
Taoka Y Okajima K Uchiba M Murakami K Harada N Johno M Naruo M Activated protein C reduces the severity of compression-induced spinal cord injury in rats by inhibiting activation of leukocytes J Neurosci 1998 18 1393 1398 9454848
Yuksel M Okajima K Uchiba M Horiuchi S Okabe H Activated protein C inhibits lipopolysaccharide-induced tumor necrosis factor-a production by inhibiting activation of both nuclear factor-κB and activator protein-1 in human monocytes Thromb Haemost 2002 88 267 273 12195699
|
16277710
|
PMC1297612
|
CC BY
|
2021-01-04 16:04:56
|
no
|
Crit Care. 2005 Jul 21; 9(5):R490-R497
|
utf-8
|
Crit Care
| 2,005 |
10.1186/cc3774
|
oa_comm
|
==== Front
Crit CareCritical Care1364-85351466-609XBioMed Central London cc37781627771110.1186/cc3778ResearchSteroid use in PROWESS severe sepsis patients treated with drotrecogin alfa (activated) Levy Howard [email protected] Pierre-Francois [email protected] Becky [email protected] Rebecca L [email protected] Medical Director, Acute Care, Eli Lilly and Co., Indianapolis, Indiana, USA2 Professor in Medicine, Head of Intensive Care Medicine, Critical Care and Emergency Department, Cliniques Universitaires St Luc, UCL, Brussels, Belgium3 Associate Senior Statistician, Eli Lilly and Co., Indianapolis, Indiana, USA4 Senior Scientific Communications Associate, Eli Lilly and Co., Indianapolis, Indiana, USA2005 27 7 2005 9 5 R502 R507 2 5 2005 25 5 2005 9 6 2005 1 7 2005 Copyright © 2005 Levy 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
In a study conducted by Annane, patients with septic shock and unresponsive to adrenocorticotropic hormone stimulation receiving low-dose steroid therapy had prolonged survival but not significantly improved 28-day mortality. The present study examines intravenous steroid use in PROWESS (Recombinant Human Activated Protein C Worldwide Evaluation in Severe Sepsis) patients meeting the Annane enrollment criteria (AEC).
Methods
Adrenocorticotropic hormone stimulation tests were not done in PROWESS. Steroids were allowed but their use was not directed. Patients were identified using AEC (all of: randomization to study drug treatment within 8 hours of shock onset; infection, fever, or hypothermia; tachycardia; systolic blood pressure <90 mmHg on vasopressors; mechanical ventilation; and one of urine <0.5 ml/kg per hour, lactic acidosis, or arterial oxygen tension/inspired fractional oxygen <280). We examined steroid use and mortality data; additional analyses were done outside the 8-hour window.
Results
Steroid-treated patients were older, had higher Acute Physiology and Chronic Health Evaluation scores and more organ dysfunctions, and were more commonly receiving mechanical ventilation. Among patients meeting AEC, regardless of steroid treatment (n = 97), mortality in the placebo and drotrecogin alfa (activated) groups was 38% (19/50) and 28% (13/47), respectively (relative risk [RR] = 0.73, 95% confidence interval [CI] 0.41–1.30). When using AEC but excluding the requirement for randomization within 8 hours of shock onset (n = 612), placebo mortality was 38% (118/313) and drotrecogin alfa (activated) mortality was 29% (88/299; RR = 0.78, 95% CI 0.62–0.98). Using AEC but excluding the 8-hour window and with steroids initiated at baseline and/or infusion (n = 228) resulted in mortality for placebo and drotrecogin alfa (activated) groups of 43% (51/118) and 33% (36/110), respectively (RR = 0.76, 95% CI 0.54–1.06).
Conclusion
Patients with severe sepsis from the PROWESS trial who were likely to respond to low-dose steroids according to the AEC were those patients at a high risk for death. However, when using the AEC, regardless of steroid use, patients exhibited a survival benefit from treatment with drotrecogin alfa (activated).
==== Body
Introduction
Corticosteroid therapy in sepsis and septic shock has been investigated for more than 50 years [1]. Over this period there have been dozens of trials examining various patient populations, assessing different corticosteroids in a wide range of dosing regimens, and employing methodologies that are diverse in form and quality [1-15]. Results have varied widely, with some studies favoring the control group and some favoring the treatment group (low-dose use); others have shown virtually no difference in outcome, and still other studies (particularly those examining high-dose steroids) have indicated that steroid therapy is harmful [3,4,9,16]. Recently, a small study of patients with community-acquired pneumonia [17] showed a positive effect of steroid treatment. However, findings from several investigators suggest that steroid treatment should be limited to patients who have adrenal insufficiency [18-21].
The hypothalamic–pituitary–adrenal axis plays an important role in the body's ability to respond to stress. Patients who develop septic shock and who are consequently maximally stressed, in response to an infection, may exhibit adrenal insufficiency. Insufficiency of the adrenal system correlates with increased risk for mortality associated with severe sepsis and/or septic shock [22]. Adrenal replacement therapy in patients with adrenal failure may be a logical addition to standard care in patients with severe sepsis and vasopressor dependent shock. Annane and colleagues [22] demonstrated that the response to a short corticotropin test could potentially be used to identify patients with relative adrenal insufficiency who are at high risk for death related to septic shock. In another recent, randomized trial of 300 patients with septic shock, Annane and coworkers [15] found that 229 patients (about two-thirds) had adrenal insufficiency, as determined using the 250 μg corticotropin stimulation test. In this subgroup mortality at 28 days was not significantly less among those who received corticosteroids (53%) than in the placebo group (63%; P = 0.10). Patients who had adrenal insufficiency appeared to have prolonged median survival (16.5 days for corticosteroid treatment versus 14 days for placebo), but these values and the difference in 28-day mortality between treatment group and placebo were not significant.
A recent review and meta-analysis [18] assessing the effects of corticosteroids on mortality in patients with severe sepsis and septic shock found that, for all published trials, use of corticosteroids did not significantly affect mortality overall. The studies by Annane and coworkers did show that corticosteroid treatment might reduce mortality in a subgroup of septic shock patients with well defined adrenal insufficiency. However, even with corticosteroid plus fludrocortisone treatment, more than half of that subgroup of patients died, clearly indicating the importance of additional therapies to reduce mortality not only in this subgroup but also in the overall population of septic shock patients, with or without adrenal insufficiency.
There are currently no published data on the use of drotrecogin alfa (activated) with corticosteroids in the treatment of severe sepsis. The Recombinant Human Activated Protein C Worldwide Evaluation in Severe Sepsis (PROWESS) trial [23] was a phase III study designed to evaluate drotrecogin alfa (activated) for the treatment of patients with severe sepsis at high risk for death (e.g. as determined by an Acute Physiology and Chronic Health Evaluation II score ≥25 and/or two or more organ dysfunctions). The present study examines steroid use in PROWESS patients with severe sepsis and septic shock.
Materials and methods
In the PROWESS trial, severe sepsis patients were randomly assigned to receive either drotrecogin alfa (activated) at a dose of 24 μg/kg per hour, or placebo, administered intravenously for 96 hours. Concomitant use of steroids was allowed but was not required or specified by the protocol in PROWESS. The duration and route but not dose of steroids was recorded. For the present analysis, patients were identified using all of the Annane enrollment criteria (AEC): randomization of treatment with drotrecogin alfa (activated) or placebo within 8 hours of onset of shock; infection, fever, or hypothermia; tachycardia; systolic blood pressure <90 mmHg on vasopressors; mechanical ventilation; and one of urine output <0.5 ml/kg per hour, lactic acidosis, or arterial oxygen tension/fractional inspired oxygen <280. In the PROWESS trial patients were classified as being in septic shock at baseline if they met any of the following criteria for at least 1 hour despite adequate fluid resuscitation or having documented adequate intravascular volume status, at any time within the 6 hours before the start of infusion of drotrecogin alfa (activated) or placebo: arterial systolic blood pressure ≤90 mmHg; mean arterial pressure ≤70 mmHg; or need for vasopressors (defined as dopamine ≥5 μg/kg per min or noradrenaline [norepinephrine], adrenaline [epinephrine], or phenylephrine at any dose) to maintain systolic blood pressure ≥90 mmHg or mean arterial pressure ≥70 mmHg.
We also analyzed data from PROWESS patients selected using the AEC but without the criterion of drotrecogin alfa (activated) or placebo treatment initiation within 8 hours of the onset of septic shock. The adrenocorticotropic hormone stimulation test was not done in PROWESS, and so subgroups related to adrenal insufficiency could not be evaluated.
The characteristics of patients receiving steroids at baseline or during infusion were compared with those of patients who did not receive steroids at baseline or during infusion. Continuous baseline characteristics (e.g. age) were analyzed using one-way analysis of variance. Categorical baseline characteristics were analyzed using Pearson's χ2 test.
Pearson's χ2 tests were used for all 28-day mortality subgroup analyses, which compared drotrecogin alfa (activated) treated patients with placebo patients. The logit methodology was used to calculate relative risks and associated 95% confidence intervals.
Results
Baseline characteristics (e.g. age, disease severity, etc.) were not different between placebo and drotrecogin alfa (activated) treated patients in the PROWESS trial [23]. The distribution of patients from PROWESS according to AEC is shown in Fig. 1. Of the 1690 PROWESS patients, 36.2% met the AEC without the 8-hour time restriction (i.e. randomization to study drug treatment within 8 hours of shock onset) and 5.7% met the AEC with the 8-hour time restriction. These two groups were then further subdivided into patients receiving steroids and those not receiving steroids. Patients receiving steroid treatment at either severe sepsis onset or during drotrecogin alfa (activated) infusion were classified as treated with steroids.
Table 1 lists the baseline disease severity measures for PROWESS patients treated or not treated with steroids (PROWESS overall; AEC not considered). Patients treated with steroids were older, and had higher mean Acute Physiology and Chronic Health Evaluation II scores and more organ dysfunctions than did patients not receiving steroids. Patients were also more likely to receive ventilator support in the steroid treatment group at baseline. PROWESS 28-day all-cause mortality by steroid exposure either at severe sepsis onset or drotrecogin alfa (activated) infusion is shown in Fig. 2.
Figure 3 illustrates the 28-day all-cause mortality for PROWESS patients meeting the AEC, either including or excluding the 8-hour time window (study treatment within 8 hours of onset of shock). A survival benefit was observed for drotrecogin alfa (activated)-treated patients regardless of whether they were treated with steroids at baseline or during infusion, or whether they met the AEC with or without the 8-hour time criteria.
Discussion
The use of steroid therapy in the treatment of sepsis and septic shock has been a controversial issue for many decades. Recent data [15] indicate that physiologic doses of hydrocortisone and fludrocortisone used in combination can reduce the risk for death in patients with relative adrenal insufficiency and septic shock. However, the patient population in that study remained at a high risk for death, as indicated by a 28-day mortality rate of 53% in the treatment group in which steroids were most effective. Guidelines from the Surviving Sepsis Campaign [24] suggested that stress dose steroid therapy should be used for septic shock; however, they further stated that there are no documented studies showing that stress doses of steroids improve the outcome of sepsis without shock unless a patient's history indicates steroid use or adrenal dysfunction. In the recent meta-analysis conducted by Annane and coworkers [18] it was concluded that steroids should be given to patients only when absolute or relative adrenal insufficiency is present. However, the definition for adrenal insufficiency has varied in the few trials in which it was used to evaluate patients for steroid treatment [12,15]. A further area of controversy is whether serum cortisol levels should be measured as total or free cortisol. It was recently reported that severe hypoproteinemia frequently results in concentrations of serum total cortisol in critically ill patients that are lower than expected, whereas free cortisol levels give a more accurate indication of response to corticotropin stimulation [25] and thus provide better identification of patients with adrenal insufficiency.
The PROWESS trial was a phase III placebo-controlled study that evaluated drotrecogin alfa (activated) for the treatment of patients with severe sepsis [23]. In that study drotrecogin alfa (activated) treatment was associated with a significant absolute reduction in mortality rate of 6.1% (relative risk reduction 19.4%; P = 0.005), and of 12.8% (relative risk reduction 29.2%; P = 0.0002) in the subpopulation of patients who were at high risk for death, which led to its approval by the US Food and Drug Administration.
This is the first report on the use of drotrecogin alfa (activated) with corticosteroids in the treatment of severe sepsis. An analysis of the PROWESS data indicates that 36.2% of the 1690 PROWESS patients met the AEC without the 8-hour time restriction and 5.7% met the criteria with the 8-hour time restriction for enrollment in the study by Annane and coworkers [15]. Limitations of our study include the fact that we did not know the dose or particular type of corticosteroid drug administered and that we did not know the responsiveness of patients to the adrenocorticotropic hormone test.
When examining data from PROWESS, mortality among placebo patients does not differ regardless of whether steroid was given at baseline or during infusion, or whether one applies the 8-hour time restriction or not. Where no steroid was given, the mortality in the two groups still does not differ, suggesting that the timing of steroid treatment alone does not affect mortality. These data further demonstrate an absence of effect of steroid treatment on the potential benefit from drotrecogin alfa (activated) treatment.
The mortality rate from severe sepsis in the PROWESS trial was substantially lower than that previously reported by Annane and coworkers [15,18]. However, the PROWESS trial employed different exclusion criteria than did Annane and coworkers; in particular, the PROWESS trial excluded moribund patients and patients not expected to survive 28 days because of an underlying medical disease.
Drotrecogin alfa (activated) reduced mortality in PROWESS patients with severe sepsis at high risk for death [23]. Patients at high risk for death were more likely to be treated with steroids. In the PROWESS trial the use of steroids did not significantly affect the treatment benefit from drotrecogin alfa (activated).
Conclusion
Drotrecogin alfa (activated) reduces mortality in patients with severe sepsis at high risk for death, as indicated by meeting the AEC for steroid use. Therefore, we conclude that severe sepsis patients with vasopressor dependent shock should be evaluated for drotrecogin alfa (activated) therapy, particularly if steroids are considered. This is because, regardless of steroid use, these patients have demonstrated survival benefit from treatment with drotrecogin alfa (activated).
Key messages
• Meeting criteria for steroid use by Annane study entry criteria identifies a patient at high risk of death.
• Patients receiving steroids in PROWESS were older, had higher APACHE II scores, more organ dysfunctions, and were more commonly receiving mechanical ventilation than those who did not receive steroids.
• Drotrecogin alfa (activated) provides a survival benefit to these high-risk patients regardless of steroid use.
Abbreviations
AEC = Annane enrollment criteria.
Competing interests
Howard Levy, Becky Bates, and Rebecca L Qualy are employees and shareholders of Eli Lilly and Company. Pierre-Francois Laterre was an investigator in the PROWESS trial, and is a paid consultant and speaker for Eli Lilly and Company.
Authors' contributions
All the authors contributed to the composition, revision and review of the manuscript, and have read and approved the final version. In addition, HL conceived the idea for this report, BB performed the statistical analysis, RLQ drafted the document and P-FL participated in obtaining the original PROWESS data.
Acknowledgements
This study was presented in part at the European Society of Intensive Care Medicine 16th Annual Congress, held on 7 October 2003 in Amsterdam, The Netherlands.
Figures and Tables
Figure 1 Patient population. AEC = Annane enrollment criteria.
Figure 2 28-Day all-cause mortality for PROWESS patients. Infusion refers to infusion period plus 1 calendar day after termination of infusion. DrotAA, drotrecogin alfa (activated); PROWESS, Recombinant Human Activated Protein C Worldwide Evaluation in Severe Sepsis.
Figure 3 28-Day all-cause mortality for PROWESS Patients meeting the Annane enrollment criteria. Steroids received at baseline and/or during the infusion period. AEC = Annane enrollment criteria; DrotAA, drotrecogin alfa (activated); PROWESS, Recombinant Human Activated Protein C Worldwide Evaluation in Severe Sepsis.
Table 1 Baseline disease severity measures for patients treated versus not treated with steroids
Characteristic/parameter Steroidsa No steroids P value
Age (years; mean ± SD [n])
Overall 62 ± 15.9 (586) 60 ± 17.3 (1104) 0.024b
Drotrecogin alfa (activated) 62 ± 16.1 (291) 60 ± 17.7 (559) 0.126b
Placebo 62 ± 15.7 (295) 60 ± 16.8 (545) 0.098b
APACHE II (mean ± SD [n])
Overall 26.4 ± 7.6 (586) 23.9 ± 7.6 (1104) <0.001b
Drotrecogin alfa (activated) 26.6 ± 7.7 (291) 23.5 ± 7.4 (559) <0.001b
Placebo 26.2 ± 7.5 (295) 24.3 ± 7.9 (545) <0.001b
Number of organ dysfunctions (mean ± SD [n])
Overall 2.5 ± 1.1 (586) 2.3 ± 1.1 (1104) 0.018b
Drotrecogin alfa (activated) 2.5 ± 1.1 (291) 2.3 ± 1.1 (559) 0.020b
Placebo 2.5 ± 1.1 (295) 2.4 ± 1.1 (545) 0.301b
Baseline shockd (% [n])
Overall 74% (432) 70% (768) 0.073c
Drotrecogin alfa (activated) 75% (217) 68% (381) 0.521c
Placebo 73% (215) 71% (387) 0.565c
Baseline vasopressor (% [n])
Overall 64% (374) 62% (683) 0.429c
Drotrecogin alfa (activated) 62% (179) 60% (337) 0.728c
Placebo 66% (195) 63% (346) 0.450c
Baseline ventilator (% [n])
Overall 80% (471) 73% (804) <0.001c
Drotrecogin alfa (activated) 79% (230) 70% (393) 0.006c
Placebo 82% (241) 75% (411) 0.037c
aPatients receiving steroids at baseline or infusion were classified as receiving steroids. bBy analysis of variance. cBy Pearson's χ2 test. dBaseline shock was shock at any time within 6 hours prior to drotrecogin alfa (activated) or placebo infusion. APACHE, Acute Physiology and Chronic Health Evaluation.
==== Refs
Hahn EO Houser HB Rammelkamp CH JrDenny FW Wannamaker LW Effect of cortisone on acute streptococcal infections and post streptococcal complications J Clin Invest 1951 30 274 281 14824277
Schumer W Steroids in the treatment of clinical septic shock Ann Surg 1976 184 333 341 786190
Cronin L Cook DJ Carlet J Heyland DK King D Lansang MA Fisher CJ Jr Corticosteroid treatment for sepsis: a critical appraisal and meta-analysis of the literature Crit Care Med 1995 23 1430 1439 7634816 10.1097/00003246-199508000-00019
Lefering R Neugebauer EA Steroid controversy in sepsis and septic shock: a meta-analysis Crit Care Med 1995 23 1294 1303 7600840 10.1097/00003246-199507000-00021
Bennett IL JnrFinland M Hamburger M Kass EH Lepper M Waisbren BA The effectiveness of hydrocortisone in the management of severe infections JAMA 1963 183 462 465
Klastersky J Cappel R Debusscher L Effectiveness of betamethasone in management of severe infections. A double-blind study N Engl J Med 1971 284 1248 1250 4929896
Sprung CL Caralis PV Marcial EH Pierce M Gelbard MA Long WM Duncan RC Tendler MD Karpf M The effects of high-dose corticosteroids in patients with septic shock. A prospective, controlled study N Engl J Med 1984 311 1137 1143 6384785
The Veterans Administration Systemic Sepsis Cooperative Study Group Effect of high-dose glucocorticoid therapy on mortality in patients with clinical signs of systemic sepsis N Engl J Med 1987 317 659 665 2888017
Bone RG Fisher CJ JrClemmer TP Slotman GJ Metz CA Balk RA A controlled clinical trial of high-dose methylprednisolone in the treatment of severe sepsis and septic shock N Engl J Med 1987 317 653 658 3306374
Luce JM Montgomery AB Marks JD Turner J Metz CA Murray JF Ineffectiveness of high-dose methylprednisolone in preventing parenchymal lung injury and improving mortality in patients with septic shock Am Rev Respir Dis 1988 138 62 68 3202402
Slusher T Gbadero D Howard C Lewison L Giroir B Toro L Levin D Holt E McCracken GH Jr Randomized, placebo-controlled, double blinded trial of dexamethasone in African children with sepsis Pediatr Infect Dis J 1996 15 579 583 8823850 10.1097/00006454-199607000-00004
Bollaert PE Charpentier C Levy B Debouverie M Audibert G Larcan A Reversal of late septic shock with supraphysiologic doses of hydrocortisone Crit Care Med 1998 26 645 650 9559600 10.1097/00003246-199804000-00010
Chawla K Kupfer Y Goldman I Tessler S Hydrocortisone reverses refractory septic shock Crit Care Med 1999 27 33A 10.1097/00003246-199901001-00022
Briegel J Forst H Haller M Schelling G Kilger E Kuprat G Hemmer B Hummel T Lenhart A Heyduck M Stress doses of hydrocortisone reverse hyperdynamic septic shock: a prospective, randomized, double-blind, single-center study Crit Care Med 1999 27 723 732 10321661 10.1097/00003246-199904000-00025
Annane D Sebille V Charpentier C Bollaert PE Francois B Korach JM Capellier G Cohen Y Azoulay E Troche G Effect of treatment with low doses of hydrocortisone and fludrocortisone on mortality in patients with septic shock JAMA 2002 288 862 871 12186604 10.1001/jama.288.7.862
Minneci PC Deans KJ Banks SM Eichacker PQ Natanson C Meta-analysis: the effect of steroids on survival and shock during sepsis depends on the dose Ann Intern Med 2004 141 47 56 15238370
Confalonieri M Urbino R Potena A Piattella Parigi P Puccio G Della Porta R Giorgio C Blasi F Umberger R Meduri GU Hydrocortisone infusion for severe community-acquired pneumonia: a preliminary randomized study Am J Respir Crit Care Med 2005 171 242 248 15557131 10.1164/rccm.200406-808OC
Annane D Bellissant E Bollaert PE Briegel J Keh D Kupfer Y Corticosteroids for severe sepsis and septic shock: a systematic review and meta-analysis BMJ 2004 329 480 15289273 10.1136/bmj.38181.482222.55
Keh D Sprung CL Use of corticosteroid therapy in patients with sepsis and septic shock: an evidence-based review Crit Care Med 2004 32 11 Suppl S527 S533 15542960 10.1097/01.CCM.0000142983.15421.11
Burry LD Wax RS Role of corticosteroids in septic shock Ann Pharmacother 2004 38 464 472 14970369 10.1345/aph.1C435
Cooper MS Stewart PM Corticosteroid insufficiency in acutely ill patients N Engl J Med 2003 348 727 734 12594318 10.1056/NEJMra020529
Annane D Sebille V Troche G Raphael JC Gajdos P Bellissant E A 3-level prognostic classification in septic shock based on cortisol levels and cortisol response to corticotropin JAMA 2000 283 1038 1045 10697064 10.1001/jama.283.8.1038
Bernard GR Vincent JL Laterre PF LaRosa SP Dhainaut JF Lopez-Rodriguez A Steingrub JS Garber GE Helterbrand JD Ely EW Efficacy and safety of recombinant human activated protein C for severe sepsis N Engl J Med 2001 344 699 709 11236773 10.1056/NEJM200103083441001
Dellinger RP Carlet JM Masur H Gerlach H Calandra T Cohen J Gea-Banacloche J Keh D Marshall JC Parker MM Surviving Sepsis Campaign guidelines for management of severe sepsis and septic shock Crit Care Med 2004 32 858 873 15090974 10.1097/01.CCM.0000117317.18092.E4
Hamrahian AH Oseni TS Arafah BM Measurements of serum free cortisol in critically ill patients N Engl J Med 2004 350 1629 1638 15084695 10.1056/NEJMoa020266
|
16277711
|
PMC1297614
|
CC BY
|
2021-01-04 16:04:56
|
no
|
Crit Care. 2005 Jul 27; 9(5):R502-R507
|
utf-8
|
Crit Care
| 2,005 |
10.1186/cc3778
|
oa_comm
|
==== Front
Crit CareCritical Care1364-85351466-609XBioMed Central London cc37861627771210.1186/cc3786ResearchRecombinant human erythropoietin therapy in critically ill patients: a dose-response study [ISRCTN48523317] Georgopoulos Dimitris [email protected] Dimitris 2Routsi Christina 3Michalopoulos Argiris 4Maggina Nina 5Dimopoulos George 6Zakynthinos Epaminondas 7Nakos George 8Thomopoulos George 9Mandragos Kostas 10Maniatis Alice 11the Critical Care Clinical Trials Greek Group1 Professor of Medicine & ICU Director, Department of Intensive Care Medicine, University Hospital of Heraklion, University of Crete, Heraklion, Crete, Greece2 ICU Director, Intensive Care Unit, Papageorgiou Hospital of Thessaloniki, Thessaloniki, Greece3 Assistant Professor of Medicine, Department of Intensive Care, Evangelismos Hospital, University of Athens, Athens, Greece4 ICU Director, Intensive Care Unit, Henry Dunan Hospital, Athens5 ICU Director, Intensive Care Unit, Saint Olga Hospital, Athens, Greece6 Intensive Care Unit, Sotiria Hospital, Athens, Greece7 Assistant Professor of Medicine & ICU Director, Intensive Care Unit, University Hospital of Larisa, University of Larisa, Larisa, Greece8 Associate Professor of Medicine, Intensive Care Unit, University Hospital of Ioannina, University of Ioannina, Ioannina, Greece9 ICU Director, Intensive Care Unit, Laiko Hospital, Athens10 ICU Director, Hellenic Red Cross Hospital, Athens, Greece11 Professor of Medicine, University Hospital of Patras, Patras, Greece2005 5 8 2005 9 5 R508 R515 28 3 2005 5 5 2005 6 6 2005 5 7 2005 Copyright © 2005 Georgopoulos 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 aim of this study was to assess the efficacy of two dosing schedules of recombinant human erythropoietin (rHuEPO) in increasing haematocrit (Hct) and haemoglobin (Hb) and reducing exposure to allogeneic red blood cell (RBC) transfusion in critically ill patients.
Method
This was a prospective, randomized, multicentre trial. A total of 13 intensive care units participated, and a total of 148 patients who met eligibility criteria were enrolled. Patients were randomly assigned to receive intravenous iron saccharate alone (control group), intravenous iron saccharate and subcutaneous rHuEPO 40,000 units once per week (group A), or intravenous iron saccharate and subcutaneous rHuEPO 40,000 units three times per week (group B). rHuEPO was given for a minimum of 2 weeks or until discharge from the intensive care unit or death. The maximum duration of therapy was 3 weeks.
Results
The cumulative number of RBC units transfused, the average numbers of RBC units transfused per patient and per transfused patient, the average volume of RBCs transfused per day, and the percentage of transfused patients were significantly higher in the control group than in groups A and B. No significant difference was observed between group A and B. The mean increases in Hct and Hb from baseline to final measurement were significantly greater in group B than in the control group. The mean increase in Hct was significantly greater in group B than in group A. The mean increase in Hct in group A was significantly greater than that in control individuals, whereas the mean increase in Hb did not differ significantly between the control group and group A.
Conclusion
Administration of rHuEPO to critically ill patients significantly reduced the need for RBC transfusion. The magnitude of the reduction did not differ between the two dosing schedules, although there was a dose response for Hct and Hb to rHuEPO in these patients.
==== Body
Introduction
Anaemia is a common problem in critically ill patients [1,2]. Indeed, it has been shown that, at intensive care unit (ICU) admission, mean haemoglobin (Hb) concentration in critically ill patients is about 11 g/dl, and in 60% and 30% of them the mean Hb is less than 12 and 10 g/dl, respectively. Thus, the majority of critically ill patients exhibit anaemia upon ICU admission, which persists throughout the duration of their ICU stay. Overt or occult blood loss and decreased production of red blood cells (RBCs) due to blunted erythropoietic response are the two main causes of anaemia in these patients [3].
Anaemia in critically ill patients results in significant RBC transfusions. Approximately 40% of critically ill patients receive at least one unit of RBCs, relatively soon after ICU admission [1,2]. It is of interest that the mean number of RBC units transfused approaches five and the pretransfusion Hb is about 8.5 g/dl, indicating that the large number of transfusions is not due to a high transfusion threshold for Hb [1,2].
It has been recognized that RBC transfusion is not without risks. The adverse effects of RBC transfusions are numerous, including transmission of infection [4], transfusion associated immunosuppression [5-9], transfusion related acute lung injury [10], disturbances in microcirculation due to blood storage [11,12] and allergic reactions [9]. Large observational studies in critically ill patients have shown that RBC transfusion is an independent risk factor for increased mortality [1,2]. Although the mechanism through which RBC transfusion may increase mortality is currently unknown, studies have shown that RBC transfusion in critically ill patients is associated with a higher incidence of infection and evidence of tissue hypoxia [11,13,14]. These data indicate that the likely contributing factors to mortality are related to immunosuppression and disturbances in microcirculation, as opposed to allergic reaction or transmission of infection.
Because of the risks associated with blood transfusion, alternative treatments and preventative strategies for anaemia in critically ill patients have been explored. Among them, exogenous administration of recombinant human erythropoietin (rHuEPO) demonstrated promising results [15-17]. The rationale underlying therapy with rHuEPO therapy in critically ill patients is that increased erythropoiesis will result in higher Hb levels and subsequently reduce the need for RBC transfusion [18]. It has been shown that exogenous rHuEPO in critically ill patients raised reticulocyte counts and Hb, and reduced considerably requirements for RBC transfusion [15-17].
The two randomized studies that showed that rHuEPO is efficacious in increasing Hb level and reducing allogeneic RBC transfusion used two different therapeutic regimens [15,16]. One study [16] used 300 units/kg rHuEPO for 5 consecutive days and then every other day to achieve a haematocrit (Hct) concentration above 38%, whereas in the other [15] the drug was administered weekly in a dose of 40,000 units. Thus, the optimal dose of rHuEPO in critically ill patients is not known, which is an issue of financial importance, given the cost of this therapy.
The aim of the present study was to assess the efficacy of two dosing schedules of rHuEPO (40,000 units once and thrice per week, respectively) in increasing Hct and Hb and in reducing exposure to allogeneic RBC transfusion in critically ill patients. These dosing regimens are comparable to those used by the two randomized studies in critically ill patients [15,16].
Materials and methods
This study was a prospective, randomized, multicentre trial conducted at 13 Greek ICUs between November 2000 and December 2003. Approval of the study was given by the institutional review committee at each participation centre and written informed consent was obtained from each patient or next of kin. Patient enrollment was done at each site and supervised by the data coordinating centre. Randomization and data analysis were done by the data coordinating centre. A stratified random sampling scheme was employed as the selection method for randomization. Acute Physiology and Chronic Health Evaluation II score and age decades were considered as distinct strata. To ensure equal allocation of individuals from each stratum (epsem scheme), the sampling fraction was considered. The sample size was calculated in order to detect a 10% difference in the Hct values between groups receiving rHuEPO at a 5% significance level and 90% power, assuming that the mean Hct for the group receiving the lowest dose would be in the range of 35% and the variance equal to 30.
All patients admitted to the ICU in each of the 13 participating centres were evaluated for study eligibility. Inclusion criteria were as follows: age at least 18 years; Hb less than 12 g/dl; no iron deficiency (defined as transferrin saturation <10% and ferritin <50 ng/ml); negative pregnancy test (for females of reproductive age); an expected ICU stay of at least 7 days; and provision of signed informed consent. The expected duration of the ICU stay was judged on clinical grounds and Acute Physiology and Chronic Health Evaluation II score by the ICU team at admittance to the unit. Exclusion criteria included chronic renal failure requiring dialysis, new onset (<6 months) seizures, life expectancy under 7 days, previous use of rHuEPO (within 3 months), recent use of cytostatics or recent radiotherapy (within 1 month) and participation in another research protocol.
The patients were randomly assigned (day 0) to receive intravenous iron saccharate alone (control group), intravenous iron saccharate and subcutaneous rHuEPO 40,000 units once per week (group A), and intravenous iron saccharate and subcutaneous rHuEPO 40,000 units three times per week (group B). In all groups iron was given at a dose of 100 mg three times per week. rHuEPO was provided by the sponsor of the trial.
rHuEPO was given for a minimum of 2 weeks or until ICU discharge or death. The maximum duration of therapy was 3 weeks. rHuEPO was temporarily withheld when Hb exceeded 12 g/dl and was resumed if Hb again fell to below 12 g/dl. rHuEPO was given intravenously if the platelet count was below 20,000 μL.
Transfusion of RBCs was standardized at a Hb of 7 g/dl and, in cases of active cardiac ischaemia and central nervous system damage, at 9 g/dl [19]. In patients with active blood loss, defined as evidence of ongoing blood loss accompanied by a decrease in the Hb concentration of 3.0 g/dl in the preceding 12 hours or a requirement for at least 3 units of packed RBCs during the same period, the need for blood transfusion was determined by the patient's attending physician. The physicians caring for the patients were instructed to administer RBC transfusions, one unit at a time, and to measure the patient's Hb concentration after each unit was transfused.
The primary outcome end-points were differences in Hct and Hb between groups and transfusion independence between study days 1 and 28. Additional data recorded included ICU length of stay and cumulative mortality through to day 28. Adverse effects were assessed daily. Nosocomial infections were diagnosed using standard criteria [20,21].
All patients were followed up for a total of 28 days from the day of randomization, unless death occurred earlier. Patients discharged from the hospital before study day 28 had final laboratory data obtained within 5 days of study day 28. Patients were followed up for 28 days, unless death occurred earlier. Patients who were discharged from the hospital before study day 28 and were not available to provide the final laboratory data (i.e. data were not available within 5 days of study day 28) were considered lost to follow up. Analysis of outcomes was on an intent-to-treat basis.
All categorical variables were summarized by frequency distribution tables and analyzed by χ2 tests. Descriptive results for continuous measurements were presented as mean ± standard deviation unless otherwise stated. The methods used for analysis were analysis of variance F tests, Scheffe tests for multiple comparisons, Kruskal–Wallis and Mann–Whitney tests where appropriate. The transfusion rate was analyzed using a zero-inflated Poisson model. All computations were done using Sigmastat-plus (SPSS, INC, Chicago, Ill). All statistical tests were two-sided, and the level of statistical significance was set at 5%.
Patients who did not receive a transfusion at the time of study withdrawal, who died, or who were lost to follow up after hospital discharge were considered nontransfused for the analysis. The number of RBC units transfused, transfusion rate, average days transfused and units per transfused patient were analysed using the Mann–Whitney test. Transfusion rate, expressed as the number of RBC units transfused per day during the study, was determined by dividing the number of transfused units for each group by the total number of days alive for the patients in the group. Average days transfused was determined by dividing the number of transfusion days for each group by the total number of days alive for the patients in the group. The average number of units transfused was determined by dividing the number of transfusions for each group by the total number of patients in the group. Units per transfused patient were determined by dividing the number of transfusions for each group by the total number of patients transfused in the group.
Results
A total of 148 patients were enrolled in the study (Fig. 1). Forty-eight patients were randomly assigned to the control group, 51 to group A (40,000 units of rHuEPO once per week) and 49 to group B (40,000 units of rHuEPO three times per week). At baseline the demographic characteristics and severity of the disease did not differ significantly between groups (Table 1). All patients were mechanically ventilated at the time of enrollment. This was because the attending physicians did not expect an ICU stay to exceed 7 days if the patient did not need mechanical ventilatory support at the time of randomization.
The pretransfusion Hct and Hb did not differ significantly between groups, averaging 24.5 ± 3.2%, 24.1 ± 2.7% and 23.5 ± 1.8%, and 7.9 ± 1.1 g/dl, 7.6 ± 0.8 g/dl and 7.7 ± 0.9 g/dl, respectively, in the control group, group A and group B. The cumulative number of RBC units transfused, the average RBC units transfused per patient and per transfused patient, and the average volume of RBC transfused per day were significantly higher in the control group than in groups A and B, whereas the differences between groups A and B were not significant. Also, the percentage of transfused patients was significantly higher in control group than in groups A and B (Table 2). Noncompliance of physicians with the transfusion strategy, as indicated by a finding of pretransfusion Hb 0.5 g/dl higher than the transfusion threshold, occurred on 10 occasions in control group (7% of the total units transfused in control group), on seven in group A (21%) and on three (13%) in group B (P > 0.05).
Transfusion rate represents the mean transfusion per patient per day. Because of the presence of many zeros, a zero-inflated Poisson distribution was deemed suitable for modelling the data [22]. This is approximately equivalent to using two separate analyses. The first is the percentage of patients with no transfusion requirement and the second is the fit of a Poisson regression to the data for the transfused patients only. The percentages of patients with no need for transfusion were 41.7%, 62.7% and 73.5% for the control group, group A and group B, respectively. The percentage in group B was statistically different from that in the control group (Fisher's exact test with Bonferroni correction, P = 0.002). The percentage in group A did not differ significantly from those in group B and the control group. Considering the transfusion rate for the transfused patients only, group A exhibited the lowest value (22.8 ml). This value was significantly different from the corresponding transfusion rates for patients of group B (32.5 ml; P < 0.001) and the control group (59.3 ml; P < 0.001).
There was a dose response of Hb and Hct to rHuEPO, which was evident from study days 14 to 28 (Table 3). The mean increase in Hct (ΔHct) and Hb (ΔHb) from baseline to the final measurement was significantly greater in group B than in the control group (Table 2). ΔHct was significantly higher in group B than in group A. ΔHct in group A was significantly higher than in control individuals, whereas ΔHb did not differ significantly between the control group and group A (Table 2).
There was no significant difference in lengths of ICU and hospital stay among the three groups. Mean ICU length of stay averaged 21.8 ± 8.2, 21.0 ± 8.3 and 19.6 ± 8.8 days in the control group, group A and group B, respectively (P > 0.05). Seven patients stayed in the ICU for less than 7 days, two of whom were in the control group and five were in group B. Exclusion of these patients did not materially alter the results (22.5 ± 7.4, 21.0 ± 8.3 and 21.3 ± 7.5 days, respectively, in the control group, group A and group B). There was a weak relationship between the total transfusion need (in ml) and length of ICU stay (r = 0.162, P = 0.05). Again, exclusion of the seven patients who stayed in the ICU for less than 7 days did not change that relationship. Also, mean ICU free days did not differ between groups, averaging 6.3 ± 8.2 days in the control group, 7.0 ± 8.3 days in group A and 8.5 ± 8.8 days in group B. There was no significant difference in mean ventilator free days among groups (10.3 ± 10.6 days in the control group, 11.1 ± 11.5 days in group A and 11.9 ± 10.4 days in group B).
Seven, five and ten patients died, respectively, in the control group, group A and group B, resulting in corresponding mortality rates of 14.6%, 9.8% and 20.4% (P > 0.05). The incidence of serious adverse events reported was comparable between the three groups (Table 4). At least one adverse event occurred in 23 patients (48.8%) in the control group, in 21 (41.2%) in group A and in 22 (45.8%) in group B.
Discussion
The main findings of our study were as follows: in critically ill patients rHuEPO administration significantly reduced the need for RBC transfusions; the magnitude of this reduction did not differ between the two dosing schedules; and there was a dose response of Hct and Hb to rHuEPO in these patients.
The study was not blinded and this might be a limitation. The designers of the study considered it unethical to administer placebo subcutaneously to critically ill patients. However, we believe that this limitation is unlikely to have influenced the results. Contrary to other studies dealing with rHuEPO administration in critically ill patients [15,16], in our study a specific transfusion threshold was applied. Indeed, the indications for transfusion were predefined and based on objective indices [19]. This is further supported by the similar value of pretransfusion Hb and Hct in the three groups of patients, indicating that our results cannot be accounted for by different transfusion strategies between groups. In addition, the rate of failure of physicians to adhere to the predefined transfusion trigger was comparable between the three groups. These factors suggest that the lack of blinding was not a significant confounding factor.
In both dosing regimens rHuEPO was given in much higher doses than are used in patients who are not critically ill [23]. However, it is well known that the requirement for erythropoietin is increased in patients with severe illness [24]. In addition, the studies that found an effect of rHuEPO on erythropoiesis in critically ill patients [15-17] used doses in the range of 40,000 to approximately 120,000 units per week, which are comparable to those given in our study.
In our study 58% of patients randomly assigned to the control group received transfusion, and on average they received approximately 5 units of RBCs. These findings are remarkably similar to those reported by other studies [1,2] and emphasize the great number of ICU patients who need RBC transfusion. We also showed that exogenous administration of rHuEPO to ICU patients at a dose of 40,000 units once or thrice per week was able to reduce the number of transfused patients by 36% and 55%, respectively. Similar results were also reported by two randomized studies showing that rHuEPO at a dose of 300 units/kg for 5 days followed by administration every other day [15] or 40,000 once weekly [16] decreased the number of transfused critically ill patients by approximately 18%. We further showed that the reduction in the proportion of transfused patients, the total and daily RBC units transfused, the transfused RBC units per patient and per transfused patient, and the average days transfused did not differ between the two dosing regimens. The similar reduction in the need of transfusion indicates that if the purpose of the administration of rHuEPO is to reduce RBC transfusion in critically ill patients, then 40,000 once weekly is probably sufficient. Higher doses increase the cost of therapy considerably, whereas on the other hand they are unlikely to have significant impact on transfusion needs. However, we should note that these results were obtained with predefined indications for transfusion based on a certain restrictive transfusion strategy that is currently recommended [19]. Different results might have been obtained if other transfusion strategies were used.
It is believed that critically ill patients have limited ability to compensate for the fall in Hb concentration [25,26]. Indeed, in these patients anaemia is associated with increased morbidity and mortality, particularly in patients with pre-existing cardiac disease [25,26]. Transfusion of RBCs may not be the ideal therapy for these patients because there is a growing body of evidence indicating that RBC transfusion in critically ill patients is an independent risk factor for increased morbidity and mortality [1,2]. In addition, it has been shown that in the majority of critically ill patients the transfused blood is relatively old (>10 days) [2] and this may limit the ability of transfused RBCs to increase the supply of oxygen to tissues [11]. The immunomodulatory effects of blood transfusion is also of great concern in critically ill patients in whom the risk for infections is high. It was recently shown in patients undergoing hip replacements that the levels of natural killer cell precursors and interferon-γ were substantially reduced by the surgery and blood loss and by transfusions of allogeneic nonleucodepleted, allogeneic leucodepleted, and autologous predeposit blood [5]. Considering that critically ill patients are often immunoparalyzed [27], the immunosuppressive effects of blood transfusion should be taken into account. Thus, preventing anaemia by administration of rHuEPO minimizes the risks for anaemia but without exposing the critically ill to the deleterious effect of RBC transfusion. Studies have shown that rHuEPO can achieve this goal [15-17]. We further demonstrated a dose response of Hct and Hb to rHuEPO; Hct and Hb increased with increasing dose of rHuEPO. It is of interest that the final values of Hb and Hct achieved with the highest dose of rHuEPO were close to normal. It follows that if the goal of rHuEPO therapy is to attain normal values of Hct and Hb, then 40,000 units thrice per week may be a reasonable strategy.
Corwin and coworkers [28] showed that administration of rHuEPO in critically ill patients at a dose of 40,000 once per week was not associated with increased side effects. These findings are further extended by our study, demonstrating that rHuEPO even at higher doses of 40,000 three times per week is probably safe; no significant difference was observed between rHuEPO groups and the control group in terms of side effects. Nevertheless, the sample size was relatively small, and so comments regarding safety should be made with great caution.
It is currently unclear whether administration of rHuEPO in critically ill patients is associated with improved outcome. Our study was underpowered to demonstrate an effect of rHuEPO on mortality or resource utilization (i.e. length of ICU or hospital stay, and ICU and ventilator free days). Nevertheless, Corwin and coworkers [28] reported that neither morbidity nor mortality differed significantly between critically ill patients receiving rHuEPO 40,000 once per week and a placebo group. However, interpretation of these results is complicated by the fact that the majority of patients receiving rHuEPO were anaemic by the end of the study, and Hb level differed slightly between groups (approximately 0.3 g/dl). Considering the relationship between the level of anaemia and morbidity and mortality [25,26], the inability of this rHuEPO regimen to increase Hb to normal levels might have influenced the morbidity and mortality data. Perhaps doses of rHuEPO that result in near normal values of Hb and Hct such as those used in the present study (i.e. 40,000 units three times per week) may be associated with improved outcome. Further studies with the appropriate power are needed to resolve this issue.
Finally, we should emphasize that the present study was designed to evaluate the efficiency of two dosing regimens of rHuEPO in increasing Hb and Hct and decreasing transfusion requirements, and not to collect pharmacoeconomic data. Although the study provides some data such as length of stay and pharmaceutical treatment, it is very difficult to estimate precisely the cost of overall medical services because Greek National Health System prices are underestimated because of the fact that the System is based on very low charges for patients, largely subsidized by taxation.
Conclusion
In conclusion, the present study showed that administration of rHuEPO reduces the need for RBC transfusion in critically ill patients. The magnitude of this reduction was similar between the two dosing schedules. On the other hand, rHuEPO increased Hb and Hct in a dose dependent manner. These results indicate that dose of rHuEPO in critically ill patients should be titrated depending on the desired goal.
Key messages
• Administration of rHuEPO reduced the need for RBC transfusion in critically ill patients.
• The magnitude of this reduction was similar between the two dosing schedules.
• rHuEPO administration increases Hb and Hct in a dose dependent manner.
• The dose of rHuEPO in critically ill patients may be titrated depending on the desired goal.
Abbreviations
Hb = haemoglobin; Hct = haematocrit; ICU = intensive care unit; RBC = red blood cell; rHuEPO = recombinant human erythropoietin.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
DG designed the study, supervised the study, analyzed the data and wrote the manuscript. DM, CR, AM, AM, GD, EZ, GN, GT, KM and AM designed the study.
Acknowledgements
This study was supported by grant from Janssen-Cilag. We would like to acknowledge the support of the other members of the EPO Critical Care Trial Greek Group (Nektaria Xirouchaki, Miranda Anastasaki, Eleni Sinnefaki, Kostas Relos, Chara Nikolaou, Maria Baka, Vassilis Koulouras, Tania Tsapoga, John Pavleas, Georgia Vasiliadou, Maria Peftoulidou, Kriton Filos) who collected the data for this study
Figures and Tables
Figure 1 Study flow chart.
Table 1 Demographics and baseline characteristics (at day 0)
Characteristic/parameter Control group Group A Group B
Number of patients 48 51 49
Age (years [median, range]) 58 (19–86) 60 (19–91) 63 (22–88)
Sex (n)
Men 33 41 39
Women 15 10 10
APACHE II score (mean ± SD) 14.4 ± 6.1 14.4 ± 5.3 14.9 ± 5.9
Admitting diagnosis (n)
Trauma 12 19 20
Surgical 9 6 11
Nonsurgical 27 26 18
Baseline laboratory values (day 0)
Hct (% [mean ± SD]) 28.3 ± 3.7 28.2 ± 3.7 28.4 ± 2.8
Hb (g/dl [mean ± SD]) 9.2 ± 1.3 9.3 ± 0.9 9.3 ± 1.2
Reticulocytes (% [mean ± SD]) 3.0 ± 4.2 3.1 ± 3.2 3.6 ± 4.4
Iron (μg/dl [mean ± SD]) 41.3 ± 23.0 46.2 ± 37.1 39.7 ± 24.4
Ferritin (ng/dl [mean ± SD]) 590.6 ± 454.3 453.5 ± 443.1 561.3 ± 489.5
Transferrin saturation (% [mean ± SD]) 23.7 ± 13.1 22.3 ± 12.6 24.8 ± 12.7
The three groups were comparable at enrollment (P > 0.05) with respect to baseline demographic characteristics, admitting diagnosis, severity score and laboratory values. APACHE, Acute Physiology and Chronic Health Evaluation; Hct, haematocrit; Hb, haemoglobin; SD, standard deviation.
Table 2 Study outcomes
Parameter Control Group A Group B
Total units transfused 138 33* 23*
% of transfused patients 58.3 37.3* 26.5*
Units transfused per patient 2.83 ± 3.9 0.64 ± 1.0* 0.47 ± 0.9*
Units transfused per transfused patient 4.86 ± 4.0 1.74 ± 0.7* 1.77 ± 0.7*
Volume of RBCs transfused per day 43.2 ± 61.1 11.3 ± 20.3* 16.1 ± 36.1*
Average days transfused 1.60 ± 2.2 0.59 ± 0.9* 0.35 ± 0.7*
Hct (%) at day 28 30.4 ± 4.5 33.3 ± 5.3* 37.5 ± 5.8* †
Hb (mg/dl) at day 28 9.9 ± 1.5 10.7 ± 1.9 11.6 ± 1.9*
ΔHct (%) 2.09 ± 5.0 5.53 ± 5.5* 8.76 ± 6.2* †
ΔHb (mg/dl) 0.69 ± 1.5 1.43 ± 1.7 2.24 ± 6.2*
*P < 0.05 versus control; †P < 0.05 versus group A. ΔHb, mean increase in Hb from baseline to final measurement; ΔHct, mean increase in Hct from baseline to final measurement; Hb, haemoglobin; Hct, haematocrit; RBC, red blood cell.
Table 3 Haematocrit and haemoglobin on different study days
Day Haematocrit Haemoglobin
Control Group A Group B Control Group A Group B
0 28.3 ± 3.7 28.2 ± 3.7 28.4 ± 2.8 9.2 ± 1.3 9.3 ± 1.2 9.2 ± 0.9
3 27.9 ± 3.7 26.7 ± 5.0 28.2 ± 3.7 9.1 ± 1.3 9.0 ± 1.1 9.1 ± 1.2
7 28.4 ± 4.1 28.4 ± 3.7 29.5 ± 4.0 9.1 ± 1.4 9.2 ± 1.2 9.5 ± 1.3
10 28.8 ± 4.5 29.4 ± 4.1 30.3 ± 4.2* 9.4 ± 1.6 9.6 ± 1.2 9.7 ± 1.4
14 28.8 ± 4.5 30.8 ± 4.5 32.3 ± 5.4* 9.4 ± 1.4 10.0 ± 1.5 10.3 ± 1.9*
21 29.4 ± 6.2 31.9 ± 5.0 36.5 ± 6.3* † 9.7 ± 2.0 10.3 ± 1.7 11.3 ± 2.1*
28 30.4 ± 4.5 33.3 ± 5.3 37.5 ± 5.8* † 9.9 ± 1.5 10.6 ± 1.9 11.6 ± 1.9*
Shown are mean ± standard deviation values for haematocrit and haemoglobin on different study days. *P < 0.05 versus control; †P < 0.05 versus group A.
Table 4 Serious adverse events
Adverse event Control Group A Group B
Deaths 7 5 10
Nosocomial infection 21 22 26
Heart rate and rhythm disorders 3 2 2
Thrombocytosis 1 - 1
Deep vein thrombosis 2 - 1
Central nervous system disorders (new ischaemic or haemorrhagic stroke) 3 2 2
==== Refs
Vincent JL Baron JF Reinhart K Gattinoni L Thijs L Webb A Meier-Hellmann A Nollet G Peres-Bota D Anemia and blood transfusion in critically ill patients JAMA 2002 288 1499 507 12243637 10.1001/jama.288.12.1499
Corwin HL Gettinger A Pearl RG Fink MP Levy MM Abraham E MacIntyre NR Shabot MM Duh MS Shapiro MJ The CRIT Study: Anemia and blood transfusion in the critically ill-Current clinical practice in the United States Crit Care Med 2004 32 39 52 14707558 10.1097/01.CCM.0000104112.34142.79
Vincent JL Sakr Y Creteur J Anemia in the intensive care unit Can J Anaesth 2003 50 6 Suppl S53 S59 14629054
Stevens CE Aach RD Hollinger FB Mosley JW Szmuness W Kahn R Werch J Edwards V Hepatitis B virus antibody in blood donors and the occurrence of non-A, non-B hepatitis in transfusion recipients. An analysis of the Transfusion-Transmitted Viruses Study Ann Intern Med 1984 101 733 738 6437308
Gharehbaghian A Haque KM Truman C Evans R Morse R Newman J Bannister G Rogers C Bradley BA Effect of autologous salvaged blood on postoperative natural killer cell precursor frequency Lancet 2004 363 1025 1030 15051282 10.1016/S0140-6736(04)15837-6
Goodnough LT Brecher ME Kanter MH AuBuchon JP Transfusion medicine. First of two parts: blood transfusion N Engl J Med 1999 340 438 447 9971869 10.1056/NEJM199902113400606
Spence RK Cernaianu AC Carson J DelRossi AJ Transfusion and surgery Curr Probl Surg 1993 30 1101 1180 8261799 10.1016/0011-3840(93)90043-G
Blumberg N Heal JM Immunomodulation by blood transfusion: an evolving scientific and clinical challenge Am J Med 1996 101 299 308 8873492 10.1016/S0002-9343(96)00124-6
Brunson ME Alexander JW Mechanisms of transfusion-induced immunosuppression Transfusion 1990 30 651 658 2205953 10.1046/j.1537-2995.1990.30790385527.x
Looney MR Gropper MA Matthay MA Transfusion-related acute lung injury: a review Chest 2004 126 249 258 15249468 10.1378/chest.126.1.249
Marik PE Sibbald WJ Effect of stored-blood transfusion on oxygen delivery in patients with sepsis JAMA 1993 269 3024 3029 8501845 10.1001/jama.269.23.3024
Fitzgerald RD Martin CM Dietz GE Doig GS Potter RF Sibbald WJ Transfusing red blood cells stored in citrate phosphate dextrose adenine-1 for 28 days fails to improve tissue oxygenation in rats Crit Care Med 1997 25 726 732 9187588 10.1097/00003246-199705000-00004
Shorr AF Duh MS Kelly KM Kollef MH Red blood cell transfusion and ventilator-associated pneumonia: a potential link? Crit Care Med 2004 32 666 674 15090945 10.1097/01.CCM.0000114810.30477.C3
Taylor RW Manganaro L O'Brien J Trottier SJ Parkar N Veremakis C Impact of allogenic packed red blood cell transfusion on nosocomial infection rates in the critically ill patient Crit Care Med 2002 30 2249 2254 12394952 10.1097/00003246-200210000-00012
Corwin HL Gettinger A Rodriguez RM Pearl RG Gubler KD Enny C Colton T Corwin MJ Efficacy of recombinant human erythropoietin in the critically ill patient: a randomized, double-blind, placebo-controlled trial Crit Care Med 1999 27 2346 2350 10579246 10.1097/00003246-199911000-00004
Corwin HL Gettinger A Pearl RG Fink MP Levy MM Shapiro MJ Corwin MJ Colton T Efficacy of recombinant human erythropoietin in critically ill patients: a randomized controlled trial JAMA 2002 288 2827 2835 12472324 10.1001/jama.288.22.2827
van Iperen CE Gaillard CA Kraaijenhagen RJ Braam BG Marx JJ van de Wiel A Response of erythropoiesis and iron metabolism to recombinant human erythropoietin in intensive care unit patients Crit Care Med 2000 28 2773 2778 10966249 10.1097/00003246-200008000-00015
Corwin HL Surgenor SD Gettinger A Transfusion practice in the critically ill Crit Care Med 2003 31 12 Suppl S668 S671 14724464 10.1097/01.CCM.0000099348.99451.84
Hebert PC Wells G Blajchman MA Marshall J Martin C Pagliarello G Tweeddale M Schweitzer I Yetisir E A multicenter, randomized, controlled clinical trial of transfusion requirements in critical care. Transfusion Requirements in Critical Care Investigators, Canadian Critical Care Trials Group N Engl J Med 1999 340 409 417 9971864 10.1056/NEJM199902113400601
Crowe MJ Cooke EM Review of case definitions for nosocomial infection – towards a consensus. Presentation by the Nosocomial Infection Surveillance Unit (NISU) to the Hospital Infection Liaison Group, subcommittee of the Federation of Infection Societies (FIS) J Hosp Infect 1998 39 3 11 9617679 10.1016/S0195-6701(98)90237-7
Bone RC Balk RA Cerra FB Dellinger RP Fein AM Knaus WA Schein RM Sibbald WJ Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. The ACCP/SCCM Consensus Conference Committee. American College of Chest Physicians/Society of Critical Care Medicine Chest 1992 101 1644 1655 1303622
Bohning DE Schlattmann P The zero-inflated Poisson model and the decayed, missing and filled teeth index in dental epidemiology J Roy Stat Soc 1999 162 195 209 10.1111/1467-985X.00130
Cody J Daly C Campbell M Donaldson C Grant A Khan I Vale L Wallace S MacLeod A Frequency of administration of recombinant human erythropoietin for anaemia of end-stage renal disease in dialysis patients Cochrane Database Syst Rev 2002 4 CD003895 12519614
Rogiers P Zhang H Leeman M Nagler J Neels H Melot C Vincent JL Erythropoietin response is blunted in critically ill patients Intensive Care Med 1997 23 159 162 9069000 10.1007/s001340050310
Hebert PC Wells G Tweeddale M Martin C Marshall J Pham B Blajchman M Schweitzer I Pagliarello G Does transfusion practice affect mortality in critically ill patients? Transfusion Requirements in Critical Care (TRICC) Investigators and the Canadian Critical Care Trials Group Am J Respir Crit Care Med 1997 155 1618 1623 9154866
Nelson AH Fleisher LA Rosenbaum SH Relationship between postoperative anemia and cardiac morbidity in high-risk vascular patients in the intensive care unit Crit Care Med 1993 21 860 866 8504653
Kox WJ Volk T Kox SN Volk HD Immunomodulatory therapies in sepsis Intensive Care Med 2000 26 Suppl 1 S124 S128 10786969 10.1007/s001340051129
Corwin HL Anemia and blood transfusion in the critically ill patient: role of erythropoietin Crit Care 2004 8 Suppl 2 S42 S44 15196323 10.1186/cc2411
|
16277712
|
PMC1297615
|
CC BY
|
2021-01-04 16:04:56
|
no
|
Crit Care. 2005 Aug 5; 9(5):R508-R515
|
utf-8
|
Crit Care
| 2,005 |
10.1186/cc3786
|
oa_comm
|
==== Front
Crit CareCritical Care1364-85351466-609XBioMed Central London cc37901627771410.1186/cc3790ResearchApplication of a population-based severity scoring system to individual patients results in frequent misclassification Booth Frank V [email protected] Mary [email protected] Andrew F [email protected] Nancy [email protected] Becky [email protected] Rebecca L [email protected] Howard [email protected] Medical Fellow, Eli Lilly and Company, Indianapolis, IN, USA2 Associate Clinical Research Scientist, Eli Lilly and Company, Indianapolis, IN, USA3 Associate Director of Pulmonary Critical Care Medicine, Pulmonary and Critical Care Medicine, Washington Hospital Center, Washington, DC, USA and Associate Professor of Medicine, Georgetown University, Washington, DC, USA4 Senior Clinical Development Associate, Eli Lilly and Company, Indianapolis, IN, USA5 Associate Senior Statistician, Eli Lilly and Company, Indianapolis, IN, USA6 Senior Scientific Communication Associate, Eli Lilly and Company, Indianapolis, IN, USA7 Medical Director, Eli Lilly and Company, Indianapolis, IN, USA+2005 9 8 2005 9 5 R522 R529 6 5 2005 15 6 2005 1 7 2005 12 7 2005 Copyright © 2005 Booth 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
APACHE II (AP2) was developed to allow a systematic examination of intensive care unit outcomes in a risk adjusted manner. AP2 has been widely adopted in clinical trials to assure broad consistency amongst different groups. Although errors in calculating the true AP2 score may not be reducible below 15%, the self-canceling effect of random errors reduces the importance of such errors when applied to large populations. It has been suggested that a threshold AP2 score be used in clinical decision making for individual patients. This study reports the AP2 scoring errors of researchers involved in a large sepsis trial and models the consequences of such an error rate for individual severe sepsis patients.
Methods
Fifty-six researchers with explicit training in data abstraction and completion of the AP2 score received scenarios consisting of composites of real patient histories. Descriptive statistics were calculated for each scenario. The standard deviations were calculated compared with an adjudicated score. Intraclass correlations for inter-observer reliability were performed using Shrout-Fleiss methodology. Theoretical distribution curves were calculated for a broad range of AP2 scores using standard deviations of 6, 9 and 12. For each curve, the misclassification rate was determined using an AP2 score cut-off of ≥25. The percentage of misclassifications for each true AP2 score was then applied to the corresponding AP2 score obtained from the PROGRESS severe sepsis registry.
Results
The error rate for the total AP2 score was 86% (individual variables were in the range 10% to 87%). Intraclass correlation for the inter-observer reliability was 0.51. Of the patients from the PROGRESS registry. 50% had AP2 scores in the range 17 to 28. Within this interquartile range, 70% to 85% of all misclassified patients would reside.
Conclusion
It is more likely that an individual patient will be scored incorrectly than correctly. The data obtained from the scenarios indicated that as the true AP2 score approached an arbitrary cut-off point of 25, the observed misclassification rate increased. Integrating our study of AP2 score errors with the published literature leads us to conclude that the AP2 is an inappropriate sole tool for resource allocation decisions for individual patients.
==== Body
Introduction
The Acute Physiology and Chronic Health Evaluation II (APACHE II) scoring system was originally developed as a tool for comparing the outcomes of acute disease in critically ill patients across multiple intensive care units in a therapy-independent fashion [1]. Although relatively few critical care units have adopted this system or its successor, APACHE III, for this purpose, APACHE II has found widespread application in clinical trials as a tool both for stratification of patient populations and as a means of demonstrating acceptable baseline balance amongst subgroups within a given trial. In large groups of patients, it has repeatedly been demonstrated that there is excellent correlation between APACHE II score and risk of death. The actual mortality risk predicted by this scoring system varies considerably with the underlying diagnosis and from country to country. The developers of APACHE II have emphasized that an accurate classification of the underlying disease state is essential for the accuracy of the predictive model [1].
The total APACHE II score is derived by summing points from three distinct categories: acute physiologic derangements (12 individual elements); age points; and points for the presence of certain specific chronic health conditions or medical situations. Within the acute physiologic score, three elements require additional decisions or preparatory calculation: the Glasgow coma score; an assessment of pulmonary function; and a decision if an abnormal value of creatinine represents acute or chronic renal failure. The difficulties of reliably determining Glasgow Coma Score have been well documented. In assessing pulmonary function, depending on the fraction of inspired oxygen (FiO2), either the arterial partial oxygen pressure (pO2) or the alveolar-arterial oxygen gradient (A-a DO2) must be used. The calculation of the latter requires the successful application of the alveolar gas equation, which in turn requires knowledge of average local atmospheric pressure. These numerous and complex data manipulations required to calculate the APACHE II score introduce many opportunities for error in the determination of an individual patient's points total. The combination of many elements into a composite score means that there are literally thousands of data permutations, which may be recorded to produce an identical APACHE II score.
This retrospective study reports the APACHE II scoring error rates for three case scenarios calculated by Clinical Research Associates and Research Coordinators involved in a large randomized placebo-controlled critical care clinical trial. We examined the effects of these scoring error rates on the ability to correctly classify an individual into either having an APACHE II score above or below a cut-off score of 25. In addition, we used a large database of patients with severe sepsis to estimate the distribution of reported APACHE II scores. Combining this known distribution of APACHE II scores and our estimated misclassification rates, we estimated the overall frequency of misclassification of individual severe sepsis patients into categories of having an APACHE II score above or below 25.
Methods
Study participants
Fifty-six individuals (clinical research associates (n = 17) and study coordinators (n = 39), associated with the ADDRESS clinical trial) returned completed case scenarios used in this study. Demographic data on these individuals were not obtained. All received explicit training in data abstraction and recording for the ADDRESS trial, a multi-institutional investigation of drotrecogin alfa (activated) in severe sepsis. Study procedures for this trial required that APACHE II score be obtained at baseline, either from the medical record if this calculation was part of the clinical routine at the specific institution or as a study-specific determination. The study coordinators came from individual participant sites in the ADDRESS trial and were either employees or associates of the principal investigators at those sites. The clinical research associates were either employees of Eli Lilly and Company or of a contract research organization engaged by Lilly to assist in the conduct of the ADDRESS trial. The case scenarios, instructions and scoring sheets for APACHE II were distributed to the participants at the beginning of a two-day study initiation meeting and were returned at its conclusion. Participants completed these forms individually. No constraints were applied on the time allowed for completion. Participants were given the option of returning the score sheets anonymously or with their names included (for the purpose of receiving feedback). Almost without exception, score sheets were returned bearing the participant's name, but were subsequently obliterated and replaced with an anonymous identifier for the purposes of data analysis for this study.
Case scenarios
Three individual case scenarios were developed using composites of real patient histories and laboratory values. Each scenario consisted of several elements but all contained at a minimum: a multi-page critical-care vital signs flow sheet (with multiple and frequent observations of pulse rate, blood pressure, respiratory rate, components of the Glasgow coma score, etc.); and a laboratory values report in the form of a spreadsheet, typically covering a 48 h period and including 18 routine chemistries, cardiac enzymes, arterial and venous blood gas values as well as routine hematology results. The third element of the scenario was a narrative summary of the patient's clinical course. In many cases this summary contained items of relevance to the calculation of an APACHE II score, such as times of landmark events, and physiologic values observed in the pre-hospital or emergency room environment. The participants were given a standardized APACHE II scoring sheet and instruction set.
Adjudicated APACHE II score
Two of the authors (MS and FVMcLB) independently scored each clinical scenario on two separate occasions approximately two weeks apart. A consensus-forming session was then held at which every individual contributing element of the APACHE II score was reviewed, agreed upon and an adjudicated point value determined. For one of the scenarios (APACHE II score = 22) the agreed aggregate point value was identical to the value calculated by the two observers independently. For the other two, an adjustment of a single point was agreed upon. These consensus values were then used as the adjudicated values against which the scores of the study participants were measured.
Statistical methodology
Descriptive statistics (mean, median, inter-quartile range) were calculated for each scenario. The standard deviations were calculated using the adjudicated APACHE II value in place of the mean reported APACHE II score.
Intraclass correlations for inter-observer reliability were performed using Shrout-Fleiss methodology [2]. The intraclass correlation used in this study assumed the same observers scored the three scenarios, although each scenario was a random subset of all possible observers. In the second phase of this study, it was assumed that for any given population of patients with an identical true APACHE II score, the distribution of possible APACHE II scores would be approximately normal. Theoretical distribution curves were calculated for each true APACHE II score using standard deviations of 6, 9 and 12. For each distribution curve, the misclassification rate was determined in the following manner. If the true score was <25, misclassification was represented by the area of the distribution curve above or equal to 25. If the true score was ≥25, misclassification was represented by the area of the distribution curve below 25.
A large sample of APACHE II scores (n = 5,253) was obtained from the PROGRESS registry, a collaborative web-based registry of severe sepsis patients admitted to over sixty intensive care units worldwide [3]. The percentage of misclassifications for each true APACHE II score estimated in the second phase of this study was applied to the corresponding scores in this large sample of APACHE II scores. An overall misclassification rate was estimated by summing the misclassifications for each APACHE II score from this sample.
Results
Not every participant completed every case scenario; the completion rate was 159/168 (94.6%). Fifteen participants returned composite scores only. The three different scenarios had widely differing adjudicated APACHE II scores. The scenario with an adjudicated score of 44 was most frequently scored incorrectly (52/56, 92.9% incorrect). The accuracy of scoring was better for the other two scenarios whose adjudicated scores were markedly lower (score = 22: 45/52, 86.5 % incorrect; score = 19: 41/43, 77.4% incorrect). In only two of the numerically correct total scores did the participant arrive at their answers by a balanced combination of errors.
In contrast to the scenario with a score of 44 in which all but one of the erroneous scores underestimated the true APACHE II score, the distribution of the erroneous scores assumed a more normal random distribution for scenarios with scores of 19 and 22. The intraclass correlation for the inter-observer reliability was 0.51, 95% CI (0.22–0.98). The results of the scoring exercise, individual scores, standard deviations and interquartile ranges are shown in Fig. 1.
Table 1 lists the error rate for each component of the APACHE II score. Fig. 2 shows the theoretical distribution curves of five true values of APACHE II scores. The areas shaded show the proportion of scores that would result in a misclassification using an APACHE II score cut-off of 25 or greater. The value of 25 was chosen because it has been suggested that this value may be used to identify a patient at high risk of death from severe sepsis. The effect of varying the assumed standard deviation is also shown. The proportion of misclassification increases as the true score approaches the cut-off score of 25. The highest rate of misclassification occurs when the true score equals the cut-off score.
Fig. 3 shows the relative frequency of APACHE II scores observed in a population of severe sepsis patients (PROGRESS Registry). The lightly shaded areas in Fig. 3 show the estimated distribution of misclassification rates of individuals with severe sepsis into groups of scores <25 and ≥25 based on the estimated misclassification rates from the theoretical distribution curves. Using this distribution of APACHE II scores from the PROGRESS registry, 50% of severe sepsis patients have APACHE II scores ranging from 17 to 28. Within this interquartile range will reside 70% to 85% of all misclassified patients (depending on the underlying standard deviation of the APACHE II scores).
Discussion
Our observations concerning the accuracy and inter-observer variability of deriving APACHE II scores in a simulated clinical setting are concordant with those of Polderman et al. [4] and Chen et al. [5], although the absolute magnitude of the errors we report is larger. Fig. 1 shows the differing distributions of erroneously calculated scores. The scenario with a score of 44 has a mean and median that substantially underestimate the adjudicated APACHE II score, whereas the distribution of errors in the other two scenarios resulted in a mean and median within three APACHE II points of the adjudicated score. This would be considered statistically indistinguishable from the adjudicated value and acceptable from a scientific sampling point of view when comparing population intensive care unit outcomes, or the success of randomizing patients into subgroups with comparable severity.
Furthermore, it should be noted that in our study, Glasgow coma scores as recorded in the flow sheets were assumed to be accurate. In nursing practice, errors in reporting of Glasgow coma scores, especially for the intubated patient, are well documented [6]. APACHE II methodology requires that scores for creatinine intervals be doubled for acute renal failure. Overall, creatinine points were assessed incorrectly 43% of the time. In one of the scenarios, however, all the creatinine values were within normal limits. Thus, on the score sheets where the creatinine score should have been doubled, this step was omitted on 64 of a possible 72 (89%) occasions, making omission of this step in practice the most frequent error seen.
In this simulation the practitioners were provided with a summary clinical abstract, which was assumed to contain all the relevant clinical and time-line information. In the clinical environment such information is frequently intimately interwoven with extraneous confusing and irrelevant matter within a complex clinical chart. This likely increases the possibility that an important point of information may be overlooked. Table 1 demonstrates that even invariable information such as age was incorrectly abstracted or assigned to the incorrect interval that determines the score for that parameter, and that for many parameters an incorrect value was more likely to be assigned than a correct value. In selecting the range of standard deviation used for the illustrative theoretical curves (Fig. 2), we relied on values observed in the case scenarios (standard deviations of 6 and 12), and arbitrarily included a standard deviation of 9.
In reviewing the technical literature of APACHE II, two distinct approaches are seen: the overall performance of the score as an outcome predictor for groups of patients; and the performance of small groups of individuals in achieving accuracy and reproducibility of the actual APACHE II score. With regard to the much more frequent reporting of overall performance of the score as an outcome predictor for groups of patients, all of these types of studies have the underlying assumption that there is a sufficiently large patient sample size to ensure that any effect of individual error in determining the APACHE II score is trivial in comparison to the underlying trend of the group as a whole. In this guise the tool has been used to predict the outcome of classes of patients as varied as those with acute pancreatitis to patients with acute community acquired pneumonia. The original authors of the APACHE II system emphasized that although the APACHE II score was highly correlated with risk of death, an individual score could not be translated into a specific risk of death without taking into account the underlying diagnosis [1]. Thus in a large group of patients, all with an APACHE II score of 22 and the same clinical diagnosis, for example pneumonia, the risk of death would be very similar. However, the risk of death would not necessarily be the same as another group of patients also with an APACHE II score of 22 who had a different underlying diagnosis, for example, ascending cholangitis. A specific example of this was cited in the original paper [1]. Despite the different weighting given to the presence of chronic health conditions in the emergency surgical patient, there was still a substantial difference in observed mortality between medical and surgical patients. APACHE II seems to perform less well in surgical patients [7]. These cited limitations clearly show that applying a single APACHE II score cut-off to determine high risk of death to all classes of patients is less than optimal.
The second type of review has focused on the performance of small groups of individuals in achieving accuracy and reproducibility of the actual APACHE II. As Rowley and Fielding [6] have shown, inter-rater reliability alone is insufficient grounds for confidence in the accuracy of real-world APACHE II scores. In studies where the accuracy of an individual APACHE II determination is the main focus of attention, the number of cases that can be studied is necessarily limited given the intensive effort required to determine what the 'gold standard' value really is. We are not aware of any studies that attempt to examine the consequences of random or systematic errors on the performances of the APACHE II predictive model.
Although the absolute rate of erroneous APACHE II score determination that we have reported appears to be higher than that reported either by Polderman et al. [4] or Chen et al. [5], this may be largely attributable to the greater severity of physiologic derangements used in our simulations. Thus, the mean and median APACHE II score in Polderman's repeat scoring exercise was 14.3 (± 4.4) and 13.9 before rigorous training and 18.9 (± 2.4) and 16.2 after training. They do not provide adjudicated or 'gold standard' values for the individual patients they studied, so that strict comparisons of accuracy as opposed to inter-rater agreement cannot be made. The simulations we used had APACHE II scores of 19, 22 and 44. The opportunity for error rises almost geometrically with the number of deranged physiologic variables, which likely explains the higher standard deviation we observed in the simulations with the higher APACHE II scores. The overall intra-class correlation which we report (0.51) lies between the worst individual component value reported by Chen et al. [5] (for Glasgow Coma Score at 0.315) and the best (for age at 0.976). We did not perform intra-class correlations for individual elements of the APACHE II score. Despite the intrinsic variability noted by Chen et al. [5], when groups of patients were compared (as was intended by the designers of the original tool) the correlation was excellent.
The inter-rater reliability noted in this investigation (0.51) can, at best, be described as only fair. From a research perspective this underscores the potential bias in any critical care study relying on the APACHE II score either for entry into a trial or for analysis of baseline severity of illness. Moreover, if in the future novel therapies are to be targeted based on such a criterion, many patients eligible for a therapy may be excluded whereas others may be treated inappropriately. That we studied only trained researchers reinforces this point, as it seems reasonable to conclude that less specifically trained personnel or clinicians will likely make more errors in the computation of the APACHE II score. Future research in critical care might include multiple measures of severity of illness to address this limitation
Recently, it has been suggested that the APACHE II score may be a useful tool to determine if a patient has a sufficient risk of death to warrant treatment with drotrecogin alfa (activated). For a population of severe sepsis patients enrolled in the PROWESS trial [8], the APACHE II score was the strongest indicator for distinguishing patients with a response to the drug from the group that did not show a positive response [8]. Explicitly, the current US package insert for drotrecogin alfa (activated) proposes an APACHE II score of 25 or greater as a way to determine if a patient is at high risk of death [9]. Even if it is assumed that APACHE II methodology is perfect for resolving the arbitrary distinction between high risk of death and not at high risk of death, the error rate in determining the APACHE II score, which others have reported and which we have confirmed, will ensure that significant numbers of patients will be misclassified (i.e. they will be assigned to one side of a 25 point threshold when their true score lies on the other). There is a fundamental practical difference between using a scoring system such as APACHE II for examining likelihood of death, and using it to determine if a severe sepsis patient lies above or below an arbitrary threshold. In any given intensive care unit population; the majority of survivors are clustered at the low end of the APACHE II score range. Deaths are concentrated at the high end. If, in a population of patients, the observed mortality is plotted against APACHE II score, at the low end of the range the curve is quite flat. A change of score from 4 to 8 makes little difference to mortality; the vast majority still survive. Likewise, at the upper end of the range, above a score of about 40, most patients die, and increasing the score by two or three points changes the mortality little. In the mid-range of the curve, however, the mortality versus APACHE II score is very steep. A change of one or two points makes a large difference in the observed change in mortality. Thus, when using a cut-off point that happens to lie in the steepest region of the curve, the significance of scoring errors is maximized. The closer a patient's true APACHE II score approaches the cut-off point of 25, the higher the misclassification rate (this trend is illustrated in Fig. 2). Unfortunately, a cut-off value of 25 sits uncomfortably close to the median APACHE II score of 22, seen in severe sepsis patients admitted to intensive care units included in the PROGRESS registry (Fig. 3). The chance of misclassification for a patient lying within the inter-quartile range (17 to 28) is estimated to be as high as 38%. This set of patients represents the population of severe sepsis patients admitted to the intensive care unit for whom the outcome is most in doubt. Because the APACHE II scoring error rate estimates are based on a normal distribution around the true APACHE II score, these misclassification rates are conservative in nature, as the maximum misclassification rate can only be 50%. The real world distributions of scoring errors, such as seen in the scenario with APACHE II score of 22, suggest that occasionally the misclassification rate can exceed 50%. If such a score is to be used in a medical decision making process, the likely error rate should be clearly understood, and serious attention should be paid to maximizing the expertise and accuracy of those responsible for the scoring process.
Conclusion
It is far more likely that an individual patient will be scored incorrectly than correctly, even by a group of individuals trained in scoring and chart abstraction. Even the scenario with an adjudicated APACHE II score that placed it many points distant from an arbitrary cut-off point of 25 was misclassified at an unacceptably high rate. Observed misclassification rate for the scenario with an adjudicated score within 3 points of the cut-off was over 50%. Integrating our study of APACHE II score errors with the published literature leads us to conclude that the APACHE II is an inappropriate sole tool for resource allocation decisions for individual patients.
Key messages
• There are typically errors in execution of a complex scoring scheme such as APACHE II.
• These errors do not have a significant effect when applied to patient populations of a sufficient size.
• If a cut-off APACHE II score in the middle range of critically ill patients is used for making decisions about individual patients, an error rate that may be considered acceptable for use with sufficiently large patient populations will produce a very high rate of misclassification in those individuals so classified.
Abbreviations
APACHE II = Acute Physiology and Chronic Health Evaluation II.
Competing interests
FVMcLB, MS, NA, BB, RLQ and HL are full-time employees and shareholders of Eli Lilly and Company. AFS has been a paid consultant and speaker for Eli Lilly and Company.
Authors' contributions
All the authors have contributed to the composition, revision and review of the manuscript and have read and approved the final version. In addition, FVMcLB and MS conceived of the idea for this manuscript, BB performed the statistical analysis, RLQ edited the document and BB participated in obtaining the original PROWESS data.
Figures and Tables
Figure 1 Results of the scoring exercise. The results of the scoring exercise completed by researchers involved in a large randomized placebo-controlled critical care trial illustrating individual scores, standard deviations and inter-quartile ranges of case scenarios with adjudicated total APACHE II scores of 44, 22 and 19. 1Correct classification is determined by the adjudicated score being either APACHE II ≥25 or APACHE II <25. 2Standard deviation is calculated using the adjudicated APACHE II score in place of the mean APACHE II score.
Figure 2 Theoretical distributions of APACHE II scores with varying SDs. A set of theoretical distributions of reported APACHE II scores based on standard deviations of 6 and 12 (which were what we observed in the case scenario data.) For the purposes of comparison, a set of curves using an intermediate standard deviation of 9 is also shown. In every curve, the shaded area illustrates the theoretical probability of misclassification based on a cut-off score of ≥25.
Figure 3 Distribution of reported APACHE II scores in the PROGRESS registry. The darker shading (outer envelope) of these plots represents the observed distribution of APACHE II scores of 5,253 severe sepsis patients in the PROGRESS registry. The lighter shading (inner envelope) is calculated by applying the probability of misclassification for each individual APACHE II score based on assumed standard deviation (SD) of (from top to bottom) 6, 9 and 12 and on an APACHE II cut-off score ≥25.
Table 1 Error rates of combined case study data for each component of the APACHE II score
Acute physiologic score (A) Error rates
Temperature (rectal/core) 48%
Mean arterial pressure 59%
Heart rate (ventricular response) 46%
Respiratory rate (non-ventilated or ventilated) 45%
Oxygenation 52%
Arterial pH 38%
Serum sodium 29%
Serum potassium 26%
Serum creatinine 43%
Hematocrit 33%
White blood count 49%
Glasgow Coma Scale 69%
Total acute physiology score (A) 87%
Age points (B) 10%
Chronic health points (C) 34%
Total APACHE II score (A+B+C) 86%
==== Refs
Knaus WA Draper EA Wagner DP Zimmerman JE APACHE II: a severity of disease classification system Crit Care Med 1985 13 818 829 3928249
Shrout PE Fleiss JL Intraclass correlations: uses in assessing rater reliability Psychological Bulletin 1979 86 420 429 10.1037//0033-2909.86.2.420
Beale R Reinhart K Silva E Dobb G Sarwat S Garg R Vincent JL Comparison of PROGRESS Severe Sepsis Registry patients to INDEPTH Integrated Severe Sepsis Clinical Trial Database placebo patients Chest Meeting Abstracts; Chest 2004 126 864S Abstract #496
Polderman KH Thijs LG Girbes AR Interobserver variability in the use of APACHE II scores Lancet 1999 353 380 9950452 10.1016/S0140-6736(05)74953-9
Chen LM Martin CM Morrison TL Sibbald WJ Interobserver variability in data collection of the APACHE II score in teaching and community hospitals Crit Care Med 1999 27 1999 2004 10507631 10.1097/00003246-199909000-00046
Rowley G Fielding K Reliability and accuracy of the Glasgow Coma Scale with experienced and inexperienced users Lancet 1991 337 535 538 1671900 10.1016/0140-6736(91)91309-I
Cerra FB Negro F Abrams J APACHE II score does not predict multiple organ failure or mortality in postoperative surgical patients Arch Surg 1990 125 519 522 2322119
Bernard GR Vincent J-L Laterre P-F LaRosa SP Dhainaut J-F Lopez-Rodriguez A Steingrub JS Garber GE Helterbrand JD Ely EW Efficacy and safety of recombinant human activated protein C for severe sepsis N Engl J Med 2001 344 699 709 11236773 10.1056/NEJM200103083441001
Drotrecogin Alfa (activated) Product Label
|
16277714
|
PMC1297617
|
CC BY
|
2021-01-04 16:04:56
|
no
|
Crit Care. 2005 Aug 9; 9(5):R522-R529
|
utf-8
|
Crit Care
| 2,005 |
10.1186/cc3790
|
oa_comm
|
==== Front
Crit CareCritical Care1364-85351466-609XBioMed Central London cc37921627771610.1186/cc3792ResearchAntifactor Xa activity in critically ill patients receiving antithrombotic prophylaxis with standard dosages of certoparin: a prospective, clinical study Jochberger Stefan [email protected] Viktoria 1Luckner Günter 1Fries Dietmar R 1Mayr Andreas J 1Friesenecker Barbara E 1Lorenz Ingo 2Hasibeder Walter R 3Ulmer Hanno 4Schobersberger Wolfgang 5Dünser Martin W 11 Department of Anesthesiology and Critical Care Medicine, Innsbruck Medical University, Innsbruck, Austria2 Professor, Department of Anesthesiology and Critical Care Medicine, Innsbruck Medical University, Innsbruck, Austria3 Professor, Department of Anesthesiology and Critical Care Medicine, Krankenhaus der Barmherzigen Schwestern, Ried im Innkreis, Austria4 Professor, Institute of Medical Biostatistics, Innsbruck Medical University, Innsbruck, Austria5 Professor, Institute for Leisure, Travel and Alpine Medicine, University for Health Sciences, Medical Informatics and Technology, Hall, Austria2005 9 8 2005 9 5 R541 R548 29 5 2005 7 6 2005 16 6 2005 17 7 2005 Copyright © 2005 Jochberger 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
Deep venous thrombosis with subsequent pulmonary embolism or post-thrombotic syndrome is a feared complication in the intensive care unit. Therefore, routine prophylactic anticoagulation is widely recommended. Aside from unfractionated heparin, low molecular weight heparins, such as certoparin, have become increasingly used for prophylactic anticoagulation in critically ill patients. In this prospective study, we evaluated the potency of 3,000 IU certoparin administered once daily to reach antithrombotic antifactor Xa (aFXa) levels of 0.1 to 0.3 IU/ml in 62 critically ill patients.
Methods
AFXa levels were determined 4, 12 and 24 h after injection of certoparin. Prothrombin time, activated partial thromboplastin time, antithrombin, fibrinogen, hemoglobin, platelet count, serum urea and creatinine concentrations were documented before and 12 and 24 h after injection of certoparin.
Results
Four hours after certoparin injection (n = 32), 28% of patients were within the antithrombotic aFXa range. After 12 and 24 h, 6% achieved antithrombotic aFXa levels. Because of a severe pulmonary embolism in one study patient, an interim analysis was performed, and the dosage of certoparin was increased to 3,000 IU twice daily. This regime attained recommended antithrombotic aFXa levels in 47%, 27%, 40% and 30% of patients at 4, 12, 16 and 24 h, respectively, after twice daily certoparin injection (n = 30). Antithrombin and fibrinogen concentrations slightly increased during the observation period. Low antithrombin concentrations before certoparin were independently correlated with underdosing of certoparin. Patients with aFXa levels <0.1 IU/ml 4 h after certoparin injection required vasopressors more often and had lower serum concentrations of creatinine and urea than patients with antithrombotic aFXa levels.
Conclusion
Standard dosages of certoparin of 3,000 IU given once or twice daily are ineffective for attaining the recommended aFXa levels of 0.1 to 0.3 IU/ml in critically ill patients. Low antithrombin levels before certoparin administration were independently associated with low aFXa levels. Renal function and vasopressor therapy may further influence the effectiveness of certoparin in ensuring adequate antithrombotic prophylaxis.
==== Body
Introduction
Deep venous thrombosis is a feared complication in the intensive care unit, occurring in up to one-third of patients without prophylactic anticoagulation [1]. Pulmonary embolism and/or post-thrombotic syndrome may significantly increase morbidity and mortality in the acute and/or chronic setting of thromboembolic complications [2,3]. Risk factors for the development of deep venous thrombosis in critically ill patients include high age and trauma, heart failure or central nervous system injury as admission diagnoses. Further factors contributing to the development of thromboembolic complications are immobilization, reduction of muscle tone due to analgosedation or relaxation, mechanical ventilation and vessel injury by catheterization of large vessels [4]. Because the inflammatory network and the coagulation cascade are interconnected [5], patients with the systemic inflammatory response syndrome or sepsis are at high risk of developing venous thrombosis [6,7]. Therefore, prophylactic anticoagulation is generally recommended in all critically ill patients [2,8,9].
Currently, unfractionated heparin is the most widely used drug for antithrombotic prophylaxis in intensive care patients. Short half-life time, low costs and availability of an effective antagonist make it an almost ideal antithrombotic agent [10,11]. Variable pharmacokinetics, irregularities in laboratory monitoring, as well as adverse side effects, including heparin-induced thrombocytopenia, however, have raised concerns about the use of unfractionated heparin [10]. Moreover, current evidence suggests that twice daily low dosage unfractionated heparin (5,000 IU) may not be effective enough to prevent thromboembolism in acutely ill medical patients [12].
These considerations, together with novel pharmacological developments, have led to the increased use of low molecular weight heparins (LMWHs) as alternative antithrombotic agents in critically ill patients. Whereas unfractionated heparin appears to decrease the incidence of deep venous thrombosis by 20%, LMWHs were able to reduce it by another 30% in one study [1]. This additional LMWH-associated benefit in thromboembolic risk reduction was particularly effective in intensive care patients at high risk for thrombotic complications [12]. Accordingly, prophylactic anticoagulation with LMWH has already been suggested to be the preferred strategy in critically ill medical patients [13].
In a recent study, however, our working group observed that standard dosages of the LMWH enoxaparin were ineffective at achieving adequate antithrombotic antiFactor Xa (aFXa) levels in critically ill patients. High body weight and the degree of multiple organ dysfunction syndrome were associated with a high probability of underdosing with enoxaparin [14]. Although certoparin is another frequently used LMWH in the intensive care unit, only limited data exist on its efficacy in preventing thromboembolism in critically ill patients.
Although measurement of aFXa levels has been the most widely used method for assessing LMWH activity and the establishment of a therapeutic range for different LMWHs, results on the relationship between aFXa levels and antithrombotic activity of LMWHs are contradictory [15]. Even if aFXa and anti IIa activities have been well correlated with the dose of subcutaneously injected LMWH [16,17], some human and experimental studies could not demonstrate a strong correlation between antithrombotic activity and in vitro aFXa plasma levels [18-20]. In contrast, clinical trials have found a significant statistical relationship between aFXa plasma levels and both thrombotic and haemorrhagic outcomes with different LMWHs [21-24].
In this prospective study, the potency of certoparin in achieving adequate antithrombotic aFXa levels is examined in 62 critically ill patients. Additionally, risk factors associated with inadequate aFXa levels under standard certoparin dosages are evaluated.
Materials and methods
The present study was performed in a 12 bed general and surgical intensive care unit in a tertiary, university teaching hospital between October 2003 and December 2004. The study protocol was approved by the institutional review board and by the Ethics Committee of the Innsbruck Medical University. Written informed consent was obtained, if possible, from all patients, or otherwise from the closest family members prior to study enrolment.
Patients
Criteria for study inclusion were indication for thromboembolic prophylaxis, age >19 years, body weight >50 kg and intensive care unit stay ≥3 days. Exclusion criteria were any contraindication for anticoagulation with heparins, treatment with anticoagulatory or antiplatelet drugs, continuous veno-venous hemofiltration or other extracorporal therapies, planned or emergency surgery during the 24 h study period, administration of unfractionated heparin within 24 h preceding study entry, and hemorrhage or hemodilution of >30% of estimated blood volume.
Study protocol
Three thousand international aFXa units (Ph.Eur. 95–130 IU/mg; aFXa/antiIIa 1.5) of certoparin (= 32 mg) in pre-filled, single-dose syringes (Sandoparin®; Sandoz, Kundl, Austria) were administered into the abdominal wall subcutis of study patients once daily at 8 a.m. Strict attention was paid to the exact emptying of the whole content of the syringe. AFXa levels, as a measurement of the LMWH's biological activity, were determined on the second day of certoparin administration. Then, aFXa levels were measured before, as well as 4, 12 and 24 h after administration of certoparin.
AFXa determinations
Blood was collected using 3.13% trisodium citrate containing tubes. In the institutional laboratory, plasma aFXa levels were determined by an amidolytic assay using the specific chromogenic substrate S-2732 and bovine factor Xa as reagents and simultaneous thermal analyzers (Coamatic Heparin®; Chromogenix, Milano, Italy). No antithrombin was added to the assay in vitro. Test results were expressed in international units per milliliter. The aFXa assay standard calibration curve ranged from 0.1 to 1.3 IU/ml with a minimum limit of quantitation of 0.1 IU/ml. To exclude test-related influences on aFXa results, quality testing of the in vitro analysis was performed using dilution series. Thus, it is unlikely that methodological, test-associated errors have altered aFXa measurements in this study.
Three groups of patients were specified according to aFXa levels; AFXa levels of 0.1–0.3 IU/ml were considered to represent effective antithrombotic activity [21]. The first group consisted of patients with aFXa levels <0.1 IU/ml, which we considered to be inadequately anticoagulated according to the predefined antithrombotic range. The second group of patients had aFXa levels within the pre-specified antithrombotic range (0.1 to 0.3 IU/ml). The third group consisted of patients with aFXa levels >0.3 IU/ml.
Study endpoints
The primary endpoint of this study was to evaluate the potency of 3,000 IU certoparin administered once daily to reach antithrombotic aFXa levels of 0.1–0.3 IU/ml in a critically ill patient population. The secondary study endpoint was to identify independent risk factors for underdosing or overdosing of standard certoparin dosages.
Measurements and documentation of study parameters
The following data were collected from all study patients at study enrolment: age, sex, body mass index, admission diagnosis, and a modified Goris multiple organ dysfunction score [25] calculated from worst clinical and laboratory parameters on the day of study entry.
Documented laboratory parameters included prothrombin time, activated partial thromboplastin time, antithrombin, fibrinogen, hemoglobin and platelet count. Variables were collected before and 12 and 24 h after the second certoparin administration. Serum urea and creatinine concentrations were determined and reported before and 24 h after study enrolment.
Statistical analysis
A sample size of 60 patients was precalculated on the basis of a previous prospective study [14]. Shapiro Wilk's tests were used to check for normality distribution of data, which was approximately fulfilled in all parameters except for vasopressor requirements and activated partial thromboplastin time, which were log-transformed. For analysis of demographic, laboratory and clinical data, descriptive statistical methods were applied. Time dependence of laboratory values was analyzed using a nonparametric Friedman ANOVA.
To identify independent risk factors for underdosing or overdosing of certoparin, demographic, laboratory and clinical data were entered into a bivariate correlation model to test for differences between patients within antithrombotic aFXa concentrations versus patients with aFXa levels <0.1 or >0.3 IU/ml 4 h after certoparin injection. In case of significant correlations (p < 0.05), variables were entered into a binary logistic regression model to identify independent risk factors. The time point of 4 h after certoparin injection was chosen because antithrombotic activity of LMWH is maximal at 3 to 4 h after subcutaneous injection [21].
For each analysis, a significance level of 5% was applied. All data are given as median values and range, or percentage.
Results
During the study period, 62 patients were enrolled in the trial. Table 1 presents demographic data of all study patients, as well as admission diagnoses, cardiovascular drug requirements, multiple organ dysfunction syndrome score counts, length of intensive care unit stay, and intensive care unit mortality.
The percentage of patients within antithrombotic aFXa range at 4, 12 and 24 h after injection of 3,000 IU certoparin once daily (n = 32) is shown in Fig. 1a. Four hours after certoparin administration, median aFXa levels were <0.1 IU/ml (range, <0.1 to 0.2 IU/ml), with 28% (9/32) of patients being within the recommended antithrombotic range of 0.1 to 0.3 IU/ml. Twelve hours after certoparin administration, median aFXa levels were <0.1 IU/ml (range, <0.1 to 0.16 IU/ml), with 6% (2/32) of patients being within the antithrombotic range. Twenty-four hours after certoparin administration, median aFXa levels were <0.1 IU/ml (range, <0.1 to 0.17 IU/ml), with 6% (2/32) of patients being within the antithrombotic range. At no time point did any study patient have aFXa levels >0.3 IU/ml or show clinical signs of bleeding.
Because of a severe pulmonary embolism in one study patient, an interim analysis was performed after inclusion of 32 patients. Following renewed appraisal of the study protocol by the ethical committee, the dosage of the study medication was increased to 3,000 IU twice daily. In this study protocol, 3,000 IU certoparin were administered twice daily at 8 a.m. and 8 p.m. Once again, patients were included into the study protocol only after having received certoparin prophylaxis at 3,000 IU twice a day for one day. AFXa levels were measured before and 4, 12, 16 and 24 h after administration of certoparin. Patients receiving 3,000 IU certoparin once daily were younger than patients receiving 3,000 IU certoparin twice daily. There were no other significant differences between the groups (Table 1).
The percentage of patients within antithrombotic range at 4, 12, 16 and 24 h after injection of 3,000 IU certoparin twice daily is shown in Fig. 1b. Four hours after certoparin administration, median aFXa levels were <0.1 IU/ml (range, <0.1 to 0.28 IU/ml), with 47% (14/30) of patients being within the recommended antithrombotic range. Twelve hours after certoparin administration, median aFXa levels were <0.1 IU/ml (range, <0.1 to 0.26 IU/ml), with 27% (8/30) of patients being within the antithrombotic range. Sixteen hours after the 8 a.m., and four hours after the 8 p.m. certoparin administration, median aFXa levels were <0.1 IU/ml (range, <0.1 to 0.24 IU/ml), with 40% (12/30) of patients being within the antithrombotic range. Twenty-four hours after the 8 a.m., and twelve hours after the 8 p.m. certoparin administration, median aFXa levels were <0.1 IU/ml (range, <0.1 to 0.26 IU/ml), with 30% (9/30) of patients being within the antithrombotic range. At no time point did any study patient develop clinically relevant pulmonary embolism, have aFXa levels >0.3 IU/ml or display clinical signs of bleeding.
Table 2 describes the laboratory results obtained during certoparin therapy in all study patients. During the 24 h observation period, antithrombin and fibrinogen concentrations increased. Although these increases were statistically significant, they occurred in a clinically non-relevant range. There were no changes in prothrombin time, activated partial thromboplastin time, hemoglobin, platelet count, serum creatinine or urea concentrations after certoparin injection. There were no differences in the response of laboratory parameters to 3,000 IU certoparin given once or twice daily.
Table 3 displays bivariate and binary models for identifying independent risk factors for aFXa levels <0.1 IU/ml at 4 h after injection of standard certoparin dosages. In the bivariate analysis, patients with aFXa levels <0.1 IU/ml had significantly lower antithrombin concentrations and higher serum creatinine and urea concentrations, as well as a higher need for vasopressor drugs, than patients within the antithrombotic range. The binary model could identify only low antithrombin concentrations at baseline as an independent risk factor for low aFXa levels 4 h after injection of standard certoparin dosages.
Discussion
In this prospective study, once and twice daily injection of 3,000 IU certoparin could achieve recommended antithrombotic aFXa levels of 0.1 to 0.3 IU/ml 4 h after administration in only 28% and 47% of patients, respectively. Low antithrombin concentrations before certoparin administration were significantly correlated with low aFXa levels. These results are in striking contrast to earlier studies reporting effective antithrombotic prophylaxis with standard dosages of certoparin (3,000 IU once daily) in high risk patients [26-29].
Despite the fact that certoparin is a frequently used anticoagulant for the prevention of thromboembolic complications in critically ill patients, certoparin proved to be highly ineffective at achieving recommended antithrombotic aFXa levels in this study population. When given at a dosage of 3,000 IU once daily, one patient developed severe pulmonary embolism. In this study patient, certoparin injection could achieve an aFXa level of 0.11 IU/ml 4 h after certoparin administration, whereas aFXa levels were not detectable (<0.1 IU/ml) at 12 and 24 h. Although many individual patient- and critical illness-related factors may have caused pulmonary embolism in this patient, no specific pathogenic factors other than insufficient anticoagulation could be clinically identified.
Similarly, when the dosage frequency of certoparin was increased from 3,000 IU once daily to 3,000 IU twice daily, only 25% to 50% of patients attained antithrombotic aFXa levels during the observation period. It may be speculated that increasing the single dosage from 3,000 IU to 6,000 IU would have resulted in higher aFXa levels during the study period. Whereas a significantly higher proportion of patients would have most likely reached adequate aFXa levels at 4 h after certoparin injection, it is difficult to state whether such an increase in the dosage given once daily would have provided better anticoagulation during the 24 h period than 3,000 IU given twice daily. Moreover, it is currently unknown whether recommended aFXa levels need to be achieved only at 4 h after injection of the LMWH or during the entire dosage interval in order to guarantee adequate antithrombotic protection.
Many pathophysiological mechanisms may have contributed to the observation in this study protocol that aFXa levels were undetectable in the majority of this critically ill patient population. Augmented total body water together with changes in fluid compartments are known to change distribution volume of water soluble drugs like LMWHs in critically ill patients [30]. Furthermore, frequently observed hypoproteinemia and acid-base disturbances can alter the concentration of free, active certoparin in these patients. Aside from such factors, numerous other pathophysiological factors have been reported to influence pharmacokinetics and pharmacodynamics in the intensive care unit patient [31,32]. Low antithrombin concentrations before certoparin injection is one of the most important factors explaining the low aFXa levels after injection of standard certoparin dosages in this study population. LMWHs, as well as unfractionated heparin, exert their anticoagulatory effects by accelerating the inhibitory effects of antithrombin on thrombin formation [33]. Although LMWH-induced bridging between antithrombin and factor Xa is less critical for aFXa activity [34], certoparin could only achieve adequate aFXa levels in this critically ill patient population if antithrombin levels were approximately >70%. Because of ongoing, multifactorial activation of the coagulation system in critical illness, antithrombin levels are often decreased in intensive care patients [35]. Furthermore, patients with low antithrombin concentrations mostly suffer from severe disease and may thus be more likely to receive vasopressor drugs. As indicated in the bivariate statistical model in this study and also in other clinical trials [36,37], patients with cardiovascular failure treated with vasopressor drugs had lower aFXa levels after subcutaneous injection of standard dosages of LMWHs. This is most likely due to reduced subcutaneous blood flow with impaired drug absorption [36].
Because LMWHs are predominantly eliminated by a non-saturable renal mechanism as active or inactive fragments [38], kidney function may substantially influence aFXa levels after certoparin administration. Although it did not reach statistical significance in this multiple regression model, patients with inadequately low aFXa levels after standard certoparin dosages had significantly lower serum creatinine and urea concentrations than patients presenting with antithrombotic aFXa levels between 0.1 and 0.3 IU/ml after certoparin injection. This might be interpreted as better renal function in these patients, which seems to have resulted in a higher clearance of certoparin. Similar effects of renal function on aFXa activity with other LMWHs have been reported [36,39,40].
Interestingly, in a recent prospective trial examining aFXa levels after injection of standard enoxaparin dosages (40 mg once daily) in critically ill patients, our working group observed a significant correlation between aFXa levels and the degree of multiple organ dysfunction syndrome as well as body mass index [14]. In the present study, however, we could not find such a relationship (univariate analysis: body mass index, p = 0.322; multiple organ dysfunction syndrome, p = 0.988). On the other hand, aFXa levels after enoxaparin injection did not correlate with renal function in the former study [14]. Different pharmacological characteristics of single LMWHs have been reported (e.g. molecular weight distribution, aFXa/antiFactor IIa activity), and could therefore explain differences between enoxaparin and certoparin. When compared with enoxaparin, however, certoparin seems to be inferior for attaining adequate antithrombotic aFXa levels 4 h after injection of standard dosages (56.5% versus 28%, p = 0.008; Fisher's exact test).
When interpreting the results of this study, some important limitations must be considered. Assessing the antithrombotic potency of LMWHs by measuring exclusively aFXa activity may underestimate the antithrombotic potency of certoparin by omitting other potential anticoagulatory effects, such as inhibition of thrombin generation, increase of fibrinolytic activity or tissue factor pathway inhibitor, by which LMWH may further influence hemostasis [41]. Therefore, it also cannot be excluded that increasing the absolute dosage of certoparin might result in more bleeding complications. Although no certoparin-associated hemorrhage was observed in this study, the protocol was underpowered with respect to reliably assessing clinical endpoints such as thrombotic or hemorrhagic complications.
Conclusion
Standard dosages of certoparin of 3,000 IU given once or twice daily are ineffective for attaining recommended antithrombotic aFXa levels of 0.1 to 0.3 IU/ml in critically ill patients. Low antithrombin levels before certoparin administration were independently associated with low aFXa levels 4 h after injection of certoparin. Renal function and vasopressor therapy may further influence the effectiveness of certoparin in ensuring adequate antithrombotic prophylaxis in critically ill patients.
Key messages
• Standard certoparin dosages are ineffective for attaining recommended antithrombotic aFXa levels in critically ill patients.
• Low antithrombin levels are associated with low aFXa levels during certoparin prophylaxis.
• Renal function and vasopressor therapy may further influence the effectiveness of certoparin.
Abbreviations
aFXa = antifactor Xa; LMWH = low molecular weight heparin.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
SJ conceived of the study protocol, participated in its design and coordination, carried out bedside sampling and documentation, and helped to draft the manuscript. VM and GL participated in the design of the study and its coordination, and carried out bedside sampling and documentation. DF conceived of the study, helped to perform statistical analysis, and contributed to the draft of the manuscript. AJM conceived of the study protocol and participated in its design and coordination. BEF participated in the study design and its coordination. IL participated in the study design and its coordination. WRH conceived of the study protocol, participated in its design, and helped to draft the manuscript. HU performed the power analysis and the statistical analysis of the data. WS conceived of the study protocol, participated in its design and coordination, and helped to draft the manuscript. MWD conceived of the study protocol, participated in its design and coordination, performed the statistical analysis, and drafted the manuscript. All authors read and approved the final version of the manuscript.
Figures and Tables
Figure 1 Percentage of patients within recommended antithrombotic range after (a) 1 × 3,000 IU/d, and (b) 2 × 3,000 IU/d certoparin.
Table 1 Demographic data of study patients
Total 1 × 3,000 IU/d 2 × 3,000 IU/d P-value
n 62 32 30
Sex (male) 45/62 (72.6%) 20/32 (62.5%) 25/30 (83.3%) 0.09
Age (years) 63 ± 12 60 ± 14 66 ± 10 0.038a
BMI 25 ± 4.2 25 ± 5 25 ± 3.3 0.924
Admission diagnoses (n/%) 0.17
Multiple trauma 10/62 (16.1) 5/32 (15.5) 5/30 (16.6)
Pulmonary disease 4/62 (6.4) 1/32 (3.1) 3/30 (10)
Cardiac disease 22/62 (35.5) 9/32 (39.1) 13/30 (43.3)
Infectious disease 5/62 (8.1) 3/32 (9.4) 2/30 (6.6)
Neoplasm 12/62 (19.4) 9/32 (39.1) 3/30 (10)
Orthopedic disease 3/62 (4.8) 2/32 (6.3) 1/30 (3.3)
Other 6/62 (9.7) 3/32 (9.4) 3/30 (10)
CV drug requirement (n/%) 25/62 (40.3) 13/32 (40.6) 12/30 (40) 1
Dopamine 6/62 (9.7) 5/32 (15.5) 1/30 (3.3)
Phenylephrine 16/62 (25.8) 10/32 (31.2) 6/30 (20)
Norepinephrine 7/62 (11.3) 2/32 (6.25) 5/30 (16.6)
Adrenaline 3/62 (4.8) 1/32 (3.1) 2/30 (6.6)
Vasopressin 2/62 (3.2) 1/32 (3.1) 1/30 (3.3)
Milrinone 13/62 (21) 5/32 (15.5) 8/30 (26.6)
MODS (pts) 4.1 ± 2.1 4.1 ± 2.4 4.2 ± 1.8 0.801
Length of ICU stay (days) 12 ± 11 10 ± 12 13 ± 11 0.301
ICU mortality 2/62 (3.2%) 1/32 (3.1%) 1/30 (3.3%) 1
aSignificant difference between group 1 × 3,000 IU/d and 2 × 3,000 IU/d. Data are given as mean values ± standard deviation, if not indicated otherwise. BMI, body mass index; CV, cardiovascular; ICU, intensive care unit; MODS, multiple organ dysfunction syndrome.
Table 2 Laboratory results during certoparin therapy in all study patients (n = 62)
Baseline 12 h 24 h P-value
Prothrombin (%) 87 ± 14 88 ± 13 90 ± 14 0.240
aPTT (sec) 39 ± 9 38 ± 7 38 ± 7 0.183
Antithrombin (%) 73 ± 17 75 ± 17 77 ± 17 0.006a
Fibrinogen (mg/dl) 497 ± 198 529 ± 201 551 ± 211 <0.001a
Hemoglobin (g/dl) 10.2 ± 1.3 10.3 ± 1.3 10 ± 1.1 0.221
Platelets (g/l) 189 ± 103 183 ± 96 199 ± 119 0.055
Serum creatinine (mg/dl) 1.35 ± 0.85 - 1.41 ± 0.95 0.536
Serum urea (mg/dl) 67 ± 40 - 72 ± 42 0.072
a, significant time effect.
PT, Prothrombin Time; aPTT, activated Partial Thromboplastin Time; AT, Antithrombin; -, not measured,
Data are given as mean values ± SD.
Table 3 Independent risk factors for aFXa <0.1 U/ml at 4 hours after injection of standard certoparin dosages
Binary model Bivariate model
OR 95% CI P-value Patients below antithrombotic range Patients within antithrombotic range P-value
Vasopressor requirement (n/%) 0.348 0.07–1.66 0.185 18/38 (47.4) 5/24 (20.8) 0.038
Antithrombin at baseline (%) 0.910 0.86–0.97 0.002 66 ± 11 86 ± 18 <0.001
Serum creatinine at baseline (mg/dl) 0.302 0.96–1.01 0.302 1.19 ± 0.57 1.7 ± 1.16 0.027
Serum urea at baseline (mg/dl) 0.630 0.17–2.39 0.497 57 ± 34 88 ± 44 0.004
Data are given as mean values ± standard deviation, if not indicated otherwise. aFXa, antiFactor Xa activity; CI, confidence interval; OR, odds ratio.
==== Refs
Attia J Ray JG Cook DJ Douketis J Ginsberg JS Geerts WH Deep vein thrombosis and its prevention in critically ill adults Arch Intern Med 2001 161 1268 1279 11371254 10.1001/archinte.161.10.1268
Hoyt DB Swegle JR Deep venous thrombosis in the surgical intensive care unit Surg Clin North Am 1991 71 811 830 1862471
Jain M Schmidt GA Venous thromboembolism: prevention and prophylaxis Semin Respir Crit Care Med 1997 18 79 90
Cook D Attia J Weaver B McDonald E Meade M Crowther M Venous thrombotic disease: an observational study in medical-surgical intensive care unit patients J Crit Care 2000 15 127 132 11138871 10.1053/jcrc.2000.19224
Saibeni S Spina L Vecchi M Exploring the relationships between inflammatory response and coagulation cascade in inflammatory bowel disease Eur Rev Med Pharmacol Sci 2004 8 205 208 15638231
Trzeciak S Dellinger RP Other supportive therapies in sepsis: An evidence-based review Crit Care Med 2004 32 11 Suppl S571 S577 15542966 10.1097/01.CCM.0000145915.53268.A8
Yoo HH de Paiva SA Silveira LV Queluz TT Logistic regression analysis of potential prognostic factors for pulmonary thromboembolism Chest 2003 123 813 821 12628883 10.1378/chest.123.3.813
Dellinger RP Carlet JM Masur H Gerlach H Calandra T Cohen J Gea-Banacloche J Keh D Marshall JC Parker MM Surviving Sepsis Campaign guidelines for management of severe sepsis and septic shock Crit Care Med 2004 32 858 873 15090974 10.1097/01.CCM.0000117317.18092.E4
Davidson BL Risk assessment and prophylaxis of venous thromboembolism in acutely and/or critically ill patients Haemostasis 2000 30 Suppl 2 77 81 11251346 10.1159/000054168
Bussey H Francis JL Heparin overview and issues Pharmacotherapy 2004 24 103S 107S 15334855 10.1592/phco.24.12.103S.36109
Dobesh PP Economics of unfractionated heparin: beyond acquisition cost Pharmacotherapy 2004 24 161S 164S 15334863 10.1592/phco.24.1.105.34811
Michota FA Venous thromboembolism prophylaxis in the medically ill patient Clin Chest Med 2003 24 93 101 12685058 10.1016/S0272-5231(02)00078-3
Geerts W Cook D Selby R Etchells E Venous thromboembolism and its prevention in critical care J Crit Care 2002 17 95 104 12096372 10.1053/jcrc.2002.33941
Mayr AJ Dunser M Jochberger S Fries D Klingler A Joannidis M Hasibeder W Schobersberger W Antifactor Xa activity in intensive care patients receiving thromboembolic prophylaxis with standard doses of enoxaparin Thromb Res 2002 105 201 204 11927124 10.1016/S0049-3848(02)00028-2
Morris TA Heparin and low molecular weight heparin: background and pharmacology Clin Chest Med 2003 24 39 47 12685055 10.1016/S0272-5231(02)00053-9
Hirsh J Ofosu FA Levine M Verstraete M, Vermylen J, Lijnen R, Leuven AJ The development of low molecular weight heparins for clinical use Thrombosis and Haemostasis 1987 Leuven: International Society on Thrombosis and Haemostasis and Leuven University Press 328 348
Frydman AM Bara L Le Roux Y Woler M Chauliac F Samama MM The antithrombotic activity and pharmacokinetics of enoxaparine, a low molecular weight heparin, in humans given single subcutaneous doses of 20 to 80 mg J Clin Pharmacol 1988 28 609 618 2851016
Bara L Planes A Samama MM Occurrence of thrombosis and haemorrhage, relationship with anti-Xa, anti-IIa activities, and D-dimer plasma levels in patients receiving a low molecular weight heparin, enoxaparin or tinzaparin, to prevent deep vein thrombosis after hip surgery Br J Haematol 1999 104 230 240 10050702 10.1046/j.1365-2141.1999.01153.x
Buchanan MR Boneu B Ofosu F Hirsh J The relative importance of thrombin inhibition and factor Xa inhibition to the antithrombotic effects of heparin Blood 1985 65 198 201 3965047
Fernandez F N'Guyen P Van Ryn J Ofosu FA Hirsh J Buchanan MR Hemorrhagic doses of heparin and other glycosaminoglycans induce of platelet effect Thromb Res 1986 43 491 495 3764805 10.1016/0049-3848(86)90094-0
Levine MN Planes A Hirsh J Goodyear M Vochelle N Gent M The relationship between anti-factor Xa level and clinical outcome in patients receiving enoxaparine low molecular weight heparin to prevent deep vein thrombosis after hip replacement Thromb Haemost 1989 62 940 944 2556813
Koller M Schoch U Buchmann P Largiader F von Felten A Frick PG Low molecular weight heparin (KABI 2165) as thromboprophylaxis in elective visceral surgery. A randomized, double-blind study versus unfractionated heparin Thromb Haemost 1986 56 243 246 3551180
Bergqvist D Burmark US Frisell J Hallbook T Lindblad B Risberg B Torngren S Wallin G Low molecular weight heparin once daily compared with conventional low-dose heparin twice daily. A prospective double-blind multicentre trial on prevention of postoperative thrombosis Br J Surg 1986 73 204 208 3512031
Bara L Leizorovicz A Picolet H Samama M Correlation between anti-Xa and occurrence of thrombosis and haemorrhage in post-surgical patients treated with either Logiparin (LMWH) or unfractionated heparin. Post-surgery Logiparin Study Group Thromb Res 1992 65 641 650 1319619 10.1016/0049-3848(92)90213-T
Goris RJ te Boekhorst TP Nuytinck JK Gimbrere JS Multiple-organ failure. Generalized autodestructive inflammation? Arch Surg 1985 120 1109 1115 4038052
Janni W Bergauer F Rjosk D Lohscheidt K Hagena FW [Prospective randomized study comparing the effectiveness and tolerance of various low-molecular-weight heparins in high risk patients] Zentralbl Chir 2001 126 32 38 11227291 10.1055/s-2001-11725
Adolf J Fritsche HM Haas S Hennig FF Horbach T Kastl S Koppenhagen K Michaelis HC Rhamanzadeh R Summa W Comparison of 3,000 IU aXa of the low molecular weight heparin certoparin with 5,000 IU aXa in prevention of deep vein thrombosis after total hip replacement. German Thrombosis Study Group Int Angiol 1999 18 122 126 10424367
Kirchmaier CM Wolf H Schafer H Ehlers B Breddin HK Efficacy of a low molecular weight heparin administered intravenously or subcutaneously in comparison with intravenous unfractionated heparin in the treatment of deep venous thrombosis. Certoparin-Study Group Int Angiol 1998 17 135 145 9821025
Kleindienst A Harvey HB Mater E Bronst J Flack J Herenz K Haupt WF Schon R Early antithrombotic prophylaxis with low molecular weight heparin in neurosurgery Acta Neurochir (Wien) 2003 145 1085 1091 14663565 10.1007/s00701-003-0142-y
Power BM Forbes AM van Heerden PV Ilett KF Pharmacokinetics of drugs used in critically ill adults Clin Pharmacokinet 1998 34 25 56 9474472
De Paepe P Belpaire FM Buylaert WA Pharmacokinetic and pharmacodynamic considerations when treating patients with sepsis and septic shock Clin Pharmacokinet 2002 41 1135 1151 12405864
Park GR Pharmacokinetics and pharmacodynamics in the critically ill patient Xenobiotica 1993 23 1195 1230 8310706
Abildgaard U Highly purified antithrombin 3 with heparin cofactor activity prepared by disc electrophoresis Scand J Clin Lab Invest 1968 21 89 91 5637480
Hirsh J Dalen J Guyatt G The sixth (2000) ACCP guidelines for antithrombotic therapy for prevention and treatment of thrombosis. American College of Chest Physicians Chest 2001 119 1 Suppl 1S 2S 11157638 10.1378/chest.119.1_suppl.1S
Wilson RF Farag A Mammen EF Fujii Y Sepsis and antithrombin III, prekallikrein, and fibronectin levels in surgical patients Am Surg 1989 55 450 456 2742228
Priglinger U Delle Karth G Geppert A Joukhadar C Graf S Berger R Hulsmann M Spitzauer S Pabinger I Heinz G Prophylactic anticoagulation with enoxaparin: Is the subcutaneous route appropriate in the critically ill? Crit Care Med 2003 31 1405 1409 12771610 10.1097/01.CCM.0000059725.60509.A0
Dorffler-Melly J de Jonge E Pont AC Meijers J Vroom MB Buller HR Levi M Bioavailability of subcutaneous low-molecular-weight heparin to patients on vasopressors Lancet 2002 359 849 850 11897286 10.1016/S0140-6736(02)07920-5
Samama MM Gerotziafas GT Comparative pharmacokinetics of LMWHs Semin Thromb Hemost 2000 26 Suppl 1 31 38 11011804 10.1055/s-2000-9497
Chow SL Zammit K West K Dannenhoffer M Lopez-Candales A Correlation of antifactor Xa concentrations with renal function in patients on enoxaparin J Clin Pharmacol 2003 43 586 590 12817521 10.1177/0091270003043006004
Sanderink GJ Guimart CG Ozoux ML Jariwala NU Shukla UA Boutouyrie BX Pharmacokinetics and pharmacodynamics of the prophylactic dose of enoxaparin once daily over 4 days in patients with renal impairment Thromb Res 2002 105 225 231 11927128 10.1016/S0049-3848(02)00031-2
The DVTENOX Study Group Markers of hemostatic system activation in acute deep venous thrombosis-evolution during the first days of heparin treatment Thromb Haemost 1993 70 909 914 8165610
|
16277716
|
PMC1297619
|
CC BY
|
2021-01-04 16:04:56
|
no
|
Crit Care. 2005 Aug 9; 9(5):R541-R548
|
utf-8
|
Crit Care
| 2,005 |
10.1186/cc3792
|
oa_comm
|
==== Front
Crit CareCritical Care1364-85351466-609XBioMed Central London cc37931627771310.1186/cc3793ResearchMedication errors: a prospective cohort study of hand-written and computerised physician order entry in the intensive care unit Shulman Rob [email protected] Mervyn [email protected] John [email protected] Geoff [email protected] ICU Pharmacist, Pharmacy Department, University College London Hospitals, Middlesex Hospital, London, UK2 Consultant, Critical Care Directorate and Professor, Department of Medicine and Wolfson Institute of Biomedical Research, University College London, Middlesex Hospital, London, UK3 Consultant, Intensive Care and Anaesthetics Department, University College London Hospitals, Middlesex Hospital, London, UK4 Consultant and Clinical Director, Critical Care Directorate, University College London Hospitals, Middlesex Hospital, London, UK2005 8 8 2005 9 5 R516 R521 11 4 2005 26 5 2005 12 7 2005 15 7 2005 Copyright © 2005 Shulman 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 study aimed to compare the impact of computerised physician order entry (CPOE) without decision support with hand-written prescribing (HWP) on the frequency, type and outcome of medication errors (MEs) in the intensive care unit.
Methods
Details of MEs were collected before, and at several time points after, the change from HWP to CPOE. The study was conducted in a London teaching hospital's 22-bedded general ICU. The sampling periods were 28 weeks before and 2, 10, 25 and 37 weeks after introduction of CPOE. The unit pharmacist prospectively recorded details of MEs and the total number of drugs prescribed daily during the data collection periods, during the course of his normal chart review.
Results
The total proportion of MEs was significantly lower with CPOE (117 errors from 2429 prescriptions, 4.8%) than with HWP (69 errors from 1036 prescriptions, 6.7%) (p < 0.04). The proportion of errors reduced with time following the introduction of CPOE (p < 0.001). Two errors with CPOE led to patient harm requiring an increase in length of stay and, if administered, three prescriptions with CPOE could potentially have led to permanent harm or death. Differences in the types of error between systems were noted. There was a reduction in major/moderate patient outcomes with CPOE when non-intercepted and intercepted errors were combined (p = 0.01). The mean baseline APACHE II score did not differ significantly between the HWP and the CPOE periods (19.4 versus 20.0, respectively, p = 0.71).
Conclusion
Introduction of CPOE was associated with a reduction in the proportion of MEs and an improvement in the overall patient outcome score (if intercepted errors were included). Moderate and major errors, however, remain a significant concern with CPOE.
See related commentary
==== Body
Introduction
Medication errors (MEs) in the intensive care unit (ICU) are common and can arise from a number of causes. A large study from two tertiary care hospitals reported the error rate was highest in medical ICUs (19.4 per 100 patient days), particularly at the prescribing stage, which accounted for 56% of errors detected [1]. The National Health Service Plan in the UK [2] states that 75% of hospitals should have implemented electronic patient record systems by 2004 in order to make information available at the point of need. Computerised physician order entry (CPOE) without decision support may have advantages over hand-written prescribing (HWP) in terms of standardisation, full audit trail, legibility, use of approved names, specification of key data fields such as route of administration, storage and recall of records.
Although the CPOE system recently installed in our ICU has access to our locally produced on-line formulary (which includes local guidelines), IV guide (advising how to safely administer intravenous medications), drug interactions, contraindications and side effects, these are for information only and decision support capability does not exist. Systems with decision support offer the ability to prevent physicians prescribing either a known allergenic drug or a toxic drug dose [3]. It can flag up drug-drug interactions, force compliance with hospital protocols, and can prevent the prescription of certain drugs, thus implementing evidence based medicine [4] and improving clinical practice [5-7]. This prospective study compares HWP with CPOE without decision support, in several ways. We compare the rates and types of MEs and the potential outcome of intercepted and non-intercepted errors.
Materials and methods
In April 2002, University College Hospitals London ICU introduced the QS 5.6 Clinical Information System (CIS) (GE Healthcare, Anapolis, MD, USA) to the ICU but not on the general wards. The new system was introduced following a program of staff training and HWP was completely changed on a single day. The system used offers a CPOE component but without decision support. Prior to this, hand-written drug charts were used. With both prescribing systems, prescribing was restricted to intensive care medical staff only. To compare both prescribing systems, details of all MEs identified by the ICU clinical pharmacist, in the course of his normal prescription review, were prospectively recorded before the change period and for four reasonably evenly spaced data collection periods after the introduction of the CPOE. The study was designed in advance to collect data over a 70 week time period to enable reliable estimates of error rates. The HWP data collection began on the following dates: 17 September 2001 for 5 days; 24 September 2001 for 4 days. CPOE data collection began on the following dates: 15 April 2002 for 5 days; 10 June 2002 for 2 days; 27 September 2002 for 5 days; and 18 December 2002 for 5 days. CPOE and HWP sample sizes were of different lengths so that an assessment of learning curve could take place. We aimed for each monitoring period to be 5 days. The first two HWP periods were consecutive and thus merged in the results. One period was curtailed due to investigator illness. The ICU medical and nursing staff were unaware that the study was being conducted. Ethical approval was not sought, because at the time audits were not within the remit of the local ethics committee. Prior to introduction of CPOE, local standards of prescribing existed specifying the tenets of good practice, including the avoidance of the use of abbreviations.
An ME was defined to have occurred when a prescribing decision or prescription writing process resulted in either an unintentional significant reduction in the probability of treatment being timely and effective or an unintentional significant increase in the risk of harm when compared with generally accepted practice [8]. During the monitoring period, details of the total number of all prescribed drugs on each day were recorded.
MEs were assessed by type and patient outcome. The type of error was categorised by the pharmacist into groups that best represented the data. A single error could be recorded as several types of error. The total numbers of MEs were also recorded. If a single drug episode was judged to be in error for multiple reasons, it was counted only once for the error rate analysis.
The patient outcome from each error were assigned by the pharmacist and the ICU clinical director, according to an adapted scale [9-11]. Minor errors were classified as those causing no harm or an increase in patient monitoring with no change in vital signs and no harm noted. Moderate errors were classified as those causing an increase in patient monitoring, a change in vital signs but without associated harm or a need for treatment or increased length of stay. Major errors were categorised as those causing permanent harm or death. In this study, intercepted errors (e.g. where an incorrect dose of a drug was prescribed but not administered) were separated from non-intercepted errors (where the patient received the drug). The intercepted errors were scored separately on the basis of their possible impact on the patient, if the prescription had been administered as prescribed.
The chi squared test for trend was used to test whether there was a learning effect over time with CPOE. A chi squared test was used to test for the error rates and outcome comparisons. A two tailed t test was used to compare means of APACHE II score for the HWP and CPOE periods. For this test, as the Levene's test was not significant, equal variance was assumed.
Results
The mean Acute Physiology and Chronic Health Evaluation (APACHE) II scores for the HWP (19.4, standard deviation 9.5, n = 56) and CPOE (20.0, standard deviation 8.0, n = 99) periods were not significantly different (p = 0.71). In the study, 134 drug charts with 1036 prescriptions were reviewed in the HWP group and 253 charts with 2429 prescriptions were assessed in the CPOE group. The proportion of MEs for each data collection period are shown in Fig. 1. The proportion of MEs before CPOE was 6.7% (69 errors from 1036 prescriptions) and 4.8% after CPOE introduction (117 errors from 2429 prescriptions) (p < 0.04). Thus, the reduction in the proportion of MEs following the introduction of CPOE was statistically significant. The proportion of MEs with CPOE varied over time after its introduction (p < 0.001). Evidence also indicated the strong linear trend of a declining proportion of MEs over time (p < 0.001). The types of error from the two systems are listed in Table 1. CPOE appeared to be associated with a high number of dosing errors, omission of the required drug and the prescriber's signature. A number of hand-written prescriptions were missing key details, for example, dose, units or frequency. Several incidences were noted with CPOE in which a drug was not prescribed; for example, caspofungin was omitted when a patient previously established on this drug was admitted to the ICU. Although we did not prospectively look for all missed prescriptions, standard care was for the pharmacist to review admissions and note discrepancies between ward and ICU prescriptions. This error occurred during the CPOE prescribing period.
The patient outcome scores are given in Tables 2 and 3. Most of the errors were minor in outcome, although two non-intercepted errors with CPOE led to an increased length of stay or increased monitoring. In the first case, an anuric patient on haemofiltration was prescribed and administered gentamicin 500 mg, which resulted in prolonged toxic levels. In the second case, a unique problem to CPOE occurred when a loading dose of phenytoin was not administered because a stage of prescription activation was not correctly carried out; the computer-generated order for the nurse to administer the drug was not triggered due to poor prescribing practice, leading to the dose being omitted. This resulted in an extended period before seizure control was achieved.
Three intercepted errors with CPOE could have caused permanent harm or death if they had been administered as prescribed. These intercepted errors were not administered to the patient because either the pharmacist intercepted the prescription before administration or the nurse recognised the error. A potentially fatal intercepted error occurred when diamorphine was prescribed electronically using the pull down menus at a dose of 7 mg/kg instead of 7 mg, which could have lead to a 70 times overdose. In a separate case, amphotericin 180 mg once daily was prescribed, when liposomal amphotericin was intended. The doses of these two products are not interchangeable and the high dose prescribed would have been nephrotoxic. In the third case, vancomycin was prescribed 1 g intravenously daily to a patient in renal failure, when the appropriate dose would have been to give 1 g and then to repeat when the plasma levels fell below 10 mg/L. The dose as prescribed would have lead to nephrotoxicity.
There were many cases of minor errors with CPOE that did not cause patient harm but did increase monitoring. With respect to the non-intercepted errors, there was no significant difference between groups (p = 0.51; Table 3). If we include intercepted errors, however, there is a difference due to the increased rate in the HWP group (p = 0.01; Table 3). It is of note that the only major errors encountered were the three major intercepted errors attributed to CPOE. It appears that CPOE was associated with more minor errors that did not cause patient harm but did increase monitoring.
Discussion
This study was designed to investigate the impact of CPOE, without decision support, on MEs in the critical care setting. The data collected were viewed in terms of proportion of errors, patient outcomes arising from the error and types of error.
The proportion of MEs reduced following the introduction of CPOE. There was also some evidence that a learning curve occurred with CPOE, as the proportion of errors appeared to decline over time. This learning curve could have included improvements made to the system in light of experience, although it is conceivable that the ME rate may have reduced by itself over time. The error rates found were less than those reported in a recent study of prescription errors in UK critical care units [12]. There was no difference in the mean APACHE II score in the HWP and CPOE periods, indicating that it is unlikely that severity of illness differed substantially in the monitored periods.
It was decided to separate the recording of non-intercepted and intercepted errors (where an error was spotted and corrected before having an impact on the patient). The intercepted errors were scored on the basis of what might have occurred if the patient received the medication as prescribed. There was a demonstrated benefit on patient outcome scores with CPOE prescribing when the intercepted errors were combined with the non-intercepted errors. It was reassuring to note that no patients suffered permanent harm or death as a result of any non-intercepted error. Three errors, which all occurred with CPOE, could have led to permanent harm or death had they been administered as prescribed. This CPOE system lacks the ability to effectively deal with drugs with variable dosage regimens such as vancomycin, gentamicin and warfarin. In addition, our impression is that prescribers often prescribed too quickly and made mistakes when using pull-down menus, as seen with the diamorphine error. A lack of product knowledge probably led to the amphotericin error. Prescribers need to develop a thorough, systematic approach to prescribing, similar to that which they employ for diagnosis. This aspect of our findings is in accordance with a recent study that identified that a CPOE system frequently increased the probability of prescribing errors [13].
Most of the errors were defined as 'minor' in outcome and, as such, did not cause the patient harm but, in some cases, may have lead to an increase in monitoring but with no change in vital signs. There were four errors, however, that caused either patient harm or increased monitoring and 34 intercepted errors that could have potentially caused harm had they been administered. The fact that these MEs were rectified before they harmed the patient underlines the value of daily prescription review by an experienced clinical pharmacist [14,15]. In contrast to other views [8], it was decided not to regard abbreviated drug names as errors, because this would have distorted the results in favour of CPOE. In justification of this treatment of the results, no abbreviated drug name led to a patient receiving the wrong drug, but it is regarded as poor prescribing practice as defined by our own prescribing guidelines and national guidelines [16]. CPOE effectively eradicated the use of abbreviations.
The study was not designed or powered to identify differences in the types of errors under the two systems. Future work should be designed to focus on these differences. Omission of key prescription details such as dose, units, frequency and signatures appeared to be much reduced with CPOE, as the computer program did not permit drug entry with missing key data entry fields. Dose errors were still prevalent with CPOE, however, as a result of physicians choosing the wrong drug template, selecting from multiple options, or as a consequence of constructing their own drug prescriptions using pull down menus.
There were also many missed prescribers' signatures with CPOE. This did not affect the patient but, in these cases, nurses administered medication without a legally valid physician order. Although an absent 'signature' with CPOE was regarded as an error, the audit facility of the Clinical Information System did record who prescribed the drug. There were several cases where necessary drugs were not prescribed with CPOE; this was probably not related specifically to the prescribing system.
The categories described were specific to the setting and systems, thus a general taxonomy of medication errors [17] was not used as it was considered that this did not adequately characterise the errors. The categories used here specifically describe the event and general taxonomies were considered to be too broad to provide a specific and useful description of the episode.
During the data collection period, key staff such as consultants, senior nurses and the pharmacist remained the same, so this did not influence the results. Pharmacist attendance at ward rounds has been associated with a reduction in adverse events [15]. In this study the pharmacist attended the ward round throughout the study. No other significant organisational changes occurred during the study period. The only possible changes were the junior medical staff who did change during the study and this may have affected the results. Ideally, the impact of this could be minimised by sampling over a longer period and more frequently, but this was beyond the scope and resources of this study. Alternatively, we could have statistically adjusted for experience level, although this is a difficult issue and has not been attempted by other researchers. Furthermore, the MEs recorded were all proactively identified from the daily pharmacist prescription chart review, and thus did not rely on the notoriously low reporting of multi-disciplinary adverse incident reports. Patient outcome was assessed by the pharmacist and clinical director, who were not blinded to the prescribing system; this could have introduced the potential for bias in the results and is a limitation of the study.
Medical errors are among the leading causes of death in the United States. In its highly publicised report, the Institute of Medicine estimates that between 44,000 and 98,000 Americans die as a result of medical errors each year, with the majority of these errors being preventable [18]. MEs are the leading type of medical error [3]. Previously, in a setting that included general wards and ICUs, a similar type of CPOE was associated with a halving of the rate of non-intercepted MEs [19]; ours is the first study identified that investigates the impact of CPOE on MEs solely in an adult ICU. CPOE is already the subject of considerable interest [20] and has already shown benefits in paediatric medicine [21-23]. A systematic review of the impact of clinical decision support systems (CDSS) [6] has demonstrated statistically significant improvements in antibiotic-associated MEs or adverse drug events and an improvement in theophylline-associated MEs, while several studies have shown non-significant results. CDSS is worthy of future study in the adult ICU in order to build on the experience gained from the limited CDSS system used in a mixed ICU and general ward setting [19].
Conclusion
These results indicate that the introduction of CPOE, without decision support, in our ICU was associated with a reduced proportion of MEs and improved patient outcome after an error (when non-intercepted and intercepted errors were combined). The limitations of this study and the potential for bias discussed previously must be borne in mind when interpreting these results.
Some of the types of errors appeared to change with CPOE; of particular concern was the finding that all three of the major intercepted errors arose with CPOE. In our study, CPOE clearly reduced the incidence of less major errors but the more serious errors are a genuine concern with this CPOE system. This is not an isolated finding [13] and should be noted by clinical directors as they review the need for CPOE on their units. As clinicians embrace CPOE, they should not make the assumption that CPOE removes errors; in fact, different types of errors emerge. We cannot abdicate our responsibility for ensuring that a prescription is correct in favour of a computer.
Key messages
• This study is the first to compare CPOE and HWP solely in the ICU.
• CPOE was associated with a reduced proportion of MEs compared with HWP and this lowered over time.
• When intercepted and non-intercepted errors were combined, CPOE was associated with an improvement in the error outcome scoring compared to HWP; however, the three intercepted errors that could have caused permanent harm or death all occurred with CPOE.
• The types of error appeared to change with the introduction of CPOE.
• The introduction of CPOE without decision support eliminated many minor types of error but introduced new types of error that may be more serious.
Abbreviations
APACHE = Acute Physiology and Chronic Health Evaluation; CDSS = clinical decision support systems; CPOE = computerised physician order entry; HWP = hand-written prescribing; ICU = intensive care unit; ME = medication error.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
RS conceived the study, collected the data, analysed the results and drafted the article. MS was involved in critically revising the draft. JG made substantial contributions to the data analysis. GB was substantially involved in the analysis, interpretation and drafting the manuscript.
Acknowledgements
To the Medical Statistics Unit, Research and Development Directorate, UCL Hospitals and to Steve Batson for providing the APACHE II data.
Figures and Tables
Figure 1 Proportion of medication errors before and after implementation of computerised physician order entry (CPOE) using the Clinical Information System with 95% confidence intervals. Hand-written prescribing (HWP) data collection began on the following dates: 17 September 2001 for 5 days; 24 September 2001 for 4 days (merged with the previous period). CPOE data collection began on the following dates: 15 April 2002 for 5 days; 10 June 2002 for 2 days; 27 September 2002 for 5 days; and 18 December 2002 for 5 days.
Table 1 Types of medication errors before and after implementing CPOE
Error type HWP (no. of errors and % of total errors)a CPOE (no. of errors and % of total errors)a
Drug prescribed on incorrect drug chart section (e.g. continuous IV infusion prescribed on 'when required' part of drug chart) 2 (2.8%) 1 (0.9%)
Drug needed but not given as not prescribed properly 3 (4.2%) 5 (4.3%)
Inappropriate/inadequate additional information on prescription to adequately administer the drug appropriately 8 (11.3%) 12 (10.3%)
Dose/units/frequency omitted on prescription 22 (31%) 1 (0.9%)
Prescription not signed or change not signed/dated 10 (14.1%) 39 (33.3%)
Still wrong next day after pharmacist recommended appropriate correction that was agreed with doctor 0 (0%) 3 (2.6%)
Dose error 12 (16.9%) 31 (26.5%)
Wrong drug prescribed 3 (4.2%) 6 (5.1%)
Incorrect route/unit 5 (7%) 8 (6.8%)
Formulary not followed without reason 3 (4.2%) 1 (0.9%)
Administration not in accordance with prescription 3 (4.2%) 3 (2.6%)
Required drug not prescribed 0 (0%) 7 (6%)
Total 71/1036 prescriptions 117/2429 prescriptions
aOne episode could be recorded here as being in error for several reasons but was only recorded once in the proportion of error analysis. This explains why the total of hand-written prescribing (HWP) error types stated here is in excess of the total number of errors stated in the results section. CPOE, computerised physician order entry.
Table 2 Error outcome categories
Error category Minor Moderate Major
HWP non-intercepted errors 43 0 0
CPOE non-intercepted errors 93 4 0
HWP intercepted errors 7 19 0
CPOE intercepted errors 2 15 3
CPOE, computerised physician order entry; HWP, hand-written prescribing.
Table 3 Error outcome category analysis
Error category None Minor Moderate/major Total
Non-intercepted errorsa
HWP 993 (95.9%) 43 (4.2%) 0 (0%) 1036
CPOE 2332 (96.0%) 93 (3.8%) 4 (0.2%) 2429
Non-intercepted plus intercepted errorsb
HWP 967 (93.3%) 50 (4.8%) 19 (1.8%) 1036
CPOE 2312 (95.2%) 95 (3.9%) 22 (0.9%) 2429
aNo significant difference with regard to errors between hand-written prescribing (HWP) and computerised physician order entry (CPOE; p = 0.51).
bIf we include intercepted errors, there was a significant difference (p = 0.01) due to increased error rate with HWP.
==== Refs
Bates DW Cullen DJ Laird N Petersen LA Small SD Servi D Laffel G Sweitzer BJ Shea BF Hallisey R Incidence of adverse drug events and potential adverse drug events. Implications for prevention. ADE Prevention Study Group JAMA 1995 274 29 34 7791255 10.1001/jama.274.1.29
Department of Health The NHS Plan, a Plan for Investment, a Plan for Reform 2000 UK 13-7-0003
Leape LL Error in medicine JAMA 1994 272 1851 1857 7503827 10.1001/jama.272.23.1851
Cook D Evidence-based critical care medicine: a potential tool for change New Horiz 1998 6 20 25 9508254
Grimshaw JM Russell IT Effect of clinical guidelines on medical practice: a systemic review of rigorous evaluations Lancet 1993 342 1317 1322 7901634 10.1016/0140-6736(93)92244-N
Kaushal R Shojania KG Bates DW Effects of computerized physician order entry and clinical decision support systems on medication safety: a systematic review Arch Intern Med 2003 163 1409 1416 12824090 10.1001/archinte.163.12.1409
Morris AH Computerized protocols and bedside decision support Crit Care Clin 1999 15 523 545 10442261 10.1016/S0749-0704(05)70069-5
Dean B Barber N Schachter M What is a prescribing error? Qual Health Care 2000 9 232 237 11101708 10.1136/qhc.9.4.232
Dean BS Barber ND A validated, reliable method of scoring the severity of medication errors Am J Health Syst Pharm 1999 56 57 62 10048880
Dean B Barber N Validity and reliability of observational methods for studying medication administration errors Am J Health Syst Pharm 2001 58 54 59 11194136
ASHP Guidelines on preventing medication errors in hospitals Am J Health Pharm 1993 50 305 314
Ridley SA Booth SA Thompson CM The Intensive Care Society's Working Group on Adverse Incidents Prescription errors in UK critical care units Anaesthesia 2004 59 1193 1200 15549978 10.1111/j.1365-2044.2004.03969.x
Koppel R Metlay JP Cohen A Abaluck B Localio AR Kimmel SE Strom BL Role of computerized physician order entry systems in facilitating medication errors. [see comment] JAMA 2005 293 1197 1203 15755942 10.1001/jama.293.10.1197
Kane SL Weber RJ Dasta JF The impact of critical care pharmacists on enhancing patient outcomes Intensive Care Med 2003 29 691 698 12665997
Leape LL Cullen DJ Clapp MD Burdick E Demonaco HJ Erickson JI Bates DW Pharmacist participation on physician rounds and adverse drug events in the intensive care unit JAMA 1999 282 267 270 10422996 10.1001/jama.282.3.267
British Medical Association and the Royal Pharmaceutical Society of Great Britain British National Formulary 2003 March London
National Coordinating Council for Medication Error Taxonomy of Medication Errors Reporting and Prevention 2004
Committee on Quality of Health Care in America, Institute of Medicine Kohn LT, Corrigan JM, Donaldson M To Err Is Human: Building a Safer Health System 2000 Washington, DC: National Academic Press
Bates DW Leape LL Cullen DJ Laird N Petersen LA Teich JM Burdick E Hickey M Kleefield S Shea B Effect of computerized physician order entry and a team intervention on prevention of serious medication errors JAMA 1998 280 1311 1316 9794308 10.1001/jama.280.15.1311
Kuperman GJ Gibson RF Computer physician order entry: benefits, costs, and issues Ann Intern Med 2003 139 31 39 12834316
Cordero L Kuehn L Kumar RR Mekhjian HS Impact of computerized physician order entry on clinical practice in a newborn intensive care unit J Perinatol 2004 24 88 93 14872207 10.1038/sj.jp.7211000
King WJ Paice N Rangrej J Forestell GJ Swartz R The effect of computerized physician order entry on medication errors and adverse drug events in pediatric inpatients Pediatrics 2003 112 506 509 12949274 10.1542/peds.112.3.506
Potts AL Barr FE Gregory DF Wright L Patel NR Computerized physician order entry and medication errors in a pediatric critical care unit Pediatrics 2004 113 59 63 14702449 10.1542/peds.113.1.59
|
16277713
|
PMC1297620
|
CC BY
|
2021-01-04 16:04:56
|
no
|
Crit Care. 2005 Aug 8; 9(5):R516-R521
|
utf-8
|
Crit Care
| 2,005 |
10.1186/cc3793
|
oa_comm
|
==== Front
Crit CareCritical Care1364-85351466-609XBioMed Central London cc37941627771710.1186/cc3794ResearchDoes cardiac surgery in newborn infants compromise blood cell reactivity to endotoxin? Schumacher Kathrin [email protected] Stefanie [email protected] Jaime F [email protected] Bernuth Götz [email protected] Jean [email protected] Marie-Christine [email protected] Fellow in pediatrics, Department of Pediatric Cardiology, Aachen University, Aachen, Germany2 Fellow in internal medicine, Department of Pediatric Cardiology, Aachen University, Aachen, Germany3 Head of department, Department of Pediatric Cardiac Surgery, Aachen University, Aachen, Germany4 Former head of department, Department of Pediatric Cardiology, Aachen University, Aachen, Germany5 Director, Department of Immunology, University Hospital Brugmann and Saint-Pierre, Free University of Brussels, Brussels, Belgium6 Head of department, Department of Pediatric Cardiology, Aachen University, Aachen, Germany2005 9 8 2005 9 5 R549 R555 20 4 2005 31 5 2005 13 7 2005 15 7 2005 Copyright © 2005 Schumacher 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
Neonatal cardiac surgery is associated with a systemic inflammatory reaction that might compromise the reactivity of blood cells against an inflammatory stimulus. Our prospective study was aimed at testing this hypothesis.
Methods
We investigated 17 newborn infants with transposition of the great arteries undergoing arterial switch operation. Ex vivo production of the pro-inflammatory cytokine tumor necrosis factor-α (TNF-α), of the regulator of the acute-phase response IL-6, and of the natural anti-inflammatory cytokine IL-10 were measured by enzyme-linked immunosorbent assay in the cell culture supernatant after whole blood stimulation by the endotoxin lipopolysaccharide before, 5 and 10 days after the operation. Results were analyzed with respect to postoperative morbidity.
Results
The ex vivo production of TNF-α and IL-6 was significantly decreased (P < 0.001 and P < 0.002, respectively), whereas ex vivo production of IL-10 tended to be lower 5 days after the operation in comparison with preoperative values (P < 0.1). Ex vivo production of all cytokines reached preoperative values 10 days after cardiac surgery. Preoperative ex vivo production of IL-6 was inversely correlated with the postoperative oxygenation index 4 hours and 24 hours after the operation (P < 0.02). In contrast, postoperative ex vivo production of cytokines did not correlate with postoperative morbidity.
Conclusion
Our results show that cardiac surgery in newborn infants is associated with a transient but significant decrease in the ex vivo production of the pro-inflammatory cytokines TNF-α and IL-6 together with a less pronounced decrease in IL-10 production. This might indicate a transient postoperative anti-inflammatory shift of the cytokine balance in this age group. Our results suggest that higher preoperative ex vivo production of IL-6 is associated with a higher risk for postoperative pulmonary dysfunction.
==== Body
Introduction
Cardiac surgery is associated with a systemic inflammatory reaction comprising activation of the complement system, stimulation of leukocytes, synthesis of cytokines, and increased interactions between leukocytes and endothelium [1,2]. In children, contact activation, ischemia/reperfusion injury and endotoxin released from the gut [3,4] are thought to be the major inductors of pro-inflammatory cytokines such as tumor necrosis factor-α (TNF-α) and IL-6 in the cardiac surgery setting. In newborn infants, morbidity after cardiac surgery is related to the importance of the intra-operative production of pro-inflammatory cytokines such as IL-6, as we have shown previously [5].
NF-κB is the main transcription factor of many inflammatory genes, such as that encoding TNF-α [6]. TNF-α induces secondary mediators of inflammation such as IL-6, the principal regulator of the acute-phase response [7]. IL-10 is an anti-inflammatory cytokine that strongly inhibits the synthesis of pro-inflammatory cytokines at the transcriptional level by controlling the degradation of the inhibitory protein of NF-κB, IκB, and thereby the nuclear translocation of NF-κB [8]. IL-10 has a central role in the control and termination of systemic inflammation. Although IL-10 is thought to have a protective role in the early postoperative period, the maintenance of normal postoperative organ function is likely to depend on an adequate balance between the production of pro-inflammatory and anti-inflammatory cytokines [9]. It has been suggested that the overproduction of IL-10 after severe injury might be associated with a hyporesponsiveness to lipopolysaccharide (LPS) that carries a higher risk for infections [10].
The ex vivo production of cytokines by whole blood is a widely accepted method of evaluating the reactivity of immunoreactive and inflammatory cells and their potential for inflammatory responses [11]. In this study, we tested the hypothesis that neonatal cardiac surgery would influence the ex vivo production of cytokines.
Materials and methods
Patients
After approval by the Human Ethical Committee of the Aachen University Hospital as well as written consent from the parents, 17 consecutive newborn infants aged 2 to 13 days (median 8 days) were included in this study. To ensure homogeneity of the patient group, the inclusion criterion was a simple transposition of the great arteries, suitable for an arterial switch operation. All patients received prostaglandin E1 infusion (0.05 μg kg-1 min-1) before the operation, to maintain patency of the ductus arteriosus. Preoperative cardiac catheterization for balloon atrioseptostomy and angiography was performed in 13 patients.
Anesthesia, operative management and postoperative care
Conventional general anesthesia was conducted with diazepam, fentanyl sulfate and pancuronium bromide. Perioperative antibiotic prophylaxis consisted of cefotiam hydrochloride (100 mg kg-1 body weight). Dexamethasone (10 mg m-2 body surface area) was administered immediately before sternotomy.
The standardized neonatal cardiopulmonary bypass (CPB) protocol included a roller pump, a disposable membrane oxygenator and an arterial filter. All patients were operated on under deep hypothermic CPB, as described previously [5]. Epinephrine (adrenaline), dopamine and sodium nitroprusside were administered systemically for weaning the patients from CPB.
Standardized postoperative care was provided. Monitoring included continuous registration of hemodynamic variables, diuresis and blood gases. Inotropic support consisted in all cases of dopamine (5 μg kg-1 min-1) and, if necessary, epinephrine (0.05 to 0.2 μg kg-1 min-1) or dobutamine (5 to 7.5 μg kg-1 min-1) and vasodilatory treatment of sodium nitroprusside (0.5 to 2 μg kg-1 min-1). Diuretics (furosemide, single dosage of 0.1 to 1 mg kg-1) and volume substitution, which consisted of fresh-frozen plasma or human albumin 5%, were administered depending on the hemodynamic variables. Postoperative clinical endpoint variables were mean arterial blood pressure, mean central venous pressure, need for inotropic support, oxygenation index expressed as the ratio of partial arterial oxygen tension to fraction of inspired oxygen, minimal diuresis, maximal serum creatinine and maximal serum glutamate oxaloacetate transaminase values during the first 72 hours after the operation, and duration of inotropic and ventilatory support.
Blood elements
Leukocyte counts were determined by a Cell-Dyn 3700 (Abbott GmbH & Co. KG, Wiesbaden, Germany).
C-reactive protein
C-reactive protein was determined by laser nephelometry. The detection limit of this method is 5 mg dl-1.
Ex vivo stimulation
Whole blood culture was performed as described previously [12]. Blood (1 ml) was withdrawn under sterile conditions from a peripheral vein and was taken in endotoxin-free tubes (Endo tube ET; Chromogenix, Haemochrom Diagnostica GmbH, Essen, Germany) before the operation (median 5 days), as well as 5 and 10 days after operation. The timing of blood samples was dictated by the fact that ex vivo production of TNF-α was reported to be decreased up to the sixth postoperative day in adults undergoing cardiac surgery [13]. Blood was mixed in a 1:10 ratio with RPMI 1640 medium containing L-glutamine and 25 mM Hepes medium (Bio Whittaker Europe, Verviers, Belgium). Cell cultures were stimulated with LPS (LPS for cell culture, Escherichia coli, lot 026.B6:L2654; Sigma, St Louis, MO, USA) at a final concentration of 1 ng ml-1. In control samples, the LPS volume was replaced with cell culture medium. Because it has been shown that ex vivo cytokine production reaches its plateau mainly between 12 and 24 hours after stimulation [14], cell cultures were incubated for 16 hours in a humidified incubator at 37°C in an atmosphere consisting of a mixture of 5% CO2 and 95% air (Heraeus HBB 2472b; Heraeus Instruments GmbH, Hanau, Germany); the supernatant was then separated after centrifugation (2,500 r.p.m. for 3 min) and frozen at -70°C until assay.
Cytokine determination
TNF-α, IL-6 and IL-10 were determined with an immunocytometric assay (Biosource International, Camarillo, CA, USA), in accordance with the manufacturer's recommendations for cell culture supernatant. It is a solid-phase, enzyme-amplified sensitivity immunoassay performed on microtiter plates based on the oligoclonal system in which several monoclonal antibodies directed against distinct epitopes of cytokines are used, permitting a high sensitivity of the assay. The minimal detectable concentrations are 3 pg ml-1 for TNF-α, 2 pg ml-1 for IL-6, and 1 pg ml-1 for IL-10. The ranges covered by the standard curve are 0 to 1,700 pg ml-1 for TNF-α, 0 to 2,100 pg ml-1 for IL-6, and 0 to 1,750 pg ml-1 for IL-10. Samples were diluted accordingly.
Statistical analysis
Results are expressed as means ± SEM. The data were analyzed with the nonparametric paired Wilcoxon rank test. The Spearman rank correlation coefficient was assessed for correlation of independent parameters. P < 0.05 was considered significant.
Results
Clinical results
Operative data and clinical results are summarized in Table 1. Seven of the 17 newborn infants showed early postoperative complications that are summarized in Table 2. Six of the seven patients with complications had a capillary leak syndrome as previously described by our group [15]. One patient developed pneumonia. There was one postoperative death 29 days after operation in a patient having developed thrombosis of the right and of the left persistent superior caval veins.
Leukocyte count
There was no statistical difference between the counts of leukocytes, granulocytes and monocytes measured before the operation, and 5 and 10 days after it (Table 3). Leukocyte counts were not different in patients with or without complications.
C-reactive protein
C-reactive protein (CRP) increased in all patients from 7.94 ± 1.27 mg dl-1 before the operation to 15.7 ± 3.7 mg dl-1 5 days after it. At that time point, CRP values were higher in patients with complications than in those without (23.8 ± 5.5 versus 10.2 ± 4.3 mg dl-1, P = 0.001). The patient with pneumonia had a CRP value of 8 mg dl-1 before the operation and 9 mg dl-1 5 days after the operation, increasing to 50 mg dl-1 12 hours later. CRP values were still elevated in all patients 10 days after the operation (16.6 ± 4.4 mg dl-1), and at that time there was no difference between patients with and without complications. The patient with pneumonia had a CRP value of 8 mg dl-1 at that time.
Ex vivo production of cytokines after LPS stimulation before and after operation
At all time points investigated in this study there was a significant production of TNF-α, IL-6 and IL-10 after stimulation by LPS in comparison with the control sample.
Concentrations of TNF-α and IL-6 in the cell culture supernatant were significantly decreased on day 5, in comparison with preoperative levels (P < 0.001 and P < 0.002, respectively).
Postoperative IL-10 concentrations on day 5 were also reduced compared with the preoperative value, although not significantly (P < 0.1). On the 10th day after the operation, concentrations of TNF-α, IL-6 and IL-10 had returned to their preoperative levels (Figs 1, 2, 3).
Correlation between ex vivo production of cytokines and outcome
In all patients preoperative IL-6 production was inversely correlated with the oxygenation index, as measured 4 and 24 hours after the operation (Spearman correlation coefficient: -0.62; P < 0.02). Figure 4 shows the relationship between preoperative ex vivo IL-6 production and the oxygenation index, as measured 24 hours after the operation. There was no correlation between the ex vivo production of TNF-α and IL-10 and postoperative morbidity, respectively. In particular, the only patient with pneumonia (patient 2 in Table 2) showed ex vivo cytokine production that was in the same range as for all other patients.
Discussion
In previous studies we have shown that neonatal cardiac surgery induces a systemic inflammatory reaction with complement activation, leukocyte stimulation and cytokine synthesis that is associated with postoperative complications such as the capillary leak syndrome and myocardial dysfunction [2,5,15]. In this study we confirm the association between systemic inflammation and postoperative morbidity. Although it has been suggested that, in the setting of cardiac surgery, parenchymatous cells such as cardiomyocytes contribute to the systemic inflammatory reaction by producing cytokines, circulating blood cells, in particular leukocytes, are considered the major source of inflammatory mediators [16,17]. This is supported by previous studies that report a clear association between uncontrolled leukocyte activation and early postoperative morbidity after cardiac surgery in newborn infants and in children [5,15].
The systemic inflammatory reaction induced by cardiac surgery is normally controlled by a natural anti-inflammatory response. Indeed, levels of IL-10 are already increased at the end of the operation and remain substantially elevated for at least 48 hours after the operation [18].
Although the anti-inflammatory response to cardiac surgery is thought to be beneficial with regard to early postoperative organ protection [17], it remains unclear whether it could impair leukocyte reactivity and thereby decrease resistance against infections.
In this study, the reactivity of circulating cells after neonatal cardiac surgery was evaluated by the ex vivo production of pro-inflammatory and anti-inflammatory cytokines after a standardized inflammatory stimulus in a homogenous patient group.
A previous study in older children who had undergone cardiac surgery for various cardiac defects showed decreased ex vivo cytokine production on the morning of the first postoperative day. However, later time points, to document the normalization of cytokine production, were not investigated [19]. One main result of our study is that neonatal cardiac surgery is associated with a transiently decreased ex vivo production of the pro-inflammatory cytokines TNF-α and IL-6, and that this is not related to a decrease in leukocyte count. This indicates impaired reactivity of inflammatory cells. In adults this phenomenon has been reported after cardiac surgery [13], severe injury and sepsis, and defined as hyporesponsiveness to LPS [20,21]. In adults who have undergone cardiac surgery, ex vivo TNF-α production and TNF-α mRNA in whole blood were still lower at the end of the study period, which was 6 days after surgery [13]. We also investigated the ex vivo production of cytokines at a later time point and show a return of TNF-α production to preoperative values 10 days after cardiac surgery. The reason for the transient impairment of leukocyte reactivity in our series could be ascribed to the exhaustion of circulating inflammatory cells due to the massive inflammatory stress due to cardiac surgery and also to the perioperative treatment applied. With this regard, drugs administered before, during and after the operation could have influenced hyporesponsiveness to LPS. Indeed, prostaglandin E1 has been shown to reduce the ex vivo production of TNF-α and IL-1β by adult monocytes [22]. However, in our patients who were all treated with prostaglandin-E1 infusion before the operation, preoperative levels of cytokines measured in the supernatant of the whole blood culture were similar to those after stimulation of cord blood in healthy newborn infants [12]. This suggests a minor effect of prostaglandin-E1 on the ex vivo production of cytokines in our study. In adults, anesthesia and heparin were shown not to influence the ex vivo production of TNF-α [13].
The course of ex vivo IL-10 production after cardiac surgery has so far not been followed for more than 6 hours after CPB. In a previous study, IL-10 production reached its lowest point 2 hours after cardiac surgery in adult patients and returned to preoperative values 6 hours later [10]. Our results, in contrast, show that in newborn infants the ex vivo production of IL-10 was decreased 5 days after the operation, even though not significantly in comparison with preoperative values. This reduction could be the result of negative feedback by IL-10, which inhibits not only pro-inflammatory cytokines but also its own production.
The exact mechanisms leading to hyporesponsiveness to LPS in newborn infants reported here are not yet clear. However, the anti-inflammatory cytokines IL-10 and tissue growth factor-β are thought to be important in its regulation [23]. In a clinical study, adult patients with sepsis or severe trauma showed a reduced expression of the active form of NF-κB [24]. In those who did not survive, the IL-10 plasma levels were inversely correlated to the ratio between the active and inhibitory forms of NF-κB, supporting the view that IL-10 might participate in the induction of LPS hyporesponsiveness by inhibiting cytokine synthesis at or upstream of the transcriptional level.
Newborn infants who have undergone cardiac surgery have been reported to show a higher natural production of IL-10 than older children [25]. For the reasons cited above, this natural anti-inflammatory cytokine imbalance could well have contributed to the hyporesponsiveness to LPS observed in our series. Furthermore, as reported by others in older children operated on with CPB [19], perioperative treatment with dexamethasone could also have contributed to hyporesponsiveness to LPS by inhibiting the activation of NF-κB and thereby the production of pro-inflammatory cytokines [26], as well as by stimulating the production of IL-10 [27,28].
Although a clear association has been demonstrated between hyporesponsiveness to LPS and poor clinical outcome in sepsis [24], we were not able to confirm such an association in our series. One reason for this might be that, in the small group of patients investigated, the overall rate of complications related to inflammation or infection was low.
In contrast, we observed a clear association between the preoperative ex vivo production of IL-6 and postoperative respiratory morbidity. This suggests that a higher preoperative potential for ex vivo production of IL-6 is a risk factor for inflammation-related postoperative complications in newborn infants.
Conclusion
Our results show for the first time that cardiac surgery in newborn infants is associated with a transient but significant decrease in the ex vivo production of the pro-inflammatory cytokines TNF-α and IL-6 together with a less pronounced decrease in IL-10 production. This suggests a postoperative anti-inflammatory shift of the cytokine balance in this age group 5 days after cardiac surgery. A higher preoperative ex vivo production of IL-6 might indicate a higher risk for postoperative pulmonary dysfunction. Further studies will address the question of whether preoperative ex vivo production of IL-6 would be a suitable predictor of postoperative complications in newborn infants with congenital cardiac defects.
Key messages
• Cardiac surgery in newborn infants decreases the reactivity of blood cells to LPS.
• Cardiac surgery in newborn infants might lead to an anti-inflammatory shift of the cytokine balance.
• In this series, postoperative complications related to decreased blood cell reactivity were not observed.
• The higher the ex vivo production of IL-6 before the operation, the worse the postoperative lung function.
• Testing the ex vivo production of IL-6 in newborn infants might help to predict postoperative pulmonary dysfunction.
Abbreviations
CPB = cardiopulmonary bypass; CRP = C-reactive protein; IL = interleukin; LPS = lipopolysaccharide; NF-κB = nuclear factor κB; TNF-α = tumor necrosis factor-α.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
KS performed whole blood cultures, ELISAs, acquisition and statistical analysis of the data, and redaction of the manuscript. SK performed ELISAs, data acquisition and analysis. JFV-J coordinated sample withdrawal and revised the manuscript. GvB drafted the manuscript and revised it critically. JD supervised the blood cultures and ELISAs, study design, data analysis and interpretation. M-CS was responsible for study conception and design, data analysis and interpretation and manuscript preparation and final revision. All authors read and approved the final manuscript.
Figures and Tables
Figure 1 Ex vivo production of tumor necrosis factor-α. Preoperative and postoperative (po) tumor necrosis factor-α (TNF-α) levels in whole blood culture supernatant. Values are expressed as means and SEM (error bars). TNF-α production was significantly increased after stimulation with lipopolysaccharide (LPS; white), in comparison with the unstimulated control (C; black) at all time points. In comparison with preoperative levels, TNF-α production after stimulation with LPS significantly decreased 5 days after operation (P < 0.001) but again reached preoperative levels 10 days after operation.
Figure 2 Ex vivo production of interleukin-6. Preoperative and postoperative (po) interleukin (IL)-6 levels in whole blood culture supernatant. Values are expressed as means and SEM (error bars). IL-6 production was significantly increased after stimulation with lipopolysaccharide (LPS; white), in comparison with the unstimulated control (C; black) at all time points. In comparison with preoperative levels, IL-6 production after stimulation with LPS significantly decreased 5 days after operation (P < 0.002) but again reached preoperative levels 10 days after operation.
Figure 3 Ex vivo production of interleukin-10. Preoperative and postoperative (po) IL-10 levels in whole blood culture supernatant. Values are expressed as means and SEM (error bars). IL-10 production was significantly increased after stimulation with lipopolysaccharide (LPS; white), in comparison with the control (C; black) at all time points. In comparison with preoperative levels, IL-10 production after stimulation with LPS tended to decrease 5 days after operation but again reached preoperative levels 10 days after operation.
Figure 4 Relationship between preoperative production of interleukin-6 (IL-6) and postoperative pulmonary dysfunction. Plot showing the correlation between preoperative IL-6 production after stimulation with lipopolysaccharide (LPS) and the oxygenation index 24 hours after operation (n = 14). Spearman correlation coefficient -0.62; P < 0.02.
Table 1 Clinical and operative data
Variable Value
Age at operation (days) 8 (2–13)
Duration of cardiopulmonary bypass (min) 58 (53–63)
Duration of aortic cross-clamping (min) 62 (54–78)
Mean blood pressure (mmHg)
4 h after operation 65 (48–80)
24 h after operation 53 (47–68)
Diuresis (ml kg-1 h-1)
4 h after operation 7.8 (1.6–17.5)
24 h after operation 7.1 (1–8)
Oxygenation index PaO2/FiO2 (mmHg)
4 h after operation 176.7 (69–283)
24 h after operation 195.5 (63–370)
Aspartate aminotransferase concentration (IU L-1)
4 h after operation 32 (13–66)
24 h after operation 33 (7–162)
Epinephrine dosage (μg kg-1 min-1)
4 h after operation 0.16 (0.02–0.36)
24 h after operation 0.12 (0.02–0.41)
Values are presented as number (n) and range. FiO2, fraction of inspired oxygen; PaO2, partial arterial oxygen tension.
Table 2 Postoperative complications
Patient Complications Time after operation Outcome
1 Cardiac arrest after blood transfusion 4 h Survived
2 Capillary leak syndromea 24 h Survived
Pneumoniab 5 d
3 Capillary leak syndrome 24 h Survived
4 Capillary leak syndrome 24 h Survived
5 Capillary leak syndrome 24 h Survived
6 Capillary leak syndrome 24 h Survived
7 Capillary leak syndrome 24 h Died
Thrombosis of the right and left persistent superior caval veins 10 d
aCapillary leak syndrome was diagnosed in accordance with our definition [15]. bDiagnosis of pneumonia was made on the basis of respiratory insufficiency, a pathological chest X-ray and a secondary increase in C-reactive protein.
Table 3 Preoperative and postoperative leukocyte, granulocyte, monocyte and lymphocyte counts
Cell type Cell count
Before operation 5 d after operation 10 d after operation
Leukocytes (Giga l-1) 12.8 ± 1.2 10.1 ± 0.8 11.9 ± 1.3
Granulocytes (%) 53.2 ± 4.2 55.9 ± 3.1 45.2 ± 4.5
Monocytes (%) 3.7 ± 1.5 3 ± 0.6 9.1 ± 1.6
Lymphocytes (%) 37.4 ± 5.9 33.4 ± 4 37.5 ± 3.8
Values are presented as means ± SEM.
==== Refs
Kirklin JK Westaby S Blackstone EH Kirklin JW Chenoweth DE Pacifico AD Complement and the damaging effects of cardiopulmonary bypass J Thorac Cardiovasc Surg 1983 86 845 857 6606084
Seghaye MC The clinical implications of the systemic inflammatory reaction related to cardiac operations in children Cardiol Young 2003 13 228 239 12903869
Jansen NJG Van Oeveren W Gu YJ Vanvliet MH Eijsman L Wildevuur CRH Endotoxin release and tumor-necrosis-factor formation during cardiopulmonary bypass Ann Thorac Surg 1992 54 744 748 1417233
Casey WF Hauser GJ Hannallah RS Midgley FM Khan WN Circulating endotoxin and tumor-necrosis-factor during pediatric cardiac surgery Crit Care Med 1992 20 1090 1096 1643887
Hovels-Gurich HH Vazquez-Jimenez JF Silvestri A Schumacher K Minkenberg R Duchateau J Messmer BJ von Bernuth G Seghaye MC Production of proinflammatory cytokines and myocardial dysfunction after arterial switch operation in neonates with transposition of the great arteries J Thorac Cardiovasc Surg 2002 124 811 820 12324741 10.1067/mtc.2002.122308
Siebenlist U Franzoso G Brown K Structure, regulation and function of NF-κB Annu Rev Cell Biol 1994 10 405 455 7888182 10.1146/annurev.cb.10.110194.002201
Heinrich PC Castell JV Andus T Interleukin-6 and the acute phase response Biochem J 1990 265 621 636 1689567
Schottelius AJG Mayo MW Sartor RB Badwin AS Jnr Interleukin-10 signaling blocks inhibitor of kappa B kinase activity and nuclear factor kappa B DNA binding J Biol Chem 1999 274 31868 31874 10542212 10.1074/jbc.274.45.31868
Hovels-Gurich HH Schumacher K Vazquez-Jimenez JF Qing M Huffmeier U Buding B Messmer BJ von Bernuth G Seghaye MC Cytokine balance in infants undergoing cardiac operation Ann Thorac Surg 2002 73 601 608 11845881 10.1016/S0003-4975(01)03391-4
Dehoux MS Hernot S Asehnoune K Boutten A Paquin S Lecon-Malas V Toueg ML Desmonts JM Durand G Philip I Cardiopulmonary bypass decreases cytokine production in lipopolysaccharide-stimulated whole blood cells: roles of interleukin-10 and the extracorporeal circuit Crit Care Med 2000 28 1721 1727 10890609 10.1097/00003246-200006000-00004
De Groote D Zangerle PF Gevaert Y Fassotte MF Beguin Y Noizat-Pirenne F Pirenne J Gathy R Lopez M Dehart I Direct stimulation of cytokines (IL-1-β, TNF-α, IL-6, IL-2, IFN-γ and GM-CSF) in whole blood. I. Comparison with isolated PBMC stimulation Cytokine 1992 4 239 248 1498259 10.1016/1043-4666(92)90062-V
Seghaye MC Heyl W Grabitz RG Schumacher K von Bernuth G Rath W Duchateau J The production of pro- and anti-inflammatory cytokines in neonates assessed by stimulated whole cord blood culture and by plasma levels at birth Biol Neonate 1998 73 220 227 9551188 10.1159/000013980
Borgermann J Friedrich I Flohe S Spillner J Majetschak M Kuss O Sablotzki A Feldt T Reidemeister JC Schade FU Tumor necrosis factor-α production in whole blood after cardiopulmonary bypass: downregulation caused by circulating cytokine-inhibitory activities J Thorac Cardiovasc Surg 2002 124 608 617 12202878 10.1067/mtc.2002.122300
Dembinski J Behrendt D Martini R Heep A Bartmann P Modulation of pro- and anti-inflammatory cytokine production in very preterm infants Cytokine 2003 21 200 206 12788309 10.1016/S1043-4666(02)00498-2
Seghaye MC Grabitz RG Duchateau J Busse S Dabritz S Koch D Alzen G Hornchen H Messmer BJ von Bernuth G Inflammatory reaction and capillary leak syndrome related to cardiopulmonary bypass in neonates undergoing cardiac operations J Thorac Cardiovasc Surg 1996 112 687 697 8800157
Wan S LeClerc JL Vincent JL Inflammatory response to cardiopulmonary bypass – mechanisms involved and possible therapeutic strategies Chest 1997 112 676 692 9315800
Qing M Vazquez-Jimenez JF Klosterhalfen B Sigler M Schumacher K Duchateau J Messmer BJ von Bernuth G Seghaye MC Influence of temperature during cardiopulmonary bypass on leukocyte activation, cytokine balance, and post-operative organ damage Shock 2001 15 372 377 11336197
Seghaye MC Duchateau J Bruniaux J Demontoux S Bosson C Serraf A Lecronier G Mokhfi E Planche C Interleukin-10 release related to cardiopulmonary bypass in infants undergoing cardiac operations J Thorac Cardiovasc Surg 1996 111 545 553 8601968
Duval EL Kavelaars A Veenhuizen L van Vught AJ van de Wal HJ Heijnen CJ Pro- and anti-inflammatory cytokine patterns during and after cardiac surgery in young children Eur J Pediatr 1999 158 387 393 10333121 10.1007/s004310051098
Keel M Schregenberger N Steckholzer U Ungethum U Kenney J Trentz O Ertel W Endotoxin tolerance after severe injury and its regulatory mechanisms J Trauma 1996 41 430 437 8810959 10.1097/00005373-199609000-00008
Appoloni O Vincent JL Duchateau J Response of tumour necrosis factor-α to delayed in vitro monocyte stimulation in patients with septic shock is related to outcome Clin Sci (Lond) 2002 102 315 320 11869172 10.1042/CS20010260
Widomski D Fretland DJ Gasiecki AF Collins PW The prostaglandin analogs, misoprostol and SC-46275, potently inhibit cytokine release from activated human monocytes Immunopharmacol Immunotoxicol 1997 19 165 174 9130003
Schroder M Meisel C Buhl K Profanter N Sievert N Volk HD Grutz G Different modes of IL-10 and TGF-β to inhibit cytokine-dependent IFN-γ production: consequences for reversal of lipopolysaccharide desensitization J Immunol 2003 170 5260 5267 12734375
Adib-Conquy M Adrie C Moine P Asehnoune K Fitting C Pinsky MR Dhainaut JF Cavaillon JM NF-κB expression in mononuclear cells of patients with sepsis resembles that observed in lipopolysaccharide tolerance Am J Respir Crit Care Med 2000 162 1877 1883 11069829
Alcaraz AJ Sancho L Manzano L Esquivel F Carrillo A Prieto A Bernstein ED Alvarez-Mon M Newborn patients exhibit an unusual pattern of interleukin 10 and interferon gamma serum levels in response to cardiac surgery J Thorac Cardiovasc Surg 2002 123 451 458 11882815 10.1067/mtc.2002.120006
van Leeuwen HJ van der Bruggen T van Asbeck BS Boereboom FTJ Effect of corticosteroids on nuclear factor-κB activation and hemodynamics in late septic shock Crit Care Med 2001 29 1074 1077 11378624 10.1097/00003246-200105000-00041
Wan S LeClerc JL Schmartz D Barvais L Huynh CH Deviere J DeSmet JM Vincent JL Hepatic release of interleukin-10 during cardiopulmonary bypass in steroid-pretreated patients Am Heart J 1997 133 335 339 9060803
Tabardel Y Duchateau J Schmartz D Marecaux G Shahla M Barvais L LeClerc JL Vincent JL Corticosteroids increase blood interleukin-10 levels during cardiopulmonary bypass in men Surgery 1996 119 76 80 8560390
|
16277717
|
PMC1297621
|
CC BY
|
2021-01-04 16:04:56
|
no
|
Crit Care. 2005 Aug 9; 9(5):R549-R555
|
utf-8
|
Crit Care
| 2,005 |
10.1186/cc3794
|
oa_comm
|
==== Front
Crit CareCritical Care1364-85351466-609XBioMed Central London cc37971627771810.1186/cc3797ResearchUrinary bladder partial carbon dioxide tension during hemorrhagic shock and reperfusion: an observational study Dubin Arnaldo [email protected] Mario O [email protected] Vanina S Kanoore [email protected] Gastón [email protected] Héctor S [email protected]án Marcelo [email protected] Bernardo [email protected] Gonzalo 8Laporte Mercedes 9Estenssoro Elisa [email protected] Medical Director, Intensive Care Unit, Sanatorio Otamendi y Miroli, Buenos Aires, Argentina2 Staff physician, Intensive Care Unit, Clínicas Bazterrica y Santa Isabel, Buenos Aires, Argentina3 Research Fellow, Cátedra de Farmacología, Facultad de Ciencias Médicas, Universidad Nacional de La Plata, La Plata, Argentina4 Staff physician, Intensive Care Unit, Clínicas Bazterrica y Santa Isabel, Buenos Aires, Argentina5 Staff physician, Intensive Care Unit, Hospital San Martín de La Plata, Argentina6 Medical Director, Renal Transplantation Unit, CRAI Sur, CUCAIBA, Argentina7 Medical Director, Intensive Care Unit, Hospital Posadas, Buenos Aires, Argentina8 Resident, Intensive Care Unit, Hospital San Martín de La Plata, Argentina9 Medical Director, Clinical Chemistry Laboratory, Hospital San Martín de La Plata, Argentina10 Medical Director, Intensive Care Unit, Hospital San Martín de La Plata, Argentina2005 17 8 2005 9 5 R556 R561 17 6 2005 12 7 2005 20 7 2005 25 7 2005 Copyright © 2005 Dubin 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
Continuous monitoring of bladder partial carbon dioxide tension (PCO2) using fibreoptic sensor technology may represent a useful means by which tissue perfusion may be monitored. In addition, its changes might parallel tonometric gut PCO2. Our hypothesis was that bladder PCO2, measured using saline tonometry, will be similar to ileal PCO2 during ischaemia and reperfusion.
Method
Six anaesthetized and mechanically ventilated sheep were bled to a mean arterial blood pressure of 40 mmHg for 30 min (ischaemia). Then, blood was reinfused and measurements were repeated at 30 and 60 min (reperfusion). We measured systemic and gut oxygen delivery and consumption, lactate and various PCO2 gradients (urinary bladder–arterial, ileal–arterial, mixed venous–arterial and mesenteric venous–arterial). Both bladder and ileal PCO2 were measured using saline tonometry.
Results
After bleeding systemic and intestinal oxygen supply dependency and lactic acidosis ensued, along with elevations in PCO2 gradients when compared with baseline values (all values in mmHg; bladder ΔPCO2 3 ± 3 versus 12 ± 5, ileal ΔPCO2 9 ± 5 versus 29 ± 16, mixed venous–arterial PCO2 5 ± 1 versus 13 ± 4, and mesenteric venous–arterial PCO2 4 ± 2 versus 14 ± 4; P < 0.05 versus basal for all). After blood reinfusion, PCO2 gradients returned to basal values except for bladder ΔPCO2, which remained at ischaemic levels (13 ± 7 mmHg).
Conclusion
Tissue and venous hypercapnia are ubiquitous events during low flow states. Tonometric bladder PCO2 might be a useful indicator of tissue hypoperfusion. In addition, the observed persistence of bladder hypercapnia after blood reinfusion may identify a territory that is more susceptible to reperfusion injury. The greatest increase in PCO2 gradients occurred in gut mucosa. Moreover, the fact that ileal ΔPCO2 was greater than the mesenteric venous–arterial PCO2 suggests that tonometrically measured PCO2 reflects mucosal rather than transmural PCO2. Ileal ΔPCO2 appears to be the more sensitive marker of ischaemia.
==== Body
Introduction
Monitoring the adequacy of tissue oxygenation in critically ill patients is a challenging task [1]. Despite extensive research, tissue capnometry remains the only clinically relevant approach to monitoring regional perfusion and oxygenation. Elevation in tissue partial carbon dioxide tension (PCO2) might represent a better surrogate of hypoperfusion than other systemic and regional parameters [2,3].
During the past 20 years a large body of clinical evidence was developed supporting the usefulness of gastrointestinal PCO2 tonometry for the monitoring of tissue perfusion [4]. Gastric tonometry can readily be performed in the critically ill and gives significant information on outcomes [5,6]. It may also be a helpful guide in therapeutic decision making [7]. Nevertheless, technical difficulties and frequent artefacts have dampened the initial enthusiasm [8]. In an attempt to overcome the limitations of gastric tonometry, sublingual capnometry was then developed [9]. Despite initial interest and potential advantages, this technique has neither been completely validated nor widely used [10].
More recently, tissue perfusion has been assessed with continuous monitoring of bladder PCO2 using fibreoptic sensor technology [11,12], yielding interesting findings in experimental models of ischaemia/reperfusion. Although the equipment required may be expensive, bladder PCO2 can readily be measured via a urinary catheter incorporating a silicone balloon. Our goal in the present study was to compare bladder PCO2 measured using saline tonometry versus other tissue and venous PCO2 values. Our hypothesis was that bladder PCO2 will track ileal PCO2 during ischaemia and reperfusion.
Materials and methods
Surgical preparation
Six sheep were anaesthetized with 30 mg/kg sodium pentobarbital, intubated and mechanically ventilated (Harvard Apparatus Dual Phase Control Respirator Pump Ventilator; South Natick, MA, USA) with a tidal volume of 15 ml/kg, a fractional inspired oxygen of 0.21, and positive end-expiratory pressure adjusted to maintain arterial oxygen saturation above 90%. The respiratory rate was set to keep the end-tidal PCO2 at 35 mmHg. Neuromuscular blockade was applied with intravenous pancuronium bromide (0.06 mg/kg). Additional pentobarbital boluses (1 mg/kg per hour) were administered.
Catheters were advanced through the left femoral vein to administer fluids and drugs, and through left femoral artery to measure blood pressure and obtain blood gases. A pulmonary artery catheter was inserted through the right external jugular vein (Flow-directed thermodilution fibreoptic pulmonary artery catheter; Abbott Critical Care Systems, Mountain View, CA, USA).
An orogastric tube was inserted to allow drainage of gastric contents. Then, a midline laparotomy and splenectomy were performed. An electromagnetic flow probe was placed around the superior mesenteric artery to measure intestinal blood flow. A catheter was placed in the mesenteric vein through a small vein proximal to the gut to draw blood gases. Tonometers (TRIP Sigmoid Catheter; Tonometrics, Inc., Worcester, MA, USA) were inserted through small ileotomy and cystostomy to measure ileal and urinary bladder intramucosal PCO2. A second catheter was placed through the same cystostomy to drain urine. Finally, after careful haemostasis, the abdominal wall incision was closed.
Measurements and derived calculations
Arterial, systemic, pulmonary and central venous pressures were measured using corresponding transducers (Statham P23 AA; Statham, Hato Rey, Puerto Rico). Cardiac output was measured by thermodilution with 5 ml saline solution at 0°C (HP OmniCare Model 24 A 10; Hewlett Packard, Andover, MA, USA). An average of three measurements taken randomly during the respiratory cycle was considered and was referenced to body weight to yield the cardiac output (Q). Intestinal blood flow was measured with the electromagnetic method (Spectramed Blood Flowmeter model SP 2202 B; Spectramed Inc., Oxnard, CA, USA) with in vitro calibrated transducers of 5–7 mm diameter (Blood Flowmeter Transducer; Spectramed Inc.). Occlusive zero was controlled before and after each experiment. Non-occlusive zero was corrected before each measurement. Superior mesenteric blood flow was referenced to gut weight (Qintestinal).
Arterial, mixed venous and mesenteric venous partial oxygen tension (PO2), PCO2 and pH were measured using a blood gas analyzer (ABL 5; Radiometer, Copenhagen, Denmark), and haemoglobin and oxygen saturation were measured using a co-oximeter calibrated for sheep blood (OSM 3; Radiometer). Arterial oxygen content (CaO2), mixed venous oxygen content (CvO2) and mesenteric venous oxygen content (CvmO2) were calculated as follows: haemoglobin × 1.34 × oxygen saturation + PO2 × 0.0031. Systemic and intestinal oxygen delivery (DO2) and oxygen consumption (VO2) were calculated as follows: systemic DO2 = Q × CaO2; systemic VO2 = Q × (CaO2 - CvO2); intestinal DO2 = Qintestinal × CaO2; and intestinal VO2 = Qintestinal × (CaO2 - CvmO2).
Arterial lactate concentration was measured using an automatic analyzer (Hitachi 912; Boehringer Mannheim Corporation, Indianapolis, IN, USA).
Bladder and ileal intramucosal PCO2 were measured using a tonometer filled with 2.5 ml saline solution. Of the solution, 1.0 ml was discarded after an equilibration period of 30 min, and PCO2 was measured in the remaining 1.5 ml. These values were corrected for the equilibration period and were used to calculate intramucosal-arterial gradients (bladder and ileal ΔPCO2). Mixed venous–arterial PCO2 (Pv–aCO2) and mesenteric venous–arterial PCO2 differences (Pvm–aCO2) were also calculated.
Experimental procedure
Basal measurements were taken after a stabilization period longer than 30 min. Then, sheep were bled to a mean arterial blood pressure of 40 mmHg for 30 min (ischaemia). This degree of arterial hypotension was maintained by extracting or returning blood, as necessary. Collected blood was heparinized (5,000 U/l) and stored in a warmed water bath (37.5°C). Then, blood was reinfused and measurements were repeated at 30 and 60 min (reperfusion).
At the end of the experiment the animals were killed with an additional dose of pentobarbital and a KCl bolus. A catheter was inserted into the superior mesenteric artery and Indian ink was instilled. Dyed intestinal segments were dissected, washed and weighed to calculate gut indices.
The local Animal Care Committee approved the study. Care of animals was in accordance with US National Institute of Health guidelines.
Statistical analysis
Data were assessed for normality and expressed as mean ± standard deviation. Differences were analyzed using repeated measures analysis of variance and Dunnett's multiple comparisons test to compare each time point with baseline. One-time comparisons between different PCO2 gradients were tested using one-way analysis of variance and Newman–Keuls multiple comparisons test.
Results
Haemodynamic and oxygen transport effects
Mean arterial pressure decreased during bleeding, as did Q, Qintestinal and systemic and intestinal DO2 and VO2. These variables returned to basal values after reinfusion of blood, with the exception of mean arterial pressure and systemic VO2, which remained higher than basal values (Table 1).
Metabolic effects
Metabolic acidosis and hyperlactataemia developed during ischaemia, and persisted after reinfusion (Table 2).
Effects on partial carbon dioxide tension gradients
Mixed and mesenteric venoarterial and urinary bladder and ileal ΔPCO2 differences increased during ischaemia. Ileal ΔPCO2 was higher than other PCO2 gradients during ischaemia (Fig. 1). The change in ileal ΔPCO2 (20 ± 10 mmHg) during ischaemia was greater than that in bladder ΔPCO2 (8 ± 7 mmHg) and in Pv–aCO2 (9 ± 5 mmHg) and Pvm–aCO2 (10 ± 3 mmHg; P < 0.05 for bladder ΔPCO2, Pv–aCO2 and Pvm–aCO2 versus ileal ΔPCO2). However, all PCO2 gradients returned to basal values after reperfusion, except for bladder ΔPCO2, which remained elevated (Fig. 1).
Discussion
The main finding in the present study is the consistent expression of hypercapnia during low flow states. High PCO2 values were evident in veins, ileum and even urinary bladder. In contrast to the other carbon dioxide gradients, bladder ΔPCO2 remained elevated after reperfusion.
The prevention, detection and correction of tissue dysoxia are main goals in the management of critically ill patients [1]. Gastric tonometry has been considered the only available method to track tissue oxygenation in the clinical arena [1]. However, tissue hypercapnia is not just a marker of dysoxia but is also an indicator of hypoperfusion. Tissue and venous PCO2 remain unchanged in states of tissue dysoxia with preserved blood flow, such as hypoxic and anaemic hypoxia [13-15]. On the other hand, in a high flow state, such as sepsis, measurements of intramucosal acidosis remain helpful because of the frequent presence of microcirculatory derangements [16]. Moreover, increased blood flow may correct tissue hypercapnia in endotoxaemia [17].
Although most studies dealing with tissue capnometry have focused on the gastrointestinal tract, others have been performed in muscle [18,19], renal parenchyma [20,21] and subcutaneous tissue [22]. Few studies have assessed urinary PCO2 for the monitoring of tissue oxygenation. Lin and coworkers [23] measured urinary PCO2 in critically ill patients to evaluate the adequacy of perfusion. Urinary PCO2 was higher in shock than in control patients (79 ± 10 mmHg versus 43 ± 2 mmHg; P < 0.0001). Lang and colleagues [11] measured urinary bladder gases using a fibreoptic sensor in a swine model of ischaemia/reperfusion. After 30 min of aortic clamping bladder PCO2 increased from 57 ± 5 mmHg to 117 ± 7 mmHg, and it returned to baseline after 60 min of reperfusion. Clavijo-Alvarez and coworkers [12] studied this issue in a model of haemorrhagic shock in which pigs were bled and kept at a mean arterial pressure of 40 mmHg until decompensation. Animals were then resuscitated with shed blood plus lactated Ringer's solution and observed for 2 hours. In contrast to our findings, those investigators found greater increases in bladder PCO2; basal PCO2 was 49 ± 6 mmHg and increased to 71 ± 7 mmHg at the end of shock. Jejunal intramucosal PCO2 exhibited similar behaviour.
These differences might be related to the use of different animal species but also, and primarily, to the longer period of shock. Because the pigs in the study by Clavijo-Alvarez and coworkers [12] reached a lower cardiac output than did the sheep in our study, changes in surrogates of hypoperfusion such as base excess bicarbonate and bladder PCO2 were more pronounced. Nevertheless, gut intramucosal acidosis was similar in both studies, which might be related to the greater vulnerability of sheep intestinal mucosa to hypoperfusion. In addition, differences might be explained by diverse surgical preparations and methods for measuring intramucosal PCO2. Clavijo-Alvarez and coworkers completely isolated the bladder, and the PCO2 sensor was encased within the mucosa so that they could avoid interference. In this way, the measurements should reflect those from the bladder wall more accurately. Furthermore, they used a more sensitive method to measure PCO2. Nevertheless, it is difficult to reproduce this type of measurement in patients, and our methodology seems more suitable for clinical application.
Although tissue and venous hypercapnia is a widespread consequence of hypoperfusion, our experiments reveal that the increase in PCO2 is higher in ileal mucosa than in bladder mucosa and mixed and mesenteric venous blood. The underlying mechanism producing this preferential elevation in ileal ΔPCO2 might be related to particular characteristics of villi microcirculation. Countercurrent circulation might induce a functional shunt that could place distal microvilli segments at ischaemic risk [24]. There is some controversy regarding the meaning of intramucosal PCO2; specifically, does it reflect whole wall or superficial mucosal perfusion? An ileal ΔPCO2 greater than the Pvm-aCO2 suggests that tonometric PCO2 reflects mucosal rather than transmural PCO2. On the other hand, the similar increase in bladder–arterial and systemic and intestinal venoarterial PCO2 gradients suggests the presence of similar degrees of hypoperfusion. As previously described [25], the fraction of cardiac output directed to gut (superior mesenteric artery blood flow/cardiac output) decreased during ischaemia (from 0.23 ± 0.06 to 0.16 ± 0.07; data not shown). However, this was not enough to produce differences between systemic and intestinal venoarterial PCO2 gradients.
Another interesting finding of this study lies in the persistence of bladder intramucosal acidosis during reperfusion. Recent studies indicated that ischaemia/reperfusion can cause acute inflammation and contractile dysfunction of the bladder [26]. Bajory and coworkers [27] demonstrated severe microcirculatory derangements such as decreased functional capillary density, red blood cell velocity, venular and arteriolar diameter, and enhanced macromolecular leakage after bladder ischaemia/reperfusion. We speculate that these microcirculatory alterations might lead to decreased carbon dioxide removal. Again, differential susceptibility to injury between species could explain differences from other studies [11,12].
Limitations of the present study could be related to the method of measurement of bladder PCO2. First, tonometric measurement of PCO2 has drawbacks [8]. Second, urine itself could potentially influence tonometric PCO2 beyond perfusion deficits. In fact, urine can have variable carbon dioxide content, resulting, for example, from different grades of carbonic anhydrase inhibition or from systemic bicarbonate administration [28]. Actually, failure to observe an appropriate increase in urinary-blood PCO2 during bicarbonate loading has been employed as an index of reduced distal nephron proton secretion in distal renal tubular acidosis [28]. Changes in systemic oxygenation can also modify urine composition. Moriguchi and coworkers [29] have showed that urinary bicarbonate, calculated from urinary PCO2 and pH, increases after anaerobic exercise. Those authors related these findings to systemic carbon dioxide production and later urinary excretion [29]. They also described a circadian rhythm in urinary bicarbonate elimination [30]. Moreover, an elevated bladder ΔPCO2 could also represent a late manifestation of renal hypoperfusion. Further studies are needed to clarify the influence of renal carbon dioxide excretion on bladder PCO2.
Conclusion
Our data suggest that bladder ΔPCO2 could be a useful indicator of tissue perfusion. However, intestinal ΔPCO2 is the more sensitive carbon dioxide gradient for monitoring low flow states. Further studies are needed to establish the definitive monitoring value of urinary PCO2.
Key messages
• Urinary bladder ΔPCO2 may be a useful indicator of tissue perfusion, but intestinal ΔPCO2 is the more sensitive carbon dioxide gradient for the monitoring of low flow states.
• The fact that the observed ileal ΔPCO2 was greater than Pvm-aCO2 suggests that tonometric PCO2 reflects mucosal rather than transmural PCO2.
Abbreviations
CaO2 = arterial oxygen content; CvmO2 = mesenteric venous oxygen content; CvO2 = mixed venous oxygen content; DO2 = oxygen transport; PCO2 = partial carbon dioxide tension; PO2 = partial oxygen tension; Pv–aCO2 = mixed venous-arterial PCO2 difference; Pvm–aCO2 = mesenteric venous–arterial PCO2 difference; Q = cardiac output; VO2 = oxygen consumption.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
AD was responsible for the study concept and design, analysis and interpretation of data, and drafting of the manuscript. MOP, VSKE, GM and HSC performed acquisition of data and contributed to drafting of the manuscript. BM and ML conducted blood determinations and contributed to drafting of the manuscript. MB and GF performed the surgical preparation and contributed to the discussion. EE helped in the drafting of the manuscript and conducted a critical revision for important intellectual content. All authors read and approved the final manuscript.
Figures and Tables
Figure 1 Behaviour of PCO2 gradients. Shown are the various partial carbon dioxide tension (PCO2) gradients in basal conditions, during ischaemia and after reperfusion.
Table 1 Haemodynamic and oxygen transport parameters at basal conditions, during ischaemia, and after 30 and 60 min of reperfusion
Reperfusion
Parameter Basal Ischemia 30 min 60 min
Mean arterial blood pressure (mmHg) 87 ± 14 38 ± 4 105 ± 10* 104 ± 10*
Cardiac output (ml/min per kg) 138 ± 10 70 ± 17* 136 ± 17 137 ± 16
Intestinal blood flow (ml/min per kg) 787 ± 181 272 ± 100* 890 ± 278 756 ± 134
Systemic oxygen transport (ml/min per kg) 19.5 ± 2.7 7.8 ± 1.9* 18.8 ± 2.8 19.3 ± 3.2
Systemic oxygen consumption (ml/min per kg) 6.8 ± 1.0 5.7 ± 1.5* 7.4 ± 1.2* 7.2 ± 0.9*
Systemic oxygen extraction ratio 0.35 ± 0.06 0.72 ± 0.08* 0.40 ± 0.09* 0.39 ± 0.09
Intestinal oxygen transport (ml/min per kg) 112.5 ± 35.2 31.1 ± 14.0* 126.1 ± 51.1 107.8 ± 28.7
Intestinal oxygen consumption (ml/min per kg) 30.3 ± 4.6 19.3 ± 7.1* 31.3 ± 6.9 31.5 ± 6.6
Intestinal oxygen extraction ratio 0.29 ± 0.09 0.65 ± 0.12* 0.28 ± 0.11 0.31 ± 0.10
*P < 0.05 versus basal.
Table 2 Arterial, mixed venous and mesenteric venous blood gases, and arterial lactate at basal conditions, during ischemia and after 30 and 60 minutes of reperfusion
Reperfusion
Parameter Basal Ischaemia 30 min 60 min
Arterial pH 7.37 ± 0.03 7.36 ± 0.05 7.33 ± 0.05* 7.36 ± 0.04
Arterial PCO2 (mmHg) 38 ± 4 35 ± 5* 36 ± 4 36 ± 5
Arterial PO2 (mmHg) 77 ± 9 80 ± 15 75 ± 10 78 ± 8
Arterial HCO3- (mmol/l) 22 ± 3 19 ± 2* 19 ± 2* 20 ± 2*
Arterial base excess (mmol/l) -3 ± 3 -5 ± 2* -6 ± 2* -4 ± 3*
Mixed venous pH 7.34 ± 0.03 7.26 ± 0.03* 7.28 ± 0.04* 7.32 ± 0.04
Mixed venous PCO2 (mmHg) 43 ± 4 48 ± 5* 43 ± 4 42 ± 3
Mixed venous PO2 (mmHg) 38 ± 4 23 ± 3* 37 ± 4 39 ± 4
Mixed venous HCO3- (mmol/l) 23 ± 3 21 ± 3 20 ± 2 21 ± 2
Mixed venous base excess (mmol/l) -3 ± 3 -6 ± 3* -7 ± 2* -5 ± 2*
Mesenteric venous pH 7.34 ± 0.03 7.26 ± 0.03* 7.30 ± 0.05* 7.32 ± 0.04
Mesenteric venous PCO2 (mmHg) 42 ± 5 49 ± 5* 41 ± 4 41 ± 4
Mesenteric venous PO2 (mmHg) 43 ± 7 26 ± 3* 42 ± 6 43 ± 5
Mesenteric venous HCO3- (mmol/l) 23 ± 3 22 ± 2 20 ± 2* 21 ± 2
Mesenteric venous base excess (mmol/l) -3 ± 3 -5 ± 2* -6 ± 2* -5 ± 2*
Arterial lactate (mmol/l) 1.6 ± 0.5 3.7 ± 1.7* 3.9 ± 2.0* 3.2 ± 1.5*
Values are expressed as mean ± standard deviation. *P < 0.05 versus basal. PCO2, partial carbon dioxide tension; PO2, partial oxygen tension.
==== Refs
Third European Consensus Conference in Intensive Care Medicine Tissue hypoxia: how to detect, how to correct, how to prevent Am J Respir Crit Care Med 1996 154 1573 1578 8912783
Hamilton-Davies C Mythen MG Salmon JB Jacobson D Shukla A Webb AR Comparison of commonly used clinical indicators of hypovolaemia with gastrointestinal tonometry Intensive Care Med 1997 23 276 281 9083229 10.1007/s001340050328
Dubin A Estenssoro E Murias G Canales H Sottile P Badie J Barán M Pálizas F Laporte M Rivas Díaz M Effects of hemorrhage on gastrointestinal oxygenation Intensive Care Med 2001 27 1931 1936 11797030 10.1007/s00134-001-1138-9
Taylor DE Gutierrez G Tonometry. A review of clinical studies Crit Care Clin 1996 12 1007 1018 8902381 10.1016/S0749-0704(05)70289-X
Marik PE Gastric intramucosal pH. A better predictor of multiorgan dysfunction syndrome and death than oxygen-derived variables in patients with sepsis Chest 1993 104 225 229 8325075
Doglio GR Pusajo JF Egurrola MA Bonfigli GC Parra C Vetere L Hernandez MS Fernandez S Palizas F Gutierrez G Gastric mucosal pH as a prognostic index of mortality in critically ill patients Crit Care Med 1991 19 1037 1040 1860328
Gutierrrez G Pálizas F Doglio G Wainsztein N Gallesio A Pacín J Dubin A Schiavi E Jorge M Pusajó J Gastric intramucosal pH as a therapeutic index of tissue oxygenation in critically ill patients Lancet 1992 339 195 199 1346170 10.1016/0140-6736(92)90002-K
Oud L Kruse JA Poor in vivo reproducibility of gastric intramucosal pH determined by saline-filled balloon tonometry J Crit Care 1996 11 144 150 8891966 10.1016/S0883-9441(96)90011-8
Marik PE Bankov A Sublingual capnometry versus traditional markers of tissue oxygenation in critically ill patients Crit Care Med 2003 31 818 822 12626990 10.1097/01.CCM.0000054862.74829.EA
Maciel AT Creteur J Vincent JL Tissue capnometry: Does the answer lie under the tongue? Intensive Care Med 2004 30 2157 2165 15650865 10.1007/s00134-004-2416-0
Lang JD JrEvans DJ deFigueiredo LP Hays S Mathru M Kramer GC A novel approach to monitor tissue perfusion: bladder mucosal PCO2, PO2, and pHi during ischemia and reperfusion J Crit Care 1999 14 93 98 10382790 10.1016/S0883-9441(99)90020-5
Clavijo-Alvarez JA Sims CA Menconi M Shim I Ochoa C Puyana JC Bladder mucosa pH and PCO2 as a minimally invasive monitor of hemorrhagic shock and resuscitation J Trauma 2004 57 1199 1210 15625450
Vallet B Teboul JL Cain S Curtis S Venoarterial CO2 difference during regional ischemic or hypoxic hypoxia J Appl Physiol 2000 89 1317 1321 11007564
Dubin A Murias G Estenssoro E Canales H Badie J Pozo M Sottile JP Baran M Palizas F Laporte M Intramucosal-arterial PCO2 gap fails to reflect intestinal dysoxia in hypoxic hypoxia Crit Care 2002 6 514 520 12493073 10.1186/cc1813
Dubin A Estenssoro E Murias G Pozo MO Sottile JP Baran M Piacentini E Canales HS Etcheverry G Intramucosal-arterial PCO2 gradient does not reflect intestinal dysoxia in anemic hypoxia J Trauma 2004 57 1211 1217 15625451
Tugtekin IF Radermacher P Theisen M Matejovic M Stehr A Ploner F Matura K Ince C Georgieff M Trager K Increased ileal-mucosal-arterial PCO2 gap is associated with impaired villus microcirculation in endotoxic pigs Intensive Care Med 2001 27 757 766 11398705 10.1007/s001340100871
Dubin A Murias G Maskin B Pozo M Sottile JP Barán M Kanoore Edul VS Canales HS Badie J Etcheverry G Estenssoro E Increased blood flow prevents intramucosal acidosis in sheep endotoxemia: A controlled study Critical Care 2005 9 R66 R73 15774052 10.1186/cc3021
Kvarstein G Mirtaheri P Tonnessen TI Detection of ischemia by PCO2 before adenosine triphosphate declines in skeletal muscle Crit Care Med 2004 32 232 237 14707584 10.1097/01.CCM.0000104948.41544.FC
Sims C Seigne P Menconi M Monarca J Barlow C Pettit J Puyana JC Skeletal muscle acidosis correlates with the severity of blood volume loss during shock and resuscitation J Trauma 2001 51 1137 1145 11740266
Murakawa K Kobayashi A Effects of vasopressors on renal tissue gas tensions during hemorrhagic shock in dogs Crit Care Med 1988 16 789 792 3135151
Tonnessen TI Kvarstein G PCO2 electrodes at the surface of the kidney detect ischaemia Acta Anaesthesiol Scand 1996 40 510 519 8792878
Venkatesh B Morgan TJ Hall J Endre Z Willgoss D Subcutaneous gas tensions closely track ileal mucosal gas tensions in a model of endotoxaemia without anaerobism Intensive Care Med 2005 31 447 453 15703895 10.1007/s00134-005-2558-8
Lin MS Lien TC Yang WC Wu SC Tsai WW Wang JH Urinary PCO2 for hemodynamically unstable patients Zhonghua Yi Xue Za Zhi (Taipei) 1996 57 112 117 8634925
Shepherd AP Kiel JW A model of countercurrent shunting of oxygen in the intestinal villus Am J Physiol 1992 262 H1136 H1142 1373572
Reilly PM MacGowan S Miyachi M Schiller HJ Vickers S Bulkley GB Mesenteric vasoconstriction in cardiogenic shock in pigs Gastroenterology 1992 102 1968 1979 1587416
Bratslavsky G Kogan BA Matsumoto S Aslan AR Levin RM Reperfusion injury of the rat bladder is worse than ischemia J Urol 2003 170 2086 2090 14532859 10.1097/01.ju.0000092144.48045.13
Bajory Z Hutter J Krombach F Messmer K The role of endothelin-1 in ischemia-reperfusion induced acute inflammation of the bladder in rats J Urol 2002 168 1222 1225 12187271 10.1097/00005392-200209000-00092
DuBose TD JrCaflisch CR Validation of the difference in urine and blood carbon dioxide tension during bicarbonate loading as an index of distal nephron acidification in experimental models of distal renal tubular acidosis J Clin Invest 1985 75 1116 1123 3921566
Moriguchi T Tomoda A Ichimura S Odagiri Y Inoue S Nagasawa T Tanaka H Nakagawa N Shimomitsu T Significance of post-exercise increment of urinary bicarbonate and pH in subjects loaded with submaximal cycling exercise Tohoku J Exp Med 2004 202 203 211 15065646 10.1620/tjem.202.203
Moriguchi T Shimomitsu T Odagiri Y Ichimura S Fukuda J Tomoda A Circadian changes in urinary bicarbonate, nitric oxide metabolites and pH in female player during handball camp involved in an exercise, rest and sleep cycle Tohoku J Exp Med 2002 196 281 291 12086156 10.1620/tjem.196.281
|
16277718
|
PMC1297623
|
CC BY
|
2021-01-04 16:04:55
|
no
|
Crit Care. 2005 Aug 17; 9(5):R556-R561
|
utf-8
|
Crit Care
| 2,005 |
10.1186/cc3797
|
oa_comm
|
==== Front
Crit CareCritical Care1364-85351466-609XBioMed Central London cc37991627771910.1186/cc3799ResearchRelation between respiratory variations in pulse oximetry plethysmographic waveform amplitude and arterial pulse pressure in ventilated patients Cannesson Maxime [email protected] Cyril 2Durand Pierre G 1Bohé Julien 3Jacques Didier 31 Anesthesiology and Critical Care Fellow, Service de Réanimation Médicale, Centre Hospitalier Lyon Sud, Pierre Bénite, France2 Intensivist, Service de Réanimation Médicale, Centre Hospitalier Lyon Sud, Pierre Bénite, France3 Intensivist, Service de Réanimation Médicale, Centre Hospitalier Lyon Sud, Pierre Bénite, France2005 23 8 2005 9 5 R562 R568 30 6 2005 29 7 2005 Copyright © 2005 Cannesson et al.; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons AttributionLicense (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Introduction
Respiratory variation in arterial pulse pressure is a reliable predictor of fluid responsiveness in mechanically ventilated patients with circulatory failure. The main limitation of this method is that it requires an invasive arterial catheter. Both arterial and pulse oximetry plethysmographic waveforms depend on stroke volume. We conducted a prospective study to evaluate the relationship between respiratory variation in arterial pulse pressure and respiratory variation in pulse oximetry plethysmographic (POP) waveform amplitude.
Method
This prospective clinical investigation was conducted in 22 mechanically ventilated patients. Respiratory variation in arterial pulse pressure and respiratory variation in POP waveform amplitude were recorded simultaneously in a beat-to-beat evaluation, and were compared using a Spearman correlation test and a Bland–Altman analysis.
Results
There was a strong correlation (r2 = 0.83; P < 0.001) and a good agreement (bias = 0.8 ± 3.5%) between respiratory variation in arterial pulse pressure and respiratory variation in POP waveform amplitude. A respiratory variation in POP waveform amplitude value above 15% allowed discrimination between patients with respiratory variation in arterial pulse pressure above 13% and those with variation of 13% or less (positive predictive value 100%).
Conclusion
Respiratory variation in arterial pulse pressure above 13% can be accurately predicted by a respiratory variation in POP waveform amplitude above 15%. This index has potential applications in patients who are not instrumented with an intra-arterial catheter.
See related commentary
==== Body
Introduction
Initial therapy in patients with sepsis-induced circulatory failure is volume expansion. However, fluid therapy is not always efficient and does not always increase stroke volume. Furthermore, fluid therapy carries major risks for complications such as volume overload, systemic and pulmonary oedema, and increased tissue hypoxia [1]. To avoid the potential deleterious effects of volume expansion, reliable predictors of fluid responsiveness are needed. In mechanically ventilated patients, respiratory variations in systemic systolic pressure (ΔPs) have been proposed to be a good indicator of fluid responsiveness [2,3]. Indeed, fluid responsiveness was found to be proportional to ΔPs. More recently, respiratory variations in systemic pulse pressure (ΔPP) were shown to be even more predictive of fluid responsiveness [4]. ΔPP above 13% could predict an increase in cardiac index of 15% or more after infusion of 500 ml colloids with positive and negative predictive values of 94% and 96%, respectively. One of the limitations of this method is that it requires an intra-arterial catheter, and catheter-related sepsis and ischaemia are well known complications of the use of such devices [5,6]. Furthermore, most patients are not equipped with such a device when the circulatory failure manifests.
Pulse oximeters are widely used in intensive care units. The pulse oximetry plethysmographic (POP) waveform depends on arterial pulsatility. Respiratory variations in POP waveform peaks are correlated with ΔPs [7] in the setting of mechanical ventilation. However, respiratory variations in POP waveform amplitude (ΔPOP) have not been evaluated. Thus, we tested the hypothesis that ΔPOP and ΔPP are correlated in mechanically ventilated critically ill patients.
Materials and methods
The protocol used in the present study was part of our routine clinical practice, and ethical approval was given by the institutional review board (Comité Consultatif de Protection des Personnes dans la Recherche Biomédicale Lyon B) of our institution (Hospices Civils de Lyon, France).
Patients
Twenty-two deeply sedated patients (14 men and 8 women) receiving mechanical ventilation were studied. Their age (mean ± standard deviation) was 64 ± 11 years (range 41–85 years). Inclusion criteria were as follows: instrumentation with an indwelling radial arterial catheter, according to the attending physician; haemodynamic stability, defined as a variation in heart rate and blood pressure of less than 10% over the 15 min preceding the period of evaluation; and pulse oximetry monitored using a pulse oximeter (M1190A; Philips, Suresnes, France) attached to the index or middle finger. Exclusion criteria were cardiac arrhythmia and low POP signal. POP waveform quality was considered suitable when POP amplitude was superior to the signal quality index displayed by the monitor.
Haemodynamic measurements
Patients were studied in supine position. The arterial pressure transducer was set at mid-axillary level for zero pressure. When available, transthoracic echocardiography was performed to assess left ventricular function. Left ventricular systolic dysfunction was defined as left ventricular ejection fraction below 40%.
Respiratory variations in arterial pulse pressure analysis
Pulse pressure (PP) was calculated at the bedside using a standard monitor (Monitor M1165A; Philips) as the difference between systolic and diastolic arterial pressures. Maximal PP (PPmax) and minimal PP (PPmin) values were determined over the same respiratory cycle. To assess ΔPP, the percentage change in PP was calculated (as described by Michard and coworkers [4]): ΔPP (%) = 100 × ([PPmax - PPmin]/ [(PPmax + PPmin)/2]). The measurements were repeated on three consecutive respiratory cycles and averaged for statistical analysis.
Respiratory variations in POP waveform amplitude analysis
A pulse oximeter was attached to the index or middle finger of either right or left hand. POP waveforms were recorded using a monitor (M3150A; Philips). The plethysmographic gain factor was held constant throughout the procedure, which was possible because the bedside monitor allows one to choose between manual and automatic gain control. POP waveform amplitude was measured on a beat-to-beat basis as the vertical distance between peak and preceding valley trough in the waveform and was expressed in millimeters (Fig. 1). Maximal POP (POPmax) and minimal POP (POPmin) were determined over the same respiratory cycle. ΔPOP were calculated using a formula similar to that for ΔPP: ΔPOP (%) = 100 × ([POPmax - POPmin]/ [(POPmax + POPmin)/2]). ΔPOP was evaluated on three consecutive respiratory cycles simultaneously with ΔPP measurements. Measurements were then averaged for statistical analysis.
Respiratory parameters
All patients received mechanical ventilation in volume controlled mode with a tidal volume of 8 ± 2 ml/kg and an inspiratory/expiratory ratio of one-third to one-half. Positive end-expiratory pressure was set at 5 ± 4 cmH2O.
Statistical analysis
Linear correlation between ΔPP and ΔPOP was tested using the Spearman rank method. ΔPP and ΔPOP were compared using Bland–Altman analyses [8]. Data are presented as mean ± standard deviation. A receiver operating characteristic curve was generated for ΔPOP, varying the discriminating threshold of this parameter to determine the ability of ΔPOP to discriminate between patients with a ΔPP above 13% and those with a ΔPP of 13% or less. P < 0.05 was considered statistically significant. Statistical analysis was performed using Statview 5.0 software (SAS Institute Inc., Cary, NC, USA).
Results
Among the 22 patients studied, acute circulatory failure (defined as a systolic blood pressure <90 mmHg or need for vasopressive drugs) was present in 14 patients (12 received vasopressor support and two had severe hypotension). Sepsis (n = 7) and bleeding (n = 3) were the main causes. Other patients had isolated acute respiratory failure.
Echocardiography was performed in 19 patients (86%), revealing left ventricular systolic dysfunction in five patients. Other demographic, haemodynamic and ventilatory parameters are presented in Table 1.
In the patients overall, there was a strong correlation between ΔPP and ΔPOP (r2 = 0.83; P < 0.001), as shown in Fig. 2. This correlation remained significant in the subgroup of 14 patients with acute circulatory failure (r2 = 0.89; P < 0.001). Using Bland–Altman analysis (Fig. 3), there was a weak bias and relatively good precision between the two methods (0.8 ± 3.5%). Figure 4 shows an example of simultaneous recording of arterial pressure and pulse oxymetry plethysmography in a patient with gastric bleeding. The threshold ΔPOP value of 15% permitted discrimination between patients with a ΔPP above 13% and those with a ΔPP of 13% or less with a sensitivity of 87%, a specificity of 100%, a positive predictive value of 100% and a negative predictive value of 94%.
Discussion
ΔPP is an invasive but accurate indicator of fluid responsiveness in mechanically ventilated patients with acute circulatory failure [4]. This study demonstrates a close relationship between the noninvasively measured parameter ΔPOP and ΔPP. A patient with a ΔPOP value above 15% was highly likely to have a ΔPP value of above 13% (positive predictive value 100%). In contrast, if ΔPOP was below 15% then the patient was unlikely to have a ΔPP value of more than 13% (negative predictive value 94%). However, it must be recalled that 13% is not a universal cutoff value and that many studies focusing on fluid responsiveness and ΔPP have found values ranging from 11% to 13%; we used 13% as a reference because it was the first cutoff value to be reported [9,10]. The pulse oximeter is a standard noninvasive monitor in intensive care units and operating rooms, and is used to monitor arterial oxygen saturation. Our data suggest that the pulse oximeter could also be used to assess fluid responsiveness, but further studies with volume expansion are needed to address this and to determine the optimal ΔPOP threshold.
Many indices have been proposed for monitoring fluid therapy in patients with acute circulatory failure induced by hypovolaemia or severe sepsis. Right or left ventricular filling pressures and cardiac volume measurements have several limitations [11]. In sedated and mechanically ventilated patients, respiratory variations in arterial pressure have been studied for more than 20 years [2,3,12,13], showing that the degree of hypovolaemia correlates closely with ΔPs. Indeed, inspiratory right ventricular stroke volume decrease is proportional to the degree of hypovolaemia and is transmitted to the left heart after two or three beats. Thus, left ventricular stroke volume and then arterial pressure decrease during expiration. More recently, ΔPP were shown to be slightly more predictive of fluid responsiveness than were ΔPs [4]. Indeed, ΔPs depend not only on respiration induced changes in stroke volume but also on respiration induced changes in intrathoracic pressure, which are transmitted to both diastolic and systolic components of blood pressure. On the other hand, PP variations do not depend on intrathoracic pressure variations and therefore are more related to stroke volume variations than variations in systolic pressure [4].
Pulse oximeters display a signal proportional to light absorption between the nail and the anterior face of the finger. Light absorption increases with the amount of haemoglobin present in the fingertip. Thus, the POP waveform depends on the arterial pulse [14]. Previous studies have shown a correlation between respiratory variations in POP waveform peaks and arterial systolic pressure [7,15], demonstrating that decreased preload resulted in waveform variation of the plethysmographic signal similar to the variation observed in the arterial waveform. However, like systolic pressure, POP waveform peaks also depend on transmission of intrathoracic pressure [14]. Hence, POP waveform amplitude analysis should be more accurate. To the best of our knowledge, a relationship between ΔPOP and ΔPP has not yet been reported. PP and POP waveform amplitude are related to stroke volume and vascular tone [14]. Vascular tone is considered unchanged between inspiration and expiration, and so respiratory variations in POP waveform amplitude mainly reflect respiratory changes in left ventricular stroke volume.
Because pulse oximeters are already widely available in intensive care units and operating rooms, they may represent a noninvasive and simple means with which to predict fluid responsiveness in patients with circulatory failure, especially if they are not instrumented with an arterial catheter. Because most patients with shock have arterial catheters, POP waveform analysis could be utilized in patients not routinely monitored with such catheters. Applications include detection and assessment of unexpected circulatory failure in patients undergoing surgery, and preliminary evaluation of patients admitted for shock to intensive care units.
Study limitations
Only 14 out of 22 patients in this series presented with circulatory failure. Seven of them seemed fluid dependent, according to their ΔPP value. It must be emphasized that our study focused on the relationship between respiratory variations in both POP waveform amplitude and arterial PP. Although there was good agreement between ΔPOP and ΔPP, the precision was quite low (3.5%), especially for the highest values (Figs 2 and 3). This means that ΔPP cannot be accurately inferred from ΔPOP. Consequently, further experiments are required to study the relationship between ΔPOP and fluid responsiveness and to determine the optimal threshold value for ΔPOP in patients with acute circulatory failure. Such studies should also focus on the evolution of ΔPOP after volume expansion. Moreover, we chose 13% as a cutoff value for ΔPP because this was the first value to be reported and because most of the studies focusing on this topic found similar values. However, even though a ΔPOP above 15% appears to be a strong predictor of a ΔPP above 13%, it must be emphasized that a prospective study using volume expansion is needed to determine the optimal ΔPOP threshold. Nevertheless, our study is the first to demonstrate a strong relationship between ΔPOP and ΔPP and our findings should be considered to provide a primary hypothesis for such experiments.
As is the case with ΔPP, ΔPOP cannot be assessed in the case of cardiac arrhythmia or in patients who trigger the respirator. Also, the POP signal can be unstable, depending on finger perfusion. Therefore, a stable and satisfactory signal is a prerequisite for assessing ΔPOP. Next, commercially available pulse oximeters use automated algorithms to display a stable signal by adjusting gain continuously. Automatic gain must therefore be disabled to allow respiratory variations to emerge. Finally, the POP waveform is a scaleless curve. Thus, only relative changes in POP waveform amplitude (ΔPOP) can be used to assess volume status, and not absolute values. Because currently there are no commercial monitors that display ΔPOP values, the POP waveform may be used by the clinician by visual inspection alone to assess volume status.
Conclusion
The results of the present study show that there is a strong correlation and a relatively good agreement between ΔPOP and ΔPP. Moreover, a ΔPOP value of more than 15% accurately discriminated patients with a ΔPP above 13% from those with a ΔPP of 13 or less. Therefore, ΔPOP has potential in assessing ΔPP in patients who are not instrumented with an intra-arterial catheter.
Key messages
• ΔPOP and ΔPP are strongly correlated.
• A ΔPOP value above 15% accurately discriminates patients with a ΔPP above 13% from those with a ΔPP of 13 or less.
• ΔPOP has potential for clinical application in the assessment of fluid responsiveness, but further studies are required to address this.
Abbreviations
ΔPOP = respiratory variations in POP waveform amplitude; ΔPP = respiratory variations in systemic pulse pressure; ΔPs = respiratory variations in systemic systolic pressure; POP = pulse oximetry plethysmographic; PP = pulse pressure.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MC conceived the study, analyzed the curves, performed the statistical analysis and drafted the manuscript. CB and PGD collected the data and helped to draft the manuscript. JB participated in the design of the study. DJ conceived the study, participated in its design and coordination, and helped to draft the manuscript. All authors read and approved the final manuscript.
Acknowledgements
This work was presented at the '23rd Congrès de réanimation de langue française' in January 2004 and at the 13th World Congress of Anesthesiologists (PO934) in April 2004.
Figures and Tables
Figure 1 Pulse oximetry plethysmographic waveform analysis. Shown is pulse oximetry plethysmographic (POP) waveform (PLETH) analysis in one illustrative patient. Beat-to-beat measurement of POP waveform amplitude allowed determination of maximal POP (POPmax) and minimal POP (POPmin) over a single respiratory cycle.
Figure 2 Relationship between ΔPOP and ΔPP. Empty circles indicate patients receiving vasopressor support, and empty squares indicate patients with severe hypotension. ΔPOP, respiratory variations in POP waveform amplitude; ΔPP, respiratory variations in systemic pulse pressure.
Figure 3 Bias and precision of ΔPP estimated from ΔPOP (Bland–Altman analysis). ΔPOP, respiratory variations in POP waveform amplitude; ΔPP, respiratory variations in systemic pulse pressure.
Figure 4 Comparison between invasive arterial pressure and pulse oxymetry plethysmography recordings. Simultaneous recording of electrocardiographic lead (II), systemic arterial pressure (PA), pulse oximetry plethysmography (PLETH) and respiratory signal (RESP) in one illustrative patient. POP, pulse oximetry plethysmographic; PP, pulse pressure.
Table 1 Demographic data and baseline values for haemodynamic, plethysmographic and respiratory parameters
Parameter Value (mean ± SD) Range
Demography
Age (year) 64 ± 11 41–85
Height (cm) 166 ± 8 155–180
Weight (kg) 65 ± 12 106–17
Arterial blood pressure and heart rate
Systolic blood pressure (mmHg) 116 ± 20 78–146
Diastolic blood pressure (mmHg) 57 ± 14 39–89
Mean arterial pressure (mmHg) 75 ± 15 54–105
Heart rate (beats/min) 97 ± 19 66–133
PPmax (mmHg) 66 ± 15 39–95
PPmin (mmHg) 60 ± 16 29–90
ΔPP (%) 12 ± 8 2–34
Pulse oximetry plethysmography
POPmax (mm) 52 ± 4 43–61
POPmin (mm) 46 ± 3 37–51
ΔPOP (%) 12 ± 8 3–37
SpO2 (%) 96 ± 3 90–100
Respiratory parameters
PaO2/FiO2 (mmHg) 233 ± 99 67–440
Vt (ml/kg) 8 ± 2 5–13
Respiratory rate (breaths/min) 19 ± 5 12–30
PEEP (cm H2O) 5 ± 4 0–15
ΔPOP, respiratory variations in pulse oximetry plethysmographic waveform amplitude; ΔPP, respiratory variations in pulse pressure; PaO2/FiO2, ratio of arterial oxygen tension to fractional inspired oxygen; PEEP, positive end-expiratory pressure; SpO2, pulse oximeter oxygen saturation; Vt, tidal volume.
==== Refs
Anonymous Practice parameters for hemodynamic support of sepsis in adult patients in sepsis. Task Force of the American College of Critical Care Medicine, Society of Critical Care Medicine Crit Care Med 1999 27 639 660 10199548
Coriat P Vrillon M Perel A Baron JF Le Bret F Saada M Viars P A comparison of systolic blood pressure variations and echocardiographic estimates of end-diastolic left ventricular size in patients after aortic surgery Anesth Analg 1994 78 46 53 8267179
Tavernier B Makhotine O Lebuffe G Dupont J Scherpereel P Systolic pressure variation as a guide to fluid therapy in patients with sepsis-induced hypotension Anesthesiology 1998 89 1313 1321 9856704 10.1097/00000542-199812000-00007
Michard F Boussat S Chemla D Anguel N Mercat A Lecarpentier Y Richard C Pinsky MR Teboul JL Relation between respiratory changes in arterial pulse pressure and fluid responsiveness in septic patients with acute circulatory failure Am J Respir Crit Care Med 2000 162 134 138 10903232
Bedford RF Wollman H Complications of percutaneous radial-artery cannulation: an objective prospective study in man Anesthesiology 1973 38 228 236 4698143
Jones RM Hill AB Nahrwold ML Bolles RE The effect of method of radial artery cannulation on postcannulation blood flow and thrombus formation Anesthesiology 1981 55 76 78 7247060
Shamir M Eidelman LA Floman Y Kaplan L Pizov R Pulse oximetry plethysmographic waveform during changes in blood volume Br J Anaesth 1999 82 178 181 10364990
Bland JM Altman DG Statistical methods for assessing agreement between two methods of clinical measurement Lancet 1986 1 307 310 2868172
Vieillard-Baron A Chergui K Rabiller A Peyrouset O Page B Beauchet A Jardin F Superior vena caval collapsibility as a gauge of volume status in ventilated septic patients Intensive Care Med 2004 30 1734 1739 15375649
Kramer A Zygun D Hawes H Easton P Ferland A Pulse pressure variation predicts fluid responsiveness following coronary artery bypass surgery Chest 2004 126 1563 1568 15539728 10.1378/chest.126.5.1563
Michard F Teboul JLL Predicting fluid responsiveness in ICU patients: a critical analysis of the evidence Chest 2002 121 2000 2008 12065368 10.1378/chest.121.6.2000
Coyle JP Teplick RS Long MC Davison JK Respiratory variations in systemic arterial pressure as an indicator of volume status Anesthesiology 1983 59 A53
Perel A Pizov R Cotev S Systolic blood pressure variation is a sensitive indicator of hypovolemia in ventilated dogs subjected to graded hemorrhage Anesthesiology 1987 67 498 502 3310740
Dorlas JC Nijboer JA Photo-electric plethysmography as a monitoring device in anaesthesia. Application and interpretation Br J Anaesth 1985 57 524 530 3994887
Partridge BL Use of pulse oximetry as a non invasive indicator of intravascular volume status J Clin Monit 1987 3 263 268 3681360
|
16277719
|
PMC1297625
|
CC BY
|
2021-01-04 16:04:56
|
no
|
Crit Care. 2005 Aug 23; 9(5):R562-R568
|
utf-8
|
Crit Care
| 2,005 |
10.1186/cc3799
|
oa_comm
|
==== Front
Crit CareCritical Care1364-85351466-609XBioMed Central London cc38011627772010.1186/cc3801ResearchEffects of reduced rebreathing time, in spontaneously breathing patients, on respiratory effort and accuracy in cardiac output measurement when using a partial carbon dioxide rebreathing technique: a prospective observational study Tachibana Kazuya [email protected] Hideaki [email protected] Muneyuki [email protected] Tomoyo [email protected] Yuji [email protected] Masaji [email protected] Staff physician, Surgical Intensive Care Unit, National Cardiovascular Center, Osaka, Japan2 Director, Surgical Intensive Care Unit, National Cardiovascular Center, Osaka, Japan3 Staff physician, Surgical Intensive Care Unit, National Cardiovascular Center, Osaka, Japan4 Staff physician, Surgical Intensive Care Unit, National Cardiovascular Center, Osaka, Japan5 Staff physician, Surgical Intensive Care Unit, National Cardiovascular Center, Osaka, Japan6 Professor, Department of Emergency and Critical Care Medicine, Tokushima University School of Medicine, Tokushima, Japan2005 7 9 2005 9 5 R569 R574 18 5 2005 24 6 2005 22 7 2005 2 8 2005 Copyright © 2005 Tachibana 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
New technology using partial carbon dioxide rebreathing has been developed to measure cardiac output. Because rebreathing increases respiratory effort, we investigated whether a newly developed system with 35 s rebreathing causes a lesser increase in respiratory effort under partial ventilatory support than does the conventional system with 50 s rebreathing. We also investigated whether the shorter rebreathing period affects the accuracy of cardiac output measurement.
Method
Once a total of 13 consecutive post-cardiac-surgery patients had recovered spontaneous breathing under pressure support ventilation, we applied a partial carbon dioxide rebreathing technique with rebreathing of 35 s and 50 s in a random order. We measured minute ventilation, and arterial and mixed venous carbon dioxide tension at the end of the normal breathing period and at the end of the rebreathing periods. We then measured cardiac output using the partial carbon dioxide rebreathing technique with the two rebreathing periods and using thermodilution.
Results
With both rebreathing systems, minute ventilation increased during rebreathing, as did arterial and mixed venous carbon dioxide tensions. The increases in minute ventilation and arterial carbon dioxide tension were less with 35 s rebreathing than with 50 s rebreathing. The cardiac output measures with both systems correlated acceptably with values obtained with thermodilution.
Conclusion
When patients breathe spontaneously the partial carbon dioxide rebreathing technique increases minute ventilation and arterial carbon dioxide tension, but the effect is less with a shorter rebreathing period. The 35 s rebreathing period yielded cardiac output measurements similar in accuracy to those with 50 s rebreathing.
==== Body
Introduction
A partial carbon dioxide rebreathing technique has been developed to estimate cardiac output (CO) in mechanically ventilated patients undergoing surgery [1,2] or intensive care [3,4]. We previously reported that 50 s carbon dioxide rebreathing resulted in increased minute ventilation (VE) and an irregular respiratory pattern [4]. Recently, an improved system with a shorter rebreathing time (35 s) was developed and is replacing the 50 s rebreathing system. We reasoned that shortening the carbon dioxide rebreathing period would lessen the increases in arterial carbon dioxide tension (PaCO2) and respiratory effort during carbon dioxide rebreathing. We were concerned, however, that measurement of CO might be compromised by a shorter rebreathing period because there would be smaller changes in the measured variables, fewer sampled breaths and incomplete equilibrium [5]. We designed the present prospective study to investigate how, in spontaneously breathing patients, the shorter carbon dioxide rebreathing period affects respiratory effort during rebreathing and how it affects the accuracy of CO measurement.
Materials and methods
The study was approved by the ethics committee of the National Cardiovascular Center (Osaka, Japan), and written informed consent was obtained from each patient.
Patients
Thirteen consecutive patients (age 39–79 years) who had undergone elective cardiovascular surgery were enrolled in the study (Table 1). Enrolment criteria were similar to those of previous studies [3,4]: insertion of a pulmonary artery catheter, stable haemodynamics in the intensive care unit (ICU) and no leakage around the endotracheal tube. We excluded those patients who had central nervous system disorders, who might be adversely affected by induced hypercapnia, or who exhibited severe tricuspid regurgitation. After admission to the ICU each patient was ventilated with an 8400STi ventilator (Bird Corp., Palm Springs, CA, USA). Initial ventilatory settings were synchronized intermittent mandatory ventilation plus pressure support ventilation (PSV), volume controlled ventilation, tidal volume (VT) 10 ml/kg, respiratory rate 10 breaths/min, inspiratory time 1.0 s, positive end-expiratory pressure 4 cmH2O, and PSV 10 cmH2O. The inspired fraction of oxygen was adjusted by attending physicians to maintain arterial oxygen tension greater than 100 mmHg. Using an inspiratory hold technique, we measured the effective static compliance and resistance of the respiratory system (Table 1) [6]. In all patients, arterial blood pressure, heart rate, pulmonary artery pressure, central venous pressure and pulse oximeter signal (PM-1000; Nellcor Inc., Hayward, CA, USA) were continuously monitored. Patients were sedated with propofol (2–3 mg/kg per hour). After waiting 1–2 hours for haemodynamics to stabilize, we decreased the dosage of propofol to 1–2 mg/kg per hour.
Study protocol
As each patient recovered spontaneous breathing, we gradually decreased synchronized intermittent mandatory ventilation rates, finally changing the ventilatory mode to continuous positive airway pressure with PSV at 10 cmH2O. The measurement protocol was started when the recruited patients satisfied the following conditions: recovery of cough reflex; VT ≥ 8 ml/kg and respiratory rate ≤ 20 breaths/min; arterial blood gas of pH 7.35–7.45; PaCO2 35–45 mmHg; and arterial oxygen tension ≥ 100 mmHg at an inspired fraction of oxygen ≤ 0.5. We applied two systems of noninvasive partial carbon dioxide rebreathing technique in a random order. After waiting for at least 15 min, we recorded respiratory and haemodynamic data. Because the stimuli of partial carbon dioxide rebreathing increased spontaneous breathing, we recorded the data as displayed on the graphic monitors of the ventilators for respiratory rate and VE at the end of the normal breathing period and at the end of the rebreathing periods (Fig. 1). At the same times arterial blood was drawn via radial artery cannulation and mixed venous blood via pulmonary artery catheter; samples were analyzed with a calibrated blood gas analyzer (ABL 505; Radiometer, Copenhagen, Denmark).
Cardiac output measurements
We randomly applied two systems of noninvasive partial carbon dioxide rebreathing technique to measure CO (CONI): 35 s rebreathing (version 4.5, fast mode; Novametrix Medical Systems Inc., Wallingford, CT, USA) and 50 s rebreathing (version 4.2, fast mode). Although the durations of carbon dioxide rebreathing were different, both the total cycle (3 min) and the calculation algorithm were the same. Sensors for noninvasive partial carbon dioxide rebreathing technique (NICO2) were placed between the tracheal tube and Y-piece. The principle underlying this technique is described in detail elsewhere [3-5]. Briefly, carbon dioxide production (VCO2) is calculated on a breath-by-breath basis and a differential Fick equation is applied to establish the relationship between VCO2 and CO as follows:
VCO2 = CO × (CvCO2 – CaCO2) (1)
Where CvCO2 is the carbon dioxide content in mixed venous blood, and CaCO2 is the carbon dioxide content in arterial blood. Assuming that both CO and CvCO2 remains constant during carbon dioxide rebreathing and that the change in CaCO2 between normal breathing and carbon dioxide rebreathing is proportional to the changes in PaCO2 and end-tidal carbon dioxide pressure (PETCO2), the following equation is substituted for the previous one:
CO = ΔVCO2/(S × ΔPETCO2) (2)
Where ΔVCO2 is the change in VCO2 and ΔPETCO2 is the change in PETCO2 between normal breathing and carbon dioxide rebreathing, and S is the slope of the carbon dioxide dissociation curve from haemoglobin. After compensating, from the pulse oximeter signal, for the intrapulmonary shunt fraction, the partial carbon dioxide rebreathing technique obtains values for CO.
After we had acquired CONI data, we measured thermodilution CO (COTD) via a 7.5-Fr pulmonary artery catheter (Abbott Laboratories, North Chicago, IL, USA; Fig. 1). During the latter half of the normal breathing period, injection of 10 ml cold saline (0°C) was done three times and the values obtained were averaged. We carefully standardized the timing of bolus injections to after the first half of the expiratory phase [7].
Statistical analysis
Data are presented as mean ± standard deviation, or as the median and interquartile range if the data were skewed. Comparison of respiratory rate, VE, PaCO2 and mixed venous partial carbon dioxide tension (PCO2) between different conditions (35 s versus 50 s rebreathing, and normal breathing versus rebreathing) were conducted using the Friedman test and the Wilcoxon signed rank test. We evaluated the agreement among CONI with 35 s rebreathing, CONI with 50 s rebreathing and COTD using Bland-Altman analysis [8]. P < 0.05 was considered statistically significant.
Results
Respiratory loads
Respiratory and blood gas results are summarized in Table 2. There was no significant difference in respiratory rate, VE, PaCO2 and mixed venous PCO2 during normal breathing between 35 s rebreathing and 50 s rebreathing (Table 2). With either duration of rebreathing, respiratory rate and VE increased during rebreathing. Similarly, the values for PaCO2 and mixed venous PCO2 were higher at the end of the rebreathing period. The changes in VE and PaCO2 due to rebreathing were significantly less with 35 s rebreathing than with 50 s rebreathing (Fig. 2).
Cardiac output
The results of Bland-Altman analysis for 35 s and 50 s rebreathing systems are summarized in Fig. 3. The CO measured using both systems exhibited similar agreement (bias and precision, respectively: 0.02 l/min and 1.06 l/min with 35 s rebreathing, and -0.34 l/min and 1.08 l/min with 50 s rebreathing) with values measured by thermodilution. When comparing the CO between 35 s rebreathing and 50 s rebreathing, bias was 0.26 l/min and precision was 0.51 l/min (Fig. 3c).
Discussion
The main findings of the present study, conducted in spontaneously breathing patients, are that respiratory rate, VE, PaCO2 and mixed venous PCO2 increased during the rebreathing period; that increases in VE and PaCO2 during carbon dioxide rebreathing were less with the shorter rebreathing period; and that the two systems, with different rebreathing periods, provided similarly accurate CO measurements.
The NICO2 system is appealing as a noninvasive method for measuring CO in patients in whom pulmonary artery catheterization is not possible or desirable. Because it is now common for ICU patients to receive partial ventilatory support that allows spontaneous breathing [9], we must determine how the reduction in carbon dioxide rebreathing time affects respiratory effort and how accurate the NICO2 system is in such patients.
Respiratory effort
One disadvantage of the partial carbon dioxide rebreathing technique is that rebreathing increases the respiratory effort of spontaneously breathing patients [4]. Consequently, the effect on respiratory effort of different durations of carbon dioxide rebreathing requires clarification. To our knowledge, no other investigations into this issue have been published. First, we found that the increase in PaCO2 during 50 s rebreathing was 5.9 mmHg (median; Fig. 2). These increases were greater than values (2–5 mmHg) previously reported in applications of controlled mechanical ventilation [10,11]. We assume that the greater metabolic rate in awake and spontaneously breathing patients accounted for the higher increase in PaCO2 during carbon dioxide rebreathing. Next, as we had conjectured, the shorter period of carbon dioxide rebreathing resulted in lesser increases in PaCO2 and, as a result, reduced the increases in VE during carbon dioxide rebreathing (Fig. 2). Although NICO2 monitoring is relatively noninvasive under controlled mechanical ventilation, it increases PaCO2 and respiratory effort under partial ventilatory support, even during 35 s rebreathing.
Accuracy of cardiac output measurement
Although we previously found this technique to be less accurate when there were spontaneous breathing efforts [4], in the present study CONI correlated moderately well with COTD. We reason that we were able to obtain more stable VT and VE findings during CO measurement in the present study by using a larger dosage of propofol (1–2 mg/kg per hour) than in the previous study (0.5 mg/kg per hour). It is likely that stable VT and VE resulted in more accurate CO measurement. Gama de Abreu and coworkers [12], using a system different from ours, also reported that results were less precise when there was irregular spontaneous breathing than when respiratory rate and VT were fixed.
Because of smaller changes in the measured variables, fewer sampled breaths and incomplete equilibrium, we expected that the shorter duration of rebreathing would lead to less accurate CO measurement [5]. However, CO measurement with 35 s rebreathing was as accurate as with 50 s rebreathing (Fig. 3). Although the exact reason is unknown, we speculate as follows; Because the CONI value is calculated from the ratio of change in VCO2 and PETCO2 during carbon dioxide rebreathing, the measurement is corrupted by noise and by variations in VT and respiratory rate [5]. Smaller carbon dioxide stimuli during 35 s rebreathing probably result in a more stable ventilatory pattern, whereas the smaller changes in VCO2 and PETCO2 during 35 s rebreathing lead to a poorer signal-to-noise ratio. In the range of durations tested, these two factors might proportionally cancel each other out, resulting in similar accuracy between 35 s rebreathing and 50 s rebreathing.
Limitations
The present study has several limitations. First, we waited for 15 min after applying each NICO2 system with 35 s and 50 s rebreathing. When spontaneous breathing effort is present and VE is changing, more time may be required to attain stable conditions and an accurate CONI. The time course of the increase in PaCO2 after a decrease in VE is much slower than the rate of decrease after an increase in VE [13]. Second, all of the patients included were sedated, but different levels of sedation may result in different responses to carbon dioxide rebreathing. Third, although the patients enrolled in this study exhibited normal lung mechanics (Table 1), critically ill patients with metabolic acidosis may respond differently to carbon dioxide rebreathing [14]. Although we speculate that our findings may be expanded to other patients with stable haemodynamics, and normal lung mechanics and gas exchange, further studies are needed to evaluate the accuracy and reproducibility of the NICO2 system with various levels of sedation and various patient populations. Fourth, the sample size in the study was small and we did not conduct a power analysis to determine the needed sample size. Because we performed multiple measurements in the same individuals, the order of measurements might have affected the results. Finally, the NICO2 algorithm assumes that mixed venous PCO2 remains constant during partial carbon dioxide rebreathing [5]. However, we found that increases in mixed venous PCO2 were larger than those previously reported (Table 2) [15,16]. When mixed venous PCO2 increases during carbon dioxide rebreathing, this must lead to an underestimation in CONI [5]. Further study is needed to clarify the effects of the change in mixed venous PCO2 on the accuracy of CO measurement.
Conclusion
When patients breathe spontaneously, CO measurement using partial carbon dioxide rebreathing technique increases PaCO2 and VE, although shortening the carbon dioxide rebreathing period causes a lesser increase. The two durations of rebreathing result in similar accuracy in measuring CO.
Key messages
• The NICO2 monitor is claimed to measure CO noninvasively using the partial carbon dioxide rebreathing technique.
• When there are spontaneous breaths, partial carbon dioxide rebreathing increases VE and PaCO2.
• Use of a shorter duration of rebreathing (35 s versus 50 s) has smaller effects on respiratory effort in spontaneously breathing patients.
• The shorter duration of carbon dioxide rebreathing system yields a CO measurement that is similar in accuracy to that obtained with the previously used, longer duration of rebreathing.
Abbreviations
CO = cardiac output; ICU = intensive care unit; NICO2 = noninvasive partial CO2 rebreathing technique; PaCO2 = arterial carbon dioxide tension; PCO2 = partial carbon dioxide tension; PETCO2 = end-tidal carbon dioxide tension; PSV = pressure support ventilation; VCO2 = carbon dioxide production; VE = minute ventilation; VT = tidal volume.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
KT designed the study, collected and analyzed the clinical data, and drafted the manuscript. HI designed the study, carried out data collection and analysis, and extensively revised the manuscript. MT designed the study and performed the statistical analysis. TN and YT participated in the analysis and interpretation of data. MN designed the study and extensively revised the manuscript. All authors read and approved the final manuscript.
Acknowledgements
Support was provided solely from departmental sources: Department of Surgical Intensive Care Unit, National Cardiovascular Center, Osaka, Japan.
Figures and Tables
Figure 1 Schedule of measurements. Respiratory rate (RR), minute ventilation (VE), arterial carbon dioxide tension (PaCO2) and mixed venous carbon dioxide tension (PvCO2) were recorded both at the end of the normal breathing period (NB) and at the end of the partial rebreathing period (RB). At the middle of normal breathing period cardiac output using partial carbon dioxide rebreathing technique (CONI) was measured; then, cardiac output using thermodilution technique (COTD) was measured in triplicate and the values were averaged.
Figure 2 Changes in respiratory values in each patient due to carbon dioxide rebreathing. (a) Increases in minute ventilation (VE) due to carbon dioxide rebreathing. (b) Increases in arterial carbon dioxide tension (PaCO2) due to carbon dioxide rebreathing. Medians (triangles) and interquartile ranges are also shown. *P < 0.05 versus 35 s rebreathing.
Figure 3 Bias analysis between cardiac output measurements. (a) Cardiac output obtained by partial carbon dioxide rebreathing of duration 35 s (CONI,35s) and thermodilution technique (COTD). (b) Cardiac output obtained by partial carbon dioxide rebreathing of duration 50 s (CONI,50s) and COTD. (c) CONI,35s and CONI,50s. Dotted lines show bias and limits of agreement between the two methods.
Table 1 Patient profile at study enrolment
Characteristic/parameter Value
Number of patients 13
Male/female 8/5
Age (years) 64 ± 12
Height (cm) 160 ± 11
Body weight (kg) 58 ± 14
Operative time (min) 252 ± 50
Intraoperative dose of fentanyl (μg/kg) 21 ± 8
Carbon dioxide production (ml/min per kg) 2.6 ± 0.2
Dead space fraction 0.48 ± 0.02
Venous admixture fraction 0.08 ± 0.02
CO with thermodilution (l/min) 5.3 ± 2.1
Compliance of the respiratory system (ml/cmH2O) 49.8 ± 14.8
Resistance of the respiratory system (cmH2O·s per l) 12.0 ± 2.9
Background disease
Coronary artery disease 6
Acquired valve disease 6
Thoracic aortic aneurysm 1
Values are expressed as mean ± standard deviation. CO, cardiac output.
Table 2 Respiratory parameters and blood gas analysis at normal breathing and rebreathing
Respiratory and blood gas parameters 35 s system 50 s system
Respiratory rate (breaths/min)
Normal breathing 16 (15–18) 17 (15–17)
Rebreathing 18* (16–22) 19* (16–22)
Minute ventilation (l/min)
Normal breathing 6.6 (5.9–7.4) 6.3 (6.2–7.3)
Rebreathing 8.8* (8.0–11.6) 9.5* (8.2–12.4)
Arterial carbon dioxide tension (mmHg)
Normal breathing 42.1 (41.0–46.9) 42.2 (39.6–48.6)
Rebreathing 46.5* (43.5–52.5) 47.2* (45.9–55.0)
Mixed venous carbon dioxide tension (mmHg)
Normal breathing 46.2 (44.4–52.2) 48.0 (43.9–52.2)
Rebreathing 47.6* (46.1–52.9) 49.0* (47.0–54.4)
Values are expressed as median (interquartile range). *P < 0.05 versus normal breathing.
==== Refs
Kotake Y Moriyama K Innami Y Shimizu H Ueda T Morisaki H Takeda J Performance of noninvasive partial CO2 rebreathing cardiac output and continuous thermodilution cardiac output in patients undergoing aortic reconstruction surgery Anesthesiology 2003 99 283 288 12883400 10.1097/00000542-200308000-00009
Rocco M Spadetta G Morelli A Dell'Utri D Porzi P Conti G Pietropaoli P A comparative evaluation of thermodilution and partial CO2 rebreathing techniques for cardiac output assessment in critically ill patients during assisted ventilation Intensive Care Med 2004 30 82 87 14652718 10.1007/s00134-003-2094-3
Tachibana K Imanaka H Miyano H Takeuchi M Kumon K Nishimura M Effect of ventilatory settings on accuracy of cardiac output measurement using partial CO2 rebreathing Anesthesiology 2002 96 96 102 11753008 10.1097/00000542-200201000-00021
Tachibana K Imanaka H Takeuchi M Takauchi Y Miyano H Nishimura M Noninvasive cardiac output measurement using partial carbon dioxide rebreathing is less accurate at settings of reduced minute ventilation and when spontaneous breathing is present Anesthesiology 2003 98 830 837 12657842 10.1097/00000542-200304000-00007
Capek JM Roy RJ Noninvasive measurement of cardiac output using partial CO2 rebreathing IEEE Trans Biomed Eng 1988 35 653 661 3139547 10.1109/10.7266
Tobin MJ Van de Graaff WB Tobin MJ Monitoring of lung mechanics and work of breathing Principles and Practice of Mechanical Ventilation 1994 New York: McGraw-Hill 967 1003
Magder S Tobin MJ Cardiac output Principles and Practice of Intensive Care Monitoring 1998 New York: McGraw-Hill 797 810
Bland JM Altman DG Statistical methods for assessing agreement between two methods of clinical measurement Lancet 1986 1 307 310 2868172
Esteban A Anzueto A Alía I Gordo F Apezteguía C Pálizas F Cide D Goldwaser R Soto L Bugedo G for the Mechanical Ventilation International Study Group How is mechanical ventilation employed in the intensive care unit? An international utilization review Am J Respir Crit Care Med 2000 161 1450 1458 10806138
Maxwell RA Gibson JB Slade JB Fabian TC Proctor KG Noninvasive cardiac output by partial CO2 rebreathing after severe chest trauma J Trauma 2001 51 849 853 11706330
van Heerden PV Baker S Lim SI Weidman C Bulsara M Clinical evaluation of the non-invasive cardiac output (NICO) monitor in the intensive care unit Anaesth Intensive Care 2000 28 427 430 10969371
Gama de Abreu M Melo MFV Giannella-Neto A Pulmonary capillary blood flow by partial CO2 rebreathing: importance of the regularity of the respiratory pattern Clin Physiol 2000 20 388 398 10971551 10.1046/j.1365-2281.2000.00271.x
Nunn JF Nunn JF Carbon dioxide Nunn's Applied Respiratory Physiology 1993 4 Oxford: Butterworth-Heinemann 219 246
Nunn JF Nunn JF Control of breathing Nunn's Applied Respiratory Physiology 1993 4 Oxford: Butterworth-Heinemann 90 116
Nilsson LB Eldrup N Berthelsen PG Lack of agreement between thermodilution and carbon dioxide-rebreathing cardiac output Acta Anaesthesiol Scand 2001 45 680 685 11421824 10.1034/j.1399-6576.2001.045006680.x
Odenstedt H Stenqvist O Lundin S Clinical evaluation of a partial CO2 rebreathing technique for cardiac output monitoring in critically ill patients Acta Anaesthesiol Scand 2002 46 152 159 11942862 10.1034/j.1399-6576.2002.t01-1-460205.x
|
16277720
|
PMC1297627
|
CC BY
|
2021-01-04 16:04:56
|
no
|
Crit Care. 2005 Sep 7; 9(5):R569-R574
|
utf-8
|
Crit Care
| 2,005 |
10.1186/cc3801
|
oa_comm
|
==== Front
Crit CareCritical Care1364-85351466-609XBioMed Central London cc38031627772110.1186/cc3803ResearchA systematic evaluation of the quality of meta-analyses in the critical care literature Delaney Anthony [email protected] Sean M 2Ferland Andre 3Manns Braden 4Laupland Kevin B 5Doig Christopher J 61 Staff Specialist, Department of Intensive Care Medicine, Royal North Shore Hospital, Sydney, NSW, Australia2 Fellow, Department of Critical Care Medicine, University of Calgary, Calgary, Alberta, Canada3 Clinical Associate Professor, Departments of Critical Care Medicine and Medicine, University of Calgary, Calgary, Alberta, Canada4 Assistant Professor, Departments of Medicine and Community Health Sciences, University of Calgary, Calgary, Alberta, Canada5 Assistant Professor, Departments of Critical Care Medicine, Medicine and Community Health Sciences, University of Calgary, Calgary, Alberta, Canada6 Associate Professor, Departments of Critical Care Medicine, Medicine and Community Health Sciences, University of Calgary, Calgary, Alberta, Canada2005 9 9 2005 9 5 R575 R582 5 7 2005 2 8 2005 8 8 2005 9 8 2005 Copyright © 2005 Delaney 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
Meta-analyses have been suggested to be the highest form of evidence available to clinicians to guide clinical practice in critical care. The purpose of this study was to systematically evaluate the quality of meta-analyses that address topics pertinent to critical care.
Methods
To identify potentially eligible meta-analyses for inclusion, a systematic search of Medline, EMBASE and the Cochrane Database of Systematic Reviews was undertaken, using broad search terms relevant to intensive care, including: intensive care, critical care, shock, resuscitation, inotropes and mechanical ventilation. Predetermined inclusion criteria were applied to each identified meta-analysis independently by two authors. To assess report quality, the included meta-analyses were assessed using the component and overall scores from the Overview Quality Assessment Questionnaire (OQAQ). The quality of reports published before and after the publication of the QUOROM statement was compared.
Results
A total of 139 reports of meta-analyses were included (kappa = 0.93). The overall quality of reports of meta-analyses was found to be poor, with an estimated mean overall OQAQ score of 3.3 (95% CI; 3.0–3.6). Only 43 (30.9%) were scored as having minimal or minor flaws (>5). We noted problems with the reporting of key characteristics of meta-analyses, such as performing a thorough literature search, avoidance of bias in the inclusion of studies and appropriately referring to the validity of the included studies. After the release of the QUOROM statement, however, an improvement in the overall quality of published meta-analyses was noted.
Conclusion
The overall quality of the reports of meta-analyses available to critical care physicians is poor. Physicians should critically evaluate these studies prior to considering applying the results of these studies in their clinical practice.
==== Body
Introduction
One of the challenges that faces critical care physicians is staying up to date with the current state of knowledge, in a field that has a broad scope of practice and time dependency for many of the interventions provided. Traditional sources of information such as narrative review articles, medical textbooks and the clinical opinion of experts are often at odds with the best current available evidence [1,2]. Systematic reviews in general, and meta-analyses in particular, have been suggested as one solution to this problem [3]. Some authorities have suggested that systematic reviews and meta-analyses are the highest form of published evidence available to clinicians [4].
There are numerous incidences, however, where meta-analyses have pooled results from small trials with disparate results, and produced conflicting evidence [5-7], as well as meta-analyses that have produced results that were in conflict with the results of subsequent large randomised clinical trials (RCTs). [8-11]. When this occurs it causes difficulties for clinicians trying to apply the best available evidence in the care of their patients, as it is not clear which is the best evidence to follow. As a result, doubts have been raised about the reliability of using meta-analyses to guide clinical practice. [12-14].
If clinicians are to have confidence that the results of meta-analyses can be used to guide clinical practice, then the reports of these studies need to be of a high quality. The Overview Quality Assessment Questionnaire (OQAQ) [15] is the only validated instrument available to grade the quality of review articles [16]. It has been used to grade the quality of reports of review articles in a number of fields related to critical care. [17-19].
There were three main aims of this study. First, to describe the quality of the reports of meta-analyses that are available to critical care clinicians using the OQAQ. Second, we hypothesized that the publication of the Quality of Reporting of Meta-analyses statement (QUOROM), published in 1999 [20], that was meant to improve reporting and performance of meta-analyses, might have resulted in an improvement in the quality of meta-analyses. As such, the effect of the publication of the QUOROM statement [20] on the quality of these reports was also examined. Finally, to place the results of this assessment in a broader context, the quality of the reports of meta-analyses in the critical care literature was compared to the quality of the reports of meta-analyses and systematic reviews published in the fields of emergency medicine, anaesthesia and general surgery.
Materials and methods
Study sample
The search for reports of meta-analyses that addressed issues pertinent to critical care medicine was conducted using the Medline database using the PubMed interface, as well as Medline, EMBASE and the Cochrane Database of Systematic Reviews using the OVID interface. Meta-analyses were considered to be any study that statistically integrated the results of a number of primary trials, randomised clinical trials or observational studies. The search terms were individualised for each database and included terms for: critical care, critical illness, intensive care, shock, resuscitation, inotropes and mechanical ventilation. This was combined with sensitive filters to identify meta-analyses [21,22]. Searches were limited to human subjects and reports published in English. The search was limited to articles published between January 1 1994 and December 31 2003, and was completed in August 2004. Full details of the search strategy are available as an additional data file (Additional file 1).
Study selection
One reviewer examined the titles and abstracts of all articles returned by the search to identify potentially eligible articles. All potentially eligible studies were then retrieved and the full-text article was reviewed to determine if it met the pre-determined inclusion criteria. Assessments were conducted independently by two reviewers, with disagreements resolved by discussion, or by resort to a third reviewer if consensus could not be reached. The inclusion criteria were: the study addressed an issue pertinent to critical care medicine; study population in the included studies were adult patients; study population in the included studies were human participants; the systematic review used statistical methods to produce a summary result; the report was published in English; the report of the study was first published between 1994 and 2003.
Data extraction
Two reviewers independently extracted data from the included studies. Data were extracted from the reports regarding the individual components of the OQAQ, and a summary score was then determined. Within the OQAQ instrument, there are nine individual items relating to the methodological quality of the meta-analysis, including the performance of a thorough search, the avoidance of bias in the inclusion of studies, appropriately referring to the validity of the included studies, appropriately combining the results and drawing appropriate conclusions from the data. Each report was assessed as to whether it clearly met the criterion, clearly did not meet the criterion, or it partially met or it was unclear whether it had met the criterion. After assessment of each of the nine component questions, a final overall score was given, based on the answers to the previous nine questions on a scale of 1 to 7, with 7 indicating no flaws, and a score of ≥ 5 indicating that the study has only minimal or minor flaws. The full details of the OQAQ scoring questionnaire are available as Additional file 2. Data were also collected regarding the date of publication. The QUOROM statement was first published in November 1999 [20], so to allow a reasonable lag time for studies in progress or under review for publication to finish and the report to be published, those reports published prior to December 31, 2000 were adjudicated as the 'pre-QUOROM' group and those published after January 1, 2001 as the 'post-QUOROM' group. The source of the publication was also classified as to whether the publication was in a critical care journal or a journal that primarily dealt with another area of medical practice.
Analysis
The primary analysis of the data was descriptive. The proportion of reports that met each of the criteria was determined and tabulated. The estimated mean overall quality summary score was calculated.
To assess whether the overall quality of reviews had improved after publication of the QUOROM statement, the overall quality score of reports published prior to the publication of the QUOROM statement was compared to the overall quality score of reports published after the QUOROM statement. Data from this study were compared with the data published in previous reports from the emergency medicine [17], anaesthesia [19] and general surgery [18] literature.
Agreement on the inclusion of studies was assessed using a kappa statistic. The results were summarized with means and standard deviations for normally distributed data and medians and interquartile ranges for non-normally distributed data. The means of normally distributed variables were compared using unpaired t-tests. Proportions were compared using Fisher's exact test. All statistical tests were two-sided with a p-value of < 0.05 considered significant unless otherwise stated. Statistical calculations were performed using STATA 8.2 (College Station, TX, USA).
Results
Search results
A total of 7,935 articles were returned by the initial search. Of these 7,723 were deemed ineligible after inspection of the titles and abstracts. A total of 212 unique reports were retrieved for further review, and 139 were considered to be eligible for inclusion. Agreement on the inclusion of articles occurred in 97.8% of cases, which gave a kappa = 0.93 (p < 0.0005). A wide range of topics were addressed by the meta-analyses, the most common of which are shown in Table 1. A full list of the references is available as Additional file 3. The reasons for exclusion of reports, and the flow of studies are shown in Fig. 1. Table 2 shows the source of publication of the reports. The reports of meta-analyses were published in a wide variety of sources, with the majority of reports being published in sources that were not classified as critical care journals.
The overall quality meta-analyses in the critical care literature
Agreement was reached on the scoring of all component scores and the overall quality scores without the need for resort to a third reviewer. Table 3 contains the summary results of the quality assessment of all meta-analyses that addressed topics relevant to critical care. The results for each individual study are shown in Additional file 4. Of note is that the weakest areas within the included meta-analyses were the failure to report whether a comprehensive literature search was conducted and failure to report how bias in the inclusion of studies was avoided, with only 35.3% of reports adequately fulfilling these criteria. Less than half of the reports referred to the validity of the included studies by appropriate criteria in the text.
The overall quality scores are shown in Table 4. The estimated mean overall quality score for meta-analyses published in the critical care literature from 1994 to 2003 was 3.3 (95% CI; 3.0–3.6). A total of 43 (30.9%) reports had minimal or minor flaws as shown by an overall score of ≥ 5, and 96 (69.1%) reports had major or extensive flaws, scoring ≤ 4 on the overall quality summary score.
Has the quality of meta-analyses in the critical care literature improved over time?
An increasing number of reports of meta-analyses were published in the later years of the study (Fig. 2). There were 59 reports of meta-analyses published on or before December 31 2000 that were classified as 'pre-QUOROM' and 80 reports of meta-analyses published on or after January 1 2001 that were classified as 'post-QUOROM'. Table 5 shows the number and proportion of reports that clearly fulfilled each of the components of the OQAQ (i.e. scored 'yes'). The failure to refer to the validity of the included studies occurred in 39% and 52.5% of reports pre- and post-QUOROM, respectively (p = 0.13 Fishers's exact test). All other components showed a significant improvement after the publication of the QUOROM statement.
The estimated mean quality score of the reports was 2.8 (95% CI; 2.3–3.2), and 3.7 (95% CI; 3.3–4.1) pre- and post-QUOROM, respectively. This represented an estimated improvement of 0.96 (95% CI; 0.4–1.6, p = 0.0018 two sided t-test).
Comparison of the quality of meta-analyses in the critical care literature and in the emergency medicine, anaesthesia and general surgical literature
Three previous published studies have assessed the quality of reports of meta-analyses in the emergency medicine, anaesthesia, and general surgery fields. These studies included 29 reports of meta-analyses published in five emergency medicine journals from 1988 to 1998. [17], 82 reports of meta-analyses that addressed issues pertinent to anaesthesia identified up until June 1999, from a Medline search not limited solely to anaesthesia journals [19], and 51 meta-analyses that addressed general surgery issues from 1997 to 2002 [18]. The estimates of the mean overall quality scores for the emergency medicine, anaesthesia, general surgery and critical care, as well as the estimates of the proportions of reports that had minimal or minor flaws only (i.e. had scored ≥ 5 on the OQAQ overall quality score) are shown in Table 6. It should be noted that the overall quality of reports was poor for each discipline, with the estimated mean OQAQ scores being <5 in each discipline and with less than 50% of all reports having a score of ≥ 5 in each discipline.
Discussion
Many reports of meta-analyses address topics pertinent to critical care available to physicians. The number of reports is increasing with time, as has been demonstrated in a number of other studies [19,23]. If critical care physicians are to use these reports to guide their clinical practice, they cannot rely on browsing solely from critical care journals, as the majority of reports of meta-analyses are not published in critical care journals. The result of this study raises questions about the quality of those reports, however, and therefore whether they can be recommended without qualification as the best evidence to guide clinical practice at the present time.
It was found that the overall quality of reports of meta-analyses in addressing critical care topics is generally poor. Studies with an overall OQAQ score of 5 or more are regarded as having minimal or minor flaws. The average score of the reports in the critical care literature was only 3.3, so clearly the majority of reports are of an inferior quality. Less than one-third of reports had a score of 5 or more. This places an important caveat on the recommendation that these reports are the highest quality evidence available. Clinicians must still critically appraise the reports prior to consideration of the recommendations made in the report of the meta-analysis [4].
While the overall quality of reports is of some interest, the results of the component scores of the OQAQ may offer more insight into the areas that should be improved. The areas that were most poorly attended to were the conduct of a comprehensive search, the avoidance of bias in the selection of studies and the assessment of the validity of all the included studies. These are crucial elements in the conduct of a meta-analysis, without which the results of the meta-analysis will be questionable. Authors contemplating conducting meta-analyses and reviewers assessing studies for publication may be able to focus on these aspects of the conduct and reporting of meta-analyses in order to have the greatest impact on improving their overall quality.
There is some cause for optimism, however. Clearly the quality of reports of meta-analyses has improved over time. While it is hard to pinpoint the exact cause of the improvement, it may be that the dissemination of guidelines such as the QUOROM statement [20] has been associated with an improvement in the quality of reports. A similar improvement in the quality of reports has been found with regards to the quality of reports of RCTs following the publication of the Consolidated Standards of Reporting Trials (CONSORT) statement [24]. It is also possible that increased attention paid to the general methodological quality of reports by journal editors and reviewers has also played a role. Both of these factors may be contributing to a general global trend for better quality research. Authors, reviewers and journal editors should be encouraged to follow these guidelines in the hope that a more standard, high quality report of this type of study will become the norm, and clinicians can spend more time considering the results of the meta-analysis, rather than scrutinizing the methodological quality of the report.
It was found that the quality of the meta-analyses in the critical care literature was comparable to the quality of reviews published in the emergency medicine [17], the anaesthesia [19] and the general surgery literature [18]. There were some differences in the conduct of this study compared to the conduct of the previous studies that makes comparing the results somewhat problematic. While this makes it difficult to draw strong conclusions regarding the comparative quality of the reviews in the different fields, the lower quality of the scores in the emergency medicine literature may reflect the temporal trend seen in the critical care literature. The slightly higher scores in the anaesthesia literature may reflect differences in implementation of the scoring system. It should suffice to note that there is ample room for improvement in the quality of the reviews in each of the fields.
There are a number of limitations to this study. Critical care is an area of medicine that covers a wide variety of fields. As such, sampling the meta-analyses that address critical care topics is difficult. While attempts were made to include a diverse range of search terms, it is possible that some studies were not identified by the search strategy employed in this study. The studies not included could have different characteristics to those included, although it is unlikely that they are systematically different. It is also important to note that while the OQAQ is the instrument most widely used to grade the quality of meta-analyses and systematic reviews, it has not had extensive validation testing, nor validation testing since the establishment of the QUORUM guidelines [16].
While it would be hoped that high quality meta-analyses would produce the results that are concordant with the results of other high quality evidence, such as well-conducted, large RCTs, this is not necessarily the case. Due to differences in the interventions tested, populations, outcomes measured and other confounding issues, it is difficult to determine when meta-analyses will agree with RCTs that address the same issue. Previous studies [12,25,26] that have examined the relationship between the results of the meta-analyses and large RCTs have not addressed the issue of the methodological quality of the meta-analyses or the RCTs, another issue that may confound this relationship. Uncertainty about when the results of meta-analyses can be used to guide clinical practice rather than a future research agenda remains and further methodological investigation in this area is still needed.
Conclusion
A large number of reports of meta-analyses address issues pertinent to critical care, and these numbers are increasing over time. These reports appear in a wide variety of sources. Physicians wishing to use the results of these studies to guide their clinical practice would need to employ strategies other than browsing critical care journals in order to access all the relevant reports. The overall quality of the reports is low, and the majority of reports of meta-analyses are not of a methodological quality whereby the results of the study could be reliably used to guide clinical practice. There is, however, some hope that improvement in the quality of the reports subsequent to the publication of the QUOROM guidelines will continue, and authors and reviewers should be encouraged to follow established methodological guidelines for the conduct and reporting of these studies.
Key messages
• The overall quality of meta-analyses that address topics pertinent to critical care medicine is poor.
• Meta-analyses need to be critically appraised prior to the results being considered applicable to guide clinical practice.
• The main areas that were reported to be deficient were the conduct of a reasonably thorough search, the avoidance of bias in the inclusion of studies and referring to the validity of the included studies appropriately. Authors should pay greater attention to these aspects of the meta-analytic process in the conduct and reporting of their study.
• Authors, reviewers and journal editors could improve the reporting of meta-analyses by more closely adhering to the established methodological guidelines such as the QUOROM statement.
Abbreviations
OQAQ = Overview Quality Assessment Questionnaire; QUOROM = Quality of Reporting of Meta-analyses; RCT = randomised clinical trial.
Competing interests
This work was part of a thesis submitted to the faculty of graduate studies in partial fulfilment of the requirement for the degree of Master of Science, Department of Community Health Science, University of Calgary.
Authors' contributions
AD and CD conceived the study. AD was responsible for the design of the study, searching for studies, selection of studies, data acquisition and analysis. CD was responsible for the design of the study, selection of studies and data acquisition. SB was responsible for the selection of studies. AF, BM and KL were all involved in the design of the study. All authors were involved in the drafting of the manuscript and gave approval of the final version.
Supplementary Material
Additional File 1
Word file (doc) providing full details of the search strategy to Identify Meta-analyses pertinent to Critical Care Medicine.
Click here for file
Additional File 2
Word file (doc) providing full details of the OQAQ scoring questionnaire.
Click here for file
Additional File 3
Word file (doc) providing a full list of the references used in this study.
Click here for file
Additional File 4
Spreadsheet (xls) listing the quality assessment results for each individual meta-analysis included in this study.
Click here for file
Acknowledgements
The authors would like to thank Glynis Hawkins and Celia Bradford for their help in revising the manuscript.
Figures and Tables
Figure 1 Flow chart showing results of search and reasons for exclusion of reports. ICU, intensive care unit.
Figure 2 Frequency histogram showing the number of reports of meta-analyses addressing critical care issues per year, 1994 to 2003.
Table 1 Common topics addressed by meta-analyses in the critical care literature
Topic Number of reports
Nutrition 13
Fluid therapy 11
Central venous catheters 10
Traumatic brain injury 10
Variceal bleeding 9
Non-invasive ventilation 8
Selective decontamination 7
Oxygen delivery 6
Intervention in sepsis and septic shock 6
Cardiac arrest 5
Therapy for acute renal failure 4
Blood transfusion 3
Sedation 3
Low tidal volume ventilation 2
Eclampsia 2
Nitric oxide 2
Deep Venous Thrombosis prophylaxis 2
Heliox for acute asthma 2
Stress ulcer prophylaxis 2
Other issues 32
Table 2 Source of publication of reports of meta-analyses that address critical care issues
Source of publication Number of reports Percentage
Cochrane database of systematic reviews 37 26.6%
Critical care journals 36 25.9%
Specialty medicine journals 29 20.9%
General medicine journals 15 10.8%
Anaesthesia journals 5 3.6%
General surgery journals 5 3.6%
Nursing journals 3 2.2%
Specialty surgery journals 1 0.7%
Other journals 8 5.8%
Table 3 Overview Quality Assessment Questionnaire component score results
OQAQ question No (n (%)) Partial or can't tell (n(%)) Yes (n(%))
Were the search methods used to find evidence on the primary question(s) stated 5 (3.6) 3 (2.2) 131 (94.2)
Was the search for evidence reasonably comprehensive? 23 (16.6) 67 (48.2) 49 (35.3)
Were the criteria used for deciding which studies to include in the overview reported? 14 (10.1) 7 (5.0) 118 (84.9)
Was bias in the selection of studies avoided? 27 (19.4) 63 (45.3) 49 (35.3)
Were the criteria used for assessing the validity of the included studies reported? 38 (27.3) 8 (5.8) 93 (66.9)
Was the validity of all the studies referred to in the text assessed using appropriate criteria? 45 (32.4) 29 (20.9) 65 (46.8)
Were the methods used to combine the findings of the relevant (to reach a conclusion) reported? 12 (8.6) 17 (12.2) 110 (79.1)
Were the findings of the relevant studies combined appropriately relative to the primary question of the overview? 14 (10.1) 37 (26.6) 88 (63.3)
Were the conclusions made by the author(s) supported by the data and/or analysis reported in the overview? 6 (4.3) 29 (20.9) 104 (74.8)
Data expressed as total number of reports with that score (percent).
Table 4 Overview Quality Assessment Questionnaire summary score results
Overall OQAQ score n (%)
1 26 (18.7)
2 37 (26.6)
3 10 (7.2)
4 23 (16.6)
5 26 (18.7)
6 10 (7.2)
7 7 (5.0)
Data expressed as total number of reports receiving that score (percent).
Table 5 Comparison of reports that fulfilled each OQAQ component pre-QUOROM and post-QUOROM
OQAQ question Pre-QUOROM (n (%)) Post-QUOROM (n (%)) p-value
Were the search methods used to find evidence on the primary question(s) stated 52 (88.1) 79 (98.8) 0.010
Was the search for evidence reasonably comprehensive? 14 (23.7) 35 (43.8) 0.019
Were the criteria used for deciding which studies to include in the overview reported? 44 (74.6) 74 (92.5) 0.007
Was bias in the selection of studies avoided? 15 (25.4) 34 (42.5) 0.048
Were the criteria used for assessing the validity of the included studies reported? 33 (55.9) 60 (75.0) 0.028
Was the validity of all the studies referred to in the text assessed using appropriate criteria? 23 (39.0) 42 (52.5) 0.13
Were the methods used to combine the findings of the relevant (to reach a conclusion) reported? 40 (67.8) 70 (87.5) 0.006
Were the findings of the relevant studies combined appropriately relative to the primary question of the overview? 29 (49.2) 59 (73.8) 0.004
Were the conclusions made by the author(s) supported by the data and/or analysis reported in the overview? 35 (59.3) 69 (86.3)0 <0.0005
Data expressed as the number of reports that scored 'yes' for each component (percent). P-values derived from Fisher's exact test. OQAQ, Overview Quality Assessment Questionnaire; QUOROM, Quality of Reporting of Meta-analyses.
Table 6 Comparison of the overall quality of reports of meta-analyses in the emergency medicine, anaesthesia and critical care literature
Emergency medicine Anaesthesia General surgery Critical care
Mean overall OQAQ score (95% CI) 2.7 (2.1–3.2) 4.3 (3.8–4.7) 3.3 (2.8–3.9) 3.3 (3.0–3.6)
Proportion of reports with an overall OQAQ score ≥5 (95% CI) 13.8 (3.9–31.6) 41.5 (30.7–52.9) 25.5 (14.3–39.6) 30.9 (23.4–39.3)
==== Refs
Mulrow CD The medical review article: state of the science Ann Intern Med 1987 106 485 488 3813259
Antman EM Lau J Kupelnick B Mosteller F Chalmers TC A comparison of results of meta-analyses of randomized control trials and recommendations of clinical experts. Treatments for myocardial infarction JAMA 1992 268 240 248 1535110 10.1001/jama.268.2.240
Cook DJ Meade MO Fink MP How to keep up with the critical care literature and avoid being buried alive Crit Care Med 1996 24 1757 1768 8874317 10.1097/00003246-199610000-00026
Guyatt GH Haynes RB Jaeschke RZ Cook DJ Green L Naylor CD Wilson MC Richardson WS Users' Guides to the Medical Literature: XXV. Evidence-based medicine: principles for applying the Users' Guides to patient care. Evidence-Based Medicine Working Group JAMA 2000 284 1290 1296 10979117 10.1001/jama.284.10.1290
McIntyre LA Fergusson DA Hebert PC Moher D Hutchison JS Prolonged therapeutic hypothermia after traumatic brain injury in adults: a systematic review JAMA 2003 289 2992 2999 12799408 10.1001/jama.289.22.2992
Jadad AR Cook DJ Browman GP A guide to interpreting discordant systematic reviews CMAJ 1997 156 1411 1416 9164400
Henderson WR Dhingra VK Chittock DR Fenwick JC Ronco JJ Hypothermia in the management of traumatic brain injury. A systematic review and meta-analysis Intensive Care Med 2003 29 1637 1644 12915937 10.1007/s00134-003-1848-2
Finfer S Bellomo R Boyce N French J Myburgh J Norton R A comparison of albumin and saline for fluid resuscitation in the intensive care unit N Engl J Med 2004 350 2247 2256 15163774 10.1056/NEJMoa040232
Cook DJ Reeve BK Guyatt GH Heyland DK Griffith LE Buckingham L Tryba M Stress ulcer prophylaxis in critically ill patients. Resolving discordant meta-analyses JAMA 1996 275 308 314 8544272 10.1001/jama.275.4.308
Cook D Guyatt G Marshall J Leasa D Fuller H Hall R Peters S Rutledge F Griffith L McLellan A A comparison of sucralfate and ranitidine for the prevention of upper gastrointestinal bleeding in patients requiring mechanical ventilation. Canadian Critical Care Trials Group N Engl J Med 1998 338 791 797 9504939 10.1056/NEJM199803193381203
Cochrane Injuries Group Albumin Reviewers Human albumin administration in critically ill patients: systematic review of randomised controlled trials BMJ 1998 317 235 240 9677209
LeLorier J Gregoire G Benhaddad A Lapierre J Derderian F Discrepancies between meta-analyses and subsequent large randomized, controlled trials N Engl J Med 1997 337 536 542 9262498 10.1056/NEJM199708213370806
Flather MD Farkouh ME Pogue JM Yusuf S Strengths and limitations of meta-analysis: larger studies may be more reliable Control Clin Trials 1997 18 568 579 9408719 10.1016/S0197-2456(97)00024-X
Celermajer DS Evidence-based medicine: how good is the evidence? Med J Aust 2001 174 293 295 11297118
Oxman AD Guyatt GH Singer J Goldsmith CH Hutchison BG Milner RA Streiner DL Agreement among reviewers of review articles J Clin Epidemiol 1991 44 91 98 1824710 10.1016/0895-4356(91)90205-N
Oxman AD Guyatt GH Validation of an index of the quality of review articles J Clin Epidemiol 1991 44 1271 1278 1834807 10.1016/0895-4356(91)90160-B
Kelly KD Travers A Dorgan M Slater L Rowe BH Evaluating the quality of systematic reviews in the emergency medicine literature Ann Emerg Med 2001 38 518 526 11679863 10.1067/mem.2001.115881
Dixon E Hameed M Sutherland F Cook DJ Doig C Evaluating meta-analyses in the general surgical literature: a critical appraisal Ann Surg 2005 241 450 459 15729067 10.1097/01.sla.0000154258.30305.df
Choi PT Halpern SH Malik N Jadad AR Tramer MR Walder B Examining the evidence in anesthesia literature: a critical appraisal of systematic reviews Anesth Analg 2001 92 700 709 11226105 10.1097/00000539-200103000-00029
Moher D Cook DJ Eastwood S Olkin I Rennie D Stroup DF Improving the quality of reports of meta-analyses of randomised controlled trials: the QUOROM statement. Quality of Reporting of Meta-analyses Lancet 1999 354 1896 1900 10584742 10.1016/S0140-6736(99)04149-5
Shojania KG Bero LA Taking advantage of the explosion of systematic reviews: an efficient MEDLINE search strategy Eff Clin Pract 2001 4 157 162 11525102
Hunt DL McKibbon KA Locating and appraising systematic reviews Ann Intern Med 1997 126 532 538 9092319
Egger M Smith GD Meta-analysis. Potentials and promise BMJ 1997 315 1371 1374 9432250
Moher D Jones A Lepage L Use of the CONSORT statement and quality of reports of randomized trials: a comparative before-and-after evaluation JAMA 2001 285 1992 1995 11308436 10.1001/jama.285.15.1992
Villar J Carroli G Belizan JM Predictive ability of meta-analyses of randomised controlled trials Lancet 1995 345 772 776 7891492 10.1016/S0140-6736(95)90646-0
Cappelleri JC Ioannidis JP Schmid CH de Ferranti SD Aubert M Chalmers TC Lau J Large trials vs meta-analysis of smaller trials: how do their results compare? JAMA 1996 276 1332 1338 8861993 10.1001/jama.276.16.1332
|
16277721
|
PMC1297628
|
CC BY
|
2021-01-04 16:04:56
|
no
|
Crit Care. 2005 Sep 9; 9(5):R575-R582
|
utf-8
|
Crit Care
| 2,005 |
10.1186/cc3803
|
oa_comm
|
==== Front
Crit CareCritical Care1364-85351466-609XBioMed Central London cc38051627772210.1186/cc3805ResearchHospital-acquired sinusitis is a common cause of fever of unknown origin in orotracheally intubated critically ill patients van Zanten Arthur RH [email protected] J Mark [email protected] Martine D [email protected] Bree Remco [email protected] Armand RJ [email protected] Kees H [email protected] Senior Consultant in Internal Medicine and Intensive Care, Department of Intensive Care, Gelderse Vallei Hospital, Ede, The Netherlands2 Senior Consultant in Anaesthesiology and Intensive Care, Department of Anesthesiology and Intensive Care, Norfolk and Norwich University Hospital, Norwich, UK3 Resident in Plastic Surgery, Hospital Hilversum, Hilversum, The Netherlands4 Professor of Intensive Care Medicine, Department of Intensive Care, VU University Medical Center, Amsterdam, The Netherlands5 Senior Consultant in Otolaryngology, Department of Otolaryngology/Head and Neck Surgery, VU University Medical Center, Amsterdam, The Netherlands6 Senior Consultant in Intensive Care, Department of Intensive Care, VU University Medical Center, Amsterdam, The Netherlands2005 13 9 2005 9 5 R583 R590 21 6 2005 27 7 2005 9 8 2005 12 8 2005 Copyright © 2005 van Zanten 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
Sinusitis is a well recognised but insufficiently understood complication of critical illness. It has been linked to nasotracheal intubation, but its occurrence after orotracheal intubation is less clear. We studied the incidence of sinusitis in patients with fever of unknown origin (FUO) in our intensive care unit with the aim of establishing a protocol that would be applicable in everyday clinical practice.
Methods
Sinus X-rays (SXRs) were performed in all patients with fever for which an initial screening (physical examination, microbiological cultures and chest X-ray) revealed no obvious cause. All patients were followed with a predefined protocol, including antral drainage in all patients with abnormal or equivocal results on their SXR.
Results
Initial screening revealed probable causes of fever in 153 of 351 patients (43.6%). SXRs were taken in the other 198 patients (56.4%); 129 had obvious or equivocal abnormalities. Sinus drainage revealed purulent material and positive cultures (predominantly Pseudomonas and Klebsiella species) in 84 patients. Final diagnosis for the cause of fever in all 351 patients based on X-ray results, microbiological cultures, and clinical response to sinus drainage indicated sinusitis as the sole cause of fever in 57 (16.2%) and as contributing factor in 48 (13.8%) patients with FUO. This will underestimate the actual incidence because SXR and drainage were not performed in all patients.
Conclusion
Physicians treating critically ill patients should be aware of the high risk of sinusitis and take appropriate preventive measures, including the removal of nasogastric tubes in patients requiring long-term mechanical ventilation. Routine investigation of FUO should include computed tomography scan, SXR or sinus ultrasonography, and drainage should be performed if any abnormalities are found.
==== Body
Introduction
A large proportion of patients admitted to the intensive care unit (ICU) are likely to develop fever of unknown origin (FUO) at some point of their stay there. Many of these episodes are due to well-recognised hospital-acquired infections such as ventilator-associated pneumonia (VAP) and central venous catheter infections [1,2]. Various diagnostic strategies have been developed to handle such infectious complications in the ICU, many of which have been laid down in hospital or national guidelines [3,4]. However, the potential role of sinusitis as a source of hospital-acquired infections has been much less well studied. It is well recognised that sinusitis can occur as a complication of nasotracheal intubation; however, the incidence of sinusitis in patients after orotracheal intubation is unclear, and the data from the literature have been conflicting [5-8]. We therefore decided to assess the role of sinusitis as a hospital-acquired infection in mechanically ventilated and orotracheally intubated patients admitted to our ICU, in a prospective study using a rigorous protocol with predefined criteria for suspecting sinusitis.
Our aim was not only to assess the incidence of hospital-acquired sinusitis in patients with FUO but also to provide a practical protocol for diagnostic work-up and treatment that could be quickly implemented and easily applied in everyday clinical practice. Diagnostic and therapeutic procedures were therefore chosen in part on the basis of feasibility in daily clinical practice in the care of critically ill patients.
The three main imaging techniques available to establish a diagnosis of sinusitis are a standard sinus X-ray (SXR), ultrasound investigation, and computed tomography (CT) of the sinuses. Of these, a CT scan of the sinus cavities is unquestionably the most accurate and reliable procedure to establish the diagnosis of sinusitis. However, it would be highly impractical and costly to perform repeated CT scans on large numbers of ICU patients on a routine basis. In addition, transporting critically ill patients from the ICU to the radiology unit to perform a CT scan involves some risks [9-11]. A relatively new and promising development is the use of ultrasound as a diagnostic tool for sinusitis in the ICU setting, especially for the detection of maxillary sinusitis [12-15]; however, the reliability of this technique is strongly operator-dependent, and its sensitivity, especially in detecting frontal sinusitis, and overall specificity are relatively low [15-20]. Varonen and associates performed a meta-analysis of studies comparing SXR and ultrasound and reported that ultrasound was slightly less accurate than radiography when compared with the gold standard of sinus puncture [21]. Engels and associates [22] also concluded that, in spite of some limitations, sinus radiography rather than ultrasonography should still be viewed as the most reliable initial screening procedure for sinusitis. The most recent European Position Paper on Rhinosinusitis and Nasal Polyps recommends a combination of SXR followed by sinus puncture and aspiration as the diagnostically most accurate procedure [23].
It should be pointed out that most of these studies were not performed in mechanically ventilated ICU patients, and some studies have suggested that ultrasound has a higher sensitivity and specificity in the ICU setting. However, ultrasound has not so far been widely adopted as a first-line diagnostic tool for sinusitis, and most ICUs use plain SXRs as a first-line screening tool. We therefore chose SXR as our initial screening technique.
Methods
Patients
The study was performed in accordance with guidelines laid down by the hospital ethics committee. All mechanically ventilated adult patients admitted to the surgical wing of our intensive care department during the 18-month study period who spent more than 48 hours in the ICU and who developed fever during their ICU stay were included in the study. Inclusion criteria were as follows: age 18 to 80 years; core temperature 38.5°C (measured in oesophagus, bladder or rectum); not admitted for infections or, if infection was the primary reason for admission, infection treated and temperature normalised for at least 72 hours before recurrence of FUO. At the time of our study, gastric tubes were inserted nasally in most patients. Sedation and analgesia were given in the context of a nurse-driven sedation protocol using the Ramsey score to guide levels of sedation. Exclusion criteria included severe head and facial injuries, skull fractures and immunocompromised patients.
FUO was defined as follows: the cause of fever not immediately clear; the patient was not admitted because of fever or sepsis, or the patient had recovered from one or more previous septic episodes or infections. This means that some patients were admitted with, for example, abdominal sepsis, and developed sinusitis in the course of their admission. Such patients were eligible for inclusion in our study.
Protocol
According to our protocol all patients who developed fever first underwent routine analysis, which included physical examination, drawing of blood cultures and analysis for white blood cell count, and a chest X-ray. Central lines were changed if they had been in place for 1 week or more, or if there were any signs of local infection [2].
An SXR was taken if a cause of fever did not become clear from the above mentioned analysis. An SXR was also taken if a cause of fever was found on routine analysis but when fever persisted for more than 48 hours in spite of antibiotic therapy to exclude sinusitis as the primary cause of fever and/or a contributing factor.
SXRs were taken in two directions, the straight anterior–posterior view (Caldwell view) and the lateral view, using portable devices in the ICU. Additional X-rays were taken if the first X-rays were difficult to interpret, in accordance with our routine for radiodiagnostic procedures [24]. Interpretations were made by the attending physician and confirmed by an independent radiologist. Three categories were used: abnormal (clouded sinuses with fluid), equivocal and normal.
Patients with an abnormal SXR were treated by an ear, nose and throat (ENT) surgeon with diagnostic and therapeutic antral sinus tap and drainage [22,23]. The procedure had to be performed as soon as possible, but a maximum interval of 12 hours was allowed if there was a need for correction of coagulopathy. To prevent accidental contamination the nares were swabbed with chlorhexidine before puncture. Macroscopic inspection of the aspirate was performed by the ENT surgeon using four categories: pus, purulent, bloody and clear. In all cases samples were taken for both aerobic and anaerobic cultures. Cultures were performed using semi-quantitative methods (no growth, 0 colonies; +, 1 to 10 colonies; ++, 10 to 100 colonies; +++, more than 100 colonies), with ++ or +++ being regarded as positive and 0 or + as negative. Repeated drainage could be scheduled at the discretion of the attending ENT surgeon on clinical grounds. Patients with equivocal and normal results on SXR were followed up. In patients with equivocal results a repeat SXR was made 48 hours later unless the fever had resolved or another cause of fever had been found. In patients with normal SXRs no repeat was indicated except at the discretion of the attending physician.
Final diagnosis for cause of fever in all 351 patients was based on blood, sputum and sinus cultures as applicable, chest X-rays and on clinical criteria (normalisation of temperature after removal of the central line, or after sinus drainage, response to antibiotic treatment, and so on).
Statistical analyses were performed with Student's t-test for unpaired groups. Results are expressed as means ± SD. Statistical significance was accepted at P < 0.05.
Results
The results are summarised in Fig. 1.
During a period of 15 months a total of 351 patients met the initial inclusion criteria. In 153 patients a probable cause of fever was found on routine analysis. Therefore 198 patients met the criteria for SXR. Patient data and the results of these X-rays are shown in Table 1.
On the basis of the results of the SXR, sinus drainage was first performed in those patients with evident abnormalities (n = 101). Drainage was performed in 98 of these 101 patients within 12 hours (mean 2.05 ± 5.7 hours). Twenty-four patients had been given platelets or fresh frozen plasma before the procedure. In three patients the procedure was delayed for a longer period because of the use of anticoagulants and/or platelet aggregation inhibitors. In these patients the procedure was performed within 48 hours.
Repeat drainage was performed in 41 patients after an average period of 52 ± 38 hours.
The initial (macroscopic) interpretation of the material obtained during the draining procedure by the ENT surgeon was pus in fluid from 17 of 101 patients (17%), purulent in 38 (38%), bloody in 13 (13%) and clear in 33 (33%). Culture results of initial sinus drainage are shown in Table 2. Many patients had more than one microorganism cultured from the sinus fluid. A total of 140 microorganisms were cultured from 84 of these 101 sinus drainage fluids (84%). All cultures that had been deemed as pus or purulent on macroscopic evaluation turned out positive for pathogenic microorganisms. However, bacteria were also cultured from a substantial proportion (18 of 33 (55%)) of the fluids that had been deemed clear on microscopic inspection. The cultured pathogens are listed in Table 2. The most predominant microorganisms in the sinus fluids were Gram-negative pathogens such as Pseudomonas and Klebsiella species.
Of the 28 patients with indeterminate or equivocal results on the initial SXR, a repeat SXR was performed in 25 patients. Ten (40%) now had obvious abnormalities, and drainage was performed. Of these the diagnosis of sinusitis was confirmed in 9 patients. Of the 69 patients with an initially normal SXR, sinus drainage was nevertheless performed in the subsequent 72-hour period in 12 patients on clinical grounds (n = 2), repeat SXR (n = 2) and following CT scan (n = 8). The diagnosis was confirmed by drainage and cultures in all 12 of these patients. Thus, a total of 21 cases (22%) of microbiological sinusitis were subsequently found in the group of 97 patients who initially had equivocal or normal findings on SXR.
On the basis of these clinical, radiological and microbiological criteria we evaluated the final diagnoses in all 351 patients with FUO initially included in our study. The results are shown in Tables 3 and 4.
Discussion
The results of our study demonstrate that sinusitis is a frequently occurring hospital-acquired infection in the ICU. Sinusitis was initially diagnosed in 84 of 351 (24%) patients with FUO, and in an additional 21 patients who had equivocal or normal findings on initial SXR, giving a total of 105 of 351 patients (30%). This underestimates the true incidence because SXRs were not taken in 153 patients who had obvious other causes of fever on initial screening, some of whom might also have had sinusitis.
Sinusitis was the sole cause of fever in 57 patients (16%) and one of several causes (for example sinusitis and purulent bronchitis) in 48 patients (14%). Pathogenic microorganisms were cultured not only when material obtained by antral sinus puncture was classified as 'purulent' but also in more than half of the patients whose puncture material was less suspect on macroscopic examination.
Previous studies on sinusitis in orotracheally intubated patients have reported a lower incidence of sinusitis than was observed in our study, ranging from 2% to 7.7% [5,7,25,26]. There are several possible reasons for this. First, the rate of antibiotic resistance in the Netherlands is low, and antibiotics are used relatively sparingly. This might have reduced the likelihood of undetected sinusitis being concomitantly treated because patients were receiving antibiotics for other infections [25,27]. Second, our patients were more severely ill than patients included in the previous studies, as demonstrated by a high average severity of disease score (Acute Physiology and Chronic Health Evaluation (APACHE)-II score of 21 ± 6.8 in our study, compared with Simplified Acute Physiology Score (SAPS) II scores of 12 ± 4.5 [5] and 11.0 ± 3.5 [7]; other studies reported no severity scores). Third, risk factors for sinusitis such as sedation and nasogastric tube feeding were present more frequently in our patients, perhaps because of the greater severity of disease.
Of the positive cultures in our patients, 77% contained Gram-negative pathogens. This rate is higher than reported in previous studies, in which about 50% of cultured pathogens were Gram-positive [8,25,28,29]. This might be explained by differences in case mix, severity of illness and length of ICU stay, as well as effects of previous antibiotic treatment on the patients' microflora [25,28,29].
Our study has some limitations. The diagnosis was based on abnormal findings on SXR and positive microbiological cultures obtained after antral drainage. However, SXRs cannot accurately distinguish purulent sinusitis from sterile fluids, so abnormal SXRs may overestimate the incidence of sinusitis [8,25]. Moreover, even positive microbiological cultures may not prove clinically relevant sinusitis, because they may indicate colonisation rather than actual infection. We tried to circumvent these problems by classifying only cultures with more than 10 colonies of bacteria as positive and by basing our diagnosis on a combination of radiological abnormalities, positive cultures, and clinical response to therapeutic measures such as drainage and targeted antibiotic treatment. We are therefore confident that our results accurately reflect the true incidence of sinusitis.
Early detection and treatment is important because delays can lead to the development of VAP, sepsis, and life-threatening complications such meningitis, mastoiditis, intra-cranial abscesses and venous thrombosis of the sinus cavernosus [24,30,31]. Early treatment of sinusitis may significantly reduce the risk of VAP and perhaps also ICU mortality [8,32].
The results of our microbiological cultures underline the close link between sinusitis and the development of VAP. Of 105 patients in whom positive sinus cultures were obtained, the same microorganisms were cultured from bronchotracheal aspirates in 40% of cases (n = 42). In some patients we were able to demonstrate that positive sinus cultures preceded positive cultures from the lungs, strongly suggesting that sinusitis can lead to infections of the lower airway. Others reported similar observations; for example, Holzapfel and co-workers found that the early detection and treatment of hospital-acquired sinusitis could prevent the occurrence of VAP and reduce mortality in nasotracheally intubated ICU patients [33].
Bacteraemia with the same microorganism as that cultured from the sinus occurred in 12 patients; in five patients the microorganism causing bacteraemia was cultured only from the blood and the sinus, making sinusitis the most likely cause of bacteraemia. However, no definite conclusions about cause and effect can be drawn because bacteraemia can also lead to sinusitis, with bacterial colonisation of sinus fluids following bacteraemia [30,34].
Various mechanisms might explain the high incidence of sinusitis in ICU patients. The first is anatomical. The paranasal sinuses secrete mucus that flows to the natural ostia located posteriorly towards the nasopharynx; this flow can be blocked by infection, inflammation, anatomic abnormalities or the presence of foreign material such as nasotracheal intubation tubes. Even tubes with smaller diameters (such as nasogastric feeding and suction tubes) can cause significant obstruction in the normal flow of sinus fluids, leading to an increased risk of bacterial colonisation and development of hospital-acquired sinusitis [25,28]. The presence of nasogastric tubes has been linked to a significant increase in the risk for sinusitis in mechanically ventilated patients (odds ratio 14.1, 95% confidence interval 1.7 to 117) [25]. Another important risk factor is the use of sedatives (odds ratio 15.9, 95% confidence interval 1.9 to 133.5) [25]. Underlying mechanisms may include the suppression of normal cleansing mechanisms such as coughing, sneezing and nose-blowing, because of sedation and analgesia; in addition, immobility precludes positional changes that improve mucous drainage under normal circumstances [24]. Remaining in a recumbent position can increase nasal congestion and obstruction of the ostia of the maxillary sinuses. This problem may be compounded by the positive inspiratory and end-expiratory pressure in ventilated patients, which also induces an increase in central venous pressure [6,35]. In addition, critically ill patients recovering from earlier episodes of sepsis may develop relative immune suppression, so-called immunoparalysis [36].
ICU patients are often unable to communicate, and complaints related to sinusitis may go unnoticed by the medical and nursing staff. Patients may have a 'runny nose', or discharge of purulent material from the nasal cavity. However, this is seen in only 27% of cases [37]. Thus elevations in white blood cell count and/or FUO may be the first presenting symptoms [24].
In theory, the use of imaging modalities such as CT scans [24,38,39] and B-mode ultrasound [12-14] could improve the diagnostic yield. As discussed above, the CT scan should be regarded as the gold standard for the diagnosis of sinusitis. Unfortunately, CT scans are not easily performed in the ICU setting, meaning that the patient has to be transported to the department of radiology for this procedure. These in-hospital transports can be risky [9,10,40,41]. The potential benefits in establishing or confirming the diagnosis should therefore be weighed against the risks of transport. The development of mobile CT scans for use at bedside would significantly reduce these problems; however, such devices are not yet available in most hospitals.
Some authors have suggested that ultrasound may provide a good or even better alternative to SXR for detecting sinusitis at bedside in critically ill patients [12,13]. However, ultrasound is not yet widely used for this purpose in routine clinical practice. Moreover, its diagnostic accuracy depends on the experience of the operator, and the costs are higher than for SXR. In addition, the literature comparing diagnostic yields of ultrasound and SXR provides conflicting results [12,13,21,22,42]. Our study was not designed to compare the two techniques; we based our choice mainly on the fact that ultrasound is not yet widely used to detect sinusitis in the ICU setting, and on our pre-existing clinical protocols. It seems unlikely that use of ultrasound for initial screening would have significantly affected our results; at best it could have increased our diagnostic yield, further strengthening our observation that sinusitis is a frequent cause of FUO in ICU patients. In addition, about 85% of the cases of hospital-acquired sinusitis associated with mechanical ventilation involve the maxillary sinuses [6]. As conventional SXRs are most reliable in detecting maxillary sinusitis (in comparison with frontal and ethmoidal sinusitis) we feel that SXRs remain the most practical diagnostic tool, with an acceptable sensitivity for detecting sinusitis in the ICU setting. Hospitals favouring ultrasonography as initial screening method could easily adapt our protocol, replacing SXR by ultrasound. The CT scan remains the radiological gold standard in the diagnosis of sinusitis.
On the basis of the results of our study we recommend that hospital-acquired sinusitis be considered in all patients with FUO in the ICU in whom a cause of fever is not immediately apparent from initial examination and chest X-ray. SXR or ultrasound or (if possible) a CT scan should be included in the diagnostic work-up, and sinus puncture with drainage should be performed in case of abnormal or equivocal findings. In our study all procedures were performed at thr bedside; 40% of patients with confirmed sinusitis required repeat drainage, but no patients required more than two procedures.
All nasal tubes should be removed if sinusitis is suspected; antibiotics should be started empirically or based on Gram staining, and adjusted for final culture results. In most patients temperature normalises within 48 hours [37]; this was also observed in our study. Radiological signs of sinusitis clear more slowly but should disappear within ±1 week [43].
The results of our study have led to the implementation of several measures to reduce the incidence of sinusitis. First, nasogastric tubes are no longer used in intubated patients unless it is expected that the endotracheal tube can be removed within 24 hours. Gastric tubes in all other patients are now inserted through the mouth. Second, patients intubated for ≥24 hours now routinely receive topical administration of saline 0.9% and/or decongestants such as xylometazoline drops in the nasal cavities. Thirdly, the nursing staff keeps a far more rigorous watch for signs of purulent nasal discharge in all patients, and diagnostic procedures such as X-sinus are performed if such discharge is observed. Finally, the routine diagnostic work-up in patients who develop fever in the ICU now includes an SXR. Drainage (both as a diagnostic and therapeutic tool) takes place in all patients with clear or equivocal signs of sinusitis. Topical decongestants are used to reduce oedema and facilitate drainage. In patients with clear SXR in whom no other diagnosis is established, SXR is repeated after 48 hours. These measures have led to a marked reduction in the incidence of sinusitis in our ICU.
Conclusion
Hospital-acquired sinusitis is a frequent cause of FUO in orotracheally intubated and mechanically ventilated critically ill patients. ICU physicians should be aware of the numerous risk factors for sinusitis simultaneously present in ICU patients and take appropriate preventive measures. We recommend including an SXR in the routine work-up for FUO in all ICU patients; drainage should take place if SXR reveals clouding, and should also be considered if the SXR is equivocal or difficult to interpret. A normal SXR does not rule out sinusitis, and when in doubt drainage or additional diagnostic procedures such as CT scan should be performed.
Key messages
• Sinusitis is a frequent cause of FUO in the ICU (in this study it was the sole cause in 16% and a contributing factor in 13% of patients with FUO).
• Bacterial colonisation of the sinuses often precedes the development of bronchitis and VAP; sinusitis may be a frequent cause of hospital-acquired bronchitis and VAP.
• Diagnostic work-up of FUO should include an SXR, ultrasound or CT scan; drainage should be performed if any abnormalities are found.
• Physicians treating critically ill patients should be aware of the high risk of sinusitis and take appropriate preventive measures, including the removal of nasogastric tubes in patients requiring long-term mechanical ventilation.
Abbreviations
CT = computed tomography; ENT = ear, nose and throat; FUO = fever of unknown origin; ICU = intensive care unit; SXR = sinus X-ray; VAP = ventilator-associated pneumonia.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
KHP, JMD and ARG designed and coordinated the study. AvZ, RdB, JMD and KHP were involved in the collection, statistical analysis and interpretation of the data. MDN performed literature analysis and assisted in the data collection. AvZ and KHP drafted and revised the manuscript. All authors read and approved the final manuscript.
Figures and Tables
Figure 1 Flowchart depicting the organisation of our study in patients with fever of unknown origin (FUO), as well as the diagnostic work-up and culture results.
Table 1 Epidemiological data, results of SXR and macroscopic evaluation of sinus fluids obtained by antral puncture
Parameter Value
Patient data (n = 351)
Sex (M:F) 193:158
Age (mean ± SD) 59 ± 21.2
APACHE II score (mean ± SD) 21 ± 6.8
ICU LOS (days) at diagnosis (mean ± SD) 5.9 ± 5.7
Results of sinus X-rays in patients with FUO (n = 198)
Sinus X-ray abnormal (two directions) 101 (51%)
Sinus X-ray equivocal 28 (14%)
Sinus X-ray normal 69 (35%)
Macroscopic evaluation of sinus fluid (n = 101)
Pus 17 (17%)
Purulent 38 (38%)
Bloody 13 (13%)
Clear 33 (33%)
APACHE, Acute Physiology and Chronic Health Evaluation; FUO, fever of unknown origin; UCI LOS, length of stay in the intensive care unit.
Table 2 Results of sinus fluid cultures of patients with gross abnormalities in their initial sinus X-ray
Bacterium Sinus fluid Same MO cultured from tracheal aspirate Same MO cultured from blood
Pseudomonas aeruginosa 32 22 8
Klebsiella oxytoca 5 3 1
Klebsiella pneumoniae 17 10 2
Enterococcus faecalis 10 2 1
Enterobacter cloacae 12 4 0
Escherichia coli 16 8 1
Staphylococcus aureus 8 3 1
Gram-positive mixture 11 - -
Gram-negative mixture 14 - -
Othera 15 8 1
Total 140 60 15
There were 101 patients with gross abnormalities in their initial sinus X-ray. Positive cultures were obtained in 84 patients, with 140 different types of microorganisms (MOs). Columns 3 and 4 show positive results of the same microorganisms (MOs) cultured from tracheal aspirate and blood, cultured in the period between 24 hours before and 48 hours after sinus drainage.
aOther pathogens included anaerobic bacteria (such as Bacteroides sp.) and fungi (Candida sp.).
Table 3 Initial diagnosis for fever of unknown origin in mechanically ventilated patients in intensive care unit
Cause of fever Sole cause One of multiple causes
Central line infection 43 1
Upper respiratory tract infection/pneumoniaa 93 42
Sinusitis 45 39
Abdominal focus 5 3
Otherb 2 1
Unknown 121
Total 188 86
Initial diagnosis was performed after initial screening with physical examination and chest X-ray in all 351 patients, sinus X-ray in 198 patients and sinus drainage in 98 patients; cultures were not yet available. All patients had fever and leucocytosis. aPurulent tracheobronchial aspirate with cultures positive for pathogenic microorganisms, combined with new or progressive pulmonary infiltrates on chest X-ray; bother causes of fever included meningitis, phlebitis and deep venous thrombosis.
Table 4 Final diagnosis for FUO at ICU discharge, with final results of all cultures known
Cause of fever Sole cause One of multiple causesa
Central line infection 44 11
Upper respiratory tract infection/pneumonia 132 58
Sinusitis 57 48
Abdominal focus 8 16
Otherb 12 28
Unknown 46
Total 253 161
See also Fig. 1. aMost patients with more than one cause of fever had sinusitis and bronchitis/pneumonia; bother causes of fever included meningitis (not related to sinusitis), phlebitis and deep venous thrombosis.
==== Refs
Chastre J Fagon J Ventilator-associated pneumonia Am J Respir Crit Care Med 2002 165 867 903 11934711
Polderman KH Girbes AR Central venous catheter use. Part 2: Infectious complications Intensive Care Med 2002 28 18 28 11818995 10.1007/s00134-001-1156-7
Anon Hospital-acquired pneumonia in adults: diagnosis, assessment of severity, initial antimicrobial therapy, and preventive strategies. A consensus statement. American Thoracic Society, November 1995 Am J Respir Crit Care Med 1996 153 1711 1725 8630626
O'Grady NP Alexander M Dellinger EP Gerberding JL Heard SO Maki DG Masur H McCormick RD Mermel LA Pearson ML Guidelines for the prevention of intravascular catheter-related infections. Centers for Disease Control and Prevention MMWR Recomm Rep 2002 51 1 29 12233868
Bach A Boehrer H Schmidt H Geiss HK Nosocomial sinusitis in ventilated patients. Nasotracheal versus orotracheal intubation Anaesthesia 1992 47 335 339 1519688
Michelson A Schuster B Kamp HD Paranasal sinusitis associated with nasotracheal and orotracheal long-term intubation Arch Otolaryngol Head Neck Surg 1992 118 937 939 1503719
Salord F Gaussorgues P Marti-Flich J Sirodot M Allimant C Lyonnet D Robert D Nosocomial maxillary sinusitis during mechanical ventilation: a prospective comparison of orotracheal versus the nasotracheal route for intubation Intensive Care Med 1990 16 390 393 2246421 10.1007/BF01735177
Holzapfel L Chastang C Demingeon G Bohe J Piralla B Coupry A A randomized study assessing the systematic search for maxillary sinusitis in nasotracheally mechanically ventilated patients. Influence of nosocomial maxillary sinusitis on the occurrence of ventilator-associated pneumonia Am J Respir Crit Care Med 1999 159 695 701 10051239
Beckmann U Gillies DM Berenholtz SM Wu AW Pronovost P Incidents relating to the intra-hospital transfer of critically ill patients. An analysis of the reports submitted to the Australian Incident Monitoring Study in Intensive Care Intensive Care Med 2004 30 1579 1585 14991102 10.1007/s00134-004-2177-9
Smith I Fleming S Cernaianu A Mishaps during transport from the intensive care unit Crit Care Med 1990 18 278 281 2302952
Waydhas C Intrahospital transport of critically ill patients Crit Care 1999 3 R83 R89 11094486 10.1186/cc362
Hilbert G Vargas F Valentino R Gruson D Chene G Bebear C Gbikpi-Benissan G Cardinaud JP Comparison of B-mode ultrasound and computed tomography in the diagnosis of maxillary sinusitis in mechanically ventilated patients Crit Care Med 2001 29 1337 1342 11445682 10.1097/00003246-200107000-00007
Puidupin M Guiavarch M Paris A Caroff P Boutin JP Le Bivic T Garcia JF B-mode ultrasound in the diagnosis of maxillary sinusitis in intensive care unit Intensive Care Med 1997 23 1174 1175 9434925 10.1007/s001340050476
Lichtenstein D Biderman P Meziere G Gepner A The 'sinusogram', a real-time ultrasound sign of maxillary sinusitis Intensive Care Med 1998 24 1057 1061 9840240 10.1007/s001340050716
Lucchin F Minicuci N Ravasi MA Cordella L Palu M Cetoli M Borin P Comparison of A-mode ultrasound and computed tomography: detection of secretion in maxillary and frontal sinuses in ventilated patients Intensive Care Med 1996 22 1265 1268 9120124
Haapaniemi J Laurikainen E Ultrasound and antral lavage in the examination of maxillary sinuses Rhinology 2001 39 39 42 11340694
Pfleiderer AG Drake-Lee AB Lowe D Ultrasound of the sinuses: a worthwhile procedure? A comparison of ultrasound and radiography in predicting the findings of proof puncture on the maxillary sinuses Clin Otolaryngol Allied Sci 1984 9 335 339 6398149
Jensen C von Sydow C Radiography and ultrasonography in paranasal sinusitis Acta Radiol 1987 28 31 34 2952138
Trigaux JP Bertrand BM Van Beers BE Comparison of B-mode ultrasound, radiography and sinuscopy in the diagnosis of maxillary sinusitis. Report on 177 cases Acta Otorhinolaryngol Belg 1988 42 670 679 3071937
Shapiro GG Furukawa CT Pierson WE Gilbertson E Bierman CW Blinded comparison of maxillary sinus radiography and ultrasound for diagnosis of sinusitis J Allergy Clin Immunol 1986 77 59 64 3511125 10.1016/0091-6749(86)90324-6
Varonen H Makela M Savolainen S Laara E Hilden J Comparison of ultrasound, radiography, and clinical examination in the diagnosis of acute maxillary sinusitis: a systematic review J Clin Epidemiol 2000 53 940 948 11004420 10.1016/S0895-4356(99)00213-9
Engels EA Terrin N Barza M Lau J Meta-analysis of diagnostic tests for acute sinusitis J Clin Epidemiol 2000 53 852 862 10942869 10.1016/S0895-4356(00)00209-2
European Academy of Allergology and Clinical Immunology European position paper on rhinosinusitis and nasal polyps Rhinol Suppl 2005 18 1 87 15847064
Seiden AM Sinusitis in the critical care patient New Horiz 1993 1 261 270 7522919
George DL Falk PS Umberto Meduri G Leeper KV JrWunderink RG Steere EL Nunnally FK Beckford N Mayhall CG Nosocomial sinusitis in patients in the medical intensive care unit: a prospective epidemiological study Clin Infect Dis 1998 27 463 470 9770141
Borman KR Brown PM Mezera KK Jhaveri H Occult fever in surgical intensive care unit patients is seldom caused by sinusitis Am J Surg 1992 164 412 415 1443363
Cars O Molstad S Melander A Variation in antibiotic use in the European Union Lancet 2001 357 1851 1853 11410197 10.1016/S0140-6736(00)04972-2
Vandenbussche T De Moor S Bachert C Van Cauwenberge P Value of antral puncture in the intensive care patient with fever of unknown origin Laryngoscope 2000 110 1702 1706 11037829 10.1097/00005537-200010000-00025
Bert F Lambert-Zechovsky N Microbiology of nosocomial sinusitis in intensive care unit patients J Infect 1995 31 5 8 8522833 10.1016/S0163-4453(95)91147-2
Bos AP Tibboel D Hazebroek FW Hoeve H Meradji M Molenaar JC Sinusitis: hidden source of sepsis in postoperative pediatric intensive care patients Crit Care Med 1989 17 886 888 2766760
O'Reilly MJ Reddick EJ Black W Carter PL Erhardt J Fill W Maughn D Sado A Klatt GR Sepsis from sinusitis in nasotracheally intubated patients. A diagnostic dilemma Am J Surg 1984 147 601 604 6721035 10.1016/0002-9610(84)90122-3
Talmor M Li P Barie PS Acute paranasal sinusitis in critically ill patients: guidelines for prevention, diagnosis and treatment Clin Infect Dis 1997 25 1441 1446 9431392
Holzapfel L Chevret S Madinier G Ohen F Demingeon G Coupry A Chaudet M Influence of long-term oro- or nasotracheal intubation on nosocomial maxillary sinusitis and pneumonia: results of a prospective, randomized, clinical trial Crit Care Med 1993 21 1132 1138 8339576
Aebert H Hunefeld C Regel G Paranasal sinusitis and sepsis in ICU patients with nasotracheal intubation Intensive Care Med 1988 15 27 30 3230197 10.1007/BF00255632
Aust R Drettner B The patency of the maxillary ostium in relation to body posture Acta Otolaryngol 1975 80 443 446 1202917
Adrie C Pinsky MR The inflammatory balance in human sepsis Intensive Care Med 2000 26 364 375 10872127 10.1007/s001340051169
Caplan ES Hoyt NJ Nosocomial sinusitis JAMA 1982 247 639 641 7054565 10.1001/jama.247.5.639
Osguthorpe JD Hadley JA Rhinosinusitis. Current concepts in evaluation and management Med Clin North Am 1999 83 27 41 9927958 10.1016/S0025-7125(05)70085-7
Zinreich SJ Rhinosinusitis: radiologic diagnosis Otolaryngol Head Neck Surg 1997 117 S27 S34 9334785
Szem JW Hydo LJ Fischer E Kapur S Klemperer J Barie PS High-risk intrahospital transport of critically ill patients: safety and outcome of the necessary 'road trip' Crit Care Med 1995 23 1660 1666 7587230 10.1097/00003246-199510000-00009
Evans A Winslow EK Oxygen saturation and hemodynamic response in critically ill, mechanically ventilated adults during intrahospital transport Am J Crit Care 1995 4 106 111 7749441
de Bock GH Houwing-Duistermaat JJ Springer MP Kievit J van Houwelingen JC Sensitivity and specificity of diagnostic tests in acute maxillary sinusitis determined by maximum likelihood in the absence of an external standard J Clin Epidemiol 1994 47 1343 1352 7730843 10.1016/0895-4356(94)90078-7
Deutschman CS Wilton P Sinow J Dibbell D JrKonstantinides FN Cerra FB Paranasal sinusitis associated with nasotracheal intubation: a frequently unrecognized and treatable source of sepsis Crit Care Med 1986 14 111 114 3943315
|
16277722
|
PMC1297630
|
CC BY
|
2021-01-04 16:04:56
|
no
|
Crit Care. 2005 Sep 13; 9(5):R583-R590
|
utf-8
|
Crit Care
| 2,005 |
10.1186/cc3805
|
oa_comm
|
==== Front
Crit CareCritical Care1364-85351466-609XBioMed Central London cc38061627772310.1186/cc3806ResearchCirculating anions usually associated with the Krebs cycle in patients with metabolic acidosis Forni Lui G [email protected] William 2Lord Gwyn A [email protected] David F 4Peron Jean-Marie R 5Hilton Philip J 61 Consultant Physician & Intensivist, Department of Critical Care, Worthing Hospital, Worthing, West Sussex, UK2 Research Fellow, Renal Laboratory, St Thomas' Hospital, London, UK3 MRC Scientist, MRC Toxicology Unit, Birkbeck College, London, UK4 Consultant Physician & Intensivist, Renal Laboratory, St Thomas' Hospital, London, UK5 Research Fellow, Department of Chemistry, Kingston University, Surrey, UK6 Consultant Physician & Research Director, Renal Laboratory, St Thomas' Hospital, London, UK2005 13 9 2005 9 5 R591 R595 29 6 2005 22 7 2005 1 8 2005 12 8 2005 Copyright © 2005 Forni 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
Acute metabolic acidosis of non-renal origin is usually a result of either lactic or ketoacidosis, both of which are associated with a high anion gap. There is increasing recognition, however, of a group of acidotic patients who have a large anion gap that is not explained by either keto- or lactic acidosis nor, in most cases, is inappropriate fluid resuscitation or ingestion of exogenous agents the cause.
Methods
Plasma ultrafiltrate from patients with diabetic ketoacidosis, lactic acidosis, acidosis of unknown cause, normal anion gap metabolic acidosis, or acidosis as a result of base loss were examined enzymatically for the presence of low molecular weight anions including citrate, isocitrate, α-ketoglutarate, succinate, malate and d-lactate. The results obtained from the study groups were compared with those obtained from control plasma from normal volunteers.
Results
In five patients with lactic acidosis, a significant increase in isocitrate (0.71 ± 0.35 mEq l-1), α-ketoglutarate (0.55 ± 0.35 mEq l-1), malate (0.59 ± 0.27 mEq l-1), and d-lactate (0.40 ± 0.51 mEq l-1) was observed. In 13 patients with diabetic ketoacidosis, significant increases in isocitrate (0.42 ± 0.35 mEq l-1), α-ketoglutarate (0.41 ± 0.16 mEq l-1), malate (0.23 ± 0.18 mEq l-1) and d-lactate (0.16 ± 0.07 mEq l-1) were seen. Neither citrate nor succinate levels were increased. Similar findings were also observed in a further five patients with high anion gap acidosis of unknown origin with increases in isocitrate (0.95 ± 0.88 mEq l-1), α-ketoglutarate (0.65 ± 0.20 mEq l-1), succinate (0.34 ± 0.13 mEq l-1), malate (0.49 ± 0.19 mEq l-1) and d-lactate (0.18 ± 0.14 mEq l-1) being observed but not in citrate concentration. In five patients with a normal anion gap acidosis, no increases were observed except a modest rise in d-lactate (0.17 ± 0.14 mEq l-1).
Conclusion
The levels of certain low molecular weight anions usually associated with intermediary metabolism were found to be significantly elevated in the plasma ultrafiltrate obtained from patients with metabolic acidosis. Our results suggest that these hitherto unmeasured anions may significantly contribute to the generation of the anion gap in patients with lactic acidosis and acidosis of unknown aetiology and may be underestimated in diabetic ketoacidosis. These anions are not significantly elevated in patients with normal anion gap acidosis.
See related commentary
==== Body
Introduction
Metabolic acidosis is a common presentation in acute medicine and in only a minority of cases can renal failure alone be considered the sole cause of the acidosis. Diabetic ketoacidosis (DKA) and 'classic' lactic acidosis (taken as a blood lactate in excess of 5 mmol l-1) account for most of the remaining cases [1,2]. Amongst clinicians there is an increasing awareness that there is often an important discrepancy between the measured acidosis and the associated base deficit, suggesting that other anions must contribute to the generation of the anion gap [3,4]. Further evidence to support the hypothesis of additional 'missing anions' is supplied by observations on patients receiving bicarbonate-buffered haemofiltration for treatment of lactic acidosis where it is possible to calculate the net rates of lactate removal and bicarbonate donation by the haemofilter [5]. In patients whose blood pH is slowly falling (thereby excluding any possible confounding effect of other blood buffers), it is not uncommon for the rate of bicarbonate donation to exceed the rate of lactate removal by as much as twofold (PJ Hilton, unpublished observations). This implies that not only lactic acid, but other 'unknown' acids are being neutralised by the bicarbonate in the haemofiltration replacement fluid.
Preliminary experiments performed in this laboratory with high performance liquid chromatography coupled to mass spectrometry on plasma ultrafiltrate obtained from patients with metabolic acidosis had suggested that anions principally associated with the Krebs tricarboxylic acid cycle were significantly elevated in patients with unexplained metabolic acidosis. We undertook an enzymatic study of anions of the Krebs cycle in patients with metabolic acidosis whose standard base deficit was 8 mmol l-1 or greater. The samples were obtained from consecutive patients who fulfilled these requirements except those with a normal anion gap, which were collected subsequently. The results obtained from the patient samples were compared to control values obtained from healthy volunteers.
Methods
This study was approved by the Ethics Committee of Guy's and St Thomas' National Health Service Trust (reference number EC03/104). Prior to the sample being taken, informed consent was obtained from the subject or, where this was not possible, their next of kin. Patient studies were undertaken on 15 ml of arterial blood taken from arterial cannulae in patients with metabolic acidosis whose standard base deficit was 8 mmol l-1 or greater. The arterial blood was drawn into a non-heparinised syringe before being rapidly transferred into SST II (KODAK) Vaccutainers (BD Vaccutainer Systems Ltd, Plymouth, UK). Control samples were obtained from venous blood of laboratory workers and treated in the same manner as that obtained from patients. Arterial blood gas levels were not measured as it was felt that arterial puncture was inappropriate.
Once obtained, the sample was chilled and rapidly transported to the laboratory. The plasma was isolated by centrifugation of the Vaccutainers (1,500 g) at 4°C for 10 minutes. The plasma was transferred to an Amicon 30,000 Da cutoff filter (Millipore, Watford, Herts, UK) where centrifugation at 1,560 g for 15 minutes produced ultrafiltrate. The generated ultrafiltrate was either immediately analysed or stored at -20°C for analysis within 24 h. Previous work had highlighted the need for rapid assay of the samples due to an observed rapid decrease in concentrations of the measured anions within plasma.
The concentration of anions in the ultrafiltrate was determined by enzymatic assay with reference to internal standards. The plasma ultrafiltrate concentrations of citrate, succinate, malate, d-lactate and l-lactate anions were estimated using commercially available kits (Roche, Glasgow, UK). The levels of isocitrate and α-ketoglutarate were measured using enzyme assays developed by ourselves using isocitrate dehydrogenase and α-ketoglutarate dehydrogenase, respectively, and their associated co-factors (Sigma Chemicals, Poole, UK). All the enzymatic assays relied upon the interconversion of NAD+ and NADH or NADP+ and NADPH and used the change in absorbance due to the reduced co-enzyme at 340 nm. Oxaloacetate concentrations could not be accurately measured as a result of its short half-life (approximately 69 s) in aqueous systems at near physiological pH [6]. All data are presented as mean ± standard deviation. Unless stated otherwise, the data are normally distributed and statistical analysis was undertaken with an unpaired t-test. Where the data was not normally distributed a Mann-Whitney non-parametric test was applied. In both cases, significance was deemed to have been attained if p was ≤0.05
Results
Samples were obtained from 28 patients in total and from 12 control subjects. Of the study patients, 13 had DKA at the time of blood sampling and 5 had a lactic acidosis (defined as a blood lactate concentration in excess of 5 mmol/l in the presence of acidosis). Five patients had a metabolic acidosis that could not be ascribed to either lactic acidosis, ketoacidosis or exogenous agents. A further five patients had an acidosis as a result of gastrointestinal or renal ion losses. In all cases, the patients were acidotic with an average pH of 7.18 (±0.11) and a base deficit of 13.4 (±4.7) mmol l-1.
The mean results obtained for the plasma ultrafiltrate concentration of citrate, succinate, malate, d-lactate, l-lactate, isocitrate and α-ketoglutarate are presented in Table 1. The mean results obtained from the data together with the maximal value for each anion measured in each type of acidosis are shown in Fig. 1.
Patients with DKA showed significant increases relative to the control values in isocitrate, α-ketoglutarate, malate and d-lactate levels. The concentrations of both citrate and succinate did not differ significantly from controls. Patients with lactic acidosis showed significant increases relative to the control values in citrate, isocitrate, α-ketoglutarate, succinate, malate and d-lactate levels. Patients with acidosis of unknown origin showed significant increases in the concentrations of isocitrate, α-ketoglutarate, succinate, malate and d-lactate. The level of citrate did not differ significantly from that of the controls. In those patients with a normal anion gap acidosis, the levels of citrate, isocitrate, α-ketoglutarate, succinate and malate did not differ significantly from the control values. The concentration of d-lactate was significantly raised compared to control values in this patient group.
Discussion
The consequences of metabolic acidosis can be catastrophic and a considerable body of literature highlights the poor outlook in patients where lactic acid is the principal component of the acidaemia [7-9]. This increase in blood lactate concentration reflects either increased lactate production, reduced lactate metabolism or, more commonly, a combination of the two [10]. In patients with DKA, 3-hydroxybutyric acid and, to a lesser extent, acetoacetic acid play the major role in the generation of the anion gap. As outlined, however, in a third patient group, neither lactate nor 3-hydroxybutyrate is responsible for the elevated anion gap and the relevant anions responsible remain unknown [11]. In the fourth patient group, acidosis is generated as a result of uncontrolled electrolyte loss either from the kidney (renal tubular acidosis) or the gut.
We have shown that the plasma concentrations of acids usually associated with the Krebs tricarboxylic acid cycle are significantly increased in patients with lactic acidosis as well as those with 'unexplained acidosis' with normal or near normal blood lactate concentrations. In DKA, although the concentrations of these acids are less strikingly elevated, they are still abnormal in the majority of patients when compared to controls. They are not, however, significantly elevated in patients with normal anion gap acidosis secondary to excess base loss. The accumulation of such acids may contribute significantly to the production of the anion gap and account, in part, for the 'missing' anions in patients with certain forms of acidosis. Recent studies, in keeping with previous work, have demonstrated the predictive value of acid-base variables on outcome in the critically ill [12]. Furthermore, the calculation of unmeasured anions appears to be a better discriminator of outcome than lactate or base deficit [13].
With the partial exception of citrate and isocitrate (97% ionised at pH 7.0) the anions examined in this study are effectively fully ionised at the measured pH. Unlike lactate, not all the anions are monobasic, with tribasic acids (citric and isocitric) contributing three protons, whilst the dibasic acids (α-ketoglutaric, malic and succinic) add two protons to the solution on ionisation. Converting the concentrations of these observed anions to mEq l-1 (Table 1) shows that, on average, the contribution to the observed anion gap due to such anions may be in excess of 3 mEq l-1 and in some cases may be over 5 mEq l-1. Thus, the contribution of these anions to the generation of the anion gap is of much greater significance than is apparent from their molarity. The large standard deviation present in these samples probably reflects their heterogeneous nature and, in many ways, the range and maxima in each group are of as much interest as the means, demonstrating the extreme ranges that can be present in patients with metabolic acidosis (Fig. 1).
The greatest deviations in the level of measured plasma ultrafiltrate acids from that observed in the control group were seen in the patients within the lactic acidosis and 'unexplained acidosis' groups. In these two groups, the concentrations of most of the acids studied were present in the plasma ultrafiltrate at a concentration significantly higher than that observed in the normal control population. In those with DKA, four of the six acids measured were significantly elevated relative to the control values. Interestingly, in normal anion gap acidosis, only d-lactate was significantly increased relative to the control values.
It has been widely reported elsewhere that plasma citrate is not elevated in acidosis [4,11] and we observed this in all the groups of patients studied except those with lactic acidosis. In this group, significantly elevated levels of citrate were observed in comparison to the normal control values. This result may be unreliable as four of the patients in this group had received an infusion of heparin (containing sodium citrate as an anticoagulant) prior to the blood sample being obtained. Isocitrate concentrations were significantly elevated in patients with ketoacidosis, lactic acidosis and those whose acidosis was of unknown origin. Consequently, the citrate:isocitrate ratio is significantly reduced compared to control values and we are unable to advance a simple explanation as to why this ratio should be so low.
Although it seems reasonable to believe that the likely source for the generation of these observed anions is the mitochondria, we have no direct evidence for this and the results may indeed reflect compartmentalised oxidative and glycolytic energy production. Indeed, studies on mitochondria in isolated rat skeletal muscle have demonstrated that lactic acidosis has differential effects on actively phosphorylating and non-phosphorylating mitochondria, suggesting that the effect of acidaemia may depend on local physiological conditions [14]. The rate of oxygen delivery to respiring tissue may also play a role in generating intermediates of Krebs acids, with several authors suggesting that hypoxia can cause an increase in intermediates of the citric acid cycle [15-18], although in the patients examined in this study none were significantly hypoxic at the time the sample was taken. Furthermore, it seems unlikely that the acidaemia per se is responsible for the increased levels of Krebs intermediates within our patient population given the normal values found in the patients with normal anion gap acidosis and the lesser elevation seen in the patients with DKA. An alternative explanation may be the generation of intermediates from anaplerotic pathways of metabolism, which may reflect enhanced protein catabolism in these patients
Other authors have proposed alternative hypotheses to explain how the 'missing' ions in the elevated anion gap may be generated. These include the choice of fluid used for resuscitation, the effects of hypoproteinaemia or the presence of other metabolites [19-21], in addition to various 'physical chemical' approaches, including the calculation of the strong anion difference [22]. This approach, which enjoys some popularity at the present time, represents an alternative way to express the principles of acidosis. The 'strong ion difference' model can be thought of as a restatement of the older concept of buffer base and thus, predictably, must approach the corrected anion gap, although some of its stronger proponents suggest that it can explain the generation of acidosis per se. We prefer to adhere to the concept that acidosis can best be regarded as an excess of protons over those normally found in physiological states, caused either from the loss of base or excessive net proton production.
We have previously referred to the difficulties of measuring plasma oxaloacetate in view of its extremely short half-life in aqueous solution. Given the increases in plasma concentrations of the other components of the Krebs cycle and the known existence of cytosolic oxaloacetate, it seems likely that some oxaloacetate does enter the plasma of these acidotic patient groups along with the other acids described here. There, it would spontaneously decarboxylate to pyruvate and its presence might only be inferred as a small deviation of the pyruvate:lactate ratio from that predicted on the basis of pH alone.
The results obtained from this study suggest that the role of anions principally associated with the Krebs cycle may be greater than previously thought in the generation of the anion gap in 'classic' lactic acidosis. In addition, these anions appear to have a significant role in the generation of the anion gap in patients with acidosis of unknown cause.
Conclusion
We have shown that the concentration of anions normally associated with the Krebs tricarboxylic acid cycle are elevated in appreciable quantities in patients with a metabolic acidosis. We propose that they may play a significant role in generating the anion gap. Further work in this laboratory is currently underway to explore the clinical implications of these findings.
Key messages
• Low molecular weight anions usually associated with intermediary metabolism are significantly elevated in the plasma ultrafiltrate obtained from patients with metabolic acidosis.
• These anions may contribute significantly to the elevated anion gap observed in patients with metabolic acidosis, in particular those of unknown aetiology.
Abbreviations
DKA = diabetic ketoacidosis.
Competing interests
This work was supported by The Special Trustees for St Thomas' Hospital, London. The authors have no financial interests relevant to the results of this research, nor are there any other circumstances that could potentially provoke a conflict of interest.
Authors' contributions
LGF conceived the study, participated in its design, collected patient samples and drafted the manuscript. WM developed and performed the enzyme assays, participated in the design of the study, performed the statistical analysis and helped to draft the manuscript. GAL participated in the design of the study, performed the initial mass spectrometry and helped in assay development. DFT collected patient samples and participated in study design. JMP helped in assay development and stability studies. PJH conceived the study, participated in its design, collected patient samples, helped in assay development and helped to draft the manuscript. All authors read and approved the final manuscript.
Figures and Tables
Figure 1 The concentration of various weak acids grouped by their underlying aetiology (mean ± upper range). DKA, diabetic ketoacidosis; NAG, normal anion gap.
Table 1 Concentrations of measured anions in plasma (μEq l-1)
Acid Patient group
Diabetic ketoacidosis Lactic acidosis Unknown origin Normal anion gap Controls
Mean SD p Mean SD p Mean SD p Mean SD p Mean SD
Citrate 454.02 194.11 ns 1,453.13 513.95 <0.01 335.63 69.82 nsa 239.1 105.1 ns 448.55 119.80
Isocitrate 421.93 352.56 0.02 704.6 347.6 <0.01 949.16 883.22 <0.01a 84.48 69.6 nsa 60.97 31.31
α-Ketoglutarate 413.41 158.48 <0.01a 547.72 344.98 <0.01a 651.51 203.06 <0.01a 72.97 67.34 nsa 79.17 106.74
Succinate 181.1 173.24 ns 358.27 112.49 <0.01a 340.04 128.74 0.02a 125.89 73.57 nsa 90.29 49.97
Malate 229.81 181.87 <0.01a 593.65 265.88 <0.01 485.16 189.67 <0.01a 95.07 117.59 ns 59.82 32.94
d-Lactate 157.34 67.85 <0.01 397.69 511.15 <0.01a 176.49 135.21 <0.01 69.27 55.39 <0.01 35.63 18.42
For each group, statistical analysis is presented relative to control value. aMann-Whitney non-parametric test. ns, not significant; SD, standard deviation.
==== Refs
Mizock BA Falk JL Lactic acidosis in critical illness Crit Care Med 1992 20 80 93 1309494
Cohen RD Woods HF Lactic acidosis revisited Diabetes 1983 32 181 191 6337898
Van Lambalgen AA Bronsveld W Van Den Bos GC Thijs LG Distribution of cardiac output, oxygen consumption and lactate production in canine endotoxin shock Cardiovasc Res 1984 18 195 205 6713447
Rackow EC Mecher C Astiz ME Goldstein C McKee D Weil MH Unmeasured anion during severe sepsis with metabolic acidosis Circ Shock 1990 30 107 115 2311201
Wright DA Forni LG Carr P Treacher DF Hilton PJ Use of continuous haemofiltration to assess the rate of lactate metabolism in acute renal failure Clin Sci (Lond) 1996 90 507 510 8697721
Tsai CS Spontaneous decarboylation of oxalacetic acid Canadian J Chem 1967 45 873 880
Stacpoole PW Wright EC Baumgartner TG Bersin RM Buchalter S Curry SH Duncan C Harman EM Henderson GN Jenkinson S Natural history and course of acquired lactic acidosis in adults. The DCA-Lactic Acidosis Study Group Am J Med 1994 97 47 54 8030656 10.1016/0002-9343(94)90047-7
Weil MH Afifi AH Experimental and clinical studies on lactate and pyruvate as indicators of the severity of acute circulatory failure (shock) Circulation 1970 41 989 1001 5482913
Smith I Kumar P Molloy S Rhodes A Newman PJ Grounds RM Bennett ED Base excess and lactate as prognostic indicators for patients admitted to intensive care Intensive Care Med 2001 27 74 83 11280677 10.1007/s001340051352
Hilton PJ Taylor J Forni LG Treacher DF Bicarbonate-based haemofiltration in the management of acute renal failure with lactic acidosis QJM 1998 91 279 283 9666950 10.1093/qjmed/91.4.279
Mecher C Rackow EC Astiz ME Weil MH Unaccounted for anion in metabolic acidosis during severe sepsis in humans Crit Care Med 1991 19 705 711 2026034
Haterwill M Waggie Z Purves L Reynolds L Argent A Correction of the anion gap for albumin in order to detect occult tissue anions in shock Arch Dis Child 2002 87 526 529 12456555 10.1136/adc.87.6.526
Kaplan LJ Kellum JA Initial pH, base deficit, lactate, anion gap, strong ion difference, and strong ion gap predict outcome from major vascular injury Crit Care Med 2004 32 1120 1124 15190960 10.1097/01.CCM.0000125517.28517.74
Tonkonogi M Sahlin K Actively phosphorylating mitochondria are more resistant to lactic acidosis than inactive mitochondria Am J Physiol 1999 277 C288 C293 10444405
Peuhkurinen KJ Takala TES Nuutinen EM Hassinen IE Tricarboxylic acid cycle metabolites during ischemia in isolated perfused rat heart Am J Physiol 1983 244 H281 H288 6824095
Folbergrova J Ljunggren B Norberg K Siesjo BK Influence of complete ischemia on glycolytic metabolites citric acid cycle intermediates and associated amino acids in the rat cerebral cortex Brain Res 1974 80 265 279 4154061 10.1016/0006-8993(74)90690-8
Hassel B Ilebekk A Tonnessen T Cardiac accumulation of citrate during brief myocardial ischaemia and reperfusion in the pig in vivo Acta Physiol Scand 1998 164 53 59 9777025 10.1046/j.1365-201X.1998.0400e.x
Taegtmeyer H Metabolic responses to cardiac hypoxia. Increased production of succinate by rabbit papillary muscles Circ Res 1978 43 808 815 709743
Morgan TJ The meaning of acid-base abnormalities in the intensive care unit: part III - effects of fluid administration Crit Care 2005 9 204 211 15774079 10.1186/cc2946
Rossing TH Maffeo N Fencl V Acid-base effects of altering plasma protein concentration in human blood in vitro J Appl Physiol 1986 61 2260 2265 3100499
Kellum JA Closing the gap on unmeasured anions Crit Care 2003 7 219 220 12793870 10.1186/cc2189
Stewart PA Modern quantitative acid-base chemistry Can J Physiol Pharmacol 1983 61 1444 1461 6423247
|
16277723
|
PMC1297631
|
CC BY
|
2021-01-04 16:04:56
|
no
|
Crit Care. 2005 Sep 13; 9(5):R591-R595
|
utf-8
|
Crit Care
| 2,005 |
10.1186/cc3806
|
oa_comm
|
==== Front
Crit CareCritical Care1364-85351466-609XBioMed Central London cc38101627772710.1186/cc3810CorrectionCorrection: Bispectral index versus COMFORT score to determine the level of sedation in paediatric intensive care unit patients: a prospective study Triltsch Andreas E 1Nestmann Grit 2Orawa Helmut 3Moshirzadeh Maryam 4Sander Michael 4Große Joachim 4Genähr Arka 4Konertz Wolfgang 5Spies Claudia D [email protected] Department of Anesthesiology and Intensive Care Medicine, Campus Benjamin Franklin, Charité University Hospital Berlin, Berlin, Germany2 Department of Pediatrics, Campus Benjamin Franklin, Charité University Hospital Berlin, Berlin, Germany3 Department of Medical Informatics, Biometry and Epidemiology, Charité University Medicine Berlin, Berlin, Germany4 Department of Anesthesiology and Intensive Care Medicine, Campus Charité Mitte, Charité University Hospital Berlin, Berlin, Germany5 Department of Cardiac Surgery, Campus Charité Mitte, Charité University Hospital Berlin, Berlin, Germany2005 2 9 2005 9 5 426 426 Copyright © 2005 Spies 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 cited.
==== Body
After the publication of this article [1] we noticed that the affiliation details were incorrect and should be as follows:
Andreas E Triltsch – Department of Anesthesiology and Intensive Care Medicine, Campus Benjamin Franklin, Charité University Hospital Berlin, Berlin, Germany
Grit Nestmann – Department of Pediatrics, Campus Benjamin Franklin, Charité University Hospital Berlin, Berlin, Germany
Helmut Orawa – Department of Medical Informatics, Biometry, and Epidemiology, Charité University Medicine Berlin, Berlin, Germany
Maryam Moshirzadeh, Michael Sander, Joachim Große and Claudia D Spies – Department of Anesthesiology and Intensive Care Medicine, Campus Charité Mitte, Charité University Hospital Berlin, Berlin, Germany
Wolfgang Konertz – Department of Cardiac Surgery, Campus Charité Mitte, Charité University Hospital Berlin, Berlin, Germany
==== Refs
Triltsch AE Nestmann G Orawa H Moshirzadeh M Sander M Große J Genähr A Konertz W Spies CD Bispectral index versus COMFORT score to determine the level of sedation in paediatric intensive care unit patients: a prospective study Crit Care 2005 9 R9 R17 15693968 10.1186/cc2977
|
0
|
PMC1297632
|
CC BY
|
2021-01-04 16:04:56
|
no
|
Crit Care. 2005 Sep 2; 9(5):426
|
utf-8
|
Crit Care
| 2,005 |
10.1186/cc3810
|
oa_comm
|
==== Front
Crit CareCritical Care1364-85351466-609XBioMed Central London cc38101627772710.1186/cc3810CorrectionCorrection: Bispectral index versus COMFORT score to determine the level of sedation in paediatric intensive care unit patients: a prospective study Triltsch Andreas E 1Nestmann Grit 2Orawa Helmut 3Moshirzadeh Maryam 4Sander Michael 4Große Joachim 4Genähr Arka 4Konertz Wolfgang 5Spies Claudia D [email protected] Department of Anesthesiology and Intensive Care Medicine, Campus Benjamin Franklin, Charité University Hospital Berlin, Berlin, Germany2 Department of Pediatrics, Campus Benjamin Franklin, Charité University Hospital Berlin, Berlin, Germany3 Department of Medical Informatics, Biometry and Epidemiology, Charité University Medicine Berlin, Berlin, Germany4 Department of Anesthesiology and Intensive Care Medicine, Campus Charité Mitte, Charité University Hospital Berlin, Berlin, Germany5 Department of Cardiac Surgery, Campus Charité Mitte, Charité University Hospital Berlin, Berlin, Germany2005 2 9 2005 9 5 426 426 Copyright © 2005 Spies 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 cited.
==== Body
After the publication of this article [1] we noticed that the affiliation details were incorrect and should be as follows:
Andreas E Triltsch – Department of Anesthesiology and Intensive Care Medicine, Campus Benjamin Franklin, Charité University Hospital Berlin, Berlin, Germany
Grit Nestmann – Department of Pediatrics, Campus Benjamin Franklin, Charité University Hospital Berlin, Berlin, Germany
Helmut Orawa – Department of Medical Informatics, Biometry, and Epidemiology, Charité University Medicine Berlin, Berlin, Germany
Maryam Moshirzadeh, Michael Sander, Joachim Große and Claudia D Spies – Department of Anesthesiology and Intensive Care Medicine, Campus Charité Mitte, Charité University Hospital Berlin, Berlin, Germany
Wolfgang Konertz – Department of Cardiac Surgery, Campus Charité Mitte, Charité University Hospital Berlin, Berlin, Germany
==== Refs
Triltsch AE Nestmann G Orawa H Moshirzadeh M Sander M Große J Genähr A Konertz W Spies CD Bispectral index versus COMFORT score to determine the level of sedation in paediatric intensive care unit patients: a prospective study Crit Care 2005 9 R9 R17 15693968 10.1186/cc2977
|
0
|
PMC1297636
|
CC BY
|
2021-01-04 16:04:56
|
no
|
Crit Care. 2005 Sep 9; 9(5):458
|
latin-1
|
Crit Care
| 2,005 |
10.1186/cc3820
|
oa_comm
|
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-11-r901627774510.1186/gb-2005-6-11-r90ResearchChromatin and siRNA pathways cooperate to maintain DNA methylation of small transposable elements in Arabidopsis Tran Robert K [email protected] Daniel [email protected] Bustos Cecilia [email protected] Renata F [email protected] Jorja G [email protected] Anders M [email protected] Jeffrey [email protected] Tom [email protected] Samson [email protected] Terri D [email protected] Steven E [email protected] Steven [email protected] Division of Basic Sciences, Fred Hutchinson Cancer Research Center, 1100 Fairview Ave N, Seattle, WA 98109, USA2 Howard Hughes Medical Institute, Fred Hutchinson Cancer Research Center, 1100 Fairview Ave N, Seattle, WA 98109, USA3 Department of Genetics and Pathology, Rudbeck Laboratory, Uppsala University, Uppsala, 751 85 Sweden4 Department of Molecular, Cell and Developmental Biology, University of California Los Angeles, Los Angeles, CA 90095, USA2005 19 10 2005 6 11 R90 R90 26 7 2005 26 8 2005 21 9 2005 Copyright © 2005 Tran 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.
Microarray-based profiling of the involvement of two DNA methyltransferases (CMT3 and DRM), a histone H3 lysine-9 methyltransferase (KYP) and an Argonaute-related siRNA silencing component (AGO4) in methylating target loci in Arabidopsis reveals that transposable elements are the targets of both DNA methylation and histone H3K9 methylation pathways, irrespective of element type and position.
Background
DNA methylation occurs at preferred sites in eukaryotes. In Arabidopsis, DNA cytosine methylation is maintained by three subfamilies of methyltransferases with distinct substrate specificities and different modes of action. Targeting of cytosine methylation at selected loci has been found to sometimes involve histone H3 methylation and small interfering (si)RNAs. However, the relationship between different cytosine methylation pathways and their preferred targets is not known.
Results
We used a microarray-based profiling method to explore the involvement of Arabidopsis CMT3 and DRM DNA methyltransferases, a histone H3 lysine-9 methyltransferase (KYP) and an Argonaute-related siRNA silencing component (AGO4) in methylating target loci. We found that KYP targets are also CMT3 targets, suggesting that histone methylation maintains CNG methylation genome-wide. CMT3 and KYP targets show similar proximal distributions that correspond to the overall distribution of transposable elements of all types, whereas DRM targets are distributed more distally along the chromosome. We find an inverse relationship between element size and loss of methylation in ago4 and drm mutants.
Conclusion
We conclude that the targets of both DNA methylation and histone H3K9 methylation pathways are transposable elements genome-wide, irrespective of element type and position. Our findings also suggest that RNA-directed DNA methylation is required to silence isolated elements that may be too small to be maintained in a silent state by a chromatin-based mechanism alone. Thus, parallel pathways would be needed to maintain silencing of transposable elements.
==== Body
Background
DNA cytosine methylation is an ancient process, found in both prokaryotes and eukaryotes, and catalyzed by a single family of methyltransferases [1]. In prokaryotes, cytosine-5 methyltransferases protect target sites from cleavage by partner restriction endonucleases, but in eukaryotes, the function of DNA methylation is less clear. In organisms that retain DNA methylation, including plants, most animals and some fungi, it has been speculated that DNA methylation provides genomic immunity against mobile elements [2,3]. This hypothesis has been difficult to test in vertebrates, because most CG dinucleotides are heavily methylated in both genic and intergenic regions [4]. In fungi and plants, however, the localized nature of DNA methylation makes it possible to identify sequences that are targeted for DNA methylation. For example, in Neurospora, DNA methylation occurs at repeated sequences that are targeted for point mutation [5]. In plants, transposable elements are heavily methylated relative to genic regions, suggesting that the silencing that accompanies DNA methylation is a means of defending against transposition [3,6,7]. An additional form of DNA methylation is found in the model plant Arabidopsis, where short dense CG methylation clusters are occasionally found in genic regions that are otherwise devoid of methylation [8].
Although many DNA methylation targets are known, it has been unclear how these sites are recognized by DNA methyltransferases. The Dnmt3 subfamily of DNA methyltransferases, which includes Arabidopsis DRM1 and DRM2, can methylate de novo [9], but there are no known sequences in common among target sites. Recent work in Arabidopsis has implicated the small interfering (si)RNA machinery in targeting de novo methylation [10-12], and a large number of transposon-directed siRNAs have been sequenced [13,14]; however, the mechanism by which siRNA production leads to de novo DNA methylation is not known. Another open question is how some forms of DNA methylation are maintained during rounds of cell division. In the case of CG sites, a member of the Dnmt1 subfamily of DNA methyltransferases maintains methylation by specifically methylating hemi-methylated sites behind the replication fork [15], but in cases of non-CG methylation, there does not appear to be a comparable reaction. Non-CG methylation in Neurospora is maintained by the action of a histone H3 lysine-9 (H3K9) methyltransferase [5], so the successive action of a histone methyltransferase and a DNA methyltransferase suffices to maintain methylation indefinitely. A similar maintenance mechanism occurs for CNG sites in Arabidopsis, where the KRYPTONITE (KYP = SUVH4) H3K9 methyltransferase directs methylation by the CHROMOMETHYLASE3 (CMT3) CNG methyltransferase [16,17]. These findings have led to a general model whereby siRNAs direct de novo methylation by DRM1 and DRM2, CG sites are maintained by the Dnmt1 ortholog, MET1, and CNG sites are maintained by the successive action of KYP and CMT3 [10,18].
These insights into DNA methylation mechanisms were obtained using sensitive reporter systems chosen because they display striking epigenetic silencing phenotypes [18]. As a result, they were not designed to reveal the spectrum of target sites acted upon by these various DNA methylation pathways. An alternative approach is to look at large numbers of sites for changes in methylation levels when mutations in various components of epigenetic silencing are introduced. In previous work, we used a microarray-based method for profiling methylation patterns to detect changes that occur in a cmt3 mutant line [7]. This analysis revealed hypomethylation of a subset of randomly chosen sites. This subset was enriched in transposon-derived sequences, consistent with DNA methylation playing a role in genome defense against transposable elements [7,19]. The small scale of the analysis did not allow us, however, to determine whether there are preferences for different types or locations of elements, as has been suggested for CMT3 [20].
Here we present a large-scale analysis of methylation patterns in mutants that are involved in CNG methylation. We found that CMT3 and KYP targets are transposons of all types and show a distribution along the chromosomes that is similar to that of the bulk of elements in the genome. In contrast, we found relatively few DRM and AGO4 targets scattered throughout the chromosomes, and these are significantly enriched in small isolated transposon-derived sequences. Our findings suggest a special role for RNA-directed DNA methylation in silencing mobile elements that are scattered along chromosome arms.
Results
Profiling of CNG methylation
To profile methylation patterns, DNA samples are treated with a methylation-sensitive restriction endonuclease and are size-fractionated by sucrose gradient centrifugation [7]. The low molecular weight fraction is collected and labeled with either of two fluorescent dyes, such that two samples can be compared by standard microarray analysis. If one sample is derived from a mutant in which methylation is reduced, then affected sites will be more frequently cleaved by the restriction endonuclease relative to wild-type. When cleavage results in an assayed fragment sedimenting faster than the 2.5 kb cutoff used in the fractionation, then there will be a stronger signal for mutant than wild-type. Conversely, if the mutant causes hypermethylation of a site, then the wild-type signal will be higher than the mutant signal. In this way, we can detect whether or not changes occur in methylation patterns from the ratio of the two dye signals, scoring as positive targets only those that are statistically significant based on repeated measurements from different biological samples [21]. Positive targets can be scored as either hypomethylated or hypermethylated.
In our original methylation profiling study, we assayed PCR-amplified fragments from loci known or suspected to be targets of DNA methylation and also PCR-amplified single-copy approximately 700 base pair (bp) fragments chosen at random from the Arabidopsis genome sequence, 360 in all [7]. In a subsequent study, we increased the size of this array to include 597 randomly chosen loci [8]. In the present study, we have used this 'random-PCR' array, as well as a 'gene-oligo' array consisting of 26,090 oligonucleotides (70-mers) representing essentially all annotated Arabidopsis genes [22], of which 10% (2,633 of 26,090) are identified as containing transposons and repetitive elements detectable by RepBase [23] and RECON [24] analysis (unpublished observations). The gene-oligo array thus samples both distal and transposon-rich pericentric chromatin regions of the genome.
To detect CNG methylation, we used the MspI restriction endonuclease to digest DNA samples from five mutant lines: cmt3, kyp, ago4, drm1/2 (double mutant) and cmt3 drm1/2 (triple mutant), each paired with its wild-type parental line. Using the random-PCR array, we detected five loci as hypomethylated in the cmt3 mutant background; these represent single-copy targets of CMT3 that are methylated on the first C of one or more CCGG sites, a modification that blocks MspI digestion.
We then asked whether any of these methyl-CNG positive loci were also detected as targets of other proteins assayed in this way. KYP yielded four (hypomethylated) targets from among the 597 randomly chosen single-copy loci, of which three were the same as the CMT3 targets (Table 1). Using the gene-oligo array, we detected 536 CMT3 targets (498 hypomethylated and 38 hypermethylated in cmt3; Figure 1a,b), and 81 KYP targets (79 hypomethylated and 2 hypermethylated in kyp; Figure 1c), of which 79 were also CMT3 targets (Figure 2) [25]. This nearly complete overlap shows that the interplay between CMT3 and KYP found for sensitive reporter loci [16,17] is also true for at least a large fraction of CMT3 targets genome-wide. All CMT3 or KYP positives on the random-PCR array and >90% of the positives on the gene-oligo array were hypomethylated in the mutant, as would be expected if these enzymes work in tandem to maintain CNG methylation. The close agreement between two very different array platforms [26] provides a high degree of confidence in our conclusions.
AGO4 yielded one hypomethylated target locus that was also a CMT3 target in the gene-oligo array and one (hypomethylated) target in the random-PCR array (Figures 1d and 2). In addition, we detected two targets of DRM1/2 in the random-PCR array (Table 1) and ten DRM1/2 targets (nine hypomethylated and one hypermethylated in drm1/2; Figure 1f,g) in the gene-oligo array (Figure 2). The low number of AGO4 and DRM1/2 targets relative to the high number of CMT3 and KYP targets on the gene-oligo array suggests that AGO4 and DRM1/2 play only a minor role in maintaining CNG methylation throughout the genome. We also assayed a triple mutant combination of cmt3 drm1/2 and observed extensive overlap for both the random-PCR array (Table 1) and the gene-oligo array (313 hypomethylated and 33 hypermethylated in cmt3 drm1/2; Figures 1e and 2).
CMT3, KYP and DRM target transposable elements
In our original study, we noticed that all four randomly chosen loci that were positive for CMT3 represent transposable elements [7]. This correspondence was especially striking considering that the loci were chosen to be single-copy in the genome, so that these represent the rarest class of transposable elements. This conclusion is confirmed in the present study. For the purposes of our analysis, we considered only loci where methylation blockage of a single site could cause a fragment to sediment more rapidly than 2.5 kb, thus resulting in its exclusion from the DNA used as probe (see Materials and methods and [8]). By this criterion, only a subset of restriction sites overlapped by a repeat or transposon would be scored as affected by the mutation. Nevertheless, we found a preponderance of transposable elements in this class for CMT3 (Table 2). Likewise, three of the four single-copy targets of KYP were scored as transposable elements. This preference for transposable elements in CMT3 targets was amply confirmed on the gene-oligo array, with 63% (104/164) of loci with sites falling within transposons, compared with only 13% (907/7,032) for all loci on the array. These 104 elements include long terminal repeat (LTR), long interspersed element (LINE) and short interspersed element (SINE) retrotransposons, DNA transposons and helitrons. KYP showed a similar preference to CMT3 with 68% (26/38) of loci falling within transposons. We conclude that transposable elements of all types are by far the predominant target of the CMT3-KYP system. We also found a preference of DRM1/2 for transposable elements (Table 2).
If the targets of CMT3 result from a general preference of these DNA methylation pathways for transposable elements, then we would predict that the distribution of targets would approximately correspond to that of transposable elements along the chromosome. We tested this by comparing the distribution of target distances for CMT3 from the centromeric gap to the locations of repetitive elements. We searched the Repbase library of consensus sequences [23] against the Arabidopsis genomic sequence to determine the distribution of repeats. Most transposable elements in Arabidopsis are located near the centromeric gap, gradually decreasing in abundance towards the telomere (Figure 3a). Similarly, CMT3 targets, whether repetitive or single-copy, are highly abundant close to the centromeric gap, decreasing as one moves distally along the chromosome arms.
We wondered whether the KYP and DRM1/2 targets also showed a transposon-like distribution. This would be expected if KYP and DRM1/2 targets mostly comprise a representative sample of CMT3 targets. Indeed, the KYP target distribution along the chromosome nearly superimposes over that for CMT3 (Figure 3b), which together with the nearly complete inclusion of KYP targets within the set of CMT3 targets, indicates that KYP and CMT3 have the same target preferences. In contrast, DRM1/2 targets are scattered throughout the chromosome arms, with only one of ten targets in the most proximal 2 Mb where about half of the CMT3 targets are found; this difference in the distribution of elements is statistically significant (p = 0.013, Fisher's exact test). We conclude that CMT3 and KYP target transposable elements in general, whereas DRM1/2 is required primarily at elements that are distally located along the chromosome.
Our conclusion that DRM1/2 targets are distinct from CMT3 targets is unlikely to have resulted from false positives in the DRM1/2 dataset. As previously mentioned, DRM1/2 targets are enriched in transposons. In addition, the close correspondence between CMT3 and KYP distributions, even considering just the 85% of CMT3 targets that are not KYP targets (Figure 3c), implies that the cutoff criteria used for target detection were very conservative. As indicated below, it appears that the large majority of KYP targets were simply too weak relative to CMT3 targets to be detected in the context of a whole-genome analysis. Furthermore, the CMT3 DRM1/2 dataset provides an independent test of the stringency of our cutoff criteria, because we would expect it to include all of the CMT3 targets; but it actually is a smaller set that only partially overlaps. This partial overlap is evidently not attributable to false positives in both datasets, because the distributions of CMT3 and CMT3 DRM1/2 targets essentially superimpose (Figure 3c), even considering just the 21% of CMT3 targets that are not CMT3 DRM1/2 targets. This indicates that the small number of DRM1/2 targets results from strict cutoff criteria that identify a subset of truly affected loci.
Bisulfite sequencing of CNG methylation targets
To confirm and quantify the array results, we performed bisulfite sequencing on a selection of target sites. We chose one positive example from the random-PCR array and five from the gene-oligo array. For locus 4:1813417-1814107 (Mu-PCR), detected as a target of CMT3, KYP and CMT3 DRM1/2 on the random-PCR array, wild-type methylation levels averaged 88% for the 11 CG sites and 47% for the 10 CNG sites assayed by bisulfite sequencing (Figure 4a; Table 3). In the cmt3 mutant background, the average level dropped to 63% for CG and to 1% for CNG methylation. This drastic decrease in CNG methylation is as expected considering that CMT3 is known to be responsible for nearly all of this modification at selected loci [27,28]. In a kyp mutant background, the average level dropped to 74% for CG and to 16% for CNG methylation. In this fragment, methylation of a single MspI site would account for a change in fragment size and its differential fractionation prior to microarray analysis. Remarkably, the kyp-induced decrease in methylation at the MspI site itself was only about one third (from 11 to 7 methyl-Cs of the 19 determined for this site; Table 3), confirming that methylation profiling on the random-PCR array is capable of detecting an intermediate drop in methylation levels.
Bisulfite sequencing of targets detected on the gene-oligo arrays confirmed that the positives detected on these arrays indeed reflect changes in the degree of methylation. For example, locus A000229 was detected as hypomethylated in both cmt3 and drm1/2 mutants, and bisulfite sequencing shows a reduction of methylation at one flanking MspI site in cmt3 and at the other flanking MspI site in drm1/2 (Figure 4b,c; Table 3). Interestingly, a reduction in methylation at the first MspI site was also seen in kyp. This partial loss of CNG methylation in kyp that was not detected on the gene-oligo array could account for the low fraction of CMT3 targets that are also targets of KYP on this array (Figure 2b,c). Of the three other loci examined in cmt3 and kyp mutants, all showed a major loss of CNG methylation in cmt3, one showed a major loss of CNG methylation in kyp, one showed a minor loss of CNG methylation, and one showed no loss (Figure 4c–e). This consistently strong effect of cmt3 and variable effect of kyp on CNG methylation at unselected sites is in agreement with studies of cmt3 and kyp/suvh4 mutants at particular loci [16,17,19,27,28].
An unexpected finding was that CG methylation levels dropped five- to tenfold at two loci when both classes of de novo/CNG methyltransferases were absent (cmt3 drm1/2 in Figure 4c,e). This effect might be caused by an occasional failure of the MET1 CG maintenance methyltransferase, leading to a dependence on methyltransferases that do not require a hemi-methylated substrate [29,30].
Methylation of small transposable elements is dependent on DRM1/2 and AGO4
We wondered whether there is an inherent difference between transposons that require DRM1/2 for methylation and those that do not. Bisulfite sequencing revealed major losses of CNG methylation in drm1/2 mutants at three loci: A000229 and two loci corresponding to SINE3 elements (Figure 4c,e,f). The approximately 160 bp size of these SINE3 targets of DRM1/2 contrasts with the >5 kb size of the three loci that were not affected by drm1/2. Only one of these SINE3 elements is present in the parental strain of ago4 (Ler), and this showed a major drop in CNG methylation, whereas all three large elements that were unaffected by drm1/2 were also unaffected by ago4. Taken together, our results are consistent with the possibility that DRM1/2 and AGO4 are required to maintain DNA methylation at small, but not large transposable elements. The small size and low abundance of DRM1/2 and AGO4 targets might explain why so few of them were detected relative to CMT3 and KYP targets.
To determine the generality of our observations on DRM1/2 and AGO4 targets, we included bisulfite sequencing data from previous studies [11,12] with our own in comparisons of levels of methylation reduction in mutants. drm1/2 and ago4 mutants show strongly correlated CNG methylation reductions (r = 0.91, p = 0.0002). In addition, methylation reduction shows a clear association with element size for both ago4 and drm1/2 (Figure 5a), with small elements preferentially affected. No significant associations with element size were seen for either cmt3 or kyp (Figure 5b). Methylation reduction is not attributable to differences in element type, because heterogeneous classes of sequences were found in both size classes, with DNA transposons and LINE elements in the large size class and SINEs, tandem and inverted repeats and genes in the small size class. Methylation reduction is also not attributable to differences in element abundance, because no association is seen between changes in methylation and the estimated copy number of elements in the genome (data not shown). It thus appears that AGO4 and DRM1/2 work together to maintain DNA methylation and silencing of small elements.
Discussion
We have used DNA methylation profiling to assay the effects of mutations in Arabidopsis genes that have been implicated in gene silencing and epigenetic inheritance. This has led to the identification of common targets of a DNA methyltransferase and a histone modifying enzyme. The original reports of connections between these two silencing paradigms were major breakthroughs in the epigenetics field, and we have shown that this connection is widespread and not confined to a few selected and unusual loci. We also have demonstrated that the targets of both DNA methylation and histone H3K9 methylation pathways are transposable elements, irrespective of element type and position. Furthermore, we have shown that the de novo methylation pathway targets a selected subset of elements, and we provide data suggesting that short elements are preferentially dependent on an RNAi-mediated de novo methylation pathway.
We have found that nearly all CNG targets of the KYP (SUVH4) H3K9 methyltransferase are also CMT3 targets. We attribute the detection of only 79 KYP targets from among the 536 CMT3 targets to the limited sensitivity of our large-scale profiling assay, where stringent criteria are needed to eliminate false positives among the 26,090 loci scored. In support of this interpretation, bisulfite sequencing revealed a consistent partial loss of CNG methylation in kyp mutants. This stronger effect of CMT3 than KYP is consistent with previous work showing that CMT3-dependent DNA methylation at the SUP and PAI2 loci requires KYP (SUVH4), whereas DNA methylation of the inverted repeat PAI1-PAI4 locus does not [16,17]. This is also consistent with the finding that three transposons showed a greater loss of CNG methylation in cmt3 than in kyp mutants [20]. A possible reason for the variable effect of kyp is that other histone methyltransferases function in this capacity, and there are about a dozen kyp homologues in the Arabidopsis genome [31].
In the case of mutants in AGO4, a member of the Argonaute family of RISC complex components, only two sites of CNG hypomethylation were seen; this is not unexpected insofar as studies of the SUP locus showed weak effects of ago4 mutants relative to cmt3 and kyp mutants [12]. Similarly, mutations in the drm1/drm2 de novo methyltransferases affected only a few loci in our assay, consistent with evidence that CNG methylation is maintained primarily by the CMT3-KYP pathway [6].
As in our original methylation profiling study, we found that transposons were primary targets of CMT3 [7], and the inclusion of KYP and DRM1/2 extends this conclusion to two of the three major pathways for DNA methylation in Arabidopsis. Furthermore, the use of a gene-oligo array that samples most of the sequenced genome thoroughly confirms that the targets are repeats of all types, including LTR and non-LTR retrotransposons, helitrons, and MuDR and other classes of DNA transposons, and not simply a limited sample of common elements.
The use of a comprehensive gene array also allowed us to detect target differences that were not apparent from studies of single loci. In particular, we detected a preferential dependence of distally located elements on DRM1/2. It is possible that this preferential dependence results simply from the elements' distal location. If so, then we would expect to find that distal elements in general would show dependence on DRM1/2. However, two elements that were chosen for bisulfite sequencing because of their distal location (one Mu-4802 and the other TA11-4217) showed no significant loss of CNG methylation in drm1/2. Therefore, it appears that some property other than distal location per se is responsible for DRM1/2 dependence.
It is possible that DRM1/2 preferentially targets elements with corresponding siRNAs. This is suggested by the strong correlation between the degree of methylation loss in drm1/2 and in ago4, a siRNA-mediated silencing component. Indeed, most of the sequences included in our analysis that depend on DRM1/2 or AGO4 have corresponding siRNAs [13,14,19,32,33]. However, transposable elements of all types have corresponding siRNAs [13,14], indicating that siRNAs, and by inference DRM1/2 and AGO4, target transposable elements in general. Therefore, some other feature must determine whether the siRNA pathway is required to maintain DNA methylation of distal elements.
Our finding that element size is strongly associated with the degree of DRM1/2- and AGO4-dependent methylation provides a rationale for the distal location preference. Most transposable elements tend to cluster in pericentric regions in plants and animals, leading to large silent heterochromatic blocks, whereas elements that insert distally tend to be isolated. Small isolated elements might be more difficult to silence than large clustered elements [34-36], and mobile elements that are not silenced can damage the genome by replicative transposition [37]. In Arabidopsis, SINEs show a distribution along the chromosome [38] that is not unlike the distal distribution that we report for DRM1/2 targets. Therefore, the preferential dependence of small distal elements on DRM1/2 and AGO4 might reflect an adaptation to defend against SINEs, which would otherwise escape silencing by the chromatin-based CMT3-KYP machinery. The dependence of small elements on the DRM1/2-AGO4 pathway for DNA methylation provides support for the hypothesis that siRNA-mediated methylation reinforces unstable silencing of such elements [39].
Materials and methods
Sample preparations
Arabidopsis thaliana mutants were previously described [10,12]. To control for background variability, lines were constructed by backcrossing parental mutant lines with either Ler (for cmt3, kyp, ago4 and drm1 drm2 cmt3) or Ws (for drm1 drm2), which served as the corresponding wild-type controls. Whole 5 week old plants were used to prepare genomic DNA using the CTAB extraction method [40]. After ethanol precipitation, DNA samples were treated with DNase-free ribonuclease (Roche, Indianapolis, IN, USA) and precipitated by addition of 3 M sodium acetate and ethanol, then pelleted by centrifugation and air-dried. Bisulfite treatment of DNA, cloning into a Topo TA vector (Invitrogen, San Diego, CA, USA) and DNA sequencing were performed as described [27]. Primer sets are listed in Table 4.
Microarray construction
Primer selection, amplification and spotting in duplicate onto glass slides have been described in our original methylation profiling study [7]. The random-PCR array in the present study consisted of 960 loci of which 597 were randomly chosen approximately 700 bp single-copy loci and 363 were selected control loci of different lengths [8]. The 597 loci were selected as random non-overlapping 1 kb single-copy fragments from a non-redundant database consisting of contigs representing A. thaliana chromosomes 2 and 4 taken from a December 1999 version of the A. thaliana TIGR assembly [41]. Chromosomes 1, 3 and 5 were pieced together from the A. thaliana genome project clone table from an August 2000 version of the A. thaliana TIGR assembly. Most of the selected loci were amplified from segments of known targets of the gene products under study [12,27]. Primers were designed and checked by BLAST searching to avoid redundancy as described [7]. The gene-oligo array consisted of 26,090 70-mer oligonucleotides from the Arabidopsis genome oligo set Version 1.0 [41], arrayed and hybridized as previously described [7].
Hybridization to microarrays
Samples were dissolved in Tris-EDTA (TE) buffer and 50 to 60 μg aliquots were subjected to digestion by addition of 200 units of restriction endonuclease for 3 to 4 h. MspI endonuclease was obtained from New England Biolabs (New England Biolabs, Ipswich, MA, USA). Digested DNAs were size-fractionated on 5% to 30% sucrose gradients as described [42]. Aliquots of DNA fractions were examined by agarose gel electrophoresis to verify DNA fragment size and concentration. Fractions in the <2.5 kb range were pooled, precipitated by addition of ethanol, and fluorescently labeled with either Cy3 or Cy5-dCTP (Amersham, Piscataway, NJ, USA) by random priming (Invitrogen) as described [42]. Oppositely labeled samples from mutant and wild-type were mixed together and hybridized to microarrays on glass slides and processed as described [42].
Data processing
Slides were scanned using a GenePix 4000 fluorescent scanner (Axon Instruments Inc, Union City, CA, USA). For each mutant comparison, three to four biological replicates were performed, all with dye-swaps. A lowess normalization was applied to the gene-oligo array to correct for non-linearity in this dataset. Methylation profiles were analyzed and p values assigned using Cyber-T microarray analysis software, which applies a Bayesian T-statistic method [21]. The data-versus-model weighting factor was adjusted to 8 for the random-PCR array and to 6 for the gene-oligo array. A window size of 161 was used for the random-PCR array and 201 for the gene-oligo array. Bayesian-derived p values were adjusted for multiple hypotheses testing using a Bonferroni correction (p = 0.05) for the random-PCR array and a false discovery rate of p = 0.05 for the gene-oligo array. Note that the use of statistical criteria to delineate targets results in greatly reduced sensitivity of the gene-oligo array relative to the much smaller random-PCR array. An additional criterion for significance was implemented using 'self versus self' control experiments to assess experimental variation within the system. Accordingly, a lower-bound threshold for the log2 methylation ratios (cy3/cy5) was defined as 3 standard deviations for the random-PCR array (4 for the gene-oligo array) from the corrected mean of the distributions of log2-transformed ratios.
Analysis of methylation profiling data
To facilitate comparison of datasets, we implemented a relational database (mysql) with a web browser display (Methprof [25]). Methprof has utilities for processing raw data and for statistical analysis by CyberT [21]. Methprof displays positive hits based on CyberT analysis for individual and combined datasets, together with a graphical chromosomal map of all the loci. Each hit in a Methprof table links to annotation data and displays user-provided descriptions, the number and identity of datasets in which it is positive, and whether the hit is hypo- or hyper-methylated.
In addition, a Javascript program (Region Viewer, developed by us) was implemented to display annotation and restriction site data for loci on the PCR-based array, and Methprof was adapted to display similar information for the oligo-based array. For each locus, a 'neighborhood' centered on a locus was defined such that blockage of a methyl-sensitive restriction site anywhere in the region could increase a fragment from less than to greater than the 2.5 kb cutoff. The blocked site was inferred as that most likely to have caused the depletion from the <2.5 kb fraction, ignoring ambiguous cases. Gene information was parsed from Genbank entries. Repeat information was generated using the program Censor4.1 [43] on the A. thaliana repeat library athrep.ref [23]. Repeat information was also obtained by BLASTN searching of an A. thaliana library of consensus sequences generated by the RECON program (Zhirong Bao, personal communication) [24]. Data and maps used in this study are available for querying, browsing or downloading using Methprof [25], and all raw data can be downloaded from the GEO database under Accession number GSE3109 [44].
Acknowledgements
We thank members of our laboratories for helpful discussions, and Ryan Basom for assistance with data processing. R.K.T. was supported by an NIH training grant (T32CA09657), C.D.B. by a graduate fellowship from the Department of Education, Universities and Research of the Basque Government, S.K. and J.G.H. by a grant from the National Science Foundation (DBI 0234960), A.M.L. and S.E.J. by NIH grant GM60398.
Figures and Tables
Figure 1 Raw data plots for the gene-oligo array. For each genotype pair, the average log2(exp/ref) ratio is plotted versus the corresponding average log2 fluorescent intensity. Each plot contains the results of six array measurements, that is dye-reversed measurements on three biological replicates. All data were lowess normalized as described in the Materials and methods section. Red dots represent statistically significant target loci, where those with positive log ratios indicate hypomethylation and those with negative log ratios indicate hypermethylation. Blue dots represent the rest of the loci.
Figure 2 CNG methylation targets of epigenetic silencing components. (a) Venn diagram summaries of positive loci using random-PCR arrays in cmt3, kyp, drm1/2 and ago4 mutant backgrounds. Loci were scored as positive if methylation was significantly changed in the indicated mutant relative to the Ler wild-type background. (b) Venn diagram summaries of positive loci using gene-oligo arrays, where cmt3, kyp, drm1/2 and ago4 were in a Ler (clk-st) and crm3 drm1/2 was in a Ws wild-type background. Gene-oligo and random-PCR datasets of targets are available with a graphical interface for browsing and for downloading [25]. (c) Table showing the number of positives and overlaps for each mutant class. Mutants are color coded for clarity in the Venn diagrams.
Figure 3 Location of transposable elements, CMT3, KYP and DRM1/2 targets along chromosome arms. (a) Transposable elements and CMT3 targets. (b) Comparison of KYP, DRM1/2 and CMT3 DRM1/2 targets to CMT3 targets. (c) Comparison of all CMT3 targets to the subset of CMT3 targets that are not also KYP targets, and comparison of CMT3 targets to the subset that are not also CMT3 DRM1/2 targets. To map repeats relative to the centromere, Repbase library sequences were searched using BLASTN with default Repbase parameters against TIGR Release 5 of the Arabidopsis genome sequence. All CMT3 targets, single-copy CMT3 targets and DRM1/2 targets from the gene-oligo array (Figure 2b) were also mapped on the same scale. The fraction of the total number of hits within each 1 Mb bin is shown. To compensate for differences in oligo abundance on the array, bins were normalized by dividing each raw fraction by the fraction of oligos in the bin.
Figure 4 Methylation occupancies of selected target loci determined by bisulfite sequencing. Elements are: (a) Mu-PCR (locus 4:1813417-1814107 on random-PCR array hypomethylated in cmt3, kyp and cmt3 drm1/2); (b) 229-R1 (left side of A000229 on gene-oligo array hypomethylated in cmt3 and drm1/2); (c) 229-R2 (right side of A000229); (d) Mu-4802 (A004802 hypomethylated in cmt3, kyp and cmt3 drm1/2); (e) TA11-4217 (A004217 hypomethylated in cmt3); (f) SINE3-5300 (A005300); (g) SINE3-11193 (A011193 hypomethylated in drm1/2). Wild-type lines are Ler, clk-st (parental line of cmt3, kyp, and ago4 derived from Ler) and Ws (parental line of drm1/2). See Table 3 for details.
Figure 5 Methylation by DRM1/2 and AGO4 is associated with the size of their targets. (a) The loss of methylation for each locus is calculated from the reduction seen in drm1/2 and ago4 when measured by bisulfite sequencing (drm1/2: correlation coefficient r = 0.82, p < 0.003; ago4: r = 0.90, p = 0.0002). The fraction methylated is the ratio of mutant to wild-type percentages listed in Table 3. Regression lines are shown for clarity. (b) A similar comparison of CMT3 and KYP reveals no significant associations (cmt3: r = -0.48, p = 0.2; kyp: r = -0.32, p = 0.5), so no regression lines are shown. The comparisons include data reported in this study supplemented with previously published data for other loci [11,12].
Table 1 Loci scored as CNG methylation targets in mutants on the random-PCR array
Genomic location* Gene ID TIGR designation drm1
drm1 drm2
cmt3 kyp ago4 drm2 cmt3
Random loci
1:14147723-14148423 Non-LTR retrotransposon family (LINE) - - -
3:18799179-18799859 At3g50620 Intergenic - - -
4:1813417-1814107 Mutator-like transposase family - - -
2:8651310-8652034 At2g20020 Expressed protein - -
2:11049695-11050396 At2g25900 Zinc finger (CCCH-type) family protein - -
3:19989359-19990040 At3g53960 Proton-dependent oligopeptide transport -
2:5627581-5628302 non-LTR retrotransposon family (LINE) -
1:8273192-8273908 At1g23320 Alliinase family protein contains Pfam -
1:22107946-22108642 At1g60020 Copia-like retrotransposon family -
2:7532595-7533279 At2g17305 Hypothetical protein -
2:13188318-13189035 At2g30970 Aspartate aminotransferase, mitochondrial -
*Chromosome number:span of fragment in TIGR map April 2004. Dashes indicate loss of methylation and blank spaces indicate no significant change.
Table 2 CNG methylation changes within transposable elements
Mutant Total targets Transposon targets* p value†
Random-PCR array
All loci 597 12
cmt3 5 2 0.01
kyp 4 3 <0.0001
ago4 1 0 1
drm1/2 2 0 1
drm1/2 cmt3 7 3 0.001
Gene-oligo array
‡
All unambiguous loci 6,950 907 (13%)
cmt3 targets 164 104 (63%) <0.0001
kyp targets 38 26 (68%) <0.0001
ago4 targets 0 0 -
drm1/2 targets 6 4 (67%) 0.005
drm1/2 cmt3 targets 103 67 (65%) <0.0001
*Total number of possible transposable element hits among the 597 loci on the random-PCR array or among the 26,090 loci on the gene-oligo array. †Using Fisher's exact test. ‡All numbers represent only those loci for which the MspI cut site responsible for differential fractionation could be unambiguously determined.
Table 3 Percent DNA methylation of elements in mutant lines1
Genotype
Ler (wt) Clk-st (wt) Ws (wt) cmt3 kyp ago4 drm1/2 drm1/2 cmt3
Locus Size CG CNG CNN CG CNG CNN CG CNG CNN CG CNG CNN CG CNG CNN CG CNG CNN CG CNG CNN CG CNG CNN
MU-48022 4.8 84 34 19 75 28 13 79 25 9 84 1 4.8 84 0.5 1 64 30 19 74 27 9.4 79 0.5 2.8
TA11-42173 6 - - - 81 62 12 88 77 12 75 7.5 10 80 42 13 82 65 12 81 60 5.5 9.6 0 0.2
SIN3-53004 0.2 80 78 38 80 70 59 88 72 58 90 10 6.8 83 73 54 58 2.5 0 58 1.7 0 15 2.5 0
MU-PCR5 >7 88 47 8 80 43 7.5 - - - 63 1.1 4.8 74 16 7 85 43 8 81 54 7.5 89 0.5 4.9
ATSN16 0.2 75 70 24 49 8.7 8.7 82 53 29 42 14 0.8 53 30 3.3 28 0 0.2
SUP7 1.2 16 55 16 8.9 0 5.7 5.9 1.7 1.9 16 20 3.6 14 39 8.7 10 0 0
MEA-ISR8 1.3 95 58 26 87 47 18 89 13 13 92 25 17 81 0 0.7 87 0 0 87 0 0
FWA9 0.5 84 16 6.8 89 18 4.2 93 2.9 6.4 87 2.5 0.7 81 0 0 88 0 0
NOS-PRO10 0.4 71* 45* 38* 54 19 24 77 21 10 77 23 3.2 61 1.3 0.3
SIN3-1119311 0.2 N/A N/A N/A N/A N/A N/A 91 90 56 N/A N/A N/A N/A N/A N/A N/A N/A N/A 81 2.5 0.3 N/A N/A N/A
229R112 ? - - - 93 58 13 93 58 7 83 1 1.7 88 17 6.3 88 54 5.1 87 49 1.9 76 0.5 0.5
229R212 ? - - - - - - 94 48 2.1 - - - - - - - - - 73 0 0 - - -
1The fraction methylated reported in Figure 5 is the ratio of mutant to wild-type. For example, the fraction of TA11-4217 that is methylated in ago4 is 65% (ago4)/62% (Clk-st) = 1.05. Numbers in bold are from previously published work. All loci except for 229R1 and 229R2 were used in the scatter plots in Figure 5. We determined the size of each element based on the Repbase consensus sequence, except for published examples, in which we used the information provided in the references below. 2MU-4802 is an isolated Mutator-like element annotated as At1g17275. 3TA11-4217 is an isolated LINE element annotated as At1g29650 with very close homology to TA11. 4SIN3-5300 is a SINE3 element located between At3g60130 and At3g60140. 5MU-PCR is a complex locus of multiple transposable elements inserted into one another present at the hk4S heterochromatic knob on chromosome 4 in Ws and in pericentric heterochromatin on chromosome 4 in Ler [19]. 6AtSN1 is a SINE1 element [33]. Bisulfite sequencing data derived from [12]. 7SUP is the Arabidopsis SUPERMAN gene [45]. Bisulfite sequencing data derived from [12,16,27,46]. 8MEA-ISR is a locus composed of tandem direct repeats located downstream of the Arabidopsis MEDEA gene [46]. Bisulfite sequencing data derived from [12,46]. 9FWA is the Arabidopsis FWA gene [47]. Bisulfite sequencing data derived from [11,46]. 10NOS-PRO is a transgenic nopaline synthase promoter [32]. Bisulfite sequencing data derived from [10,29]. 11SIN3-11193 is a SINE3 element located between At3g22060 and At3g22070. 12229R1 and 229R2 are intergenic sequences within 2 kb of one another located between At1g36940 and At1g36950. Dashes indicate that it is not done; N/A, sequence absent in Ler; asterisks indicate that the numbers are for the transgenic NOS-PRO line, Col ecotype; question marks indicate that the size of the methylated sequence cannot be accurately predicted.
Table 4 Primers used for bisulfite sequencing
SIN3-11193
(A011193) G->A: 5'-CCTCCTTCGTTGACCTGTCTTCATCGCAATGACTCAGCATAG-3'
C->T: 5'- GTCTTCTAATCAAGTTTAGTTATGTTAATGTTTTTGGATAGAAC-3'
SIN3-5300
(A005300) G->A: 5'-TTCATTTGTTACCTACTATCATTTTCAAGAACGAAACAATG-3'
C->T: 5'-TAGTAGTTGTTCTCATCTTGTTTTTGGCAACTGGACGTGTC-3'
229R1
(A000229) G->A:5'- CACCATGTTCTAGCCCTTGTTCGGTCGTCGTTCCTTCCGTGG-3'
C->T: 5'- AAAAGAAAGGCGTCGTGGAATCACCACTAGCTACAACCGC-3'
229R2
(A000229) G->A:5'- TTAGAGCTTGTTTTCATTACCTTCTTCACACAACCTCCAAG-3'
C->T: 5'- TTTCAGGGTATCATGGTTCTCGACAAAGTAGGGTTATTATC-3'
TA11-4217
(A004217) G->A:5'- CAACATAAGATTGTAGCCTTCCATCCTTGACCACGCTTTG-3'
C->T: 5'- TCTTAAGATAGGAGATGATGTGTAGGAATGGTTTCTGGCAC-3'
MU-4802
(A004802) G->A:5'- AGCCATTATCATGTCCATCTGATCCTTCTACATGCCCTTG-3'
C->T: 5'- TATGTGAACGACTCATACACAAGAAATAGGTGGCGAGAAAC-3'
MU-PCR
(57802433) G->A:5'- CACCAGCTCGAACACCACCAACAGATTCCTTGTAAATCTG-3'
C->T: 5'- GATGGAGCGAGTGACGGGGATGAAGAGTCTAGTGTGTGCAC-3'
To amplify bisulfite-treated DNA, primers were synthesized with G→A (first sequence of the pair or C→T (second sequence), except for CGs and CNGs, which were synthesized with G→R or C→Y, respectively.
==== Refs
Chen T Li E Structure and function of eukaryotic DNA methyltransferases. Curr Top Dev Biol 2004 60 55 89 15094296
Yoder JA Walsh CP Bestor TH Cytosine methylation and the ecology of intragenomic parasites. Trends Genet 1997 13 335 340 9260521 10.1016/S0168-9525(97)01181-5
Martienssen R Colot V DNA methylation and epigenetic inheritance in plants and filamentous fungi. Science 2001 293 1070 1074 11498574 10.1126/science.293.5532.1070
Bird A DNA methylation patterns and epigenetic memory. Genes Dev 2002 16 6 21 11782440 10.1101/gad.947102
Tamaru H Selker EU A histone H3 methyltransferase controls DNA methylation in Neurospora crassa. Nature 2001 414 277 283 11713521 10.1038/35104508
Kato M Miura A Bender J Jacobsen SE Kakutani T Role of CG and non-CG methylation in immobilization of transposons in Arabidopsis. Curr Biol 2003 13 421 426 12620192 10.1016/S0960-9822(03)00106-4
Tompa R McCallum CM Delrow J Henikoff JG van Steensel B Henikoff S Genome-wide profiling of DNA methylation reveals transposon targets of CHROMOMETHYLASE3. Curr Biol 2002 12 65 68 11790305 10.1016/S0960-9822(01)00622-4
Tran R Henikoff J Zilberman D Ditt R Jacobsen S Henikoff S DNA methylation profiling identifies CG methylation clusters in Arabidopsis genes. Curr Biol 2005 15 154 159 15668172 10.1016/j.cub.2005.01.008
Okano M Bell DW Haber DA Li E DNA methyltransferases Dnmt3a and Dnmt3b are essential for de novo methylation and mammalian development. Cell 1999 99 247 257 10555141 10.1016/S0092-8674(00)81656-6
Cao X Aufsatz W Zilberman D Mette MF Huang MS Matzke M Jacobsen SE Role of the DRM and CMT3 methyltransferases in RNA-directed DNA methylation. Curr Biol 2003 13 2212 2217 14680640 10.1016/j.cub.2003.11.052
Chan SW Zilberman D Xie Z Johansen LK Carrington JC Jacobsen SE RNA silencing genes control de novo DNA methylation. Science 2004 303 1336 14988555 10.1126/science.1095989
Zilberman D Cao X Jacobsen SE ARGONAUTE4 control of locus-specific siRNA accumulation and DNA and histone methylation. Science 2003 299 716 719 12522258 10.1126/science.1079695
Gustafson AM Allen E Givan S Smith D Carrington JC Kasschau KD ASRP: the Arabidopsis Small RNA Project Database. Nucleic Acids Res 2005 33 Database issue D637 D640 15608278 10.1093/nar/gki127
Lu C Tej SS Luo S Haudenschild CD Meyers BC Green PJ Elucidation of the small RNA component of the transcriptome. Science 2005 309 1567 1569 16141074 10.1126/science.1114112
Li E Bestor TH Jaenisch R Targeted mutation of mouse DNA methyltransferase results in embryonic lethality. Cell 1992 69 915 926 1606615 10.1016/0092-8674(92)90611-F
Jackson JP Lindroth AM Cao X Jacobsen SE Control of CpNpG DNA methylation by the KRYPTONITE histone H3 methyltransferase. Nature 2002 416 556 560 11898023 10.1038/nature731
Malagnac F Bartee L Bender J An Arabidopsis SET domain protein required for maintenance but not establishment of DNA methylation. EMBO J 2002 21 6842 6852 12486005 10.1093/emboj/cdf687
Bender J DNA methylation and epigenetics. Annu Rev Plant Biol 2004 55 41 68 15725056 10.1146/annurev.arplant.55.031903.141641
Lippman Z Gendrel AV Black M Vaughn MW Dedhia N McCombie WR Lavine K Mittal V May B Kasschau KD Role of transposable elements in heterochromatin and epigenetic control. Nature 2004 430 471 476 15269773 10.1038/nature02651
Lippman Z May B Yordan C Singer T Martienssen R Distinct mechanisms determine transposon inheritance and methylation via small interfering RNA and histone modification. PLoS Biol 2003 1 E67 14691539 10.1371/journal.pbio.0000067
Baldi P Long AD A Bayesian framework for the analysis of microarray expression data: regularized t -test and statistical inferences of gene changes. Bioinformatics 2001 17 509 519 11395427 10.1093/bioinformatics/17.6.509
Tian L Fong MP Wang JJ Wei NE Jiang H Doerge RW Chen ZJ Reversible histone acetylation and deacetylation mediate genome-wide, promoter-dependent, and locus-specific changes in gene expression during plant development. Genetics 2005 169 337 345 15371352 10.1534/genetics.104.033142
Jurka J Repbase update: a database and an electronic journal of repetitive elements. Trends Genet 2000 16 418 420 10973072 10.1016/S0168-9525(00)02093-X
Bao Z Eddy SR Automated de novo identification of repeat sequence families in sequenced genomes. Genome Res 2002 12 1269 1276 12176934 10.1101/gr.88502
Methylation Profiler
Jordan B How consistent are expression chip platforms? Bioessays 2004 26 1236 1242 15499585 10.1002/bies.20128
Lindroth AM Cao X Jackson JP Zilberman D McCallum CM Henikoff S Jacobsen SE Requirement of CHROMOMETHYLASE3 for maintenance of CpXpG methylation. Science 2001 292 2077 2080 11349138 10.1126/science.1059745
Bartee L Malagnac F Bender J Arabidopsis cmt3 chromomethylase mutations block non-CG methylation and silencing of an endogenous gene. Genes Dev 2001 15 1753 1758 11459824 10.1101/gad.905701
Zilberman D Cao X Johansen LK Xie Z Carrington JC Jacobsen SE Role of Arabidopsis ARGONAUTE4 in RNA-directed DNA methylation triggered by inverted repeats. Curr Biol 2004 14 1214 1220 15242620 10.1016/j.cub.2004.06.055
Xie Z Johansen LK Gustafson AM Kasschau KD Lellis AD Zilberman D Jacobsen SE Carrington JC Genetic and functional diversification of small RNA pathways in plants. PLoS Biol 2004 2 E104 15024409 10.1371/journal.pbio.0020104
Springer NM Napoli CA Selinger DA Pandey R Cone KC Chandler VL Kaeppler HF Kaeppler SM Comparative analysis of SET domain proteins in maize and Arabidopsis reveals multiple duplications preceding the divergence of monocots and dicots. Plant Physiol 2003 132 907 925 12805620 10.1104/pp.102.013722
Aufsatz W Mette MF van der Winden J Matzke AJM Matzke M RNA-directed DNA methylation in Arabidopsis. Proc Natl Acad Sci USA 2002 99 16499 16506 12169664 10.1073/pnas.162371499
Hamilton A Voinnet O Chappell L Baulcombe D Two classes of short interfering RNA in RNA silencing. EMBO J 2002 21 4671 4679 12198169 10.1093/emboj/cdf464
Dorer DR Henikoff S Expansions of transgene repeats cause heterochromatin formation and gene silencing in Drosophila. Cell 1994 77 993 1002 8020105 10.1016/0092-8674(94)90439-1
Garrick D Fiering S Martin DI Whitelaw E Repeat-induced gene silencing in mammals. Nat Gen 1998 18 56 59 10.1038/ng0198-56
Michalowski SM Allen GC Hall GE JrThompson WF Spiker S Characterization of randomly-obtained matrix attachment regions (MARs) from higher plants. Biochemistry 1999 38 12795 12804 10504249 10.1021/bi991142c
Berg DE Howe MM Mobile DNA 1989 Washington: American Society of Microbiology
Lenoir A Lavie L Prieto JL Goubely C Cote JC Pelissier T Deragon JM The evolutionary origin and genomic organization of SINEs in Arabidopsis thaliana. Mol Biol Evol 2001 18 2315 2322 11719581
Zilberman D Henikoff S Silencing of transposons in plant genomes: kick them when they're down. Genome Biol 2004 5 249 15575975 10.1186/gb-2004-5-12-249
Saghai-Maroof MA Soliman KM Jorgensen RA Allard RW Ribosomal DNA spacer-length polymorphisms in barley: mendelian inheritance, chromosomal location, and population dynamics. Proc Natl Acad Sci USA 1984 81 8014 8018 6096873
The Arabidopsis Information Source
van Steensel B Delrow J Henikoff S Chromatin profiling using targeted DNA adenine methyltransferase. Nat Gen 2001 27 304 308 10.1038/85871
Genetic Information Research Institute
Gene Expression Omnibus
Sakai H Medrano LJ Meyerowitz EM Role of SUPERMAN in maintaining Arabidopsis floral whorl boundaries Nature 1995 378 199 203 7477325 10.1038/378199a0
Cao XF Jacobsen SE Locus-specific control of asymmetric and CpNpG methylation by the DRM and CMT3 methyltransferase genes. Proc Natl Acad Sci USA 2002 99 Suppl 4 16491 16498 12151602 10.1073/pnas.162371599
Soppe WJ Jacobsen SE Alonso-Blanco C Jackson JP Kakutani T Koornneef M Peeters AJ The late flowering phenotype of fwa mutants is caused by gain-of-function epigenetic alleles of a homeodomain gene. Mol Cell 2000 6 791 802 11090618 10.1016/S1097-2765(05)00090-0
|
16277745
|
PMC1297646
|
CC BY
|
2021-01-04 16:05:36
|
no
|
Genome Biol. 2005 Oct 19; 6(11):R90
|
utf-8
|
Genome Biol
| 2,005 |
10.1186/gb-2005-6-11-r90
|
oa_comm
|
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-11-r911627774610.1186/gb-2005-6-11-r91ResearchDo universal codon-usage patterns minimize the effects of mutation and translation error? Marquez Roberto [email protected] Sandra [email protected] Rob [email protected] Department of Computer Science, New Mexico State University, MSC CS, Las Cruces, NM 88003, USA2 Department of Chemistry and Biochemistry, University of Colorado, Boulder, CO 80309, USA2005 19 10 2005 6 11 R91 R91 7 7 2005 24 8 2005 21 9 2005 Copyright © 2005 Marquez 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 analysis of codon usage in nearly 900 species of the three domains of life suggests that codon usage patterns in mRNA messages do not minimize the effects of translation error.
Background
Do species use codons that reduce the impact of errors in translation or replication? The genetic code is arranged in a way that minimizes errors, defined as the sum of the differences in amino-acid properties caused by single-base changes from each codon to each other codon. However, the extent to which organisms optimize the genetic messages written in this code has been far less studied. We tested whether codon and amino-acid usages from 457 bacteria, 264 eukaryotes, and 33 archaea minimize errors compared to random usages, and whether changes in genome G+C content influence these error values.
Results
We tested the hypotheses that organisms choose their codon usage to minimize errors, and that the large observed variation in G+C content in coding sequences, but the low variation in G+U or G+A content, is due to differences in the effects of variation along these axes on the error value. Surprisingly, the biological distribution of error values has far lower variance than randomized error values, but error values of actual codon and amino-acid usages are actually greater than would be expected by chance.
Conclusion
These unexpected findings suggest that selection against translation error has not produced codon or amino-acid usages that minimize the effects of errors, and that even messages with very different nucleotide compositions somehow maintain a relatively constant error value. They raise the question: why do all known organisms use highly error-minimizing genetic codes, but fail to minimize the errors in the mRNA messages they encode?
==== Body
Background
Genetic codes are arranged in a way that is highly resistant to errors, but whether the mRNAs that genomes encode also resist errors has been largely untested. The standard genetic code is found in most nuclear and mitochondrial genomes, although some genomes have slight variations in the genetic code (see [1] for review). The biochemical basis for many of these variations is known, but their purpose remains unclear. The extent to which a genetic code is resistant to errors (in replication, transcription, or translation) can be defined by an 'error value' [2,3], which is the sum of the differences in amino-acid properties when changing from each codon to each other codon that can be reached by a single-base substitution (see Materials and methods). The standard genetic code and all known variants resist error better (have a lower error value) than do random codes for a wide range of different amino-acid properties and models of random code generation [4-9], although the extent to which natural selection has reached the best of all codes remains somewhat controversial [10-13]. We now test the idea that organisms optimize their codon usage as well as their genetic code: codons with low error values might be used in preference to those with high error values, to reduce the overall probability of error.
Different organisms use the four bases in varying amounts at each of the three positions within the codon (that is, the average counts of each of the four bases in all the first positions of all the codons in a genome are different from the counts in all the second positions and the third positions) [1]. In particular, the first position is heavily biased towards purines, and the second position is somewhat biased towards A and C. These trends hold for all organisms in all three domains of life. In addition, organisms vary extensively in GC content (the fraction of bases that are G or C, as opposed to A or T) at each of the three codon positions, which also affects the amino-acid usage [1,14-16]. These features might be related to the code's error-minimizing properties: organisms might choose their codon and/or amino-acid usages in ways that reduce errors during translation [17-20].
Previous research has suggested that the GC content of a sequence can greatly affect its error-minimizing properties [20], and that amino-acid and/or codon usage may be optimized in Drosophila and mouse [19] but not in Escherichia coli [18], but no global survey has yet been performed. If mRNA messages are arranged in ways that minimize error, as has been comprehensively established for the genetic code itself (see for example [2,3,7]), this error minimization might arise by adjusting the usage of individual codons or amino acids, or by adjusting the overall base frequencies at each of the three codon positions. In particular, the error values might be especially stable against change in GC content, since organisms have mRNAs that vary over a wide range of GC content but vary little over the other two orthogonal axes of nucleotide composition. However, it is also possible that the genetic code was shaped under different selection pressures than those acting in modern organisms, resulting in codon-usage patterns that are random with respect to error minimization.
Codon and amino-acid usage statistics are now available for thousands of species from the Codon Usage Tabulated from GenBank (CUTG) database [21]. We tested whether species preferentially use codons with low error values; that is, codons that, if misread, would tend to substitute a more similar amino acid. To do this, we compared the error value of the code weighted by the actual codon usages against the error values of codes in which the codon or amino-acid usages had been randomized. Thus, we tested three specific hypotheses: first, that organisms choose codon usages that produce fewer errors than permuted or randomly chosen codon usages; second, that organisms choose amino-acid usages that produce fewer errors than permuted or randomly chosen amino-acid usages; and third, that the discrepancy in composition in the three nucleotide positions is caused by selection of codons that minimize errors in translation.
Results and discussion
Messages are not optimized
We used two different methods to compare the actual codon usages to randomized codon usages. First, we used 'shuffled' codon usages. In shuffled codon usages, the codons, amino acids, or positional-base frequencies were randomly permuted. This method preserves the relative frequencies of the the different codons, amino acids, or positional-base frequencies, but changes their meanings. For example, if the original amino-acid usage was 5%A, 10%G, and 2%W, the usage after shuffling might be 5%A, 2%G, and 10%W. Second, we used random codon usages that did not preserve the relative frequencies of codons, amino acids, or positional-base frequencies, but instead assigned each codon, amino acid, or positional-base frequency a random number from a uniform distribution, followed by normalization so that the frequencies summed to one (see Materials and methods). We analyzed species in the three domains of life separately: 33 archaea, 457 bacteria, and 264 eukaryotes for which at least 50 genes were available.
From the distributions of code-error values for real and randomized codon usages (Figure 1 first column, and Table 1), we make three observations. First, the actual distribution of error values in organisms was much tighter than in any of the randomized usages (63.8 ≤ mean ≤ 67.7 and standard deviation ≤ 3.42 for all domains). Second, both the permuted and random codon usages produced code-error values significantly lower than the corresponding values for actual codon usages (P ≤ 0.05 by two-tailed paired t-test between actual and shuffled or random codon usages). Finally, the shuffled and random codon usages produced almost identical results (P > 0.05 in all cases by two-tailed paired t-test).
The variance of the actual codon usages is significantly smaller than the shuffled and random usages under each randomization model and for all domains of life. The P-value ranges are as follows: for archaea from 7.7 × 10-9 to 0.59 (where 0.59 is the only non-significant value), for bacteria from 3.9 × 10-257 to 1.1. × 10-43, and for eukaryotes from 8.5 × 10-131 to 5.5 × 10-10. The significance of the difference in variance between a shuffled and random usage varies considerably (no consistent trend in P-values), probably depending on each specific random sample.
The pattern was similar for shuffled and random amino-acid usages, and for shuffled and random positional-base usages. In all cases, the means for the shuffled and random distributions were similar to each other and lower than the mean for the actual distribution (Figure 1, columns 2 and 3). The similarities across domains are striking: the error values for codon usages in all three domains of life fall in the same narrow region.
Code error is not correlated with composition
To test whether the error value varied systematically with nucleotide composition, we plotted the error value as a function of position in the tetrahedron of possible base compositions (see Materials and methods for discussion). If the error value of a message depended on the composition of the codons, we would expect to see no correlation along the GC axis, because the amount of natural variation along this axis suggests that all values are selectively neutral and that therefore the code error is approximately the same. In contrast, we would expect to see increasing error values with increasing distance from the GC axis, constraining the biological variation in these other directions. However, contrary to these predictions, we find that for the real, permuted, and random positional-base usages, there are clear differences both in composition and in error at the three positions, but there is no systematic variation of error with composition.
Figure 2 shows the composition of each of the three codon positions and of the total in composition space, where the volume of a sphere is proportional to its error value. As expected, we observe clear differences in composition between the three codon positions. We can also see that the different codon positions contribute very differently to the total error value of the message. The second codon position determines about 70% of the total error value, the first codon position another 29%, and the third codon position less than 1%.
To highlight possible changes in code-error value along the three compositional axes, which are difficult to see in the simplex, we plotted code-error value versus composition along each of the three axes separately. Figure 3 shows the code-error values for the actual codon usages of bacteria along the UC, UG, and UA axes. In the left column, the error values have been scaled relative to the maximum value for each codon position independently to demonstrate relative changes, while in the right column the absolute values are displayed. Results for archaea and eukaryotes are very similar to those for bacteria (data not shown).
We applied the same analysis to permuted and random positional-base usages, which allowed us to examine the correlations along a wider compositional range on all of the axes. These codon usages form spherical distributions around the center of the tetrahedron (Figure 4). For permuted usages, the original compositional values are redistributed over the three axes; the random usages show equal distributions for each of the three codon positions with equal variation along each axis. Figure 5 shows the corresponding scatterplots for the permuted and random usages.
We found highly significant correlations between (total) code error and position on each of the three orthogonal composition axes, except for the eukaryotes along the UG axis (Table 2). For total code error, the significant P-values averaged 0.0042 (range 1 × 10-6 to 0.03), explaining an average of 0.19 (range 0.020 to 0.37) of the variance in code error. However, the correlation along the GC axis was not, in general, less than the correlation along the other axes. In addition, we found no significant correlations along the UG and UA axes for random and permuted data sets (in a single case the correlation was significant, but only explained 0.023 of the variation). Along the UC axis, the correlations in random and shuffled bacterial and eukaryotic usages are of similar magnitude to the correlations in the natural usages. Together with the observation that actual usage errors are typically higher than random usage errors, these observations suggest that selection against errors caused by variation along the different composition axes cannot explain observed trends in codon usage.
Conclusion
If organisms were under strong selection to minimize errors in replication and translation, we would expect them to choose codons that are less prone to error. Consequently, we would expect that the actual codon, amino-acid, and positional-base usages would have lower error values than would permuted versions. However, we found exactly the opposite: the actual codon, amino-acid, and positional-base usages produce more errors than randomly chosen compositions.
Consequently, our hypothesis that genetic messages (as well as genetic codes) are optimized for error minimization was not supported by the data. However, the low variance in codon-usage error values in organisms suggests the intriguing alternative possibility that mRNAs are selected for a specific level of errors, rather than to minimize errors overall. Because the rate of evolution is limited by mutation, it is possible that the ability to tune the rate of protein sequence evolution by using error-prone codons has provided a selective advantage to modern organisms. Intriguingly, recent research suggests that the canonical genetic code allows target protein sequences to evolve far more rapidly than do the alternative genetic codes [22]. Codon usage may also be tuned for evolvability rather than for error minimization.
Another possible explanation for the limited variability in error-minimization properties is that the genetic code was shaped under very different selection pressures than those acting in modern organisms. Today, other factors, including directional mutation or selection for translation speed, may greatly outweigh the benefits that could be obtained by using error-minimizing codons or amino acids. However, such an explanation would predict that modern usages would be random with respect to code error, and would not predict the near constancy of error values in actual organisms. This work is consistent with the previous observations that messages within E. coli are not optimized for error minimization at the codon level [18] and that codon usage can greatly influence error minimization [20], and extends the analysis to a sample of over 700 bacterial, archaeal, and eukaryotic species. However, it does not confirm the observation that the amino-acid usage in some species is chosen in a way that minimizes errors [17,19]. This latter discrepancy could be due to the different sampling of genes or the different methods used to calculate the error value (single-step versus multi-step mutations).
As previously observed, we confirm that the three nucleotide positions differ greatly in nucleotide composition [1] and in error minimization [3]. However, we find no evidence for a relationship between these two properties. The universal maintenance of these patterns across species suggests that some kind of selection is involved, but the factors influencing this selection remain undefined. In particular, positional base-composition patterns orthogonal to the actual base-composition patterns, and occupying regions of composition space in which no organism has ever been observed, have errors no worse than do the actual usage patterns. This similarity strongly suggests that selection for error minimization does not play a role in keeping genomes within a narrow region of composition space. The nucleotide composition of a message has relatively little effect on its error value, suggesting that other factors maintain the systematic biases in composition at the three codon positions that are observed in all species and domains of life.
Thus, organisms do not choose their codon, amino-acid, or nucleotide composition in a way that minimizes the effects of errors. This observation is highly unexpected in light of the great extent to which the genetic code itself is arranged in an error-minimizing fashion, and suggests that some factor underlying the near-constant error values of codon usage across genomes in all three domains of life remains to be discovered.
Materials and methods
We addressed our first and second hypotheses, that genetic messages are optimized for error minimization either at the codon or amino-acid level, by comparing the actual codon usages from organisms to first, permuted codon usages, in which the codon counts were preserved but the codons to which those counts applied were randomized, and second, to completely random codon usages. We addressed our third hypothesis, that the code error is robust to variation in GC content but not robust to other compositional variation, by examining the correlation between composition along each of the three compositional axes (GC, GU, and GA) and the code-error values for real, permuted, and random codon usages.
Data source
We used the CUTG database as source for codon usages found in organisms [21]. We repeated the analysis separately for the three domains of life (archaea, bacteria, and eukaryotes). The species were classified according to the NCBI Taxonomy. We analyzed the 754 species for which at least 50 genes were available: 33 archaea, 457 bacteria, and 264 eukaryotes. Mitochondrial sequences were excluded.
Calculating the error value of a message
The process of calculating an error value for a message (or codon usage) uses the basic method for calculating an error value for a genetic code [2,3], with the addition that the error value of a change from one codon to another is weighted by the frequency of the starting codon [18]. To maintain consistency with previous work [2,3], we measured the distance between amino acids using polar requirement, a measure of hydrophobicity [23].
The error value of a code is given by:
For all possible mutations b at each of the three codon positions p in all 64 codons c, we sum the weighted size of the change in amino-acid property, for example, hydrophobicity. The change is given by the difference in the amino-acid property of the amino acids encoded by the old and new codons, νold - νnew, weighted by the abundance of the codon wc, the effect of the base position wp, and the probability of mutation to the new base given the codon and position wb|(c,p). A 'mutation' from a codon to itself does not add to the error value, because the same amino acid is present before and after the 'mutation'. Stop codons are excluded from the calculation. Codon frequencies were taken from the codon usage database or assigned at random. We used a range of transition/transversion biases from 1:1 to 10:1, although there was no qualitative effect on the results. Results shown are for a transition/transversion bias of 4:3, and equal weighting for the three base positions.
Creating permuted and random codon usages
We can calculate the amino-acid usage and positional-base usage from a given codon usage. The frequency of an amino acid is the sum of the frequencies of each of its codons. A positional-base usage is the frequency of each of the four bases at each of the three codon positions. For example, the frequency of U at the first codon position is the sum of the frequencies of all codons that start with a U. Thus, each codon usage is associated with one unique amino-acid usage and one positional-base usage.
However, many different codon usages correspond to the same amino-acid usage. To predict the codon usage associated with an amino-acid usage, we used the assumption that all codons coding for the same amino acid occur with equal frequencies, so that each gets an equal share of the amino-acid frequency. Consequently, blocks of codons (coding for the same amino acid) are assigned the same frequency. The prediction of the frequency of a codon from a positional-base usage is calculated as the product of the positional-base frequencies of its bases at the three codon positions. This method reflects the idea that if a species were under selection for amino-acid usage only, there would be no a priori reason to assign different frequencies to the different codons for a given amino acid. Similarly, to predict the codon usage associated with a particular positional-base usage, we take the product of the frequency of the appropriate base at each of the three codon positions. For example, the frequency of the codon AUG is the product of the frequency of A at the first position, U at the second position, and G at the third position.
With the above transformations in mind, we can shuffle frequencies or choose random frequencies at three levels: codons, amino acids, and positional bases. After creating a permuted or random amino-acid usage or positional-base usage, we calculate the corresponding codon usage as described above (because the error value calculations require codon usages as input).
Statistics
We used the two-tailed paired t-test to compare the means of the various distributions, because we examined the same sample before and after randomization. Differences in variance between the error values of the actual usages and the permuted and random usages were calculated by a two-tailed F-test.
Visualization
The (positional) composition of the codon usages can be conveniently visualized with the program MAGE [24], using a presentation scheme in which the volume of a sphere is proportional to the error value at a particular codon position. The base frequency of a set of bases, such as a sequence of nucleotides or all bases at a particular codon position, can be visualized as a point in composition space. The base frequency is described as a vector of the fraction of each of the four bases (U, C, A, and G) in the set. These fractions form the four coordinates to describe sequence composition. When visualizing the space of all possible compositions, we only have three dimensions to work with. Three unique ways divide the four bases into sets of two, which provide an orthogonal coordinate system. The three axes are the lines where G+C equals A+U, G+U equals A+C, and G+A equals U+C. The GC (or AU) axis is also called Chargaff's axis, because it is the line where all perfectly Watson-Crick base-paired regions would reside. Composition space can thus be visualized as a tetrahedral unit simplex [25].
Additional data files
The Python code and the raw data to perform the described code-error analysis are available as an Additional data file with the online version of this paper. Additional data file 1 is a tar archive containing the used CUTG records, separated for archaea, bacteria, and eukaryotes, the data used to produce the histograms in Figure 1, the kinemages used to produce Figures 2 and 4, and the data used to produce the scatterplots in Figures 3 and 5.
Supplementary Material
Additional data file 1
A tar archive containing the used CUTG records, separated for archaea, bacteria, and eukaryotes, the data used to produce the histograms in Figure 1, the kinemages used to produce Figures 2 and 4, and the data used to produce the scatterplots in Figures 3 and 5.
Click here for file
Acknowledgements
R.M. and S.S. contributed equally to this work, and should be considered joint first authors. We thank Michael Yarus, Noboru Sueoka, and members of the Knight and Yarus labs for critical discussion of the manuscript. R.M. was supported by a SMART scholarship.
Figures and Tables
Figure 1 Code-error values for actual and permuted codon usages. The usages are displayed for three randomization algorithms and each domain of life. Rows: archaea, bacteria, and eukaryotes. Columns (randomization algorithms): codon, amino acid, positional base. Black, biological (unpermuted); red, permuted; green, random. Variability is always much less in the biological codon usages (black lines) than in any of the random or randomized usages, and the mean is always higher, suggesting that the biological codon usages are constrained to a narrow band but are not optimized for error minimization.
Figure 2 Relationship between base composition and code error. Bacterial codon usages are chosen to illustrate this relationship by plotting the base composition and code-error value for each codon position in the tetrahedral simplex (composition space). The error value for each species is plotted as a sphere with volume proportional to the error. Two perspectives are given. On the left is an oblique view to show variation along Chargaff's axis (G = C and A = T) and the relative contribution of each codon position to the error value. On the right is a view down Chargaff's axis to show the bias of each codon position. First position, yellow; second position, red; third position, blue; and total, green. As expected, the error value is always lowest at the third position (blue) as result of interconversion among synonymous codons and codons for similar amino acids.
Figure 3 Variation in code error along the three axes in composition space: G+A, G+C, G+U. Scatterplots of variation in code-error value along each of the three axes that make up the composition space. Top row, UC content; middle row, UG content; bottom row, UA content. Left column, error value at each codon position individually scaled relative to the maximum value for that position (maximum = 1.0). Right column, absolute error values for each codon position. First position, yellow; second position, red; third position, blue; and total, green. Data shown are for bacteria, though results were similar for the other two domains (data not shown). Although substantial correlations are revealed in the scaled data, these correlations contribute little to the overall error value, which is dominated by the second codon position.
Figure 4 Base composition by codon position for randomized base usages. Left: permuted by positional bases, where the variability at each position is preserved, but the direction of the variability is rotated by 90 degrees around an arbitrary axis. Right: randomly chosen positional bases, where the amount of variability and the size of the correlations between axes at each position are destroyed. First position, yellow; second position, red; third position, blue; and total, green. Compare this figure with biological codon usages in Figure 2.
Figure 5 Absolute error values for permuted bacterial codon usages. The variation in code-error values is shown along the three compositional axes. Compare this figure with biological codon usages in Figure 3. Top row, UC content; middle row, UG content; bottom row, UA content. Left column, permuted positional-base usages. Right column, random positional-base usages. First position, yellow; second position, red; third position, blue; and total, green. Lack of correlation along any axis and wide range suggests that constraints on positional-base usage do not explain the pattern of codon usage error values in organisms.
Table 1 Error values for biological and random codon usages
Archaea Bacteria Eukaryotes
Natural codon usages 67.7 ± 3.42 64.7 ± 1.77 63.8 ± 2.14
Codon permuted 52.4 ± 4.92 52.2 ± 5.15 52.7 ± 3.61
Codon random 52.6 ± 3.76 52.6 ± 3.47 52.4 ± 3.16
Amino acid permuted 61.6 ± 8.74 61.0 ± 6.95 61.1 ± 6.35
Amino acid random 61.0 ± 7.37 61.8 ± 6.96 61.7 ± 6.72
Positional base permuted 51.7 ± 6.49 52.3 ± 6.91 52.2 ± 5.44
Positional base random 52.1 ± 10.5 53.4 ± 12.6 52.1 ± 12.9
Mean ± standard deviation for each set of codon usages. The natural codon usages invariably have higher error values and lower standard deviations than any of the random or randomized codon usages: this pattern is consistent for all three domains of life.
Table 2 Correlations between composition and code-error value
UC (or AG) UG (or AC) UA (or GC)
Bacteria Natural 0.23 (1 × 10-6) 0.14 (1 × 10-6) 0.023 (0.0012)
Permuted 0.017 (0.0055) 0.0020 (0.35) 0.023 (0.0011)
Random 0.23 (1 × 10-6) 0.0026 (0.28) 0.00064 (0.59)
Eukaryotes Natural 0.21 (1 × 10-6) 0.0021 (0.46) 0.12 (1 × 10-6)
Permuted 0.14 (1 × 10-6) 0.00012 (0.86) 0.0033 (0.35)
Random 0.20 (1 × 10-6) 0.0014 (0.55) 0.0069 (0.18)
Archaea Natural 0.14 (0.029) 0.28 (0.0016) 0.37 (0.00017)
Permuted 0.073 (0.13) 0.016 (0.49) 0.029 (0.34)
Random 0.10 (0.071) 0.00056 (0.90) 0.025 (0.38)
Coefficient of determination (r2) and P-value for natural and representative randomized usages. Because of the much smaller sample size in archaea, the significance of the correlations is generally much lower than in the other two domains (n = 33 for archaea, 264 for eukaryotes, and 457 for bacteria).
==== Refs
Knight RD Freeland SJ Landweber LF A simple model based on mutation and selection explains trends in codon and amino-acid usage and GC composition within and across genomes Genome Biol 2001 2 research0010.1 0010.13 11305938 10.1186/gb-2001-2-4-research0010
Haig D Hurst LD A quantitative measure of error minimization in the genetic code J Mol Evol 1991 33 412 417 1960738 10.1007/BF02103132
Freeland SJ Hurst LD The genetic code is one in a million J Mol Evol 1998 47 238 248 9732450
Woese CR On the evolution of the genetic code Proc Natl Acad Sci USA 1965 54 1546 1552 5218910
Alff-Steinberger C The genetic code and error transmission Proc Natl Acad Sci USA 1969 64 584 591 5261915
Ardell DH On error minimization in a sequential origin of the standard genetic code J Mol Evol 1998 47 1 13 9664691
Freeland SJ Knight RD Landweber LF Hurst LD Early fixation of an optimal genetic code Mol Biol Evol 2000 17 511 518 10742043
Ardell DH Sella G No accident: genetic codes freeze in error-correcting patterns of the standard genetic code Philos Trans R Soc Lond B Biol Sci 2002 357 1625 1642 12495519 10.1098/rstb.2002.1071
Freeland SJ Wu T Keulmann N The case for an error minimizing standard genetic code Orig Life Evol Biosph 2003 33 457 477 14604186 10.1023/A:1025771327614
Crick FH The origin of the genetic code J Mol Biol 1968 38 367 379 4887876 10.1016/0022-2836(68)90392-6
Di Giulio M The extension reached by the minimization of the polarity distances during the evolution of the genetic code J Mol Evol 1989 29 288 293 2514270 10.1007/BF02103616
Di Giulio M Genetic code origin and the strength of natural selection J Theor Biol 2000 205 659 661 10931760 10.1006/jtbi.2000.2115
Judson OP Haydon D The genetic code: what is it good for? An analysis of the effects of selection pressures on genetic codes J Mol Evol 1999 49 539 550 10552035
Sueoka N Compositional correlation between deoxyribonucleic acid and protein Cold Spring Harb Symp Quant Biol 1961 26 35 43 13918160
Lobry JR Influence of genomic G+C content on average amino-acid composition of proteins from 59 bacterial species Gene 1997 205 309 316 9461405 10.1016/S0378-1119(97)00403-4
Foster PG Jermiin LS Hickey DA Nucleotide composition bias affects amino acid content in proteins coded by animal mitochondria J Mol Evol 1997 44 282 288 9060394
Gilis D Massar S Cerf NJ Rooman M Optimality of the genetic code with respect to protein stability and amino-acid frequencies Genome Biol 2001 2 research0049.1 0049.12 11737948 10.1186/gb-2001-2-11-research0049
Zhu CT Zeng XB Huang WD Codon usage decreases the error minimization within the genetic code J Mol Evol 2003 57 533 537 14738311 10.1007/s00239-003-2505-7
Archetti M Selection on codon usage for error minimization at the protein level J Mol Evol 2004 59 400 415 15553093 10.1007/s00239-004-2634-7
Archetti M Codon usage bias and mutation constraints reduce the level of error minimization of the genetic code J Mol Evol 2004 59 258 266 15486699 10.1007/s00239-004-2620-0
Nakamura Y Gojobori T Ikemura T Codon usage tabulated from the international DNA sequence databases: status for the year 2000 Nucleic Acid Res 2000 28 292 10592250 10.1093/nar/28.1.292
Zhu W Freeland S The standard genetic code enhances adaptive evolution of proteins J Theor Biol
Woese CR Dugre DH Dugre SA Kondo M Saxinger WC On the fundamental nature and evolution of the genetic code Cold Spring Harb Symp Quant Biol 1966 31 723 736 5237212
Richardson DC Richardson JS The kinemage: a tool for scientific communication Protein Sci 1992 1 3 9 1304880
Schultes E Hraber PT LaBean TH Global similarities in nucleotide base composition among disparate functional classes of single-stranded RNA imply adaptive evolutionary convergence RNA 1997 3 792 806 9214661
|
16277746
|
PMC1297647
|
CC BY
|
2021-01-04 16:05:36
|
no
|
Genome Biol. 2005 Oct 19; 6(11):R91
|
utf-8
|
Genome Biol
| 2,005 |
10.1186/gb-2005-6-11-r91
|
oa_comm
|
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-11-r921627774710.1186/gb-2005-6-11-r92ResearchThe interferon-inducible p47 (IRG) GTPases in vertebrates: loss of the cell autonomous resistance mechanism in the human lineage Bekpen Cemalettin [email protected] Julia P [email protected] Christoph [email protected] Iana [email protected] Libby [email protected] Diane M [email protected] Eva [email protected] Maria [email protected] Jonathan C [email protected] Institute for Genetics, University of Cologne, Zülpicher Strasse 47, 50674 Cologne, Germany2 Eccles Institute of Human Genetics, University of Utah, Salt Lake City, UT 84112-5330, USA3 Informatics & Systems Groups, Sanger Centre, The Wellcome Trust Genome Campus, Hinxton, Cambridge, CB10 1SA UK4 Department of Structural Biology, Stanford University Medical School, Stanford, CA 94305, USA5 Institute for Microbiology and Immunology, University of Cologne Medical School, 50935 Cologne, Germany2005 31 10 2005 6 11 R92 R92 4 6 2005 7 9 2005 7 10 2005 Copyright © 2005 Bekpen 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 survey of p47 GTPases in several vertebrate organisms shows that humans lack a p47 GTPase-based resistance system, suggesting that mice and humans deploy their immune resources against vacuolar pathogens in radically different ways.
Background
Members of the p47 (immunity-related GTPases (IRG) family) GTPases are essential, interferon-inducible resistance factors in mice that are active against a broad spectrum of important intracellular pathogens. Surprisingly, there are no reports of p47 function in humans.
Results
Here we show that the p47 GTPases are represented by 23 genes in the mouse, whereas humans have only a single full-length p47 GTPase and an expressed, truncated presumed pseudo-gene. The human full-length gene is orthologous to an isolated mouse p47 GTPase that carries no interferon-inducible elements in the promoter of either species and is expressed constitutively in the mature testis of both species. Thus, there is no evidence for a p47 GTPase-based resistance system in humans. Dogs have several interferon-inducible p47s, and so the primate lineage that led to humans appears to have lost an ancient function. Multiple p47 GTPases are also present in the zebrafish, but there is only a tandem p47 gene pair in pufferfish.
Conclusion
Mice and humans must deploy their immune resources against vacuolar pathogens in radically different ways. This carries significant implications for the use of the mouse as a model of human infectious disease. The absence of the p47 resistance system in humans suggests that possession of this resistance system carries significant costs that, in the primate lineage that led to humans, are not outweighed by the benefits. The origin of the vertebrate p47 system is obscure.
==== Body
Background
It is generally assumed that the immune system of the mouse is a good experimental model for that in humans. However, several studies suggest that immune mechanisms have been evolving rather differently in the human and mouse lineages (for review, see Mestas and Hughes [1]). The p47 (immunity-related GTPases (IRG) family; see Nomenclature, below) GTPases present a uniquely striking example of this divergence.
In mice the interferon-γ-inducible p47 GTPases constitute one of the most powerful resistance systems against several important intracellular pathogens [2-4]. The proteins localize on intracellular membrane systems in interferon-induced cells, some (IGTP, IIGP1) favoring the endoplasmic reticulum [5,6] and others (LRG-47, GTPI) the Golgi membranes [6,7] (for names of individual IRG GTPases see Additional data file 1). Infection or phagocytosis, however, initiates redistribution of the p47 GTPases to the phagocytic vacuole [6-8]. The p47 GTPases probably act specifically against vacuolar pathogens. Thus, Gram-positive and Gram-negative bacteria, mycobacteria, and protozoal pathogens are all resisted by the p47 GTPases, whereas no viral target has yet been confirmed.
The p47 GTPase IIGP1 is a low-affinity nucleotide binding protein with a slow GTP turnover [9]. At high protein concentrations and in the presence of GTP, IIGP1 oligomerizes and increases GTP turnover by up to 20-fold. These properties are distinct from those of the classical signaling GTPases and are reminiscent of the dynamins and p65 (GBP-1) GTPases [10,11]. The crystal structure of IIGP1 exhibits a H-Ras-1-like nucleotide-binding domain flanked by amino-terminal and carboxyl-terminal helical domains that are unknown in other GTPases [12]. This basic structure is probably common to the whole family. However, the divergent sequences of published p47 GTPases [13] and the patterns of susceptibility in knockout strains (for reviews, see Taylor [2] and MacMicking [3,4]) show that the proteins are highly diversified. Thus, a subgroup of three proteins (the GMS GTPases) have a radical substitution (the substitution of Methinine (M) for Lysine (K)) in the conserved P-loop G1 motif of the nucleotide binding site (Walker A motif) and correlated sequence variation elsewhere in the G-domain [13], implying a distinct catalytic mechanism for GTP hydrolysis. In the case of IIGP1 and LRG-47, the cell biology of the two proteins is distinct; IIGP1 associates with the endoplasmic reticulum membrane primarily through an amino-terminal myristoylation sequence, whereas LRG-47 associates with Golgi membrane via an amphipathic helix in the subterminal domain [6]. We recently showed that IIGP1 participates in a novel effector mechanism in Toxoplasma gondii infected astrocytes involving vesiculation and ultimately destruction of the parasitophorous vacuole membrane [8]. In contrast, there is evidence that LRG-47 is involved in accelerated acidification of the phagocytic vacuole containing Mycobacterium tuberculosis [8].
The p47 GTPases are thus a functionally diverse resistance system with many signs of adaptive divergent evolution. Surprisingly, there are no reports of p47 GTPase function in humans. To address this imbalance, we analyzed the p47 GTPase gene family in depth. We conclude that although the mouse has 23 p47 GTPases, of which up to 20 may be functional in resistance, the resistance system is entirely absent from humans. This finding carries important implications for our understanding of human and mouse immunity to vacuolar pathogens.
Results
Genomic organization of the p47 GTPase (Irg) genes of the C57BL/6 mouse
There are 23 p47 GTPase (Irg) genes in the C57BL/6 mouse, including the six previously known members of the family [13], localized on chromosomes 7, 11 and 18 (Figure 1a,b; also see Figure 7a). (For the nomenclature of the Irg genes, see Nomenclature (below) and Additional data file 1). Two of the mouse Irg sequences, namely Irga5 and Irgb7, are clearly pseudo-genes (see legend to Figure 1b). The remaining 21 Irg genes are intact across the GTP-binding domain, although Irga1, Irga8, and Irgb10 are carboxyl-terminally truncated relative to the majority, and no transcripts of Irga7 and Irgb8 have yet been found. Thus, the number of potentially functional Irg genes is not six but rather 21 in the C57/BL6 mouse. The nucleotide and protein sequences of these genes can be found on our home page [14].
The complex block of 13 genes on chromosome 11 contains the most divergent sequences (Figure 1c,d; Additional data file 2), including all three GMS (Irgm) GTPases [13], suggesting that this cluster is relatively ancient. In contrast, the eight Irga genes clustered on chromosome 18 are also clustered phylogenetically, suggesting more recent divergence, probably from a translocated member of the Irgb (TGTP) cluster on chromosome 11. The isolated Irg gene on chromosome 7, Irgc, is an ancient root with no obvious systematic relationship to the other subfamilies. Within the chromosomal clusters, more recent duplication events are apparent. The sibling pair Irgb3 and Irgb4 differ by only nine nucleotides in the open reading frame. The genes Irgb1, Irgb3, Irgb4, and Irgb8 appear to have been duplicated in tandem with Irgb2, Irgb5, and Irgb9, respectively. The pattern of divergence in the mouse p47 tree suggests an old gene family that has undergone a succession of duplication-divergence cycles over time - a pattern of evolution that is still actively continuing in several of the subfamilies.
The structure of p47 GTPase genes and their splicing patterns
The open reading frame of Irg genes is typically encoded on a single long 3' exon (Figure 2a) behind one or more 5'-untranslated exons. However, in one splice form of Irgm1 and one splice form of Irgm2 the initial methionine is encoded at the 3' end of the penultimate exon (also see the legend to Figure 2). The closely related Irgb1 and Irgb4 genes are exceptional in apparently occurring only as tandem transcripts in-frame with their respective closely linked upstream genes Irgb2 and Irgb5. If translated, such transcripts would generate 94 kDa polypeptides containing two distinct full-length p47 GTPase units. For the sequence phylogenies and alignments (Figure 1c,d; also see Figure 4, below), we provisionally treat these separate p47 units as independent genes. It remains to be seen whether the third tandem gene pair, Irgb9 and Irgb8, is also expressed as a tandem transcript. That Irgb1, Irgb3, and possibly Irgb8 are normally expressed in tandem with an upstream gene is also consistent with the absence both of autonomous transcripts of these exons and of interferon-inducible promoter elements (see below).
Identification of interferon-stimulatable elements in putative promoters of Irg genes
The basis for interferon-inducible expression of the mouse p47 GTPases has previously been investigated only for Irgd (IRG47) [15], in which an active interferon-stimulated response element (ISRE) was found upstream of the putative transcriptional start point. A GAS (γ-activated sequence) site was predicted in the putative promoter region of Irgm1 (LRG-47) [8]. Most of the transcribed p47 genes on chromosomes 11 and 18 exhibit multiple perfect interferon-inducible genomic motifs, both ISRE and GAS elements (Figure 2b; Additional data file 3). The sequences and relative positions of the GAS and ISRE elements vary, both classes of site are not present in all promoters, and the orientations of the two components are also variable. Thus, the association of interferon-inducible elements with Irg genes is presumably ancient and has been retained against the disruptive forces of spontaneous genome evolution. No further immunity-related inducible elements such as NFκB sites were found to be associated with the ISRE/GAS motifs. Irgd and Irga6 are both transcribed from alternative 5'-untranslated exons, each furnished with an independent promoter. In both genes the initial methionine is encoded at the beginning of the long 3' exon, so that the two transcripts of each gene generate identical proteins. Both putative promoters of Irgd and Irga6 have interferon-inducible elements. As noted above, genes Irgb1, Irgb4, and Irgb8 are probably expressed only as the 3' ends of tandem transcripts with Irgb2, Irgb5, and Irgb9, respectively. No dedicated 5'-untranslated exons could be identified for these downstream domains. Using RT-PCR we were able to show clear induction of eight further genes (Irga2, Irga3, Irga4, Irga8, Irgb1, Irgb2, Irgb5, and Irgb10) in addition to the six (Irga6 (IIGP), Irgb6 (TGTP), Irgd (IRG-47), Irgm1 (LRG-47), Irgm2 (GTPI) and Irgm3 (IGTP)) assayed by Boehm and coworkers [13] in L929 fibroblasts stimulated with interferon-γ in vitro (Figure 3a).
The isolated p47 gene, Irgc, on chromosome 7 is a clear exception. No clustered or isolated ISRE or GAS elements could be identified up to 10 kilobases (kb) 5' of the putative transcription start of this transcribed gene, and Irgc was not induced in interferon-stimulated fibroblasts (Figure 3b, panel i left). A weak Sox-related element was detected in the proximal promoter region. In view of the close homology of Irgc to the interferon-inducible Irg genes, we considered whether Irgc is induced in tissues of mice 24 hours after infection with Listeria monocytogenes [13,16]. No induction of Irgc was detected in liver, lung, or spleen after 50 cycles of amplification, whereas Irga2, used as a positive control, was induced in all three tissues (Figure 3b; panel i right). However, Irgc, unlike Irga2, was constitutively expressed in the mature mouse testis (Figure 3b; unpublished data). We conclude that mouse Irgc is expressed in a tissue-specific manner and is not induced by infection.
The coding sequences of the p47 GTPases
In Figure 4 we present the predicted translation products of the 21 intact p47 GTPase genes, and reconstructed partial sequences of the two pseudo-genes, Irga5ψ and Irgb7ψ, aligned on the secondary structures of Irga6 [12] and H-Ras-1 [17]. The full alignment confirms a number of major features that are already apparent from the previously published alignment of six family members [13] and consolidates the definition of the p47 GTPases as a distinctive sequence family. Especially noteworthy are novel features of the amino- and carboxyl-termini, which were not apparent before. Eleven of the proteins, including six of chromosome 18 Irga gene products and Irgb2, Irgb5, Irgb9 and Irgb10, carry the amino-terminal myristoylation signal MGxxxS [18]. This sequence in Irga6 (IIGP1) is indeed myristoylated in vitro [19] and in vivo, and, as expected, favors binding of the protein to membranes [6]. No other membrane attachment sequences or lipid modification motifs are apparent in p47 GTPase sequences, despite the documented attachment of several of these proteins to membranes [5,6,16]. Irgb2, Irgb5, Irgb7, Irgb9 and Irgc have carboxyl-terminal extensions up to 65 residues in length compared with the canonical IIGP1 sequence.
The p47 GTPase genes of the human genome
Only two IRG sequences, both transcribed, are present in humans (or chimpanzee), one (IRGC) on chromosome 19 (19q13.31) and the other (IRGM) on chromosome 5 (5q33.1). Human IRGC is more than 85% identical at the nucleotide level and 90% at the amino acid level to the isolated mouse gene Irgc. IRGM encodes an amino- and carboxyl-terminally truncated G-domain homologous to the Irgm (GMS) subfamily of mouse p47 GTPases. Predicted protein products of IRGC and the IRGM gene fragment are included in an extended phylogeny (Figure 5) and alignment (Figure 6) of the vertebrate IRG proteins.
The IRGC mouse and human genes sit in chromosomal regions syntenic between chromosomes 7 and 19, respectively (Figure 7a) and are clearly orthologous. The proximal promoter region of human IRGC is largely conserved with that of mouse Irgc. However, as in the mouse, no interferon response elements are found either in the proximal conserved region or in divergent regions up to 10 kb upstream of the transcriptional start (data not shown). Human IRGC, like mouse Irgc, is not inducible in vitro by interferons, is not expressed detectably in brain or liver, but is strongly expressed in adult testis (Figure 3b, panel ii). As in the mouse, a weak Sox element is present in the proximal promoter of human IRGC.
The human genomic segments syntenic to the mouse chromosome 11 and chromosome 18 IRG gene clusters both mapped to human 5q33.1, suggesting that the interferon-inducible IRG proteins were once encoded in a single block ancestral to the human chromosome 5 region (Figure 7b). IRGM maps only 80 kb away from the closest syntenic marker DCTN4. IRGM is transcribed in unstimulated human tissue culture lines HeLa and GS293 (Figure 8a), with no increase after interferon induction. Polyadenylated transcripts of IRGM occur with five 3' splicing isoforms extending more than 50 kb 3' of the long coding exon (Figure 8b). The transcripts have a 5'-untranslated region of more than 1,000 nucleotides that corresponds largely to the U5 region of an ERV9 repetitive element [20]. The promoter region corresponds to the ERV9 U3 LTR (long terminal repeat) without interferon response elements, and three of the five splice forms have exon-intron boundaries downstream of the putative termination codon, normally a signal for rapid RNA degradation [21].
At the protein level the shortest isoform of IRGM is shorter than a canonical G-domain, being truncated in the middle of β-strand 6 just before the G5 sequence motif, which interacts with the guanine base of the bound nucleotide (Figures 6 and 8b; also see Ghosh and coworkers [12]). The longer isoforms are terminated by short sequence extensions that are unrelated to known GTPase domains. A rabbit antiserum raised against recombinant human IRGM produced in Escherichia coli failed to detect signal by immunofluorescence or Western blot in human cell lines (data not shown).
IRG genes of the dog
Is the mouse (order Rodentia) or the human (order Primata) the exception? We looked for IRG genes in a third order of mammals, the Carnivora. We recovered a total of eight IRG genes from the public genome database of the dog Canis familiaris (Figures 5 and 6) as well as a partial sequence of a 9th gene (not shown). Of these, one (not shown) is a pseudo-gene by a number of criteria, another is clearly dog IRGC, whereas the partial sequence is novel but most closely related to IRGC. The remainder assort into segments of the phylogeny already established for the interferon-inducible mouse IRG genes (Figure 5). Both GMS and GKS genes are represented and are inducible by interferon in dog MDCK epithelial cells (Additional data file 4). The three dog GMS genes appear to have diversified independently from the mouse GMS genes (Figure 5). As in humans and mouse, dog IRGC was not induced by interferon-γ (Additional data file 4). Overall, the IRG gene status of the dog clearly resembles that of mouse rather than that of humans.
IRG genes in fish genomes
IRG GTPases are at least as old as the vertebrates. We have identified at least two distinct irg genes in the freshwater pufferfish Tetraodon nigriviridis, a closely linked pair of irg genes in the saltwater pufferfish Fugu rubripes, and at least 11 partially clustered irg genes in the zebrafish Danio rerio (Figures 5 and 6, and Additional data file 5). The fish irg genes fall into separate clades from the mammalian genes (Figure 5). A specific IRGC homolog is not immediately apparent. GMS subfamily IRGM genes are absent from fish. The pufferfish and zebrafish irgf genes have one intron identically positioned at the end of helix 4 of the G-domain (indicated on Figure 6; also see Additional data file 5). This intron is 81 bp long in both pufferfish species but is substantially longer in the zebrafish genes. The distinct irge subfamily of the Danio irg genes are intronless in the open reading frame, like mammalian IRG genes.
IRG homologs with divergent nucleotide-binding regions: the quasi-GTPases
The mouse, human and zebrafish genomes encode proteins that are homologous to the IRG GTPases but are radically modified in the GTP-binding site. The mammalian protein FKSG27 (IRGQ), a protein of unknown function that is 70% conserved between man and mouse, is extended amino-terminally relative to a p47 GTPase by about 100 residues encoded on three short exons. The remaining 420 residues, encoded on a single long exon, are clearly homologous to and colinear with the IRG proteins (Figure 6 and Additional data file 6), especially in the amino- and carboxyl-terminal parts of the exon. The region of lowest similarity is in the G-domain, and conserved GTP-binding motifs are lacking (Figure 6, and Additional data files 6 and 7). Thus, FKSG27 (IRGQ) is not a GTPase despite its phylogenetic relationship to the IRG proteins. FKSG27 (IRGQ) is closely linked to IRGC in humans and mouse (Figure 7a).
The zebrafish genome contains three IRG homologs with more or less modified GTP-binding motifs (irgq1-irgq3; Figures 5 and 6, and Additional data file 7). Their homology to IRG genes is stronger than that of FKSG27 (IRGQ), but as with FKSG27 (IRGQ) their function as GTPases is doubtful. The irgq1 gene is clustered on a single BAC clone with four apparently normal irge genes and immediately downstream of a truncated p47 gene, irgg, with which irgq1 is transcribed as the carboxyl-terminal half of a tandem transcript. Thus, the hypothetical protein product would be a carboxyl-terminally truncated p47 GTPase, linked at its carboxyl-terminus to a similarly truncated p47 homolog probably without GTPase function.
We propose to term the modified IRG proteins without GTPase function 'quasi IRG' proteins, hence IRGQ. IRGQ sequences reveal their phylogenetic relationship to the IRG proteins, but they are nevertheless more or less radically modified, primarily in the nucleotide binding site. In view of the substantial divergence between the IRGQ genes and functional p47 GTPases, it was unexpected not to find close homologs of the Danio irgq sequences in either the Fugu or Tetraodon genomes. The evolution and diversity of the Danio irgq genes is apparently linked to the evolution and diversity of the GTPase-competent IRG sequences.
IRG homologs outside the vertebrates
No unambiguous IRG homologs have been found outside the vertebrates. However, two possibly related sequence were recovered from the Caenorhabditis elegans genome, and several groups of putative GTPases of unknown function exist in the bacteria that have sequence features reminiscent of IRG GTPases. Perhaps the most striking of these are found in the Cyanobacteria (see Additional data file 1 for accession numbers for these sequences). Among other features, all of these sequences have in common with the IRG GTPases the presence of a large hydrophobic residue in place of the familiar catalytic Q61 of H-Ras-1, but this feature is far from diagnostic for the IRG GTPases [22]. Despite several suggestive characteristics of these invertebrate and bacterial GTPase sequences, it is not possible on the basis of sequence criteria alone to establish their phylogenetic relationship with vertebrate IRG proteins.
Discussion
The p47 GTPases (IRG proteins) are an essential resistance system in the mouse for immunity against pathogens that enter the cell via a vacuole. In this study we reached several unexpected conclusions about the evolution of the system. First, the IRG resistance system, despite its importance for the mouse, is absent from humans because it has been lost during the divergent evolution of the primates. Second, the IRG resistance system is at least as old as the bony fish but missing in the invertebrates. Finally, the IRG proteins appear to be accompanied phylogenetically by homologous proteins, here named IRGQ proteins, that probably lack nucleotide binding or hydrolysis function, and that may form regulatory heterodimers with functional IRG proteins. We consider these points in order.
The argument for the absence of the IRG resistance system in humans relies on several findings. The system is reduced from 23 genes in mouse to one full-length gene and a transcribed G-domain in humans, and the residual genes lack the character of functional resistance genes. Thus, IRGC is highly conserved in humans, dog and mouse, is not interferon or infection inducible, and is expressed constitively in mature testis. IRGM, although clearly derived from a typical GMS subfamily resistance gene, is transcribed constitutively from an endogenous retroviral LTR, is unresponsive to interferon, and appears to be structurally damaged in several ways.
We argue that the IRG resistance system has been lost from primates (the situation in chimpanzee is identical to that in humans; unpublished data) rather than gained by the murine rodents (including rat; unpublished data) on the following grounds. First, like the mouse, the dog genome has several complete, interferon-inducible IRG genes in addition to IRGC. Second, humans and chimpanzees possess a degraded member of the GMS subfamily of IRG proteins, confirming that this distinctive subfamily, present and functional as resistance genes in dog and mouse, was widely distributed at the origin of the mammalian radiation. Finally, the IRG system is present in bony fish, representing ancient vertebrates. Rapid expansion and contraction of multigene families associated with pathogen resistance has frequently been documented in both animals and plants [23-28]. In all of these cases, however, the resistance mechanism itself has been retained as its protein mediators have evolved or even, in the natural killer receptor case, been replaced by a different molecular species [29]. The IRG case may be different because here the resistance mechanism itself has apparently been lost during primate evolution.
It will be of interest and of considerable importance to analyze the different strategies by which humans, dog and mouse deploy resistance mechanisms effective against vacuolar pathogens. None of the known mechanisms active in humans against vacuolar pathogens, namely nitric oxide and oxygen radicals [30-32], tryptophan depletion [33,34], accelerated acidification by Rab5a [35], cation depletion [36-38] or autophagy [39,40], is missing from the mouse. Nevertheless, it remains possible that the distinctive resistance actions of the p47 GTPases [7,8] are performed by an unrelated and thus far unidentified molecular machine in primates.
The loss of a highly evolved and complex resistance system that is active against vacuolar pathogens needs an adaptive explanation. The evolution of a successful avoidance strategy by the pathogens is unlikely, because many different pathogens and pathogen classes are controlled by IRG proteins in the mouse [2,4]. Nevertheless very recent evidence suggests that Chlamydia spp. divergence between humans and mouse may indeed be partially driven by differences in the deployment of p47 GTPases, in this case Irga6 (IIGP1) [41]. Human Chlamydia trachomatis is controlled by IIGP1 in interferon-treated mouse oviduct epithelial cells, whereas control of the extremely closely related mouse C. muridarum is independent of interferon. However, a more plausible general model is the evolution in the primate lineage of improvements to the battery of parallel mechanisms, rendering the IRG system redundant. Interestingly, in this context it was noted in the Chlamydia study quoted above that interferon-stimulated mouse oviduct epithelial cells did not express the important resistance factor indoleamine deoxygenase, which is responsible for tryptophan depletion, whereas this is well expressed in interferon-stimulated human HeLa cells [41]. In general, pathogen resistance mechanisms also carry costs, for example autoimmunity and allergy, arising from the adaptive immune system and many others arising from innate immunity [42-46]. Indeed, the interferon-inducible dynamin-like GTPases, the Mx proteins, which confer on mice strong resistance to certain RNA viruses and have been considered functionally related to the p47 GTPases [4,47-49], exist in a balanced polymorphism in the wild over null alleles [50,51], and have been lost spontaneously from all except two laboratory mouse strains [52]. It is not yet obvious what specific costs might be associated with possession of the Mx or IRG resistance systems.
The IRG proteins are well represented in the bony fish but, although they are abundant and diverse in the zebrafish, in Fugu there are only two very closely linked and similar genes that are, indeed, annotated as a single tandem gene in ENSEMBL (although we judge this not to be the case). Thus, the available annotated fish genomes seem to mirror the IRG situation in the mammals, with Fugu and Tetraodon reflecting the reduced human case and Danio the complex dog and mouse. However, it has not yet been reported whether any fish IRG genes respond to infection [53]. Both Danio and the pufferfish are Actinopterygian fish, in which where there is increasing agreement that the genome has been amplified by three rounds of duplication [54,55]. It is plausible that the complex IRG representation in Danio may be attributed to preservation of these genes on more than one of the potential eight paralogons, whereas only a single copy carries IRG genes in the pufferfish. Further clarification of this issue awaits the completion of the genomes.
The phylogenetic origin of the IRG proteins is obscure. Because the family is conserved at least down to the bony fishes with little structural modification, the IRG genes are not strictly fast-evolving and their basic conservatism makes them easy to identify. Thus, their apparent absence from most known invertebrate genomes is probably real. Although many components of the adaptive immune system appear to have evolved close to the chordate-vertebrate boundary [56], this is not generally the case for innate immune mechanisms [57,58] There seems to be no reason in principle why the IRG system should not work in invertebrates because it is cell-autonomous. However, the putative GTPase sequences that we have recovered from C. elegans and from Cyanobacteria are too distantly related outside the G-domain for a clear phylogenetic relationship to IRG proteins to be established from sequence similarity alone, whereas the similarities within the G-domains, although occasionally striking, are to some extent forced by the maintenance of a highly conserved function, namely regulated GTP hydrolysis. A stronger case for a meaningful phylogenetic relationship between these proteins and the vertebrate IRG proteins would follow from structural evidence that they display the distinctive IRG fold exemplified by mouse Irga6 (IIGP1) and from a detailed analysis of their catalytic mechanism.
The basic unit of IRG protein function may be a dimer because several genes we have identified occur in pairs in a head-to-tail arrangement, are expressed as tandem transcripts, and are presumably expressed as dimeric proteins. This conclusion is consistent with the dimer of IIGP1 (Irga6) observed in the crystal, shown by site-directed mutagenesis of the dimer interface to be essential for GTP-dependent oligomerization and cooperative hydrolysis [12]. However, a second dimer interface is also required for oligomerization (unpublished data), and which of the two dimer structures the constitutive IRG dimers represent is of considerable interest. Unlike the observed homodimer of IIGP1, the products of the two putative tandem genes in the mouse Irgb2/b1 and Irgb5/b4 are heterodimers, implying that the two IRG subunits serve distinct functions in the protein. The same may be true for the tandem pair of irg genes of Fugu, which are annotated as a single tandem gene in ENSEMBL. These latter genes have diverged very recently, because with three exceptions they are identical at the nucleotide level over the first 290 amino acids. However, they have diverged substantially in the carboxyl-terminal region (Figure 6), suggesting a recent selective force. If the two tandem IRG domains are indeed expressed as a tandem protein (as favored by the ENSEMBL annotation), then it will be as a heterodimer with significant sequence variation at the carboxyl-terminus. The extreme case of heterodimer differentiation in IRG tandem genes may occur in Danio, in which gene irgg, a canonical (although truncated) IRG gene, is apparently expressed in tandem with the adjacent downstream gene irgq1, which is a modified (and also truncated) quasi-GTPase gene that is unlikely to function as a GTPase (Additional data file 7). In this case the role of the irgq1 domain may be regulatory for the amino-terminal irgg domain. Other IRGQ proteins may also be regulators of IRG proteins, interacting with the functional IRG proteins with a symmetry resembling one or other of the two dimer structures. Thus IRGQ proteins would coevolve with IRG proteins. This would explain why the three irgq genes of the zebrafish have no homologs in the pufferfishes, with their single tandem pair of irg proteins, and recalls the recent observation that the GAP protein of the small GTPase, Rap1, is itself probably derived from a GTPase ancestor, retaining the G-domain structure but not the sequence to reveal its origin [59].
Better understanding of the mechanism of action and regulation of the p47 GTPases is needed before their complex evolutionary history can be put in context. From the evidence we present here, however, it is already clear that effective resistance to vacuolar pathogens in humans and mouse must be organized on radically different principles.
Nomenclature
We introduce here a general nomenclature on phylogenetic principles for the p47 GTPases, based on the stem name IRG (immunity-related GTPases). This stem was favored over other possibilities because the name IRG-47 has priority in the literature as the first description of a p47 GTPase [60]. The phylogenetic basis of the nomenclature is apparent from Figure 5; each deep monophyletic clade is identified by a single-letter suffix as IRGM or IRGC. The nomenclature proposed here in Figure 5 and in Additional data file 1 has been accepted by the gene nomenclature committees of human and mouse, and by the zebrafish sequencing project. We have tried throughout to use the different forms of gene name accepted by the nomenclature committees for mouse (Irg) human (IRG), dog (IRG) and zebrafish (irg). The nomenclature of the IRGQ genes departs from the phylogenetic principle. The IRGQ nomenclature simultaneously recognizes the affinity of these sequences to IRG genes and stresses their anomalous GTP-binding domain features. It is, however, highly unlikely that the IRGQ sequences of humans, mouse, and fish represent a monophyletic group. It is more likely that the IRGQ genes of each taxonomic group derive from IRG genes of other clades of that group. This pattern is already apparent by inspection of the irgq protein sequences of the zebrafish in Figure 6, but it cannot be discerned from the G-domain-based phylogeny shown in Figure 5 because of the specific divergence of the G-domains of the irgq proteins.
Materials and methods
Use of database resources
All available public databases were extensively screened by BLAST and related searches for sequences belonging to the IRG family. In the case of the mouse, transcript sequences derived from the C57BL/6 strain were given preference over sequences of other and undefined strain origin, and compared in all cases with genomic sequence available via the ENSEMBL (v28.33d.1, February 2005) array of websites [61]. A systematic study of polymorphism has not yet been completed, but it is already clear that nearly all IRG sequences derived from the CZECHII cDNA libraries (Mus musculus musculus) differ from C57BL/6 sequences. These differences make allocation of many CZECHII sequences to individual clade members of the C57BL/6 mouse problematic. Identification of certain Irg sequences with recognized gene symbols was achieved through the Mouse Genome Initiative web resources [62]. Where ambiguities persist in the mouse genomic map, especially on chromosome 18 in the region of IrgA6-IrgA8 (Mb60.878-60.958), and on chromosome 11 in the region from PA28βψ to IrgB7ψ (Mb57.570-57.700), we used primary BAC and cosmid sequences to reach a consensus view.
Human and dog IRG sequences were identified from the available public databases (ENSEMBL, National Center for Biotechnology Information) and confirmed wherever possible by multiple sequence comparisons at transcriptional and genomic levels. Fugu material was obtained and analyzed through [63-65]. Tetraodon sequence was initially assembled from the GSS sequence database at National Center for Biotechnology Information and subsequently from the University of California at Santa Cruz compiled genome database [66] via the BLAST server. Zebrafish sequence was obtained from zebrafish genome resources at the Sanger Centre [67] and analyzed in an Acedb database using the Spandit annotation tool.
Chromosomal locations and synteny analysis of mouse and human chromosomes was initiated through ENSEMBL [68]. Further details were obtained through the Sanger Centre [69]. Nucleotide sequences and translated open reading frames of IRG family members used in this paper are given in Additional data files 9 and 10, and can also be accessed at the IRG family database at our laboratory [14].
Phylogeny and alignment protocols
Routine sequence analysis and local sequence database management was handled using DNA-Strider 1.3f12, Vector-Nti, and MacVector 7.2. The identity and similarity matrix on protein and nucleotide sequences (Additional data file 2) are based on GeneDoc (version number 2.6.002). Phylogenetic analysis was conducted using the neighbor-joining method [70], as implemented in the MEGA2 program [71]. We used p-distances for constructing the phylogenetic trees. Reliability of the neighbor-joining trees was examined using the bootstrap test [72].
Alignments were performed via the BCM multiple alignment programme suite [73] and EBI-ClustalW [74] using the default options and manipulated according to the crystal structure of IIGP1 [12]. Shading of alignments was performed using Boxshade [75] and additional sequences were shaded manually according to the default options of Boxshade.
Identification of transcription factor binding sites
Promoter regions (2 kb upstream of putative transcription start point) were screened for putative transcription factor binding sites with the Transcription Element Search System [76,77], and the results were further analysed and confirmed manually. Additional promoter analysis of Irgc (mouse Cinema) and IRGC (human CINEMA) was performed with ConSite [78] based on phylogenetic footprinting [79].
RT-PCR on cells and tissues
C57BL/6J mice were obtained from the animal house at the Institute for Genetics, University of Cologne. Listeria monocytogenes infection was performed as described previously [13]. Twenty-four hours after infection, the mice were killed, and liver, lung and spleen were removed and snap frozen in liquid nitrogen. Mouse L929 fibroblasts were stimulated for 24 hours with 200 U/ml interferon-γ or 200 U/ml interferon-β (R&D System GmBH, Weisbaden-Nordenstadt, Germany and Calbiochem-Novabiochem Corparation La Jolla, CA, respectively). Human cell lines (Hela, GS293 (GeneSwitch™ -293, Invitrogen GmbH Karlsruhe, Germany), HepG2, T2, THP1, MCF-7, SW-480, Primary foreskin fibroblast-HS27) were stimulated for 24 hours with 2,000 U/ml interferon-β or 200 U/ml interferon-γ (PBL Biomedical Laboratories, NY, USA and Peprotech/Cell concepts GmbH UmKirsh, Germany, respectively). Total RNA was extracted from tissues and cells using the RNAeasy mini kit (QIAGEN, Hilden, Germany), except for testis, for which the RNAeasy Lipid Tissue Kit (QIAGEN) was used. Poly(A) RNA was isolated from total RNA using the Oligotex mRNA kit (QIAGEN). Total RNA from human tissues was purchased from Biochain (Hayward, CA, USA). cDNA was generated from mRNA and total RNA using the Super Script First-Strand Synthesis System for RT-PCR (Invitrogen, Carlsbad, CA, USA). The generated cDNAs were screened for the presence of p47 (IRG) GTPase transcripts by PCR. A list of the primers used is given in Additional data file 8. The amplified fragments were confirmed by sequencing.
Additional data files
The following additional data are included with the online version of this paper: A list of all IRG gene family members described in this paper (gives names, synonyms, accession numbers and further information for each IRG gene; Additional data file 1); nucleotide and amino acid identities based on G-domain of mouse Irg family (gives percentage of identity on both protein and nucleotide level within the mouse Irg family; Additional data file 2); ISRE and GAS elements of mouse IRG family genes (contains the positions and exact sequences of all ISRE and GAS elements found in putative promoters of mouse IRG genes; Additional data file 3); inducibility of Dog p47 (IRG) GTPases (shows interferon inducibility of members of the p47 (IRG) GTPases present in the dog; Additional data file 4); genomic organization of Danio rerio p47 (irg) GTPases (illustrates the genomic organization of all p47 (irg) GTPases found in zebrafish to date; Additional data file 5); protein similarity matrix of Irgc and Irgq (contains comparison between the mouse p47 GTPase Irgc and the long coding exon of the closely linked quasi-GTPase Irgq (FKSG27); Additional data file 6); divergent nucleotide-binding motifs in quasi-GTPases (compares the nucleotide binding motifs of quasi-GTPases to those of the classical mouse p47 GTPases; Additional data file 7); a list of the primers used (contains the sequences of all primers used in this study; Additional data file 8); nucleotide sequences of all IRG family members (Additional data file 9); protein sequences of all IRG family members (Additional data file 10).
Supplementary Material
Additional data file 1
A list of all IRG gene family members described in this paper (gives names, synonyms, accession numbers and further information for each IRG gene)
Click here for file
Additional data file 2
Nucleotide and amino acid identities based on G-domain of mouse Irg family (gives percentage of identity on both protein and nucleotide level within the mouse IRG family)
Click here for file
Additional data file 3
ISRE and GAS elements of mouse IRG family genes (contains the positions and exact sequences of all ISRE and GAS elements found in putative promoters of mouse IRG genes)
Click here for file
Additional data file 4
Inducibility of Dog p47 (IRG) GTPases (shows interferon inducibility of members of the p47 (IRG) GTPases present in the dog)
Click here for file
Additional data file 5
Genomic organization of Danio rerio p47 (IRG) GTPases (illustrates the genomic organization of all p47 (IRG) GTPases found in zebrafish to date)
Click here for file
Additional data file 6
Protein similarity matrix of Irgc and Irgq (contains comparison between the mouse p47 GTPase Irgc and the long coding exon of the closely linked quasi-GTPase Irgq (FKSG27)
Click here for file
Additional data file 7
Divergent nucleotide-binding motifs in quasi-GTPases (compares the nucleotide binding motifs of quasi-GTPases to those of the classical mouse p47 GTPases)
Click here for file
Additional data file 8
A list of the primers used (contains the sequences of all primers used in this study)
Click here for file
Additional data file 9
Nucleotide sequences of all IRG family members
Click here for file
Additional data file 10
Protein sequences of all IRG family members
Click here for file
Acknowledgements
We are greatly indebted to Lois Maltais of the Mouse Genome Database at The Jackson Laboratory; Ruth Lovering, Gene Nomenclature Advisor, HUGO Gene Nomenclature Committee (HGNC); and Yvonne Edwards of the Fugu Genomics Project at the UK Human Genome Mapping Project (HGMP) Resource Centre for their time and effort in developing a useful nomenclature for the p47 GTPases. We are grateful to Kerstin Jekosch, Informatics & Systems Groups, Sanger Centre, for help with analyzing and annotating the zebrafish genes; to Cornelia Stein, Institute for Genetics, Cologne for communicating unpublished zebrafish material; and to Natasa Papic, Institute for Genetics, Cologne for assistance in editing the long sequence alignments. This study was supported by the Centre for Molecular Medicine, Cologne, and DFG grants SPP1110, SFB243 and SFB635. Iana Parvanova was supported by the DFG Graduate College 'Genetics of Cellular Systems'; and Cemalettin Bekpen and Julia Hunn were supported by the Cologne Graduate School in Genetics and Functional Genomics. We are particularly grateful to the anonymous referee who drew our attention to the candidate p47 GTPase sequence C46E1.3 in C. elegans.
Figures and Tables
Figure 1 Genomic positioning and phylogenetic relationship of mouse Irg GTPases. (a) Disposition of the 23 Irg genes on the mouse karyotype. Individual Irg genes are listed in correct gene order in each cluster. (b) Positioning and orientation of Irg genes in the mouse chromosome 11 and 18 clusters. Positions of genes refer to the location in Mouse ENSEMBL release (v28.33d.1, February 2005) [61] of the first G of the glycine codon of the G1 motif (GKS or GMS) of the GTP-binding domain of each gene. The segments of the chromosome 11 cluster indicated with square brackets are regions of uncertain structure. Gene orientation is given by black arrows. The shaded region of the chromosome 11 map is a duplication introduced in Mouse ENSEMBL v28.33d.1 (February 2005) in an attempt to resolve a region of high ambiguity indicated by the longer square bracket. In our view this duplication does not resolve the ambiguities consistently, and we see no justification at present for the duplicated Irgb5 and Irgb6 genes. The sibling genes Irgb3 and Irgb4 differ by only nine nucleotides; in this case, however, the independent existence of the two genes is proved by the proximity of the PA28βψ retropositioned pseudo-gene to Irgb3 but not to Irgb4, in addition to consistent sequence differences. We have left the duplication of the Irgb5/Irgb6 region in the map for consistency of the base numbering with this release of ENSEMBL. *Indicates minor sequence differences presumably due to sequencing errors. (c) Unrooted tree (p-distance based on neighbour-joining method) of nucleotide sequences of the G-domains of the 23 mouse Irg GTPases, including the two presumed pseudo-genes Irga5 and Irgb7. The sources of all Irg sequences are given in Additional data file 1, and the nucleotide and amino acid sequences themselves are collected in the p47 (IRG) GTPase database from our laboratory website [14]. (d) Phylogenetic tree of the amino acid sequences of the G-domains of 21 mouse Irg GTPases rooted on the G-domain of H-Ras-1 (accession number: P01112). The products of the two presumed pseudo-genes Irga5 and Irgb7 are excluded from the analysis.
Figure 2 Genomic and promoter structure of mouse Irg GTPases. (a) Genomic structure of mouse Irg genes. Green blocks indicate coding exons and blue blocks indicate 5'-untranslated exons. Orange arrows identify putative promoter regions. Stars identify exons shown to be excluded in alternative splice forms. The scale bar is measured in base pairs up to the first base of the long coding exon. Note the presence of two promoters for Irga6 and Irgd. (b) Interferon response elements in the promoter regions of mouse Irg genes. γ-Activated sequences (GAS; pale blue blocks) and interferon-stimulated response element (ISRE; red blocks) sequences were identified in the promoters shown in panel a (also see Additional data file 7). Dark blue blocks downstream of each promoter represent the most 5' exon. The yellow block identifies a putative Sox1 transcription factor binding site in the proximal promoter region of Irgc. The scale bar is measured in base pairs from the first base of the 5' exon.
Figure 3 Interferon responsiveness of mouse and human p47 (IRG) GTPase. (a) Interferon (IFN)-γ responsiveness of eight new mouse Irg genes. Inducibility of eight further Irg genes (also see Boehm and coworkers [13]) in L929 fibroblasts induced for 24 hours with IFN-γ, demonstrated by RT-PCR. D refers to a positive control genomic DNA template; O refers to a negative control of the same genomic template after DNAse1 treatment; and + and - refer to RT-PCR on DNAse1-treated RNA templates from IFN-γ-induced and IFN-γ-noninduced cells, respectively. The sibling genes of the Irgb series could not be individually amplified because of their close sequence similarity. The identities of the amplified genes responding to interferon induction, indicated by vertical arrows, were subsequently established by sequencing of multiple clones from the PCR product. (b) Irgc is not induced by interferon or infection but is constitutively expressed in testis. (i, left) Mouse L929 fibroblasts were induced for 24 hours with IFN-β or IFN-γ or left uninduced (-). Irgc could not be detected by RT-PCR even after 50 amplification cycles in L929 cells. Irga2 after 50 cycles was used as a positive control for the interferon-induced L929 RNA. RNA from mouse testis provided a positive control for Irgc. (i, right) RT-PCR for Irgc and Irga2 (50 and 30 amplification cycles respectively) on RNA from tissues of uninfected mice (-) or mice infected 24 hours previously with Listeria monocytogenes (+). Irga2 was induced in all tissues and Irgc in none. RNA from mouse testis provided a positive control for Irgc, which is detected after 50 cycles. Testis expression of Irga2 was barely detected after 30 cycles (compare with i, left, showing Irga2 in testis after 50 cycles). (Panel ii, left) Human IRGC is not induced by 24 hours of stimulation with IFN-β or IFN-γ in human cell lines (induction of GBP-1 [accession number P32455] was assayed as a positive control) and (Panel ii, right) is constitutively expressed only in human testis. GAPDH was used as control.
Figure 4 Amino acid alignment of the mouse Irg GTPases. Sequences of all 23 mouse Irg GTPases showing the close homology extending to the carboxyl-terminus, aligned on the known secondary structure of Irga6 (indicated in blue above sequence alignment). The sequences of notional products of the two pseudo-genes Irga5 and Irgb7 have been partially reconstructed; premature terminations are indicated by red highlighting. In the C57BL/6 mouse the sequence of the Irga8 gene is damaged by an adenine insertion, indicated by the red highlighted K at position 204. (The sequence given after this point is that given after correcting the frameshift, and is identical to that of the CZECHII [Mus musculus musculus] sequence BC023105 that lacks the extra adenine.) The turquoise-highlighted M in M1 and M2 are initiation codons that are dependent on alternative splicing (also see Figure 2a); the unusual methionine residues in the G1 motif of GMS proteins are highlighted in green. The blue background Q residue of Irgb5 and Irgb2 at positions 405 and 396 indicate the point at which tandem splicing occurs to Irgb4 and Irgb1, respectively. Canonical GTPase motifs are indicated by red boxes.
Figure 5 Extended phylogeny of the G domains of IRG and related proteins. The phylogeny relates all of the IRG sequences described in this report and reveals the distinct clades on which the nomenclatural fine structure is based. All except the mouse sequences are labeled with the species of origin. Dog IRG sequences are found in the B, C, D and M clades, and human sequences only in clades C and M. The mouse and human quasi-IRG proteins, IRGQ (FKSG27), could not be included in the phylogeny because they are so deviant in the G-domain (see Figure 6 and Additional data file 6).
Figure 6 Extended alignment of the vertebrate IRG proteins. Individual sequences are given in full and are labeled as in Figure 5. Unusual residues in the G1 motif are highlighted (M of the GMS proteins in green and two deviant residues in the zebrafish irgq sequences in pink). The essential structural relationship between IRG genes and quasi-IRG genes is apparent in the alignment despite the modified G-domains. For mouse and human IRGQ the long carboxyl-terminal coding exons that contain the p47 homology were used for the alignment. In human IRGQ the sequence ENPKGESLKNAGGGGLENALSKGREKCSAGSQKAGSGEGP was removed from the alignment between positions 210 and 211 (highlighted in turquoise) to prevent extensive gap formation. The position of the intron present in pufferfish and zebrafish irgf genes is indicated by two adjacent residues highlighted in blue.
Figure 7 Synteny relationships between the human and mouse IRG genes (a) Synteny between mouse chromosome 7 and human chromosome 19 in the region of the IRGC and IRGQ genes. The figures indicate distances from the centromere in megabases. The locations of three further syntenic markers are given. Gene orientation is given by black arrows. (b) Complex synteny relationship between human chromosome 5 and mouse chromosomes 11 and 18 in the regions containing the mouse Irg genes. Figures indicate distances from the centromere in megabases. The locations of IRG genes are shown in the yellow panels. Positions of diagnostic syntenic markers are also indicated. Syntenic blocks are given in full color, and the rest is shaded.
Figure 8 Structure and expression of the human IRGM gene. (a) (left panels) RT-PCR analysis of expression of IRGM in HeLa and GS293 cells. The b and c splice variants were amplified simultaneously by the same primer pair (IRGMs1-rGMS). A different downstream primer (IRGMs1-r1) internal to all the 3' splice forms was used to show differences in the overall expression level of IRGM in the two cell lines. No RT' indicates that no reverse transcriptase is included in cDNA preparation. The band immediately below the IRGMc band in GS293 cell material, indicated with an asterisk, is a nonspecific band amplified only in this cell line. The band was sequenced and is unrelated to IRGM. (right panel) Analysis of IRGM expression in human brain, liver and testis by RT-PCR. GAPDH was used as a control. (b) Five splice forms of the IRGM gene have been identified, as indicated: IRGM(a)-IRGM(e). The promoter and 5'-untranslated regions of the gene are associated with an ERV9 retroviral LTR. Scale-bar is given in base pairs.
==== Refs
Mestas J Hughes C Of mice and not men: differences between mouse and human immunology. J Immunol 2004 172 2731 2738 14978070
Taylor GA Feng CG Sher A p47 GTPases: regulators of immunity to intracellular pathogens. Nat Rev Immunol 2004 4 100 109 15040583 10.1038/nri1270
MacMicking JD Immune control of phagosomal bacteria by p47 GTPases. Curr Opin Microbiol 2005 8 74 82 15694860 10.1016/j.mib.2004.12.012
MacMicking JD IFN-inducible GTPases and immunity to intracellular pathogens. Trends Immunol 2004 25 601 609 15489189 10.1016/j.it.2004.08.010
Taylor GA Stauber R Rulong S Hudson E Pei V Pavlakis GN Resau JH Vande Woude GF The inducibly expressed GTPase localizes to the endoplasmic reticulum, independently of GTP binding. J Biol Chem 1997 272 10639 10645 9099712 10.1074/jbc.272.16.10639
Martens S Sabel K Lange R Uthaiah R Wolf E Howard JC Mechanisms regulating the positioning of mouse p47 resistance GTPases LRG-47 and IIGP1 on cellular membranes: retargeting to plasma membrane induced by phagocytosis. J Immunol 2004 173 2594 2606 15294976
Martens S Parvanova I Zerrahn J Griffiths G Schell G Reichman G Howard JC Disruption of Toxoplasma gondii parasitophorous vacuoles by the mouse p47 resistance GTPases. PLoS Pathogens 2005
MacMicking J Taylor GA McKinney J Immune control of tuberculosis by IFN-gamma-inducible LRG-47. Science 2003 302 654 659 14576437 10.1126/science.1088063
Uthaiah R Praefcke GJM Howard JC Herrmann C IIGP-1, a interferon-γ inducible 47 kDa GTPase of the mouse, is a slow GTPase showing co-operative enzymatic activity and GTP-dependent multimerisation. J Biol Chem 2003 278 29336 29343 12732635 10.1074/jbc.M211973200
Tuma PL Collins CA Activation of dynamin GTPase is a result of positive cooperativity. J Biol Chem 1994 269 30842 30847 7983015
Prakash B Renault L Praefcke GJ Herrmann C Wittinghofer A Triphosphate structure of guanylate-binding protein 1 and implications for nucleotide binding and GTPase mechanism. EMBO J 2000 19 4555 4564 10970849 10.1093/emboj/19.17.4555
Ghosh A Uthaiah R Howard JC Herrmann C Wolf E Crystal structure of IIGP1: a paradigm for interferon-inducible p47 resistance GTPases. Mol Cell 2004 15 727 739 15350217 10.1016/j.molcel.2004.07.017
Boehm U Guethlein L Klamp T Ozbek K Schaub A Fütterer A Pfeffer K Howard JC Two families of GTPases dominate the complex cellular response to interferon-γ. J Immunol 1998 161 6715 6723 9862701
p47 (IRG) GTPase Database
Gilly M Damore MA Wall R A promoter ISRE and dual 5' YY1 motifs control IFN-gamma induction of the IRG-47 G-protein gene. Gene 1996 179 237 244 8972906 10.1016/S0378-1119(96)00366-6
Zerrahn J Schaible UE Brinkmann V Guhlich U Kaufmann SH The IFN-inducible Golgi- and endoplasmic reticulum-associated 47-kDa GTPase IIGP is transiently expressed during listeriosis. J Immunol 2002 168 3428 3436 11907101
Pai EF Krengel U Petsko GA Goody RS Kabsch W Wittinghofer A Refined crystal structure of the triphosphate conformation of H-ras p21 at 1.35 A resolution: implications for the mechanism of GTP hydrolysis. EMBO J 1990 9 2351 2359 2196171
Resh MD Fatty acylation of proteins: new insights into membrane targeting of myristoylated and palmitoylated proteins. Biochim Biophys Acta 1999 1451 1 16 10446384 10.1016/S0167-4889(99)00075-0
Maurer-Stroh S Gouda M Novatchkova M Schleiffer A Schneider G Sirota FL Wildpaner M Hayashi N Eisenhaber F MYRbase: analysis of genome-wide glycine myristoylation enlarges the functional spectrum of eukaryotic myristoylated proteins. Genome Biol 2004 5 R21 15003124 10.1186/gb-2004-5-3-r21
Ling J Pi W Bollag R Zeng S Keskintepe M Saliman H Krantz S Whitney B Tuan D The solitary long terminal repeats of ERV-9 endogenous retrovirus are conserved during primate evolution and possess enhancer activities in embryonic and hematopoietic cells. J Virol 2002 76 2410 2423 11836419 10.1128/jvi.76.5.2410-2423.2002
Singh G Lykke-Andersen J New insights into the formation of active nonsense-mediated decay complexes. Trends Biochem Sci 2003 28 464 466 13678954 10.1016/S0968-0004(03)00176-2
Mishra R Gara SK Mishra S Prakash B Analysis of GTPases carrying hydrophobic amino acid substitutions in lieu of the catalytic glutamine: implications for GTP hydrolysis. Proteins 2005 59 332 338 15726588 10.1002/prot.20413
Delarbre C Jaulin C Kourilsky P Gachelin G Evolution of the major histocompatibility complex: a hundred-fold amplification of MHC class I genes in the African pigmy mouse Nannomys setulosus. Immunogenetics 1992 37 29 38 1428009 10.1007/BF00223542
Mashimo T Glaser P Lucas M Simon-Chazottes D Ceccaldi PE Montagutelli X Despres P Guenet JL Structural and functional genomics and evolutionary relationships in the cluster of genes encoding murine 2',5'-oligoadenylate synthetases. Genomics 2003 82 537 552 14559211 10.1016/S0888-7543(03)00176-9
Trowsdale J Barten R Haude A Stewart CA Beck S Wilson MJ The genomic context of natural killer receptor extended gene families. Immunol Rev 2001 181 20 38 11513141 10.1034/j.1600-065X.2001.1810102.x
Noel L Moores TL van der Biezen EA Parniske M Daniels MJ Parker JE Jones JDG Pronounced intraspecific haplotype divergence at the RPP5 complex disease resistance locus in Arabidopsis. Plant Cell 1999 11 2099 2111 10559437 10.1105/tpc.11.11.2099
Angata T Margulies EH Green ED Varki A Large-scale sequencing of the CD33-related Siglec gene cluster in five mammalian species reveals rapid evolution by multiple mechanisms. Proc Natl Acad Sci USA 2004 101 13251 13256 15331780 10.1073/pnas.0404833101
Leister D Tandem and segmental gene duplication and recombination in the evolution of plant disease resistance gene. Trends Genet 2004 20 116 122 15049302 10.1016/j.tig.2004.01.007
Barten R Torkar M Haude A Trowsdale J Wilson MJ Divergent and convergent evolution of NK-cell receptors. Trends Immunol 2001 22 52 57 11286693 10.1016/S1471-4906(00)01802-0
Nathan C Shiloh MU Reactive oxygen and nitrogen intermediates in the relationship between mammalian hosts and microbial pathogens. Proc Natl Acad Sci USA 2000 97 8841 8848 10922044 10.1073/pnas.97.16.8841
Murray HW Nathan CF Macrophage microbicidal mechanisms in vivo: reactive nitrogen versus oxygen intermediates in the killing of intracellular visceral Leishmania donovani. J Exp Med 1999 189 741 746 9989990 10.1084/jem.189.4.741
Fang FC Antimicrobial reactive oxygen and nitrogen species: concepts and controversies. Nat Rev Microbiol 2004 2 820 832 15378046 10.1038/nrmicro1004
Adams O Besken K Oberdorfer C MacKenzie CR Takikawa O Daubener W Role of indoleamine-2,3-dioxygenase in alpha/beta and gamma interferon-mediated antiviral effects against herpes simplex virus infections. J Virol 2004 78 2632 2636 14963171 10.1128/JVI.78.5.2632-2636.2004
Daubener W MacKenzie CR IFN-gamma activated indoleamine 2,3-dioxygenase activity in human cells is an antiparasitic and an antibacterial effector mechanism. Adv Exp Med Biol 1999 467 517 524 10721095
Prada-Delgado A Carrasco-Marin E Pena-Macarro C Del Cerro-Vadillo E Fresno-Escudero M Leyva-Cobian F Alvarez-Dominguez C Inhibition of Rab5a exchange activity is a key step for Listeria monocytogenes survival. Traffic 2005 6 252 265 15702993 10.1111/j.1600-0854.2005.00265.x
Forbes JR Gros P Divalent-metal transport by NRAMP proteins at the interface of host-pathogen interactions. Trends Microbiol 2001 9 397 403 11514223 10.1016/S0966-842X(01)02098-4
Flo TH Smith KD Sato S Rodriguez DJ Holmes MA Strong RK Akira S Aderem A Lipocalin 2 mediates an innate immune response to bacterial infection by sequestrating iron. Nature 2004 432 917 921 15531878 10.1038/nature03104
Schaible UE Kaufmann SH Iron and microbial infection. Nat Rev Microbiol 2004 2 946 953 15550940 10.1038/nrmicro1046
Ogawa M Yoshimori T Suzuki T Sagara H Mizushima N Sasakawa C Escape of intracellular Shigella from autophagy. Science 2005 307 727 731 15576571 10.1126/science.1106036
Gutierrez MG Master SS Singh SB Taylor GA Colombo MI Deretic V Autophagy is a defense mechanism inhibiting BCG and Mycobacterium tuberculosis survival in infected macrophages. Cell 2004 119 753 766 15607973 10.1016/j.cell.2004.11.038
Nelson DE Virok DP Wood H Roshick C Johnson RM Whitmire WM Crane DD Steele-Mortimer O Kari L McClarty G Caldwell HD Chlamydial IFN-γ immune evasion is linked to host infection tropism. Proc Natl Acad Sci USA 2005 102 10658 10663 16020528 10.1073/pnas.0504198102
Modlin RL Activation of toll-like receptors by microbial lipoproteins: role in host defense. J Allergy Clin Immunol 2001 108 4 Suppl S104 S106 11586275 10.1067/mai.2001.118299
Tian D Traw MB Chen JQ Kreitman M Bergelson J Fitness costs of R-gene-mediated resistance in Arabidopsis thaliana. Nature 2003 423 74 77 12721627 10.1038/nature01588
Burdon JJ Thrall PH The fitness costs to plants of resistance to pathogens. Genome Biol 2003 4 227 12952527 10.1186/gb-2003-4-9-227
Weatherall DJ Miller LH Baruch DI Marsh K Doumbo OK Casals-Pascual C Roberts DJ Malaria and the red cell. Hematology (Am Soc Hematol Educ Program) 2002 35 57 12446418
Schmid-Hempel P Variation in immune defence as a question of evolutionary ecology. Proc R Soc Lond B Biol Sci 2003 270 357 366 10.1098/rspb.2002.2265
Haller O Kochs G Interferon-induced MX proteins: dynamin-like GTPases with antiviral activity. Traffic 2002 3 710 717 12230469 10.1034/j.1600-0854.2002.31003.x
Praefcke GJK McMahon HT The dynamin superfamily: universal membrane tubulation and fission molecules? Nat Rev Mol Cell Biol 2004 5 133 147 15040446 10.1038/nrm1313
Klamp T Boehm U Schenk D Pfeffer K Howard JC A giant GTPase, very large inucible GTPase-1, is inducible by IFNs. J Immunol 2003 171 1255 1265 12874213
Haller O Acklin M Staeheli P Influenza virus resistance of wild mice: wild-type and mutant Mx alleles occur at comparable frequencies. J Interferon Res 1987 7 647 656 3681017
Jin H Yamashita T Ochiai K Haller O Watanabe T Characterization and expression of the Mx1 gene in wild mouse species. Biochem Genet 1998 36 311 322 9919357 10.1023/A:1018741312058
Staeheli P Grob R Meier E Sutcliffe J Haller O Influenza virus-susceptible mice carry Mx genes with a large deletion or a nonsense mutation. Mol Cell Biol 1988 8 4518 4523 2903437
Trede NS Langenau DM Traver D Look AT Zon LI The use of zebrafish to understand immunity. Immunity 2004 20 367 379 15084267 10.1016/S1074-7613(04)00084-6
Christoffels A Koh EG Chia JM Brenner S Aparicio S Venkatesh B Fugu genome analysis provides evidence for a whole-genome duplication early during the evolution of ray-finned fishes. Mol Biol Evol 2004 21 1146 1151 15014147 10.1093/molbev/msh114
Meyer A Van de Peer Y From 2R to 3R: evidence for a fish-specific genome duplication (FSGD). Bioessays 2005 27 937 945 16108068 10.1002/bies.20293
Flajnik MF Du Pasquier L Evolution of innate and adaptive immunity: can we draw a line? Trends Immunol 2004 25 640 644 15530832 10.1016/j.it.2004.10.001
Kimbrell DA Beutler B The evolution and genetics of innate immunity. Nat Rev Genet 2001 2 256 267 11283698 10.1038/35066006
Fujita T Matsushita M Endo Y The lectin-complement pathway: its role in innate immunity and evolution. Immunol Rev 2004 198 185 202 15199963 10.1111/j.0105-2896.2004.0123.x
Daumke O Weyand M Chakrabarti PP Vetter IR Wittinghofer A The GTPase-activating protein Rap1GAP uses a catalytic asparagine. Nature 2004 429 197 201 15141215 10.1038/nature02505
Gilly M Wall R The IRG-47 gene is IFN-gamma induced in B cells and encodes a protein with GTP-binding motifs. J Immunol 1992 148 3275 3281 1578148
EnsembleArchive
The Mouse Genome Informatics
MRC RFCGR: The Fugu Genomics Project
ENSEMBL: Fugu Genome Browser
IMCB Fugu Genome Project
Human Genome Browser Gateway
The Danio rerio Sequencing Project (Sanger)
Mouse-human synteny alignments (Sanger)
Ensembl Genome Browser
Saitou N Nei M The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol Biol Evol 1987 4 406 425 3447015
Kumar S Tamura K Jakobsen IB Nei M MEGA2: molecular evolutionary genetics analysis software. Bioinformatics 2001 17 1244 1245 11751241 10.1093/bioinformatics/17.12.1244
Felsenstein J Confidence limits on phylogenies: an approach using the bootstrap. Evolution 1985 39 783 791
BCM search Launcher
EMBL-EBI clustal-w
BOXSHADE 3.21
Schug J Overton GC TESS: Transcription Element Search Software on the WWW: Technical Report CBIL-TR-1997-1001-V00 1997 Pennsylvania: Computational Biology and Informatics Laboratory, School of Medicine, University of Pennsylvania
Transcription Element Search System
Lenhard B Sandelin A Mendoza L Engstrom P Jareborg N Wasserman WW Identification of conserved regulatory elements by comparative genome analysis. J Biol 2003 2 13 12760745 10.1186/1475-4924-2-13
Tools for Phylogenetic Footprinting Purposes
Li Y Chambers J Pang J Ngo K Peterson PA Leung WP Yang Y Characterization of the mouse proteasome regulator PA28b gene. Immunogenetics 1999 49 149 157 9914329 10.1007/s002510050476
Lafuse WP Brown D Castle L Zwilling BS Cloning and characterization of a novel cDNA that is IFN-gamma-induced in mouse peritoneal macrophages and encodes a putative GTP-binding protein. J Leukoc Biol 1995 57 477 483 7884320
Carlow D Marth J Clark-Lewis I Teh H-S Isolation of a gene encoding a developmentally regulated T cell specific protein with a guanine nucleotide triphosphate-binding motif. J Immunol 1995 154 1724 1734 7836757
Sorace JM Johnson RJ Howard DL Drysdale BE Identification of an endotoxin and IFN-inducible cDNA: possible identification of a novel protein family. J Leukoc Biol 1995 58 477 484 7561525
Taylor GA Jeffers M Largaespada DA Jenkins NA Copeland NG Vande Woude GF Identification of a novel GTPase, the inducible expressed GTPase, that accumulates in responses to interferon γ. J Biol Chem 1996 271 20399 20405 8702776 10.1074/jbc.271.34.20399
|
16277747
|
PMC1297648
|
CC BY
|
2021-01-04 16:05:35
|
no
|
Genome Biol. 2005 Oct 31; 6(11):R92
|
utf-8
|
Genome Biol
| 2,005 |
10.1186/gb-2005-6-11-r92
|
oa_comm
|
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-11-r931627774810.1186/gb-2005-6-11-r93ResearchTranscriptome analysis of antigenic variation in Plasmodium falciparum - var silencing is not dependent on antisense RNA Ralph Stuart A [email protected] Emmanuel [email protected] Denise [email protected] Odile [email protected] Marie-Agnès [email protected] Ghislaine [email protected] Jean-Yves [email protected] Peter H [email protected] Artur [email protected] Institut Pasteur, Unit of Biology of Host-Parasite Interactions, Centre National de la Recherche Scientifique, Unité de Recherche Associée 2581, 25 Rue du Docteur Roux, F-75724 Paris Cedex 15, France2 Institut Pasteur, Plate-Forme 2 - Puces à ADN, 28 Rue du Docteur Roux, F-75724 Paris Cedex 15, France3 Institut Pasteur, Unité d'Immunologie Moléculaire des Parasites, 28 Rue du Docteur Roux, F-75724 Paris Cedex 15, France4 The Walter and Eliza Hall Institute of Medical Research, 1G Royal Parade, Parkville, Melbourne 3050, Victoria, Australia5 Institut Pasteur, Plate-Forme 8 - CNR/Santé Publique, 28 Rue du Docteur Roux, F-75724 Paris Cedex 15, France2005 31 10 2005 6 11 R93 R93 29 4 2005 12 7 2005 21 9 2005 Copyright © 2005 Ralph 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 microarray analysis of Plasmodium falciparum selected to express different var genes suggests that antisense transcripts are not responsible for the transcriptional silencing of non-expressed var genes.
Background
Plasmodium falciparum, the causative agent of the most severe form of malaria, undergoes antigenic variation through successive presentation of a family of antigens on the surface of parasitized erythrocytes. These antigens, known as Plasmodium falciparum erythrocyte membrane protein 1 (PfEMP1) proteins, are subject to a mutually exclusive expression system, and are encoded by the multigene var family. The mechanism whereby inactive var genes are silenced is poorly understood. To investigate transcriptional features of this mechanism, we conducted a microarray analysis of parasites that were selected to express different var genes by adhesion to chondroitin sulfate A (CSA) or CD36.
Results
In addition to oligonucleotides for all predicted protein-coding genes, oligonucleotide probes specific to each known var gene of the FCR3 background were designed and added to the microarray, as well as tiled sense and antisense probes for a subset of var genes. In parasites selected for adhesion to CSA, one full-length var gene (var2csa) was strongly upregulated, as were sense RNA molecules emanating from the 3' end of a limited subset of other var genes. No global relationship between sense and antisense production of var genes was observed, but notably, some var genes had coincident high levels of both antisense and sense transcript.
Conclusion
Mutually exclusive expression of PfEMP1 proteins results from transcriptional silencing of non-expressed var genes. The distribution of steady-state sense and antisense RNA at var loci are not consistent with a silencing mechanism based on antisense silencing of inactive var genes. Silencing of var loci is also associated with altered regulation of genes distal to var loci.
==== Body
Background
Plasmodium falciparum is a parasite belonging to the phylum apicomplexa, a group characterized by intracellular parasitism. A striking feature of apicomplexans' intracellular lifestyle is their ability to modify host cells though export of macromolecules. P. falciparum parasitizes erythrocytes, which it proceeds to alter via the secretion of a large number of proteins. Much of this protein content is represented by the Plasmodium falciparum erythrocyte membrane protein 1 (PfEMP1) molecules, ligands that span the erythrocyte membrane and mediate cytoadhesion to human receptors exposed to circulating parasites. PfEMP1 proteins are encoded by var genes, and field isolates possess approximately 60-70 distinct var genes. Each var gene consists of a large variable 5' exon (around 4-9 kb in length), and a smaller, more conserved 3' exon (around 1 kb in length) that encodes the intracellular portion of the PfEMP1 protein. Individual parasites do not express all PfEMP1 isoforms simultaneously, but rather change from one var to another successively. The adaptive pressure that selects such behavior is controversial, but plausible hypotheses include avoidance of host antibody response, and changes in cytoadherence ligand in response to tissue environment.
Switching of transcription from one var gene to another does not require genetic rearrangements [1,2] (unlike antigenic variation in Trypanosoma brucei), but is instead associated with epigenetic changes [3-5]. Parasites can change from expressing one PfEMP1 molecule to another both in vivo and in vitro. The rate at which parasites switch away from their parental phenotype is difficult to measure, and different methods have resulted in estimates varying from less than 1% per generation in vitro [6], to more than 16% per generation in vivo [7].
The switching of active var genes in vitro means that cloned parasites expressing individual var genes will eventually drift in the absence of immune pressure to heterogeneous populations. This makes it difficult to assess how many var genes are being expressed in individual parasites. However, parasites selected for binding to different host receptors express distinct var genes and such populations have previously been described to transcribe single dominant var genes [2]. Nevertheless many contentious questions remain about how var genes are transcriptionally regulated. Some studies have suggested that mutually exclusive expression is developmentally controlled, with a number of var genes being transcribed in ring-stage parasites, but only a single var transcribed in the later trophozoite stage [2,8]. Other studies suggest that transcription is initiated at a number of var loci, but that only a single var gene produces complete transcripts [9]. Another puzzling phenomenon is the so-called sterile transcripts that are apparently produced from the 3' exon of many var genes [10].
Analysis of the var introns shows that they contain a promoter that is responsible for the sterile transcripts. The same cryptic promoter was also shown to be bi-directionally functional in reporter assays [11], raising the intriguing prospect that antisense transcripts may play a role in var regulation. Antisense transcription has been suggested as a general control mechanism for Plasmodium transcription [12-14], with a global transcription profile indicating an inverse correlation between abundance of sense and the ratio of sense-to-antisense for many loci. Additionally, artificially introduced antisense molecules have been used to specifically downregulate some genes in P. falciparum [15-17]. Widespread antisense transcripts are also believed to be involved in the modulation of gene expression in humans [18], rice [19], and Arabidopsis [20]. Although antisense is commonly seen as a means of downregulating expression of the protein-coding strand, several global transcriptional studies indicate some sense and antisense RNAs are co-regulated, with transcription of both strands up- or downregulated in certain conditions or tissues [21,22].
To address these important outstanding questions concerning regulation of var genes we constructed a customized oligonucleotide array containing sense and antisense probes to all known var genes of the P. falciparum FCR3 strain, in addition to probes to all other predicted protein coding genes of the sequenced 3D7 strain [23]. Individual parasites have approximately 60 var genes, and of these, 36 have been identified so far in FCR3. For a subset of eight var genes, we made tiled probes against both strands, spanning from the 5'UTR to the 3'UTR. Parasites were panned on either CD36 or chondroitin sulfate A (CSA) to select for parasites expressing distinct var genes, then compared at three points through the asexual intraerythrocytic life cycle. We hypothesized that upregulation of a var gene would be accompanied by decreased abundance of complementary antisense, while downregulated var genes would be associated with an increase in corresponding antisense RNA. Instead, we found that no consistent positive or negative correlation existed between abundance of sense and antisense transcripts. Notably, the very strong upregulation of var2csa gene (Genbank: AY372123) in CSA-selected parasites was accompanied with substantially increased abundance of antisense RNA throughout the same gene. These data indicate that antisense RNAs do not control antigenic variation in Plasmodium. We failed to find any evidence for var transcripts that included only the 5' end, and we also show that 3' sterile transcripts are limited to a subset of var genes.
Parasite adhesion phenotypes also correlate with some specific patterns of physiopathology so other non-var genes upregulated in association with specific binding types are of interest. We detected several genes that are differentially transcribed between CSA and CD36 parasites, including mature parasite-infected erythrocyte surface antigen (MESA - known to bind to the erythrocyte membrane cytoskeleton) and other proteins predicted to be exported to the infected erythrocyte.
Results and discussion
Transcriptional changes in var genes
Arrays containing specific var gene probes for the FCR3 strain allowed us to assay steady-state RNA changes between CSA-panned and CD36-panned parasites. Total RNA was harvested from three time points through the parasite life cycle, at 12 hours, 24 hours and 36 hours post invasion. Parasites from these time points are referred to as ring, trophozoites and schizont stage parasites, respectively. Previous analyses have shown that the peak of var transcription is in ring stages [24,25] and this was confirmed by our analysis, with highest total var transcripts present in ring stages (Figure 1) for both FCR3-CSA and FCR3-CD36. A comparison of the two populations revealed that several var genes are expressed in the FCR3-CD36 population, while only one dominant var, known as var2csa (or PFL0030c) is apparent in the FCR3-CSA population. Multiple probes from this gene detected transcripts at an abundance 150 to 200-fold higher in FCR3-CSA than in FCR3-CD36 parasites (Figures 1 and 2). This could reflect the almost total absence of var2csa transcripts in FCR3-CD36 parasites. Peak transcript abundance for this gene was in ring stages, with the fold-difference between populations falling markedly in trophozoite (60 to 80-fold) (Figure 2) and schizont parasites (6 to 10-fold) (Figure 2). Only hybridization ratios and not levels of hybridization are appropriate to consider when interpreting results obtained with this type of glass spotted microarrays. However, the absolute values obtained for each RNA population (we will refer to these as 'apparent absolute transcript levels' or AATLs), also strongly suggest that peak transcript abundance for this gene was in ring stages. Considering all var and non-var genes, var2csa was the most highly upregulated gene found in FCR-CSA relative to FCR3-CD36 and had one of the highest AATLs detected in these parasites (Figure 1). These data are consistent with previous reports that find a correlation between CSA binding and expression of var2csa in different strains [26-28]. Northern analysis of FCR3-CSA and FCR3-CD36 parasites prepared in our laboratory also shows a very high expression of var2csa in CSA binding parasites and none in CD36-binding parasites [29]. Cross-reactive probes directed against var exon 2, which detect most (but not all) var genes detect no other var transcripts in CSA-binding parasites [29]. Additionally, FCR3 parasites with the var2csa gene disrupted can no longer bind to CSA. Although our array covers all currently known var genes for the FCR3 strain, not every var gene has been sequenced. We therefore cannot exclude that another unknown var gene is involved in CSA binding, although evidence from transcription, proteomic, serological and biochemical studies now indicates that upregulation of var2csa is central to CSA binding [26,27].
In addition to the major var2csa transcript, the microarray analysis detected a less pronounced upregulation of a second full-length var transcript in the CSA-binding population - the A4-tres gene. The probes corresponding to this open reading frame (ORF) indicated a 5 to 9-fold upregulation of this gene in FCR3-CSA parasites compared with FCR3-CD36, but the AATL for this gene is still relatively low (Additional data file 1), and varA4-tres transcript is not detected in CSA-panned parasites by Northern blot using cross-reactive var probes [29]. The A4tres protein is unable to mediate CSA binding in var2csa knockout parasites, so it is unclear whether A4tres has a role in CSA binding.
Unlike CSA binding, multiple var genes are known to participate in CD36 interactions [30]. It is therefore unsurprising that several var genes are upregulated in the FCR3-CD36 population (Figure 1, Additional data files 1 and 2). No var gene in this population exhibits the same fold change or the same AATL as the var2csa gene in FCR3-CSA. This suggests that the FCR3-CD36 population is not homogenous, but rather a heterogeneous mix of parasites each expressing one of a select subset of var genes. The molecular basis for CD36 binding is relatively well understood, and the domains responsible for the interaction have been identified in several strains [31-33]. The upregulated var genes in FCR3-CD36 include domains that have been previously demonstrated to encode CD36-binding PfEMP1 proteins (for example, varFCR3S1.2), as well as several poorly characterized var genes (for example, var_clone_70, var_cDNA11).
The current paucity of panning systems for selecting monomorphic populations prevents us from determining if the behavior of the var2csa-expressing parasites is representative of all var types. Both the characterization of additional receptor-ligand interactions and the development of selectable markers in or adjacent to var genes should generate valuable tools to address this in the future.
Antisense RNAs
Global and specific transcriptional profiles of P. falciparum indicate extensive transcription from the antisense strand of many genes [12]. Nuclear run-on assays show that antisense production is highly alpha-amanitin sensitive, implying a dependence on RNA polymerase II activity [14]. As in some other organisms, the distribution of Plasmodium antisense transcripts suggests a role in regulation of sense strands, with abundance of sense and antisense frequently inversely related for given loci [13]. The availability of genes specifically up- or downregulated at the same life stage, and in genetically identical parasites, creates an ideal system to test the importance of antisense RNAs for Plasmodium gene expression. To investigate this mechanism, we designed specific oligonucleotides probes for antisense RNAs derived from all known var genes of the FCR3 strain. For eight of these genes we also printed multiple oligonucleotide probes tiling the sense and antisense strands of eight different var genes (see Additional data file 1). These include var genes strongly upregulated (var2csa), weakly upregulated (varA4tres), downregulated (varFCR3S1.2) or with no change (varITOR29, varITO4A4) in FCR3-CSA relative to FCR3-CD36.
Our data reveal a pattern for var antisense transcripts that is not consistent with direct antisense transcriptional inhibition (Figure 3). For var loci with high upregulation of sense transcript, the corresponding antisense was sometimes downregulated and sometimes upregulated. Similarly, downregulation of some sense transcripts was seen in conjunction with downregulation of complementary antisense but for other var genes was accompanied with upregulation of antisense (Figure 3). It is noteworthy that for the most highly upregulated sense transcripts (for example, the var2csa gene in CSA panned parasites), strong upregulation of antisense was also seen (Figure 2). The abundance of these antisense molecules is comparable with that produced from other genes known to have highly abundant antisense (for example, MSP2 [34]) (Figure 2). For var loci, these antisense RNA molecules were distributed throughout the gene, although their apparent absolute abundance was much more variable than that of the corresponding sense strand. For example, sense probes throughout the var2csa gene detected consistently strong upregulation throughout the ORF, while antisense RNAs were highly upregulated at some positions in the same gene and not at all in other positions (Figure 2). The large changes in both apparent absolute abundance, and in fold change for neighboring probes against antisense, suggests that antisense RNAs may not be large molecules spanning the entire gene, but rather multiple short transcripts initiating and terminating several times within several kb. Although promoter elements in var introns have been described that appear to drive reverse strand transcription (at least on plasmids) [11], the scattered production of antisense RNA that we observe points to weak promoter-like activity dispersed throughout the var genes. Our failure to detect antisense for the var loci that are silenced does not conclusively prove that they cannot play a role in var silencing, but the presence of abundant antisense molecules that coincide with highly transcribed (and translated) mRNA molecules strongly argues against their having a direct role in gene silencing.
Both the interspersed distribution of antisense RNA molecules and their coincident high abundance with a strongly upregulated protein-coding gene are evocative of a non-specific induction that can correspond with activation of a var gene. Our current understanding of var gene activation is that var genes are activated through disassociation from silencing molecules, subsequent local histone modification and decondensation of the local chromatin environment [3-5]. Indeed this has been shown for the var2csa gene itself. Such modifications make the DNA more accessible to initiation factors and to RNA polymerase. This increased accessibility is consistent with the concept of relaxed non-specific transcription from both strands in the surrounding environment. We hypothesize that the production of antisense RNA, at least in the case of var genes, is not a mechanism for silencing the protein coding strand, but is rather a consequence of an open chromatin configuration and greater concentration of transcription factors required for expression of the active var gene (Figure 4). A similar explanation has been advanced for some human loci, where sense and antisense RNAs are co-ordinately regulated [22]. Long transcripts simultaneously produced from both strands are physically implausible, as one polymerase complex would displace the other. This is consistent with our finding that antisense fragments appear to be small, or alternatively, that sense and antisense are produced simultaneously but in different cells.
Full length or incomplete transcripts?
Various studies of var transcription have been able to detect transcripts corresponding to multiple var genes from parasite populations [2,8] or from single cells [35]. Most of these studies have used degenerate primers targeted to the conserved DBL region found at the 5' of most var genes. These results have led to the widespread understanding that transcription initiates at many var genes, but full-length var genes are produced from only one or very few loci [9]. Unfortunately the size of these molecules has never been thoroughly investigated and we find no data in the literature to suggest that these RNA species are in fact prematurely truncated. Indeed where RT-PCR has been used to assay transcription of the 3' end of var genes (across the splice site) multiple transcripts are still detected, even in adhesion-restricted lines [36]. Certainly, sensitive RT-PCR amplifications do produce evidence of multiple var transcripts, but these multiple transcripts are undetectable by Northern analysis. Our data do not support the existence of truncated 5' transcripts resulting from multiple var loci, although it is possible that some transcript exists below the limits of detection. Additionally, our experiments are unable to address whether some transcripts from multiple loci might be produced but very quickly degraded. This is still a possible additional means of var regulation, although the only published nuclear run-on experiments (which can still only partially address this issue) found no evidence of 'leaky' transcription from multiple var loci [2].
Although there are no quantitative data available regarding the existence of truncated transcripts originating at the 5' end of var genes, Northern blots using a probe from the 3' exon do consistently detect abundant RNA, often referred to as 'sterile transcript'. These probes cross react with the large pf60 family of genes and pseudogenes, which are transcribed in late-stage parasites and are approximately 3 kb in length. Other transcripts of around the same size appear to emanate from var introns themselves [10], though it is unknown at which stage these intron-derived fragments are produced. These intron-derived fragments, and perhaps pf60 transcripts too, may be involved in var silencing. Assays conducted with luciferase reporter driven by a var promoter indicated that the presence of a flanking var intron is required for proper silencing [11]. Mutations perturbing the promoter activity within this intronic sequence also disrupt silencing, indicating the sterile transcripts may themselves play a role in var silencing. We investigated the distribution of these var intron-derived transcripts using var genes for which we had probes for exon 1 and exon 2 transcripts. Our data show that transcripts do originate from the var introns, but only for a subset of var genes. For several var genes in the FCR3-CSA parasites, probes throughout exon 1 indicate the gene is silenced, but exon 2 is strongly upregulated. For example, exon 1 of varFCR3S1.2 is downregulated 5 to 25-fold in FCR3-CSA parasites, but exon 2 probes show a 10 to 25-fold upregulation. For other silenced var genes (for example, var2csa in FCR3-CD36 parasites or varFCR3 T11-1 in FCR3-CSA parasites) no sterile transcript is apparent in the same parasites, nor is it upregulated at any of the life-stages sampled. For some loci, intron-derived transcript was most abundant in ring transcripts, while at other loci exon 2 transcript was more abundant in later-stage parasites (Additional data file 1). The confusing overlap and cross hybridization of the var exon 2 transcript with pf60 transcript makes it difficult to clarify the relative abundance of either RNA species by Northern blot.
The absence of sterile transcripts corresponding to some silenced genes indicates that continuous presence of sterile transcript is not an absolute requirement for var silencing. Calderwood and colleagues have speculated that the promoter for sterile transcripts may participate in silencing by acting as a buffer for chromatin spreading [11]. Alternatively, sterile transcripts may flag complementary genomic regions as targets for chromatin condensation. If either of these possibilities is true, the promoter activity might be required to initiate the silencing chromatin state, but not to maintain it. Our discovery that transcripts are produced from the introns of some silenced var genes but not others requires a rethinking of the involvement of sterile transcript in silencing.
The var1csa gene
One var gene that has been implicated in CSA adhesion through serological and binding assays is the var1csa gene [37-39]. Consistent with recent reports [35,40], we find that this gene does not appear to be upregulated at a transcriptional level in CSA-binding parasites. A previous study indicated that this gene is transcribed throughout the erythrocytic life cycle, apparently irrespective of adherence phenotypes [40]. This pattern is confirmed by our data, which show apparently continuous low-level expression of the var1csa gene in both CSA- and CD36-panned populations (Additional data file 1). Our data do not exclude a role for the Var1CSA protein in CSA binding, but they do suggest that the transcription status of var1csa is not in itself indicative of CSA binding.
Steady-state RNA changes in non-var genes
Several non-var genes encoding parasite proteins predicted to be exported to the infected erythrocyte [41] are differentially abundant in our analysis (Additional data file 1). The most dramatic difference is seen for the pfe0040c gene, encoding the mature parasite-infected erythrocyte surface antigen (MESA - also known as PfEMP2). Three independent probes consistently registered 16-24 times greater abundance of this transcript in ring and trophozoite stages of the FCR3-CD36 parasites compared with FCR3-CSA (Figure 1). It is worth noting that MESA seems to be negatively co-regulated with var2csa (mean of Pearson R = -0.87 for a var2csa random sample of 6 of 30 values for each time point with the 6 mesa values available with 10,000 iterations). This was confirmed by Western blot (Figure 5a) and immunofluorescence (Figure 5b) with a monoclonal antibody specific for the MESA protein. Substantially more MESA is present in FCR3-CD36 than in FCR3-CSA parasites. The localization of MESA is unchanged between the two parasite types, with immunofluorescence showing a distribution at the erythrocyte periphery. In both populations, over 95% of mature parasites are positive for MESA by indirect immunofluorescence assay, so differences in transcript abundance are not due to a gene deletion in FCR3-CSA (as can sometimes happen with subtelomerically-located MESA). MESA is known to bind to the erythrocyte membrane skeletal protein 4.1 [42], and is thought to alter host cell membrane stability. However, erythrocytes infected by mutant parasites lacking MESA are able to adhere normally to CD36-presenting cells [43,44], indicating MESA is not required for cytoadhesion, at least in vitro. This does not exclude a role in vivo and the observation of major differences in levels of MESA expression between parasites expressing PfEMP1 with different adhesive properties is intriguing.
Transcripts representing several hypothetical proteins are differentially abundant in FCR3-CSA and FCR3-CD36, and their localization and function deserve further attention. Several possess targeting motifs predicted to direct their export out of the parasite and into the red blood cell (RBC) [41] (notable examples include PFC1080c, PFA0615w and PFD0080c) (Additional data file 1). Other annotated genes that are differentially regulated include the exported RBC protein GARP, and MAEBL, a predicted invasion ligand. The differential expression of genes not involved in cytoadhesion suggests that receptor use may actually trigger other changes that might be more involved in adaptations to tissue environment or local pH. Our data do not reveal any obvious candidates for signaling molecules involved in detection of or reaction to the parasites' external environment.
Conclusion
The past three years have seen an increasing number of global transcriptional analyses of P. falciparum. Experiments have compared transcriptional changes between the vertebrate life stages [23,45], between genetically distinct strains [46,47], and in response to drug treatment [48] or glucose deprivation [49]. Despite high-quality, reproducible data demonstrating that a very high proportion of genes are rigidly and specifically regulated, recent reviews highlight our scant understanding of transcriptional control in Plasmodium [50,51]. Very few transcription factors have been identified, and genetic regulatory elements are not well described. This deficit has suggested to some that gene regulation in Plasmodium is post-transcriptionally controlled, perhaps by antisense-mediated repression [13].
Our analysis of parasite cytoadhesion shows that differences in receptor use are associated with limited specific transcriptional differences for both var and non-var genes. We find no changes in known transcription factors that associate with the observed differences. This is consistent with previous studies, which suggest that var transcription is regulated by histone modification and chromatin condensation. Silencing of var genes was not associated with increased antisense production at silenced loci, but rather, antisense abundance was in some cases coincident with high sense strand transcription. This indicates that var regulation is not mediated by antisense inhibition. Instead, antisense transcription may be a product of relaxation in the local chromatin structure (as reported in [3] and [5]), accompanied by loci moving to pro-transcription nuclear zones that may allow promiscuous conditions for transcription [3]. High-resolution microarrays offer very promising avenues for the investigation of such interactions between chromatin-mediated events and transcriptional regulation. Future studies will reveal DNA regions that are controlled by chromatin remodeling factors by superimposing array transcriptional information over data from 'ChIP-on-chip' analyses that use microarrays of immunoprecipitated chromatin to map specific chromatin features to the genome.
Materials and methods
Parasite culture
FCR3 parasites were cultured using modifications to the method described by Trager and Jensen [52]. Parasites were grown in a gas environment of 5% CO2, 1% O2 and 94% N2. Media was supplemented with 5% v/v human serum and 5% v/v Albumax II (Invitrogen SARL Cergy Pontoise, France).
Panning of infected erythrocytes
P. falciparum strain FCR3 was panned on endothelial cells expressing either CSA (SBEC-17 line) or CD36 (SBEC-CS2 line) as described previously [2]. The resulting populations are hereafter referred to as FCR-CSA and FCR-CD36, respectively. Panning was repeated twice more, and parasites were tested for their ability to bind purified CSA (Sigma) or soluble recombinant CD36 (Affymax Research Institute) immobilized with monoclonal antibody 179 (Affymax Research Institute). After panning, parasites were expanded for 4-6 generations to generate sufficient quantities for analysis. Mature stages were eliminated using 0.3 M alanine in 10 mM HEPES [53]. Parasites were allowed to reinvade and were synchronized with 0.3 M alanine twice with an interval of eight hours to obtain tightly synchronous parasites. Parasites were allowed to reinvade once again, and were harvested at 12 hours, 24 hours and 36 hours post invasion. FCR3-CD36 parasites appeared to have a slight but consistently shorter life cycle than the FCR3-CSA parasites. For this reason, the schizont stage comparison was slightly asynchronous (2-4 h) with the CD36 parasites slightly more mature than the CSA. Subsets of parasites were assayed for their adhesion to CD36 and CSA immediately before and after each harvesting to confirm specificity of binding. Non-specific binding was at the level of the bovine serum albumin negative control for all populations.
Total RNA preparation
Infected erythrocytes were washed in PBS, permeabilized with 0.05% saponin in PBS, washed three times in PBS, and lysed in 10 pellet volumes of Trizol (Gibco) before freezing at -80°C. Total RNA was prepared from thawed samples as per the manufacturer's instructions. RNA quality was assessed with an Agilent 2100 Bioanalyser (Additional data file 4).
Oligonucleotides
The Malaria Oligo Set (Qiagen-Operon), designed by DeRisi [54], containing 7,393 optimized 70-mers corresponding to 4,644 annotated genes and to putative ORFs, was completed with 1,477 new oligos we designed using ArrayOligoSelector [54,55]. These new oligonucleotides corresponded to annotated genes in PlasmoDB that lacked oligos in the set, and also, sense and antisense probes to all known var genes of the P. falciparum FCR3 strain; for a subset of var genes, tiled probes were designed against both strands, spanning from the 5'UTR to the 3'UTR.
Microarray spotting, cDNA target labeling hybridization and scanning
Oligonucleotides were resuspended in 3X SSC at 40 μM and printed onto UltraGAPS glass slides (Corning) using a Chipwriter Pro Virtek arrayer (Biorad). After printing, arrays were treated as per the instructions of the slide manufacturer (Corning).
RNA samples (5 μg) were indirectly labeled using Atlas PowerScript Fluorescent Labeling kit (Clontech) with a mixture of random hexamer (pdN6), according to the conditions recommended by the manufacturer, with the following modifications: after reverse-transcription, RNA was digested with RNAse H for 45 minutes at 37°C. cDNAs were coupled with cyanines using Cy3 Mono-Reactive Dye or Cy5 Mono-Reactive Dye (Amersham Bioscience). Fluorescent cDNA was then purified with QIAquick PCR Purification Kit (Qiagen). Target quality and concentration were determined by spectroscopy at 260 nm, 280 nm and 550 nm (Cy3) or 650 nm (Cy5). Cy3 and Cy5 target quantities were normalized at 250 pmol, mixed and thereafter concentrated by Microcon YM-30 (Millipore). Sample volumes were adjusted to 50 μl in 5X SSC, 0.1 mg/ml fragmented Salmon sperm DNA (Sigma), 30% formamide and 0.1% SDS.
Microarrays were pre-hybridized in 5X SSC, 1 mg/ml BSA and 0.1% SDS for 1 hour at 42°C, and then washed by immersion in dH2O for 1 minute, followed by isopropanol and dried by centrifugation for 2 minutes at 1,500 rpm. Fluorescent targets were denatured 3 minutes at 95°C, incubated at RT for 5 minutes prior to hybridization and briefly spun, then loaded onto the array under a LifterSlip (Erie Scientific) and incubated in a humid chamber (Telechem) for 16-18 hours at 42°C. After hybridization, slides were washed twice in 2X SSC and 0.1% SDS at 42°C for 5 minutes, twice in 0.1X SSC and 0.1% SDS at RT for 10 minutes and four times in 0.1X SSC for 1 minute at RT, and then dried by centrifugation at 1,500 rpm for 2 minutes. Arrays were scanned with an Axon 4000a scanner with fixed PMT (PMT = 550 for Cy3 and 650 for Cy5). Data were acquired and analyzed by Genepix Pro 5.0 (Axon Instrument).
Statistical analysis
For each developmental stage, dye swaps with two technical replicates and two biological replicates were performed to compensate dye effect and to assess technical and biological reproducibility, leading to eight hybridized slides. Each biological replicate was analyzed separately using R functions (The R project) and Bioconductor package [56]. After logarithm transformation of ratio of the median of the intensities (without background subtraction) in the two channels, an intensity-dependent normalization was applied to each slide. A Loess curve (locally weighted least squares regression) was fitted to (1/2)log2(Cy5×Cy3) versus log2(Cy5/Cy3) plot (MA plot), where 40% of the data was used to calculate the Loess fit at each point [57]. This curve was used to adjust log2(ratio) for each spot. Empty and flagged spots were excluded from the analysis. A paired Student t test was used to assess differentially expressed spots. After exclusion of the values presenting too much or not enough variation, the common variance was used for all genes to improve the robustness of the test. The raw p values were then corrected using the Bonferroni method with a type I error of 0.05. All log2 ratios are presented as CSA-panned condition over CD36-panned condition. Our data have been submitted to the publicly available ArrayExpress database [58].
Immunofluorescence
FCR3-CSA and -CD36 P. falciparum-infected erythrocytes were taken from asynchronous cultures and processed for indirect immunofluorescence assay as previously described [59]. Slides of air-dried blood films were incubated with the MAb Pf12.8B7.4 [60] for 30 minutes at RT, washed and incubated with Alexa-labeled F(ab') fragment of goat anti-mouse IgG (Molecular Probes) in the same conditions. The nuclei were counterstained with 10 ng/μl DAPI (Molecular Probes). The slides were mounted in 50% glycerol in PBS containing 0.1% p-phenylenediamine (Sigma) as anti-fading. Mouse Mab89 anti-PfHRPI (or PfKAHRP) [61] and guinea pig anti-ATS domain from PfEMP1 (D Mattei, unpublished data) were used as positive controls. Labeled erythrocytes were visualized under UV light in an E800 Nikon Microscope. Images were acquired under identical exposure conditions and processed with Adobe Photoshop 7.0.
Western blot
Total parasite SDS extracts were subjected to 7.5% SDS-PAGE and were transferred onto nitrocellulose. Membranes were incubated with MAb Pf12.8B7.4 [60] and processed for chemiluminescence detection according to the manufacturer (SuperSignal West Pico Chemiluminescent Substrate, Pierce). Mab1C11 anti-PfHsp70 was used as control [62]. Pre-stained molecular weight markers were obtained from BioRad.
Additional data files
The following additional data are included with the online version of this article: a table showing normalized array data for all FCR3 and 3D7 sense and antisense oligos included in the analysis, with data from 12 hours (ring stage), 24 hours (trophozoite stage) and 36 hours (early schizont stage) timepoints. The table shows data from biological and dye repeats, in addition to dye swap replicates (Additional data file 1); a table with a subset of the microarray expression data showing normalized array data for the oligos corresponding to sense and antisense strands of var genes from 3D7 and FCR3 (Additional data file 2); histograms showing apparent absolute abundance of the varA4tres and varFCR3s1.2 transcript in CD36 (grey) and CSA (white) panned parasites. Different columns show the apparent absolute abundance for oligonucleotides at individual positions along the genes. Left panels show probes corresponding to sense transcript, right panels show probes corresponding to antisense transcripts. Separate histograms show data for ring, trophozoite and schizont stages. Standard deviation is shown. The antisense patterns for both genes show a pattern that is inconsistent with a var silencing role for antisense, with antisense just as high for all life stages in the active population as in the silenced populations. As in other genes, adjacent probes for antisense are much more variable than in the corresponding sense strand, suggesting antisense transcripts are small and interspersed (Additional data file 3); Agilent 2100 bioanalyzer analysis of total RNA used for microarrays. Virtual gel images and electrophereograms are shown for all timepoints for both treatments and replicates (Additional data file 4).
Supplementary Material
Additional data File 1
A table showing normalized array data for all FCR3 and 3D7 sense and antisense oligos included in the analysis, with data from 12 hours (ring stage), 24 hours (trophozoite stage) and 36 hours (early schizont stage) timepoints. The table shows data from biological and dye repeats, in addition to dye swap replicates
Click here for file
Additional data File 2
A table with a subset of the microarray expression data showing normalized array data for the oligos corresponding to sense and antisense strands of var genes from 3D7 and FCR3
Click here for file
Additional data File 3
Histograms showing apparent absolute abundance of the varA4tres and varFCR3s1.2 transcript in CD36 (grey) and CSA (white) panned parasites. Different columns show the apparent absolute abundance for oligonucleotides at individual positions along the genes. Left panels show probes corresponding to sense transcript, right panels show probes corresponding to antisense transcripts. Separate histograms show data for ring, trophozoite and schizont stages. Standard deviation is shown. The antisense patterns for both genes show a pattern that is inconsistent with a var silencing role for antisense, with antisense just as high for all life stages in the active population as in the silenced populations. As in other genes, adjacent probes for antisense are much more variable than in the corresponding sense strand, suggesting antisense transcripts are small and interspersed
Click here for file
Additional data File 4
Agilent 2100 bioanalyzer analysis of total RNA used for microarrays. Virtual gel images and electrophereograms are shown for all timepoints for both treatments and replicates
Click here for file
Acknowledgements
The authors thank Marta Coelho Nunes (Institut Pasteur, Paris, France) for assistance with parasite adhesion assays, Z Bozdech (Nanyang Technological University, Singapore) for his precious help in setting up the microarray platform, and Benoit Gamain (Institut Pasteur, Paris, France) for critical reading of the manuscript. The project was funded by grants from the Délégation Générale pour l'armement (DGA n°22120/DSP/SREAF), the Programme PAL+/Fonds National pour la Science, the Institut Pasteur, the Programme Génopole, and the BioMalPar network of excellence, supported by the European Union Sixth Framework Programme BioMalPar Grant LSHPCT-2004-503578. S.A.R. is supported by an Australian National Health and Medical Research Council C. J. Martin Fellowship (no. 251775).
Figures and Tables
Figure 1 A dominant var gene is upregulated in CSA binding parasites. Plots of log2 ratio of expression (M) against average log intensity (A) for ring, trophozoite and schizont stages for CSA versus CD36 panned parasites. Only statistically differential data giving a Bonferroni corrected p value (alpha = 0.05) have been displayed. This graph excludes probes corresponding to antisense transcripts and oligos to 3D7 var genes (whose orthologs in FCR3 diverge in sequence). Biological replicates were pooled. The plots reveal a single dominant var transcript (var2csa-marked in orange) that is much more abundant in CSA than in CD36-panned parasites at all life stages. Green dots represent all other oligos corresponding to FCR3 var genes. Several var genes are over-represented in CD36 as compared with CSA-panned parasites. Both log2 ratios of expression and apparent average intensities for var genes decrease through the life cycle.
Figure 2 Consistent sense transcript and interspersed antisense transcript in var2csa gene. Histograms showing apparent absolute abundance of both sense and antisense transcript at the var2csa locus in CD36 (grey) and CSA (white) panned parasites. Different columns show the apparent absolute abundance for oligonucleotides at individual positions along the whole var2csa gene. Left panels show probes corresponding to sense transcript, right panels show probes corresponding to antisense transcripts. Separate histograms show data for ring, trophozoite and schizont stages. Standard deviation is shown. No truncated 5' transcript of the var2csa gene is apparent in CD36 panned parasites, suggesting regulation is not controlled by premature termination of transcription. In ring stages, where var2csa transcript is most abundant in CSA parasites, apparent absolute abundance is also increased for antisense transcripts throughout the gene. Unlike sense transcription, apparent absolute abundance for all antisense transcripts varies greatly between adjacent probes, perhaps indicative of multiple short antisense transcripts initiating throughout the locus. Abundance of sense and antisense transcript in both populations is also shown for a non-var locus, msp2, for which high antisense transcription has previously been measured [34]. Both steady-state sense and antisense levels for the var2csa locus are comparable with those found at the msp2 locus.
Figure 3 No inverse correlation between sense and antisense ratio changes. Scatter plots of log2 ratio of expression (M) (CSA-panned parasites over CD36-panned) for antisense oligonucleotides against sense oligonucleotides for var genes. Data are shown for ring, trophozoites and schizont stages from biological replicate 1. Oligonucleotides corresponding to var2csa are represented by open triangles and the other var genes from the FCR3 strain are displayed as black dots. Oligonucleotides with the highest log2 ratio of expression in CSA- compared with CD36-panned parasites often correspond to those with the highest corresponding ratios for antisense abundance (upper right datapoints). Similarly, several sense transcripts apparently highly upregulated in CD36 correspond to upregulated antisense oligos at the same loci (lower left datapoints). These data are not consistent with a direct transcriptional silencing role for antisense transcription.
Figure 4 A hypothetical model for antisense transcription from var loci. Sense and antisense RNA at several var loci appear to be coordinately regulated. This may result from the altered chromatin state of the encoding genomic DNA, which is differentially modified between silent and active var loci [3]. Silencing factors such as the SIR complex (indicated by blue spheres) bind to inactive var genes, maintaining the chromatin in a condensed state. In the absence of SIR, the active var assumes a relaxed chromatin conformation that makes the surrounding locus competent for transcription. While a stable transcription complex with appropriate assembly of elongation factors generates abundant sense mRNA of full length, transcription from the opposite strand initiates and quickly terminates to produce fragments of antisense. Simultaneous transcription of the same bases from opposite directions is unviable, but in a population, both transcription events may occur at the same time. A chromatin barrier located in the intron [11] may maintain the first exon in a silencing conformation while allowing relaxation of the second exon, leading to partial 3' transcripts from a subset of otherwise silenced var genes.
Figure 5 MESA overexpression in CD36 parasites. (a) Western blot of non-synchronized parasites from FCR3-CD36 and FCR3-CSA parasites. PfHsp70 protein is included as a loading control. A monoclonal antibody (Pf12.8B7.4) against MESA [60] detects approximately 2-4 times more protein in CD36 compared with CSA panned parasites. (b) Immunofluorescence for MESA protein in FCR3-CD36 and FCR3-CSA parasites. The 488-labeled secondary shows that MESA is considerably more abundant in CD36-compared with CSA-panned parasites. The intracellular distribution of MESA is the same in both parasite populations - with most labeling localizing to the periphery of infected erythrocytes.
==== Refs
Smith JD Chitnis CE Craig AG Roberts DJ Hudson-Taylor DE Peterson DS Pinches R Newbold CI Miller LH Switches in expression of Plasmodium falciparum var genes correlate with changes in antigenic and cytoadherent phenotypes of infected erythrocytes. Cell 1995 82 101 110 7606775 10.1016/0092-8674(95)90056-X
Scherf A Hernandez-Rivas R Buffet P Bottius E Benatar C Pouvelle B Gysin J Lanzer M Antigenic variation in malaria: in situ switching, relaxed and mutually exclusive transcription of var genes during intra-erythrocytic development in Plasmodium falciparum. EMBO J 1998 17 5418 5426 9736619 10.1093/emboj/17.18.5418
Freitas-Junior LH Hernandez-Rivas R Ralph SA Montiel-Condado D Ruvalcaba-Salazar OK Rojas-Meza AP Mancio-Silva L Leal-Silvestre RJ Gontijo AM Shorte S Scherf A Telomeric heterochromatin propagation and histone acetylation control mutually exclusive expression of antigenic variation genes in malaria parasites. Cell 2005 121 25 36 15820676 10.1016/j.cell.2005.01.037
Ralph SA Scheidig-Benatar C Scherf A Antigenic variation in Plasmodium falciparum is associated with movement of var loci between subnuclear locations. Proc Natl Acad Sci USA 2005 102 5414 5419 15797990 10.1073/pnas.0408883102
Duraisingh MT Voss TS Marty AJ Duffy MF Good RT Thompson JK Freitas-Junior LH Scherf A Crabb BS Cowman AF Heterochromatin silencing and locus repositioning linked to regulation of virulence genes in Plasmodium falciparum. Cell 2005 121 13 24 15820675 10.1016/j.cell.2005.01.036
Horrocks P Pinches R Christodoulou Z Kyes SA Newbold CI Variable var transition rates underlie antigenic variation in malaria. Proc Natl Acad Sci USA 2004 101 11129 11134 15256597 10.1073/pnas.0402347101
Peters J Fowler E Gatton M Chen N Saul A Cheng Q High diversity and rapid changeover of expressed var genes during the acute phase of Plasmodium falciparum infections in human volunteers. Proc Natl Acad Sci USA 2002 99 10689 10694 12142467 10.1073/pnas.162349899
Chen Q Fernandez V Sundstrom A Schlichtherle M Datta S Hagblom P Wahlgren M Developmental selection of var gene expression in Nature 1998 394 392 395 9690477 10.1038/28660
Taylor HM Kyes SA Harris D Kriek N Newbold CI A study of var gene transcription in vitro using universal var gene primers. Mol Biochem Parasitol 2000 105 13 23 10613695 10.1016/S0166-6851(99)00159-0
Su XZ Heatwole VM Wertheimer SP Guinet F Herrfeldt JA Peterson DS Ravetch JA Wellems TE The large diverse gene family var encodes proteins involved in cytoadherence and antigenic variation of Plasmodium falciparum-infected erythrocytes. Cell 1995 82 89 100 7606788 10.1016/0092-8674(95)90055-1
Calderwood MS Gannoun-Zaki L Wellems TE Deitsch KW Plasmodium falciparum var genes are regulated by two regions with separate promoters, one upstream of the coding region and a second within the intron. J Biol Chem 2003 278 34125 34132 12832422 10.1074/jbc.M213065200
Patankar S Munasinghe A Shoaibi A Cummings LM Wirth DF Serial analysis of gene expression in Plasmodium falciparum reveals the global expression profile of erythrocytic stages and the presence of anti-sense transcripts in the malarial parasite. Mol Biol Cell 2001 12 3114 3125 11598196
Gunasekera AM Patankar S Schug J Eisen G Kissinger J Roos D Wirth DF Widespread distribution of antisense transcripts in the Plasmodium falciparum genome. Mol Biochem Parasitol 2004 136 35 42 15138065 10.1016/j.molbiopara.2004.02.007
Militello KT Patel V Chessler AD Fisher JK Kasper JM Gunasekera A Wirth DF RNA polymerase II synthesizes antisense RNA in Plasmodium falciparum. RNA 2005 11 365 370 15703443 10.1261/rna.7940705
Wanidworanun C Nagel RL Shear HL Antisense oligonucleotides targeting malarial aldolase inhibit the asexual erythrocytic stages of Plasmodium falciparum. Mol Biochem Parasitol 1999 102 91 101 10477179 10.1016/S0166-6851(99)00087-0
Gardiner DL Holt DC Thomas EA Kemp DJ Trenholme KR Inhibition of Plasmodium falciparum clag9 gene function by antisense RNA. Mol Biochem Parasitol 2000 110 33 41 10989143 10.1016/S0166-6851(00)00254-1
Noonpakdee W Pothikasikorn J Nimitsantiwong W Wilairat P Inhibition of Plasmodium falciparum proliferation in vitro by antisense oligodeoxynucleotides against malarial topoisomerase II. Biochem Biophys Res Commun 2003 302 659 664 12646219 10.1016/S0006-291X(03)00246-8
Yelin R Dahary D Sorek R Levanon EY Goldstein O Shoshan A Diber A Biton S Tamir Y Khosravi R Widespread occurrence of antisense transcription in the human genome. Nat Biotechnol 2003 21 379 386 12640466 10.1038/nbt808
Osato N Yamada H Satoh K Ooka H Yamamoto M Suzuki K Kawai J Carninci P Ohtomo Y Murakami K Antisense transcripts with rice full-length cDNAs. Genome Biol 2003 5 R5 14709177 10.1186/gb-2003-5-1-r5
Yamada K Lim J Dale JM Chen H Shinn P Palm CJ Southwick AM Wu HC Kim C Nguyen M Empirical analysis of transcriptional activity in the Arabidopsis genome. Science 2003 302 842 846 14593172 10.1126/science.1088305
Reis EM Nakaya HI Louro R Canavez FC Flatschart AV Almeida GT Egidio CM Paquola AC Machado AA Festa F Antisense intronic non-coding RNA levels correlate to the degree of tumor differentiation in prostate cancer. Oncogene 2004 23 6684 6692 15221013 10.1038/sj.onc.1207880
Cawley S Bekiranov S Ng HH Kapranov P Sekinger EA Kampa D Piccolboni A Sementchenko V Cheng J Williams AJ Unbiased mapping of transcription factor binding sites along human chromosomes 21 and 22 points to widespread regulation of noncoding RNAs. Cell 2004 116 499 509 14980218 10.1016/S0092-8674(04)00127-8
Bozdech Z Llinas M Pulliam BL Wong ED Zhu J DeRisi JL The transcriptome of the intraerythrocytic developmental cycle of Plasmodium falciparum. PLoS Biol 2003 1 E5 12929205 10.1371/journal.pbio.0000005
Smith JD Kyes S Craig AG Fagan T Hudson-Taylor D Miller LH Baruch DI Newbold CI Analysis of adhesive domains from the A4VAR Plasmodium falciparum erythrocyte membrane protein-1 identifies a CD36 binding domain. Mol Biochem Parasitol 1998 97 133 148 9879893 10.1016/S0166-6851(98)00145-5
Kyes S Pinches R Newbold C A simple RNA analysis method shows var and rif multigene family expression patterns in Plasmodium falciparum. Mol Biochem Parasitol 2000 105 311 315 10693754 10.1016/S0166-6851(99)00193-0
Salanti A Staalsoe T Lavstsen T Jensen AT Sowa MP Arnot DE Hviid L Theander TG Selective upregulation of a single distinctly structured var gene in chondroitin sulphate A-adhering Plasmodium falciparum involved in pregnancy-associated malaria. Mol Microbiol 2003 49 179 191 12823820 10.1046/j.1365-2958.2003.03570.x
Salanti A Dahlback M Turner L Nielsen MA Barfod L Magistrado P Jensen AT Lavstsen T Ofori MF Marsh K Evidence for the involvement of VAR2CSA in pregnancy-associated malaria. J Exp Med 2004 200 1197 1203 15520249 10.1084/jem.20041579
Duffy MF Byrne TJ Elliott SR Wilson DW Rogerson SJ Beeson JG Noviyanti R Brown GV Broad analysis reveals a consistent pattern of var gene transcription in Plasmodium falciparum repeatedly selected for a defined adhesion phenotype. Mol Microbiol 2005 56 774 788 15819631 10.1111/j.1365-2958.2005.04577.x
Viebig NK Gamain B Scheidig C Lepolard C Przyborski J Lanzer M Gysin J Scherf A A single member of the Plasmodium falciparum var multigene family determines cytoadhesion to the placental receptor chondroitin sulphate A. EMBO Rep 2005 6 775 781 16025132 10.1038/sj.embor.7400466
Robinson BA Welch TL Smith JD Widespread functional specialization of Plasmodium falciparum erythrocyte membrane protein 1 family members to bind CD36 analysed across a parasite genome. Mol Microbiol 2003 47 1265 1278 12603733 10.1046/j.1365-2958.2003.03378.x
Gamain B Gratepanche S Miller LH Baruch DI Molecular basis for the dichotomy in Plasmodium falciparum adhesion to CD36 and chondroitin sulfate A. Proc Natl Acad Sci USA 2002 99 10020 10024 12096191 10.1073/pnas.152321599
Miller LH Hudson-Taylor D Gamain B Saul AJ Definition of the minimal domain of CIDR1alpha of Plasmodium falciparum PfEMP1 for binding CD36. Mol Biochem Parasitol 2002 120 321 323 11897140 10.1016/S0166-6851(02)00011-7
Baruch DI Ma XC Singh HB Bi X Pasloske BL Howard RJ Identification of a region of PfEMP1 that mediates adherence of Plasmodium falciparum infected erythrocytes to CD36: conserved function with variant sequence. Blood 1997 90 3766 3775 9345064
Kyes S Christodoulou Z Pinches R Newbold C Stage-specific merozoite surface protein 2 antisense transcripts in Plasmodium falciparum. Mol Biochem Parasitol 2002 123 79 83 12165392 10.1016/S0166-6851(02)00135-4
Duffy MF Brown GV Basuki W Krejany EO Noviyanti R Cowman AF Reeder JC Transcription of multiple var genes by individual, trophozoite-stage Plasmodium falciparum cells expressing a chondroitin sulphate A binding phenotype. Mol Microbiol 2002 43 1285 1293 11918813 10.1046/j.1365-2958.2002.02822.x
Noviyanti R Brown GV Wickham ME Duffy MF Cowman AF Reeder JC Multiple var gene transcripts are expressed in Plasmodium falciparum infected erythrocytes selected for adhesion. Mol Biochem Parasitol 2001 114 227 237 11378202 10.1016/S0166-6851(01)00266-3
Costa FT Fusai T Parzy D Sterkers Y Torrentino M Douki JB Traore B Petres S Scherf A Gysin J Immunization with recombinant duffy binding-like-gamma3 induces pan-reactive and adhesion-blocking antibodies against placental chondroitin sulfate A-binding Plasmodium falciparum parasites. J Infect Dis 2003 188 153 164 12825185 10.1086/375800
Lekana Douki JB Traore B Costa FT Fusai T Pouvelle B Sterkers Y Scherf A Gysin J Sequestration of Plasmodium falciparum-infected erythrocytes to chondroitin sulfate A, a receptor for maternal malaria: monoclonal antibodies against the native parasite ligand reveal pan-reactive epitopes in placental isolates. Blood 2002 100 1478 1483 12149234 10.1182/blood-2002-01-0315
Gamain B Smith JD Avril M Baruch DI Scherf A Gysin J Miller LH Identification of a 67-amino-acid region of the Plasmodium falciparum variant surface antigen that binds chondroitin sulphate A and elicits antibodies reactive with the surface of placental isolates. Mol Microbiol 2004 53 445 455 15228526 10.1111/j.1365-2958.2004.04145.x
Kyes SA Christodoulou Z Raza A Horrocks P Pinches R Rowe JA Newbold CI A well-conserved Plasmodium falciparum var gene shows an unusual stage-specific transcript pattern. Mol Microbiol 2003 48 1339 1348 12787360 10.1046/j.1365-2958.2003.03505.x
Marti M Good RT Rug M Knuepfer E Cowman AF Targeting malaria virulence and remodeling proteins to the host erythrocyte. Science 2004 306 1930 1933 15591202 10.1126/science.1102452
Lustigman S Anders RF Brown GV Coppel RL The mature-parasite-infected erythrocyte surface antigen (MESA) of Plasmodium falciparum associates with the erythrocyte membrane skeletal protein, band 4.1. Mol Biochem Parasitol 1990 38 261 270 2183050 10.1016/0166-6851(90)90029-L
Magowan C Coppel RL Lau AO Moronne MM Tchernia G Mohandas N Role of the Plasmodium falciparum mature-parasite-infected erythrocyte surface antigen (MESA/PfEMP-2) in malarial infection of erythrocytes. Blood 1995 86 3196 3204 7579415
Cooke BM Glenister FK Mohandas N Coppel RL Assignment of functional roles to parasite proteins in malaria-infected red blood cells by competitive flow-based adhesion assay. Br J Haematol 2002 117 203 211 11918556 10.1046/j.1365-2141.2002.03404.x
Le Roch KG Zhou Y Blair PL Grainger M Moch JK Haynes JD De La Vega P Holder AA Batalov S Carucci DJ Winzeler EA Discovery of gene function by expression profiling of the malaria parasite life cycle. Science 2003 301 1503 1508 12893887 10.1126/science.1087025
Gissot M Refour P Briquet S Boschet C Coupe S Mazier D Vaquero C Transcriptome of 3D7 and its gametocyte-less derivative F12 Plasmodium falciparum clones during erythrocytic development using a gene-specific microarray assigned to gene regulation, cell cycle and transcription factors. Gene 2004 341 267 277 15474309 10.1016/j.gene.2004.07.004
Daily JP Le Roch KG Sarr O Fang X Zhou Y Ndir O Mboup S Sultan A Winzeler EA Wirth DF In vivo transcriptional profiling of Plasmodium falciparum. Malar J 2004 3 30 15296511 10.1186/1475-2875-3-30
Gunasekera AM Patankar S Schug J Eisen G Wirth DF Drug-induced alterations in gene expression of the asexual blood forms of Plasmodium falciparum. Mol Microbiol 2003 50 1229 1239 14622411 10.1046/j.1365-2958.2003.03787.x
Fang J Zhou H Rathore D Sullivan M Su XZ McCutchan TF Ambient glucose concentration and gene expression in Plasmodium falciparum. Mol Biochem Parasitol 2004 133 125 129 14668019 10.1016/j.molbiopara.2003.09.004
Wilson RJ The transcriptome: malariologists ride the wave. Bioessays 2004 26 339 342 15057931 10.1002/bies.20025
Llinas M DeRisi JL Pernicious plans revealed: Plasmodium falciparum genome wide expression analysis. Curr Opin Microbiol 2004 7 382 387 15358256 10.1016/j.mib.2004.06.014
Trager W Jensen J Human malaria parasites in continuous culture. Science 1976 193 673 675 781840
Ginsburg H Kutner S Krugliak M Cabantchik ZI Characterization of permeation pathways appearing in the host membrane of Plasmodium falciparum infected red blood cells. Mol Biochem Parasitol 1985 14 313 322 3887158 10.1016/0166-6851(85)90059-3
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
ArrayOligoSelector 2003
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
Yang YH Dudoit S Luu P Lin DM Peng V Ngai J Speed TP Normalization for cDNA microarray data: a robust composite method addressing single and multiple slide systematic variation. Nucleic Acids Res 2002 30 e15 11842121 10.1093/nar/30.4.e15
EMBL-EBI: ArrayExpress
Hinterberg K Scherf A Gysin J Toyoshima T Aikawa M Mazie JC da Silva LP Mattei D Plasmodium falciparum: the Pf332 antigen is secreted from the parasite by a brefeldin A-dependent pathway and is translocated to the erythrocyte membrane via the Maurer's clefts. Exp Parasitol 1994 79 279 291 7957750 10.1006/expr.1994.1091
Howard RJ Lyon JA Uni S Saul AJ Aley SB Klotz F Panton LJ Sherwood JA Marsh K Aikawa M Transport of an Mr approximately 300,000 Plasmodium falciparum protein (Pf EMP 2) from the intraerythrocytic asexual parasite to the cytoplasmic face of the host cell membrane. J Cell Biol 1987 104 1269 1280 2437128 10.1083/jcb.104.5.1269
Taylor DW Parra M Chapman GB Stearns ME Rener J Aikawa M Uni S Aley SB Panton LJ Howard RJ Localization of Plasmodium falciparum histidine-rich protein 1 in the erythrocyte skeleton under knobs. Mol Biochem Parasitol 1987 25 165 174 2444884 10.1016/0166-6851(87)90005-3
Blisnick T Lema F Mazie JC Pereira da Silva LP Plasmodium falciparum: analysis of B epitopes of a polypeptide antigen expressed in Escherichia coli, using monoclonal antibodies. Exp Parasitol 1988 67 247 256 2461317 10.1016/0014-4894(88)90072-0
|
16277748
|
PMC1297649
|
CC BY
|
2021-01-04 16:05:35
|
no
|
Genome Biol. 2005 Oct 31; 6(11):R93
|
utf-8
|
Genome Biol
| 2,005 |
10.1186/gb-2005-6-11-r93
|
oa_comm
|
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-11-r941627774910.1186/gb-2005-6-11-r94ResearchGenome-wide gene expression in response to parasitoid attack in Drosophila Wertheim Bregje [email protected] Alex R [email protected] Eugene [email protected] Eric [email protected] Meirion [email protected] Scott D [email protected] Michael R [email protected] Linda [email protected] H Charles J [email protected] Centre for Evolutionary Genomics, Department of Biology, University College London, Darwin Building, Gower Street, London WC1E 6BT, UK2 NERC Centre for Population Biology, Division of Biology, Imperial College London, Silwood Park Campus, Ascot, Berkshire SL5 7PY, UK3 European Bioinformatics Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge CB10 1SD, UK4 Huffington Center on Aging and Molecular and Human Genetics, Baylor College of Medicine, One Baylor Plaza, Houston, TX 77030, USA5 Department of Entomology, 420 Biological Sciences, University of Georgia, Athens, GA 30602-2603, USA2005 31 10 2005 6 11 R94 R94 14 7 2005 20 9 2005 30 9 2005 Copyright © 2005 Wertheim 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.
Expression profiling of the transcriptional response at 9 time points of Drosophila larvae attacked by insect parasites revealed 159 genes that were differentially expressed between parasitized and control larvae. Most genes with altered expression following parasitoid attack had not previously been associated with immune defense.
Background
Parasitoids are insect parasites whose larvae develop in the bodies of other insects. The main immune defense against parasitoids is encapsulation of the foreign body by blood cells, which subsequently often melanize. The capsule sequesters and kills the parasite. The molecular processes involved are still poorly understood, especially compared with insect humoral immunity.
Results
We explored the transcriptional response to parasitoid attack in Drosophila larvae at nine time points following parasitism, hybridizing five biologic replicates per time point to whole-genome microarrays for both parasitized and control larvae. We found significantly different expression profiles for 159 probe sets (representing genes), and we classified them into 16 clusters based on patterns of co-expression. A series of functional annotations were nonrandomly associated with different clusters, including several involving immunity and related functions. We also identified nonrandom associations of transcription factor binding sites for three main regulators of innate immune responses (GATA/srp-like, NF-κB/Rel-like and Stat), as well as a novel putative binding site for an unknown transcription factor. The appearance or absence of candidate genes previously associated with insect immunity in our differentially expressed gene set was surveyed.
Conclusion
Most genes that exhibited altered expression following parasitoid attack differed from those induced during antimicrobial immune responses, and had not previously been associated with defense. Applying bioinformatic techniques contributed toward a description of the encapsulation response as an integrated system, identifying putative regulators of co-expressed and functionally related genes. Genome-wide studies such as ours are a powerful first approach to investigating novel genes involved in invertebrate immunity.
==== Body
Background
Drosophila melanogaster is an important model organism for studying the mechanistic basis and evolution of immunity and pathogen defense. The two main classes of parasites against which it must defend itself in the wild are pathogenic microorganisms (bacteria, viruses, microsporidia and fungi) and parasitoids. Parasitoids are insects whose larvae develop by destructively feeding in (endoparasitoids) or on (ectoparasitoids) the bodies of other insects, eventually killing their hosts. They are ubiquitous in natural and agricultural ecosystems and can have major impacts on the population densities of their host, which makes them a valued agent for biocontrol. Most species that parasitize Drosophila are endoparasitic wasps (Hymenoptera) that attack the larval stage, or are species that feed externally on the pupae but inside the puparium. It is well known that host insects including Drosophila have evolved potent immunologic defense responses against parasitoid attack, and that parasitoids have evolved counter-strategies to subvert host defenses [1]. How these defense and counter-defense responses are regulated is not well understood, however. Here we report a microarray study of the transcriptional response of Drosophila to parasitoid attack. It is the first global expression analysis of the immunologic defense of a host insect against parasitoids, and aims to provide a comprehensive description of the timing and sequence of genes that signal during this innate immune response.
Like most animals, the innate immune response of Drosophila consists of both humoral and cellular defense mechanisms. Humoral defenses against bacterial and fungal infection have been intensely investigated over the past decade and are now relatively well understood [2,3]. These humoral defenses are activated when pathogen recognition molecules detect conserved surface molecules on microorganisms. This in turn activates the Toll and imd signaling pathways, which upregulate expression of antimicrobial peptides and many other genes [4,5]. Homologous signaling pathways regulate antimicrobial defense in other animals including vertebrates [6]. Cellular immune responses such as phagocytosis and nodule formation are also very important in defense against microorganisms [7]. The Janus kinase (JAK)/signal transducer and activator of transcription (STAT) pathway is closely involved in the cellular and humoral responses as well [8].
The chief invertebrate defense against macroparasites such as parasitoids is a cellular immune response called encapsulation (Figure 1) [1]. An encapsulation response begins when blood cells (hemocytes) recognize and bind to the foreign body. Additional hemocytes then adhere to the target and one another, which results in the formation of a capsule comprised of overlapping layers of cells. This response typically begins 4-6 hours after parasitism and is completed by about 48 hours [9]. Capsules often melanize, 24-72 hours after parasitism, and parasitoids are probably killed by asphyxiation or through necrotizing compounds associated with the melanization pathway [10,11].
In Drosophila larvae three types of mature hemocytes are recognized: plasmatocytes, lamellocytes and crystal cells. Plasmatocytes and crystal cells are present in the hemolymph of healthy larvae, whereas lamellocytes are only produced after attack by parasitoids [10-12]. Capsules consist primarily of lamellocytes, although crystal cells and plasmatocytes are present. Crystal cells also release phenoloxidase and possibly other factors that result in melanization of the capsule [13]. After parasitism the numbers of hemocytes increase via proliferation of cells in the hematopoietic organs (lymph glands) and hemocytes already in circulation. However, hematopoietic responses vary with the species of parasitoid and the stage of the host attacked [14-16]. The molecular basis for recognition of parasitoids is unknown, although experiments with mutant stocks implicate a number of signaling pathways (Toll, JAK/STAT and ras/raf/mitogen-activated protein kinase [MAPK]) in hemocyte proliferation and capsule formation [8,17,18].
Parasitoids have evolved several different strategies to overcome host immune responses [1]. Wasps in the genus Asobara (Braconidae) are important parasitoids of larvae of Drosophila, including D. melanogaster. They evade encapsulation by laying eggs that adhere to the fat body and other internal organs of the host [19,20]. This often results in incomplete formation of a capsule, which allows the parasitoid egg to hatch and escape encapsulation [9]. The parasitoid larva then suspends development while its host grows in size and only starts its destructive feeding during the host's pupal period. The growth of parasitized Drosophila larvae is normal until pupariation, irrespective of whether they successfully encapsulate the parasitoid, except that the investment in immune responses may incur slight delays in their speed of development [21,22]. The fraction of D. melanogaster surviving parasitism varies with larval age at the time of attack, temperature, geographic strain and parasitoid species [9,23]. D. melanogaster can also be selected in the laboratory for increased resistance to its parasitoids. For example, five generations of selection for resistance against Asobara tabida increased the frequency of larvae that successfully encapsulated parasitoid eggs from about 5% to about 60% [24,25]. Increased resistance was associated with higher densities of circulating hemocytes, but also reduced larval competitiveness [26]. There are also differences in the degree to which different Drosophila spp. can defend themselves against parasitism, and this too appears to be correlated with hemocyte densities [27].
Previous genome-wide studies of Drosophila immunity all investigated responses against microbial pathogens [28-34]. Defenses against macroparasites such as parasitoids are likely to be very different, and their study, like that of responses to microbial pathogens, may reveal conserved components of the innate immune system. As a first step toward unraveling the genetic control of defenses against parasitoids, we designed a large-scale experiment to monitor the involvement and timing of differentially expressed genes during the entire immune response. We used the Affymetrix Drosophila Genome 1 Array chip (Affymetrix, Santa Clara, CA, USA) to study the transcriptional response of D. melanogaster to attack by A. tabida. Larvae of a Southern European strain of fly that is partially resistant to this parasitoid were exposed to parasitoid attack and then RNA was harvested at nine subsequent time points (from 10 minutes to 72 hours) and compared with RNA from control larvae of the same age. We used bioinformatic techniques to look for patterns of co-expression and for shared regulatory sequences. We also used current knowledge of the molecular basis of defense against parasitoids to identify a set of candidate genes and molecular systems that might be involved in defense against parasitoids, and explored whether they were present in our transcription set.
Comparison with previous studies revealed many differences in gene expression patterns between the antimicrobial and antiparasitoid responses, and implicated several new genes in insect immunity. Clusters of co-expressed genes were identified that we believe may be functional related components of the immune response (for example, a series of serpins and serine-type endopeptidases that may be involved in a proteolytic cascade). We identified a putative transcription factor binding site motif that has not hitherto been linked to any known transcription factor. The transcription factor binding sites of three known regulators of immunity were strongly associated with several clusters of co-expressed genes. Some genes known to be involved in encapsulation were identified in our screens whereas others were not, indicating that they are post-transcriptionally regulated.
Our work increases our understanding of the immunologic defense responses in hosts to parasitoid attack, and paves the way for further experiments to investigate the roles of genes and pathways of particular interest. It suggests a variety of new approaches to understanding the encapsulation process and should help us to move toward a systems level description of innate immunity in insects.
Results
The expression profiles of 159 probe sets differed significantly between parasitized and control larvae. Because we accepted a 1% false discovery rate (see Materials and methods, below), a small number of these probe sets (probably one or two) could have been incorrectly identified. Our assignment of genes to these probe sets, and the functional and structural annotation of these genes are provided in Additional data file 1. Note that some probe sets matched more than one gene (see Materials and methods, below) and some genes are represented by more than one probe set; thus, there are sometimes differences between (sub)totals or percentages calculated for probe sets and genes. Of all the differentially expressed genes, 55% had some information on 'molecular function', 55% on 'biologic process', and 46% on both in the GeneOntology database. For 59 genes (37%) there was no functional annotation in GeneOntology. These percentages did not differ significantly from their equivalents calculated for the full set of genes represented on the Affymetrix Drosophila microarray (P > 0.05, Fisher exact test). Thirty-three genes had GeneOntology annotations that included immunity and defense functions, which, as expected, was significantly more than expected by chance (P < 0.001, EASE analysis). However, more than 80% of the differentially expressed genes had not previously been associated with an immune or defense response in GeneOntology, whereas many known immunity genes were not differentially expressed (Figure 2).
Patterns of co-expression
The pattern of expression of the 159 probe sets that responded to parasitoid attack is shown in Figure 3a. The clustering algorithm sorted the probe sets into a gene tree, from which we defined 16 clusters that varied in size from one to 35 probe sets. Of these clusters, seven contained five or fewer genes, and because of this there is low statistical power to detect over-represented annotation terms. However, 83% of the probe sets were placed in eight clusters that each included more than five genes. The mean expression profile of genes in these clusters, as well as the GeneOntology annotation terms that were significantly over-represented, are shown in Figure 4; the individual gene expression profiles and the full details of the annotation are provided in Additional data files 1 and 2.
In six of these clusters (clusters 1, 2, 4, 11, 12 and 14 in Figure 4; 92 genes in all) the genes tended to have higher expression levels in parasitized than in control larvae, whereas in the remaining two (clusters 9 and 10; 39 genes) the reverse pattern was found. The clustering algorithm uses information from both temporal changes in expression and differences between treatment and control. The clusters with upregulated genes in parasitized larvae fall into a group in which the genes tend to be expressed more strongly for 3-6 hours after parasitism before returning to the same levels as controls (clusters 1, 2 and 4; 32 genes) and one in which the greatest differences occur 6-72 hours after parasitism (clusters 11 and 12; 44 genes), with the genes in the remaining more heterogeneous cluster 14 (16 genes) tending to be differentially expressed at some of the intermediate time points. Of the two clusters of downregulated genes, cluster 10 (21 genes) is largely defined by reduced expression levels in parasitized larvae throughout the course of the experiment, whereas cluster 9 (18 genes) contains genes that are expressed at the end of the experiment, and more strongly in control larvae.
We found highly significant over-representation of annotation terms in four clusters. Half of the genes in cluster 1 (six genes), which were expressed within 1-3 hours of parasitism, are annotated as involved in both immune response and response to bacteria. They included the two antimicrobial peptides AttA and AttB. Cluster 2 (20 genes) had highly significant over-representation of the category immune response (five genes: CG15066, nec, Mtk, hop, dome) and of its parent category defense response (including a further four genes: IM1, IM2, CG13422, CG3066).
Cluster 12 (32 genes) contained a highly significant over-representation of genes for the GeneOntology terms proteolysis and peptidolysis (eight genes) and enzyme regulator activity (seven genes), and the InterPro terms peptidase, trypsin-like serine and cysteine proteases (12 genes), as well as proteins with putative α2-macroglobulin domains (three genes), which may be involved in protease inhibition. These genes are upregulated relative to controls, in particular between 6 and 24 hours after parasitism. Their annotations suggest that they may be involved in a proteolytic cascade that might regulate part of the immune response, such as the formation of the melanized capsule. This hypothesis is supported by the occurrence of clip domains, which enable activation of proteinase zymogens, in several of the serine-type endopeptidases (CG16705, CG11313, CG3505).
Finally, cluster 9 contained a highly significant over-representation of genes with the GeneOntology annotations molting cycle and puparial adhesion (six genes) and the InterPro terms hemocyanin (N-terminal and C-terminal; three genes). This cluster comprises genes expressed at 72 hours after parasitism, by which time the third-instar larva is preparing to pupate; hence, the appearance of genes associated with molting and pupariation is not surprising. What is more interesting is the relatively reduced expression of these genes in parasitized larvae. Even hosts that have successfully been parasitized pupate (the parasitoid emerges from the puparium) and the low expression probably reflects delayed development caused by parasitism. Two of the genes with hemocyanin domains have monophenol mono-oxygenase activity (CG8193, Bc), and the latter of these has been associated with the melanization stage of encapsulation. In our assay, however, the expression profile suggests a closer involvement in pupation than in capsule melanization.
Regulatory sequences
Our analysis identified a set of six putative transcription factor DNA-binding motifs (TFBMs) that were significantly associated with genes in the different clusters. To these we added the STAT motif, which did not quite meet all of our criteria but which is known to be involved in the encapsulation response [8]. The pattern of association of these seven motifs is shown in Figure 3b. Three of the six putative TFBMs matched sequences associated with known transcription factors: serpent and related GATA-factors, Relish and similar nuclear factor-κB (NF-κB) factors, and TATA transcription factors. Both serpent and Relish were previously associated with the Drosophila immune response [35,36] and serpent with hematopoiesis [37].
Table 1 shows in which clusters and at which times the seven TFBMs are most strongly over-represented, and detailed quantitative information is provided in Additional data files 2, 3, 4. We found strong associations between the serpent/GATA-type motifs and the genes in cluster 2, many of which had been annotated as being involved in immunity, and the Relish/NF-κB-type motifs and the genes in cluster 12 associated with proteolysis and peptidolysis. A number of genes that shared the Relish/NF-κB-like binding site motif are all located in a cluster on the 2R chromosome (IM1, IM2, CG15065, CG15066, CG15067, CG15068, CG16836, CG16844, CG18107). The single most significant association, however, was with the motif CCARCAGRCCSA (using IUPAC Ambiguous DNA Characters [38]), which has not hitherto been associated with any transcription factor. It was found to be particularly often associated with genes in clusters 2 and 12, both upstream and in the first 50 base pairs after the start codon.
We tested whether the genes for the transcription factors associated with the TFBMs were themselves upregulated or downregulated after parasitoid attack. The NF-κB-like factor Relish was significantly upregulated 1 hour after parasitism before returning to the same levels as controls. There was no evidence of changed expression for serpent or any of the other GATA-like factors, Stat92E, or TATA factors. Interestingly, serpent/GATA-type motifs were found to be over-represented in clusters 1, 2 and 12 (upregulated genes that tend to be associated with immunity) as well as in clusters 9 and 10 (downregulated genes that tend to be associated with development and metabolism). The lack of differential expression of this transcription factor might thus be explained by it being present in both parasitized and unparasitized larvae but performing different functions.
Candidate genes
We explored whether a variety of genes known to be involved in the response to parasitoid attack had differential patterns of expression. In particular, we looked for genes associated with hemocyte proliferation and differentiation; cellular defense, in particular capsule formation and melanization; and the humoral response to microorganism infection and in regulating coagulation and melanization (Table 2). The gene expression profiles of a selection of candidate genes that were differentially expressed are shown in Figure 5. The expression profiles of all differentially expressed genes are provided in Additional data file 2.
The most dramatic initial response to parasitoid infection involves proliferation of hemocytes and differentiation of lamellocytes in the larval lymph glands, and recent work has shown that this involves the Toll and the JAK/STAT signaling pathways, which are both also implicated in responses to microorganism infection [8,39]. Activation of the Toll pathway in the lymph glands results in hemocyte proliferation, whereas in the fat body it results in the transcription of antimicrobial peptides [39]. Because relatively little is known about this pathway in the lymph glands, we discuss the Toll pathway in relation to its antimicrobial humoral response (see below). The os and Upd-like genes for the ligands that activate the JAK/STAT pathway in flies were not differentially expressed in our assay. The receptor dome and a similar but shortened version of this receptor, CG14225, as well as the Drosophila Jak hop, were all significantly upregulated 2-6 hours after attack. The transcription factor Stat92E (for discussion of the STAT TFBM, see above) is associated with proteins in the Tep and Tot families, whose functions are involved respectively in enzyme regulation and severe stress responses [8]. The genes TepI, TepII, TepIV and TotB were differentially expressed after attack by parasitoids (with the peak of expression later than dome and hop), whereas TotM and TepIII were not. The other Tot genes (including the best characterized TotA [40]) were not represented on the Affymetrix Drosophila Genome 1 Array. The JAK/STAT pathway is also thought to crosstalk with the ras/raf/MAPK pathway during hemocyte proliferation [41], but no genes associated with the latter were significantly affected by parasitism.
The encapsulation process that results in the death of the parasitoid egg involves cell adhesion and melanization [9]. Lectins and integrins are two important classes of protein that mediate cell adhesion in immune responses [42]. The gene lectin-24A was massively upregulated in parasitized larvae 6-48 hours after parasitization at the time when the capsule is formed (10-fold to 16-fold at the peak of expression). Lectins can function as adhesion ligands for invertebrate hemocytes [42]. The gene αPS4, which encodes an α integrin subunit, was upregulated at 48-72 hours, at about the time when the multilayered capsules are completed and melanization occurs [9]. Also at this time, a gene for an immunoglobulin-like protein with haemocyanin domains and predicted monophenol mono-oxygenenase activity (Dox-A3), and a serine-type endopeptidase with predicted monophenol mono-oxygenase activator activity (CG11313) were upregulated. Both are likely to be involved in melanin deposition. Two other genes in our list encode proteins with predicted monophenol mono-oxygenase activity (Bc, CG8193), but the expression profiles of these genes suggested a role in pupation and/or metamorphosis rather than in melanization.
The Drosophila response to microorganism attack involves, among others, the production of antimicrobial peptides controlled by the Toll and imd pathways [43]. The primary receptors are peptidoglycan recognition proteins that bind specifically to different classes of microorganism, and of the 13 genes of this family known in Drosophila two were significantly upregulated (PGRP-LB, PGRP-SB1). However, only a few components of the Toll (Tl and nec) and imd (Rel) pathways were significantly upregulated in response to parasitoid attack. Out of the 14 antimicrobial peptides known in Drosophila, only three (Mtk, AttA and AttB) showed significantly increased expression in parasitized larvae. The first of these acts against filamentous fungi and Gram-positive bacteria, and the latter two against Gram-negative bacteria [44]. All of these genes showed their greatest relative increase in expression soon after parasitism. Parasitoid attack involves wounding and penetration, and it is possible that the production of antimicrobial peptides is associated with damage to the exoskeleton and low-level exposure to microbial factors on the surface of the fly larvae or ovipositor of the wasp.
Discussion
We still have a relatively poor understanding of the genetic mechanisms that underlie host defense to parasitoid attack, despite the immense importance of parasitoids to the population dynamics and control of many insects. A full understanding will require extensive experimental investigation, but we believe that the dataset described here provides a first and important step toward unraveling the genes and pathways involved and their sequence of action.
We investigated the transcriptional profile of D. melanogaster larvae during the 72 hours after they had been parasitized by A. tabida. The Drosophila strain we used was highly immunocompetent and was able to encapsulate about 75% of parasitoid eggs. Furthermore, the counter-resistance strategy of the parasitoid species we used is thought to consist of evasion rather than manipulation of host defenses [19,20]. We were thus able to study a strong and uninterrupted defense response to parasitoid attack. The 72-hour period we studied, which covers the full immune response from detection of the parasitoid egg to completion of the capsule, lasts from the late second instar to just before pupation, which is just over half the length of the host's total larval stage. As expected, a very large number of genes exhibited differences in expression over time (over 8,000 genes with a 1% false discovery rate). A much more restricted set of genes (represented by 159 probe sets) differed significantly in their transcription profiles between the control and parasitized groups. We analyzed patterns of co-expression and shared regulatory motifs within this set of genes, and then asked whether they encoded proteins previously associated with defenses involved in the response to parasitoid attack. The majority of differentially expressed genes in our study had not previously been associated with innate immunity, which is a reflection of the substantial differences in immunologic responses to pathogens and macro-parasites.
Based on our clustering algorithm we identified 16 clusters, the eight largest of which contained 83% of the probe sets. Six included genes (70%) that tended to be more highly expressed in parasitized larvae, whereas two contained genes (30%) that tended to be more highly expressed in nonparasitized larvae. Not all clusters had a clear temporal signature, but we identified groups of genes expressed during the first few hours after parasitoid attack and then later at the time of capsule formation. One cluster contained genes that had reduced levels of expression in parasitized larvae only at the final sampling point, 72 hours after attack. The genes in almost all clusters exhibited significant changes in expression through time in both parasitized and control larvae, which reinforces the importance of having controls of the same age rather than comparing transcription profiles before and after parasitoid attack. Moreover, it indicates that most genes are not exclusively involved in immunity and defense, but also in other processes while the fly larva grows and readies for pupation.
We annotated all the genes in each cluster and then tested statistically for over-represented GeneOntology and InterPro terms. For relatively small clusters of genes, as present here, this procedure does not have very great statistical power, yet we were able to associate potential functions with four clusters: (i) two clusters of genes expressed soon after parasitoid attack were associated with immune functions, (ii) a cluster of genes that were expressed later after parasitism was associated with functions involved in proteolytic cascades, and (iii) the cluster of reduced-expression genes at 72 hrs was associated with preparation for pupation. The first two observations are consistent with an initial "front-line" reaction to parasite challenge, followed by a slower response, perhaps involving the consolidation of the capsule. The last observation is probably a reflection of another consequence of parasitism, a reduction in the rate of development, perhaps a cost of mounting the defensive response [21,22]. At the last sampling point unparasitized larvae were further developed and had begun to express genes associated with pupation.
Our search for potential TFBMs identified six potential sequences, three of which represented already well known transcription factors. The most significant sequence, CCARCAGRCCSA, was not associated with a currently recognized factor, and might represent a new regulatory mechanism involving a novel transcription factor. To screen for such a transcription factor, one could use a yeast 1-hybrid system and protein purification with affinity columns. Interestingly, two clusters of relatively highly expressed genes with significant annotation associations also had strong associations with TFBMs; an immune-related cluster and the possible regulatory-cascade cluster were both significantly associated with serpent/GATA-type motifs, Relish/NF-κB-like motifs, the STAT motif, and the novel sequence just discussed. The transcription factor Rel itself was significantly upregulated immediately after parasitism, but not any of the other transcription factors identified in our screen. These data contribute toward a description of the encapsulation response as an integrated system rather than a simple collection of individual genes.
A number of biochemical systems and signaling pathways are known to be involved in the response to parasitoid attack or the formation of melanotic capsules. The JAK/STAT and Toll pathways have been implicated in regulating hemocyte proliferation. Several components of these signaling pathways, as well as a number of target genes they regulate, exhibited significantly increased expression levels in parasitized larvae compared with controls. We hypothesize that upregulated expression of lectins and integrins, and genes with functions associated with melanin deposition are involved in capsule formation. The Toll and imd pathways have a well known association with microbial defense, and Toll has also been implicated in regulating immune responses toward macro-parasites [18]. Two peptidoglycan recognition proteins and three antimicrobial peptides were significantly upregulated soon after parasitism. Because parasitoid attack involves puncturing the body wall, with the obvious possibility of microbial infection, we suggest that upregulation of these genes reflects low level exposure to microorganisms at parasitoid oviposition. Overall, however, parasitism by A. tabida induced relatively few changes in expression of antimicrobial effector genes under Toll and imd pathway control.
As with other microarray studies, there are limitations to what our work can tell us about the Drosophila response to parasitoid attack. Although the Affymetrix Drosophila Genome 1 Array chip contains a large fraction of Drosophila genes, about 8.5% are missing and so cannot be included in any analysis. More seriously, much of the response to parasitoid attack likely does not involve de novo gene expression but post-transcriptional and translational events. This may be particularly true of any initial, rapid response to parasitoid attack, where any delay in protein synthesis would be maladaptive. Several genes previously implicated in melanization were not differentially expressed, which also indicates the importance of post-transcriptional and post-translational regulation of gene expression. Finally, there is always the danger of false-positives in testing numerous hypotheses simultaneously. Fortunately, because of the large number of microarrays used in this study, we had relatively high statistical power, and we corrected for multiple hypothesis testing using Storey's false discovery rate method. This meant that of the 159 probe sets we identified for further study, we estimate that only one or two are likely to have been erroneously included.
In interpreting our results, two further more specific issues must be considered. First, with the combination of host and parasitoid strains used here, we estimate that about three-quarters of the flies parasitized in the experiment will mount a successful immune response and survive parasitism, but that about a quarter will succumb. Some hosts fail to encapsulate completely the parasitoid egg because it is partially embedded in host tissue. However, parasitized host larvae almost always show some signs of capsule formation and melanization, irrespective of whether they succeed in killing the parasitoid egg (unpublished data). This suggests that much of the transcriptional response to parasitoid attack will be the same in hosts that will or will not survive, although we cannot exclude the possibility that especially some of the later differences in gene expression are pathologic responses to parasitoid attack.
Second, it was not feasible to dissect out the parasitoid eggs from the larvae. Were we to have done this in live larvae, it would have resulted in changes in gene expression due to the major trauma involved, whereas in frozen larvae the eggs become firmly attached to larval host tissue and are very difficult to remove. It is thus possible that there might have been cross-reactivity between parasitoid transcripts and the probe sets on the microarray. However, we think this unlikely, both because the volume of RNA in the parasitoid egg is small compared with that in the host larvae, and because the specificity of the probes means that they are unlikely to cross-react with nucleic acid from an insect as evolutionarily distant as a hymenopterous wasp. The high specificity of the probes was substantiated when we blasted the sequences of the 159 Drosophila probe sets from our study to the genome scaffold of honeybee (Apis melifera, another hymenopteran). Over 75% of the probe sets gave no match at all, and those in the remaining probe sets were very poor (one or two probes per probe set, with at least three errors to the perfect match (PM) sequences).
Microarrays have been used to study the transcription profile of Drosophila adults or cells subject to attack by microbial pathogens. DeGregorio [5,30], Irving [28] and Boutros [32] and their coworkers challenged flies by wounding them with needles dipped in suspensions of bacteria or by shaking them with spores of the pathogenic fungus Beauveria bassiana. Roxström-Linquist and colleagues [31] compared the transcription profiles of adult flies orally infected by bacteria, microsporidia (Octosporea muscaedomesticae) and Drosophila C virus per os, or through shaking them with Beauveria spores. Irving [34] and Johansson [33] and their coworkers recently measured gene expression at 5-6 hours after microbial infection in, respectively, the hemocytes of third-instar larvae and a hemocyte-like cell line of Drosophila. Overall, 43% of the genes in our study appeared in one or more of the lists of genes identified as being involved in immunity in the microbial pathogen studies in adults, and only 8-10% of the genes in our study were also listed as upregulated or downregulated in the studies of cells. The overlap with individual studies was low, ranging from 8% to 32%. The genes that did consistently appear in the antimicrobial studies were predominantly those in the Toll and imd pathways, and some of the serine-type endopeptidases. However, the signaling in the Toll and imd pathways in response to parasitoid attack was atypical compared to the antimicrobial response, with the expression of many intracellular signaling elements and effector genes remaining unaffected. Thus, although there appears to be limited overlap, the innate humoral response to microorganisms and the innate cellular response to macroparasites are substantially different.
Irving and coworkers [34] also explored the transcriptional profile of larval hemocytes from mutant stocks differing in the abundances of plasmatocytes, crystal cells, and lamellocytes. Interestingly, some of the genes we identified as upregulated after parasitoid attack (for example, the integrin αPS4, the monophenol monooxygenase Dox-A3 and the G-protein coupled receptor mthl2) were associated in their study with the presence of lamellocytes, specialized hemocytes that are involved in capsule formation.
Genes involved in immunity against microbial pathogens and parasites have also been studied in genome-wide screens of the mosquito Anopheles gambiae, which is one of the main vectors of the human malaria parasite Plasmodium [45]. The Anopheles genome contains families of immunity genes that are partly orthologous to those in Drosophila [46]. Mosquitoes can mount a melanotic encapsulation response against the ookinete stage of Plasmodium in the insect's gut. This kills the parasite and disrupts the transmission cycle [47,48]. In contrast to the cellular encapsulation response by Drosophila, the melanotic encapsulations of the single-celled malaria parasites by Anopheles do not contain hemocytes and result from a humoral melanization of the ookinete [49,50] Gene silencing studies in the mosquito revealed that two C-type lectins and a leucine rich-repeat immunity protein were pivotal in the melanization response, with the former two averting melanization and the latter inducing it [51]. Parasitoid attack induced strong upregulation of a gene encoding a C-type lectin (lectin-24A) and the slight downregulation of a leucine-rich repeat gene (Pxn).
Compared with our results from Drosophila, there is a greater overlap in Anopheles between genes involved in microbial challenge and parasite infection [45]. A probable explanation for this is the difference between Plasmodium and parasitoids as targets for the immune system. Mosquito immunity against Plasmodium is mostly a noncellular response [49,50], and indeed there is evidence in Anopheles for pattern recognition receptors that both respond to bacteria and Plasmodium [46,52]. In addition, the natural history of the infection is different, with Plasmodium having a variable but relatively minor affect on mosquito fitness [53], whereas the parasitoid is invariably fatal if it is not destroyed. There may also be differences between the defense response of larval and adult insects.
Previous work has shown that there are geographic clines in the degree to which Drosophila melanogaster can defend itself against Asobara tabida [23], and that it is possible to artificially select D. melanogaster for enhanced resistance against this parasitoid [24]. Orr and Irving [54] demonstrated that differences in parasitoid resistance between several field populations were largely restricted to genes on chromosome 2. Three of our clusters with upregulated expression in parasitized larvae contained significantly more genes located on chromosome 2 than expected by chance (for clusters 2, 4 and 12, χ2 with five degrees of freedom: P < 0.05) and, more specifically, a significant over-representation of differentially expressed genes located at chromosomal band 55C (EASE analysis: P < 0.001). Previous studies have suggested the occurrence of two loci in this region that might be related to parasitoid resistance, although the genes at these loci await further characterization [55]. An interesting evolutionary question is whether the differences in resistance, both geographical and before and after selection, are reflected in changes in transcription profile, and whether the genes involved are the same as those identified in the present study. Much evolutionary theory of host-parasite interactions predicts complex dynamics of alleles at loci involved in host defense, but has proved hard to test in the absence of firm information about the genes involved. Microarray studies offer a valuable tool for identifying these genes and making progress on this question.
Drosophila are attacked by several groups of parasitoids in addition to A. tabida and its relatives. In particular, parasitoids in the genus Leptopilina (Figitidae = Eucoilidae) have widespread distributions and can cause high levels of mortality in field populations of Drosophila [56-58]. Leptopilina boulardi is more specialized than A. tabida and exclusively parasitizes species of the melanogaster group. Artificial selection experiments showed that enhanced resistance to L. boulardi (increasing from about 0.5% to about 45% over five generations) also confers better resistance to A. tabida but not vice versa [59]. Leptopilina spp. have evolved a very different strategy to overcome the host immune response compared with that of A. tabida. At oviposition virus-like particles from the long (or venom) gland are injected into the host, and disrupt the immune system by altering hemocyte function [15,60,61]. Comparative microarray studies of flies exposed to the two parasitoids might help to explain the asymmetric cross-resistance and may also tell us whether the apparently very different counter-resistance mechanisms of Asobara and Leptopilina are reflected in different responses to parasitism by the host. Comparative microarray studies may also help to explain the curious observation that some species of Drosophila (D. subobscura is the best known example) never mount a defense response against a parasitoid egg, despite suffering high levels of attack in the field [62]. Finally, the strong selection pressure found in parasitoid-host interactions, in which one of the two participants invariably perishes, has resulted in a wide diversity of defense and counter-defense strategies in different species [1]. Comparative gene expression profiling of different parasitoid-host systems may help to reveal the unique and shared processes that underlie these defense and counter-defense strategies.
Conclusion
We believe that this is the first genome-wide study of the immune response of a host insect to attack by a parasitoid. Our study is relatively unusual in that we used 90 microarrays to produce a highly replicated and densely sampled time series in order to study the events that follow parasitoid attack. In Figure 6 we summarize our results and compare the expression profiles, functional annotations, and transcription factor binding motifs of the major gene clusters we identified. Different groups of co-expressed genes are associated with distinct phases of the response to parasitism identified by morphologic and previous molecular studies. We believe that further investigation of the genes identified here will help us to understand invertebrate cellular defense. Most genes whose expression changed in response to parasitoid attack differed from those induced during the antimicrobial immune response, and had not previously been associated with immunity and defense functions. We applied a combination of bioinformatic techniques to analyze our data, which contributed toward a description of the encapsulation response as an integrated system, identifying putative regulators of co-expressed and functionally related genes.
Parasitoids are major sources of mortality for Drosophila as well as many other types of insects. They are also of significant economic importance as biocontrol agents, and largely because of this the physiology of defense against parasitoids has been intensively studied for many years. Genome-wide expression studies such as ours provide a uniquely powerful approach to investigating new genes involved in invertebrate immunity and will complement these earlier approaches. Much current molecular work on insect immunity has concentrated on the humoral response to microorganisms, and our molecular understanding of cellular immunity is not as well developed. Improving the latter is important if we are to achieve a more balanced appreciation of how insects defend themselves from pathogens and parasites. Invertebrates do not have an adaptive immune system, as in vertebrates, but elements of the innate immune system are strongly conserved across the two groups of animals [6,63,64]. This is clearly so for the humoral immune response, but recent work has revealed unexpected homologies involving components of cellular innate immunity [65,66]. A better understanding of cellular defense in Drosophila thus may also be useful in the investigation of topics such as vertebrate lymphopoiesis and hematopoiesis.
Materials and methods
Insect strains
Drosophila melanogaster used in the study were collected in Avigliano, Italy, in July 2001, and were subsequently cultured in the laboratory on yeast-sugar Drosophila medium [25], at 20°C under a 16:8 light:dark cycle. The parasitoid strain was originally collected in Sospel, France and had been maintained in the laboratory for over 20 years on D. subobscura. On average, 73% of Sospel parasitoid eggs were successfully encapsulated by our experimental strain of fly.
Collection of parasitized and control hosts
A single parasitoid was observed searching for 30 host larvae in a patch of yeast placed on an agar base in a Petri dish. The host larvae were in their late-second instar, and the parasitoid had had experience of oviposition during the previous 24 hours. When a larva was seen to be parasitized, it was transferred to a fresh Petri dish, where it was allowed to develop at 20°C for a fixed period of time before harvesting for RNA extraction. Ten parasitized larvae were collected per female, and larvae attacked within a short time frame (within 1-30 minutes, depending on the time point that was being collected) were reared together in the same dish. Larvae attacked by the parasitoid but rejected (defined by the ovipositor inserted for <10 s [67]) were not used in the study. We collected larvae at nine different times after parasitism: 10-15 minutes, 1 hour, 2 hours, 3 hours, 6 hours, 12 hours, 24 hours, 48 hours and 72 hours. To control for handling trauma, any variation in developmental stage across replicates, and the effect of the circadian rhythm on gene expression, a second pair of Petri dishes was set up in parallel, and the larvae treated identically except that they were not exposed to the parasitoid. At harvest, larvae were carefully teased from the medium with a spatula, snap-frozen in liquid nitrogen, and then stored at -80°C until RNA extraction. Sample collection for the study took 7 weeks.
RNA isolation and array hybridizations
Microarray hybridizations (Affymetrix Drosophila Genome 1 Array) were performed for five biologic replicates per time point for both parasitized and control larvae (90 chips used in total). Because of circadian patterns in gene expression and possible changes in experimental conditions over the 7 weeks, the RNA used for each hybridization was pooled from flies harvested at different times of day and from over the complete collection period. In preparing samples, the paired sets of control and parasitized larvae continued to be handled together. To avoid large differences in RNA concentrations in the sample pools, the number of fly larvae used per biologic replicate depended on their age (less than 12 hours post-parasitism, 120 larvae; 12 hours, 100 larvae; 24 hours, 50 larvae; 48 and 72 hours, 30 larvae).
Preparation of material for the microarray analysis largely followed the Affymetrix manual. Briefly, samples were homogenized in 1 ml Trizol in FastPrep tubes (Lysing Matrix D; Q-Biogene, Morgan Irvine, CA, USA) using a bead mill (Hybaid RiboLyser; Hybaid, Teddington, UK). Total RNA was isolated using Trizol reagent (Invitrogen, Carlsbad, CA, USA) and the RNeasy (Qiagen, Hilden Germany) kit, following the manufacturers' instructions. For the RNA precipitation step in the Trizol protocol, 700 μl 70% diethyl pyrocarbonate-treated H2O-ethanol was used, and this volume was then applied directly onto RNeasy mini columns. The RNeasy protocol was then followed from the RW1 wash step onward. For each sample, double-stranded cDNA was synthesized from 20 μg total RNA using a commercially available kit (Roche Biochemicals, Basel, Switzerland). Biotin-labeled cRNA was then transcribed using T7 RNA polymerase and the BioArray Transcript labelling kit (Enzo, Farmingdale, NY, USA), followed by probe hydrolysis in 5 μl buffer (200 mmol/l Tris-acetate, pH 8.1, 500 mmol/l KOAc, 150 mmol/l MgOAc). The quality of total RNA and cRNA, and the fragmentation were checked using an Agilent Bio-analyzer (Agilent Technology, Palo Alto, CA, USA). The fragmented cRNA samples were stained, hybridized, and scanned by the Affymetrix microarray service at MRC Geneservices (Hinxton, UK).
Microarray analysis
Initial manipulation of the raw intensity data from the hybridizations was performed using the 'affy package' [68] of the Bioconductor Project [69,70]. An estimate of the logarithmically transformed expression level of each gene based on the intensity of the different probe sets was obtained using the RMA method (robust multi-array analysis [71]) with standard settings (for example quantile normalization and calculation of expressions levels using median polish).
We analyzed gene expression levels using the R statistical package [72]. For each of the 14,010 probe sets on the Affymetrix chip, we had 90 data points representing five replicate measurements of expression levels in (paired sets of) control and parasitized larvae at each of nine time points (after parasitism). We knew that expression levels would vary with time because host larvae molted from the second to the third instar and initiated metamorphosis during the 72 hours of study. To detect effects of parasitism, we therefore carried out a mixed-model analysis of variance for each gene by first fitting a nine-level fixed 'time factor' and a random 'pair factor', and then testing for significance by adding the nested treatment × time interaction. This nested interaction term allowed us to test whether variation in expression values could be attributed to treatment (that is, attack by a parasitoid) across all time points or during a subset of time points. Analysis of variance makes specific assumptions about the distribution of the statistical error terms, and we confirmed that this method was appropriate by checking the form of residual plots of all genes with a significant treatment interaction effect and, for a subset of 25 genes, by repeating the analysis using an empirical F distribution constructed using random permutation [73]. Because we were conducting a large number of statistical tests, we could not rely on simple P values as a measure of statistical significance. Instead we used the positive false discovery rate method of Storey [74] and Storey and Tibshirani [75] and identified a set of significant genes while accepting a rate of false positives of 1%.
Within the set of genes that exhibited a significant response to parasitoid attack, we identified subsets with common patterns of expression using a clustering algorithm based on Pearson correlation coefficients and implemented in GeneSpring (version 6.2; Silicon Genetics, now acquired by Agilent Technology). Greater weights were assigned to later time points and to parasitized samples, which is where the largest variation in expression patterns was observed. The threshold for defining clusters was initially chosen by eye, although we checked that the clusters were reasonably robust by varying the parameters of the clustering algorithm.
The complete set of raw and normalized microarray data from this study is accessible through the public repository ArrayExpress at the European Bioinformatics Institute (accession number E-MAXD-6) [76]. Data produced during this project is also catalogued in EnvBase (accession number egcat:000031) [77]. The normalized data of the probe sets that exhibited a significant response to parasitoid attack are provided in Additional data file 5.
Bioinformatics
We used bioinformatic tools to annotate the probe sets with significantly different expression profiles in parasitized and control larvae, and to look for patterns indicating functional relations and co-regulation in the major clusters of co-expressed genes.
The probes on Affymetrix microarray chips are arranged in probe sets, and we first associated these with the genes listed in the Drosophila genome project (FlyBase [78]), which we accessed through the Ensembl project web portal (version 16.3 [79]). Every individual probe sequence (usually 14 per probe set) was aligned against all available transcript sequences and matches (allowing for one error) recorded. Cases in which four probes from a probe set matched more than one gene, and those in which fewer than 10 probes matched the same gene were excluded, which meant that some probe sets remained unannotated. For 22% of the probe sets whose expression was influenced by parasitism, the probe set matched more than one transcript sequence, and in these cases annotation information from all peptides is provided. In our analyses, we used information on molecular function and biologic process from GeneOntology (September 2004 annotation [80]), and protein families, domains and functional sites from InterPro (version 7.1 [81]).
To determine whether sets of co-expressed genes identified using the clustering algorithm shared structural or functional traits, we asked whether genes in a cluster shared a particular annotation more often than expected by chance (using the program EASE [82]). EASE calculates the exact probability of randomly sampling a given number of genes with any particular annotation in relation to the total number of genes with this functional or structural annotation on the gene chip. Thus, it searches for annotations or 'biologic themes' that are statistically enriched in a group of genes as compared with the whole genome. Using EASE annotations for each probe set, the one-tailed Fisher exact probabilities and Bonferroni corrections were used to determine which particular annotation categories were over-represented.
We explored whether genes in the same cluster shared upstream motifs, including TFBMs, which might indicate coordinated expression. To do this we used the program MotifRegressor [83,84] to define a set of candidate motifs in the -1,000 to +50 base pair region of the differentially expressed genes after parasitoid attack, and then used the program Clover [85,86] to test for significant over-representation of these motifs in co-expressed genes. To define the set of candidate motifs information from individual time points were analyzed separately, and from each we retained the 20 top motifs that the program MotifRegressor identified using a regression strategy based on differential gene expression and the number and strength of match of the motif. Basically, the program searches for any sequence that is significantly associated with upregulated (or downregulated) genes. To test for over-representation of these motifs in gene clusters, the program Clover generates a score for each motif/cluster combination based on data on presence and strength of association. Initial screening identified more than 100 candidate motifs at the different time points. However, this list was reduced to six in a two-step approach: first, by merging degenerate motifs that aligned at more than half of the DNA bases per sequence, using IUPAC Ambiguous DNA Characters [38] to designate more than one DNA base at a given position within a sequence; and second, by requiring that motifs should be over-represented at multiple time points. Matches to known binding sites were identified using the TFBM databases Transfac 8.1 [87] and Jaspar [88,89], and from the list of immunity-related TFBMs presented by Senger and coworkers [36]. The significance of the associations was tested using scores generated from the -1,000 to +50 base pair regions of 1,000 randomly selected genes present on the Affymetrix Drosophila Genome 1 Array. Only clusters with more than five genes were included in the analysis.
Additional data files
The following additional data are included with the online version of this article: A table annotating the probe sets with significantly different expression profiles in parasitized and control larvae (Additional data file 1); a figure showing the expression profile and upstream motifs of all genes per cluster (Additional data file 2); a table providing a full list of putative regulatory motifs that were significantly over-represented in our clusters of genes (Additional data file 3); a diagrammatic representation of the degenerate motifs of the putative TFBMs (Additional data file 4); and a table providing normalized data for the probe sets with significantly differential expression profiles in parasitized and control larvae (Additional data file 5).
Supplementary Material
Additional data file 1
Annotation of probe sets with significantly different expression profiles in parasitized and control larvae. The annotation includes a matching score describing the fraction of the 14 probes that matched perfectly to the annotated transcript sequence, the gene name(s) assigned to probe sets, and functional and structural annotations for each gene. Whenever more than one gene was assigned to a probe set, an asterisk indicates which annotation was used in the EASE analysis.
Click here for file
Additional data file 2
Expression profile and upstream motifs of all genes per cluster. The log2 expression values at the time points (hours) after parasitoid attack are shown for all replicates (blue circles for the control larvae; red triangles for the parasitized larvae) and the lines denote the average expression at each time point. The strength of match for the putative regulatory motifs in the upstream sequences is indicated by the height of the bars.
Click here for file
Additional data file 3
A full list of putative regulatory motifs that were significantly over-represented in our clusters of genes. Motifs were identified by MotifRegressor and over-representation was calculated in Clover. The Representative Motifs denote the degenerate motifs, using IUPAC Ambiguous DNA Characters. The Raw Score measures the strength of match and the frequency of occurrence. Significance (denoted in the last two columns) is based on the comparison with the upstream sequences of 1,000 randomly chosen genes represented on the Affymetrix Drosophila 1 Genome Array, respectively - all genes on Drosophila chromosome 2.
Click here for file
Additional data file 4
A diagrammatic representation of the degenerate motifs of the putative TFBMs. The size of the letters represents the likelihood of its occurrence at each position in the sequence.
Click here for file
Additional data file 5
Normalized data of the probe sets with significantly differential expression profiles in parasitized and control larvae. The expression values (log2 transformed) for the five biologic replicates at each of nine time points after parasitism are provided for the paired sets of parasitized and control larvae.
Click here for file
Acknowledgements
We thank Franco Pennacchio for collecting the Drosophila strain; Martha Kotzen for culturing parasitoids; Kees Hofker and Leiden University for the pictures in Figures 1a–d; Stephen Henderson for advice on statistical permutation; Joe Wood for his help in submitting the data to Array Express; and Stuart Reynolds and two anonymous referees for comments. This research and the authors were funded by NERC Environmental Genomics Programme (B.W., A.R.K. H.C.J.G. and M.H.) and partly by BBSRC (S.D.P. and L.P.) and the Wellcome Trust (E.S., E.B. and L.P.).
Figures and Tables
Figure 1 The Drosophila immune response after attack by parasitoids. (a) The parasitoid Asobara tabida stabs a second instar Drosophila melanogaster larvae with her ovipositor and inserts a single egg. (b) The parasitoid egg is susceptible to nonself recognition by membrane-bound and noncellular pattern recognition proteins in the larval hemolymph. (c) Hemocyte proliferation and differentiation is triggered, and the blood cells aggregate around the parasitoid egg. (d) The hemocytes form a multilayered capsule around the parasitoid egg and melanin is deposited on the capsule. (e) The parasitoid egg dies when it becomes fully melanized.
Figure 2 Venn diagrams of genes that changed expression after parasitoid attack and known immunity genes. The differentially expressed genes after parasitoid attack differed largely from those with a GeneOntology (GO) annotation for immunity or defense (GO database September 2004; the GO codes are also shown in the figure). Some of the probe sets in our set matched to multiple genes (see additional data files), thus reporting on the expression of potentially all of these genes. We included the multiple gene annotations per probe set to define our set of differentially expressed genes for the comparisons.
Figure 3 Gene expression levels and distribution of regulatory motifs for the genes differentially expressed after parasitoid attack. (a) Expression levels for genes (rows) at different sample time points (columns: 1-9 parasitized larvae; 10-18 unparasitized larvae). The expression levels are given as multiples of the median for that gene, using a color code illustrated at top right. At the left the dendrogram produced by the clustering algorithm is shown, with the 16 clusters discussed in the text depicted in different colors (with their code numbers; the final column on the right shows the clusters again using the same color key). (b) The distribution of putative regulatory motifs in the -1,000 to +50 base pair upstream regions of the genes. The colors indicate the number and strength of the matches for each motif (see code on upper right, in which a score of 0 is equivalent to no matches, 10 is equivalent to one strong or two weak matches, and 20 is equivalent to multiple strong matches).
Figure 4 Gene expression profiles and functional annotations for the eight largest clusters of co-expressed genes. On the left-hand side the average expression levels for the genes in the eight clusters are shown (log2-transformed expression values, divided by the median for that gene across all time points and treatments). Dashed lines represent parasitized and unbroken lines represent unparasitized larvae, and the bars indicate standard errors. Functional annotations associated with clusters are shown along the top, and colors in the matrix indicate the strength of association (yellow = Ease scores (see text) <0.05; red = after Bonferroni correction at P < 0.05; grey = at least one gene with this annotation). The full annotation for all probe sets is provided in Additional data file 1.
Figure 5 Expression profiles of genes from pathways and processes known to be involved in immunity. Each graph depicts the log2 expression values for a single gene at different time points (in hours) after parasitoid attack. The blue circles and red triangles show the individual replicates of the control and parasitized larvae, respectively. The lines denote the average expression at each time point. See text for a discussion of the selected genes.
Figure 6 Overview and summary of our findings. The two left-hand columns show the time elapsed since parasitoid attack and a diagrammatic summary of major cellular and metabolic consequences of parasitism. The three right hand columns show the results of this study and the gene clusters that we hypothesize are associated with the different processes sketched on the far left. These three columns show the following: over-represented transcription factor binding motifs arranged by cluster (with code number) ordered by their time of maximum expression; average expression profiles of genes in these clusters (parasitized larvae in red, unparasitized larvae in blue) with marked temporal restricted expression; and functional annotations associated with genes in these clusters, in the same order as in the first of the three columns. A group of genes with relatively constant levels of reduced expression in parasitized larvae is shown separately at the bottom.
Table 1 Putative regulatory motifs that were over-represented in the eight major clusters of differentially expressed genes
Motif Time point (hours) Cluster, raw score and significance†
Relish/NF-κB-like 1, 3, 48 Cluster 1 8.54 P < 0.001
Cluster 2 8.01 P = 0.002
Cluster 11 5.09 P < 0.006
Cluster 12 17.3 P < 0.001
Cluster 14 12.4 P = 0.001
serpent/GATA-like 1, 2, 3, 6, 72 Cluster 1 7.13 P < 0.001
Cluster 2 21.2 P < 0.001
Cluster 9 17.5 P < 0.001
Cluster 10 8.43 P = 0.009
Cluster 12 10.5 P = 0.001
STAT - Cluster 2 4.88 P < 0.001
Cluster 12 4.83 P < 0.001
TATA-like 72 Cluster 1 5.57 P = 0.001
Cluster 9 13.9 P < 0.001
Cluster 10 6.21 P = 0.002
CCARCAGRCCSA 1, 2, 3, 6 Cluster 2 56.1 P < 0.001
Cluster 12 27.8 P < 0.001
Cluster 14 14.3 P = 0.001
CAWTSKATTC 2, 3 Cluster 2 17.5 P < 0.001
Cluster 14 8.39 P = 0.008
AMTCAGT 2, 3, 6, 12, 72 Cluster 2 16.6 P < 0.001
Cluster 12 10.9 P < 0.001
Cluster 14 8.99 P = 0.001
Putative motifs were identified as described in the text. The table shows the motifs identified, the time points at which they were significantly associated, and the clusters in which they appeared. For each cluster we give the raw score (a measure of the average occupancy in a set of sequences) and the associated significance value. †Only for clusters with more than five genes.
Table 2 Survey of candidate genes previously implicated in Drosophila defense and immunity
Functional classification of proteins or genes Differentially expressed candidate gene Cluster number
Hemocyte proliferation and differentiationa
JAK/STAT pathway
Ligands -
Receptors dome (CG14226), CG14225 2
JAK hop (CG1594) 2
STAT -
Possible effector molecules TepI (CG18096), TepII (CG7052), TepIV (CG10363) 12
TotB (CG5609) 8
Toll pathway (in lymph glands)
Ligands -
Regulators of pathway nec (CG1857) 2
Receptors Tl (CG5490) 3
Intracellular signaling elements -
NF-κB transcription factor Relish (CG11992) 4
Ras/Raf/MAPK pathway -
Notch pathway -
VEGF receptor pathway -
GATA factor homologs (e.g. srp) -
RUNX/AML1-like proteins (lz) -
Cellular defense, in particular encapsulationb
Recognition/surface binding factors
Extracellular matrix (ECM) proteins (e.g. laminin, collagen IV, fibronectin) dome (CG14226) 2
prc (CG5700) 14
Hml (CG7002) 10
CG6788/CG32496 11
Integrins αPS4 (CG16827) 11
Immunoglobulin superfamily members Pxn (CG12002) 6
CG8100 10
CG14225 13
Scavenger receptors (CD36-like) CG12789 4
CG2736 10
Tequila (CG4821) 12
Possible pattern recognition receptors lectin-24A (CG3410) 12
G-protein type receptors mthl2 (CG17795) 11
Surface helper molecules
Vinculin, talin, paxillins -
Surface-associated signaling molecules
Integrin-linked focal adhesion kinases (FAKs) -
Integral membrane proteins rost (CG9552) 4
Tsp42Ek (CG12841) 9
Intracellular signaling pathway factors
Phosphotidylinositol 3-kinase (PI3K) -
GTP-binding proteins (Ras/Rho family members) -
Protein kinase C (PKCs) or PKC regulators CG5958 (PKC transporter) 10
Protein tyrosine phosphatase (PTPs) dome (CG14226) 2
Serine/threonine kinases -
Scaffolding proteins (RACK) -
Cytoskeletal proteins (actins, tubulins, for example) αTub85E (CG9476), αTub84D (CG2512), αTub84E (CG1913), βTub60D (CG3401) 11
Eicosinoid pathway elements -
Effector molecules
NO pathway factors -
PPO pathway factors Dox-A3 (CG2952), CG11313, 11
G8193, Bc (CG5779), 9
Fmo-2 (CG3174) 15
Porferins or related molecules -
Tumor necrosis factor (TNFs) CG13559 2
Humoral defenseb
Humoral pattern-recognition receptors PGRP-SB1 (CG9681) 1
lectin-24A (CG3410) 12
Hml (CG7002) 10
Serine proteases CG3066 2
CG30414, CG30086, CG30090, Tequila (CG4821), CG16705, CG31780 / BG:DS07108.1 (CG18477), CG6639, CG3117, CG31827/BG:DS07108.5 (CG18478), CG18563, CG4793, CG4259 12
CG11313 11
CG16713 4
Serpins and other protease inhibitors nec 2
CG6687, CG16712, CG16705, TepI (CG18096), TepII (CG7052), TepIV (CG10363) 12
BcDNA:SD04019 (CG17278) 14
CG16704 1
Known ligand-like molecules (e.g. spz) -
Surface receptors
Toll and associated family members TI (CG5490) 3
Toll or imd pathway (in fatbody)
Intracellular signalling elements (e.g., tube, Pelle, DTRAF, DECSIT) -
NF-κB transcription factor Rel (CG11992) 4
Effector molecules or antimicrobial peptides AttA (CG10146), AttB (CG18372) 1
Mtk (CG8175), IM1 (CG18108), IM2 (CG18106), CG13422, CG15066 2
IM4 (CG15231), CG18279, CG16844 14
Related apoptotic regulators
Dredd -
Ubiquitins -
PPO and associated pathway molecules Dox-A3 (CG2952) 11
Melanin and free-radical intermediates Fmo-2 (CG3174) 15
The table lists the different functional classes of genes and protein surveyed, any genes in these classes that were differentially expressed, and the cluster the gene was assigned to. Note that some genes with multiple annotations can appear in more than one category.
aBased on [17,66,90,91]; bbased on [11,92] (MR Strand, personal communication).
==== Refs
Strand MR Pech LL Immunological basis for compatibility in parasitoid-host relationships. Annu Rev Entomol 1995 40 31 56 7810989 10.1146/annurev.en.40.010195.000335
Hoffmann JA The immune response of Drosophila. Nature 2003 426 33 38 14603309 10.1038/nature02021
Brennan CA Anderson KV Drosophila: the genetics of innate immune recognition and response. Annu Rev Immunol 2004 22 457 483 15032585 10.1146/annurev.immunol.22.012703.104626
Bulet P Hetru C Dimarcq JL Hoffmann D Antibacterial peptides in insects; structure and function. Dev Comp Immunol 1999 23 329 344 10426426 10.1016/S0145-305X(99)00015-4
DeGregorio E Spellman PT Tzou P Rubin GM Lemaitre B The Toll and Imd pathways are the major regulators of the immune response in Drosophila. EMBO J 2002 21 2568 2579 12032070 10.1093/emboj/21.11.2568
Khush RS Lemaitre B Genes that fight infection. What the Drosophila genome says about animal immunity. Trends Genet 2000 16 442 449 11050330 10.1016/S0168-9525(00)02095-3
Braun A Hoffmann JA Meister M Analysis of the Drosophila host defense in domino mutant larvae, which are devoid of hemocytes. Proc Natl Acad Sci USA 1998 95 14337 14342 9826701 10.1073/pnas.95.24.14337
Agaisse H Perrimon N The roles of JAK/STAT signaling in Drosophila immune responses. Immunol Rev 2004 198 72 82 15199955 10.1111/j.0105-2896.2004.0133.x
Eslin P Prévost G Racing against host's immunity defenses: a likely strategy for passive evasion of encapsulation in Asobara tabida parasitoids. J Insect Physiol 2000 46 1161 1167 10818243 10.1016/S0022-1910(99)00227-9
Lanot R Zachary D Holder F Meister M Postembryonic hematopoiesis in Drosophila. Dev Biol 2001 230 243 257 11161576 10.1006/dbio.2000.0123
Lavine MD Strand MR Insect hemocytes and their role in immunity. Insect Biochem Mol Biol 2002 32 1295 1309 12225920 10.1016/S0965-1748(02)00092-9
Meister M Lagueux M Drosophila blood cells. Cell Microbiol 2003 5 573 580 12925127 10.1046/j.1462-5822.2003.00302.x
Rizki TM Rizki RM Bellotti RA Genetics of a Drosophila phenoloxidase. Mol Gen Genet 1985 201 7 13 3932822 10.1007/BF00397978
Russo J Brehelin M Carton Y Haemocyte changes in resistant and susceptible strains of D. melanogaster caused by virulent and avirulent strains of the parasitic wasp Leptopilina boulardi. J Insect Physiol 2001 47 167 172 11064023 10.1016/S0022-1910(00)00102-5
Rizki TM Rizki RM Lamellocyte differentiation in Drosophila larvae parasitized by Leptopilina. Dev Comp Immunol 1992 16 103 110 1499832 10.1016/0145-305X(92)90011-Z
Nappi AJ Cellular immune-response of Drosophila melanogaster against Asobara tabida. Parasitology 1981 83 319 324
Asha H Nagy I Kovacs G Stetson D Ando I Dearolf CR Analysis of Ras-induced overproliferation in hemocytes. Genetics 2003 163 203 215 12586708
Sorrentino RP Melk JP Govind S Genetic analysis of contributions of Dorsal group and JAK-Stat92E pathway genes to larval hemocyte concentration and the egg encapsulation response in Drosophila. Genetics 2004 166 1343 1356 15082553 10.1534/genetics.166.3.1343
Kraaijeveld AR van Alphen JJM Geographical variation in resistance of the parasitoid Asobara tabida against encapsulation by Drosophila melanogaster larvae: the mechanism explored. Physiol Entomol 1994 19 9 14
Monconduit H Prevost G Avoidance of encapsulation by Asobara tabida, a larval parasitoid of Drosophila species. Norw J Agric Sci Suppl 1994 16 301 310
Fellowes MDE Kraaijeveld AR Godfray HCJ The relative fitness of Drosophila melanogaster (Diptera, Drosophilidae) that have successfully defended themselves against the parasitoid Asobara tabida (Hymenoptera, Braconidae). J Evol Biol 1999 12 123 128 10.1046/j.1420-9101.1999.00018.x
Tiën NSH Boyle D Kraaijeveld AR Godfray HCJ Competitive ability of parasitized Drosophila larvae. Evolutionary Ecology Research 2001 3 747 757
Kraaijeveld AR van Alphen JJM Geographical variation in encapsulation ability of Drosophila melanogaster larvae and evidence for parasitoid-specific components. Evol Ecol 1995 9 10 17 10.1007/BF01237692
Kraaijeveld AR Godfray HCJ Trade-off between parasitoid resistance and larval competitive ability in Drosophila melanogaster. Nature 1997 389 278 280 9305840 10.1038/38483
Fellowes MD Kraaijeveld AR Godfray HC Trade-off associated with selection for increased ability to resist parasitoid attack in Drosophila melanogaster. Proc Biol Sci 1998 265 1553 1558 9744107 10.1098/rspb.1998.0471
Kraaijeveld AR Limentani EC Godfray HCJ Basis of the trade-off between parasitoid resistance and larval competitive ability in Drosophila melanogaster. Proc Biol Sci 2001 268 259 261 11217895
Eslin P Prevost G Hemocyte load and immune resistance to Asobara tabida are correlated in species of the Drosophila melanogaster subgroup. J Insect Physiol 1998 44 807 816 12769876 10.1016/S0022-1910(98)00013-4
Irving P Troxler L Heuer TS Belvin M Kopczynski C Reichhart JM Hoffmann JA Hetru C A genome-wide analysis of immune responses in Drosophila. Proc Natl Acad Sci USA 2001 98 15119 15124 11742098 10.1073/pnas.261573998
DeGregorio E Han SJ Lee WJ Baek MJ Osaki T Kawabata SI Lee BL Iwanaga S Lemaitre B Brey PT An immune-responsive Serpin regulates the melanization cascade in Drosophila. Dev Cell 2002 3 581 592 12408809 10.1016/S1534-5807(02)00267-8
DeGregorio E Spellman PT Rubin GM Lemaitre B Genome-wide analysis of the Drosophila immune response by using oligonucleotide microarrays. Proc Natl Acad Sci USA 2001 98 12590 25595 11606746 10.1073/pnas.221458698
Roxström-Lindquist K Terenius O Faye I Parasite-specific immune response in adult Drosophila melanogaster: a genomic study. EMBO Rep 2004 5 207 212 14749722 10.1038/sj.embor.7400073
Boutros M Agaisse H Perrimon N Sequential activation of signaling pathways during innate immune responses in Drosophila. Dev Cell 2002 3 711 722 12431377 10.1016/S1534-5807(02)00325-8
Johansson KC Metzendorf C Söderhäll K Microarray analysis of immune challenged Drosophila hemocytes. Exp Cell Res 2005 305 145 155 15777795 10.1016/j.yexcr.2004.12.018
Irving P Ubeda JM Doucet D Troxler L Lagueux M Zachary D Hoffmann JA Hetru C Meister M New insights into Drosophila larval heamocyte functions through genome-wide analysis. Cell Microbiol 2005 7 335 350 15679837
Waltzer L Ferjoux G Bataillé L Haenlin M Cooperation between the GATA and RUNX factors Serpent and Lozenge during Drosophila hematopoiesis. EMBO J 2003 22 6516 6525 14657024 10.1093/emboj/cdg622
Senger K Armstrong GW Rowell WJ Kwan JM Markstein M Levine M Immunity regulatory DNAs share common organizational features in Drosophila. Mol Cell 2004 13 19 32 14731391 10.1016/S1097-2765(03)00500-8
Rehorn KP Thelen H Michelson AM Reuter R A molecular aspect of hematopoiesis and endoderm development common to vertebrates and Drosophila. Development 1996 122 4023 4031 9012522
Comnish-Bowden A Nomenclature for incompletely specified bases in nucleic acid sequences: recommendations 1984. Nucleic Acids Res 1985 13 3021 3030 2582368
Qiu P Pan PC Govind S A role for the DrosophilaToll/Cactus pathway in larval hematopoiesis. Development 1998 125 1909 1920 9550723
Agaisse H Petersen UM Boutros M Mathey-Prevot B Perrimon N Signaling role of hemocytes in Drosophila JAK/STAT-dependent response to septic injury. Dev Cell 2003 5 441 450 12967563 10.1016/S1534-5807(03)00244-2
Luo H Rose PE Roberts TM Dearolf CR The Hopscotch Jak kinase requires the Raf pathway to promote blood cell activation and differentiation in Drosophila. Mol Genet Genomics 2002 267 57 63 11919715 10.1007/s00438-001-0632-7
Vasta GR Quesenberry M Ahmed H O'Leary N C-type lectins and galectins mediate innate and adaptive immune functions: their roles in the complement activation pathway. Dev Comp Immunol 1999 23 401 420 10426431 10.1016/S0145-305X(99)00020-8
Hoffmann JA Reichhart JM Drosophila innate immunity: an evolutionary perspective. Nat Immunol 2002 3 121 126 11812988 10.1038/ni0202-121
Hoffmann JA Reichhart JM Drosophila immunity. Trends Cell Biol 1997 7 309 316 10.1016/S0962-8924(97)01087-8
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 8819 12077297 10.1073/pnas.092274999
Christophides GK Zdobnov E Barillas-Mury C Birney E Blandin S Blass C Brey PT Collins FH Danielli A Dimopoulos G Immunity-related genes and gene families in Anopheles gambiae. Science 2002 298 159 165 12364793 10.1126/science.1077136
Collins FH Sakai RK Vernick KD Paskewitz SP Seeley DC Miller LH Collins WE Campbell CC Gwadz RW Genetic selection of a Plasmodium-refractory strain of the Malaria vector Anopheles gambiae. Science 1986 234 607 610 3532325
Kumar S Christophides GK Cantera R Charles B Han YS Meister S Dimopoulos G Kafatos FC Barillas-Mury C The role of reactive oxygen species on Plasmodium melanotic encapsulation in Anopheles gambiae. Proc Natl Acad Sci USA 2003 100 14139 14144 14623973 10.1073/pnas.2036262100
Michel K Kafatos FC Mosquito immunity against Plasmodium. Insect Biochem Mol Biol 2005 35 677 689 15894185 10.1016/j.ibmb.2005.02.009
Paskewitz SM Brown MR Lea AO Collins FH Ultrastructure of the encapsulation of Plasmodium cynomolgi (B strain) on the midgut of a refractory strain of Anopheles gambiae. J Parasitol 1988 74 432 439 3379524
Osta MA Christophides GK Kafatos FC Effects of mosquito genes on Plasmodium development. Science 2004 303 2030 2032 15044804 10.1126/science.1091789
Dimopoulos G Seeley D Wolf A Kafatos FC Malaria infection of the mosquito Anopheles gambiae activates immune-responsive genes during critical transition stages of the parasite life cycle. EMBO J 1998 17 6115 6123 9799221 10.1093/emboj/17.21.6115
Ferguson HM Mackinnon MJ Chan BH Read AF Mosquito mortality and the evolution of malaria virulence. Evolution Int J Org Evolution 2003 57 2792 2804 14761058
Orr HA Irving S The genetics of adaptation: the genetic basis of resistance to wasp parasitism in Drosophila melanogaster. Evolution 1997 51 1877 1885
Poirie M Frey F Hita M Huguet E Lemeunier F Periquet G Carton Y Drosophila resistance genes to parasitoids: chromosomal location and linkage analysis. Proc Biol Sci 2000 267 1417 1421 10983825 10.1098/rspb.2000.1158
Rouault J Influence of entomophagous parasites in the competition between Drosophila sibling species: Experimental approach [in French]. CR Acad Sc Paris 1979 t. 289, série D 643 646
Driessen G Hemerik L van Alphen JJM Drosophila species, breeding in the stinkhorn (Phallus impudicus Pers.) and their larval parasitoids. Neth J Zool 1990 40 409 427
Carton Y Boulétreau M van Alphen JJM van Lenteren JC Ashburner M, Carson HL, Thomson JN The Drosophila parasitic wasps. The Genetics and Biology of Drosophila 1986 3e Sunderland, MA: Sinauer Associates 347 394
Fellowes MDE Kraaijeveld AR Godfray HCJ Cross-resistance following artificial selection for increased defence against parasitoids in Drosophila melanogaster. Evolution 1999 53 966 972
Chiu H Govind S Natural infection of D. melanogaster by virulent parasitic wasps induces apoptotic depletion of hematopoietic precursors. Cell Death Differ 2002 9 1379 1381 12478476 10.1038/sj.cdd.4401134
Labrosse C Eslin P Doury G Drezen JM Poirie M Haemocyte changes in D. melanogaster in response to long gland components of the parasitoid wasp Leptopilina boulardi: a Rho-GAP protein as an important factor. J Insect Physiol 2005 51 161 170 15749101 10.1016/j.jinsphys.2004.10.004
Kraaijeveld AR van der Wel NN Geographical variation in reproductive success of the parasitoid Asobara tabida in larvae of several Drosophila species. Ecol Entomol 1994 19 221 229
Kimbrell DA Beutler B The evolution and genetics of innate immunity. Nat Rev Genet 2001 2 256 267 11283698 10.1038/35066006
Vasta GR Lambris JD Innate immunity in the Aegean: ancient pathways for today's survival. Dev Comp Immunol 2002 26 217 225 11755671 10.1016/S0145-305X(01)00072-6
Crozatier M Ubeda J Vincent A Meister M Cellular immune response to parasitization in Drosophila requires the EBF orthologue collier. PLoS Biol 2004 2 E196 15314643 10.1371/journal.pbio.0020196
Evans CJ Hartenstein V Banerjee U Thicker than blood: Conserved mechanisms in Drosophila and vertebrate hematopoiesis. Dev Cell 2003 5 673 690 14602069 10.1016/S1534-5807(03)00335-6
van Alphen JJM Janssen ARM Host selection by Asobara tabida Nees (Braconidae: Alysiinae): a larval parasitoid of fruit inhabiting Drosophila species. Neth J Zool 1982 32 194 214
Gautier L Cope L Bolstad BM Irizarry RA affy: analysis of Affymetrix GeneChip data at the probe level. Bioinformatics 2004 20 307 315 14960456 10.1093/bioinformatics/btg405
BioConductor open source software for bioinformatics
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
The R Project for Statistical Computing
Slonim DK From patterns to pathways: gene expression data analysis comes of age. Nat Genet 2002 32 Suppl 502 508 12454645 10.1038/ng1033
Storey JD A direct approach to false discovery rates. J Roy Stat Soc Ser B 2002 64 479 498 10.1111/1467-9868.00346
Storey JD Tibshirani R Statistical significance for genomewide studies. Proc Natl Acad Sci USA 2003 100 9440 9445 12883005 10.1073/pnas.1530509100
EBI databases: ArrayExpress
EnvBase: the NERC environmental genomics data catalogue
FlyBase: a database of the Drosophila genome
Ensembl Drosophila genome server
The Gene Ontology
InterPro: home
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
Conlon EM Liu XS Lieb JD Liu JS Integrating regulatory motif discovery and genome-wide expression analysis. Proc Natl Acad Sci USA 2003 100 3339 3344 12626739 10.1073/pnas.0630591100
OTTL: MotifRegressor software
Frith MC Fu Y Yu L Chen JF Hansen U Weng Z Detection of functional DNA motifs via statistical over-representation. Nucleic Acids Res 2004 32 1372 1381 14988425 10.1093/nar/gkh299
Clover: Cis-eLement OVERrepresentation
TRANSFAC professional
Lenhard B Sandelin A Mendoza L Engstrom P Jareborg N Wasserman WW Identification of conserved regulatory elements by comparative genome analysis. J Biol 2003 2 13 12760745 10.1186/1475-4924-2-13
The JASPAR database
Evans CJ Banerjee U Transcriptional regulation of hematopoiesis in Drosophila. Blood Cells Mol Dis 2003 30 223 228 12732186 10.1016/S1079-9796(03)00028-7
Govind S Control of development and immunity by Rel transcription factors in Drosophila. Oncogene 1999 18 6875 6887 10602463 10.1038/sj.onc.1203223
Gillespie JP Kanost MR Trenczek T Biological mediators of insect immunity. Annu Rev Entomol 1997 42 611 643 9017902 10.1146/annurev.ento.42.1.611
|
16277749
|
PMC1297650
|
CC BY
|
2021-01-04 16:05:35
|
no
|
Genome Biol. 2005 Oct 31; 6(11):R94
|
utf-8
|
Genome Biol
| 2,005 |
10.1186/gb-2005-6-11-r94
|
oa_comm
|
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-11-r951627775010.1186/gb-2005-6-11-r95MethodA computational investigation of kinetoplastid trans-splicing Gopal Shuba [email protected] Saria [email protected] Terry [email protected] George AM [email protected] Laboratory of Computational Genomics, Rockefeller University, 1230 York Avenue, New York, NY 10021, USA2 Department of Biological Sciences, Rochester Institute of Technology, 85 Lomb Memorial Drive, Rochester, NY 14623, USA3 Laboratory of Molecular Parasitology, Rockefeller University, 1230 York Avenue, New York, NY 10021, USA2005 17 10 2005 6 11 R95 R95 23 5 2005 28 7 2005 7 9 2005 Copyright © 2005 Gopal 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 novel computational approach is presented and applied to predicting trans-splicing sites in 2 chromosomes of Leishmania major.
Trans-splicing is an unusual process in which two separate RNA strands are spliced together to yield a mature mRNA. We present a novel computational approach which has an overall accuracy of 82% and can predict 92% of known trans-splicing sites. We have applied our method to chromosomes 1 and 3 of Leishmania major, with high-confidence predictions for 85% and 88% of annotated genes respectively. We suggest some extensions of our method to other systems.
==== Body
Background
RNA splicing is a key process in the transformation of genomic instructions into functional proteins and may play a critical role in regulating gene expression in a variety of eukaryotes. Two forms of splicing have been documented in eukaryotes. Many eukaryotes use cis-splicing, the process of removing introns from precursor RNAs, to generate mature mRNAs. A related and less understood process, trans-splicing, appears most commonly in a family of protozoa known as the Kinetoplastida, although recent evidence suggests it might be quite widespread as well [1].
While much effort has been focused on identifying the sites of cis-splicing [2-4], a rigorous and thorough analysis of the likely sites for trans-splicing has been slower to appear. Yet the two processes may have common mechanisms, because many of the spliceosomal components are shared [5,6]. Indeed, in Caenorhabditis elegans it appears that both trans-splicing and cis-splicing occur in a coordinated fashion [7]. It is therefore possible that through consideration of the signals involved in trans-splicing, new insights can be gained regarding RNA splicing processes in all eukaryotes. As a first step in this direction, we present a computational analysis of trans-splicing signals from Leishmania major, a member of the Kinetoplastida family.
The Kinetoplastidae diverged approximately 800 million years ago from other eukaryotic lineages [8]. Perhaps as a consequence of this long divergence time, the various species of kinetoplastids exhibit features rarely seen in other eukaryotes. Many genes in kinetoplastids are co-transcribed as polycistronic pre-mRNAs [9-11]. A striking feature of these polycistronic transcripts is their sheer size; in L. major, polycistronic units have been identified that extend nearly half the length of a chromosome [12,13]. Cleavage to monocistronic transcripts is accomplished by the addition of a short spliced leader (SL, or mini-exon) sequence to the 5' untranslated region (UTR) of each transcript through a process known as trans-splicing. As in other eukaryotes, polyadenylation occurs at the 3' end of each mRNA.
In this paper, we use statistical methods to identify those regions most likely to be involved in trans-splicing and predict the most likely splice site(s). Specifically, we have observed that the AG dinucleotide that is most often used as the splice acceptor site is isolated from other AG dinucleotides by long stretches of non-AG dinucleotides. We applied this observation to develop a novel computational approach to predicting trans-splicing acceptor sites in the genus Leishmania. The method and results are presented here and some potential applications for the approach are discussed.
Results
The canonical trans-splicing signal is believed to be composed of four elements: the branch-point adenine (A), a polypyrimidine (C, T-rich) tract, a short variable spacer region, and a downstream 3' splice acceptor site (AG) [9,14]. Of these signals, the easiest to detect computationally are the polypyrimidine tracts. However, these tracts can be highly variable in length (from 5 to well over 100 nucleotides in our datasets) and in composition (entirely pyrimidine or interspersed with purines).
In the analysis we describe here, we used two forms of sequence data. We use the term 'trans-splicing region' to describe data that contain the upstream sequence region with the signals for trans-splicing. This region would include all four of the known signals for trans-splicing and possibly additional sequence information. In our datasets, approximately 400 nucleotides of upstream sequence are designated as the trans-splicing region for each sequence considered. Kinetoplastid 5' UTRs are fairly short, ranging in length from 40 to 200 nucleotides (based on a survey of GenBank entries). By utilizing 400 nucleotides of sequence, we could be reasonably confident that we had included all of the signals associated with trans-splicing. When we wish to refer to the trans-splicing splice site, we will use the term splice junction. This refers to the specific AG dinucleotide that will serve as the 3' splice acceptor site.
Problem statement
In developing a computational method for predicting trans-splice junctions, we can take one of two approaches. We can identify putative coding regions within a genomic sequence, and then search upstream of these genes to locate putative splice junctions. This is not an ideal approach, because it is predicated on accurate gene prediction. No gene-prediction method to date is 100% accurate [15,16], and therefore we will inevitably miss some genes and their associated splice junctions. A second issue is that many gene-finding programs will find the longest possible open reading frame, even if the actual coding start of the gene is internal to the predicted start [15]. The longest open-reading-frame approach was used in the gene-prediction phase of genome annotation in L. major [17], and there are already some instances where the annotated start is known to be upstream of the functional start (AC Ivens, personal communication). When using predicted coding regions as an anchor for searching for splice junctions, we cannot account for such errors in predicting the start of the coding region. Clearly there are some disadvantages to relying on gene prediction as a means for anchoring the search for splice junctions within a genomic sequence.
Alternatively, we can develop a method that attempts to identify splice junctions independently of the presence or absence of genes. In essence, we would take the reverse approach from the one outlined above: we begin by finding all putative splice junctions, and then search downstream for regions that are likely coding regions. The first advantage of this approach is that we can potentially identify genes that were missed by a gene-prediction method. More importantly, we can refine the starts of predicted coding regions based on splice-site predictions. The coding start of a gene must per force be downstream of the trans-splicing junction, so this approach can both predict splice sites and refine gene predictions at the same time.
The disadvantage, of course, is that we must search long stretches of genomic sequence without a clear means for anchoring the search to likely regions of the genome. For this approach to yield reliable predictors of splice junctions, we must first find regions of the genome that are likely to contain splice junctions, and then attempt to identify the putative splice junction itself.
The approach we describe here follows the latter plan of action. In essence, the problem is twofold. There is a classification problem, namely classifying a sequence region as containing trans-splicing signals (what we term trans-splicing regions) or containing other signals. We can use any of a number of well known methods for sequence classification in this phase of the analysis.
Once we have identified sequence regions that are likely to contain a trans-splicing signal, we turn to the second half of the problem. We must now specifically identify the most likely splice junction or junctions within this putative trans-splicing region. This is a separate problem from classification of sequences, and we use a simple metric to identify the most likely splice junction(s) in a given sequence region.
Dinucleotide composition is a reliable indicator of trans-splicing regions
The nucleotide composition of known trans-splicing regions is heavily skewed in favor of pyrimidines in general, even outside of the polypyrimidine tract believed to be part of the trans-splicing signal [18]. In previous work, we used the dramatic shifts in nucleotide composition between known trans-splicing regions and known coding regions as the basis for identifying likely coding regions. This was done in the related organism, Trypanosoma brucei. At the time, we were able to correctly identify 90% of known trans-splicing regions based on dinucleotide composition, for an overall accuracy of 93% in T. brucei [18]. We used linear discriminant analysis (LDA) to classify sequences based on dinucleotide composition.
We have now extended this analysis to L. major. We used 214 expressed-sequence-tag (EST)-mapped trans-splicing regions and 198 known, experimentally verified coding regions as described in Materials and methods. Because these datasets are relatively small, we used tenfold cross-validation to evaluate our models. As described in Materials and methods, tenfold cross-validation involves training on 90% of the data and testing on 10%. This is done over ten iterations, with each iteration involving a random split of both known positives (trans-splicing regions in this instance) and known negatives (coding regions) into the relevant training and testing datasets. Performance of the model is then averaged across the ten test datasets and reported [19]. We also note the range of values across the ten test sets to enable a more fine-grained evaluation of performance.
After tenfold cross-validation, we obtained the results shown in Table 1. Our LDA model has, on average, an accuracy of 96% (range of 90% to 100%). The sensitivity, or ability to identify known trans-splicing regions, is 97% (range of 91% to 100%), and the specificity, or ability to identify known coding regions, is 96% (range of 88% to 100%).
The high accuracy of the LDA model allowed us to reliably classify regions of genomic sequence that were likely to contain trans-splicing signals. To further improve our accuracy at this phase of the analysis, we considered only those predictions that had an individual confidence level of 95% or better. In other words, we only selected those sequence regions where the likelihood that the region was a trans-splicing region was 95% or better. Such predictions have an overall accuracy of 99% (data not shown).
Identifying putative splice junctions within a trans-splicing region requires other metrics
Our LDA model is useful for locating the regions of the genome that are likely to contain a splice junction. However, the LDA model cannot on its own identify the actual splice junction. This is where the second phase of the problem must be addressed. We need a way to identify the most likely splice junction, specifically the AG dinucleotide that will serve as the 3' splice acceptor site. Below, we describe two such metrics and outline why we have selected the use of inter-AG distance as the primary metric for identifying splice junctions.
Polypyrimidine tracts are not reliable indicators of splice junctions
Our first hypothesis was that identifying the longest polypyrimidine tract would enable us to reliably identify the known splice junction. We used pattern matching to identify pyrimidine tracts and allowed variable numbers of purines to be interspersed. After extensive empirical testing, we determined that the fewest false positives were generated when up to two purines were allowed for every six pyrimidines identified. We defined a false positive as a polypyrimidine tract that could be found with equal probability in known trans-splicing regions and known coding regions. Thus the sequence YYRYYYYRYY (Y = pyrimidine, R = purine) would be accepted, but YYRRRRRYYY would not be accepted as a pyrimidine tract.
We applied this approach to 214 known trans-splicing sites derived from EST mappings (see Materials and methods). Unfortunately, it appears that the longest polypyrimidine tract does not directly correlate with the true splice site. In fact, only 51% of known splice sites in the test dataset were correctly predicted using this approach (see Additional data file 1).
We highlight this finding because many previous efforts at identifying the splice junction have focused exclusively on this signal [9,14,20]. Our data would suggest that while the polypyrimidine tracts are certainly necessary for trans-splicing, they do not appear to be sufficient to computationally pinpoint the specific splice junction with a high degree of accuracy.
The distance between AG dinucleotides seems to be a good indicator of splice junctions
While analyzing our training data, we observed that AG dinucleotides show an unusual distribution in known trans-splicing regions (Figure 1). When compared with known coding regions, the inter-AG distances are significantly greater in known trans-splicing regions than in known coding regions. In addition, the longest inter-AG segments seem to correlate well with the known splice sites.
We therefore proposed as a second hypothesis that the inter-AG distance may be a good indicator of splice sites. Surprisingly, up to 60% of known splice sites can be exactly identified simply by selecting the longest inter-AG segment in known trans-splicing regions (data not shown). This is strikingly effective, given the simplicity of the measure.
Overview of method and performance
As described below, we have been able to identify splice sites by combining the evaluation of dinucleotide composition with the inter-AG distance. The method has a mean accuracy of 82%, with a range of 74% to 93% in tenfold cross-validation (Table 2). The sensitivity, or ability to identify splice junctions in known trans-splicing regions, has a mean of 80% (71% to 93%), and a specificity, or ability to eliminate coding regions from predictions, of 85% (75% to 100%). Within trans-splicing regions, on average 92% (15.7 out of 17 predicted) of known splice junctions were correctly predicted (Table 3). Of these, 81% (12.6 out of 15.7) were exact predictions.
Approach to identifying splice junctions
These results were obtained by first extracting all possible inter-AG segments from the training datasets. Each training dataset had on average 192 trans-splicing regions (five sequences), which yielded on average 3,468 inter-AG segments (± 100 segments).
To evaluate the performance of the method, we extracted inter-AG segments from coding regions as well. We chose these sequences because we can reasonably expect that trans-splicing signals will not exist within a functional protein-coding region. However, there is a small probability that these coding regions will contain signals for cis-splicing, as there is some evidence of cis-splicing in kinetoplastids [21]. Therefore, some predictions within coding regions might be functional cis-splicing sites. Nevertheless, in the absence of a better set of negative controls, we have relied on coding regions as our best representative of sequences that do not contain splicing signals.
For each inter-AG segment, regardless of whether it was from a trans-splicing or a coding region, we then used LDA to classify the sequence based on dinucleotide composition. We retained any inter-AG segment that had a 95% confidence or better likelihood of being a trans-splicing region (see Materials and methods).
With the set of inter-AG segments that had the best dinucleotide composition for each trans-splicing region, we next evaluated the inter-AG length. The distribution of inter-AG lengths seen in Figure 1 is quite long-tailed, so we log-transformed the data to approximate a normal curve (Figure 2). We then used the z score as a measure of how a given inter-AG length compares to the mean of the distribution. The larger the z score value, the more standard deviations lie between that inter-AG length and the mean of the distribution [19,22].
We could also assign a confidence value to each individual prediction of a splice site by considering the number of false positives likely to occur at a given z score. We used the training data and results from both trans-splicing regions and known coding regions to determine confidence levels. This would allow us to estimate the likelihood that a given inter-AG length was indicative of a known splice site. We wished to identify an optimal z score such that the false-positive rate could be as low as possible. In Figure 3, a z score of +0.6 yields a false-positive rate of just 5%. Therefore, any inter-AG segment with a z score of 0.6 or greater would have a 95% confidence in the prediction. In all of our analyses, we consider only these high-confidence predictions in assessing the validity of our method.
Splice junction identification in sequences
On average, the model predicts 1.25 high-confidence splice junctions per trans-splicing region. In other words, most trans-splicing segments have one splice junction prediction, with a few having two predictions per sequence. We only have evidence for one splice junction per sequence in this dataset, so at first glance, this might suggest that the false-positive rate is higher than our estimates from coding regions would suggest. However, there is experimental evidence that trans-splicing may occur at multiple sites upstream of some coding regions, and it may be specific to certain stages within the life cycle [23-29]. Thus, it is possible that multiple, valid trans-splicing sites exist for any given transcript. Any computational method for identifying splice sites may therefore identify sites that are functional in some limited context, but for which we do not currently have experimental validation. Predictions that might currently be construed as false positives may prove to be functional as experimental evidence accrues.
Similar performance on data with multiple, known splice junctions
To test the ability of our method to identify multiple known trans-splicing sites, we assembled a small dataset of 21 genes from Leishmania species. These genes each contain experimental evidence for more than one trans-splicing site within their upstream regions. A total of 36 splice sites have been experimentally confirmed within the upstream regions of these genes, and 27 (75%) were identified with high confidence by the method. Of these predictions, 19 (70%) mapped exactly to known splice sites. Allowing for a small window of error of ten nucleotides (error of 0.002 given the length of the sequences analyzed), the method had near exact predictions for 85% (23 out of 27) of the known sites that were identified (full details in Additional data file 2). These findings are very similar to the results obtained from our EST-mapped set of trans-splicing signals.
Application to genomic sequence
We have applied our method to the entire genomic sequence of chromosomes 1 and 3 of L. major. We would emphasize that this represents an example of how our method might be applied and is not presented as a mechanism for evaluating the performance of our method.
There are four challenges to analyzing these data. First, the chromosomal sequences are much longer than any of our other datasets, on the order of several hundred kilobase pairs (kbp). This dramatically raises the statistical noise present in the data. Second, both strands of the genomic sequence must be analyzed for accurate predictions. In all previous datasets, only the strand known to contain the trans-splicing site was evaluated. Third, few of the genes annotated on these chromosomes have been experimentally evaluated for trans-splicing. Therefore, we must compare our predictions with the approximate region in which a prediction would yield a reasonable transcript, as described below. Finally, not all genes annotated on these chromosomes have a clear biological function assigned. Some of the genes will have annotations such as 'hypothetical' or 'conserved hypothetical.' For our purposes, we decided to accept each annotation in the publicly released data, but in Tables 3 and 4 we break down performance based on the category of annotation.
For both chromosomes, we evaluated the entire genomic sequence in both the forward and reverse complement directions. After predicting splice sites regardless of the location of annotated coding regions, we considered the number that were within a reasonable distance upstream of an annotated gene. The key constraint here is the length of the 5' UTR that might result if trans-splicing occurred at the predicted site. Kinetoplastid 5' UTRs are fairly short, ranging in length from 40 to 200 nucleotides (based on a survey of GenBank entries). We therefore considered a prediction as being a reasonable prediction if it was within 400 nucleotides of the annotated start of the coding region. This would yield a UTR that would be within the observed range of lengths.
On chromosome 1, 71 of the 84 (85%) genes have a high-confidence prediction within 400 nucleotides of the annotated start of the coding region (Table 4). Of the remainder, ten genes had low-confidence predictions, and only three genes were missed entirely by the method (i.e. no prediction within 400 nucleotides of the annotated start). Of the missed genes, one was annotated as 'hypothetical' and the other two were annotated as 'conserved hypothetical'.
The results for chromosome 3 are similar: of 98 annotated genes, 86 (88%) had high-confidence predictions (Table 5). The other 12 genes had low-confidence predictions. No genes were missed for this chromosome.
These results are very consistent with the method's performance on other datasets, suggesting that the method is robust and can be applied to long genomic sequences. The few instances where we have missed genes may be instances in which the annotated start of the open reading frame varies from the functional start of the coding region. We explore this issue in more detail in the Discussion. A more detailed version of these results is included (see Additional data file 4), and a graphical representation of predicted splice site locations along each chromosome is also available [30].
Discussion
The ability to identify a small signal in much longer sequences is a critical issue in the computational identification of both trans-splicing and cis-splicing sites. The results seen here mirror similar results from methods attempting to identify the 3' acceptor site in cis-splicing [16]. While intronic sequences are typically longer than the upstream trans-splicing regions used here, a comparison in performance is still valid. This is because our method attempts to identify trans-splicing sites without prior knowledge of coding region location. As a result, our method scanned the entire length of each chromosome or genomic region available for analysis. These lengths are more than equivalent to the intron-length scans used by many other gene-prediction methods [16]. Given the nature of the signal, our method performs as well as most existing tools that identify 3' acceptor sites in cis-splicing.
In contrast to other methods for identifying splicing signals, however, our method takes advantage of two relatively simple statistical measures. Nucleotide composition and inter-AG distance seem to be almost too simple, and it would appear that a more powerful method would yield better results. Indeed, most tools for cis-splicing use complex probabilistic models such as hidden Markov models to identify splice sites effectively [2,31]. Such methods could indeed further our ability to identify trans-splicing sites, if sufficient data were available to correctly train such methods. As the complexity of a statistical model increases, so does the quantity of data required for accurate and reliable modeling. In our case, the paucity of known trans-splicing sites limits the applicability of complex statistical models. Our method represents, to our knowledge, the best available option given the need for predicting these regions and the limited data available for modeling the features of the trans-splicing junction.
A key advantage of our method is that it can identify splice sites without a priori knowledge of the location of coding regions. One of the challenges with gene prediction in the kinetoplastids is that the longest open reading frame present in a genomic segment does not necessarily correlate with the true open reading frame in the mRNA. Currently, the most popular tool for gene prediction in the kinetoplastids will always predict the longest possible open reading frame [15].
Our method could be used to refine the identification of the true open reading frame. If a genomic segment predicted to contain a gene also contained a high-confidence splice site within the putative open reading frame, that would be strong evidence for an internal start rather than the furthest upstream start codon. Conversely, if the high-confidence splice site were upstream of the furthest upstream start codon, then it would argue in favor of retaining that start codon as the functional start of the gene. In this manner, true coding regions might be more effectively identified. This could in turn assist in subsequent experimental studies of gene function.
A second advantage of our method is that it can predict more than one splice junction in a given sequence region. Given the possibility of multiple splice sites for trans-splicing in kinetoplastid genes, any method for splice-site prediction must be able to account for this phenomenon. Since our method provides a confidence estimate with each prediction, it is possible to evaluate multiple splice-site predictions for the likelihood that they are functional. This should provide others with the means to evaluate these predictions in an experimental context.
Our method also suggests interesting avenues for further research into the phenomenon of trans-splicing. Two models have been proposed for the mechanism by which splicing might occur. The first model would argue for a set of highly conserved motifs that direct the spliceosome to a specific location and a target AG that will serve as the 3Â' splice acceptor site. For example, cis-splicing sites in Saccharomyces cerevisiae include highly conserved sequences immediately upstream of the 3Â' splice acceptor site that allow for precise splicing at the correct location [32]. In the second model, the spliceosome would employ a scanning technique, evaluating each AG as a potential splice site [33-36]. In this model, strongly conserved signals are not required, but consistent nucleotide context, including a polypyrimidine bias, would be critical for the spliceosome to identify candidate splice acceptor sites.
The success of our method, which does not rely on strict consensus sequence features, favors a scanning model, where the trans-splicing process can occur at any AG dinucleotide that satisfies the general requirements for a splice site. The experimentally observed utilization of multiple trans-splicing sites in a given upstream region is consistent with this view of the trans-splicing mechanism. Indeed, splicing probably occurs sequentially at multiple upstream sites, with differing efficiencies, until no further site is recognized and a stable mRNA is generated almost by default. In effect, the final mRNA is formed by lack of further recognizable splice sites as the spliceosome, perhaps linked to the transcription machine, passes along the nascent RNA. We would encourage the use of the findings presented here not only in their predictive capacity, but as an impetus for further study of this intriguing process.
The method presented here is easily extensible to other members of the kinetoplastid family, and perhaps to other organisms that exhibit trans-splicing. It may also be possible to generalize this approach to improve on the prediction of 3' splice acceptor sites in cis-splicing. Preliminary results with human introns and exons indicate a distribution of inter-AG lengths that is almost identical to the distribution shown in Figure 1 (based on human intron and exon sequence data from [37]; data not shown). Thus, it is possible that the study of signals underlying cis-splicing might benefit from cross-fertilization with the methods we have developed for trans-splicing.
Conclusions
We present a method for identifying the 3' splice acceptor site during trans-splicing, an unusual process in which two RNAs are spliced together to yield a mature mRNA. Our method is able to predict 92% of known splice sites with high confidence, and 81% of these predictions map exactly to the known splice site. Based on the statistical measures we use, it appears that our method would support the scanning model of trans-splicing rather than the model of site-specific binding by the spliceosomal apparatus.
We propose that our method might be applied to refine gene prediction in kinetoplastid genomes by assisting in the identification of the true starts of coding regions. In addition, our method could possibly be extended to other organisms and may even be relevant to the study of 3' acceptor site selection in cis-splicing.
Materials and methods
Data
Our data consisted of upstream sequences known to be involved in trans-splicing and experimentally verified coding regions from L. major. We used tenfold cross-validation to train and test our method. In addition, we had an independent test dataset of experimentally verified splice junctions derived from other members of the Leishmania genus. The latter data are described in more detail in Additional data file 2.
Primary dataset
For our primary dataset, we started with 527 5' ESTs from L. major. These were mapped to the completed genome sequence of this protozoan, based on the genome sequence release of July 2005 [17]. Each of these ESTs was mapped to the genomic sequence using BLAST [38], with the requirement that the EST had to match the genomic sequence at 95% or greater sequence identity. In addition, the EST had to match no other genomic sequence with greater than 50% identity. By setting these very stringent levels on the EST mappings, we were able to obtain a set of 266 unique, nonredundant, highly accurate mappings of ESTs to the genomic sequence of L. major. Of these, 214 had a clear AG splice acceptor site immediately upstream of the mapped region, and these sequences were used for all aspects of the analysis described here.
While ESTs do tend to have higher rates of sequencing error, our stringent mapping to the genomic sequence allowed us to use the latter for all analyses. That is, we primarily used the genomic sequence that the EST mapped to, rather than the EST sequence itself. Therefore, we can be reasonably confident that the sequences do not contain the high error rates of the original EST sequence.
For each EST mapping, we obtained 400 nucleotides of sequence upstream of the 5' end of the EST mapping. This would be expected to contain the actual AG dinucleotide used as the splice acceptor site, as well as signals upstream of this AG dinucleotide.
In addition to these trans-splicing regions, we identified 198 coding regions from GenBank that had experimental evidence of function. The full set of GenBank accession numbers for these sequences is provided in Additional data file 3. We noted that four of these coding regions were present on chromosome 1 or 3 of L. major by using BLAST to map coding sequences to the genome.
With our dataset of 214 uniquely mapped trans-splicing regions and 198 coding regions, we generated ten sets of training and testing data for tenfold cross-validation. Using an ad hoc Perl script, we split the trans-splicing regions such that 90% were used for training in each set and 10% were used for testing. Similarly, we split the coding regions to generate ten sets of training and testing data. By training on each of the ten sets and then evaluating performance on the associated testing data, we were able to obtain reliable estimates of the method's performance.
Other data
We also used the publicly available genomic sequences for chromosomes 1 and 3 of L. major [12,13]. These were downloaded from GenBank ([GenBank:AE001274, GenBank:AC125735]), and the analysis was run on the entire genomic sequence. After the prediction of splice sites for the entire forward and reverse strands, any splice site that was within 400 nucleotides of the annotated start of a predicted gene was retained as a splice-site prediction. Those predictions were then evaluated for their confidence level as reported in Results.
Algorithm development
The algorithm described here has three main stages of analysis. In the first stage, all inter-AG segments are extracted from a FASTA formatted file [39] of sequences using an ad hoc Perl script [40]. In the second phase of the analysis, the nucleotide content of each inter-AG segment is evaluated using linear discriminant analysis (LDA). Finally, the inter-AG length is assessed using the z score. For the LDA analysis, we needed a means of comparing sequences of differing lengths and composition. We compared the trans-splicing regions to coding regions at the dinucleotide level using transition probabilities. We applied maximum likelihood estimation (MLE) to estimate these probabilities, since this method allows for estimation from a relatively small sample size. MLE-based transition probabilities are calculated by the formula:
where akl is the transition probability that the nucleotide l follows the nucleotide k, and ckl is the number of times the dinucleotide combination kl occurs. In the denominator, we calculate the sum of the transition probabilities of all nucleotides that could follow k, represented by l' as any of the four nucleotides [41].
We used these transition probabilities to train LDA, which, as implemented in the statistical package R, was used for this analysis [42]. We then tested performance of the LDA approach on the test datasets. The training and testing occurred over ten iterations.
Each prediction by LDA is accompanied by a posterior probability value, a measure of the likelihood that a given individual prediction is correct [42]. We used these posterior probabilities to select those inter-AG segments that scored at the 95% confidence level or higher. We have previously shown that selecting sequences with a 95% confidence in the individual predictions leads to an overall classification accuracy of 98% [18]. Therefore, selecting inter-AG segments that had such high posterior probability values ensured that true trans-splicing regions were selected with a high degree of confidence.
The selected inter-AG segments were then assessed based on the inter-AG segment length. For each inter-AG segment, a z score was calculated based on the mean and standard deviation of the log-transformed distribution from the training data.
We calculated z scores as follows:
where z is the z score, x is the log-transformed inter-AG distance, μ is the mean of the log-transformed distribution, and σ is the standard deviation of the log-transformed distribution [19]. For the L. major training dataset, μ was 3.351 and σ was 0.881 averaged across the ten cross-validation datasets.
To estimate the likelihood that a given prediction was a true assignment, we used the false-positive rate, determined from the training dataset, as a function of z scores. That is, for each z score from -6 to +6, we determined the proportion of predictions that were false positives (known nonsplice sites predicted to be functional splice sites) assigned at each z score. This led to the plot shown in Figure 3. From this, we determined that a z score of 0.6 or higher would be likely to represent a false-positive rate of just 5%. In other words, we could be 95% confident that a prediction made at this z score represented a true positive. Thus, any prediction with a z score of 0.6 or greater is considered a high-confidence prediction. High-confidence predictions for the various datasets are reported in Results.
Additional data files
The following additional data are available with the online version of this paper. Additional data file 1 is a PDF file describing the method, data, and detailed results for the identification of polypyrimidine tracts within known trans-splicing regions. Additional data file 2 is a PDF file containing the data and detailed results for our dataset of multiple splice sites with experimental confirmation. Additional data file 3 is a PDF file containing all the data used in this analysis; the sequence data are presented in the FASTA format. Additional data file 4 is a PDF file containing the full set of predictions for chromosomes 1 and 3 of L. major.
Supplementary Material
Additional data File 1
PDF file describing the method, data, and detailed results for the identification of polypyrimidine tracts within known trans-splicing regions
Click here for file
Additional data File 2
PDF file containing the data and detailed results for our dataset of multiple splice sites with experimental confirmation
Click here for file
Additional data File 3
PDF file containing all the data used in this analysis; the sequence data are presented in the FASTA format
Click here for file
Additional data File 4
PDF file containing the full set of predictions for chromosomes 1 and 3 of Leishmania major
Click here for file
Acknowledgements
The authors would like to thank Allison Griggs for developing the LDA model used here. A portion of this work was supported by a grant to T.G. from the Burroughs Wellcome Fund New Investigator in Molecular Parasitology award (no. 1001530). The authors would like to thank the Leishmania Genome Network for making genomic sequence data publicly available.
Figures and Tables
Figure 1 Inter-AG lengths in known splicing and coding regions. Inter-AG distances in known coding and trans-splicing regions show different distributions. The distance between AG dinucleotides is significantly greater in known trans-splicing regions than in known coding regions. Distances are shown for 214 known trans-splicing regions and 198 coding regions. The mean inter-AG distance for the coding region data is 42 nucleotides, compared with a mean inter-AG distance of 81 nucleotides for known trans-splicing regions.
Figure 2 Log-normal transform of inter-AG lengths. Inter-AG distances after log-normal transform show a roughly normal curve. To better evaluate the inter-AG distances in known trans-splicing regions, we transformed the long-tailed distribution seen in Figure 1 using a log-transform. The result is a good approximation to a normal curve, allowing us to use the full panoply of statistical analyses available for manipulations of normally distributed data.
Figure 3 False-positive rate as a function of z score. False-positive rate as a function of z score can be used to measure the confidence of an individual prediction. The rate of false positives predicted by the method is shown as a function of the z scores used to evaluate inter-AG distances. False-positive rates were estimated for a range of z scores from -6 to +6 based on known splice sites in the training data. The dotted lines indicate that a z score of 0.6 or greater will yield a false positive rate of just 5%. In other words, inter-AG segments with a z score of 0.6 or greater will have a 95% confidence of being trans-splicing regions.
Table 1 Classification of sequences by linear discriminant analysis (LDA)
Known trans-splicing regions Known coding regions
Predicted trans-splicing region True positive False positive
20 0.9
(14-24) (0-3)
Predicted coding region False negative True negative
0.7 18.5
(0-2) (10-31)
Sensitivity: 0.97 Specificity: 0.96
(0.91-1.00) (0.88-1.00)
Accuracy: 0.96
(0.90-1.00)
The overall performance of the LDA method after tenfold cross-validation using 214 known trans-splicing regions and 198 coding regions is shown here. The average across all ten testing sets is reported, with the range of values indicated in parentheses for each class of sequence. Each test dataset had on average 20.7 known trans-splicing regions and 19.4 known coding regions.
Table 2 Identification of splice junctions
Known trans-splicing regions Known coding regions
Predicted splice sites True positive False positive
17 2.5
(10-22) (0-4)
Predicted nonsplice sites False negative True negative
4.5 13.9
(1-8) (11-16)
Sensitivity: 0.80 Specificity: 0.85
(0.71-0.93) (0.75-1.00)
Accuracy: 0.82
(0.74-0.93)
The overall performance of the method in identifying splice junctions was determined by comparing the number of known splice junctions that were identified by the method in known trans-splicing regions versus those in known coding regions. These results are from tenfold cross-validation, and each test dataset had on average 21.5 known trans-splicing regions and 16.4 known coding regions.
Table 3 High-confidence predictions for known trans-splicing regions
Distance from known site (nucleotides) Number of regions with sites predicted (n = 17 regions)
Exact matches 12.6 (7-16)
10 1.38 (1-2)
25 1.5 (1-3)
50 1.38 (1-2)
Missing predictions 4.5 (1-8)
Overall performance of the method on a set of known trans-splicing regions (tenfold cross-validation of 214 EST mapped trans-splicing sites). Each test dataset had on average 21.5 known trans-splicing regions, of which on average 17 had predictions. Missing predictions indicate those sequences for which no high-confidence prediction was available or where the nearest prediction was more than 50 nucleotides away. The mean of all ten datasets is reported with the range of values in parentheses.
Table 4 Predictions for chromosome 1
Public annotation Totals High confidence Low confidence No prediction
Forward strand
Protein function assigned 22 20 2 0
Conserved hypothetical 18 16 0 2
Hypothetical 13 9 3 1
Total 53 45 5 3
Reverse strand
Protein function assigned 9 8 1 0
Conserved hypothetical 18 14 4 0
Hypothetical 4 4 0 0
Total 31 26 5 0
Comparison of predicted splice sites with annotation of chromosome 1 of Leishmania major. A total of 84 genes have been annotated on chromosome 1 of L. major [12]. Of these, the method finds a splice site with a high-confidence score in all but 13 instances (85%). Only three genes were missed entirely by the method, with no prediction within the 400 nucleotide window upstream of the annotated start of the gene.
Table 5 Predictions for chromosome 3
Public annotation Totals High confidence Low confidence No prediction
Forward strand
Protein function assigned 18 14 4 0
Conserved hypothetical 5 5 0 0
Hypothetical 44 40 4 0
Total 67 59 8 0
Reverse strand
Protein function assigned 7 6 1 0
Conserved hypothetical 2 2 0 0
Hypothetical 22 19 3 0
Total 31 27 4 0
Comparison of predicted splice sites with annotation of chromosome 3 of Leishmania major. A total of 98 genes have been annotated on chromosome 3 of L. major [13]. Of these, the method finds a splice site with a high confidence score in all but 12 instances (88%). A splice site was predicted for every gene annotated on this chromosome.
==== Refs
Hastings KE SL trans-splicing: easy come or easy go? Trends Genet 2005 21 240 247 15797620 10.1016/j.tig.2005.02.005
Burge C Karlin S Prediction of complete gene structures in human genomic DNA. J Mol Biol 1997 268 78 94 9149143 10.1006/jmbi.1997.0951
Fickett J The gene identification problem: an overview for developers. Comput Chem 1996 20 103 118 10.1016/S0097-8485(96)80012-X
Zhang MQ Identification of protein coding regions in the human genome by quadratic discriminant analysis. Proc Natl Acad Sci USA 1997 94 565 568 9012824 10.1073/pnas.94.2.565
Liang Xh Haritan A Uliel S Michaeli S trans and cis splicing in trypanosomatids: mechanism, factors, and regulation. Eukaryot Cell 2003 2 830 840 14555465 10.1128/EC.2.5.830-840.2003
Palfi Z Lucke S Lahm H Lane W Kruft V Bragado-Nilsson E Seraphin B Bindereif A The spliceosomal snRNP core complex of Trypanosoma brucei : cloning and functional analysis reveals seven Sm protein constituents. Proc Natl Acad Sci USA 2000 97 8967 8972 10900267 10.1073/pnas.150236097
Blumenthal T Evans D Link CD Guffanti A Lawson D Thierry-Mieg J Thierry-Mieg D Chiu WL Duke K Kiraly M Kim SK A global analysis of Caenorhabditis elegans operons. Nature 2002 417 851 854 12075352 10.1038/nature00831
Fernandes AP Nelson K Beverley SM Evolution of nuclear ribosomal RNAs in kinetoplastid protozoa: perspectives on the age and origins of parasitism. Proc Natl Acad Sci USA 1993 90 11608 11612 8265597
Lee MGS Van der Ploeg LH Transcription of protein-coding genes in trypanosomes by RNA polymerase I. Annu Rev Microbiol 1997 51 463 489 9343357 10.1146/annurev.micro.51.1.463
Martinez-Calvillo S Yan S Nguyen D Fox M Stuart K Myler P Transcription of Leishmania major Friedlin chromosome 1 initiates in both directions within a single region. Mol Cell 2003 11 1291 1299 12769852 10.1016/S1097-2765(03)00143-6
Monnerat S Martinez-Calvillo S Worthey E Myler P Stuart K Fasel N Genomic organization and gene expression in a chromosomal region of Leishmania major. Mol Biochem Parasitol 2004 134 233 243 15003843 10.1016/j.molbiopara.2003.12.004
Myler P Audleman L deVos T Hixon G Kiser P Lemley C Magness C Rickel E Sisk E Sunkin S Leishmania major Friedlin chromosome 1 has an unusual distribution of protein-coding genes. Proc Natl Acad Sci USA 1999 96 2902 2906 10077609 10.1073/pnas.96.6.2902
Worthey E Martinez-Calvillo S Schnaufer A Aggarwal G Cawthra J Fazelinia G Fong C Fu G Hassebrock M Hixson G Leishmania major chromosome 3 contains two long convergent polycistronic gene clusters separated by a tRNA gene. Nucleic Acids Res 2003 31 4201 4210 12853638 10.1093/nar/gkg469
Vanhamme L Pays E Control of gene expression in trypanosomes. Microbiol Rev 1995 59 223 240 7603410
Delcher A Harmon D Kasif S White O Salzberg S Improved microbial gene identification with GLIMMER. Nucleic Acids Res 1999 27 4636 4641 10556321 10.1093/nar/27.23.4636
Rogic S Mackworth AK Ouellette FB Evaluation of gene-finding programs on mammalian sequences. Genome Res 2001 11 817 832 11337477 10.1101/gr.147901
Ivens AC Peacock CS Worthey EA Murphy L Aggarwal G Berriman M Sisk E Rajandream MA Adlem E Aert R The genome of the kinetoplastid parasite, Leishmania major. Science 2005 309 436 442 16020728 10.1126/science.1112680
Gopal S Cross GA Gaasterland T An organism-specific method to rank predicted coding regions in Trypanosoma brucei. Nucleic Acids Res 2003 31 5877 5885 14530436 10.1093/nar/gkg798
Ewens WJ Grant GR Statistical Methods in Bioinformatics: An Introduction 2001 New York: Springer-Verlag
Huang J Van der Ploeg L Requirement of a polypyrimidine tract for trans-splicing in trypanosomes: discriminating the PARP promoter from the immediately adjacent 3' splice acceptor site. EMBO J 1991 10 3877 3885 1935907
Mair G Shi H Li H Djikeng A Aviles HO Bishop JR Falcone FH Gavrilescu C Montgomery J Santori MI A new twist in trypanosome RNA metabolism: cis-splicing of pre-mRNA. RNA 2000 6 163 169 10688355 10.1017/S135583820099229X
Mendenhall W Sincich T A Second Course in Statistics: Regression Analysis 1996 Upper Saddle River, NJ: Prentice Hall
Erondu NE Donelson JE Differential expression of two mRNAs from a single gene encoding an HMG1-like DNA binding protein of African trypanosomes. Mol Biochem Parasitol 1992 51 111 118 1565127 10.1016/0166-6851(92)90206-Y
Graham SV Mechanisms of stage-regulated gene expression in kinetoplastida. Parasitol Today 1995 11 217 223 15275346 10.1016/0169-4758(95)80081-6
Kapotas N Bellofatto V Differential response to RNA trans-splicing signals within the phosphoglycerate kinase gene cluster in Trypanosoma brucei. Nucleic Acids Res 1993 21 4067 4072 8371981
Layden RE Eisen H Alternate trans-splicing in Trypanosoma equiperdum: implications for splice site selection. Mol Cell Biol 1988 8 1352 1360 2835672
Pelle R Murphy NB Stage-specific differential polyadenylation of mini-exon derived RNA in African trypanosomes. Mol Biochem Parasitol 1993 59 277 286 7688101 10.1016/0166-6851(93)90225-M
Revelard P Lips S Pays E Alternative splicing within and between alleles of the ATPase gene 1 locus of Trypanosoma brucei. Mol Biochem Parasitol 1993 62 93 101 8114831 10.1016/0166-6851(93)90181-V
Vassella E Braun R Roditi I Control of polyadenylation and alternative splicing of transcripts from adjacent genes in a procyclin expression site: a dual role for polypyrimidine tracts in trypanosomes? Nucleic Acids Res 1994 22 1359 1364 8190625
Kinetoplastid Genome Features
Krogh A Two methods for improving performance of an HMM and their application for gene finding. Proc Int Conf Intell Syst Mol Biol 1997 5 179 186 9322033
Burge CB Tuschl T Sharp PA Gesteland RF, Cech TR, Atkins JF Splicing of precursors to mRNAs by the spliceosomes. The RNA World 1999 2 Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press 525 560
LeBowitz JH Smith HQ Rusche L Beverley SM Coupling of poly(A) site selection and trans-splicing in Leishmania. Genes Dev 1993 7 996 1007 8504937
Lopez-Estrano C Tschudi C Ullu E Exonic sequences in the 5' untranslated region of α-tubulin mRNA modulate trans-splicing in Trypanosoma brucei. Mol Cell Biol 1998 18 4620 4628 9671472
Matthews KR Tschudi C Ullu E A common pyrimidine-rich motif governs trans-splicing and polyadenylation of tubulin polycistronic pre-mRNA in trypanosomes. Genes Dev 1994 8 491 501 7907303
Ullu E Matthews KR Tschudi C Temporal order of RNA-processing reactions in trypanosomes: rapid trans splicing precedes polyadenlyation of newly synthesized tubulin transcripts. Mol Cell Biol 1993 13 720 725 8417363
Burset M Guigo R Evaluation of gene structure prediction programs. Genomics 1996 34 353 367 8786136 10.1006/geno.1996.0298
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
Pearson WR Using the FASTA program to search protein and DNA sequence databases. Methods Mol Biol 1994 24 307 331 8205202
Wall L Christiansen T Orwant J Programming Perl 2000 3 Sebastopol, CA: O'Reilly
Durbin R Eddy S Krogh A Mitchison G Biological Sequence Analysis 1998 Cambridge, UK: Cambridge University Press
Venables W Ripley B Modern Applied Statistics with S-Plus 1999 3 Heidelberg: Springer-Verlag
Siegel TN Tan KS Cross GA A systematic study of sequence motifs for RNA trans-splicing in Trypanosoma brucei. Mol Cell Biol 2005
The Leishmania major Friedlin Genome Project
|
16277750
|
PMC1297651
|
CC BY
|
2021-01-04 16:05:35
|
no
|
Genome Biol. 2005 Oct 17; 6(11):R95
|
utf-8
|
Genome Biol
| 2,005 |
10.1186/gb-2005-6-11-r95
|
oa_comm
|
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-11-r961627775110.1186/gb-2005-6-11-r96MethodChipper: discovering transcription-factor targets from chromatin immunoprecipitation microarrays using variance stabilization Gibbons Francis D [email protected] Markus [email protected] Kevin [email protected] Frederick P [email protected] Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Longwood Avenue, Boston, MA 02115, USA2 Instituto de Biología Molecular y Celular de Plantas (IBMCP), Universidad Politécnica de Valencia, Camino de Vera s/n, 46022 Valencia, Spain2005 1 11 2005 6 11 R96 R96 23 3 2005 1 8 2005 30 9 2005 Copyright © 2005 Gibbons 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 new method, implemented in software as 'Chipper', is described that
allows genome-wide determination of protein-DNA binding sites from
chromatin immunoprecipitation microarrays.
Chromatin immunoprecipitation combined with microarray technology (Chip2) allows genome-wide determination of protein-DNA binding sites. The current standard method for analyzing Chip2 data requires additional control experiments that are subject to systematic error. We developed methods to assess significance using variance stabilization, learning error-model parameters without external control experiments. The method was validated experimentally, shows greater sensitivity than the current standard method, and incorporates false-discovery rate analysis. The corresponding software ('Chipper') is freely available. The method described here should help reveal an organism's transcription-regulatory 'wiring diagram'.
==== Body
Background
A major goal in understanding cellular behavior is to reveal the 'wiring' of transcriptional regulation, through which transcription factors (TFs) bind target-gene promoters to control gene expression. Promoter regions contain sequence elements - typically 5 to 12 nucleotides (nt) in length - at which TFs bind specifically. By enhancing/inhibiting transcription or recruiting complexes that remodel chromatin structure, TFs regulate expression of the genes whose promoters they bind. Chromatin immunoprecipitation (ChIP) is an experimental technique for identifying those regions of DNA bound by a particular protein, and is, therefore, a useful method for determining which genes have their promoters bound by a TF. In outline, the method consists of the following steps. The TF under study is crosslinked to DNA which is subsequently extracted and sheared into fragments approximately 400 nt long (1,000 nt resolution is usually sufficient to assign binding to the regulation of a specific gene, so it is rare to exceed this length [1]). The fragments are immunoprecipitated with an antibody specific to that TF (or to a peptide affinity tag fused to that TF), whereupon the crosslinks are reversed, the DNA precipitate amplified, and the intergenic regions (IGRs) containing the binding site(s) are determined by examining the relative abundance of each immunoprecipitated DNA fragment. The combination of ChIP with microarray technology is often called 'ChIP-chip' [1] and is referred to here as 'Chip2'. It has turned ChIP into a high-throughput technique for efficiently mapping gene regulatory networks [2-9].
Two-channel microarrays use hybridization to compare the abundance of specific nucleic acid sequences in one mixture to abundance of the same sequences in another control mixture. The choice of control mixture may greatly affect the outcome of the experiment. A typical choice is fragmented genomic DNA, which controls for the relative abundance and non-specific hybridization potential of genomic DNA fragments. Genomic DNA may be purified from 'whole-cell extract', which itself is sometimes used as a control. As some DNA fragments may be 'stickier' than others, a more stringent and laborious mock control (containing fragments recovered nonspecifically by immunoprecipitation (IP)) is sometimes performed, in which the TF does not have a fused affinity tag.
The change in abundance of a particular sequence between two mixtures is often measured in terms of 'fold-change' between the two channels (ratio) or, alternatively, the logarithm of fold-change (log-ratio). The IP channel serves as numerator, while the control is the denominator. The array surface between regions with spotted DNA is never completely 'dark', due to the combined effects of residual DNA fragments bound non-specifically to the array surface, and the experimentalist's control of the visual amplification ('gain') in the image analysis software. It is customary to subtract this 'background' from each spot because it reveals nothing about the protein-DNA binding. This subtraction raises the possibility, however, that the denominator could become negative or zero, in which case the log-ratio is not useful. Common strategies for handling zero or negative values are either to threshold or to discard data points altogether, neither of which is entirely satisfactory. A further, and perhaps more serious, problem is the practice of interpreting this fold-change as a measure of significance, when it provides no such statistical basis. Small random fluctuations in signals close to background, particularly in the denominator, are amplified, leading to spuriously high levels of 'fold-change' [10]. In other words, we should reduce our confidence in a twofold change between signals that are each near the background noise, compared to a twofold change between strong signals. Because we are generally more interested in whether a region is specifically bound at all than we are in the degree of its binding (occupancy), there is a need for an accurate measure of confidence in each measurement.
A statistical approach for analysis of mRNA abundance microarrays has been developed in which a 'single-array' error model accounts for variation in the background level for each microarray, while a 'gene-specific' error model describes variation of a single gene across replicate arrays. These two complementary models can be combined to estimate the error in each log-ratio measurement [10]. A variant of the single-array approach (in which there is gene-specific normalization) has been applied to transcription-factor binding site identification by means of Chip2 in yeast [2]. Unfortunately, it requires one or more separate control experiments to determine error model parameters, in which identical nucleic acid mixtures are compared. This adds to the expense of the experiment; furthermore, error model parameters derived from a separate microarray are potential sources of systematic error, since quality can vary between microarrays.
Results and discussion
Here we describe a new approach for assessing statistical significance of TF-binding from Chip2 data. We illustrate our method using a Chip2 analysis of Sko1 (also known as Acr1), a TF of the basic leucine zipper (bZIP) family (CREB sub-family) that regulates the expression of osmotic stress inducible genes [11-13]. We also use independent confirmation experiments of individual IGRs to validate our method.
Combining replicates
We distinguish two kinds of repeated experiment. When the same IGR is spotted onto an array in more than one location, we term these measurements 'duplicates,' and we consider them as two spatially separated parts of the same 'spot'. Though other approaches have been described [14], for simplicity we average duplicate signals before analyzing them, giving us a single value that is less susceptible to physical blemishes on the slide. When the same IGRs are spotted onto two or more distinct microarrays, we term them 'replicates.' We consider each replicate as an independent measurement of the binding affinity or 'occupancy' of the IGRs.
Variance stabilization
It is common to replicate genome-wide experiments several times, to improve confidence in the results, which may be degraded by array imperfections or by handling errors. Additional replicates can compensate for random error in individual measurements, and the typical number of replicates is likely to increase as the cost of microarrays falls [1]. Sometimes the most significantly enhanced IGRs are those with low signal-to-noise ratio, yet applying log-ratios to such signals has the potential to introduce many false positives because minor variations in a small denominator value can have a large effect on a ratio. A single-array error model can account for this variation in calculating significance for each IGR. The log-ratios themselves are difficult to interpret, however, because two IGRs with the same log-ratio may differ in significance, and a greater log-ratio does not indicate increased significance. An alternative approach, the method of variance stabilization, was described by two groups [15,16] and made available as part of the BioConductor project [17] in the package 'vsn' [15]. It uses a regression algorithm that is robust to outliers to scale and offset each channel independently, in such a way that the variance between channels is independent of signal strength. The transformation of the signal yi in the ith channel (i = 1 for IP, or i = 2 for control) can be expressed as:
where αi and λi represent the background and noise in the ith channel, respectively. Because ln(a) - ln(b) = ln(a/b), the difference between the two transformed channels (Δh ≡ hi - h2) is then a generalized log-ratio that is asymptotically equivalent to the log-ratio of the original channels when both are high (yi >> αi), yet transforms smoothly to the difference between channels when both are low. This allows direct comparison between any two datapoints, even when they belong to opposite ends of the microarray's dynamic range. Two IGRs with the same Δh are equally significant, and greater Δh implies a more significantly bound IGR.
Deriving error model parameters internally
Binding of protein to DNA is a dynamic, stochastic process in equilibrium. While every TF is likely to be bound to every IGR at least some fraction of the time, our goal here is to perform binary classification of the IGRs. We therefore consider IGRs to fall into two categories: those that are specifically bound by the TF and those that are not. We wish to compute a p value that expresses our degree of surprise at seeing a particular Δh score for a given IGR, under the null hypothesis that the IGR is not bound. The 'vsn' package can be used to variance-stabilize each array separately, or all of them simultaneously; we used the former method. Having computed the inter-channel variance-stabilized difference (Δh) for each spot, we may plot a histogram of all scores from a chip. We expect that most regions are not bound. Therefore, the distribution of Δh scores should be largely determined by random binding and measurement errors [18]. A smaller number of regions are bound, and those will tend to have positive scores, indicating higher occupancies in the IP channel than the whole-cell extract/mock control. Measurements in the negative portion of the Δh distribution should, therefore, be more completely dominated by unbound IGRs. By fitting a parametric curve to the region of the observed Δh distribution left of the mode, we obtain an estimate of the null distribution in the positive region of the Δh distribution. This is an essential feature of our method, because it allows us to estimate the distribution expected of unbound IGRs without performing an external control experiment in which an identical mixture is examined in both channels of a separate microarray. It is this null distribution that permits calculation of significance for each observed Δh value. The symmetric nature of the null distribution is an assumption of our model, and is based on our own experience and that of others [19].
Specifically, a parametric distribution is fitted by minimizing the negative log-likelihood of the data to the left of the mode (found after smoothing the data using gaussian kernel-based density estimation) [20,21]. Three possible distributions were initially considered (normal, Cauchy, and Gumbel), but the normal distribution consistently obtains the best log-likelihood score. Goodness-of-fit for the fitted normal distributions was verified with a χ2 test, and all passed with p < 10-20. The Δh scores from all replicates are standardized (centered to have zero mean and re-scaled to have unit variance) yielding a score zi = (Δhi - μi)/σi, where μi and σi represent the mean and standard deviation, respectively, of the Δh values obtained from replicate i. Figure 1a-c shows Δh distributions for three replicates [22]. We expect the distribution of Δh scores to be centered about zero; as shown by the vertical dotted lines in Figure 1, this is true to a very good approximation. Variance stabilization attempts to transform the data such that measurement error is uniform for each spot on a given array, and if replicate arrays were identical, one would expect to see the same variance in each array; large discrepancies between arrays might indicate problems with the quality of some of the arrays. Standardization is necessary to account for minor (on the order of 10%) differences in variance between arrays. Standardized scores are averaged to give an overall score (), the distribution of which is shown in Figure 1d. This distribution is again smoothed with a gaussian kernel, and fitted as described above. Finally, a p value for each IGR is computed on the score, according to the null hypothesis that all IGRs are described by this fitted normal distribution, that is, they are not bound by the TF.
Experimental verification of our dataset and evaluation of p value accuracy
The distribution of computed p values is shown in Figure 2a. It clearly shows near-ideal behavior: uniform distribution across most of the interval (0,1) arising from the vast majority of unbound IGRs, and a peak close to p = 0, arising from bound IGRs. Figure 2b shows the distribution of q values. As expected, most IGRs have a high q value, consistent with the assumption that most are unbound. False discovery rates, as represented by q values [23], are particularly useful when the goal is discovery of TF-bound IGRs. For example, the q values for Sko1 (see Additional data file 1) indicate that scientists willing to accept a list of targets in which 33% are false positives should examine the top 224 entries using a more-accurate experimental method, while those only willing to tolerate a false-positive rate of 20% should restrict themselves to the top 91.
We independently validated 35 target genes spread widely across the top 350 in our list using targeted ChIP analysis. Considering only the 35 targets for which follow-up testing was performed, ranking of IGRs by the p values of Lee et al. [2] (see Additional data file 4) shows an ability similar to our method ('Chipper') at placing true positives above false positives. When considering all IGRs, however, there is little correlation between rank by our method and rank by the Lee et al. approach. In other words, top-ranking targets by one method are not top-ranking by the other. Thus, although our validation experiments are consistent with Chipper achieving the same sensitivity at a lower false-positive rate, it is also possible that the two methods are each adept at identifying different subsets of targets. The discrepancy may be due to some systematic error in determination of the parameters of the error model. As the error model parameters are not provided explicitly with their data, we could not investigate this possibility further. Inaccurate determination of error-model parameters can lead to unjustified confidence in differences based on noisy measurements. Therefore, in the task of ranking IGRs by the likelihood of being TF-bound, Chipper is on par and complementary to the Lee et al. approach and may outperform it. Furthermore, the Chipper algorithm uses an internally determined error model and thus is not subject to systematic errors that may arise via the separate control experiments required of the methods in Lee et al. [2]. Below we show that Chipper allows increased sensitivity at a given significance threshold.
Chip2 experiments cannot distinguish the strand on which binding occurs, only the location at which it takes place. When binding is assigned to an IGR less than 2,000 nt in size, which happens to separate two genes on opposite strands, it is not possible to determine, on the basis of Chip2 alone, which one is the target of a TF. For example, as illustrated in Table 1, FAA1 and COT1 are divergently transcribed genes separated by a 1,800 nt IGR. The IGR is split into FAA1-proximal and COT1-proximal IGR segments. The primers used for targeted ChIP (about 200 nt) are smaller than the sheared fragments used in the microarray experiments (500 nt), which gives them a greater spatial resolution. As the primers are designed for a specific promoter, and amplified by polymerase chain reaction, they are strand-specific. Only FAA1 is found to bind Sko1 in a targeted ChIP experiment, yet because both IGR segments overlap Sko1-bound fragments in the Chip2 experiment, a spurious positive result is generated for COT1. We score correctly identified IGRs as true positives, even when only a single gene is verified in the targeted experiment. The Sko1 data, along with further study of Sko1 targets, are published elsewhere in the context of a focused study of Sko1 [22].
False discovery rate analysis
A common measure of significance used in hypothesis testing is the p value. In large-scale experiments like these, random chance can cause some IGRs to have p values that will be considered significant. Multiple hypothesis corrections (that is, corrections for the fact that a hypothesis is being tested multiple times, once for each IGR) are a popular approach in which the significance threshold is raised (or the p value lowered) as a function of the number of IGRs. Bonferroni-type [24] corrections are often conservative, in that many positives may be classified as non-significant ('false negatives'). This is borne out in our analysis of Sko1 Chip2 data, in which, after multiple-hypothesis correction, only a small number of IGRs (<10) were significant, at an experimentwise p value = 0.05 or lower (equivalent to p = 1.06 × 10-5 before multiple-hypothesis correction). However, the motivation of most Chip2 users is not to cautiously establish a list of binding sites that are known with near-certainty. The attraction of Chip2 is its high-throughput nature, which allows the experimentalist to rapidly generate a list of potential binding sites for subsequent study. A relatively recent alternative to the p value is the q value, which is a measure of false discovery rate (FDR) that has proven useful when the aim of an experiment is hypothesis generation rather than hypothesis testing [23,25,26]. Despite the fact that Chip2 experiments are typically used for hypothesis generation, no previously reported analysis of Chip2 experiments has employed an FDR approach. Figure 3 shows that the q values computed from our p values (broken line) agree quite well with our empirical FDR (solid line). As the first verified false positive ranks just above 100, our empirical FDR is zero to that point. Thereafter, it tracks the computed FDR quite closely until all true positives have been discovered.
Validation with publicly available datasets
We obtained the raw data used by Lee et al. [2] and compared the p values produced by our algorithm with the published p values. The 7,200 IGRs were ranked using the appropriate score for each method, and the ranked lists were evaluated for the presence of targets annotated as bound by the TF of interest in the Yeast Proteome Database (YPD) [27,28]. Data for two TFs (Ino4 and Sko1) are shown in Figure 4, and analysis of another six TFs is shown in Additional data file 5. In Figure 4a we show the receiver-operating characteristic (ROC) curve for Ino4, which tracks the sensitivity of an algorithm (its ability to find true positives (TPs)) as a function of its tendency to turn up false positives (FPs). An optimal algorithm would rank all TPs at the top. Its ROC curve would begin at the lower left-hand corner (FP = 0, TP = 0), move vertically to the upper left-hand corner (FP = 0, TP = 1), and then across the top of the chart to the upper right-hand corner (FP = 1, TP = 1). As this is a hypothesis-generation technique, only those targets near the top of a ranked list are likely to be of interest; we therefore show only the region from FP = 0 to FP = 0.1. The ranking performance of each algorithm is good in this case, and there appears little to choose between methods: either one can achieve a sensitivity of almost 1.0 with a false-positive rate of about 0.05.
In practice, however, it is common to consider only those IGRs passing a standard threshold of significance (p < 10-3 in Lee et al. [2] and Harbison et al. [8]). Therefore, we evaluated the same data, but rather than focusing on simple ranking ability, we examined the p value of each call (results for Ino4 shown in Figure 4b). We constructed the graph by choosing a significance threshold (α) and asking what fraction of the known true positives exceed the threshold (that is, have p values less than α). At α = 1, any algorithm will have perfect sensitivity because it calls all IGRs significant; this comes at the cost of specificity, as it is unable to distinguish between true and false positives. The p values reported by Lee et al. [2] are shown in green, those by our method are shown in black. The vertical dotted line indicates a threshold α7 = 10-3 at which we would expect approximately 7 out of 7,200 intergenic regions to achieve significant scores purely by chance, even if none were bound by the TF. The vertical dashed line indicates the threshold α1 = 1.6 × 10-4, which we expect to be exceeded by chance for only one out of 7,200 IGRs. The unshaded area to the right of α1 indicates the region in which fewer than one IGR would be expected to exceed the threshold by chance. The higher an algorithm's sensitivity in this region (that is, the more true positives it puts here), the better. As we decrease the threshold, the sensitivity decreases slowly at first, for both methods. For the p values of Lee et al. [2], there is then a rapid reduction in sensitivity. At an α threshold such that only one false positive is expected, our method can recover more than half the known targets while Lee et al. [2] find none.
In Figure 4c, we show an ROC curve for the transcription factor Sko1, for which nine targets are annotated in the YPD. The error model of Lee et al. [2] ranks the targets slightly better than our method of average z scores. Yet, as shown in Figure 4d, for any given significance threshold, our algorithm returns more of those targets. Ino4 showed the most striking improvement in sensitivity (Figure 4b) for all TFs examined. However, for each of the eight TFs we examined (Figure 4 and Additional data file 5) our method called an equal or greater number of targets significant at the level of α1 than did the method of Lee et al. [2]. Thus, for all TFs examined, our method yields sensitivity either markedly better than or similar to that of the de facto standard method.
Conclusions
We have developed a method for analyzing results from chromatin-immunoprecipitation/microarray (Chip2) experiments that computes p values without needing a separate control for developing a model of measurement error. The method proposed here successfully combines multiple replicates (separate arrays) and duplicates (same array) to produce a single overall p value for each IGR. By using variance stabilization rather than log ratios, we eliminate the need to threshold low-signal spots obtaining an alternative measure, Δh, which interpolates between a difference and a log-ratio and is monotonically related to significance. In addition, by averaging the resulting z score over replicates, an IGR that scores highly in a single replicate, but has no usable data in other replicates, may score well in the overall rankings. This is desirable in hypothesis generation: the algorithm should not be conservative, rather it should be sensitive and provide accurate p values by which the false positive rate can be judged. The p values produced by our algorithm behave as one would expect p values to: a broadly uniform distribution over the full range, but with enrichment near p = 0. Experimentalists can use the q values computed from these p values to generate a short list that is customized to their tolerance for false discoveries. We have evaluated our algorithm using the transcription factor Sko1 by performing targeted ChIP on 35 selected genes. Additionally, we have compared performance of our algorithm with that of a previous error model [2], using data from a public database of transcription-factor targets [28,29]. Generally, discrimination of true positives, as measured by ROC curves, is comparable for both methods. However, our method returns targets with more significant p values. We find that the observed false-discovery rate on these putative targets generally tracks that predicted by the q values, therefore validating the accuracy of the p values and q values produced by our method. To parameterize error models, the method presented here requires no external control microarray experiments (which may introduce systematic error), giving it a distinct advantage over others in current use. Software implementing the algorithm is available either in web-based form for online use, or for download by non-commercial users, from our website [30].
Materials and methods
Chip2 analysis on Sko1 was performed using three microarrays, each with duplicate spots. Genomic DNA was used as a negative control. We used targeted ChIP experiments on 35 putative targets of Sko1 to validate how well our algorithm finds TF binding sites. We selected targets distributed throughout the top-ranking 350 IGRs. Primers were specifically designed for each IGR, and each region was assayed three times both with and without the hemagglutinin (HA) epitope tag, and the results averaged. The POL1 open reading frame (ORF) and an ORF-free region were used as negative controls, since Sko1 is not expected to bind there. Each IGR was scored according to the ratio of its IP efficiency with the HA epitope tag compared to that of POL1 ORF (non-specific control). Based on prior experience, we chose a threshold of 2.0, above which we considered Sko1 to have bound to the IGR, and below which we considered it not to have bound. By this criterion, we found 21 bound IGRs, with the remaining 7 tested IGRs not bound. (The number of IGRs tested is less than the number of target genes because some IGRs are associated with more than one gene.) Of those scoring >2.0, we found that six (ICY1, HOR7, YPR127W, DPM1, POS5, and RSN1) also scored highly (above 2.0) without the tag, indicating that they bind non-specifically. In fact, only POS5 scored in the top 100 by our method. Further details on Chip2 analysis of Sko1 and validation experiments are published elsewhere in the context of a focused study of Sko1 [22]. The complete dataset is available from the Gene Expression Omnibus (GEO) [31] under series accession number GSE3335.
Additional data files
The following additional data are available with the online version of this paper. Additional data file 1 is a tab-delimited file containing the results of our analysis for all IGRs studied in our experiments. Additional data file 2 contains a detailed description of the comparison between the targets of Sko1 identified by Chipper when applied both to the data presented here and to other Chip2 data [2], and previously published p values using a single-array error model [2]. Additional data files 3 and 4 are figures illustrating these comparisons. Additional data file 5 is a figure comparing the two methods as applied to results from six additional transcription factors. Additional data file 6 lists the IGRs identified as targets [29].
Supplementary Material
Additional data File 1
A tab-delimited file containing the results of our analysis for all intergenic regions studied in our experiments.
Click here for file
Additional data File 2
A detailed description of the comparison between the targets of Sko1 identified by Chipper when applied both to the data presented here and to other Chip2 data [2], and previously published p values using a single-array error model [2].
Click here for file
Additional data File 3
A figure illustrating the comparisons made in Additional data file 2.
Click here for file
Additional data File 4
A figure illustrating the comparisons made in Additional data file 2.
Click here for file
Additional data File 5
A figure comparing the two methods described in Additional data file 2 as applied to results from six additional transcription factors.
Click here for file
Additional data File 6
A list of the intergenic regions identified as targets [29].
Click here for file
Acknowledgements
We thank J Geisberg, M Damelin, P Silver, Z Moqtaderi and J Wade for helpful discussions, and J Geisberg and J Casolari for 'beta-testing' the website and algorithm. F.D.G. and F.P.R. were supported in part by Funds for Discovery provided by John Taplin and by an institutional grant from the HHMI Biomedical Research Support Program for Medical Schools. M.P., F.D.G., and K.S. were supported by NIH/NIGMS grants GM30186, GM53720, and NIH/NHGRI grant HG003147. M.P. was supported by an EMBO Long Term Fellowship and the 'Ramón y Cajal' program of the Spanish Ministry of Science.
Figures and Tables
Figure 1 Three replicate two-channel Chip2 experiments performed on Sko1 [22] were variance-stabilized. (a-c) Distributions of the Δh values obtained. Shaded gray areas indicate kernel-smoothed densities estimated from data. Magenta curves estimate the distribution of scores expected of unbound intergenic regions (IGRs) by fitting a normal distribution to the negative Δh side of the distribution. Sufficient statistics (mean, variance) of each fitted distribution are used to standardize the Δh distributions to a score zi for each replicate. (d) The distribution of the average score over all three replicates. We computed a p value for each IGR under the null hypothesis that it is unbound, using the curve fitted to the negative portion of the empirical distribution.
Figure 2 Observed distributions of p and q values. (a) The distribution of p values for the same data as in Figure 1. They are relatively uniformly distributed on the interval (0,1), except for a slight peak close to p = 0, indicating a small fraction of specifically bound intergenic regions (IGRs). (b) Corresponding q values, but with a log scale on the vertical axis. As one descends the ranked list of IGRs the q value rapidly approaches unity. That most IGRs have q close to 1 is expected given that the list of tested IGRs is long, and the number of true targets is generally small.
Figure 3 Agreement between predicted and empirical false-discovery rate for Sko1. The broken curve shows q values computed from the ranked list of p values, using QVALUE software [32]. The solid curve shows the false-discovery rate (FDR) computed using only targeted chromatin immunoprecipitation experiments (35 targets).
Figure 4 Performance of our algorithm on publicly available Chip2 data [2] is evaluated using the Yeast Proteome Database collection of transcription factor targets [28,29] and compared with another popular means of computing p values [2]. (a) Receiver-operating characteristic curves for our method (black, 'Chipper') and that of Lee et al. [2] (green, 'Lee') using three replicate experiments for the transcription factor Ino4, made publicly available by Lee et al. (b) Sensitivity as a function of significance threshold. The broken line represents the performance of choosing potential targets at random. (c,d) Analogous curves for the transcription factor Sko1. FP, false positive; TP, true positive.
Table 1 Divergently transcribed genes, grouped in pairs of which at least one is a target of Sko1, according to a targeted ChIP assay
Gene Promoter Target?
FAA1 -827/-576 Yes
COT1 -1,743/-1,561 No
PUT4 -617/-372 Yes
CIN1 -1,007/ - No
RPI1 -606/-451 Yes
RHO3 -1,611/-1,336 No
SPO20 -449/-211 Yes
SOK2 -1,896/- No
Promoter distances are measured in nucleotides from the start codon of gene 1. Both genes of a pair are counted as positives in evaluating the algorithm described here, since distinguishing members of these pairs is beyond the resolution of Chip2 experimental technology. ChIP, chromatin immunoprecipitation.
==== Refs
Buck MJ Lieb JD ChIP-chip: considerations for the design, analysis, and application of genome-wide chromatin immunoprecipitation experiments. Genomics 2004 83 349 360 14986705 10.1016/j.ygeno.2003.11.004
Lee TI Rinaldi NJ Robert F Odom DT Bar-Joseph Z Gerber GK Hannett NM Harbison CT Thompson CM Simon I Transcriptional regulatory networks in Saccharomyces cerevisiae. Science 2002 298 799 804 12399584 10.1126/science.1075090
Iyer VR Horak CE Scafe CE Botstein D Snyder M Brown PO Genomic binding sites of the yeast cell-cycle transcription factors SBF and MBF. Nature 2001 409 533 538 11206552 10.1038/35054095
Lieb JD Liu X Botstein D Brown PO Promoter-specific binding of Rap1 revealed by genome-wide maps of protein-DNA association. Nat Genet 2001 28 327 334 11455386 10.1038/ng569
Ren B Robert F Wyrick JJ Aparicio O Jennings EG Simon I Zeitlinger J Schreiber J Hannett N Kanin E Genome-wide location and function of DNA binding proteins. Science 2000 290 2306 2309 11125145 10.1126/science.290.5500.2306
Pugh BF Gilmour DS Genome-wide analysis of protein-DNA interactions in living cells. Genome Biol 2001 2 reviews1013.1 1013.3 11305945 10.1186/gb-2001-2-4-reviews1013
Ng HH Robert F Young RA Struhl K Genome-wide location and regulated recruitment of the RSC nucleosome-remodeling complex. Genes Dev 2002 16 806 819 11937489 10.1101/gad.978902
Harbison CT Gordon DB Lee TI Rinaldi NJ Macisaac KD Danford TW Hannett NM Tagne JB Reynolds DB Yoo J Transcriptional regulatory code of a eukaryotic genome. Nature 2004 431 99 104 15343339 10.1038/nature02800
Cawley S Bekiranov S Ng HH Kapranov P Sekinger EA Kampa D Piccolboni A Sementchenko V Cheng J Williams AJ Unbiased mapping of transcription factor binding sites along human chromosomes 21 and 22 points to widespread regulation of noncoding RNAs. Cell 2004 116 499 509 14980218 10.1016/S0092-8674(04)00127-8
Hughes TR Marton MJ Jones AR Roberts CJ Stoughton R Armour CD Bennett HA Coffey E Dai H He YD Functional discovery via a compendium of expression profiles. Cell 2000 102 109 126 10929718 10.1016/S0092-8674(00)00015-5
Nehlin JO Carlberg M Ronne H Yeast SKO1 gene encodes a bZIP protein that binds to the CRE motif and acts as a repressor of transcription. Nucleic Acids Res 1992 20 5271 5278 1437546
Proft M Serrano R Repressors and upstream repressing sequences of the stress-regulated ENA1 gene in Saccharomyces cerevisiae: bZIP protein Sko1p confers HOG-dependent osmotic regulation. Mol Cell Biol 1999 19 537 546 9858577
Vincent AC Struhl K ACR1, a yeast ATF/CREB repressor. Mol Cell Biol 1992 12 5394 5405 1448073
Smyth GK Michaud J Scott H Use of within-array replicate spots for assessing differential expression in microarray experiments. Bioinformatics 2005 21 2067 2075 15657102 10.1093/bioinformatics/bti270
Huber W von Heydebreck A Sültmann H Poustka A Vingron M Variance stabilization applied to microarray data calibration and to the quantification of differential expression. Bioinformatics 2002 18 Suppl 1 S96 S104 12169536
Durbin BP Harin JS Hawkins DM Rocke DM A variance-stabilizing transformation for gene-expression microarray data. Bioinformatics 2002 18 Suppl 1 S105 S110 12169537
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
Rocke DM Durbin B A model for measurement error for gene expression arrays. J Comput Biol 2001 8 557 569 11747612 10.1089/106652701753307485
Huber W von Heydebreck A Sueltmann H Poustka A Vingron M Parameter estimation for the calibration and variance stabilization of microarray data. Stat Appl Genet Mol Biol 2003 2 3.1 3.22
Dennis JE Schnabel RB Numerical Methods for Unconstrained Optimization and Nonlinear Equations 1983 Englewood Cliffs, NJ: Prentice-Hall
Press WH Flannery BP Teukolsky SA Vetterling WT Numerical Recipes 1986 1 Cambridge, UK: Cambridge University Press
Proft M Gibbons FD Copeland M Roth FP Struhl K Genomewide identification of Sko1 target promoters reveals a regulatory network that operates in response to osmotic stress in Saccharomyces cerevisiae. Eukaryotic Cell 2005 4 1343 1352 16087739 10.1128/EC.4.8.1343-1352.2005
Storey JD The positive false discovery rate: a Bayesian interpretation and the q-value. Ann Statistics 2003 31 2013 2035 10.1214/aos/1074290335
Sokal RR Rohlf FJ Biometry: The Principles and Practice of Statistics in Biological Research 1995 3 New York: WH Freeman & Company
Benjamini Y Hochberg Y Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Ser B 1995 57 289 300
Storey JD Tibshirani R Statistical significance for genomewide studies. Proc Natl Acad Sci USA 2003 100 9440 9445 12883005 10.1073/pnas.1530509100
Payne WE Garrels JI Yeast Protein Database (YPD): a database for the complete proteome of Saccharomyces cerevisiae. Nucleic Acids Res 1997 25 57 62 9016505 10.1093/nar/25.1.57
Costanzo MC Hogan JD Cusick ME Davis BP Fancher AM Hodges PE Kondu P Lengieza C Lew-Smith JE Lingner C The yeast proteome database (YPD) and Caenorhabditis elegans proteome database (WormPD): comprehensive resources for the organization and comparison of model organism protein information. Nucleic Acids Res 2000 28 73 76 10592185 10.1093/nar/28.1.73
Csank C Costanzo MC Hirschman J Hodges P Kranz JE Mangan M O'Neill K Robertson LS Skrzypek MS Brooks J Three yeast proteome databases: YPD, PombePD, and CalPD (MycoPathPD). Methods Enzymol 2002 350 347 373 12073323
Chipper
Barrett T Suzek TO Troup DB Wilhite SE Ngau WC Ledoux P Rudnev D Lash AE Fujibuchi W Edgar R NCBI GEO: mining millions of expression profiles - database and tools. Nucleic Acids Res 2005 33 Database issue D562 D566 15608262
QVALUE: The Manual. Version 1.0
|
16277751
|
PMC1297652
|
CC BY
|
2021-01-04 16:05:35
|
no
|
Genome Biol. 2005 Nov 1; 6(11):R96
|
utf-8
|
Genome Biol
| 2,005 |
10.1186/gb-2005-6-11-r96
|
oa_comm
|
==== Front
Genome BiolGenome Biology1465-69061465-6914BioMed Central London gb-2005-6-11-r971627775210.1186/gb-2005-6-11-r97SoftwareChIPOTle: a user-friendly tool for the analysis of ChIP-chip data Buck Michael J [email protected] Andrew B [email protected] Jason D [email protected] Department of Biology and Carolina Center for Genome Sciences, CB 3280, 202 Fordham Hall, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-3280, USA2 Department of Statistics, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-3260, USA2005 19 10 2005 6 11 R97 R97 7 6 2005 2 8 2005 22 9 2005 Copyright © 2005 Buck 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.
ChIPOTle is a new software tool designed specifically for the analysis of ChIP-chip data.
ChIPOTle (Chromatin ImmunoPrecipitation On Tiled arrays) takes advantage of two unique properties of ChIP-chip data: the single-tailed nature of the data, caused by specific enrichment but not specific depletion of genomic fragments; and the predictable enrichment of DNA fragments adjacent to sites of direct protein-DNA interaction. Implemented as a Microsoft Excel macro written in Visual Basic, ChIPOTle uses a sliding window approach that yields improvements in the identification of bona fide sites of protein-DNA interaction.
==== Body
Rationale
Interactions between proteins and DNA facilitate and regulate many basic cellular functions, including transcription, DNA replication, recombination, and DNA repair. For example, the process of transcription is regulated by a class of proteins referred to as transcription factors, which often bind to specific DNA sequences upstream of gene coding regions. This control mechanism allows cells to respond to developmental or environmental signals by using the same transcription factor to coordinate expression of many genes. Therefore, it is of interest to determine where regulatory proteins of this and other types are bound to the genome.
The genomic-binding location of transcription factors can be determined using chromatin immunoprecipitation (ChIP) followed by detection of the enriched fragments by DNA microarray hybridization. This procedure, also known as ChIP-chip, has been reviewed extensively [1-5]. To appreciate the unique properties of the data generated by the ChIP-chip procedure, it is useful to review briefly the main points of the experimental procedure (Figure 1).
After growing the cells of interest under the desired conditions, chromatin is usually cross-linked with formaldehyde to preserve sites of interaction between proteins and DNA. The cross-linked chromatin is then sheared by sonication or enzymatic digestion. Shearing creates a population of chromatin fragments of varying size, generally ranging from 200 to 1,000 base-pairs. The protein of interest, along with the DNA associated with it, is then isolated by using an antibody specific to that protein or by affinity purification utilizing an epitope or affinity tag fused to the protein. The ChIPed DNA is then purified. Because yields from most samples are low, amplification is often required. DNA fragments enriched in the procedure are then detected by comparative hybridization to a DNA microarray. Standard technical recommendations common to all microarray experiments (for example, the need for dye swaps) apply equally to ChIP-chip experiments. The result of the hybridization allows one to identify which segments of the genome were bound by the protein of interest during immunoprecipitation.
The interpretation of data generated by a ChIP-chip experiment is in many respects similar to interpretation of traditional gene expression microarrays, but it differs in two important ways. First, in traditional expression experiments, each element on the microarray measures the abundance of RNA molecules of a fixed length. (Note that we shall use the term 'arrayed elements' hereafter to describe DNA fragments that are deposited on the surface of the array; the term 'probe' is sometimes used by others.) In contrast, with ChIP-chip experiments each element measures the abundance of a population of fragments of various lengths due to the effects of chromatin shearing. As a consequence, arrayed elements representing genomic regions both at the binding site and near the binding site will detect enrichment (Figure 2).
Depending on the method and degree of chromatin shearing, and the resolution of the arrayed elements, this effect produces a 'peak' of signal centered over the binding site, which may span several arrayed elements representing genomically adjacent DNA. This 'neighbor effect' is not an expected property of noise or other spuriously high ratio measurements, and thus is a source of information that can be used for analysis.
The second difference in the interpretation of ChIP-chip and traditional gene expression data is that in expression experiments, the data are two-tailed and roughly symmetric. That is, there is biological significance associated with both low and high ratio measurements, and these measurements often occur with similar frequencies. In contrast, the measurements derived from ChIP-chip experiments arise as a mixture of two distributions. The first corresponds to the population of genomic fragments specifically enriched by the ChIP, and the second corresponds to the remaining population of genomic DNA that is not ChIP enriched and therefore represents background, or noise. The observed distribution of the log2 ratios is therefore asymmetric about zero, with a distinct, positively oriented skew (Figure 3a). The left-hand side of the distribution (the negative log ratios) is approximately Gaussian, but the positive log ratios exhibit a heavier non-Gaussian tail. For the vast majority of ChIP-chip experiments, the genomic regions of biological interest will be confined to the positive side of the distribution, and the negative log ratios will arise solely from fragments that are considered to be background. Under the additional assumption that the distribution of unenriched fragments is symmetric about zero, we can estimate the distribution of background ratios using only the observed negative log ratios as a guide [6].
The type of microarray used in a ChIP-chip experiment affects how the data can be analyzed. Two array designs are typically used for ChIP-chips: tiled or promoter-specific arrays. Promoter-specific arrays generally contain a single arrayed element to represent each regulatory region of interest. These arrays are valuable when binding is known to be confined to regulatory sequences close to transcriptional start sites of the selected genes [7], but they become less powerful when binding is not as well characterized or is spread over a large genomic area. The other type, namely tiled arrays, are best suited to ChIP-chip. The term 'tiled array', or sometimes 'tiling-path array', refers to arrays containing DNA fragments designed to cover large genomic regions or whole chromosomes with few or no gaps between arrayed elements [8,9]. Tiled arrays are advantageous because they do not require prior knowledge of potential binding targets, and they allow one to utilize the 'neighbor effect' in data analysis.
In this report we describe ChIPOTle (Chromatin ImmunoPrecipitation On Tiled arrays), software created expressly for the analysis of ChIP-chip data obtained using tiled arrays, which allow us to exploit both the 'single-tail' and 'neighbor effect'. ChIPOTle uses a sliding window approach to identify potential sites of enrichment, and then estimates the significance of enrichment for a genomic region using a standard Gaussian error function. ChIPOTle is delivered as a Microsoft Excel macro written in Visual Basic, which should facilitate widespread adoption and provide a platform for custom applications. Before ChIPOTle, to our knowledge the only publicly available program designed expressly for ChIP-chip data analysis was PeakFinder [10]. ChIPOTle offers several improvements, including accurate and powerful P value estimation and improved usability. ChIPOTle is available online (Additional data file 1) [11].
The ChIPOTle algorithm
ChIPOTle first sorts the arrayed elements by genomic location. To find potential areas of ChIP enrichment, a window of user-defined size (default 1 kilobase) is then moved stepwise (user-defined step size; default 0.25 kilobase) along the tiled region. At each step the average log2 ratio for the window is calculated by taking the simple average of all ratios reported by arrayed elements that overlap with the window to any degree. The average is unweighted, and therefore it is not dependent on the proportion of the element within the window; it depends only on whether it is present or absent. The window is then moved unidirectionally along the chromosome by the step size and the same calculation is repeated for each distinct window, until the end of the chromosome is reached. The arrayed elements need not be evenly spaced or of equal lengths. ChIPOTle can be used with any genome.
As described in more detail below, the resulting sliding window averages can be represented as a graph, with genomic position on the horizontal axis and average log2 ratio on the vertical axis. In this way, genomic binding locations are represented as a series of peaks (Figure 3b). Averaging the log2 ratios of elements in a window accounts for the neighbor effect, because the peak generated by a spuriously high signal will be reduced by averaging its value with the ratios of neighboring elements, which are very unlikely also to be high purely by chance.
ChIPOTle assigns a P value to the average log ratio within each window, under the null hypothesis that the observed log ratios are independent, identically distributed, and random variables, having a Gaussian distribution with a mean of zero. The variance of the observations is estimated by the average sum of the squared negative log ratios. Under the null hypothesis, the distribution of the average log2 ratio within each window is again Gaussian, with mean zero and variance equal to the variance of a single log ratio divided by the number of elements in the window. Thus, the nominal P value for a window with average ratio w can be calculated using the standard error function (ERF) as follows:
where σ is the standard deviation for the background distribution and n is the number of microarray elements used in the window. The P values reported by ChIPOTle are corrected for multiple comparisons using the conservative Bonferroni correction. As an alternative to using a Gaussian distribution for the background, ChIPOTle can estimate the P value for a region using a permutation-based approach (Additional data file 2).
Using ChIPOTle
Detailed instructions for the installation and use of ChIPOTle are available in the read-me file that accompanies the program (Additional data file 2). Once ChIPOTle has been correctly added to the Excel Add-Ins menu or opened manually, a new menu option will appear in the Excel Tools menu. ChIPOTle must be run from an active Excel spreadsheet containing five columns: the name of each arrayed element, chromosome name, start coordinate in base-pairs, end coordinate in base-pairs, and the log2 ratio from the ChIP-chip experiment(s). The ratio values supplied to ChIPOTle can be a single measurement from a single experiment or an average, weighted average, or median of ratio values calculated from multiple replicates. When using data from multiple replicates, before combining the data each array must be appropriately normalized to remove systematic nonbiological effects that might otherwise influence the results [1]. For single channel experiments, pseudo-ratios must be created before using ChIPOTle. Pseudo-ratios may be created by dividing the intensity value at each arrayed element by the median intensity value for all arrayed elements.
Through a dialog window, ChIPOTle will ask for the location of each data column. The user will also be prompted to provide the window size, step size, and the desired technique for determining peak significance. For the latter parameter, the user can choose (1) a simple peak height cutoff; (2) assume a Gaussian background distribution for calculation of window average P values; or (3) estimate the background distribution for calculation of window average P values via a permutation-based simulation. If option 1 is selected then the user is prompted to enter the peak height; for option 2 the user is prompted to provide the significance P value cutoff; and for option 3 the user is prompted to provide the number of simulations and the significance P value cutoff to be used in the permutation analysis. Any region with a P value lower then the selected cutoff will be recorded and summarized in the "Significant Regions" and "Peaks" worksheets.
Parameter optimization
As described above, ChIPOTle has three important user-defined parameters: P value cutoff, window size, and step size. These parameters will affect the output, and can be adjusted according to the experiment and the array design. The P value cutoff should be set at a level that produces a false discovery rate with which the user is comfortable. The "Significant Negative Regions" sheet provides an empirical estimate of the number of false-positive findings for the selected P value cutoff, and so the user can use this information to estimate the false-positive rate and adjust the P value cutoff (see below). The numbers of acceptable false-positive and false-negative findings will vary depending on the goals of the study.
The next parameter to set is window size. Ideally, for a given protein-DNA interaction, one would like to capture the maximal amount of ChIP signal associated with a single binding event, and none of the noise, in a single window. Therefore, in most ChIP-chip experiments the window size should be adjusted to approximately the average shear size of the chromatin. The average shear size is suggested because the size of the window must be balanced against making it so large that noise from adjacent genomic regions is included in the measurement, and against making the window so small that data from adjacent spots is excluded, diminishing the power of windowing to utilize the neighbor effect. Although this parameter is largely independent of array platform or array resolution, slightly smaller windows may be more effective on higher resolution arrays.
Optimization of step size depends on both the array resolution and the window size. The step size should be adjusted such that it is less than half of the array resolution, with array resolution defined as the distance between the start of one arrayed element and the start of the next. Thereby, the measurement recorded at each arrayed element will be used in the calculation of at least three windows, ensuring that every arrayed element has the opportunity to be centered under a peak. Window size is also an important factor because some overlap of windows is desirable in order to detect peaks at unknown locations. Taken together, we suggest setting the step size to the maximum value that is both less than half of the array resolution and less than or equal to one-quarter of the window size. For very high-resolution arrays (less than about 50 base-pairs), step sizes smaller than the array resolution may not improve results.
ChIPOTle output
ChIPOTle creates several output sheets with the following names: SummarySheet, Significant Regions, Significant Negative Regions, Chromosomes aveP, Peaks, and Description. The SummarySheet contains all the input data used to run ChIPOTle, now sorted by chromosome and start coordinate. For each window that meets the significance criteria specified by the user, the Significant Regions sheet contains the following: chromosome assignment, center coordinate, number of independent arrayed elements within each window, and names of the arrayed elements that comprise the window. Significant Negative Regions is similar to Significant Regions, but instead it contains all of the windows that meet the significance criteria but are sign-flipped. The number of windows reported in this sheet can be used as an estimate of the number of false-positive findings expected for the selected or estimated cutoff. Chromosome aveP contains the names of the arrayed elements that comprise each window, and the chromosome, center coordinate, and value of all windows, regardless of whether they meet the significance criterion. The values from this sheet, for example, were used to make Figure 3b.
The data written to the "Peaks" sheet are similar to those reported in "Significant Regions", except that all neighboring windows meeting the significance criteria are collapsed into a single peak. Therefore, a peak is defined as any window with a P value that meets the significance criterion defined by the user and all neighboring windows that also meet the significance criteria. In this sheet, each peak is listed in order of its occurrence along the chromosome, along with the highest window for each peak, highest raw log2 ratio for any element within the peak, start coordinate of the peak, the width of portion of the peak above the significance cutoff, 'array density' of the peak, and the P value for that peak. The array density value is defined as the average number of arrayed elements used to calculate the window values for all windows that comprise the peak. Therefore, the array density value provides an estimate of the number of actual raw data measurements that underlie each peak.
The last sheet, Description, contains a summary of the ChIPOTle execution parameters, which include the date and time, the selected window size, the step size, the significance method chosen and corresponding parameters, the number of significantly enriched peaks, and the total number of windows.
Properties of ChIP-chip data
A plot of the sliding window values generated by ChIPOTle for a Rap1p ChIP-chip reveals two important characteristics of this type of data (Figure 3b). The first is an absence of deep negative peaks. In ChIP-chip experiments, negative log ratios are not caused by specific depletion of genomic fragments but by noise. Therefore, after averaging with neighboring genomic elements, their window average will tend to be small. The second is the presence of tall positive peaks that extend well above background.
Comparing ChIPOTle with other techniques used to analyze ChIP-chip data
We compared ChIPOTle with three other analysis techniques commonly used to analyze ChIP-chip experiments: the single array error model (SAEM) [6,7,12], percentile rank analysis [13], and PeakFinder (smoothing settings: n = 5, rounds = 7) [10]. All four techniques were used to analyze four biological replicates (experiments 5, 6, 8, and 9) from the Rap1p binding dataset in yeast reported by Lieb and coworkers [13]. To compare the power of the four techniques quantitatively, they were judged by their ability to identify the 127 promoters of the ribosomal protein genes (RPGs) as targets of Rap1p binding. As a group, these promoters are known targets of Rap1p, and almost all contain consensus Rap1p-binding sites [14]. By using this functionally defined set, we avoided using any particular ChIP dataset to define our 'gold standard'. The targets identified by each technique were sorted by P value (ChIPOTle and SAEM), median percentile rank (percentile rank), or ySmooth value (PeakFinder). We then used receiver operator characteristic (ROC) plots to show how true positives (sensitivity) were captured in relation to false positives (specificity) for all values output by each method (Figure 4a). The power of each technique was then quantitated as the area under the ROC curve (AUC). An analysis technique that selected targets randomly would have an AUC of about 0.5; higher values are better (maximum = 1).
In using the Rap1p ChIP-chip data to identify the promoters of RPGs, all of the techniques worked well, but ChIPOTle (Figure 4a, black line; AUC = 0.963) performed considerably better then the other techniques (SAEM: AUC = 0.906, percentile rank AUC = 0.897; PeakFinder: AUC = 0.838). The 95% confidence interval for each AUC value (Figure 4b) was estimated by bootstrap resampling of RPG occurrence and enrichment value as measured in each technique (P value, percentile rank, or ySmooth) [15].
We next compared the ability of ChIPOTle, SAEM, and PeakFinder to identify accurately the RPG promoters from a ChIP-chip hybridization to a single microarray. This analysis cannot be performed with the percentile rank analysis because this technique requires experimental replicates. We analyzed each individual experiment independently and determined the average true-positive rate versus the false-positive rate (Figure 4c). All three techniques performed extremely well, but ChIPOTle (AUC = 0.885) outperformed both SAEM (AUC = 0.835) and PeakFinder (AUC = 0.833). In addition, ChIPOTle produced higher AUC values than both SAEM and PeakFinder for each individual experiment (data not shown).
Discussion
ChIPOTle is a Microsoft Excel macro that is designed for use in the analysis of data from ChIP-chip experiments. ChIPOTle exploits the unique characteristics of ChIP-chip data, including enrichment of DNA genomically adjacent to sites of protein-DNA interaction, and the single-tailed nature of the data, to define peaks of enrichment and their significance. ChIPOTle is very quick and easy to use. The user is prompted to select the five columns containing their data and the significance technique to be used. The program then returns the genomic regions that were enriched by the ChIP according to the data and the specified statistical parameters. In its current implementation, ChIPOTle is restricted in functionality by the limitations of Excel worksheets to 65,536 rows by 256 columns. Therefore, if the dataset of interest is derived from an array containing more then 65,536 unique elements or if the total number of windows generated exceeds 5.5 million, then the data will have to be separated into subsets (for example, individual chromosomes) if they are to be analyzed using ChIPOTle.
As currently implemented, the significance analysis in ChIPOTle is carried out under the assumption that the log2 ratios of the arrayed elements are independent and Gaussian distributed, with mean zero and common variance. Under this assumption, a nominal P value may be assigned to each window using the standard Gaussian cumulative distribution function, or an appropriate bound having closed form. Multiple comparisons can then be addressed via a Bonferroni correction or through an estimated false-discovery rate. In either case, the tail behavior of the Gaussian distribution will have a strong effect on the corrected P values.
As a more conservative alternative to the Gaussian approach, one could derive nominal P values from each window using a null distribution with heavier tails than the Gaussian. A natural choice, consistent with the observed histogram of log2 ratios, is a t-type distribution. Formally, one may adopt the null hypothesis that the observed log2 ratios are independent and distributed as cT, where c is a positive scaling factor and T has a standard t distribution with v degrees of freedom. In order to obtain nominal P values, one then needs estimates of c and v, and bounds on the probability that a sum of independent t-distributed random variables exceeds a threshold. Estimates of c and v can be obtained through moment-based methods. Suitable probability bounds with good small-sample properties are currently under investigation.
ChIPOTle, while using novel approaches, identifies a set of sites similar to that defined by other techniques (PeakFinder, SAEM, and percentile rank analysis) used for analysis of data from ChIP-chip experiments. However, the use of a sliding window allows ChIPOTle to identify enriched regions more accurately, especially after only one experiment. This is useful because when one is performing a ChIP-chip experiment for the first time with a new protein or antibody, it is often difficult to determine whether the ChIP was successful, especially for a protein with an undefined binding pattern. The ability to determine binding sites correctly using fewer replicates will be very important for larger, more complex genomes. Complete high-density tiled arrays for mammalian genomes require many arrays for each experiment, meaning that performing the ideal number of replicates can be prohibitively expensive. In mammalian systems, instead of performing all of the replicates of a ChIP-chip experiment on whole-genome arrays, preliminary experiments using whole-genome arrays can be used to find likely targets. Once these likely targets are identified, the array could be redesigned to include all prospective targets and appropriate controls on a single array. In addition to its utility as a general ChIP-chip analysis tool, ChIPOTle will make prescreening more accurate and will enhance the power and accuracy of this approach.
Additional data files
The following additional files are included with the online version of this paper: The Excel Add-In ChIPOTle v 1.0 (Additional data file 1), a pdf file containing detailed instructions for the installation and use of ChIPOTle (Additional data file 2), and an Excel file containing the Rap1p binding data used to make the comparisons between the different techniques (Additional data file 3).
Supplementary Material
Additional data file 1
An Excel Add-In ChIPOTle v 1.0.
Click here for file
Additional data file 2
A PDF file containing detailed instructions for the installation and use of ChIPOTle.
Click here for file
Additional data file 3
An Excel file containing the Rap1p binding data used to make the comparisons between the different techniques.
Click here for file
Acknowledgements
This work was supported by NIH grants to M.J.B. (F32HG002989) and J.D.L. (R01GM072518) and by an NSF grant to A.B.N. (DMS-0406361).
Figures and Tables
Figure 1 A summary of the ChIP-chip procedure. See the text for details.
Figure 2 The neighbor effect and calculation of P values. (a) After ChIP, purified DNA fragments bound by the protein of interest will be of various lengths. (b) Actual log2 ratios reported by arrayed elements for Rap1p binding to promoter region of RPL1B (array element 'A') from the Rap1p binding dataset reported by Lieb and coworkers [13]. Arrayed element 'A' contains the actual site of protein-DNA interaction, and so this spot will have the highest ratio (red = high positive ratio; yellow = low ratio; green = negative ratio). Arrayed elements 'B' (RPL1B open reading frame [ORF]) and 'C' (MRM2 ORF), which are within about 1 kilobase (kb) of the binding site, are also enriched above noise. Arrayed element 'B' has a higher ratio then spot 'C', because the binding site is located closer to element 'B'. The arrayed elements 'D', 'E', 'F', and 'G' are too far from the binding site to be enriched. (c) Using a 1 kb window with a 0.25 kb step, the value of each window is plotted. The location of each window is defined by its central coordinate. (d) The P value of each window is plotted. The Bonferroni corrected P values were calculated based on the observed data, which had a log2 background standard deviation of 0.32 with 21,208 comparisons. Note that the window with the smallest P value (about 10-30) does not correspond to the highest window average. This is due to the fact that the most significant window contains three arrayed elements (A, B, and C), whereas the windows with the highest average contain only two elements (A and B). In this case, the center of the window with the highest P value is located about 80 bases from the actual binding site.
Figure 3 Characteristics of ChIP-chip data. (a) A quantile-quantile plot (QQ plot) for one representative Rap1p ChIP-chip experiment (red) against Gaussian distribution with a standard deviation of 0.35 and a mean of 0 (black bars). The upper and lower bounds of the black dashed line represent extreme values for 10,000 simulated Gaussian distributions with the above parameters. For Rap1 about 92% of the data fit the Gaussian distribution. The top 8% is skewed away from the simulated data. (b) A sliding window analysis for yeast chromosome VI produced by ChIPOTle for four Rap1p replicates [13]. Window size is 1 kilobase (kb) with 0.25 kb step size. The Rap1p binding sites are identified with arrows.
Figure 4 Comparison of ChIPOTle with other ChIP-chip analysis approaches. (a) ChIPOTle, the single-array error model (SAEM), median percentile rank, and PeakFinder were used to analyze the same four Rap1p ChIP-chip replicates reported by Lieb and coworkers [13], and judged by their ability to determine enrichment of ribosomal protein gene (RPG) promoters. The binding site for Rap1p is found in most (>90%) RPG promoters [14], which represent approximately half of Rap1p's total in vivo targets. Receiver operating characteristic (ROC) curves summarize the power of each technique and are equivalent to a plot of the true-positive rate (fraction of ribosomal promoters) versus the false-positive rate (fraction of all genomic elements other than ribosomal promoters). Each technique is judged by means of the area under the ROC curve (AUC). An AUC value of 0.5, corresponding to a diagonal ROC curve, is expected by chance, whereas a value of 1.0 indicates a technique that predicts targets perfectly. ChIPOTle (AUC = 0.963) outperformed the other techniques tested here (SAEM: AUC = 0.906; median percentile rank: AUC = 0.897; and PeakFinder: AUC = 0.823). When comparing ChIPOTle with PeakFinder, we used the default settings for smoothing (n = 5 [11-point] smoothing with 7 rounds). In addition, we attempted to optimize the settings by trying varying levels of smoothing, including 7-point and 13-point, which produced similar results. Rap1p's strongest binding sites are located at the telomeres, which are not included with our defined 'true positive' set of RPG promoters. Therefore, the false-positive rate will be somewhat inflated, which will decrease the AUC for all techniques. This is reflected in the ROC curves by the low true-positive rate at the extreme left of the plot. (b) The 95% confidence interval for the AUC for each analysis technique was estimated by bootstrap resampling of RPG occurrence and enrichment value (1,000 iterations) as measured in each technique (P value, percentile rank, or ySmooth). Boostrapping of raw data was not practical because of inability to automate all four analysis methods. (c) ROC curves comparing ChIPOTle, SAEM, and PeakFinder with respect to their ability to identify enrichment of RPG promoters from a single experiment. The average true-positive rate (fraction of ribosomal promoters) versus false-positive rate (fraction of all genomic elements other than ribosomal promoters) for the four individual experiments is plotted. The three techniques performed extremely well, but ChIPOTle (AUC = 0.885) outperformed both SAEM (AUC = 0.835) and PeakFinder (AUC = 0.833).
==== Refs
Buck MJ Lieb JD ChIP-chip: considerations for the design, analysis, and application of genome-wide chromatin immunoprecipitation experiments. Genomics 2004 83 349 360 14986705 10.1016/j.ygeno.2003.11.004
Kurdistani SK Grunstein M In vivo protein-protein and protein-DNA crosslinking for genomewide binding microarray. Methods 2003 31 90 95 12893178 10.1016/S1046-2023(03)00092-6
Wells J Farnham PJ Characterizing transcription factor binding sites using formaldehyde crosslinking and immunoprecipitation. Methods 2002 26 48 56 12054904 10.1016/S1046-2023(02)00007-5
Lieb JD Genome-wide mapping of protein-DNA interactions by chromatin immunoprecipitation and DNA microarray hybridization. Methods Mol Biol 2003 224 99 109 12710669
Hanlon SE Lieb JD Progress and challenges in profiling the dynamics of chromatin and transcription factor binding with DNA microarrays. Curr Opin Genet Dev 2004 14 697 705 15531167 10.1016/j.gde.2004.09.008
Li Z Van Calcar S Qu C Cavenee WK Zhang MQ Ren B A global transcriptional regulatory role for c-Myc in Burkitt's lymphoma cells. Proc Natl Acad Sci USA 2003 100 8164 8169 12808131 10.1073/pnas.1332764100
Lee TI Rinaldi NJ Robert F Odom DT Bar-Joseph Z Gerber GK Hannett NM Harbison CT Thompson CM Simon I Transcriptional regulatory networks in Saccharomyces cerevisiae. Science 2002 298 799 804 12399584 10.1126/science.1075090
Kapranov P Cawley SE Drenkow J Bekiranov S Strausberg RL Fodor SP Gingeras TR Large-scale transcriptional activity in chromosomes 21 and 22. Science 2002 296 916 919 11988577 10.1126/science.1068597
Horak CE Mahajan MC Luscombe NM Gerstein M Weissman SM Snyder M GATA-1 binding sites mapped in the beta-globin locus by using mammalian chIp-chip analysis. Proc Natl Acad Sci U S A 2002 99 2924 2929 11867748 10.1073/pnas.052706999
Glynn EF Megee PC Yu HG Mistrot C Unal E Koshland DE DeRisi JL Gerton JL Genome-wide mapping of the cohesin complex in the yeast Saccharomyces cerevisiae. PLoS Biol 2004 2 E259 15309048 10.1371/journal.pbio.0020259
ChIPOTle: a user-friendly tool for the analysis of ChIP-chip data
Hughes TR Marton MJ Jones AR Roberts CJ Stoughton R Armour CD Bennett HA Coffey E Dai H He YD Functional discovery via a compendium of expression profiles. Cell 2000 102 109 126 10929718 10.1016/S0092-8674(00)00015-5
Lieb JD Liu X Botstein D Brown PO Promoter-specific binding of Rap1 revealed by genome-wide maps of protein-DNA association. Nat Genet 2001 28 327 334 11455386 10.1038/ng569
Lascaris RF Mager WH Planta RJ DNA-binding requirements of the yeast protein Rap1p as selected in silico from ribosomal protein gene promoter sequences. Bioinformatics 1999 15 267 277 10320394 10.1093/bioinformatics/15.4.267
Efron B Gong G A leisurely look at the bootstrap, the jackknife, and cross-validation. Am Stat 1983 37 36 48
|
16277752
|
PMC1297653
|
CC BY
|
2021-01-04 16:05:36
|
no
|
Genome Biol. 2005 Oct 19; 6(11):R97
|
utf-8
|
Genome Biol
| 2,005 |
10.1186/gb-2005-6-11-r97
|
oa_comm
|
==== Front
BMC BiochemBMC Biochemistry1471-2091BioMed Central London 1471-2091-6-231627114510.1186/1471-2091-6-23Research ArticleA strategy using NMR peptide structures of thromboxane A2 receptor as templates to construct ligand-recognition pocket of prostacyclin receptor Ruan Cheng-Huai [email protected] Jaixin [email protected] Ke-He [email protected] From the Vascular Biology Research Center and Division of Hematology, Department of Internal Medicine, The University of Texas Health Science Center, Houston, 6431 Fannin St., Houston, Texas 77030, USA2005 4 11 2005 6 23 23 19 5 2005 4 11 2005 Copyright © 2005 Ruan et al; licensee BioMed Central Ltd.2005Ruan 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:
Prostacyclin receptor (IP) and thromboxane A2 receptor (TP) belong to rhodopsin-type G protein-coupling receptors and respectively bind to prostacyclin and thromboxane A2 derived from arachidonic acid. Recently, we have determined the extracellular loop (eLP) structures of the human TP receptor by 2-D 1H NMR spectroscopy using constrained peptides mimicking the individual eLP segments. The studies have identified the segment along with several residues in the eLP domains important to ligand recognition, as well as proposed a ligand recognition pocket for the TP receptor.
Results:
The IP receptor shares a similar primary structure in the eLPs with those of the TP receptor. Forty percent residues in the second eLPs of the receptors are identical, which is the major region involved in forming the ligand recognition pocket in the TP receptor. Based on the high homology score, the eLP domains of the IP receptor were constructed by the homology modeling approach using the NMR structures of the TP eLPs as templates, and then configured to the seven transmembrane (TM) domains model constructed using the crystal structure of the bovine rhodopsin as a template. A NMR structure of iloprost was docked into the modeled IP ligand recognition pocket. After dynamic studies, the segments and residues involved in the IP ligand recognition were proposed. A key residue, Arg173 involved in the ligand recognition for the IP receptor, as predicted from the modeling, was confirmed by site-directed mutagenesis.
Conclusion:
A 3-D model of the human IP receptor was constructed by homology modeling using the crystal structure of bovine rhodopsin TM domains and the NMR structures of the synthetic constrained peptides of the eLP domains of the TP receptor as templates. This strategy can be applied to molecular modeling and the prediction of ligand recognition pockets for other prostanoid receptors.
G protein-coupled receptorGPCRprostacyclin (prostaglandin I2 (PGI2) receptorprotein modelingthromboxane A2 receptorNMR structuresynthetic peptide.
==== Body
Background
Prostanoids including thromboxane A2 (TXA2) and prostaglandins D2 (PGD2), E2 (PGE2), F2 (PGF2) and I2 (PGI2) act as local hormones in the vicinity of their production sites to regulate hemostasis and smooth muscle functions, which are mediated by specific prostanoid receptors in the plasma membrane. The receptors are classified into five basic types based on their cognate prostanoid (PGD2, PGE2, PGF2α, PGI2 or TXA2) and are termed DP, EP, FP, IP and TP receptors, respectively [1]. In addition, the EP receptors are subdivided into four subtypes (EP1, EP2, EP3 and EP4) based on their different signaling responses to PGE2. Human TP was first purified from a platelet in 1989 and its cDNA was cloned from placenta in 1991 [2,3]. Other prostanoid receptor cDNAs, including those for DP, EP1, EP2, EP3, EP4, FP and IP have also been cloned, and their primary biologic functions have been identified.
Prostanoid receptors can be divided into two functional groups using smooth muscle assays: relaxant receptors (including DP, EP2, EP4, and IP) and excitatory receptors (including EP1, FP, and TP). All known prostanoid receptors belong to the rhodopsin-type G protein-coupled receptor (GPCR) super-family, which has seven conserved TM domains. The diverse and/or opposite biological functions of the individual prostanoid receptors involve selective ligand binding on their extracellular and/or TM domains and the interaction with different heterotrimeric G proteins on their intracellular domains [1].
For over a decade, structural and functional studies of prostanoid receptors have focused on the identification of the specific sites for ligand binding and G protein coupling. The homology alignment-based mutagenesis studies of the receptor TM domains have suggested that the conserved regions in the third and seventh TM domains are involved in binding the core structures of the prostanoids, which consist of a carboxylic acid, a hydroxyl group on carbon 15 and two aliphatic side chains [4-6]. However, the TM domains belong to the conserved regions in all known prostanoid receptors, and therefore, probably do not define the differences among the receptors in specific interactions with their ligands. Current focus has been directed toward the involvement of the extracellular loops (eLPs) of prostanoid receptors in the selective ligand recognition, similar to some other GPCRs [7-15]. However, interpretation of the eLP specific determinants response to the specific ligand binding among prostanoid receptors is limited by the lack of experimental 3-D structural information for the extramembrane domains. Structural characterization of both the extracellular and intracellular domains of the prostanoid receptors needed to elucidate the molecular mechanisms of specific ligand recognitions and G protein couplings remains a major challenge.
Molecular modeling has been widely used to create a working model of the GPCR using a known x-ray crystal structure of rhodopsin. Electron diffraction and electron microscopy studies have succeeded in determining the structure of the GPCR-like protein, bacteriorhodopsin (BR), at low and medium resolution [16,17]. The high-resolution 3-D structure of BR has also been determined by X-ray crystal studies [18,19]. Using the BR structure as a template, a number of working models of the conserved TM domains for the rhodopsin-type GPCRs have been constructed by homology modeling [6,20-32]. Most of these models assumed that the GPCRs had the same spatial arrangement for the seven helices as BR. However, it has recently been suggested that BR is not a suitable template for the construction of GPCR TM models because BR does not function as a GPCR and there is no overall significant sequence similarity between BR and the GPCRs [33]. Further evidence is given from the structural maps of bovine, frog and squid rhodopsins [34-39], which indicate that the arrangement of the helices in the rhodopsins is indeed different from that in BR. As such, it is believed that the crystal structure of the TM domains of bovine rhodopsin (bRH) is a more suitable template for constructing working models of mammalian GPCR TM domains, which include the prostanoid receptors.
It is almost impossible to build the structural extra- and intra-cellular loop models for the prostanoid receptors using the corresponding structures of BR or bRH due to a number of differences, including size and sequence variation. However, understanding the structural features of the extracellular loops of the prostanoid receptors is the key step toward uncovering the ligand binding selectivity. To overcome the difficulty recently, we have successfully determined the solution structures of the individual eLP domains for the human TP receptor, and identified that its ligand (antagonist) recognition pocket is mainly formed by the second eLP (eLP2) and the disulfide bond between the eLP2 and the first eLP (eLP1). This was done using newly-developed concepts including, computation-guided constrained peptide synthesis and NMR experiment-guided mutagenesis approaches. These results offer a basis for understanding the ligand recognition sites for the other prostanoid receptors through the homology modeling approach using the NMR structures of the eLP segments, which share many similarities with the other prostanoid receptors, especially the IP receptor. In this paper, we describe the approaches used to construct the working model for the IP receptor by homology modeling using x-ray TM structures of bRH and NMR structures of the eLPs of the TP receptor as templates. We also discuss its value in predicting the residues involved in ligand recognition for the IP receptor through modeling and mutagenesis.
Results
Homology modeling for the seven TM domains of the human IP and TP receptors using the crystal structure of the TM domains of the BR as a template
The amino acid sequences of the human IP and TP receptors and bRH were downloaded from Gene Bank. The first step is to highlight the multiple sequence alignments between the TM domains of bRH, TP, and IP receptors. The putative TM domains of the IP and TP receptors are based on the hydropathy analysis described in the original cloning papers [3,40]. The significant similarities (data not shown) in the TM domains reflect the similarity of the general backbone structures between the TP and IP receptors with bRH. To construct the homology models of the TM domains of the IP and TP receptors, the crystal structures of the TM domains of bRH were used as a template, and its backbone conformations were adopted for the each TM domains of the IP and TP receptors using the Insight II package on a SGI Octane workstation. The well-developed commercial software, Insight II, has been widely used in various academic fields for protein modeling. The detailed steps have been described previously [41,42]. The backbone structural model of the TM domains of the IP (Figure 1) and TP (data not shown) were obtained. It shall be indicated that the gaps between the template and objects were fixed by the molecular annealing computation.
Figure 1 Homology modeling of the seven TM domains of human IP receptor. 3-D backbone structure of the seven TM domains of the human TP receptor were created by homology modeling using the crystal structural backbones of the TM domains of the bovine rhodopsin as templates.
Structural features of the reported 3-D NMR structures of the eLPs of the TP receptor
Synthetic peptides corresponding to the eLP domains of the human TP receptor were mimicked by a constrained peptide technique with the N- and C-termini connected by a homocysteine disulfide bond [43,44]. The constrained eLP2 peptide was likely able to adopt an active conformation, which was evidenced by the binding of the peptide to the receptor ligand using fluorescence and CD spectroscopic studies [43-45]. Recently, by way of 2-D 1H NMR experiments, complete 1H NMR assignments for the 2-D spectra including, nuclear Overhauser effect correlation spectroscopy (NOESY), total correlation spectroscopy (TOCSY), double-quantum-filtered correlation spectroscopy (DQF-COSY), and structural construction were used to determine the overall 3-D structures of the constrained peptides mimicking the three TP eLPs (eLP1, [43], eLP2 [44] and eLP3 [45]). All of the NMR structures indicate the presence of β-turns in the loops [43]. The distance between the N- and C-termini of the peptides shown in the NMR structures was 10–14.2 Å, which matched the distance between the two TM helices connecting the eLPs in the TM domain model of the TP receptor. These structural features allowed the NMR structures of the constrained TP eLP peptides to be grafted onto the region of the TP receptor model in a configuration without further modifications [43,45]. This study suggests that the approach using the constrained loop peptide greatly increases the likelihood of characterization of the structural features of the extracellular domains of the TP receptor. It offers a structural template to model the extracellular domains of other prostanoid receptors, which as of yet have no defined crystallographic structures. The IP receptor is one example which shares significant similarities in its extracellular domains with those of the TP receptors.
Sequence alignment of the three eLPs between the human IP and TP receptors
Sequence analysis has indicated that the amino acid residues in the eLP domains are not conserved between bRH and prostanoid receptors, such as the IP receptor. This has limited the use of the crystal structure of the eLP domains of bRH for homology modeling of the prostanoid receptors. Thus, the experimental NMR eLP structures of the TP receptor become useful in constructing a model of the IP receptor because they are conserved in the eLP domains. The eLP domains of the IP and TP receptors were identified from the hydropathy analysis described by the original cloning papers [3,40]. The three eLPs between the TP and IP were aligned and are shown in the Figure 2. The lengths of the three IP eLP domains are similar to those of the TP receptor, which allows them to align with no major gaps. The Cys residues (Cys105 in the eLP1 and Cys183 in the eLP2) of the TP receptor that form a disulfide bond [43], thereby controlling the conformation of the eLPs, are conserved in the IP receptor (Figure 2). Among the three eLPs, the most conserved region is localized in the eLP2 (40% identity, Figure 2), which has been identified to play a major structural role in forming the ligand recognition site for the TP receptor [10]. In addition, there are 33% and 36% of the sequence homologies for the eLP1 and eLP3, respectively (Figure 2). Thus, using the NMR structures of the TP eLPs as a template for modeling the IP eLPs will be useful in rationalizing the structural and functional features of the eLPs in the IP receptor. The individual similarities of the alignment are listed in the Figure 2.
Figure 2 Sequence alignment of the eLPs between the human IP and TP receptors. The sequences of the putative eLP domains of the human IP (45) and TP (3) receptors were aligned by a sequence alignment program in Insight II software package and manual adjustment. The identical and highly similar residues between the TP and IP are shaded. The residues in the eLP2 regions previously identified important to the ligand binding for the TP receptor (10) and their corresponding residues in the IP receptor are underlined.
Homology modeling for the eLP domains of human IP receptor
The backbone structures of the individual IP eLPs were constructed by homology modeling using the NMR structures of the TP eLPs as templates with respect to the disulfide bond formation between Cys92 in the eLP1 and Cys170 in the eLP2 of the IP receptor (Figure 3). After energy minimization, each eLP structure was configured to the TM domain model. Each configuration was made based on the three following considerations: 1) placing the N- and C-termini of the eLP1, eLP2 and eLP3 to the C- and N-termini of the TM2 and TM3, TM4 and TM5, and TM6 and TM7; 2) matching the distances of the N- and C-termini of the eLPs with the corresponding distance between those of the TM helix (Figure 4A), 3) connecting the eLPs to the TMs through chemical bonds by Insight II calculation using Discover program (Figure 4B), and 4) forming a disulfide bond between Cys92, at the end of the eLP1, and Cys170, in the center position of the eLP2 (Figure 4B). After completing the configuration, 500 steps of energy minimization were used to refine the conformation of the model eLPs. As shown in Figure 4B, the constructed eLPs could be fitted into the conserved TM domains with respect to the cysteine disulfide bond between eLP1 and eLP2, without major violations in the structural calculation or significantly altering the original backbone structures in the original conformations before the connection (Figure 4A). Dynamic studies for the parent NMR structures of the TP eLPs have been described in our recent publication [43], in which 20 structural conformations have been generated and used to evaluate the loop folding. Limited conformational changes (rmsd = 1.2 Å) were observed [43]. The modeled structures of the IP eLPs described above adopted a similar variation of the conformations (data not shown). This information indicated that the folding of the IP eLPs is in a reasonable conformation.
Figure 3 The predicted backbone structures of the three eLP domains of the human IP receptor. The 3-D structural models of the eLPs were constructed by homology modeling with the NMR structures of the eLP2 (48), the eLP3 (49) and the eLP1 (unpublished data) of the human TP receptor as templates using the molecular modeling package of Insight II and Discover software packages. The conformation of the eLP2 is placed in a position with respect to the formation of a disulfide bond between the Cys92 in the eLP1 and Cys170 in the eLP2.
Figure 4 Configuration of the modeled 3-D backbone structures of the three IP eLP domains (Figure 4) onto the working model of the seven TM domains of the IP receptor. Before (A) and after (B) the connections of the eLP structures to the seven TMs through chemical bonds are displayed for their comparison.
Ligand docking in the IP receptor model
To understand ligand selectivity and the action mechanisms of the IP receptor, it is very important to identify its specific ligand recognition/binding site. The IP receptor was cloned a decade ago, many attempts have been made to localize the segments and residues important to its ligand binding activities. However, little structural information is available. Through molecular modeling, we have made attempts to solve this problem.
The IP receptor ligand, prostacyclin, is not stable in solution and does not have 3D structural information. Recently, we have determined the solution structure of a stable IP receptor agonist, iloprost, using NMR spectroscopy [42]. It offers an experimental structure for molecular docking between the ligand and the IP receptor. On the other hand, based on the NMR experimental and mutagenesis results obtained for the TP receptor, the four residues (Val176, Leu185, Thr186 and Leu187) identified as being involved in ligand recognition are mainly localized in the eLP2 region [43]. The results indicated that perhaps the corresponding residues in the conserved IP eLP2 region are also likely to be involved in ligand recognition. This hypothesis has been supported by our recent NMR spectroscopic studies using a synthetic eLP2 fragment [42]. This information provided the basis for our study of the interaction between the IP receptor and its ligand using a molecular docking approach. The NMR structure of iloprost was docked into the pocket of the IP model corresponding to the ligand recognition pocket of the TP receptor as identified by NMR and mutagenesis studies [10] using Insight II and Discover computation procedures. The residues crucial to the TP ligand recognition [10] were used as a basis for localizing the possible residues involved in ligand contact in the IP receptor model (Figure 5). The orientation of the iloprost structure was positioned in contact with the side chains of the three of the four residues (Gln162, Leu172, Arg173 and Meth174, Figure 2 and 5) in the IP eLP2, corresponding to the previously identified contact sites among the residues of the TP eLP2 at Val176, Leu185, Thr186 and Leu187 with SQ29,548 (a TP receptor ligand) (10). In addition, it has been found that the residues Ala177 and Gln178 in the IP eLP2 fragment could interact with iloprost by our NMR spectroscopic studies [42]. It has also been taken for the consideration of the iloprost docking with the IP receptor (Figure 5). The structural complex of iloprost and the IP receptor was then subjected to energy minimization to find the best-fitting position of the iloprost conformation in the proposed ligand recognition site of the receptor (Figure 5). During the minimization, the main changes of the structures were the side chains of the loops, but no significant structural alternation for iloprost and the loop backbones were observed.
Figure 5 Ligand docking with the eLP domains of the IP receptor. The four residues including Gln162, Leu172, Arg173 and Met174 in the IP eLP2 (eLP1, bleu color; eLP2, red color and eLP3, yellow color) in contact with iloprost were predicted by the sequence alignment (Figure 1) using the identified four residues (Val176, Leu185, Thr186 and Leu187) in the TP eLP2 contacted with SQ29,548 (10) as a template. 3-D NMR structure of the IP receptor agonist, iloprost (42) was docked into the putative ligand recognition pocket formed by the three eLPs with respect to the contacts with Leu172, Arg1173 and Met174 in the opening of the pocket. In addition, the two residues, Ala177 and Gln178 involved in contacts with iloprost predicted by NMR spectroscopic studies (42) were also used as constraints for the iloprost docking to the recognition pocket. The configuration of the model was minimized using 1,000-step energy minimization after the iloprost was docked into the pocket. The TM domains of the IP receptor are showed with green colors.
Recently, we have published the structural information of the ligand recognition site of TP receptor [43]. The disulfide bond involved in forming the ligand recognition pocket in the TP receptor has been confirmed by the reducing of the disulfide bond in the presence and absence of the ligand binds and the Cys mutations [43]. From this IP receptor model, we have also learned that the disulfide bond between the eLP1 and eLP2 is involved in forming the ligand recognition pocket (Figure 5). Reduction of the disulfide bond will alter the active conformation of the pocket of the receptor.
Confirmation of the NMR experiment-based modeling for prediction of the residues involved in IP receptor ligand recognition using recombinant IP receptor
Distinct characteristics are noted among the four predicted residues important to ligand recognition for the IP receptor. Leu172 is both conserved between IP and TP receptors. Gln162 does not lie within close proximity of the opening of the pocket (Figure 5), and the side chain of Met174 is likely not centered enough to make contact with the ligand (Figure 5). Thus in looking at the model, Arg173 becomes the residue with the highest potential for the specific iloprost recognition (Figure 5). Based on this hypothesis, the residue, Arg173, was subjected to mutagenesis studies. First, the Arg residue in the IP eLP2 was replaced with Ala to eliminate the charged side chain. After transfection of the mutated IP receptor cDNA into COS-7 cells, a similar expression level of the mutant and wild type IP receptors was confirmed by Western blot (Figure 6A). The binding of the recombinant receptor to its ligand was then performed using [3H]-iloprost and unlabeled (cold) iloprost was used as a competitive ligand (Figure 6B). The mutant with the single replacement of the Arg residue with Ala lost its binding activity to the receptor agonist as compared with the wild type (Figure 6B). These data indicate that the Arg residue, as predicted from the IP model, is indeed important in ligand recognition. The results also confirm that the model is indeed useful in providing information for structure and function relationship studies of the IP receptor.
Figure 6 Analysis for the mutation of Arg173 to Ala residue of the recombinant human IP receptor. A). Western blot analysis. Fifty micrograms of COS-7 cells transfected with wild-type (WT) or a mutant IP receptor (R173A) cDNA was subjected to SDS-PAGE and transferred onto a nitrocellulose membrane. The membrane was probed with rabbit anti-IP peptide antibody. (B) The ligand binding activities of wild-type and mutant TP receptor. 300 μg of the protein prepared from the COS-7 cells transfected with cDNA of the wild-type (WT) or the R173A mutant was incubated with 4 nM [3H]-iloprost (30,000 cpm) in the absence or presence of unlabeled iloprost (1 μM) in a reaction volume of 100 μl. After 1 h incubation, the reaction was stopped and the binding activity of the recombinant IP receptor was measured as described in the methods. The binding activity of wild-type receptor was considered as 100% (2,000 cpm).
To further identify the specificity of the Arg residue in the IP receptor, the Arg173 was then mutated to Thr that lies in the corresponding position of the TP receptor (Figure 1). After expression of the mutant in COS-7 cells, confirmed by Western blot (Figure 7A) as described above, the binding of the mutated IP receptor to iloprost was tested. The Arg173Thr mutant retained about 40% ligand binding activity as compared with the wild type IP receptor (Figure 7B). In contrast, the control mutant of the IP receptor, Ser168Thr, which is highly conserved in the all of the prostanoid receptors, retained full activity binding to iloprost (Figure 7B). The impairment of the Arg173 mutants binding to iloprost was further concluded by kinetic studies as shown in Figure 8. These results indicate that Arg173 is specifically important for ligand recognition in the IP receptor. The Arg residue in the IP and Thr residue in the TP may be involved in the determination of their ligand selectivities. If so, this will provide important clues for further characterizing ligand selectivity of other prostanoid receptors using the modeling and mutagenesis approaches.
Figure 7 Analysis of the mutation of Arg173 to Thr residue and Ser168 to Thr residue of the recombinant IP receptors. A). Western blot. B). Ligand binding activity. The methods are described in the Figure 6.
Figure 8 Kinetic properties of [3H] iloprost binding to the recombinant IP receptors expressed in COS-7 cells. The cell membrane protein prepared from COS-7 cells that transiently expressed the wild-type (squares), R173A mutant (circles) or R173T mutant (triangles) of the IP receptor was incubated with the increasing concentration of the [3H] iloprost.
Discussion
The lack of the 3-D structural models of the prostanoid receptors has become a major obstacle in both further understanding their molecular mechanisms and in the design of pharmacological intervention strategies. Thus, developing useful approaches for further structural and functional characterization of the GPCRs is crucial. The crystal structure of rhodopsin offers a structural template for the conserved TM helices of other GPCRs, including the prostanoid receptors. The TM domain modeling for the IP and TP receptors, using the x-ray structures of the TM domains of bRH described in the paper is also suitable for the modeling of the TM domains of other prostanoid receptors. However, as described above, the bRH crystal structure provides few structural and functional clues for the extracellular and intracellular domains of the prostanoid receptors. One useful way to characterize the GPCR functions is to assemble information obtained from studies using receptor fragments. Synthetic peptides have been used as important tools in mimicking the functional domains of GPCRs. Peptides corresponding to the C-terminal extramembrane domains of the angiotensin II AT1A receptor [46], natriuretic peptide receptor C [47], testicular follicle stimulating hormone receptor [48] and BR [49,50] have functional activities, which indicate that these peptides can adopt similar structures in the cognate parts of the intact receptors. The synthetic peptides corresponding to the intracellular domains of the M4 subtype muscarinic, cholinergic and α 2-adrenergic receptors could directly bind to and activate their specific G proteins [51]. Also, Yeagle et al have used the NMR structures of the synthetic peptides getting 3D structural information for the intracellular loops of bRH before its crystal structure was available [52].
However, the synthetic peptide studies, giving only fractional information of the interested proteins, have limited use in detailed structural and functional characterization of the interested proteins. Our group has recently been focusing on developing a link between peptide and protein studies into one system to further enhance our ability to characterize the structure/function relationship of proteins. Integrated high-resolution NMR techniques with synthetic peptide and recombinant protein approaches have led to the development of "computation-guided constrained peptide synthesis" and "NMR-experiment-guided mutagenesis" for aiding in the structural and functional characterization of the TP receptor. Interestingly, protein-modeling using the TP eLP NMR structures of other prostanoid receptors has become a new exploration. It is particularly important since it is not likely that crystal structures will available for any prostanoid receptors in the near future, and the high resolution NMR structural determination for the membrane-bound receptor proteins is unlikely to be solved any time soon, and in addition, the NMR instrumentation necessary for protein structural determination is not available in many labs.
In general, the location of the ligand recognition site of the IP receptor is likely similar to the TP receptor because it is known that all prostanoids have cross-binding activities to their receptors. In addition, it is also supported by the fact of that TP eLP2 important to ligand recognition is highly conserved in IP (Figure 2). The achievement of this paper in regards to the successful modeling of the eLPs of the IP receptor using the NMR structures of the TP eLPs has supported the hypothesis in which NMR structures can be applied to the modeling of other prostanoid receptors and use them as a working model for the prediction of ligand recognition sites in general. It is particularly important to note that the modeling has also been tested by key residue mutagenesis using the recombinant IP receptor protein. The remaining work, including further mapping the residues involved in the ligand recognition in the eLPs for the IP receptor by mutagenesis analysis, is currently under progress in our laboratories.
It shall also be noted that in our previous publication [42], the prediction of the residues, Ala177 and Gln178, in the eLP2 important to ligand recognition was supported by our NMR structural studies using synthetic peptide. These two residues are likely not enough to cover the pocket. In this manuscript the additional residues are predicted from the modeling studies based on the NMR structural model of the extracellular loops of the TP receptor. In combination of the two separated studies, we have more confidences that the IP eLP2 provided major residues to form ligand recognition site in the extracellular domains of the IP receptor. In addition, the model also provides important clues for the molecular mechanisms of the interaction between the ligand and the receptor. Based on the docking model showing in Figure 5, several interactions including the charge contact (between the residue Arg173 and the C1 carboxylate of Iloprost), the hydrophobic contact (between the residues A172/ M174/Leu177 and the side chain of Iloprost), and the hydrogen bond contacts (between the residues Q172/Q178 and the C11-OH/C15-OH of Iloprost) are predicted. Of course, the predictions are needed to be further tested by mutagenesis and structural studies.
Finally, identification of the ligand recognition site on the extracellular domain of the IP receptor has no conflict with the identified residues important to the ligand binding in TM domains for GPCRs. As described in previous studies, we have proposed two stages of ligand binding to a prostanoid receptor in which ligand is specifically recognized by the key residues in the extracellular domains of the receptor first, and then deposited into the receptor TM domains [43]. For instance, the highly conserved Arg residue in the TM VII is important to the ligand binding, but it is not likely to be important in determination of the ligand selectivities for the different prostanoid receptors. Thus, identification of the new residues determined the specific ligand recognition in the extracellular domain is important, and has not controversial with the identified ligand binding residues in the TM domain.
Conclusion
We have constructed a 3D working model for the human IP receptor by homology modeling using the crystal structure of the bovine rhodopsin TM domains and the NMR peptide structures of the extracellular loops of the TP receptor. The residues in the eLP2 domain involved in forming ligand recognition site were proposed. One of the key residues, Arg173 important to the ligand recognition was predicted from the model and confirmed by mutagenesis. The strategy used for the studies is suitable for modeling and prediction of ligand recognition pockets for other prostanoid receptors.
Materials and methods
Materials
COS-7 cells were purchased from ATCC (Manassas, VA). Medium for culturing COS-7 cells was from Invitrogen. [3H]-iloprost and iloprost were purchased from Amersham Pharmacia Biotech (Piscataway, NJ). DNA polymerase and DpnI endonuclease were obtained from Stratagene (La Jolla, CA).
Antibody production
HPLC-purified synthetic peptides corresponding to the three extracellular loops were mixed and coupled to keyhole-limpet haemocyanin (KLH) using glutaldehyde. The peptide antibody was produced in Female New Zealand White rabbits by Research Genetic Inc. The specific peptide antibody was obtained by affinity chromatography using the appropriate peptide immobilized on CNBr-activated Sepharose 4B.
X-ray crystal structure of bRH
The recently reported X-ray crystal structure of bRH with a resolution at 2.6 Å (1L9H) [53] was downloaded from Protein Data Bank (PDB), and the 3-D structures of the TM domains were extracted using Insight II software on a SIG workstation. These structures were used for homology modeling of the TM domains of the IP and TP receptors.
NMR structures of the synthetic peptides mimicking the extracellular domains of TP receptors
We have solved all of the NMR structures of the constrained peptides mimicking the three eLPs of human TP receptor [43]. These are saved in our database. The NMR structures were converted to PDB format and then used for the homology modeling of the IP eLP domains using the Insight II software on a SIG workstation.
NMR structure of iloprost
The 3-D structure of iloprost, the IP receptor agonist, was solved by high-resolution 2-D NMR spectroscopy [42], and the conformation was used for directly docking with the IP receptor.
Molecular modeling and ligand docking
Molecular modeling, dynamic and ligand docking studies were performed on a Silicon Graphics Octane workstation using the software packages Insight II and Discover [54]. The package includes the software for sequence alignment, secondary structural calculation, hydropath analysis, protein modeling, energy minimization, molecular dynamics, molecular annealing and others.
Site-directed mutagenesis
A pAcSG-IP wild-type cDNA was first subcloned into EcoRI/XbaI sites of pcDNA3.1(+) expression vector. The IP receptor mutants were then constructed using standard PCR. The procedures included the use of a pcDNA3.1(+) vector containing a wild-type IP receptor cDNA as a template, and two synthetic oligonucleotide primers containing the desired mutation for the reaction. The primers, which were complementary to opposite strands of the template, extended during the temperature cycling of 95°C for 30s, 53°C for 1 min 30s, and 68°C for 13 min for a total of 25 cycles with an additional extension cycle of 68°C for 10 min using Pfu DNA polymerase. The mutant products were treated with DpnI endonuclease to digest the parental DNA template and confirmed by DNA sequencing. The plasmids were then prepared using a Midiprep kit (Qiagen) for transfection into COS-7 cells for expression.
Expression of the recombinant IP receptor in COS-7 cells
COS-7 cells were cultured at 37°C in a humidified 5% CO2 atmosphere in high glucose Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, antibiotics and antimycotics. The cells, which were placed on 100-mm dishes at a density of 1.0 × 106 were cultured overnight and then transfected with 10 μg of purified cDNA of pcDNA3.1(+)/IP wild-type or each mutant by the Lipofectamine method [55], as outlined by the manufacture's instructions (Invitrogen). Approximately 48 hours after transfection, the cells were harvested for further protein purification.
Western blot analysis
The transfected COS-7 cells were scraped from the plates into ice-cold PBS buffer, pH 7.4, and collected by centrifugation. After washing three times, the pellet was resuspended in a small volume of the same buffer. The protein was separated by 12% polyacrylamide gel electrophoresis under denaturing conditions and then transferred to a nitrocellulose membrane. A band recognized by primary antibodies against the peptides mimicking the human IP eLPs was visualized with horseradish peroxidase substrate as previously described [10].
Ligand binding assay
Ligand binding assay for the IP receptor was performed using the method as described by [40]. Briefly, the cell membrane (0.1 mg) in binding buffer was incubated with 4 nM [3H]-iloprost (30,000 cpm) in a 0.1 ml reaction volume at room temperature for 40 min. The reaction was terminated by the addition of 5 ml of ice-cold washing buffer (0.025 M Tris-HCl, pH 7.4). The unbound ligand was then filtered through a Whatman GF/B glass filter (Whatman) under a vacuum. The radioactivity of the receptor-bound [3H]-iloprost remaining on the filter was counted in 10 ml of scintillation cocktail using a Beckman β Counter.
Abbreviations
IP, prostacyclin (prostaglandin I2 (PGI2)) receptor; TP, thromboxane A2 receptor; GPCR, G protein-coupled receptor; NMR, nuclear magnetic resonance; eLP, extracellular loop; eLP1, the first eLP; eLP2, the second eLP; and eLP3, the third eLP.
Authors' contributions
CHR carried out the homology modeling and the receptor mutagenesis work. JW participated in the final figure preparations. KHR provided modeling design and drafted the manuscript. All authors read and approved the final version of the manuscript
Acknowledgements
We thank Dr. Guangxion Huang for the initial preparation of the recombinant IP receptor, and Dr. Xialain Go in Biochemistry Department, the University of Houston, for access to the NMR facility. This work was supported by NIH grants (U.S.A.) of HL56712, HL079389 and NS23327 (for KHR).
==== Refs
Coleman RA Kennedy I Humphrey PPA Bunce K Kumley P Hansch C, Sammes PG, Taylor JB, Emmett JC Prostaglandin and their receptor in comprehensive medicinal chemistry 1990 3 Pergamon Press, Oxford 643 714
Ushikubi F Nakajima M Hirata M Okuma M Fujiwara M Narumiya S Purification of the thromboxane A2/prostagiandin H2 receptor from human blood platelets J Biol Chem 1989 264 16496 16501 2528545
Hirata M Hayashi Y Ushikubi F Yokata Y Kageyama R Nakanishi S Narumiya S Cloning and expression of cDNA for a human thromboxane A2 receptor Nature 1991 349 617 620 1825698 10.1038/349617a0
Negishi M Sugimoto Y Ichikawa A Molecular mechanisms of diverse actions of prostanoid receptors Biochim Biophys Acta 1995 1259 109 120 7492609
Funk CD Furci L Moran N Fitzgerald GA Point mutation in the seventh hydrophobic domain of the human thromboxane A2 receptor allows discrimination between agonist and antagonist binding sites Mol Pharmacol 1993 44 934 939 8246916
Yamamoto Y Kamiya K Terao S Modeling of human thromboxane A2 receptor and analysis of the receptor-ligand interaction J Med Chem 1993 36 820 825 8464036 10.1021/jm00059a005
Chiang N Kan WM Tai HH Site-directed mutagenesis of cysteinyl and serine residues of human thromboxane A2 receptor in insect cells Arch Biochem Biophys 1996 334 9 17 8837733 10.1006/abbi.1996.0423
Breyer RM Audoly L The second extracellular loop of the prostaglandin EP3 receptor is an essential determinant of ligand selectivity J Biol Chem 1997 272 13475 13478 9153190 10.1074/jbc.272.21.13475
Turek JW Halmos T Sullivan NL Antonakis K Le Breton GC Mapping of a ligand-binding site for the human thromboxane A2 receptor protein J Biol Chem 2002 277 16791 16797 11877412 10.1074/jbc.M105872200
So SP Wu J Huang G Huang A Li D Ruan KH Identification of residues important for ligand binding of thromboxane A2 receptor in the second extracellular loop using the NMR experiment-guided mutagenesis approach J Biol Chem 2003 278 10922 10927 12551898 10.1074/jbc.M209337200
Perlman JH Colson AO Jain R Czyzewski B Cohen LA Osman R Gershengorn MC Role of the extracellular loops of the thyrotropin-releasing hormone receptor: evidence for an initial interaction with thyrotropin-releasing hormone Biochemistry 1997 36 15670 15676 9398295 10.1021/bi9713310
Hoffmann C Moro S Nicholas RA Harden TK Jacobson KA The role of amino acids in extracellular loops of the human P2Y1 receptor in surface expression and activation processes J Biol Chem 1999 274 14639 14647 10329657 10.1074/jbc.274.21.14639
Olah ME Jacobson KA Stiles GL Role of the second extracellular loop of adenosine receptors in agonist and antagonist binding: Analysis of chimeric A1/A3 adenosine receptors J Biol Chem 1994 269 24692 24698 7929142
Kim J Jiang Q Glashofer M Yehle S Wess J Jacobson KA Glutamate residues in the second extracellular loop of the human A2a adenosine receptor are required for ligand recognition Mol Pharmacol 1996 49 683 691 8609897
Zhao MM Hwa J Perez DM Identification of critical extracellular loop residues involved in alpha 1-adrenergic receptor subtype-selective antagonist binding Mol Pharmacol 1996 50 1118 1126 8913343
Henderson R Baldwin JM Ceska TA Zemlin F Beckmann E Downing KH Model for the structure of bacteriorhodopsin based on high-resolution electron cryomicroscopy J Mol Biol 1990 213 899 929 2359127
Gregorieff N Ceska TA Downing KH Baldwin JM Henderson R Electron crystallographic refinement of the structure of bacteriorhodopsin J Mol Biol 1996 259 393 421 8676377 10.1006/jmbi.1996.0328
Edman K Nollert P Royant A Belrhali H Pebay-Peyroula E Hajdu J Neutze R Landau EM High resolution x-ray structure of an early intermediate in the bacteriorhodopsin photocycle Nature 1999 401 822 826 10548112 10.1038/44623
Luecke H Schobert B Richter HT Cartailler JP Lanyi JK Structure of bacteriorhodopsin at 1.55Angstrom resolution J Mol Biol 1999 291 899 911 10452895 10.1006/jmbi.1999.3027
Findlay J Eliopoulos E Three-dimensional modeling of G protein-linked receptors Trends Pharmacol Sci 1990 12 492 499 1964255 10.1016/0165-6147(90)90050-I
Grotzinger J Engels M Jacoby E Wollmer A Strassburger W A model for the C5a receptor and for its interaction with the ligand Protein Eng 1991 7 767 771 1665906
Hibert MF Trumppkallmeyer S Bruinvels A Hoflack J 3-dimensional models of neurotransmitter G-binding protein-coupled receptors Mol Pharmacol 1991 40 8 15 1649965
Ijzerman AP Galen JV Jacobson KA Molecular modeling of adenosine receptors. I.The ligand binding site on the A1 receptor Drug Des Discov 1992 9 49 67 1457698
Lewell XQ A model of the adrenergic beta-2 receptor and binding sites for agonist and antagonist Drug Des Discov 1992 9 29 48 1360841
Cronet P Sander C Vriend G Modeling of transmembrane 7helix bundles Protein Eng 1993 6 59 64 8381957
Sylte I Edvardsen O Dahl SG Molecular dynamics of the 5-HT1a receptor and ligands Protein Eng 1993 6 691 700 8248092
Zhang DQ Weinstein H Signal transduction by a 5-HT2 receptor – a mechanistic hypothesis from molecular-dynamics simulations of the 3-dimensional model of the receptor complexed to ligands J Med Chem 1993 36 934 938 8464048 10.1021/jm00059a021
Van Rhee AM Fischer PJM Van G Jacobson KA Modeling the P2y purinoceptor using rhodopsin template Drug Des Discov 1995 13 133 137 8872457
Pogozheva ID Lomize AL Mosberg HI The transmembrane 7-alpha-bundle of rhodopsin: distance geometry calculations with hydrogen bonding constraints Biophys J 1997 72 1963 1985 9129801
Baldwin JM The probable arrangement of the helices in G protein-coupled receptors EMBO J 1993 12 1693 1703 8385611
Herzyk P Hubbard RE Combined biophysical and biochemical information confirms arrangement of transmembrane helices visible from the three-dimensional map of frog rhodopsin J Mol Biol 1998 281 741 754 9710543 10.1006/jmbi.1998.1981
Konvicka K Guarnieri F Ballesteros JA Weinstein H A proposed structure for transmembrane segment 7of G protein-coupled receptors incorporating an Asn-Pro/Asp-Pro motif Biophys J 1998 75 601 611 9675163
Attwood TK Findlay JBC Design of a discriminating fingerprint for G-protein-coupled receptors Protein Eng 1993 6 167 176 8386361
Schertler GFX Villa C Henderson R Projection structure of rhodopsin Nature 1993 362 770 772 8469290 10.1038/362770a0
Schertler GFX Hargrave PA Projection structure of frog rhodopsin in two crystal forms Proc Natl Acad Sci USA 1995 92 11578 11582 8524807
Davies A Schertler GFX Gowen BE Saibil HR Projection structure of an invertebrate rhodopsin J Struct Biol 1996 117 36 44 8776886 10.1006/jsbi.1996.0067
Krebs A Villa C Edwards PC Schertler GFX Characterization of an improved two dimensional p22121 crystal from bovine rhodopsin J Mol Biol 1998 282 991 1003 9753549 10.1006/jmbi.1998.2070
Unger VM Schertler GFX Low-resolution structure of bovine rhodopsin determined by electron cryomicroscopy Biophys J 1995 68 1776 1786 7612819
Unger VM Hargrave PA Baldwin JM Schertler GFX Arrangement of rhodopsin transmembrane alpha-helices Nature 1997 389 203 206 9296501 10.1038/38316
Katsuyama M Sugimoto Y Namba T Irie A Negishi M Narumiya S Ichikawa A Cloning and expression of a cDNA for the human prostacyclin receptor FEBS Lett 1994 344 74 88 7514139 10.1016/0014-5793(94)00355-6
Ruan KH Milfeld K Kulmacz RJ Wu KK Comparison of the construction of a 3-D model for human thromboxane synthase using P450 cam and BM-3 as templates: implications for the substrate binding pocket Protein Eng 1994 7 1345 1351 7700866
Ruan KH Wu J So SP Jenkins LA Evidence of the residues involved in ligand recognition in the second extracellular loop of the prostacyclin receptor characterized by high resolution 2D NMR techniques Arch Biochem Biophys 2003 418 25 33 13679079 10.1016/S0003-9861(03)00401-6
Ruan KH Wu J So SP Jenkins LA Ruan CH NMR structure of the thromboxane A2 receptor ligand recognition pocket Eur J Biochem 2004 271 3006 16 15233797 10.1111/j.1432-1033.2004.04232.x
Ruan KH So SP Wu J Li D Huang A Kung J Solution structure of the second extracellular loop of human thromboxane A2 receptor Biochemistry 2001 40 275 280 11141080 10.1021/bi001867c
Wu J So SP Ruan KH Solution structure of the third extracellular loop of human thromboxane A2 receptor Arch Biochem Biophys 2003 414 287 293 12781781 10.1016/S0003-9861(03)00192-9
Franzoni L Nicastro G Pertinhez TA Tato M Nakaie CR Paiva AC Schreier S Spisni A Structure of the C-terminal fragment 300–320 of the rat angiotensin II AT1A receptor and its relevance with respect to G-protein coupling J Biol Chem 1997 272 9734 9741 9092505 10.1074/jbc.272.15.9734
Anand-Srivastava MB Sehl PD Lowe DG Cytoplasmic domain of natriuretic peptide receptor-C inhibits adenylyl cyclase. Involvement of a pertussis toxin-sensitive G protein J Biol Chem 1996 271 19324 19329 8702617 10.1074/jbc.271.32.19324
Dias JA Human follitropin heterodimerization and receptor binding structural motifs: identification and analysis by a combination of synthetic peptide and mutagenesis approaches Mol Cell Endocrinol 1996 125 45 54 9027342 10.1016/S0303-7207(96)03947-0
Takemoto DJ Morrison D Davis LC Takemoto LJ C-terminal peptides of rhodopsin. Determination of the optimum sequence for recognition of retinal transducin Biochem J 1986 235 309 312 3461782
Konig B Arendt A McDowell JH Kahlert M Hargrave PA Hofmann KP Three cytoplasmic loops of rhodopsin interact with transducin Proc Natl Acad Sci USA 1989 86 6878 6882 2780545
Okamoto T Nishimoto I Detection of G protein-activator regions in M4 subtype muscarinic, cholinergic, and alpha 2-adrenergic receptors based upon characteristics in primary structure J Biol Chem 1992 267 8342 8346 1314825
Yeagle PL Alderfer JL Albert AD Three-dimensional structure of the cytoplasmic face of the G protein receptor rhodopsin Biochemistry 1997 36 9649 9289017 10.1021/bi970908a
Okada T Fujiyoshi Y Silow M Navarro J Landau EM Shichida Y Functional role of internal water molecules in rhodopsin revealed by X-ray crystallography Proc Natl Acad Sci USA 2002 30 5982 7 11972040 10.1073/pnas.082666399
Biosym/MSI, Insight User Guide 1995 San Diego, CA
Hatae N Yamaoka K Sugimoto Y Negishi M Ichikawa A Augmentation of receptor-mediated adenylyl cyclase activity by Gi-coupled prostaglandin receptor subtype EP3 in a Gbetagamma subunit-independent manner Biochem Biophys Res Commun 2002 290 162 168 11779148 10.1006/bbrc.2001.6169
|
16271145
|
PMC1298286
|
CC BY
|
2021-01-04 16:26:24
|
no
|
BMC Biochem. 2005 Nov 4; 6:23
|
utf-8
|
BMC Biochem
| 2,005 |
10.1186/1471-2091-6-23
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2641626907710.1186/1471-2105-6-264SoftwaremaxdLoad2 and maxdBrowse: standards-compliant tools for microarray experimental annotation, data management and dissemination Hancock David [email protected] Michael [email protected] Giles [email protected] Norman [email protected] Andrew [email protected] Helen [email protected] A Joseph [email protected] Karim [email protected] Douglas B [email protected] Andy [email protected] School of Computer Science, The University of Manchester, Kilburn Building, Oxford Road, Manchester, UK2 School of Chemistry, The University of Manchester, Faraday Building, PO Box 88, Sackville Street, Manchester, UK3 Faculty of Life Sciences, The University of Manchester, Oxford Road, Manchester, UK4 NERC Environmental Bioinformatics Centre, Oxford Centre for Ecology and Hydrology, Oxford, UK2005 3 11 2005 6 264 264 29 7 2005 3 11 2005 Copyright © 2005 Hancock et al; licensee BioMed Central Ltd.2005Hancock 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
maxdLoad2 is a relational database schema and Java® application for microarray experimental annotation and storage. It is compliant with all standards for microarray meta-data capture; including the specification of what data should be recorded, extensive use of standard ontologies and support for data exchange formats. The output from maxdLoad2 is of a form acceptable for submission to the ArrayExpress microarray repository at the European Bioinformatics Institute. maxdBrowse is a PHP web-application that makes contents of maxdLoad2 databases accessible via web-browser, the command-line and web-service environments. It thus acts as both a dissemination and data-mining tool.
Results
maxdLoad2 presents an easy-to-use interface to an underlying relational database and provides a full complement of facilities for browsing, searching and editing. There is a tree-based visualization of data connectivity and the ability to explore the links between any pair of data elements, irrespective of how many intermediate links lie between them. Its principle novel features are:
• the flexibility of the meta-data that can be captured,
• the tools provided for importing data from spreadsheets and other tabular representations,
• the tools provided for the automatic creation of structured documents,
• the ability to browse and access the data via web and web-services interfaces.
Within maxdLoad2 it is very straightforward to customise the meta-data that is being captured or change the definitions of the meta-data. These meta-data definitions are stored within the database itself allowing client software to connect properly to a modified database without having to be specially configured. The meta-data definitions (configuration file) can also be centralized allowing changes made in response to revisions of standards or terminologies to be propagated to clients without user intervention.
maxdBrowse is hosted on a web-server and presents multiple interfaces to the contents of maxd databases. maxdBrowse emulates many of the browse and search features available in the maxdLoad2 application via a web-browser. This allows users who are not familiar with maxdLoad2 to browse and export microarray data from the database for their own analysis. The same browse and search features are also available via command-line and SOAP server interfaces. This both enables scripting of data export for use embedded in data repositories and analysis environments, and allows access to the maxd databases via web-service architectures.
Conclusion
maxdLoad2 and maxdBrowse are portable and compatible with all common operating systems and major database servers. They provide a powerful, flexible package for annotation of microarray experiments and a convenient dissemination environment. They are available for download and open sourced under the Artistic License.
==== Body
Background
Rich and accurate experimental annotation is important to the analysis and understanding of experimental results. This is especially true amongst the post-genomic techniques such as transcriptomics where, due to the scale of data produced, relatively minor changes in laboratory procedure can have profound effects on results and their interpretation. For this reason it is crucial that all relevant protocols and parameters, i.e. the experimental meta-data, are recorded.
The microarray community has agreed standards regarding both the data that should be captured (MIAME, [1]) and how these data should be modelled and exchanged (MAGE-OM [2], MAGE-ML, [3]). However, these standards are still evolving. Similarly, post-genomic technologies are continuing to evolve and are being applied to news areas, for example in environmental genomics and toxicogenomics. These new applications have very different data capture needs to those encountered in more standard model organisms. Indeed, simply using arrays with different classes of organisms can add to the complexity of the annotation required.
Ensuring compliance with standards means that the meta-data capture step should not be seen as an insurmountable barrier to progress. The standards for meta-data capture and exchange are still not readily usable or even understood by much of the biological community (for example the level of detail required for MIAME compliance [1] and the complexity of models used to store microarray data). A consequence of this complexity is that the model [2] can be seen to be confusing by bench biologists, a principal target audience of experimental annotation software. In addition, the associated MGED ontology [4] that provides controlled terms for annotation is continually evolving to cover new biological fields, new applications of the technology and to correct oversights. This must be reflected in the software so that the user can annotate their experiment with the fullest set of terms available.
The volume of data needing to be captured is growing rapidly. This increase in volume occurs in the numbers of experiments and replicates performed, as well as in the numbers of spots on each array. The increasing level of interest is due to more people becoming involved in transcriptomics research, whether they are collaborators or clients, or independent researchers browsing public data. Both of these factors necessitate software that can handle large amounts of data and present it to the end user in the most efficient way possible.
To meet these challenges, we believe that tools for annotation of transcriptomics experiments must offer four key features:
• flexibility, to adapt to changing standards and technologies;
• scalability, to cope with the increasing amounts of production and use of data;
• usability, to reflect the way users consider experiments and handle their data (including provision of an simplified view of the underlying data model at a level of abstraction suitable for biologists);
• distributability, to allow users to disseminate their data to colleagues and to public repositories in standards-compliant format for publication.
There are already a number of free software tools available to the community. MADAM: MicroArray Data Manager [5] provides a Java data entry tool over a relational database. BASE [6], Longhorn Array Database [7] and MARS [8] provide data capture and browsing tools through a Web server. MIAMExpress [9] provides web tools for entering data directly into ArrayExpress.
However, although each of these software tools are MIAME supportive and generally capable of providing MAGE-ML output, they did not meet the specific requirements we had identified for the creation of customised attributes to meet the needs of specific communities, support for bulk loading data from spreadsheets, or the ability to produce MAGE-ML that was sufficiently expressive to cope with situations in which users wished to customise their meta-data and meet the standards of repositories such as ArrayExpress. We summarise these tools and their capabilities in table 1.
Table 1 A comparison of the different products available to handle transcriptome experimental data
Software Platforms MIAME Compliant Imports Exports Fully customisable attributes Comments
maxdLoad2 Any that runs Java Yes maxd-ML, MAGE-ML, bulk loading from CSV files using XML maxd-ML, text and fully expressive MAGE-ML Yes
maxdBrowse Server requires Apache and PHP5. Clients any web browser or program/script that can be configured to retrieve data via SOAP (if remote) or from the command-line (if run on the server) Same as maxdLoad2 No import Exports plain-text, HTML and XML Yes – runs off maxdLoad2 attributes
MIAMExpress Any (web-based) Yes Web-based forms N/A No No bulk load facility, although it is being extended to allow uploading from spreadsheets in the future
BASE Server is installed on GNU/Linux, with a web-based front-end Optional Web-based forms, bulk loading for raw data only MAGE-ML Potentially, via user definable fields Plug-ins available for data normalisation, analysis and viewing
MADAM MS Windows and GNU/Linux Yes Web-based forms, no bulk loading Its own format (.mev) Potentially, by editing the database schema Is being adapted to read and write MAGE-ML in the future
LAD Server is installed on GNU/Linux, with a web based front end Yes Batch imports using tab delimited files Tab delimited file No Very limited experimental annotation. Open source version of the Stanford Microarray Database
Here we present two components of the maxd software suite, maxdLoad2 and maxdBrowse. The former is a robust, fully featured database management, annotation and export tool, available as a downloadable application. The latter is a dissemination and data-mining tool, available as a server-side application presenting browser-based, command-line and web-service interfaces to the data.
Implementation
In contrast to many other transcriptomics database solutions, maxdLoad2 is a standalone application rather than a web-based tool. This allows a higher degree of interactivity whilst mitigating the considerable cost of data upload from browser to server. It is written in the Java programming language making it highly platform independent. The database back-end can be provided by any relational database server which implements the SQL92 standard, including Oracle, PostgreSQL and MySQL. The server can be accessed remotely, or for maximum performance can be run on the same machine as maxdLoad2. Whilst standalone application installation is rarely as simple as accessing a web based service, considerable effort has been put into making maxdLoad2 easy to get started. In the majority of cases installation is a matter of downloading and running a single file. The maxdLoad2 website contains full instructions on installation and configuration of Java as well as an appropriate SQL server product. In addition several commonly used microarray chip descriptions are available for download.
The ability to disseminate data stored in maxdLoad2 databases via a browser is important. Our solution was to develop one independently in a language built for this environment, PHP, which is also platform independent. Being a web-application, it is less easy to install, but once installed very easy to use on a browser.
Bioinformatics data-analysis demands keep on changing, but access to data and meta-data in a scriptable form is a must. maxdBrowse thus provides quick access to the contents of maxdLoad2 databases via the command-line and as a web-service.
User interface, navigation and browsing
The complexity of the schema, simple as it is compared to the full MAGE-OM, still represents a challenge for visualisation and navigation. A graphical representation of the schema appears at the top-level of the user interface and is used as the principal navigation tool (figure 1(a)). Users click on elements of the schema to jump directly to data entry or browsing forms for that element. The representation also attempts to make clear the relationship between schema elements. An optional alternative visualisation is a version of the schema flattened into a tree rooted at the Experiment table. Although table-interconnectivity is harder to portray in the tree paradigm, it has the advantage of being more readily able to display information about specific entries. An interactive highlighting scheme which displays the linkage of individual items as the mouse passes over them further enhances the usefulness of this representation.
Figure 1 The maxdLoad2 interface. Clockwise from the top-left, this figure shows (1) the main schema interface window (2) an example of a create mode manual entry form, (3) Browse Mode example, for viewing instances in the database, (4) the Load Mode interface for populating tables from spreadsheets.
maxdLoad2 has five principle interaction modes, 'Create', 'Browse', 'Load', 'Edit' and 'Find'. Each uses the same top-level schema-based navigation method to select particular database elements for manipulation.
Create mode displays blank forms, built from the XML specification for the table, to allow for annotation of a single entry (figure 1(b)). The data-types specified translate into standard interface elements such as free text entry boxes and drop-down menus. The Browse, Edit and Find modes allow interaction with existing entries (figure 1(c)). Navigation between entries is facilitated by the 'Find Linked' function which allows interconnected entries to be identified regardless of the underlying schema and the number of links between them. It finds the shortest path between the tables and issues the relevant series of queries to determine which entries are related. An example of this is identifying all 'Experiment' instances that use a particular 'TreatmentProtocol'.
As large amounts of microarray data and meta-data are in the form of tab-delimited or spreadsheet files, for example array descriptions, genome information and raw or processed data files, maxdLoad2 has specific features for handling such data. Load mode uses the familiar graphical interface of Create mode (figure 1(d)) but instead of providing values one-by-one, a flexible data description language is used to specify how to extract data from a file, as can be seen in figure 2. Various data manipulation tools are available including on-the-fly translation and regular expression pattern matching. Once the extraction rules have been specified, the data that will be entered into the database can be previewed by a built-in spreadsheet interface before loading.
Figure 2 Column specification Syntax for data loading. The column specification syntax controls the destination attribute field that a column of data should be entered into in the database. The data can be pre-processed before they are loaded into the fields including data enumerations to convert values, combinations to allow multiple fields to be recomposed into an attribute and regular expressions to perform complex operations such as separating values from units, or removing punctuation, amongst other procedures.
maxdBrowse presents multiple interfaces to maxd databases built by maxdLoad2. Its web interface emulates the browsing capabilities of maxdLoad2 (figure 3), with a simpler table navigator, entry selection and various formatting options. maxdBrowse also provides a command-line SOAP server interface.
Figure 3 The maxdBrowse interface. This shows maxdBrowse linking an entry on the 'Measurement' table to entries in the 'LabelledExtract' and 'Source' tables. At the very top are site navigation tabs and a banner indicating that the database page is being accessed. Underneath this are tabs to switch between navigation modes, with link highlighted as the current mode. Underneath this is a selection form, where entries can be selected for linking to other tables, additional formatting options are available by clicking the (+) link on top of the 'go' button. Below this is the data presentation form showing the results of the query, in this case the 'Measurement' DrugX-(4 h) (Cy3) vs gDNA (Cy5) has been linked down to entries in LabelledExtract and Source. On the right is the TableNavigatorTree, highlighting the tables that are being linked. Help links (?) pop-up help about maxdBrowse features, and information links (i) pop-up help about maxdLoad2 schema elements displayed on the page.
Several kinds of query are available, and these are reflected by links at the top to (currently) 'browse', 'link', 'search', 'enumerate' and 'basket'. Each of these modes gives a customised form for that kind of query. The primary query, used to navigate and inspect database, is 'browse'. Selected entries will be retrieved and formatted, displaying links to other parent and child entries in the schema. These links can be used to navigate through the schema to discover aspects of the experiment currently being browsed and various formatting options are available. It also allows the creation of simple structured documents, via use of the recursive 'View descendant' option, which trawls down the schema retrieving a selected entry's children, and those children's children etc.
The link query allows a selected entry in one table to be linked to entries in another, provided there is a direct path between the two tables in the schema, or a path can be made across ArrayType (which is linked to by Array up to Experiment on one side, and by Feature down to Gene on the other). The search query allows searching for entries in either one or all tables. There is a basic search, allowing a quick search of entry names or all the attributes for each entry, and an advanced search, where individual database fields can be searched. Advanced search provides the ability to search ontologies used in the database, including the standard MGED ontology classes used in the maxdLoad2 schema. The web interface provides ontology term selection forms for these. This allows relations between entries to be explored using browse and link-modes, and entry attributes to be searched using free-text and fixed-term strings in search-mode. The enumerate query allows retrieval of numerical data; when run it initially provides column meta-data for selected Measurements, allowing the user to identify the columns of interest for retrieval. The user is then given the option of linking the results to the relevant Feature, Reporter and Gene attributes, and also to select a subset of genes for which to return results.
Each query can be saved to the basket, and retrieved at a later time. This allows the user the opportunity to build a report of useful meta-data, in much the same way online-shoppers select items that they wish to buy. The basket has an option to display command-line and XML format query syntax (see figure 4). Thus if the user finds that reports are getting large, or if the same kinds of reports are being repeatedly generated, then by providing these examples maxdBrowse shows the user how to perform them in command-line syntax. Programmers who wish to use maxdBrowse as an API can try out example queries 'by hand' on the web-page first, and then retrieve them in the basket for use in their SOAP client/command-line code.
Figure 4 The maxdBrowse query syntax. This shows maxdBrowse in basket-mode, retrieving one of several queries previously performed and saved to basket. The query shown is an advanced search of the Reporter table, where the Type. Physical ontology field is being searched for 'unknown_sequence' and the 'Name' free-text field is being searched for 'SCO001'. Also shown are examples of the command-line and XML query syntax that can be used to generate the same result
Data storage
The schema is loosely based on the ArrayExpress database model [10] and it maps onto the MAGE-OM [12,3], with modifications for reasons of simplicity, efficiency and future configurability. The maxd schema is made up of relational and xml-specified components. The relational component covers the general more static table-structure (figure 5), representing both the links between and names of entries in various tables. The dynamic meta-data is stored in attribute columns in a more flexible structure of xml-defined name/value pairs. The attribute specifications are stored in a file which defines the collection of meta-data to be stored for each table in the database. The standard attribute definition is built from the MIAME specifications and MGED ontology, but this can be easily customised from directly within maxdLoad2. This allows attributes to be modified to any degree without requiring a change to the underlying database structure, allowing the software to adapt to changes in details of the model; this is seen in figure 6.
Figure 5 Entity relationship diagram. This shows the relationships between the tables in the maxd database schema.
Figure 6 Attribute Definition Mechanism. Clockwise from the top-left, this figure shows (1) a portion of the standard attribute definition file, (2) the section of the user interface that is generated from this definition, (3) customising the attribute definition by adding a new item "Growth Medium" and (4) how the user interface changes in response to the change.
Data import and export
Other than the form-based annotation and loading maxdLoad2 has bulk export and import facilities to assist in administration and data sharing. These include a native XML based format for transferring data between databases and disseminating complex items such as array descriptions to users. In addition there is a tab-delimited export mode for database entries, a tabular export option for measurement data and the ability to generate structured documents, including, but not limited to, MAGE-ML, via a template based export schema.
Programmatic access to the database is possible via a library API which exposes much of maxdLoad2's functionality to other applications capable of interacting with Java methods.
maxdBrowse opens up maxd databases to web dissemination, and provides additional export facilities, including tab-delimited, flat text, HTML and XML output, while a recursive search over links between entries allows for experimental report generation. The use of maxdBrowse on the command-line and in scripts allows the contents of maxd databases to be made available to a variety of tools and interfaces, for example R/Bioconductor [11] or MATLAB. maxdBrowse is also accessible via SOAP, allowing its extension as a web service to clients across the internet. Example SOAP client interfaces, initially in Perl and PHP, are included with the distribution to help those who wish to build their own interfaces, and web-services description by WSDL is also in progress.
Usability
The software has undergone several rounds of cyclical testing and modification in usability studies with lab biologists using standard cooperative evaluation methodologies [12]. The results of one of these studies are freely available [13]. This has led to several major enhancements to the software's ease of use and applicability to the process.
Results & discussion
The requirements for flexibility and usability could be seen as mutually exclusive. It is typically the case that software which is very configurable requires effort to set up and use. Conversely, software which is very easy to use typically achieves this by restricting flexibility. The maxd suite of software is therefore aimed at three specific user groups, summarised in figure 7.
Figure 7 A Use-case Diagram for maxdLoad2 and maxdBrowse. The 'system administrator' is the individual in charge of maintaining the lab's database and web server. The 'lab scientist (data provider)' is typically someone who does a lot of microarray experiments and manages his experiments using maxdLoad2. The 'lab scientist (data consumer)' is typically someone who may not perform as many microarray experiments, and may not even work in a microarray facility, but is interested in the data and could use maxdBrowse or maxdLoad2 for its access. The statistician would be interested in retrieving data and meta-data for analysis, and may prefer to access the data programmatically either with scripts using maxdBrowse on the command-line or via its SOAP interface. The 'bioinformatician (stand-in)' may be involved at any level with any of these activities.
Deciding what meta-data should be captured for a particular series of experiments and then capturing it effectively in MAGE-ML requires a user group with considerable experience – typically a bioinformatician working with the bench biologists. maxdLoad2 provides such users with a variety of tools for configuring the interface and the MAGE-ML output relatively straightforwardly. If a lot of the meta-data has already been captured in the form of spreadsheets they can also configure the software to take data directly from them.
Once the database and interfaces developed by the bioinformaticians have been set up, they are relatively straightforward to use by the bench biologists (as evidenced from the usability studies). A useful feature is the way in which the data can be provided to the software in the form of spreadsheets. In practice we have seen that this task is also often undertaken, not by the bench biologists, but by a bioinformatician responsible for capturing all the meta-data from experiments run within a particular facility. maxdLoad2 provides good tools for allowing efficient bulk loading of meta-data to help with this. This is also typically the point at which data are exported from maxdLoad2 into ArrayExpress.
Once the data have been captured within the database there is then a requirement to allow co-workers or collaborators to readily explore which experiments have been run and download some of the data if required. These users will probably not want to learn maxdLoad2 and their needs are met by maxdBrowse.
A final set of users that we should be considering are software programs wishing to access array data for use in larger, integrated genomics studies. These needs could potentially be met through the command-line interfaces to maxdBrowse and will be considerably expanded in the future.
Complex models
Annotation of transcriptomics experiments is a hard problem, partly owing to the large amounts of both data and annotation and also to the complexity of the underlying model, which does not necessarily reflect the thinking of a lab biologist. For this reason the maxd suite has been designed with usability by laboratory experimentalists in mind. The first concession to this is that the maxdLoad2 database model is not based on the MAGE-OM itself, but an abstraction of the model, hiding much of the complexity, and addressing itself to how users represent their experiments, whilst still maintaining the overall structure of the model.
Separating the functionality of the standalone-application from the web-application is also crucial, by targeting different groups of users who vary in their ability to invest time and effort into learning new tools. maxdLoad2 and maxdBrowse form a synergistic suite, with maxdLoad2 being a very powerful and flexible tool for annotation and management of experiments, whilst maxdBrowse is a powerful tool for dissemination of experiment data sets, in an easy-access and easy-to-learn web browser based environment which avoids the need for collaborators or facility clients from having to learn to use the full standalone tool.
The large amount of data available from microarray experiments as well as the increasing numbers of people generating and using the data provides challenges for transcriptomics annotation software. The maxd suite provides several features for day-to-day users to ease use of the software. maxd databases can be populated with all standard laboratory protocols and arrays, freeing the end-user of this task and ensuring consistency and completeness of annotation within laboratories. All annotation and data can be bulk-loaded from Excel spreadsheets and other tab-delimited formats into maxdLoad2 databases with ease.
The maxdBrowse web-interface allows collaborators or clients to review annotated experiments in a simple manner at any time, without needing maxdLoad2. Its recursive browse ('View descendant') and basket functions help users create reports in a flexible manner. When run as a command-line tool maxdBrowse is capable of extracting large amounts of data from a maxd database. For instance, it is quite easy to write web pages that send requests from users to maxdBrowse and return their results by e-mail. Alternatively maxdBrowse's recursive browse function enables structured representations of an experiment to be created as an overview with associated annotation and data, though in a more restricted form than maxdLoad2's structured document generation. The advantage of doing this however, is the potential for scripting a range of different export queries into analysis environments, including R/Bioconductor and MATLAB. maxdBrowse also offers the potential for exposing the underlying functionality as a web-service. An example of a useful set of maxdBrowse scripted queries would be to use the link mode (as in figure 3, but not via the GUI) to identify each Measurement's HybridisationProtocol across all experiments in a database, and reclassify them based on this. The enumerate function could then retrieve the data from the 'Flags' column for each of these Measurements, and quickly work out the percentage failure rate, to determine if correlations can be made based on the protocol used. This could also be rapidly encapsulated in a PERL script to automate the process.
The maxd suite has been designed from the ground up to be highly flexible, allowing administrators to tailor the software for the particular needs of users. For example, the set of attributes associated with the database can be altered to suit the biological domain being annotated a feature that has been exploited in the development of MIAME/Env [14]; an extended MIAME specification where new attributes have been added to support annotation of microarray experiments carried out by the NERC Environmental Genomics community [15]. This is achieved by editing the XML-based attribute definition files using an editor built into the application. This mechanism also allows the software to adapt to changes in annotation standards as the definition of the standard terms is held on a globally visible central server which is updated to track changes in both the MIAME specification and the MGED Ontology. This centralized administration stretches to other areas of the software, such as Unicode definition files to change the range of symbols that can be included in annotation description, and to loading preset files, which control maxdLoad2's parser settings for loading data from spreadsheet files. These repositories are being filled with settings for both array layout and measurement files for the broad majority of independent manufacturers over time and can be added to to cater for custom laboratory formats.
The maxd software suite also offers flexibility in the sense that it enables data-producing centres to quickly establish their own microarray databases and publish them with web-based front-ends. This allows users and clients to interact with central repositories in a far more integrated and organized manner than sending raw scanner output files with an experimental description in an e-mail. Examples of web-based repositories that embed maxdBrowse for access to transcriptomics data include StreptoBASE 16 at the University of Manchester and EnvBrowse 17 at the Natural Environment Research Council (NERC) Environmental Bioinformatics Centre.
Capturing transcriptomics data is only the first part of the process; to be useful they need to be distributed to collaborators or to public repositories. Distribution of data is complicated by the varied backgrounds of individuals involved, the size of the data sets and the distributed nature of the resources. Web-based applications tend to be easy to use, but can only present a limited subset of data to the user, due to the speed of connection to the server and issues surrounding the capabilities of web clients. On the other hand, standalone applications tend to be more complicated to install and use, but are far more robust and powerful in the specific tasks they can perform. The maxd suite has been designed to maximize the advantages that these platform decisions offer, for example data entry can only be done in the standalone application, as this is a far better environment with instantaneous interface and high throughput data handling that experimental data loading requires. Data dissemination on the other hand can be performed in either maxdLoad2 or maxdBrowse, depending on the audience. Users who are comfortable with maxdLoad2 and trusted with the data can use it to browse the contents of databases, whereas other users can be given access to the database using maxdBrowse. Both components can perform data export; maxdLoad2 can export whole experiments, and any part thereof, whereas maxdBrowse focuses only on exporting selected entries and their descendants because of web-browser and service limitations (e.g., if an Experiment and 'View descendant' are selected, everything but array design would be exported).
By ensuring that the solutions are flexible, scalable and usable, organizing efficient data distribution networks becomes far more tractable. We believe maxdLoad2 is the application of choice for communication within and between data-centres, whether it be annotating experiments or exporting structured documents such as MAGE-ML. To this end, the European Bioinformatics Institute accepts maxdLoad2-generated MAGE-ML submissions into ArrayExpress. maxdBrowse, however, can act as a general dissemination medium for non-experts, by providing a simple web-interface for selecting, retrieving and formatting the contents of maxd databases. On the command-line, its ability to query maxd databases with a simple set of parameters allows mining of these data by a variety of possible interfaces and analysis solutions, which may or may not be located at the same site as the database itself.
Planned future development
Currently, the maxdLoad2 security model is based on the rather coarse-grained security provided by the default underlying database server. This implies that a user with sufficient database privileges could perform actions on any data that they can see. Due to the lack of "row-level" security, it is not possible to control access to a subset of a particular database. Our interim solution is to advise users to disseminate data using the maxdBrowse web application, which adds a layer of control as to who can see data and can prevent changes being made. In the medium to long term adding full support for row-level security is a planned development, but care must be taken that this does not adversely impact usability or performance too significantly.
Additional forthcoming features include support for hierarchical collections of objects such as clusters of genes or results, additional import and export options, more sophisticated database visualization tools and, eventually, a fully configurable schema. We will also be further developing the web-services interface to the maxd databases through maxdBrowse to allow incorporation of microarray data into the bioinformatics workflows supported through the Taverna component of myGrid 18. This should make it much more straightforward for users to develop complex queries combining microarray data with other resources (such as Ensmart 19).
Conclusion
maxdLoad2 is a free, highly-tested standards-compliant, infinitely configurable, post-genomic experimental annotation tool, presently directed at transcriptomics array based experiments. Developed to the needs of lab based experimentalists, it can act as a pipeline for annotation into public repositories such as ArrayExpress. maxd databases can be disseminated over the internet using maxdBrowse, either in a human-friendly manner to users through a web browser, or via web-services or command-line scripts into popular analysis packages. This allows both local and remote users to browse and download data, providing a security model for dissemination of private datasets to collaborators. maxdBrowse can be implemented either standalone or can be implemented as part of maxd-based microarray data repositories and can provide a PHP library for more complex web-based maxd-related queries in the future such as search, filtering and data processing.
Availability and requirements
Project name: maxdLoad2
Project home page:
Operating system(s): Platform independent
Programming language: Java
Other requirements: e.g. Java (1.4 or higher recommended), access to SQL database
License: Perl Artistic
Any restrictions to use by non-academics: None
Project name: maxdBrowse
Project home page:
Operating system(s): Platform independent
Programming language: PHP 5
Other requirements: e.g. Web server (tested with Apache 2, IIS5), access to SQL database
License: Artistic
Any restrictions to use by non-academics: None
List of abbreviations
API – Application Programmer Interface
MAGE-ML – Microarray and Gene Expression Markup Language
MAGE-OM – Microarray and Gene Expression Object Model
MGED – Microarray Gene Expression Data (Society)
MO – MGED Ontology
NERC – Natural Environment Research Council
RDBMS – Relational Database Management System
SQL – Structured Query Language
XML – Extensible Markup Language
Authors' contributions
DH wrote the maxdLoad2 software and documentation; MW wrote the manuscript; GV wrote the maxdBrowse software and documentation; NM was responsible for development planning, schema design and testing maxdLoad2; AH was involved in design and testing of maxdLoad and providing ongoing use cases; HH was involved in feature design and testing of maxdLoad2 and training users; AJW was involved in feature design and testing of maxdLoad2/maxdBrowse, training of users, promoting the software and providing use cases; KN designed the usability studies and analysed the results; DK supervised, promoted and coordinated the development of maxdBrowse; AB supervised, promoted and coordinated the development of maxdLoad2. All authors read and approved the final manuscript.
Acknowledgements
maxdLoad2 development was funded by the NERC Environmental Bioinformatics Centre (NEBC), and maxdBrowse by the Biotechnology and Biological Sciences Research Council Investigating Gene Function Initiative.
The authors would like to acknowledge: Helen Parkinson and Tim Rayner, from the European Bioinformatics Institute (EBI) for debugging the MAGE-ML output, and advising on best practice for import into the ArrayExpress repository; Lucy Bridges (NERC) for testing the maxdLoad2 and maxdBrowse software, contributing tools and documentation and providing the information in table 1; Georgina Moulton from the Northwest Institute for Bio-Health Informatics for providing training, publicity and feedback on maxdLoad and maxdBrowse; Afsaneh Maleki, Irena Spasic, Andy Tseng and Ben Routley from the University of Manchester (BBSRC, DTI and MRC) for many useful technical discussions. The authors would also like to thank the following: Jill Wishart (BBSRC), Nianshu Zhang (BBSRC), Juan Castrillo (NERC), Studel Gato (NERC) and Justin Warne (NERC) from the University of Manchester; E. Jane Fraser and Andrew Cossins from the Univeristy of Liverpool (NERC); Colin Smith, Graham Hotchkiss, Vassilis Mersinias and Emma Laing from the University of Surrey (BBSRC & EC); Keith Chater, Sofoklis Lekkas, Eriko Takano, Sandor Biro, Aleksandra Smulczyk and Govind Chandra from the John Innes Centre, Norwich (BBSRC & EC); Alan Ward (BBSRC), Nickolas Allenby (BBSRC), Dan Swan (BBSRC), Peter Olive (NERC) and Kim Last (NERC) from the University of Newcastle; Matthew Hegarty from the University of Bristol (NERC); Mike Allen from Plymouth Marine Laboratory (NERC); Cas Kramer and Thierry Bailhache from the University of Leicester (NERC); Jonathan Reeves from the Centre for Ecology and Hydrology, Oxford (NERC); Darius Armstrong-James from Imperial College, London (BBSRC); and Bregje Wertheim, from University College London (NERC) for helping with metadata and data-collection vital for developing and testing maxdLoad2 and maxdBrowse.
==== Refs
Brazma A Hingamp P Quackenbush J Sherlock G Spellman P Stoeckert C Aach J Ansorge W Ball CA Causton HC Gaasterland T Glenisson P Holstege FC Kim IF Markowitz V Matese JC Parkinson H Robinson A Sarkans U Schulze-Kremer S Stewart J Taylor R Vilo J Vingron M Minimum information about a microarray experiment (MIAME)-toward standards for microarray data Nat Genet 2001 29 365 371 11726920 10.1038/ng1201-365
Group OM Gene Expression Specification v1.1 The Object Management Group, Inc 3 AD
Spellman PT Miller M Stewart J Troup C Sarkans U Chervitz S Bernhart D Sherlock G Ball C Lepage M Swiatek M Marks WL Goncalves J Markel S Iordan D Shojatalab M Pizarro A White J Hubley R Deutsch E Senger M Aronow BJ Robinson A Bassett D Stoeckert CJJ Brazma A Design and implementation of microarray gene expression markup language (MAGE-ML) Genome Biology 2002 3 RESEARCH0046 12225585 10.1186/gb-2002-3-9-research0046
Stoeckert CJ Parkinson H The MGED ontology: a framework for describing functional genomics experiments Comparative and Functional Genomics 2003 4 127 132 10.1002/cfg.234
MADAM 2005
Saal LH Troein C Vallon-Christersson J Gruvberger S Borg A Peterson C BioArray Software Environment (BASE): a platform for comprehensive management and analysis of microarray data Genome Biol 2002 3 SOFTWARE0003 12186655 10.1186/gb-2002-3-8-software0003
Killion PJ Sherlock G Iyer VR The Longhorn Array Database (LAD): an open-source, MIAME compliant implementation of the Stanford Microarray Database (SMD) BMC Bioinformatics 2003 4 32 12930545 10.1186/1471-2105-4-32
Maurer M Molidor R Sturn A Hartler J Hackl H Stocker G Prokesch A Scheideler M Trajanoski Z MARS: microarray analysis, retrieval, and storage system BMC Bioinformatics 2005 6 101 15836795 10.1186/1471-2105-6-101
MIAMExpress 2005
Rocca-Serra P Brazma A Parkinson H Sarkans U Shojatalab M Contrino S Vilo J Abeygunawardena N Mukherjee G Holloway E Kapushesky M Kemmeren P Lara GG Oezcimen A Sansone SA ArrayExpress: a public database of gene expression data at EBI Comptes Rendus Biologies 2003 326 1075 1078 14744115 10.1016/j.crvi.2003.09.026
Gentleman RC Carey VJ Bates DM Bolstad B Dettling M Dudoit S Ellis B Gautier L Ge Y Gentry J Hornik K Hothorn T Huber W Iacus S Irizarry R Leisch F Li C Maechler M Rossini AJ Sawitzki G Smith C Smyth G Tierney L Yang JY Zhang J Bioconductor: open software development for computational biology and bioinformatics Genome Biol 2004 5 R80 15461798 10.1186/gb-2004-5-10-r80
Monk AF Wright PC Davenport L Haber J Improving your human-computer interface: A practical technique 1993 New York, Prentice Hall
maxd Usability Results 2005
Morrison N Wood AJ Hancock D Shah S Hakes L Tiwari B Kille P Cossins A Hegarty M Allen MJ Wilson WH Olive P Last K Kramer C Bailhache T Reeves J Pallett D Warne J Nashar K Parkinson H Sansone SA Rocca-Serra P Stevens R Snape J Field D Brass A Development of the ENV specification for environmental biology and its application to transcriptomics as MIAME/Env BMC Bioinformatics 2005
MIAME/Env Specification 2005
SteptoBASE 2005
EnvBrowse 2005
Stevens RD Robinson AJ Goble CA myGrid: personalised bioinformatics on the information grid Bioinformatics 2003 19 Suppl 1 i302 i304 12855473 10.1093/bioinformatics/btg1041
Kasprzyk A Keefe D Smedley D London D Spooner W Melsopp C Hammond M Rocca-Serra P Cox T Birney E EnsMart: a generic system for fast and flexible access to biological data Genome Res 2004 14 160 169 14707178 10.1101/gr.1645104
|
16269077
|
PMC1298287
|
CC BY
|
2021-01-04 16:27:47
|
no
|
BMC Bioinformatics. 2005 Nov 3; 6:264
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-264
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2661627447610.1186/1471-2105-6-266Research ArticleAlternative splicing and protein function Neverov AD [email protected] II [email protected] RN [email protected] D [email protected] MS [email protected] AA [email protected] State Scientific Center GosNIIGenetika, 1st Dorozhny proezd 1, Moscow, 117545, Russia2 Institute for Bioinformatics/MIPS, GSF – National Research Center for Environment and Health, Ingolstädter Landstraße 1, 85764 Neuherberg, Germany3 Department of Bioengineering and Bioinformatics, M.V.Lomonosov Moscow State University, Vorobievy Gory 1–73, Moscow, 119992, Russia4 Department of Genome Oriented Bioinformatics, Technical University of Munich, Wissenschaftszentrum Weihenstephan, 85350 Freising, Germany5 Institute for Information Transmission Problems RAS, Bolshoi Karetny pereulok 19, Moscow, 127994, Russia2005 7 11 2005 6 266 266 15 6 2005 7 11 2005 Copyright © 2005 Neverov et al; licensee BioMed Central Ltd.2005Neverov 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
Alternative splicing is a major mechanism of generating protein diversity in higher eukaryotes. Although at least half, and probably more, of mammalian genes are alternatively spliced, it was not clear, whether the frequency of alternative splicing is the same in different functional categories. The problem is obscured by uneven coverage of genes by ESTs and a large number of artifacts in the EST data.
Results
We have developed a method that generates possible mRNA isoforms for human genes contained in the EDAS database, taking into account the effects of nonsense-mediated decay and translation initiation rules, and a procedure for offsetting the effects of uneven EST coverage. Then we computed the number of mRNA isoforms for genes from different functional categories. Genes encoding ribosomal proteins and genes in the category "Small GTPase-mediated signal transduction" tend to have fewer isoforms than the average, whereas the genes in the category "DNA replication and chromosome cycle" have more isoforms than the average. Genes encoding proteins involved in protein-protein interactions tend to be alternatively spliced more often than genes encoding non-interacting proteins, although there is no significant difference in the number of isoforms of alternatively spliced genes.
Conclusion
Filtering for functional isoforms satisfying biological constraints and accountung for uneven EST coverage allowed us to describe differences in alternative splicing of genes from different functional categories. The observations seem to be consistent with expectations based on current biological knowledge: less isoforms for ribosomal and signal transduction proteins, and more alternative splicing of interacting and cell cycle proteins.
==== Body
Background
The current estimates of the prevalence of alternative splicing in the human genome fall into the interval 35–60% [1-7], whereas the estimated number of human protein-coding genes has decreased from more than 100 thousand [8] through 30–35 thousand [3,9,10] to 20–25 thousand [11,12]. Thus alternative splicing emerges as a major mechanism of generating protein diversity. Continuing sequencing of ESTs, whose number currently approaches 4 million, uncovers rare, tissue- and stage-specific isoforms. On the other hand, a considerable number of ESTs seem to arise from experimental artifacts (genome contamination, unspliced transcripts, computational errors leading to mis-alignment and clustering ESTs from paralogous genes, etc.) or errors of the cellular splicing machinery itself (so-called aberrant splicing). The latter might be a relatively frequent event, as there exists a special mechanism for surveillance of splicing errors, leading to elimination of aberrant mRNA isoforms by nonsense-mediated decay [13].
The algorithms for construction and enumeration of full-length isoforms should take into account as many sources of errors as possible. In early studies EST contigs were constructed as consensus exon sequences so that each exon was used only once. This precluded combinatorial explosion, but led to generation of short single-exon contigs, and besides did not allow for enumeration of isoforms. The use of contigs for estimation of the number of isoforms [14] leads to estimates that depend on EST coverage [15]. Recent algorithms construct splicing graph whose vertices correspond to sites and edges to sequence fragments in such a way that each path in this graph corresponds to a possible isoform [16-18]. Here we apply the IsoformCounter algorithm that constructs the splicing graph aligned to the genomic sequence and computes the number of possible protein isoforms. The latter procedure employs coverage-dependent thresholds for filtering artifacts.
We used IsoformCounter to compute the number of isoforms of alternatively spliced genes from the EDAS database [19] for different functional classes of proteins from GO [20]. We estimated the probability of spliceosome error (1.2%) and suggested a simple probabilistic model for filtering exons and introns obtained by EST-genome spliced alignment. As a result we have obtained a robust method for enumeration of protein isoforms independent of EST coverage. We observed that the fraction of genes with less alternative splicing (one or two protein isoforms per gene) is higher in "Small GTPase-mediated signal transduction" and "Ribosome" classes, and lower in the "DNA replication and chromosome cycle" class, compared to the average distribution. We also analyzed the correlation between alternative splicing and protein-protein interactions and demonstrated that interacting proteins are more likely to be encoded by alternatively spliced genes.
Results
Algorithm for counting alternatively spliced isoforms
The following terms will be used. Exons and introns are genome fragments that correspond to exons and introns respectively in spliced alignment of some EST, mRNA, or protein with genomic DNA. Initial and terminal exons correspond respectively to first and last exons in spliced alignment. Support of an exon is the number of clone libraries containing ESTs whose spliced alignments contain exactly this exon; specific cases are spliced alignments with mRNA and proteins that generate exons of mRNA and protein support respectively. Protein-supported exons are ascribed the reading frame derived from protein spliced alignment, whereas EST- and mRNA-supported exons are considered as triples with the same splicing sites and all possible reading frames.
For each gene (genomic fragment and corresponding ESTs, mRNAs and proteins) the algorithm constructs the splicing graph. Each splicing site corresponds to three vertices of this graph (for three possible positions relative to the reading frame), and its edges are exons and introns. The reading frames of vertices and corresponding edges are consistent. Thus each path through this graph corresponds to a candidate mRNA isoform. There is also a special type of vertices, start and stop codons, that open and close a reading frame respectively. A protein isoform is a path starting at a start codon or 5'-boundary of an initial exon and ending at a stop codon, with an additional condition that initial and terminal exons are supported by at least two clone libraries.
The IsoformCounter algorithm filters isoforms (paths) unlikely to be expressed as a functional protein. The filters are listed below.
(1) Start codons (Fig. 1)
Figure 1 Representation of start and stop codons in the splicing graph. Rectangles: exons. Angle lines: introns. Circles: stop codons. Arrows: start codons. Filled arows: start codons generated by alignment with proteins. Crossed arrows: start codons in-frame with an upstream acceptor site or a start codon. The following situations are considered and marked on the scheme:. (1) In each exon, for each possible reading frame, a stop codon closing this reading frame generates a stop-vertex. (2) If an exon contains a start codon preceded by a stop codon in the same reading frame, a start-vertex is generated. If a start codon coincides with the beginning of a protein-genome alignment, it generates a start-vertex irrespective of upstream stop codons (gray exon). In the latter case, no additional filters are applied (see the text). (3) Any start-vertex that is in-frame with an upstream acceptor site or a start codon is removed (crossed arrow).
An ATG codon generates a vertex only if it is confirmed by spliced alignment with a protein or if it is preceded in an exon by a stop codon in the same reading frame. Thus a protein isoform starts with the leftmost methionine in a given reading frame or by a methionine supported by a protein spliced alignment. To account for a possibility of insufficient coverage of a gene 5'-region, we also allow a protein isoform to start at a 5'-end of an initial exon, if this isoform does not contain in-frame protein-supported methionine codon.
(2) Initiation of translation (Fig. 2)
Figure 2 Filters on translation initiation for start-vertices and nonsense-mediated decay for stop-vertices. Circles: stop-vertices. Arrows: start-vertices. (1) Filters are not applied to vertices supported by protein-genome alignment (filled symbols). (2) Translation initiation filter: start-vertices preceded by at least two ATG codons in at least one path are removed. (3) Nonsense-mediated decay filter: stop-vertices for which the distance to the nearest donor site exceeds 55 nucleotides are removed.
Translation initiation of most eukaryotic mRNAs uses the so-called linear scanning mechanism: 40S ribosomal subunit binds the cap structure at the 5'-terminus of the mRNA and slides to the proximal ATG codon, where it initiates translation, if the codon is in a favorable context [25]. Stable hairpins and upstream ATG codons decrease the efficiency of the linear scanning. A minor fraction of mRNAs (2–8%) contain internal sites of translation initiation [26,27]. To emulate this mechanism, IsofomCounter considers only isoforms with at most two upstream ATGs, unless the start codon is supported by a protein.
(3) Short protein isoforms
We assume that alternative splicing may eliminate up to 50% of the average protein length, that is, the average length of proteins from RefSeq corresponding to the given gene. If the gene does not have RefSeq entries, the average length of corresponding proteins in EDAS is used. Further, no proteins shorter than 33 amino acids were considered.
(4) Consistency with proteins
We required that an isoform had at least one amino acid overlapping with a known protein encoded by the gene. This filter is sufficient to remove long open reading frames in 5'-untranslated regions. On the other hand, selecting a stricter threshold would lead to the loss of some known isoforms. Further, we require that there are no conflicts between reading frames generated by intersections with proteins.
(5) Premature termination of translation (Fig. 2)
It is known that transcripts containing premature stop-codons due, in particular, to spliceosomal errors, are degraded by a specific mechanism of nonsense-mediated decay (NMD) [13]. As we have no criterion for distinguishing between functional and aberrant alternative splicing, we implemented a filter imitating the NMD action, requiring that the last exon-exon junction in an isoform were at most 55 nucleotides downstream its stop-codon. As above, this filter was not applied to protein-supported isoforms.
Counting protein isoforms
To compute the number of protein isoforms, IsoformCounter finds the number of all paths in the splicing graph and subtracts the number of paths not consistent with any known protein, see filter (4). The former value is computed in linear time (respective the number of edges) by dynamic programming. To compute the latter value, IsoformCounter constructs a subset A of acceptor sites of protein-supported exons and exons that overlap the former in consistent reading frame. Then the complement set A* of acceptor sites is determined, and the number of paths coming through A* (and not through any site from A) is calculated by dynamic programming. By definition, these paths are not consistent with any known protein.
Computing the numer of alternative regions in the longest protein isoform
Constitutive regions are exon fragments whose genome projections never overlap with introns or intergenic spacers. The procedures described above allow for segmentation of the longest protein isoform into constitutive and alternative regions so that only valid isoforms that have passed all filters are taken into account.
Normalization procedure for EST-derived exons and introns
The following model was considered. Let α be the probability of splicing error (loss of a site), and let ξ be the expression level (average number of gene transcripts per cell). We assume that ξ = f(N), where N is the observed number of ESTs. Let P(N) be the probability that the cell contains at least one aberrantly spliced transcript, P = 1 - (1 - α)ξ. Then the probability that the error occurred in an interval supported by k clone libraries is Pk ≤ β, where β denotes the significance level, that is, the probability with which we accept a splicing error. Solving this inequality with respect to k, we obtain k(ξ) ≥ ln(β) / ln(1 - (1 - α)ξ). We estimate the expression level as ξ = N/5. To estimate the probability of a splicing error, we considered losses of one or two sites corresponding to a protein-supported intron. The number of such events Noverlap can be estimated as the number of ESTs which overlap the intron sites and are not spliced at these sites. This is an overestimate, as real alternative splicing events also are counted. Further, let Nsplice be the number of ESTs whose spliced alignment contain protein-supported introns. The probability of the spliceosome error was estimated as α = Noverlap/(Noverlap+Nsplice) computed as 0.012. Note that one EST could be counted both for Noverlap and Nsplice, so the above value could be an underestimate. Finally, we used the following threshold on the number of clone libraries required to accept an exon (dependent on the EST coverage of the gene): k( N) = [- 1/ln(1-0,988N/5)]. Exons and introns confirmed by alignment with mRNA or protein are always accepted.
The restricted set of protein isoforms
For each gene in EDAS we constructed the restricted set of protein isoforms that are amino acid sequences are available at [37]. Define isoform support as the minimum support of exons and introns forming the isoform. For each edge in the splicing graph (exon or intron) we construct the longest isoform passing through the edge, whose support is not less than the edge support. At each level of support, the restricted set consists of such longest isoforms for all edges of the given support.
Analysis of alternative splicing in functional categories of genes
When IsoformCounter was applied to all genes from EDAS, in 431 cases (4%) no isoforms were found. This could happen in one of there cases. (1) Protein-DNA spliced alignment does not end at stop codon, and there is no downstream terminal exon where the induced reading frame contains a stop codon. (2) Protein-DNA spliced alignment does not start at methionine, and there is no upstream exon where the induced open reading frame contains an ATG codon preceded by a stop codon (see filter 1 in Methods). (3) Protein-DNA spliced alignment does not start at methionine and the candidate ATG found by the algorithm is eliminated by the filter on translation initiation (filter 2), or the alignment does not end at a stop codon, and the candidate stop codon is eliminated by the filter on premature termination of translation (nonsense-mediated decay, filter 5). The first and second cases are due to incomplete proteins where no candidate start or stop codon could be assigned. In the third case a start or stop codon could be assigned based on spliced alignment with incomplete protein and ESTs, but the obtained reading frame was eliminated by the filters. It may happen if the gene has an internal translation initiation site or a special mechanism to keep the isoform from the NMD degradation. In both cases algorithm needs an alignment with a complete protein. All these genes were ignored.
The obtained distribution of the isoform numbers is shown in Fig. 3 (blue columns). As it is known that the predicted number of isoforms may depend on the EST coverage, we analyzed the dependence between the EST coverage of a gene and the number of isoforms (Fig. 4). The blue plot, corresponding to the initial (raw) data demonstrates that genes with high EST coverage (>900 ESTs per gene) have a large number of isoforms (fall in the tail of the distribution). In particular, this tail contains many genes of ribosomal proteins (Fig. 5), which seems to be an artifact. Indeed, it is highly likely that the high expression level of a gene leads to appearance of relatively rare aberrant isoforms that are not seen for weakly expressed genes.
Figure 3 Blue columns: raw data. Red columns: normalized data (see Results). The difference between histograms before and after normalization is weak, because the fraction of highly expressed genes (>400 ESTs) is small (approximately 4%).
Figure 4 Correlation between the isoform number and EST coverage. Blue: raw data (all ESTs). Red: normalized data (coverage-dependent filter on the number of clone libraries supporting exons, see Results). Each dot represents the average EST coverage for genes with the given number of isoforms. The peak in the normalized plot corresponds to the gene "eukaryotic translation elongation factor 1 alpha 1", represented by 18841 EST.
Figure 5 Influence of normalization on the isoform number of proteins from the "Ribosome" GO category. Color code as in Fig. 1.
Thus we believe that the threshold level of EST support for exons and introns should depend on the EST coverage of a gene, so that weakly and highly expressed genes could be comparable. Red columns in Fig. 3 show the number of isoforms with this more stringent threshold. The distributions with raw and normalized data do not differ much, but comparison of two plots in Fig. 4 shows that normalization removes the dependence between EST coverage and isoform number. One functional category strongly affected by the normalization procedure is "ribosome". Further we consider results obtained after normalization.
Most genes (91%) have a relatively small number of isoforms (1 through 15). The number of genes with an even number of isoforms is higher than the number of genes with odd number of isoforms. Indeed, the algorithm assumes independence of individual elementary alternatives, and thus the number of paths between two alternatives is roughly the product of the number of variants. Moreover, most local alternatives preserve the reading frame. Thus, to have an even number of alternatives, it is sufficient to have a frame-preserving local alternative with two variants (e.g. a short cassette exon). 23% of genes had only one functional isoform.
Fig. 5 shows the distribution of the number of alternative and constitutive regions in the longest isoform. The fraction of genes not containing constitutive fragments at all is ~1%, which shows that the applied filters remove a considerable fraction of aberrant events. Without these filters most genes would contain only alternative regions. The fraction of constitutive genes represented in Fig. 6 is 24%, which is higher than the above estimate. It is caused by the fact that introns not overlapping with the longest isoform are not taken into account. The average number of all (alternative and constitutive) fragments per gene is 3.7.
Figure 6 Number of constitutive (blue) and alternative (red) regions in the longest isoform. The fraction of completely alternative genes is ~1%.
We considered the link between protein function and the isoform number. The following functional categories from GO were considered: "Small GTPase-mediated signal transduction" (145 genes), "Catabolism" (512 genes), "DNA replication and chromosome cycle" (99 genes), "Ribosome" (123 genes). Significant differences from the distribution for all genes (p = 0.003 according to the Mann-Whitney U test) were observed for "Ribosome" and "Small GTPase-mediated signal transduction" categories. Both of them contain fewer than expected genes with a large number of isoforms. In particular, there are 46% constititutive genes in "Ribosome", although to very high EST coverage this is observable only after normalization (Fig. 5). Genes from the "DNA replication and chromosome cycle" have more isoforms that the average (p = 0.07 according to the Mann-Whitney U test). In particular, there is a higher fraction of genes with two or more isoforms. The distribution of the isoform number for genes from these categories is shown in Fig. 7.
Figure 7 Distribution of the isoform number in GO functional clusters.
Of 452 human interaction pairs in the MPPI database, 332 pairs were heterogeneous and different (excluding protein contacts with itself and pairs that differ only by the protein order). LocusLink information was available for 312 of these pairs, containing 386 proteins. Of these proteins, 262 are encoded by genes from EDAS, and pairs with both members present in EDAS form 198 interacting pairs.
No correlation was observed between the number of contacts for a given protein and the number of isoforms or alternative regions (data not shown). However, the probability to be alternatively spliced was higher for genes encoding proteins participating in at least one protein-protein interaction (Table 1). This observation was significant for alternative splicing with all considered levels of support (only proteins, proteins and mRNA, ESTs from different number of clone libraries, ESTs with normalized threshold) at the level <0.1–1% (the highest χ2 = 11.2 for protein-supported alternatives, χ2 from 6.5 through 7.8 for EST-supported alternatives), although the difference between observed and expected numbers is not large (17–30% deficit of constitutive and 10–25% excess of alternatively spliced genes among those encoding interacting proteins). No significant correlations were observed for larger, but noisier non-curated protein-protein interaction datasets DIP and OPHID (data not shown).
Table 1 Correlation between alternative splicing (AS) and protein-protein interactions (PPI). Expected numbers under independency assumption are given in parentheses. "EST-N" denotes isoforms with each exon supported by ESTs from at least N clone libraries.
protein data, χ2 = 11.17 No PPI At least one PPI TOTAL
No AS 5434 (5408) 122 (= 83% of 148) 5556
At least two AS-isoforms 3692 (3718) 127 (= 125% of 101) 3819
TOTAL 9126 249 9375
mRNA data, χ2 = 9.05 No PPI At least one PPI TOTAL
No AS 4879 (4856) 107 (= 82% of 130) 4986
At least two AS-isoforms 4360 (4383) 141 (= 120% of 118) 4501
TOTAL 9239 248 9487
EST-5 data, χ2 = 6.57 No PPI At least one PPI TOTAL
No AS 4463 (4443) 99 (= 83% of 119) 4562
At least two AS-isoforms 4804 (4324) 149 (= 115% of 129) 4953
TOTAL 9267 248 9515
EST-3 data, χ2 = 7.79 No PPI At least one PPI TOTAL
No AS 4001 (3979) 85 (= 80% of 107) 4086
At least two AS-isoforms 5302 (5324) 164 (= 115% of 142) 5466
TOTAL 9303 249 9552
EST-2 data, χ2 = 7.57 No PPI At least one PPI TOTAL
No AS 3299 (3278) 68 (= 77% of 89) 3367
At least two AS-isoforms 6063 (6084) 185 (= 113% of 164) 6248
TOTAL 9362 253 9615
normalized data, χ2 = 7.23 No PPI At least one PPI TOTAL
No AS 2275 (2257) 42 (= 70% of 60) 2317
At least two AS-isoforms 7066 (7084) 206 (= 110% of 188) 7272
TOTAL 9341 248 9589
Discussion and conclusion
IsoformCounter is a system of filters aiming at distinguishing functional isoforms from non-functional ones. Unlike other programs for isoform generation [18,28] it assumes independence of variants selected at elementary alternatives.
The prevalence of genes with a relatively small number of isoforms agrees with the observation of [29], where all genes had less than 18 isoforms. On the other hand, the prevalence of genes with the even number of isoforms not observed in [29], where the minimal number of isoforms required to explain all local alternatives was computed, as opposed to all isoforms. The fact that 73–77% genes had more than one isoform also is consistent with previous estimates: the fraction of single-exon human genes is 20% [30], and, as the latter are not covered by EDAS, the fraction of alternatively spliced genes is approximately 60% (75% of 80%) [14,31].
It has been reported that alternative splicing tends to affect genes involved in signal transduction [31,32], although no estimates on the significance of these findings was done. We have also expected that there would be considerable differences between such GO categories as "Metabolism" and "Signal transduction". However, this was not observed, probably because these categories are too large and contain many genes with diverse functions and properties. Still, we observed significantly lower number of isoforms for genes from the "Small GTPase-mediated signal transduction" (compared to the distribution for all genes)
The correlation between alternative splicing of genes and protein-protein interactions of encoded proteins was somewhat unexpected, especially given previously described lack of correlation between alternative and contacting regions [33]. Further, although it has been reported that alternative splicing often targets domains involved in protein-protein interactions [34,35], there was no increase in the rate of alternative splicing of such domains.
The effect observed here was not particularly strong (~20–25%), but still statistically significant, and was independent from the support level of alternatively spliced isoforms. On the other hand, the result crucially depended on the reliability of protein-protein interaction data: no correlation was observed for non-curated, large-scale experimental (DIP) or inferred (OPHID) data.
Methods
The data about alternative splicing of human genes were taken from the EDAS database [19,36]. EDAS contains information about 9986 human genes (9914 with LocusLink identifiers) of which 8324 (83%) show at least some evidence of alternative splicing. The criteria for inclusion of a gene into EDAS were as follows: at least one linked protein sequence, at least one intron in the coding region, and at least 25 ESTs.
The data about protein-protein interactions (PPI) were taken from the manually curated MPPI database [21,39] containing 452 pairs of interacting human proteins. We also considered two PPI datasets, non-curated database of PPI interactions from large-scale experiments (DIP) [22] and predicted PPI (OPHID) [23]. Functional categories of genes were taken from GeneOnthology [20,40].
Spliced alignments were constructed by Pro-Frame [24] (protein-DNA) and Pro-EST [1] (mRNA-DNA and EST-DNA).
Availability
The software (IsoformCounter) for generating of alternative mRNA isoforms is available for download at [38].
Authors' contributions
AAM, DF and MSG conceived the project. RNN developed the EDAS database. AAM and ADN developed the algorithm for counting isoforms. ADN analyzed the correlation with functional categories. IIA and DF analyzed the correlation with protein-protein interactions. ADN and MSG wrote the draft. All authors edited the final text.
Acknowledgements
We are grateful to Vassily Ramensky and Dmitry Malko for useful discussions. This study was partially supported by grants from the Howard Hughes Medical Institute (grant 55000309), the Ludwig Institute for Cancer Research (grant CRDF 1268), The Russian Fund of Basic Research (grant 04-04-49440), The Russian Science Support Fund, The Russian Academy of Sciences (programs "Molecular and Cellular Biology" and "Origin and Evolution of the Biosphere"), and the FP6 Programme of the European Commission (BioSapiens project, contract number LHSG-CT-2003-503265).
==== Refs
Mironov AA Fickett JW Gelfand MS Frequent alternative splicing of human genes Genome Res 1999 9 1288 1293 10613851 10.1101/gr.9.12.1288
Brett D Hanke J Lehmann G Haase S Delbruckn S Krueger S Reich J Bork P EST comparison indicates 38% of human mRNAs contain possible alternative splice forms FEBS Lett 2000 474 83 86 10828456 10.1016/S0014-5793(00)01581-7
Lander ES Linton LM Birren B Nusbaum C Zody MC Baldwin J Devon K Dewar K Doyle M FitzHugh W Funke R Gage D Harris K Heaford A Howland J Kann L Lehoczky J LeVine R McEwan P McKernan K Meldrim J Mesirov JP Miranda C Morris W Naylor J Raymond C Rosetti M Santos R Sheridan A Sougnez C Stange-Thomann N Stojanovic N Subramanian A Wyman D Rogers J Sulston J Ainscough R Beck S Bentley D Burton J Clee C Carter N Coulson A Deadman R Deloukas P Dunham A Dunham I Durbin R French L Grafham D Gregory S Hubbard T Humphray S Hunt A Jones M Lloyd C McMurray A Matthews L Mercer S Milne S Mullikin JC Mungall A Plumb R Ross M Shownkeen R Sims S Waterston RH Wilson RK Hillier LW McPherson JD Marra MA Mardis ER Fulton LA Chinwalla AT Pepin KH Gish WR Chissoe SL Wendl MC Delehaunty KD Miner TL Delehaunty A Kramer JB Cook LL Fulton RS Johnson DL Minx PJ Clifton SW Hawkins T Branscomb E Predki P Richardson P Wenning S Slezak T Doggett N Cheng JF Olsen A Lucas S Elkin C Uberbacher E Frazier M Gibbs RA Muzny DM Scherer SE Bouck JB Sodergren EJ Worley KC Rives CM Gorrell JH Metzker ML Naylor SL Kucherlapati RS Nelson DL Weinstock GM Sakaki Y Fujiyama A Hattori M Yada T Toyoda A Itoh T Kawagoe C Watanabe H Totoki Y Taylor T Weissenbach J Heilig R Saurin W Artiguenave F Brottier P Bruls T Pelletier E Robert C Wincker P Smith DR Doucette-Stamm L Rubenfield M Weinstock K Lee HM Dubois J Rosenthal A Platzer M Nyakatura G Taudien S Rump A Yang H Yu J Wang J Huang G Gu J Hood L Rowen L Madan A Qin S Davis RW Federspiel NA Abola AP Proctor MJ Myers RM Schmutz J Dickson M Grimwood J Cox DR Olson MV Kaul R Raymond C Shimizu N Kawasaki K Minoshima S Evans GA Athanasiou M Schultz R Roe BA Chen F Pan H Ramser J Lehrach H Reinhardt R McCombie WR de la Bastide M Dedhia N Blocker H Hornischer K Nordsiek G Agarwala R Aravind L Bailey JA Bateman A Batzoglou S Birney E Bork P Brown DG Burge CB Cerutti L Chen HC Church D Clamp M Copley RR Doerks T Eddy SR Eichler EE Furey TS Galagan J Gilbert JG Harmon C Hayashizaki Y Haussler D Hermjakob H Hokamp K Jang W Johnson LS Jones TA Kasif S Kaspryzk A Kennedy S Kent WJ Kitts P Koonin EV Korf I Kulp D Lancet D Lowe TM McLysaght A Mikkelsen T Moran JV Mulder N Pollara VJ Ponting CP Schuler G Schultz J Slater G Smit AF Stupka E Szustakowski J Thierry-Mieg D Thierry-Mieg J Wagner L Wallis J Wheeler R Williams A Wolf YI Wolfe KH Yang SP Yeh RF Collins F Guyer MS Peterson J Felsenfeld A Wetterstrand KA Patrinos A Morgan MJ de Jong P Catanese JJ Osoegawa K Shizuya H Choi S Chen YJ International Human Genome Sequencing Consortium Initial sequencing and analysis of the human genome Nature 2001 409 860 921 11237011 10.1038/35057062
Brett D Pospisil H Valcarcel J Reich J Bork P Alternative splicing and genome complexity Nature Genet 2002 30 29 30 11743582 10.1038/ng803
FANTOM2 Consortium and the RIKEN GSC Genome Exploration Group Phase I & II Team Analysis of the mouse transcriptome based upon functional annotation of 60,770 full length cDNAs Nature 2002 420 563 573 12466851 10.1038/nature01266
Zavolan M van Nimwegen E Gaasterland T Splice variation in mouse full-length cDNAs identified by mapping to the mouse genome Genome Res 2003 12 1377 1385 12213775 10.1101/gr.191702
Imanishi T Itoh T Suzuki Y O'Donovan C Fukuchi S Koyanagi KO Barrero RA Tamura T Yamaguchi-Kabata Y Tanino M Integrative annotation of 21,037 human genes validated by full-length cDNA clones PloS Biology 2004 2 1 20 10.1371/journal.pbio.0020162
Liang F Holt I Pertea G Karamycheva S Salzberg SL Quackenbush J Gene index analysis of the human genome estimates approximately 120,000 genes Nat Genet 2000 25 239 240 10835646 10.1038/76126
Ewing B Green P Analysis of expressed sequence tags indicates 35,000 human genes Nat Genet 2000 25 232 234 10835644 10.1038/76115
Venter JC Adams MD Myers EW Li PW Mural RJ Sutton GG Smith HO Yandell M Evans CA Holt RA Gocayne JD Amanatides P Ballew RM Huson DH Wortman JR Zhang Q Kodira CD Zheng XH Chen L Skupski M Subramanian G Thomas PD Zhang J Gabor Miklos GL Nelson C Broder S Clark AG Nadeau J McKusick VA Zinder N Levine AJ Roberts RJ Simon M Slayman C Hunkapiller M Bolanos R Delcher A Dew I Fasulo D Flanigan M Florea L Halpern A Hannenhalli S Kravitz S Levy S Mobarry C Reinert K Remington K Abu-Threideh J Beasley E Biddick K Bonazzi V Brandon R Cargill M Chandramouliswaran I Charlab R Chaturvedi K Deng Z Di Francesco V Dunn P Eilbeck K Evangelista C Gabrielian AE Gan W Ge W Gong F Gu Z Guan P Heiman TJ Higgins ME Ji RR Ke Z Ketchum KA Lai Z Lei Y Li Z Li J Liang Y Lin X Lu F Merkulov GV Milshina N Moore HM Naik AK Narayan VA Neelam B Nusskern D Rusch DB Salzberg S Shao W Shue B Sun J Wang Z Wang A Wang X Wang J Wei M Wides R Xiao C Yan C Yao A Ye J Zhan M Zhang W Zhang H Zhao Q Zheng L Zhong F Zhong W Zhu S Zhao S Gilbert D Baumhueter S Spier G Carter C Cravchik A Woodage T Ali F An H Awe A Baldwin D Baden H Barnstead M Barrow I Beeson K Busam D Carver A Center A Cheng ML Curry L Danaher S Davenport L Desilets R Dietz S Dodson K Doup L Ferriera S Garg N Gluecksmann A Hart B Haynes J Haynes C Heiner C Hladun S Hostin D Houck J Howland T Ibegwam C Johnson J Kalush F Kline L Koduru S Love A Mann F May D McCawley S McIntosh T McMullen I Moy M Moy L Murphy B Nelson K Pfannkoch C Pratts E Puri V Qureshi H Reardon M Rodriguez R Rogers YH Romblad D Ruhfel B Scott R Sitter C Smallwood M Stewart E Strong R Suh E Thomas R Tint NN Tse S Vech C Wang G Wetter J Williams S Williams M Windsor S Winn-Deen E Wolfe K Zaveri J Zaveri K Abril JF Guigo R Campbell MJ Sjolander KV Karlak B Kejariwal A Mi H Lazareva B Hatton T Narechania A Diemer K Muruganujan A Guo N Sato S Bafna V Istrail S Lippert R Schwartz R Walenz B Yooseph S Allen D Basu A Baxendale J Blick L Caminha M Carnes-Stine J Caulk P Chiang YH Coyne M Dahlke C Mays A Dombroski M Donnelly M Ely D Esparham S Fosler C Gire H Glanowski S Glasser K Glodek A Gorokhov M Graham K Gropman B Harris M Heil J Henderson S Hoover J Jennings D Jordan C Jordan J Kasha J Kagan L Kraft C Levitsky A Lewis M Liu X Lopez J Ma D Majoros W McDaniel J Murphy S Newman M Nguyen T Nguyen N Nodell M Pan S Peck J Peterson M Rowe W Sanders R Scott J Simpson M Smith T Sprague A Stockwell T Turner R Venter E Wang M Wen M Wu D Wu M Xia A Zandieh A Zhu X The sequence of the human genome Science 2001 291 1304 1351 11181995 10.1126/science.1058040
International Human Genome Sequencing Consortium Finishing the euchromatic sequence of the human genome Nature 2004 431 931 45 15496913 10.1038/nature03001
Southan C Has the yo-yo stopped? An assessment of human protein-coding gene number Proteomics 2004 4 1712 1726 15174140 10.1002/pmic.200300700
Lejeune F Li X Maquat LE Nonsense-mediated mRNA decay in mammalian cells involves decapping, deadenylating, and exonucleolytic activities Mol Cell 2003 12 536 538 14527400 10.1016/S1097-2765(03)00349-6
Kim H Klein R Majewski J Ott J Estimating rates of alternative splicing in mammals and invertebrates Nature Genet 2004 36 915 916 15340420 10.1038/ng0904-915
Harrington ED Buoe S Valcarcel J Reich JG Bork P Estimating rates of alternative splicing in mammals and invertebrates Nature Genet 2004 916 917 10.1038/ng0904-916
Kan Z States D Gish W Selecting for functional alternative splices in ESTs Genome Res 2002 12 1837 1845 12466287 10.1101/gr.764102
Heber S Alekseyev M Sze SH Tang H Pevzner PA Splicing graphs and EST assembly problem Bioinformatics 2002 18 S181 S188 12169546
Lee C Generating consensus sequences from partial order multiple sequence alignment graphs Bioinformatics 2003 19 999 1008 12761063 10.1093/bioinformatics/btg109
Nurtdinov RN Kosmodemyansky I The EDAS (EST-Derived Alternative Splicing) database Proceedings of the International Moscow Conference on Computational Molecular Biology MCCMB'03: July 22–25 2003 2003 233 235
Ashburner M Ball CA Blake JA Botstein D Butler H Cherry JM Davis AP Dolinski K Dwight SS Eppig JT Harris MA Hill DP Issel-Tarver L Kasarskis A Lewis S Matese JC Richardson JE Ringwald M Rubin GM Sherlock G Gene ontology: tool for the unification of biology. The Gene Ontology Consortium Nat Genet 2000 25 25 29 10802651 10.1038/75556
Pagel P Kovac S Oesterheld M Brauner B Dunger-Kaltenbach I Frishman G Montrone C Mark P Stümpflen V Mewes HW Ruepp A Frishman D The MIPS mammalian protein-protein interaction database Bioinformatics 2005 Advance Access published online. Bioinformatics, doi:10.1093/bioinformatics/bti115 15531608
Salwinski L Miller CS Smith AJ Pettit FK Bowie JU Eisenberg D The Database of Interacting Proteins Update NAR Database issue 2004 32 D449 D451
von Mering C Krause R Snel B Cornell M Oliver Stephen G Fields S Bork P Comparative assessment of large-scale data sets of protein-protein interactions Nature 2002 417 399 403 12000970 10.1038/nature750
Mironov AA Novichkov PS Gelfand MS Pro-Frame: Similarity-based gene recognition in eukaryotic DNA sequences with errors Bioinformatics 2001 17 13 15 11222258 10.1093/bioinformatics/17.1.13
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
Pestova TV Kolupaeva VG Lomakin IB Pilipenko EV Shatsky IN Agol VI Hellen CU Molecular mechanisms of translation initiation in eukaryotes Proc Natl Acad Sci USA 2001 98 7029 7036 11416183 10.1073/pnas.111145798
Kochetov AV Pichueva AG Kondrakhin YA Titov SE Kolchanov NA The contextual features of higher plant mRNA 5'-untranslated regions as related to the trasnlation initiation mechanisms Biophysics (Moscow) 2003 48 76 80
Kan Z Rouchka EC Gish WR States DJ Gene structure prediction and alternative splicing analysis using genomically aligned ESTs Genome Res 2001 11 889 900 11337482 10.1101/gr.155001
Xing Y Resch A Lee C The multiassembly problem: reconstructing multiple transcript isoforms from EST fragment mixtures Genome Res 2004 14 426 441 14962984 10.1101/gr.1304504
Burge C Karlin S Prediction of Complete Gene Structures in Human Genomic DNA J Mol Biol 1997 268 78 94 9149143 10.1006/jmbi.1997.0951
Modrek B Resch A Grasso C Lee C Genome-wide detection of alternative splicing in expressed sequences of human genes Nucleic Acids Res 2001 29 2850 2859 11433032 10.1093/nar/29.13.2850
Johnson JM Castle J Garrett-Engele P Kan Z Loerch PM Armour CD Santos R Schadt EE Stoughton R Shoemaker DD Genome-wide survey of human alternative pre-mRNA splicing with exon junction microarrays Science 2003 302 2141 2144 14684825 10.1126/science.1090100
Offman MN Nurtdinov RN Gelfand MS Frishman D No statistical support for correlation between the positions of protein interaction sites and alternatively spliced regions BMC Bioinformatics 2004 5 41 15096275 10.1186/1471-2105-5-41
Zavolan M van Nimwegen E Gaasterland T Splice variation in mouse full-length cDNAs identified by mapping to the mouse genome Genome Res 2003 12 1377 1385 12213775 10.1101/gr.191702
Resch A Xing Y Modrek B Gorlick M Riley R Lee C Assessing the impact of alternative splicing on domain interactions in the human proteome J Proteome Res 2004 3 76 83 14998166 10.1021/pr034064v
EDAS: EST-Derived Alternative Splicing Database
EDAS Summary information about Homo sapiens genes
EDAS IsoformCounter page
The MIPS Mammalian Protein-Protein Interaction Database
The Gene Ontology
|
16274476
|
PMC1298288
|
CC BY
|
2021-01-04 16:27:47
|
no
|
BMC Bioinformatics. 2005 Nov 7; 6:266
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-266
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2721628865210.1186/1471-2105-6-272Methodology ArticleAutomated methods of predicting the function of biological sequences using GO and BLAST Jones Craig E [email protected] Ute [email protected] Alfred L [email protected] Australian Centre for Plant Functional Genomics, Waite Campus, University of Adelaide, South Australia, 5064, Australia2 School of Computer Science, University of Adelaide, South Australia, 5001, Australia2005 15 11 2005 6 272 272 15 6 2005 15 11 2005 Copyright © 2005 Jones et al; licensee BioMed Central Ltd.2005Jones 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
With the exponential increase in genomic sequence data there is a need to develop automated approaches to deducing the biological functions of novel sequences with high accuracy. Our aim is to demonstrate how accuracy benchmarking can be used in a decision-making process evaluating competing designs of biological function predictors. We utilise the Gene Ontology, GO, a directed acyclic graph of functional terms, to annotate sequences with functional information describing their biological context. Initially we examine the effect on accuracy scores of increasing the allowed distance between predicted and a test set of curator assigned terms. Next we evaluate several annotator methods using accuracy benchmarking. Given an unannotated sequence we use the Basic Local Alignment Search Tool, BLAST, to find similar sequences that have already been assigned GO terms by curators. A number of methods were developed that utilise terms associated with the best five matching sequences. These methods were compared against a benchmark method of simply using terms associated with the best BLAST-matched sequence (best BLAST approach).
Results
The precision and recall of estimates increases rapidly as the amount of distance permitted between a predicted term and a correct term assignment increases. Accuracy benchmarking allows a comparison of annotation methods. A covering graph approach performs poorly, except where the term assignment rate is high. A term distance concordance approach has a similar accuracy to the best BLAST approach, demonstrating lower precision but higher recall. However, a discriminant function method has higher precision and recall than the best BLAST approach and other methods shown here.
Conclusion
Allowing term predictions to be counted correct if closely related to a correct term decreases the reliability of the accuracy score. As such we recommend using accuracy measures that require exact matching of predicted terms with curator assigned terms. Furthermore, we conclude that competing designs of BLAST-based GO term annotators can be effectively compared using an accuracy benchmarking approach. The most accurate annotation method was developed using data mining techniques. As such we recommend that designers of term annotators utilise accuracy benchmarking and data mining to ensure newly developed annotators are of high quality.
==== Body
Background
Genomics research is generating enormous quantities of DNA and protein sequence data. GenBank, a major repository of genomic data, reports an exponential increase in sequence data, in the last 10 years the quantity of data has increased more than two-hundred-fold [1]. Sequence data alone is of limited use to biologists, being simply a linear array of base or amino acid codes. To make the most of the data biologists need to be able to place the sequence within a biological context, that is, they require information concerning the biological properties and functions that the DNA or protein sequence might be considered to have from an expert's point of view. This has led to the need for software able to conduct high-throughput, accurate function prediction.
The creation of the Gene Ontology, GO, has provided a rich resource for describing the functional characteristics of sequences [2]. GO is widely used, both in the analysis of microarray data and generally in comparative genomics, to abstract above the level of sequences to that of function. Essentially GO contains three ontologies, describing the biological process, cellular compartment and molecular function properties of sequences. Each ontology is a directed acyclic graph of functional term nodes where edges between nodes describe relationships between them. GO is now the defacto-standard for annotating sequences with functional information.
The Basic Local Alignment Search Tool, BLAST, is the most commonly used sequence alignment application [3,4]. It allows the user to find sequences with high degrees of local similarity to query sequences. Furthermore, it supports the creation of custom sequence databases. For these reasons BLAST has been employed to assign GO terms to novel sequences, the assumption being that GO terms belonging to similar sequences will have a high likelihood of also belonging to the query sequence.
BLAST assigns an expect value to each sequence found in the sequence database based on a local alignment between that sequence and the input sequence. The expect value is based on a score assigned to gapped alignments between sequences, the size of the database, and the lengths of both sequences. Expect values less than 0.01 can be considered to be the same as the probability that two sequences match purely by chance [5]. Therefore the lower the expect value the more significant the match between sequences.
Recently a number of accounts of automated BLAST-based GO term prediction applications have been published. GOblet [6,7] is a web-based system allowing users to find GO terms for Gene Ontology Annotation database [8] curated sequences. GOFigure [9] is also a web-based system that uses BLAST to find matching sequences with existing GO term annotations, and then constructs a minimum covering graph of term nodes. Terms are assigned a score based on the expect value of the matching sequence to which they were assigned. Parents are then assigned scores associated with their child nodes. GOEngine [10] utilises a variety of data sources from literature mining to BLAST homolog analysis. In addition, data-source oriented function annotation projects utilise BLAST as a way of predicting GO terms based on sequence similarity. Such projects include the Gene Ontology Annotation Database associated with the European Bioinformatics Institute [8], NetAffx [11] associated with Affymetrix microarray probe-sets, as well as species-specific investigations [12,13].
Most published accounts of automated approaches to BLAST-based GO term prediction have demonstrated their accuracy using ad hoc methods following implementation. It is our aim to demonstrate how well planned accuracy benchmarking can be used in a decision-making process evaluating competing designs – a useful first step towards producing a high accuracy biological function predictor. Furthermore, because of the fact that terms are united by an ontology, some researchers [9,10] have allowed terms within a given number of edges to be counted as correct. As our approach is heavily dependent on precision and recall measures, we examine the impact of increasing the allowed distance between correct and predicted term nodes on the reliability of these measures.
Results
Background
Data collection and preparation
The March 2004 Gene Ontology data [14] was downloaded and imported into a MySQL database. This data consists of both protein sequence data and their GO term associations. Only proteins and their GO term associations were included in this study if term annotations were made manually, i.e. did not have the GO evidence code [15] of inferred from sequence similarity or ISS.
The resulting "manually curated" protein term associations were broken into two distinct groups:
1. UniProt [16] annotations – proteins and their GO term associations that were submitted by UniProt. This data, consisting of 7071 proteins with high quality annotations, was referred to as the 'training set'.
2. Non UniProt annotations – proteins and their GO term associations that were submitted by FlyBase, Mouse Genome Informatics (MGI), Sanger GeneDB, Saccharomyces Genome Database, and The Institute for Genome Research (TIGR). This data was referred to as the 'test set'. It consists of 19965 annotated proteins, and can be assumed to have greater variation in annotation quality.
This data provided us with a set of known 'correct' annotations for proteins, and was used to assess the effectiveness of various 'annotation methods'. Generally speaking non-data-mining approaches used the 'training set' to assess their annotation accuracy, while data-mining approaches used the 'training set' for model creation, and the 'test set' for model assessment.
Two BLAST-able databases were created using NCBI-BLAST's formatdb command, one for the training and test sets respectively. Also, all protein sequences were written to individual fasta format text files to allow for BLAST searches where they were the query sequence.
Accuracy metrics
We modified definitions of precision and recall measures to be applicable to assessing the accuracy of term assignment predictions. Given that a protein sequence has a set of correct term associations, and that an annotation method will provide a set of predicted term associations, precision and recall were defined here as:
P = c/p
where precision (P) is the proportion of correct predicted term assignments (c) of the total number of predicted assignments (p), i.e. a measure of the accuracy of predicted terms.
R = c/t
and recall (R) is the proportion of correct predicted term assignments (c) of the total number of correct terms (t), i.e. a measure of how many of the possible correct terms were returned by the method.
Furthermore, similar to Karaoz et al [17] we have adopted the harmonic mean of the precision and recall as an overall accuracy measure. This can be defined as:
H = 2/(1/P+1/R)
where H is the harmonic mean of precision and recall for a predicted term assignment.
To examine the definition of 'a correct term' we modified the allowed edge distance between a predicted term and a curator-assigned term to assess the overall impact on accuracy scores. All UniProt submitted proteins were assigned a number of terms randomly equal to the number that had been assigned by curators. A 'correct term assignment' was defined as being when a predicted term id was within an 'allowed edge-distance' to the curator-assigned term id. The allowed edge distance between these terms was increased from equality (0) to 4, and the recall and precision calculated. This analysis demonstrated that increasing the allowed edge distance between a predicted and a curator-assigned term when deciding on which predicted terms were correct would grossly increase the perceived accuracy of annotation methods. For this reason all accuracy metrics used here required that the predicted term id and curator-assigned term id had to be equal if the predicted association was to be considered correct.
Annotation methods
BLAST was used to find matching proteins between the training and test sets. Two BLAST output datasets were generated: the output of matching training-set proteins to test-set proteins with an expect value cut-off of 1e-10, and the reciprocal search (test against training set). The command line used per query (or input) protein was similar to:
./blastall -p blastp -d blastable_db_proteins -i query-sequence -o output_file -e 1e-10
Several automated annotation methods were developed using a variety of approaches to assign terms based on BLAST output. We assigned these methods the following descriptive names based on how terms were assigned: Best BLAST, Covering Graph, Term Distance Concordance, and Discriminant Function. All methods used as input the best five BLAST-matched proteins, based on descending order of expect value, and returned a set of predicted terms, for a query-protein. Initially, work utilised all BLAST-matched proteins, but subsequent testing of these annotation methods showed that using greater than five did not result in further increases in annotation accuracy. Annotation methods were then compared using precision, recall and their harmonic mean to determine the most accurate method of automatically annotating protein sequences with GO terms. Pseudocode detailing the Covering Graph and Term Distance Concordance methods, and the Term Covariance Filter, is given in the Methods section.
Best BLAST method
The best matching protein sequence returned by BLAST for each input protein was selected, and terms assigned to it by curators were then assigned as predicted term for the input protein. This method is treated as the benchmark against which other methods listed below are compared.
Covering Graph method
The terms assigned by curators to BLAST-matched proteins for a query protein were pooled. These were then broken into groups based on term ontologies (i.e. biological process, cellular location, or molecular function). For the terms within an ontology the GO directed-acyclic-graph was examined to find the closest common-ancestor term. The paths from this term to all curator-assigned terms then defined the covering graph. All of the terms along paths within the covering graph, including the common-ancestor term and curator-assigned terms, were then assigned a 'concordance score'. This score was defined in such a way as to assign higher scores to terms that are related to a greater number of curator-assigned terms. To find this, each curator-assigned term was given a concordance score based on the expect value, and the number of times the term was associated with the best five matching proteins. This concordance score was then assigned to the term's ancestors recursively upwards in the covering graph, i.e. from the curator assigned term, then assigned to their ancestors along the covering graph, stopping when the common ancestor term is reached. A variety of methods were examined in terms of using this score to select a small number of terms. These included simply selecting the ten best scoring terms, selecting all terms with a score > 0.1, and using the Term Covariance Filter (see below). We also examined the issue of whether a maximal score cut-off could be employed to increase the accuracy of the approach. We did this by declaring a cut-off threshold value, where terms with a concordance score greater than this were excluded, and finding the accuracy score given to annotations where this was varied.
Term distance concordance method
The terms assigned by curators to BLAST-matched proteins for a query protein were pooled. We then defined 'term distance' to be the number of edges present in the shortest possible path between two terms. A matrix of term distances was calculated. The matrix comprised of essentially a table showing the distance of each term to every other term assigned by curators. Terms were assigned scores by calculating the sum of the product of the inverse log of the ascending expect value rank of the BLAST match and the maximum term depth divided by the distance between terms. In cases where terms were associated with more than one BLAST-matched sequence the highest ranking match was used for this calculation. Terms were selected for annotation by either selecting the 10 best ranking terms or using the Term Covariance Filter.
Discriminant function method
Discriminant analysis [18] was undertaken to create a model for correct term assignment. Term and result data obtained from the first five BLAST results for each query sequence were examined. Result and term data were incorporated. Results were ranked in descending order of expectation value. Duplicate terms were excluded but contributed to a count of the number of times the term appeared among the results (term-result count). Attributes included were term-result count, term depth, term usage frequency (the number of annotations using the term), the ascending rank value of the highest matching result the term was found in, BLAST score (bits) and expectation value. Two-fold cross-validation was undertaken, i.e. the data was broken into 2 groups with models built on one group and tested on the other and vice-versa. Box's M test [18] for homogeneity of covariance measures was significant. Box's M test is prone to being over sensitive for large sample sizes (N = ~16000). Log determinants were low and all attributes had high tolerance scores, indicating that covariance assumptions were not violated. Stepwise analysis indicated that all attributes were significant. Models of both training sets were significant and had highly similar discriminant functions and structure matrices, and high cross-validation accuracies (78.9% and 79.3% respectively). Due to the similarity of both models, test and training sets were combined to create an average model. This had a post hoc accuracy of 79.1%, a true negative rate of 94% and a true positive rate of 60%. The canonical discriminant function coefficients were used to score potential term associations for query sequences. Essentially terms associated with the first five BLAST matching proteins were assigned a discriminant function score. As there were only two classes to distinguish between (either a correct or incorrect prediction) those potential term associations with greater than a cut-off score (that being the Mahalanobis distance midway between the correct and incorrect points) were assigned to the sequence.
Term covariance filter
The Covering Graph and Term Distance Concordance annotation scoring methods outlined above have no intrinsic capacity to assign terms to protein sequences. They simply assign a score to potential term assignments that it is hoped corresponds to an increased probability that that potential term assignment is correct. Approaches to automatically assigning predicted terms to sequences were examined that utilise the scores output by these methods. As outlined above a simple method was to assign a maximum of the highest scoring ten terms to a sequence. Ten was found to be the most accurate number to assign based on the harmonic mean of the precision and recall (data not shown). A statistical approach was also examined that used a chi-square based decision-making algorithm, that we called the Term Covariance Filter. All terms returned by a scoring method were broken into groups based on ontology. All possible combinations of five or less terms were created in descending order of the sum of the scores given to their composite terms. A combination was checked against the database of term annotations to find the observed number of instances that this combination of terms had been assigned to protein sequences by curators (i.e. all term annotations that did not have an ISS evidence code). If the observed number was greater than five, then the chi-square test statistic was calculated, which was defined as:
chi-square test statistic = (o-e)2/e
Where o is the observed number of instances for a term combination, and e is the expected number of instances of this term combination, calculated as the product of the proportion of the total number of annotations for each term. If the chi-square test statistic was greater than the critical value (3.84), then the term combination was accepted as valid. Otherwise the next combination was examined, until only combinations consisting of a single term remained. When that occurred the best scoring term was selected.
Analysis
Impact of distance on accuracy
The precision and recall measures for term assignments made by the random term assigner at different permitted distances between the predicted and correct term for a sequence were calculated. Table 1 illustrates that increasing the allowed distance between predicted and correct term results in an exponential increase in recall and precision measures, with recall more sensitive to this than precision. The relative impact on harmonic mean is the harmonic mean at a given distance divided by the harmonic mean at distance 0. The effect on the harmonic mean of the accuracy of increasing the allowed distance to 1 is to increase the accuracy value 3-fold. As a result of this all accuracy measures described in this paper use a permitted distance between predicted and correct term nodes of 0.
Table 1 Impact on accuracy estimates of varying allowed distance between predicted and curator-assigned terms.
Distance Relative Impact on Harmonic Mean
0 0.00
1 3.08
2 10.32
3 40.65
4 97.06
Accuracy of using BLAST results for term annotation
Figure 1 demonstrates the impact on accuracy measures of increasing the number of BLAST results used for term assignment for 4,710 UniProt query sequences. Essentially as the number of BLAST results is increased the recall increases and the precision decreases. The overall impact on the harmonic mean is a slight decrease. As the number of results used increases so does the average number of term associations per query sequence (data not shown).
Figure 1 Prediction accuracy based on terms associated with a given number of best matching BLAST results.
Accuracy of term assignment approaches
No combination of approaches using the Covering Graph method had a better overall harmonic mean than the Best BLAST method. Indeed the only case where any accuracy metric is higher than that of the Best BLAST method is the recall when term assignments are based on covering graph normalised concordance scores >0.1. However, this increase in recall, and respective drop in precision, was due to this method assigning approximately four times as many terms as the Best BLAST method.
The Term Distance Concordance method had a greater precision in all cases than the Covering Graph method. Where Term Covariance Filter selection was used, recall was higher than all Covering Graph annotation approaches except where assignment was made where terms had >0.1 normalised concordance. In the case of Term Distance Concordance annotations using the ten highest scoring terms, recall was higher than any Covering Graph approach. Term Covariance Filter selection increased the precision of the Term Distance Concordance method slightly while decreasing the recall by more than double this difference.
The Discriminant Function method had the best overall performance with a higher precision than all other methods including the Best BLAST method, and a recall comparable to the best recall of any other method shown. The training set was the UniProt data set used by all other methods. Compared to the other methods used, the Discriminant Function method assigned terms to less than half the total number of query sequences and assigned the least number of terms to those that it did assign terms to. This is because in many cases a query sequence had no potential term associations that had a discriminant function score greater than the cut-off. This results in a highly conservative pattern of term prediction.
Discussion
Many approaches to GO term prediction utilise BLAST in some way. This might involve using LocusLink entries returned by online BLAST output to identify existing GO term annotations [19]. It is also common practice for researchers to use enzyme commission (EC) numbers to search for GO terms in the Kyoto Encyclopedia of Genes and Genomes database [12]. Furthermore, several recently published accounts of annotators [6,9] utilise the GO database and BLAST to find matching sequences with existing GO term annotations.
Bearing in mind that there are a great many different ways of creating a sequence function predictor based on GO and BLAST, it becomes important to demonstrate that the new system is more accurate than those already in use. Perhaps the default system of choice for use by researchers is to simply select the best matching sequence returned by BLAST that also has a GO term annotation. The advantages of this approach are speed and simplicity, biological merit in assigning function based on sequence, and that it mirrors patterns of term assignment from other existing annotations. In order to demonstrate their usefulness new approaches to assigning GO terms based on BLAST output should be benchmarked against this default approach. Unfortunately as far as the authors are aware, no other published accounts of function annotators have compared their effectiveness against simply assigning terms associated with the best matching BLAST sequence.
Annotation systems routinely address accuracy issues in an indirect or incomplete manner often using small, handpicked samples and do not demonstrate accuracy relative to techniques commonly employed by biologists. A common feature of the ad hoc manner in which accuracy has been described for some annotation systems is to increase the permitted distance between predicted and correct terms allowed before a term assignment is declared incorrect. For instance, both GOFigure and GOEngine allowed a distance of 1 between predicted and correct terms. Table 1 shows the overall impact of increasing the permitted distance between predicted and correct terms on accuracy measures for a very bad annotation method (random term assignment). At a permitted distance of 1 the precision, recall and harmonic mean are around 3 times as high as an estimate based on simple term matching. As such, accuracy estimates that do not use exact matching must be viewed sceptically. All accuracy measures used here require exact matching between terms (i.e. distance permitted is zero).
We have evaluated the accuracy of the defacto standard approach of assigning GO terms to a novel sequence based on sequence similarity to another sequence, and used this to benchmark new approaches to GO term prediction. In doing this we found that using GO terms associated with the best matching BLAST sequence for function prediction is a very effective method in and of itself. This approach is more accurate in terms of precision and recall than most of the various methods implemented here (Table 2). Furthermore, by simply increasing the number of results used the recall can be increased but with a decrease in precision.
Table 2 Accuracies of BLAST-based automated GO term predictors
Method Precision Recall Harmonic Mean Mean Term Assignments N
Best BLAST 0.41 0.56 0.48 3.9 4710
Covering Graph
Ten highest scoring terms 0.20 0.48 0.28 7.7 4710
Terms with >0.1 normalised concordance 0.19 0.61 0.29 16.1 4710
Terms with >0.1 and <0.2 normalised concordance 0.09 0.10 0.09 2.7 4710
Terms with >0.1 and <0.9 normalised concordance 0.11 0.29 0.16 7.0 4710
Term Covariance Filter selection 0.24 0.35 0.28 3.4 4710
Term Distance Concordance
Ten highest scoring terms 0.33 0.63 0.43 6.3 4710
Term Covariance Filter selection 0.36 0.51 0.42 3.9 4710
Discriminant Function
Training set (post hoc) 0.70 0.59 0.64 2.4 2070
Test set (apriori) 0.61 0.51 0.55 2.0 10689
Precision and recall are inextricably associated by the error rate associated with a new term association. It is a well-known property of information retrieval systems that as the recall increases the precision will decrease. Term annotation methods will have a precision and recall based on the error rate associated with assigning each new term. The probability of mistakenly assigning a term when it is actually incorrect to do so (false positive rate) will determine the precision, while the probability of rejecting a correct term association (false negative rate) will determine the recall. The challenge of developing BLAST-based GO annotation methods then becomes that of constructing an approach that has a lower false negative and false positive rate than simply choosing terms associated with the best BLAST result.
The Covering Graph method was able to associate terms to sequences even though these terms were not directly associated with matching BLAST sequences by making the assumption that ancestral terms (i.e. parent nodes to terms in the GO directed acyclic graph) could be assigned a score based on children associated with BLAST result sequences. Unfortunately the precision of this approach is generally very low, indicating a high false negative rate. In some instances its recall is higher than that of the Best BLAST method but this is achieved by increasing the average number of term associations per query sequence, and as such, the precision is very low. Note that in the case where the Covering Graph method has its highest recall (0.61), the same recall could be obtained by simply selecting terms associated with the first two best matching BLAST results with a much higher precision (precision 0.36, recall 0.61).
GoFigure [9] utilises a minimum covering graph approach to BLAST based GO term annotation. The researchers state that a significant problem in the use of a minimum covering graph approach is that as the tree is traversed upwards ancestors accumulate a higher score such that the root node (closest common ancestor) will have the highest score in all cases. To counteract this effect they employ a maximal cut-off to scores where terms with greater than this value are not included. We examined the utility of using a maximal cut-off to see whether this improved the precision and recall of term assignments. The Covering Graph method term assignments using a cut-off of 0.2 normalised concordance had a far lower precision and recall than where a cut-off of 0.9 normalised concordance was used, and this had a lower precision and recall than where no cut-off was used (i.e. a cut-off of >= 1.0). The reason for this is that when assigning proteins manual curators tended to prefer terms higher in the GO term hierarchy, i.e. closer to the root node. Utilising a cut-off threshold means that these terms are more likely to be omitted.
The Term Distance Concordance method is a term annotation system based on finding a concordance among BLAST results for GO terms. This is achieved by constructing a matrix of term-by-term distances (the number of edges between two terms). The highest scoring terms will have the largest number of siblings, ancestors and descendants in the set of terms associated with the best 5 BLAST results. Each term instance is assigned a score based on its total relatedness to other terms and the BLAST result in which it was found. As such the system is based on a number of assumptions: terms associated with better matching sequences, that appear in multiple results and have a smaller total tree distance to other terms associated with BLAST results are more likely to be correct. On face value these assumptions appear to be fairly safe, however the overall accuracy of the Term Distance Concordance method is slightly below that of simply selecting terms associated with the best BLAST result. Selecting up to the best 10 ranked terms for this approach does yield a better recall, but with a slightly poorer precision, due to annotating more terms to a sequence on average.
The Term Covariance Filter was applied to both the Covering Graph method and the Term Distance Concordance method. The advantage of this approach is that it can be applied to any BLAST-based GO term annotation scoring system, assigning the largest permuted combination of terms for a single ontology that were used by manual curators. As might be expected the overall impact of this method of selecting terms for sequences is to increase the precision by exclusion of poor term combinations while decreasing recall. In the two cases that this is applied the precision increases and recall decreases relative to simply choosing the best 10 terms. Overall the harmonic mean remains the same. As such an approach like this may be useful in cases where precision needs to be increased at the expense of recall.
The Discriminant Function method essentially assigns potential term associations a score based on a linear model of correct versus incorrect term assignments. It is a simple and easily extendible approach to utilising BLAST data for term assignment. The accuracy of the approach as demonstrated on the UniProt dataset is well in excess of that produced by the selecting terms associated with the best BLAST result. However this data was used as the training set for model fitting and as such any accuracy measures will be overestimates of that found when applied post hoc. A better estimate of model accuracy is obtained by applying the Discriminant Function method to annotate terms for all non-UniProt sequences matched against UniProt sequences using BLAST. Simply utilising the Best BLAST method for this data led to accuracy measures less than the Discriminant Function method (recall 0.55, precision 0.52, harmonic mean 0.54). Furthermore, the Discriminant Function method is far more concise than other methods, selecting fewer terms for each sequence that any other method, e.g. the Discriminant Function method selected nearly half as many terms for sequences than did the Best BLAST approach. This fact may make it invaluable to curators as a first step in a wide scale annotation project.
Conclusion
There are a great many possible approaches to the design of systems of GO term assignment based on BLAST output, however, new approaches need to be adequately benchmarked to demonstrate their effectiveness. We have shown that an approach of simply selecting GO terms associated with the first returned BLAST matching sequence is a fairly accurate way of predicting functions for novel sequences. As such new approaches need to demonstrate that they are at least able to out perform this default approach. At the time of writing, most published accounts of functional annotators tend to provide overly generous accounts of their accuracies. To facilitate rapid developments in this area common benchmarking protocols would be useful.
We examined three approaches to GO term assignment based on BLAST output. The Covering Graph approach was able to infer associations between sequences and GO terms even though they were not directly associated with matching sequences. Unfortunately this approach has the fundamental problem where higher-level terms will always have higher scores, are more likely to be correct than lower terms, but the capacity to differentiate between correct and incorrect high level terms is lost. Maximal thresholds, where terms with a score greater than a given value are not used for annotations, are not a solution as they tend to decrease the accuracy of the system. The system preferring terms with closely related terms in higher ranked results, the Term Distance Concordance method, performed reasonably well, with accuracy measures comparable to selecting terms associated with the best BLAST result. However the approach is computationally intensive while the Best BLAST method is not. This system is probably not worth using in its current form.
A scoring system arising from discriminant analysis, the Discriminant Function method, had a higher accuracy then all methods shown here. In particular its precision was far higher than the Best BLAST method. Annotations were conservative and of high quality, and as such, this system may be of use to curators undertaking large annotation projects. This approach will be further investigated with the aim of producing a high accuracy automated functional annotation system.
In conclusion we found that accuracy benchmarking is an absolute requirement in appropriately assessing the suitability of the design of BLAST-based GO term annotators. Of the various approaches examined, a simple data-mining-oriented application is able to provide high quality, conservative functional predictions for novel sequences. To ensure that biological function annotators are of high quality, we recommend that developers utilise accuracy benchmarking and data mining techniques where possible. Future work will focus on using data mining techniques to incorporate a range of data sources to see if this increases the accuracy of function prediction.
Methods
Pseudocode for term prediction algorithms
A number of algorithms were developed to allow for term predictions based on BLAST-matches to protein sequences that had been assigned terms manually by curators. The more complex of these that could not be fully described in text are detailed below in pseudocode.
Covering graph method
For each sequence
SELECT best 5 matching sequences from results database For each term belonging to matching-sequences
Term concordance-score = descending rank of expect value +
number of times term found in BLAST output for this sequence.
End for
Construct list of unique terms from matching-sequence
Assign matching-terms to ontology
For each ontology
/*Covering graph construction*/
Find closest-common-ancestor-term
Find all paths from ancestor term to matching terms
Add all terms along these paths to analysis
/*Term-score assignment*/
For each matching-term
While matching-term has ancestors on path to closest-common-
ancestor-term
Assign matching-term concordance score to ancestor
End while
End for
End for
End for
Term distance concordance method
For each sequence
SELECT best 5 matching sequences from results database
For each term belonging to matching-sequences
Construct matrix to all other terms in the same ontology where
value is the distance between terms.
For each term in term-matrix
Concordance score = inverse log ((ascending expect value
rank) x (ontology tree depth/distance to matching-term)
End for
End for
End for
Where 'ontology tree depth' is the distance from the root to the leaves for a given ontology.
Term covariance filter
Assign terms to ontologies
For each ontology with number of terms > 0
If number of terms == 1
Assign this term to sequence
Else
Do
Construct array of permuted term-combinations where elements
are ordered in descending order of combination size and sum
of annotations, and where the maximum size of a term-
combination is 5.
observed = number of occurrences where term-combination [i]
terms were all assigned to a single sequence
While observed number < 5
Drop lowest scoring term based on assignment method
observed = number of occurrences where term-
combination [i] terms were all assigned to a single
sequence
End while
expected = ((n1/max)*(n2/max)... *(nn/max)) * max [where max
is the total number of term observations for that ontology]
chi-square test statistic = (observed-expected)2/expected
While test statistic < chi-square critical value AND terms > 1
Assign these terms to sequence
End If
End for
Authors' contributions
CEJ undertook initial study design, software implementation, statistical analysis and interpretation, and drafted the initial manuscript. UB and ALB participated in the final study design, coordinated the study and contributed to the final manuscript.
Acknowledgements
The researchers wish to thank the Australian Centre for Plant Functional Genomics and the Adelaide Computer Science Bioinformatics Group for support during this research project.
==== Refs
GenBank statistics
Ashburner M Ball CA Blake JA Botstein D Butler HJ Cherry M Davis AP Dolinski K Dwight SS Eppig JJ Harris MA Hill DP Issel-Tarver L Kasarskis A Lewis S Matese JC Richardson JE Ringwald M Rubin GM Sherlock G Gene Ontology: tool for the unification of biology Nature Genetics 2000 25 25 29 10802651 10.1038/75556
Altschul SF Madden TL Schaffer A Zhang J Zhang Z Miller W Lipman DJ Gapped BLAST and PSI-BLAST: a new generation of protein database search programs Nucleic Acids Research 1997 25 3389 3402 9254694 10.1093/nar/25.17.3389
NCBI Handbook
Statistics of sequence similarity scores
Hennig S Groth D Lehrach H Automated Gene Ontology annotation for anonymous sequence data Nucleic Acids Research 2003 31 3712 3715 12824400 10.1093/nar/gkg582
Groth D Lehrach H Hennig S GOblet: a platform for Gene Ontology annotation of anonymous sequence data Nucleic Acids Research 2004 32 W313 W317 15215401
Camon E Magrane M Barell D Lee V Dimmer E Maslen J Binns D Harte N Lopez R Apweiler R The Gene Ontology Annotation (GOA) Database: sharing knowledge in UniProt with Gene Ontology Nucleic Acid Research 2004 32 D262 D266 10.1093/nar/gkh021
Khan S Situ G Decker K Schmidt CJ GoFigure: Automated Gene Ontology annotation Bioinformatics 2003 19 2484 2485 14668239 10.1093/bioinformatics/btg338
Xie H Wasserman A Levine Z Novik A Grebinskiy V Shoshan A Mintz L Large-scale protein annotation through Gene Ontology Genome Research 2002 12 785 794 11997345 10.1101/gr.86902
Liu G Loraine AE Shigeta R Cline M Cheng J Valmeekam V Sun S Kulp D Siani-Rose MA NetAffx: Affymetrix probesets and annotations Nucleic Acids Research 2003 31 82 86 12519953 10.1093/nar/gkg121
McCarter JP Mitreva MD Martin J Dante M Wylie T Rao U Pape D Bowers Y Theising B Murphy CV Kloek AP Chiapelli B Clifton SW Bird DM Waterston RH Analysis and functional characterisation of transcripts from the nematode Meloidogyne incognita Genome Biology 2003 4 R26 12702207 10.1186/gb-2003-4-4-r26
Mungall CJ Misra S Berman BP Carlson J Frise E Harris N Marshall B Shu S Kaminker JS Prochnik SE Smith CD Smith E Tupy JL Wiel C Rubin GM Lewis SE An integrated computation pipeline and database to support whole-genome sequence annotation Genome Biology 2002 3 1 0081 10.1186/gb-2002-3-12-research0081
GODB: Database of curated sequences and functional terms
GO Evidence Codes
Apweiler R Bairoch A Wu CH Barker WC Boeckmann B Ferro S Gasteiger E Huang H Lopez R Magrane M Martin MJ Natale DA O'Donovan C Redaschi N Yeh LS UniProt: the Universal Protein Knowledgebase Nucleic Acids Research 2004 32 D115 D119 14681372 10.1093/nar/gkh131
Karaos U Murali TM Letovsky S Zheng Y Ding C Cantor CR Kasif S Whole-genome annotation by using evidence integration in functional-linkage networks Proceedings of the National Academy of Sciences 2004 101 2888 2893 10.1073/pnas.0307326101
Flury B Riedwyl H Multivariate statistics: a practical approach 1988 London: Chapman and Hall
Pruitt KD Maglott DR RefSeq and LocusLink: NCBI gene-centered resources Nucleic Acids Research 2001 29 137 140 11125071 10.1093/nar/29.1.137
|
16288652
|
PMC1298289
|
CC BY
|
2021-01-04 16:27:46
|
no
|
BMC Bioinformatics. 2005 Nov 15; 6:272
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-272
|
oa_comm
|
==== Front
BMC Evol BiolBMC Evolutionary Biology1471-2148BioMed Central London 1471-2148-5-601627114210.1186/1471-2148-5-60Research ArticleAvoid, attack or do both? Behavioral and physiological adaptations in natural enemies faced with novel hosts Vacher Corinne [email protected] Sam P [email protected] Michael E [email protected] Equipe Biologie des Populations en Interaction, Institut National de la Recherche Agronomique (UMR1112), 06903 Sophia-Antipolis Cedex, France2 Section of Integrative Biology, University of Texas at Austin, Austin TX 78712, USA3 Laboratoire Génétique et Environnement, Institut des Sciences de l'Evolution (UMR5554), Université Montpellier II, 34095 Montpellier Cedex 5, France2005 4 11 2005 5 60 60 4 2 2005 4 11 2005 Copyright © 2005 Vacher et al; licensee BioMed Central Ltd.2005Vacher 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
Confronted with well-defended, novel hosts, should an enemy invest in avoidance of these hosts (behavioral adaptation), neutralization of the defensive innovation (physiological adaptation) or both? Although simultaneous investment in both adaptations may first appear to be redundant, several empirical studies have suggested a reinforcement of physiological resistance to host defenses with additional avoidance behaviors. To explain this paradox, we develop a mathematical model describing the joint evolution of behavioral and physiological adaptations on the part of natural enemies to their host defenses. Our specific goals are (i) to derive the conditions that may favor the simultaneous investment in avoidance and physiological resistance and (ii) to study the factors that govern the relative investment in each adaptation mode.
Results
Our results show that (i) a simultaneous investment may be optimal if the fitness costs of the adaptive traits are accelerating and the probability of encountering defended hosts is low. When (i) holds, we find that (ii) the more that defended hosts are rare and/or spatially aggregated, the more behavioral adaptation is favored.
Conclusion
Despite their interference, physiological resistance to host defensive innovations and avoidance of these same defenses are two strategies in which it may be optimal for an enemy to invest in simultaneously. The relative allocation to each strategy greatly depends on host spatial structure. We discuss the implications of our findings for the management of invasive plant species and the management of pest resistance to new crop protectants or varieties.
==== Body
Background
In natural antagonistic systems such as host-parasite, plant-herbivore, and predator-prey systems (hereafter called 'host-enemy'), enemies may frequently be confronted with hosts expressing novel defenses. For instance, in agro-ecosystems, herbivorous insects are confronted with novel plant defenses each time a new chemical pesticide or each time a new toxic cultivar is introduced. This situation is frequent in natural areas too, when an herbivorous insect's foraging area is invaded by a novel toxic plant variety or species (e.g. invasion of Rocky mountain meadows by Thlaspi arvense, a crucifer lethal to Pieris larvae [1]). 'Avoid' and 'attack' are two basic strategies that enemies may evolve to cope with these novel defenses. In other words, enemies may evolve the ability to discriminate between defended and undefended hosts and preferentially avoid defended ones (behavioral adaptation), or develop a direct counter-adaptation allowing the successful attack of defended hosts (physiological adaptation). Both adaptation modes have been extensively studied and reviewed, especially in arthropod systems [2-5].
Although simultaneous investment in both adaptations may first appear to be redundant (since avoiding defended hosts is unnecessary if defenses have been overcome anyway), some empirical studies have suggested a reinforcement of physiological resistance to host defenses with additional avoidance behaviors. The most convincing example is perhaps the study of Pluthero & Threlkeld (1981) [2]. These authors measured the levels of behavioral avoidance and physiological resistance in eight strains of wild-caught Drosophila melanogaster tested for their responses to the insecticide malathion. They showed that the most resistant line had also the highest degree of avoidance, suggesting a reinforcement of physiological resistance with an additional avoidance behavior in a wild population. However they did not find any significant correlation between these two modes of insecticide resistance in the eight strains, which indicates that the mechanisms involved were genetically independent from each other. The independence between the two adaptation modes is also suggested by other studies that have demonstrated the evolution of either physiological resistance or behavioral avoidance, but not both. For instance, in Plutella xylostella, exposure to transgenic plants expressing toxins from Bacillus thuringiensis (Bt) induced physiological resistance to Bt toxins without discrimination between transgenic and non-transgenic plants [6], whereas exposure to toxic baits altered the behavior of German cockroaches, but not their physiological resistance [7].
To explain the paradox that apparently redundant adaptations may evolve, a number of authors [8-10] have employed population genetics models. These models assume that physiological and behavioral responses are governed by two independent loci each bearing two co-dominant alleles. Nine strategy sets (i.e. nine genotypes) corresponding to the combination of three physiological adaptation levels (high, mild or null) and three behavioral adaptation levels (high, mild or null) are assumed. The results highlight the impact of population-genetic and population-dynamic factors on behavioral and physiological adaptations. In particular, a mixed strategy may be stable depending on the relative fitnesses of the nine genotypes and the initial allelic frequencies [8,9] and the mode of population regulation can have a striking impact on the likelihood of behavioral adaptation to evolve [10]. Simulations by Rausher [10] reveal that under the "hard selection" mode (i.e. regulatory factors act on the population as a whole [11]) the pure behavioral strategy evolves almost as frequently as the pure physiological strategy, whereas under "soft selection" (i.e. the subpopulations of each host are regulated independently [11]) the pure behavioral strategy never evolves. A potential explanation for this difference is that contrary to "hard selection", "soft selection" leads to overcrowding on the most suitable host and undercrowding on the other host [10]. Behavioral generalists therefore have a higher fitness than choosy enemies, because the former suffer lower levels of intra-specific competition. A less frequent outcome (c. 10% of the runs) is the evolution of a mixed strategy (i.e. a strategy where one of the loci at least is polymorphic). Under "hard selection", this outcome is favored by the absence of fitness costs to both traits [10].
The objective of this study is to (i) explore in more detail the conditions favoring the occurrence of a mixed strategy and (ii) study the factors that govern the relative investment into each adaptation mode. Specifically, we study the impact of the shape of the fitness cost functions and the impact of spatial heterogeneity in host defenses. Indeed, while several theoretical models have investigated the impact of the abundance and spatial distribution of suitable hosts on physiological resistance evolution [e.g. [12-17]] or on the evolution of host selection behavior [e.g. [18-21]], the impact of space on the joint evolution of physiological and behavioral adaptations to host defenses has been largely ignored [5]. In our model, physiological resistance and behavioral avoidance are represented as quantitative traits and we adopt an optimality approach to identify the conditions favoring the investments in both traits. Contrary to the population genetics models listed above, this approach permits quantitative predictions of the optimal relative investments in both forms of adaptation. Fitness costs of the adaptations and host spatial structure are assumed to be constant.
Below we show that (i) simultaneous investment may be optimal if the fitness costs of the adaptive traits are accelerating and the probability of encountering defended hosts is low. When (i) holds, we find that (ii) the more that defended hosts are rare and/or spatially aggregated, the more behavioral adaptation is favored.
Results
What conditions favor the simultaneous investment in physiological and behavioral adaptations to host defenses?
Accelerating fitness costs of adaptations
We find that the fitness costs of physiological and behavioral adaptations to novel host defenses can have an important effect on the optimal enemy strategy (Fig. 1). In the case of linear and decelerating maintenance costs of adaptations (Fig. 1B and 1C), simultaneous investment in physiological and behavioral adaptation to host defenses is never selected, whatever the spatial configuration of hosts. Either a pure physiological adaptation, a pure behavioral adaptation, or a "no investment" strategy is selected depending on initial investment levels. In the case of accelerating maintenance costs of adaptations (Fig. 1A), we find that a mixed strategy can be optimal.
Figure 1 Null clines for physiological (solid line) and behavioral (dotted line) adaptation to host defenses as a function of the shape of the cost function. Black points are stable steady states. Arrows represent schematic phase trajectories. kR = 0.1, kB = 0.1, e = 0.1, a = 0.4, f = 20%.
Rare and/or spatially aggregated defended hosts
When considering this latter case in more detail, interestingly, our results show that the spatial configuration of hosts with and without defenses has a strong impact on the occurrence of the mixed strategy (Fig. 2). A mixed investment is only optimal when the frequency of defended hosts in the environment is low or when their aggregation level is high (Fig. 2, white plane). It is also noteworthy that under the canonical set of parameters, the absolute investment in physiological and behavioral adaptations can be very different (Fig. 2B and 2C). In the following, we focus on the factors governing these differences in investment.
Figure 2 Co-equilibrium (R*, B*) between physiological (solid line) and behavioral (dotted line) adaptation to host defenses as a function of the frequency f and the spatial aggregation level a of well-defended hosts. Equilibrium is polymorphic in the white plane and monomorphic in the gray plane ((R*, B*) = (1,0) or (0,1)). kR = 0.1, kB = 0.1, e = 0.1, x = 2
What factors govern the relative investment in each adaptation mode?
Spatial configuration of hosts
As Fig. 2 illustrates, investment in each adaptation mode depends on the spatial configuration of hosts. For a given abundance of defended hosts, the absolute level of physiological adaptation is maximal for lower levels of host spatial aggregation than is the absolute level of behavioral adaptation (Fig. 2B: a = 0 vs. a = 0.4; Fig. 2C: a = 0.45 vs. a = 0.7). In contrast, looking at the relative investments in each adaptation mode gives a different picture (Fig. 3): a major result of this study is that the relative allocation to behavioral adaptation is maximal when defended hosts are rare and spatially aggregated (Fig. 3B). Under these conditions, we find that the total level of investment approaches zero, because defended hosts are encountered increasingly rarely (Fig. 3A).
Figure 3 Total investment in adaptation to host defenses (R*+B*) and relative allocation to behavioral adaptation B*/(R*+B*) as a function of the frequency f and the spatial aggregation level a of defended hosts in the case of a polymorphic equilibrium (R*, B*). Model parameters are the same than in Figure 2.
Magnitude of search costs
Finally, we investigate the effects of the costs associated with the active search of undefended hosts on the differences in investment in each adaptation mode (Fig. 4). Partial derivatives of enemy fitness with respect to the level of physiological adaptation and the level of behavioral adaptation are symmetric when search costs are zero (see Methods, Eq 7a and 7b for e = 0). Therefore, under this assumption, optimal investments in each resistance mode are equal (Fig. 4A). Interestingly, the effects of search costs are not uniform across the range of host spatial aggregation levels. Search costs strongly favor physiological over behavioral resistance when defended hosts are randomly distributed, but have a low impact on the relative allocation to each resistance mode when defended hosts are spatially aggregated (Fig. 4B and 4C).
Figure 4 Co-equilibrium (R*, B*) between physiological (solid line) and behavioral (dotted line) adaptation to host defenses as a function of the search cost coefficient e and the spatial aggregation level a of well-defended hosts. Equilibrium is polymorphic. kR = 0.1, kB = 0.1, x = 2, f = 20%
Discussion
In agreement with previous studies [8-10], we find that the simultaneous investment in avoidance and physiological resistance can be an optimal strategy despite interference between both adaptations (i.e., investment in one adaptation mode decreases the efficiency of investment in the other). We identify two conditions that must be fulfilled: maintenance costs of both adaptations must be accelerating (Fig. 1) and the probability of encountering defended hosts must be low (Fig. 2). Under all other conditions, pure strategies are favored. This parallels the results of Poitrineau and colleagues on host investment in defenses against multiple enemies in the case where defenses interfere with each other [22].
Our results show that in the case where a pure strategy is optimal, both the pure physiological resistance strategy and the pure behavioral avoidance strategy may evolve depending on initial investment levels (Fig. 1). This result is in agreement with previous population genetics models that explicitly assume "hard selection" [10], and could be due to our implicit assumption of "hard selection". Indeed, since there is no explicit function of population regulation in our model and because the fitness functions are not dependent on the number or strategy of local competitors, any regulation must occur on the global level.
Moreover, when the mixed investment is optimal, we found that the absolute and relative investments in each adaptation mode are sensitive to the spatial configuration of hosts (Fig. 2). When defended hosts are abundant and/or randomly distributed (i.e., when the probability of encountering defended hosts is high), it is optimal for the enemy to invest mainly in physiological resistance. We have shown that the low investment in behavioral adaptation is due to the costs of actively searching undefended hosts (Fig. 4). As the probability of encountering defended hosts decreases, the absolute investment in behavioral adaptation increases and goes through a maximum. Thereafter, optimal investments in both adaptation modes decrease. In the extreme case, total investment approaches zero because defended hosts are almost never encountered. Looking at the relative investments in each adaptation mode gives a different picture (Fig. 3). Relative investment in physiological resistance is always greater, but the difference tends to disappear with increasing rarity and/or spatial aggregation of defended hosts.
We describe the consequences of interference between two adaptive traits on the joint evolution of these traits in the first part of the Discussion below. In the second and third parts, we highlight the relevance of these findings to plant-herbivore interactions in natural and managed ecosystems.
Simultaneous investment in avoidance of and resistance to host defenses: a paradox?
Interference between physiological and behavioral adaptations is an emergent property of our model, clearly identifiable from the net fitness benefits of investment in both adaptation modes. Simplifying the enemy fitness W for null fitness costs of adaptation (see Methods, Eq. 1a for k = e = 0) gives
W = W0+dP(1+T-I)
where T = R+B is the total investment in physiological and behavioral adaptations (respectively R and B) and I = BR is the interference between physiological and behavioral adaptations. Thus, for a given total investment T, the net benefits are maximal when all resources are invested in one mode of adaptation only (i.e., interference I is zero).
However, the optimality of a strategy also depends on the fitness costs associated with the adaptive traits. Consider an enemy investing an intermediate amount of resources in one adaptation mode and facing an increase in the frequency of encounters with defended hosts. When fitness costs are decelerating, investing in this same mode of adaptation is not only the most efficient but also the least costly way to reinforce adaptation to host defenses. Thus, continuing to invest in the prevailing adaptation mode is better than developing another from zero (i.e., pure resistance strategies are favored). In contrast, for accelerating costs, investing in the prevailing mode of adaptation is more efficient but also more costly. Thus, investing in another adaptation mode can be an optimal choice (i.e., mixed resistance strategies can be optimal). Mixed resistance strategies tend to disappear when the frequency of encounters with defended hosts increases, because investing simultaneously in two interfering adaptive traits is increasingly wasteful. In the extreme case (R•1, B•1), one half of the investment is useless because of interference.
Avoidance of aggregated novel hosts: a factor in biological invasions?
Our findings are relevant to adaptation in ecological communities, for example when a habitat is invaded by a plant variety or species that is toxic to a resident herbivore. One commonly observed life history trait in invasive plant species is clonal reproduction [23]. This reproductive mode leads to the spatial aggregation of invaders. Thus, during the initial steps of a biological invasion, clonally invasive plants are rare and spatially aggregated. Our model suggests that natural enemies should invest a low amount of resources into adaptations to these novel hosts, and allocate non-trivial amount of these resources to behavioral adaptation. Consequently, we suggest that selection for avoidance of the toxic compounds produced by rare, clumped invasive plants could be a cause for the so-called "ecological release" experienced by these plants. The enemy release hypothesis states that plant species, on introduction to an exotic region, experience a decrease in regulation by herbivores (in particular, specialist herbivores) and other natural enemies, resulting in a rapid increase in distribution and abundance [24,25]. Comparisons of the parasitic load and the number of pathogens in native versus introduced regions support this hypothesis [26,27], as well as comparison of the plant anti-herbivore compounds [25]. One approach to test our hypothesis would be to compare herbivore behavior in native and introduced ranges of invasive plants.
Pest management: how to limit physiological resistance to new crop protectants or varieties
Finally, our results are relevant to certain forms of pest management, where one attempts to conserve the efficiency of a new toxic cultivar or a new chemical pesticide. Although models of pest resistance evolution to chemical pesticides or genetically-engineered toxins have long been acknowledged as a tool for pest management [28], host preferences have rarely been incorporated in theoretical developments (but see [10,29,30]). Our model suggests that using rare and aggregated treated/toxic plants during the first years of commercialization may curtail a pest's investment in physiological resistance, and favor the evolution of avoidance of treated/toxic areas. By reducing the frequency of encounters with the new pesticide/toxin, this initial step of behavioral adaptation might delay the evolution of physiological resistance if the treated/toxic plants are subsequently used more extensively (see also [10]). Refuges (i.e., non-treated/toxic host plants maintained in close proximity to treated/toxic crops to delay physiological resistance evolution [31]) would then serve as insect traps. This potential role of refuges has rarely been studied (but see [30]), since in population genetics models refuges are considered to be a source of susceptible insects. Moreover, it is noteworthy that since the commercialization of insect-resistant GM crops, the optimal spatial distribution of refuges for sustainable pest control has received much attention [11-13], but their optimal temporal distribution has rarely been investigated (but see [32]). Based on our findings, we suggest that more research should be conducted to define this optimal temporal distribution of refuges, when taking into account the evolution of pest specialization.
Conclusion
The originality of our study is to have linked together physiology, behavior and landscape structure into a general model describing the adaptation of natural enemies to their hosts. Our model predicts that the optimal strategy for a natural enemy when confronted with well-defended, novel hosts subtly depends on the fitness costs of the adaptations to host defenses and the spatial distribution of defended hosts. Interestingly, under certain conditions (i.e., maintenance costs of the adaptations are accelerating and the probability of encountering defended hosts is low), a reinforcement of physiological resistance to host defenses by avoidance of these same defenses may be optimal. In this latter case, investment in physiological resistance is favored when the novel hosts are abundant at regional scales because the active search of undefended hosts is costly. It is also favored when the host type that might be encountered is difficult to predict by the enemy because the host from which the enemy emerges cannot be used as a cue (i.e., the aggregation level of host types is low).
Although they remain to be confirmed by empirical data, our theoretical results could have important implications for the management of invasive plant species and the management of pest resistance to new crop protectants or varieties. A logical next step in the model analysis would be to enable the host level evolve (see [33]). Indeed, since the host level is currently assumed to be non-evolving, our predictions are only relevant to enemy evolution over short time scales (e.g. evolution of herbivorous insects during the initial steps of invasion of their foraging range by new plants) or to systems in which host levels can be managed (e.g. agricultural systems). Relaxing this assumption may be useful to analyze adaptive patterns in coevolving systems having contrasting spatial structures, such as plant-herbivore interactions in tropical and temperate forests [33-36].
Methods
The model considers a single species of natural enemy confronted with two host types: initial hosts, called 'undefended hosts', and novel hosts, called 'defended hosts'. A simple life-cycle for the enemy that should apply to a range of natural antagonistic systems is assumed: enemies leave their host of emergence (Fig. 5, step 1), engage in a foraging behavior (Fig. 5, step 2) and finally attack a single host individual (Fig. 5, step 3). The host of emergence belongs to the initial host type whereas the selected host may belong to either host type. During the foraging step (Fig. 5, step 2), enemies either move at random over their environment or engage in the active search of undefended hosts. Enemies can therefore adopt three types of offensive behavior (Fig. 5, step 3):
Figure 5 Enemy life cycle. Parameters W0, H and S are defined in Table 1.
1. attack of an undefended host randomly encountered in the foraging area
2. attack of a defended host randomly encountered in the foraging area
3. attack of an undefended host encountered by active search in the foraging area.
Physiological and behavioral adaptations to host defenses are assumed to be quantitative traits that respectively decrease the deleterious effects of host defenses during a type (2) attack and increase the frequency of type (3) attacks.
The enemy's fitness W is stated as
W = WG*(1-CR)*(1-CB) (1a)
where WG is a function of R and B representing the average fitness gain resulting from host attack and CR and CB are increasing functions of R and B representing the fitness costs of adaptations. The average fitness gain resulting from host attack WG is stated as
WG = (1-B)(1-P)W0+ (1-B)P(W0-H) +B(W0-S) (1b)
where the three terms respectively reflect fitness gains resulting from attacks of type (1), (2) and (3). Each term is detailed below.
The first term corresponds to the attack of an undefended host encountered by a random move in the foraging area. The fitness gain resulting from this type of attack equals W0, which is the maximal potential fitness gain resulting from a host attack. The probability of this event is (1-B)(1-P), where B is the probability for an enemy to engage in the active search of undefended hosts during the foraging step and P is the probability for an enemy to randomly encounter a defended host in its foraging area (Fig. 5). B corresponds to the investment in behavioral adaptation whereas P depends on the spatial configuration of defended and undefended hosts. This latter is described by the frequency f of defended hosts and their level a of spatial aggregation at the scale of an enemy's foraging range, which is the scale of spatial aggregation the most relevant to our study [37-39]. Host aggregation at the scale of the enemy's foraging range is described by the average frequencies of the three possible host-pair types (i.e. defended-defended, undefended-defended and undefended-undefended), when considering only host-pairs between which distance is inferior to the maximal foraging distance [40]. The higher the frequency of homologous pairs, the greater the spatial aggregation. Making a parallel with Wright's inbreeding coefficient in population genetics [41], we define the aggregation level a of defended hosts at the scale of the enemy's foraging range as the deficit in heterologous host pairs. Defended-defended, undefended-defended and undefended-undefended host-pair average frequencies, fdd, fud and fuu respectively, are defined as
fdd = f2+af(1-f) (2a)
fud = 2f(1-f)(1-a) (2b)
fuu = (1-f)2+af(1-f). (2c)
The mean probability P that an enemy emerging from an undefended host encounters a defended host after a random move in its foraging area is
P = 1/2 fud/(fuu + 1/2 fud) (3a)
which simplifies to
P = f(1-a). (3b)
The second term of equation (1b) corresponds to the attack of a defended host encountered by a random move in the foraging area. The probability of this event is (1-B)P. The fitness gain W0 is decreased by H, H being the fitness loss due to the deleterious effects of host defenses (Fig. 5). H is assumed to decrease with the level R of physiological enemy resistance and increase with the level d of host defense. We chose the simplest function to describe H [42]. Hence
H = d(1-R). (4)
The third term of equation (1b) corresponds to the attack of an undefended host encountered by active search. The probability of this event is B. The fitness gain W0 is decreased by S, S being the fitness loss due to active searching [43,44] (Fig. 5). S is assumed to increase with the probability P of encountering a defended host by any random move. We also chose the simplest function to describe S. Hence
S = eP (5)
where e is the search cost coefficient.
Let us now describe the constitutive fitness costs of adaptations, CR and CB (Eq. 1a). Evidence of fitness costs of adaptation to host defenses is scarce [45] and a fortiori, the shape of the fitness cost functions (i.e., variations in the cost magnitude with the level of investment in the adaptive trait) is largely unknown [22,45]. However, it is reasonable to assume that physiological and behavioral adaptations to host defenses have constitutive fitness costs: a few studies show that physiological resistance to toxic compounds results from permanent metabolic changes that can reduce fitness [47-49] and obviously, discrimination between defended and undefended hosts involves energy allocation to sensors and neural cells allowing the detection and treatment of signals. Consequently, we chose simple functions to describe the constitutive fitness costs of physiological and behavioral adaptations to host defenses and we assessed the robustness of model predictions to the shape of these functions. The cost functions are taken as
CR = k Rx (6a)
CB = k Bx (6b)
where k is the cost coefficient and x controls the form of the function. If x > 1 then the cost is accelerating, if x = 1 then it is linear, whereas if x < 1 then it is decelerating.
Finally, note that we assume that the two adaptive traits are independent [8-10,50]. Partial derivatives of enemy fitness W (Eq. 1a) with respect to the level of physiological adaptation R and the level of behavioral adaptation B give
∂W/∂R = (1-CB)((1-B)dP(xCR+R(1-CR-xCR))-xCRW0)/R + ePB(1-CB)xCR/R (7a)
∂W/∂B = (1-CR)((1-R)dP(xCB+B(1-CB-xCB))-xCBW0)/B - ePB(1-CR)(1-(1+x)CB)/B (7b)
The optimal strategy (R*, B*) is assessed based on the position of the null clines ∂W/∂R = 0 and ∂W/∂B = 0 [51]. All the analyses were done with Mathematica 4 [52]. All the model parameters are summarized in Table 1.
Table 1 Model parameters, their definitions, range of values employed, and notes on their use.
Parameter Definition Range Comments
Host
d Defense level of defended hosts Held at 1
f Frequency of defended hosts in the environment 0–1
a Spatial aggregation level of defended hosts 0–1 a = 0 when defended and undefended hosts are randomly distributed and a→1 when defended and undefended hosts form two distinct patches.
Enemy
P Probability of encountering a defended host during a random move in the foraging area 0–1 Increases with the frequency f of defended hosts and decreases with their aggregation level a P = f(1-a)
W0 Maximal potential fitness gain when attacking an host Held at 1 Corresponds to the fitness gain when attacking a randomly encountered undefended host
H Fitness loss due to host defense 0–1 Decreases with the level of physiological resistance R of the enemy and increases with the level of defense d of the host H = d(1-R)
S Fitness loss due to active searching of undefended hosts 0–1 Increases with the difficulty in finding undefended hosts, i.e., with the probability P of encountering a defended host during a random move S = eP
e Search cost coefficient 0–1
R Physiological adaptation level 0–1 Physiological adaptation reduces the fitness loss H when attacking a defended victim.
CR Physiological adaptation maintenance cost 0–1 Increases with the level of physiological adaptation R CR = kRx
B Behavioral adaptation level 0–1 Behavioral adaptation corresponds to the probability of engaging in the active search of undefended hosts.
CB Behavioral adaptation maintenance cost 0–1 Increases with the level of behavioral adaptation B CB = kBx
k Maintenance cost coefficient 0–1
x Shape coefficient of the maintenance cost functions Held to 1/2, 1 or 2 Maintenance costs increase with the level of adaptation in an accelerating (x = 2), linear (x = 1) or decelerating (x = 1/2) way
Authors' contributions
CV and SPB conceived the study and designed the model. CV performed the model analysis and wrote the paper. MEH coordinated the study and obtained funding to finance the research. All authors read and commented on drafts of the manuscript, and approved the final manuscript.
Acknowledgements
We thank Anthony R. Ives, Laurent Lapchin and Thomas Guillemaud for helpful comments. We acknowledge financial support from the French Ministry of Research and the CNRS (ACI "Impact of Biotechnologies on Agro-Ecosystem Functioning"). Sam P. Brown was supported by a Marie Curie Fellowship. Corinne Vacher is supported by an INRA postdoctoral fellowship.
==== Refs
Chew FS Coevolution of pierid butterflies and their cruciferous foodplants. II. The distribution of eggs on potential foodplants Evolution 1977 31 568 579
Pluthero FG Threlkeld SFH Genetic differences in malathion avoidance and resistance in Drosophila J Econ Entomol 1981 74 736 740 6799555
Pluthero FG Sing RS Threlkeld SFH The behavioural and physiological components of malathion resistance in Drosophila melanogaster Can J Genet Cytol 1982 24 807 815
Lockwood JA Sparks TC Story RN Evolution of insect resistance to insecticides: a reevaluation of the roles of physiology and behavior Bull Entomol Soc Am 1984 41 51
Hoy CW Head GP Hall FR Spatial heterogeneity and insect adaptation to toxins Ann Rev Entomol 1998 43 571 94 15012398 10.1146/annurev.ento.43.1.571
Ramachandran S Buntin GD Tabashnik BE Raymer PL Adang MJ Pulliam DA Stewart CN Survival, development, and oviposition of resistant diamondback moth (Lepidoptera: Plutellidae) on trangenic canola producing a Bacillus thuringiensis toxin J Econ Entomol 1998 91 1239 1244
Negus TF Ross MH The response of German cockroaches to toxic baits: strin differences and the effects of selection pressure Entomol Exp Appl 1997 82 247 253 10.1023/A:1002978101698
Gould F Role of behaviour in the evolution of insect adaptation to insecticides and resistant host plants Bull Entomol Soc Am 1984 30 34 41
Castillo-Chavez C Levin SA Gould F Physiological and behavioral adaptation to varying environnements: a mathematical model Evolution 1988 42 986 994
Rausher MD Kim KC, McPheron BA The evolution of habitat preference: avoidance and adaptation Evolution of insect pests: patterns of variation 1994 259 283
Wallace B Topics in Population Genetics 1968 New-York: W. W. Norton
Peck SL Ellner SP The effect of economic thresholds and life-history parameters on the evolution of pesticide resistance in a regional setting Am Nat 1996 149 42 63
Lenormand T Raymond M Resistance management: the stable zone strategy Proc R Soc Lond B Biol Sci 1998 265 1985 1990 10.1098/rspb.1998.0529
Peck SL Gould F Ellner SP Spread of resistance in spatially extended regions of trangenic cotton: implications for management of Heliothis virescens (Lepidoptera: Noctuidae) J Econ Entomol 1999 92 1 16
Caprio MA Source-sink dynamics between transgenic and non-transgenic habitats and their role in the evolution of resistance J Econ Entomol 2001 94 698 705 11425026
Vacher C Bourguet D Rousset F Chevillon C Hochberg ME Modelling the spatial configuration of refuges for a sustainable control of pests: a case study of Bt cotton J Evol Biol 2003 16 378 387 14635838 10.1046/j.1420-9101.2003.00553.x
Cerda H Wright DJ Modeling the spatial and temporal location of refugia to manage resistance in Bt transgenic crops Agric Ecosyst Environ 2004 102 163 174 10.1016/j.agee.2003.08.004
Diehl SR Bush GL Otte D, Endler JA The role of habitat preference in adaptation and speciation Speciation and its consequences 1989 Sunderland, MA: Sinauer 345 355
Cunningham JP West SA Host selection in phytophageous insects: a new explanation for learning in adults Oikos 2001 95 537 543 10.1034/j.1600-0706.2001.950319.x
Onstad DW Spencer JL Guse CA Levine E Isard SA Modeling evolution of behavioral resistance by an insect to crop rotation Entomol Exp Appl 2001 100 195 201 10.1023/A:1019289030248
West SA Cunningham JP A general model for host plant selection in phytophagous insects J Theor Biol 2002 214 499 513 11846605 10.1006/jtbi.2001.2475
Poitrineau K Brown SP Hochberg ME Defence against multiple enemies J Evol Biol 2003 16 1319 1327 14640423 10.1046/j.1420-9101.2003.00585.x
Sakai AK Allendorf FW Holt JS Lodge DM Molofsky J With KA Baughman S Cabin RJ Cohen JE Ellstrand NC McCauley DE O'Neil P Parker IM Thompson JN Weller SG The population biology of invasive species Annu Rev Ecol Syst 2001 32 305 332 10.1146/annurev.ecolsys.32.081501.114037
Keane RM Crawley MJ Exotic plant invasions and the enemy release hypothesis TREE 2002 17 164 169
Joshi J Vrieling K The enemy release and EICA hypothesis revisited: incorporating fundamental difference between specialist and generalist herbivores Ecol Lett 2005 8 704 714 10.1111/j.1461-0248.2005.00769.x
Mitchell CE Power AG Release of invasive plants from fungal and viral pathogens Nature 2003 421 625 627 12571594 10.1038/nature01317
Torchin ME Lafferty KD Dobson AP McKenzie VJ Kuris AM Introduced species and their missing parasites Nature 2003 421 628 630 12571595 10.1038/nature01346
Tabashnik B Computer simulations as a tool for pesticide resistance management Pesticide Resistance: Strategies and Tactics for Management 1986 Washington, USA: National Academy Press 194 206
Mallet J Porter P Preventing insect adaptation to insect-resistant crops: are seed mixtures or refugia the best strategy? Proc R Soc Lond B Biol Sci 1992 250 165 169
Alstad DN Andow DA Managing the evolution of insect resistance to transgenic plants Science 1995 268 1894 1896
Gould F Sustainability of transgenic insecticidal cultivars. Integrating pest genetics and ecology Annu Rev Entomol 1998 43 701 26 15012402 10.1146/annurev.ento.43.1.701
Livingston MJ Carlson GA Fackler PL Managing resistance evolution in two pests to two toxins with refugia Am J Agric Econ 2004 86 1 13 10.1111/j.0092-5853.2004.00558.x
Levin DA Alkaloid-bearing plants: an ecogeographic perspective Am Nat 1976 110 261 284 10.1086/283063
Coley PD Barone JA Herbivory and plant defenses in tropical forests Annu Rev Ecol Syst 1996 27 305 335 10.1146/annurev.ecolsys.27.1.305
Gaston KJ Global patterns in biodiversity Nature 2000 405 220 227 10821282 10.1038/35012228
Condit R Ashtton PS Baker P Bunyavejchewin S Gunatilleke S Gunatilleke N Hubbel SP Foster RB Itoh A LaFrankie JV Lee HS Losos E Manokaran N Sukumar R Yamakura T Spatial patterns in the distribution of tropical tree species Science 2000 288 1414 1418 10827950 10.1126/science.288.5470.1414
Wiens JA Spatial scaling in ecology Funct Ecol 1989 3 385 397
Lima SL Zollner PA Towards a behavioral ecology of ecological landscapes TREE 1996 11 131 135
Kawata M Agawa H Perceptual scales of spatial heterogeneity of periphyton for freshwater snails Ecol Lett 1999 2 210 214 10.1046/j.1461-0248.1999.00071.x
Law R Dieckmann U Dieckmann U, Law R, Metz JAJ Moment approximations of individual-based models The geometry of ecological interactions: simplifying spatial complexity 2000 Cambridge: Cambridge University Press
Hartl DL Clark AG Principles of Population Genetics 1989 2 Sunderland, MA: Sinauer Associates
Kawecki TJ Red queen meets Santa Rosalia: arms races and the evolution of host specialization in organisms with parasitic lifestyles Am Nat 1998 152 635 651 10.1086/286195
Lucas JR Time constraints and diet choice: different predictions from different constraints Am Nat 1985 126 680 705 10.1086/284447
Crawley MJ Krebs JR Crawley MJ Foraging theory Natural enemies: the population biology of predators, parasites and diseases 1992 Oxford: Blackwell Scientific Publications 90 114
Coustau C Chevillon C Ffrench-Constant R Resistance to xenobiotics and parasites: can we count the cost? TREE 2000 15 378 383 10931680
Boots M Haraguchi Y The evolution of costly resistance in host-parasite systems Am Nat 1999 153 359 370 10.1086/303181
McKenzie JA Purvis A Chromosomal localization of fitness modifiers of diazon resistance genotypes of Lucilia cuprina Heredity 1984 53 625 634
Groeters FR Tabashnik BE Finson N Johnson MW Fitness costs of resistance to Bacillus thuringiensis in the diamondback moth (Plutella xylostella) Evolution 1994 48 197 201
Raymond M Chevillon C Guillemaud T Lenormand T Pasteur N An overview of the evolution of overproduced esterases in the mosquito Culex pipiens Philos Trans R Soc Lond B Biol Sci 1998 353 1707 1711 10021771 10.1098/rstb.1998.0322
Hochberg ME Hide or Fight? The competitive evolution of concealment and encapsulation in parasitoid-host associations Oikos 1997 80 342 352
Murray JD Murray JD Continuous models for interacting populations Mathematical Biology, second, corrected edition 1993 Berlin/Heidelberg: Springer-Verlag
Wolfram S Mathematica Version 4010 1999 Champaign, Ill: Wolfram Research
|
16271142
|
PMC1298290
|
CC BY
|
2021-01-04 16:29:17
|
no
|
BMC Evol Biol. 2005 Nov 4; 5:60
|
utf-8
|
BMC Evol Biol
| 2,005 |
10.1186/1471-2148-5-60
|
oa_comm
|
==== Front
BMC Evol BiolBMC Evolutionary Biology1471-2148BioMed Central London 1471-2148-5-611627448810.1186/1471-2148-5-61Research ArticleGenetic covariance between indices of body condition and immunocompetence in a passerine bird Gleeson Deborah J [email protected] Mark W [email protected] Ian PF [email protected] School of Integrative Biology, University of Queensland, St Lucia, Brisbane Queensland 4072, Australia2 Division of Biology and NERC Centre for Population Biology, Imperial College London, Silwood Park, Ascot, Berkshire SL5 7PY, UK2005 7 11 2005 5 61 61 9 2 2005 7 11 2005 Copyright © 2005 Gleeson et al; licensee BioMed Central Ltd.2005Gleeson 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
Condition-dependence is a ubiquitous feature of animal life histories and has important implications for both natural and sexual selection. Mate choice, for instance, is typically based on condition-dependent signals. Theory predicts that one reason why condition-dependent signals may be special is that they allow females to scan for genes that confer high parasite resistance. Such explanations require a genetic link between immunocompetence and body condition, but existing evidence is limited to phenotypic associations. It remains unknown, therefore, whether females selecting males with good body condition simply obtain a healthy mate, or if they acquire genes for their offspring that confer high immunocompetence.
Results
Here we use a cross-foster experimental design to partition the phenotypic covariance in indices of body condition and immunocompetence into genetic, maternal and environmental effects in a passerine bird, the zebra finch Taeniopygia guttata. We show that there is significant positive additive genetic covariance between an index of body condition and an index of cell-mediated immune response. In this case, genetic variance in the index of immune response explained 56% of the additive genetic variance in the index of body condition.
Conclusion
Our results suggest that, in the context of sexual selection, females that assess males on the basis of condition-dependent signals may gain genes that confer high immunocompetence for their offspring. More generally, a genetic correlation between indices of body condition and imuunocompetence supports the hypothesis that parasite resistance may be an important target of natural selection. Additional work is now required to test whether genetic covariance exists among other aspects of both condition and immunocompetence.
==== Body
Background
Body condition is central to animal life histories because the expression of many traits critical to survival and reproductive success is condition-dependent [1,2]. Condition-dependence is, therefore, a topic of broad interest in both natural and sexual selection. One particularly striking example of the fundamental role of condition-dependence is in the context of mate choice [3-5]. Females often choose among males on the basis of condition-dependent signals, which honestly advertise male quality as the expression of these signals may trade-off with other life-history traits [6,7]. Many explanations have been put forward to explain the ubiquity of condition-dependent life histories and signals, with one influential theory predicting that the adaptive significance of condition-dependent signals may arise from the large number of genes that may influence variation in condition, thereby offering females the opportunity to assess a substantial proportion of male genomes in determining male quality [8]. Under this hypothesis, selection favours females who based their mate choice decisions on condition-dependent signals because such behaviour increases the females' chances of obtaining good genes for their offspring. This line of reasoning can be extended to predict that one class of genes that may be of particular interest to females are those loci that contribute to variation in parasite resistance [9], a major determinant of reproductive success and survival in many species [10]. The condition-mediated immunocompetence-handicap hypothesis (CMIH) [7,11-14], proposes that females base their mate choice decisions on condition-dependent male signals in order to obtain genes that confer high immunocompetence for their offspring.
A key requirement of the CMIH hypothesis, and other related life history hypotheses [15-19], is the presence of positive genetic covariance between body condition and immune response. The CMIH hypothesis proposes that it is this genetic covariance that enables a condition-dependent signal to advertise the quality of the parasite resistance genes that a male carries [14]. Although there is abundant evidence for positive phenotypic associations between body condition and immunocompetence [12,15,16,18,20-22], phenotypic analyses are insufficient to validate the CMIH hypothesis because it remains unknown whether females selecting males with good body condition simply obtain a healthy mate, or if they actually acquire genes for their offspring that confer high immunocompetence. An additive genetic component has been established in several experimental systems for both body condition [[23], but see [24]] and immune response [[25-29]; but see [30-32]]. There is also evidence of a genetic correlation between immune function and sexual signals [33,34], between immune function and life history traits [35] and between body condition and male signal [36]. As far as we are aware, however, it has not been empirically demonstrated that variation in immune response is mediated by genetic variation in body condition, a key element of the CMIH hypothesis [14].
The overall aim of this study was to test directly for genetic covariance between indices of body condition and immunocompetence in a small passerine bird, the zebra finch Taeniopygia guttata. Zebra finches provide an ideal opportunity to determine if this critical genetic association exists for two reasons. First, this species is a model system for the study of sexual selection, in which female choice is based on a number of condition-dependent male signals that include song rate and bill colour [37-40]. Second, there is phenotypic evidence of condition-dependent expression of immunocompetence in this species [4,16]. We therefore used this system to investigate the genetic basis of covariation between an index of body condition and an index of immunocompetence using a cross-fostering experiment. Here we implement the cross-fostering experimental design of Riska et al. [42] to estimate additive genetic components of variance. An important advantage of this method was that it allowed us to partition the genetic covariance between these traits into sources attributable to direct additive genetic covariance, additive maternal genetic covariance, and the covariance between these sources.
Results
Genotype-environment interaction
The interaction term (Iijk) testing for a genotype-environmental interaction for Equation (1) was not significant for our indices of either immune response (F19,42 = 0.86; p = 0.63) or body condition (F19,42 = 0.81, p = 0.69). This showed that chicks from the two sampling sites did not respond differently to nest environments from the other population. Similarly, there were no significant differences between the sites for the phenotypic means of our indices of either immunocompetence (F1,119 = 0.88, p = 0.35) or body condition (F1,119 = 3.28, p = 0.08), or the breeding values of broods for either trait (Immunocompetence F1,39 = 1.85, p = 0.18: Body condition F1,39 = 3.32, p = 0.08; Figure 1). As explained in the Methods section, the marginally non-significant difference between the populations in our index of body condition is not sufficient to cause a spurious correlation between our indices of body condition and immunocompetence. As shown in Figure 1, although one population tends to have slightly higher breeding values for our index of body condition, both populations span almost the full range of values for both traits. Site was not, therefore, considered in the subsequent genetic models.
Figure 1 Genetic correlation between our indices of immune response and body condition. Each point represents the average breeding value across a single breeding pair. Filled and hollow circles show breeding pairs derived from the Alice Springs and Townsville sites respectively. Breeding values were estimated as BLUPs from the linear model described in Equation (1) in the Methods section.
Genetic covariance
When we based our analyses on comparisons between siblings alone, we detected positive genetic covariance between our indices of immune response and body condition. A plot of the breeding values, based on BLUP methodology, for our indices of immune response and body condition visually demonstrates the positive trend indicating a positive genetic correlation (Fig. 1).
The mixed-model approach detected positive covariances for observational components of covariance y1 and y3 for our indices of both immune response and body condition (Table 1). These components accounted for the majority of additive genetic variation in the design matrix (Table 2), which was relatively free of confounding non-additive and environmental effects. The estimates of additive direct genetic variance for our indices of immune response and body condition were 0.034 ± + 0.016 (p = 0.036) and 0.149 ± + 0.156 (p = 0.342), respectively. Most importantly, however, we detected significant positive genetic covariance (0.053 ± + 0.024, p = 0.030) between our indices of immunocompetence and body condition (Table 3). The estimate of the genetic correlation (rA) between our indices of immune response level and body condition was 0.75 ± + 0.41 (Table 3), suggesting that approximately 56% of the additive genetic variance in body condition can be explained by genetic variation in our index of immune response, which was calculated as square of the genetic correlation [43]. No other causal components were significant in any of the analyses.
Table 1 Observed components of variance for ten types of relatives.
Observed component (yi) Method and types of relatives used to obtain component Immune response Body condition Covariance
y1 covariance between sire and offspring 0.003 0.106 0.019
y2 covariance between nurse and offspring where the nurse is not the genetic dam 0.002 -0.122 -0.013
y3 covariance between dam and offspring where the offspring was nursed by an unrelated dam 0.01 0.119 0.002
y4 covariance between dam and offspring, where the dam is also the nurse. 0.006 0.022 0.022
y5 covariance between full sibs raised by different nurses 0.010 0.328 0.025
y6 covariance between unrelated sibs where the offspring were nursed by an unrelated dam 0.225 0.179 0.066
y7 covariance between full sibs raised by different nurses 0.002 0.239 -0.002
y8 covariance between unrelated sibs raised the same nurse 0.002 0.076 0.012
Y9 covariance between unrelated sibs, each nursed by the genetic dam of the other -0.215 0.149 -0.042
Y10 variance among full sibs all with the same nurse 0.031 0.609 0.011
Definitions modified from Riska et al. [42]. Further details are in the Methods.
Table 2 Design matrix (X) displaying theoretical causal components of observed variances and covariances.
Observed component Causal components
σ2AO σ2DO σAOAM σ2AM σ2DM+C σ2E
Y1 0.5 0 0.25 0 0 0
Y2 0 0 0.75 0.5 0 0
Y3 0.5 0 0.25 0 0 0
Y4 0.5 0 1.25 0.5 0 0
Y5 0.5 0.25 1 1 1 0
Y6 0.5 0.25 0 1 1 0
Y7 0.5 0.25 0.5 0 0 0
Y8 0 0 0.5 1 1 0
Y9 0 0 1 0 0 0
Y10 0.5 0.75 0 0 0 1
σAO2 = additive direct genetic variance, σDO2 = dominance direct genetic variance, σAOAM = direct-maternal additive genetic covariance, σAM2 = additive maternal (or 'indirect') genetic variance, σ2DM+C = dominance maternal genetic variance and common environmental variance (maternal and cage), and σE2 = residual environmental variance. Further details are in the Methods.
Table 3 Genetic and non-genetic sources of variance and covariance.
source of variation immune response bi ± SE body condition bi ± SE covariance bi ± SE
σAO2 0.034 ± 0.016* 0.149 ± 0.156 0.053 ± 0.024*
σDO2 0.069 ± 0.062 0.386 ± 0.526 0.010 ± 0.110
σAOAM -0.044 ± 0.026 0.142 ± 0.178 -0.035 ± 0.029
σAM2 0.079 ± 0.046 -0.457 ± 0.346 0.051 ± 0.055
σ2DM+C -0.055 ± 0.033 0.465 ± 0.327 -0.018 ± 0.049
σE2 -0.038 ± 0.049 0.245 ± 0.385 -0.023 ± 0.080
σP2 0.045 ± 0.101 0.929 ± 0.841 0.037 ± 0.159
σAO2 = additive direct genetic variance, σDO2 = dominance direct genetic variance, σAOAM = direct-maternal additive genetic covariance, σAM2 = additive maternal (or 'indirect') genetic variance, σ2DM+C = dominance maternal genetic variance and common environmental variance (maternal and cage), and σE2 = residual environmental variance. Asterisks denote statistical significance (* p < 0.05)
Discussion
The relative roles of genetic and non-genetic factors in determining immunocompetence in birds is controversial. Although theoretical models of sexual selection tend to assume that such traits have a high genetic component, previous empirical evidence has often proven equivocal. The results of our cross-fostering experiment support claims that a part of the variation in at least one aspect of immune response is caused by genes [25-29]. Since males therefore differ in genetically-based levels of this aspect of immunocompetence, females could conceivably target immunocompetence during mate choice as predicted by the CMIH hypothesis.
In addition to indicating that immunocompetence is heritable in zebra finches, our experiment found substantial positive genetic covariance between our indices of immunocompetence and body condition. We found that approximately 56% of genetic variation in our index of body condition may be explained by genetic variation in our index of immune response. Since secondary sexual ornaments are typically condition-dependent in zebra finches [39,40], females that select males on the basis of well developed ornaments are likely to gain genes for their offspring which confer higher levels of immunocompetence. Therefore, our findings are consistent with the hypothesis that the genes that determine parasite-resistance may be a major target of sexual selection in this species.
The significant genetic covariance between our indices of immunocompetence and condition implies that the same genes underlie a proportion of the variation in both traits. However, although females selecting males with well developed ornaments are likely to gain genes that confer higher levels of immunocompetence, our genetic analysis is not sufficient to conclusively demonstrate that there are some genes that affect our indices of both immunocompetence and body condition. It is possible that linkage disequilibrium, generated by selection for both traits, may also contribute to the genetic covariance we have found between these two traits. Distinguishing pleiotropy from linkage disequilibrium in such a species is difficult; one approach would be the development of a pleiotropic quantitative trait loci map of both traits, but such techniques are yet to be applied in wild passerine populations.
The magnitude of the genetic correlation between our indices of immunocompetence and condition is surprising because theory predicts that such heritable genetic variation should be eroded through selection. What factors could maintain such variation? One possible explanation is that a third trait not included in this analysis trades off with our indices of immunocompetence and condition [44]. For example, one aspect of growth rate displays a negative genetic correlation with our index of immunocompetence (DJG unpublished data), suggesting that a more complex model of resource allocation than the simpler two-trait system of our indices of immunocompetence and condition might need to be considered to understand the maintenance of genetic variance in these traits.
Our results should be interpreted cautiously as this study suffers from a number of limitations. First, although our reciprocal cross-fostering design is efficient at detecting additive genetic effects, it is less powerful in estimating other quantitative genetic components. Our results indicated that none of the other causal components of covariance that we estimated were found to be significant. On first inspection, this suggests that non-additive genetic and environment effects do not play a role in generating covariance between our indices of immunocompetence and body condition, but we caution that this interpretation would be premature. Because the cross-fostering experimental design and the subsequent analytical method that we have used are primarily designed to detect additive genetic components of variance and covariance [42], we cannot exclude the possibility that environmental and/or non-additive components do exist and that we have simply failed to detect to them. This possibility is highlighted by the covariance estimates in Table 3, many of which are large in magnitude and are only non-significant because of the very large standard errors. Under these circumstances, no firm conclusions about the absence of environmental or non-additive covariance between these two traits can be drawn.
A second limitation of cross-fostering designs that use full sib cross-fostering is that they can only control for those aspects of environmental variance that occur after the cross-foster manipulation itself. In the case of avian studies, chicks are typically cross-fostered within 48 hours of hatching. Hence, although such studies can estimate variation associated with later incubation and parental feeding, they cannot deal with variance in factors such as the way mothers provision eggs or anything that happens in first few hours in the nest. There is always a risk, therefore, that cross-foster studies will inflate the estimate genetic components because these also include pre-cross-foster environmental components. Nevertheless, in the case of our study such environmental covariance is unlikely to be solely responsible for the high positive genetic covariance between our indices of immunocompetence and condition because the sire-offspring observation component is large and positive, which is unaffected by this source of variation.
The third limitation of our study is the lack of information about the adults used to establish the breeding experiment. Because the adults were caught from wild populations we do not know whether they were related to one another, their immunological history, or whether the parental generation experienced selection, all of which could effect the pattern of variation among individuals. In addition, although field parent and laboratory offspring relationships have been used to estimate genetic components of variance, it is not clear how differences between the lab and field environments would affect the genetic estimates in a model as complex as the one employed here.
Finally, our study is also limited by the fact that we have only used a single index of body condition and a single index of immunocompetence. Using residual values of body weight on skeletal size is a widely used index of body condition in birds, but it is well established that this method is not without its limitations and dangers [24,45]. Such an index cannot, for instance, differentiate between different aspects of condition, such as fat deposition and muscle size, and it has been shown that such an index can retain an element of body size itself. We nevertheless used this measure for our genetic tests, firstly, because it the index of body condition that has been used widely in previous phenotypic studies and is therefore of particular interest to avian studies, secondly, because it has repeatedly been shown to be under selection in avian populations [46], and thirdly, because it is not currently possible to perform multivariate analyses (with body size and tarsus length as separate covariates) using the genetic framework employed here. It would nevertheless be interesting in the future to test for genetic correlations among alternative measures of body composition, and to perform multivariate analyses if the statistical techniques are developed. Similarly, like any other index of immune response, our single measure of cell-mediate immunity cannot provide information on all elements of vertebrate immunocompetence [18,47]. Again, we used this index in our genetic study because it has been widely employed in phenotypic tests [15,20,22,30,40], has been shown to be associated with important components of fitness such as survival [20,30], and there is even limited evidence that variation in this measure may be positive associated with variation humoral immune response [[48], but see [20]]. But this should not obscure the facts that future studies of other elements of immunity are required to obtain a comprehensive understanding of the genetic basis of parasite resistance, and that multivariate models would help to tease these apart. Ideally, a full genetic variance-covariance matrix is required for a series of indices of body condition and a series of indices of immunocompetence, but this is well beyond the scope of the current study.
Conclusion
Our results support a key prediction of the CMIH hypothesis; that there is positive genetic covariance between an index of body condition and an index of immunocompetence. More generally, although we have primarily been concerned here with the link between our indices of immunocompetence and body condition in the context of the CMIH hypothesis, a genetic link between indices of condition and immune function would also have implications outside sexual selection theory. Condition-dependence is a general feature of many aspects of life histories in many animal species [1,2]. The genetic correlation between our indices of immunocompetence and body condition that we have found in zebra finches suggests that the inter-relationships between many such traits may also prove to be parasite-mediated [12,19]. The prevalence of condition-dependent life histories may therefore arise, in part at least, through parasite resistance being a target of both natural and sexual selection. The ongoing challenge is to test the generality of these findings, that is whether there is significant positive genetic covariance between other indices of body condition and immunocompetence. More generally, it would be interesting to know the pattern of genetic covariance between various measures of condition and a suite of fitness-related traits.
Methods
Populations and experimental design
To sample the genetic variance present in field populations, we caught eighty zebra finches from the wild and then conducted a partial cross-fostering experiment [42,49], under standardised laboratory conditions. Zebra finches were caught between August 1998 and February 1999 using mist nets at sites near Alice Springs and Townsville, Australia. Twenty birds of each sex were collected from each site, and transported by plane. For the breeding experiments, birds were kept outdoors in two large free-flight breeding aviaries at the University of Queensland. Each aviary housed twenty pairs of finches, which were introduced at the same time and allowed to form pairs naturally. All pairs bred during the course of this experiment. Ad libitum food and water were provided during the study period, including fresh green material. Birds from the two collection sites were housed separately, but offspring were cross-fostered between aviaries.
A reciprocal partial cross-fostering design was used to maximise the opportunity to estimate maternal and non-additive genetic components of variance [42]. For each reciprocal cross-foster event, the two nests between which chicks were transferred was referred to as a 'block' of nests. Broods typically consisted of four offspring, with two offspring being transferred between a pair broods. Broods within a block were matched with respect to hatching date and clutch size. Chicks were selected randomly for cross-fostering except for 'runts', which were not cross-fostered and always died before measurement. The identity of individual chicks was monitored by clipping their downy head-tufts until they were large enough to carry individually-numbered leg bands. Offspring were cross-fostered immediately after hatching, and body condition and immune response were measured 17 days later.
Measurement of traits
The element of immunocompetence that we measured was experimentally-induced T-lymphocyte cell-mediated immune response, an acquired component of the avian immune system[18]. We induced a cell mediated immune response through intradermal injection of phytohemagglutinin-P [15]. For each bird, 0.1 mg of phytohemagglutinin-P in 0.02 ml of phosphate buffered saline was intradermally injecting the right wing web, with 0.02 ml of phosphate buffered saline being injected into the opposite wing web as a control. The thickness of each wing web was measured at the injection site both immediately before and 24 hours after the injections. Twenty four hours is the standard reaction period in avian studies and is the point at which the swelling is typically maximum [15]. Measurements were taken three times to the nearest 0.001 mm using a digital micrometer and 'before' and 'after' averages were calculated for each wing. We then calculated the swelling for each wing, which was the difference between the 'before' and 'after' averages. Finally, cell-mediated immune response was calculated for each individual as the difference in swelling between the phytohemagglutinin-P injected wing and the control wing.
In this study body condition is defined as, and was measured as, the residual value from the regression of body mass on tarsus length [46]. To enhance the clarity of our writing we refer to this measure throughout this study as an index of body condition, although the reader should keep in mind that this index is derived from an estimate of residual body mass. Body mass was measured to 0.01 g using a Petit Precision balance (model MK-200 200 g × 0.01 g). Tarsus was measured to the nearest 0.5 mm using digital callipers. For parental individuals, all measurements were taken immediately before they began a breeding cycle.
Genotype-environment interactions
Because we collected birds from two different sites, we tested for possible genotype-environment interactions for our indices of both immune response and body condition. A two-way factorial ANOVA comprising nest of origin, nest of rearing and the interaction term for each block of nests was used [49-51]:
Zijkl = μ + Pi + Mij + Nik + Iijk + eijkl (1)
where, Pi = average effect of the ith cross-fostered block of nests, Mij = direct effect of the jth (genetic) mother within the ith block (j = 1 or 2), Nik = kth (unrelated) nurse within the ith block (k = 1 or 2), Iijk = M × N interaction within the ith block, and eijkl = residual error for the ith offspring of the jth mother raised by the kth nurse within the lth block of nests.
The Iijk term of Eq. 1 tested for the presence of a genotype-environment interaction in this experiment as nest of origin also represented genetic population of origin and nest of rearing also represented the population of rearing. Significance of nest of origin (Mij) and nest of rearing (Nik) was tested using the interaction term (Iijk) as the error, type III sums of squares for unbalanced designs. In all genetic models genetic relationships were inferred on the basis of the male and female providing care at the nest in question, as extra-pair paternity in zebra finch colonies is low (2.4% of chicks) [52].
When using individuals from two different populations there is also the risk that such populations could differ with respect to the parameters under study. Specifically, if the populations differed with respect to both traits then pooling individuals from the two populations might generate spurious covariance between the traits. It is important to note here that the populations need to differ for both traits and not just one of them. This is because, if populations only differ with respect to in one variable, this would just increase variation along a single axis. To assess these possibilities we therefore tested for differences between the source populations in both the phenotypic means and breeding values of our indices of both immune response and body condition.
Genetic covariance
We used two related methods to test for a genetic correlation between our indices of immunocompetence and body condition. To facilitate comparisons with other studies, we first used the standard method for analysing cross-foster experiments, which is based on using offspring values alone [49]. We used BLUP methodology to calculate breeding values [49] and plotted these against each other to visualise the pattern of covariance. However, analyses based on full-siblings alone may lead to biased estimates of additive genetic components since pre-fostering maternal effects and dominance genetic variance cannot be partitioned out from additive genetic effects [49,53]. The confounding of genetic and environmental factors may be of particular concern in birds, where the egg environment may provide a source of direct maternal effects [20].
To avoid the potential problems associated with genetic estimates based on comparisons between full-sibs alone, we used an under-utilised second method based on a mixed model [42] to test for a genetic correlation. An important advantage of this type of model is that it allows both offspring and parental trait values to be used to distinguish a number of sources of variation, thereby enabling the separation of additive genetic variance from dominance genetic variance and direct maternal effects [42,49]. The degree of similarity between ten types of relatives (Table 1) was then used to estimate six genetic and non-genetic causal components contained in each of these observational components, which are displayed in the design matrix X (Table 2).
The observation vector y comprised ten observational components of variance (yi) that were estimated by the various methods listed in Table 1. All components were computed using methodology taken directly from Riska et al. [42], with the exception of y9. The estimation of the direct-maternal additive genetic covariance which is isolated by observational component 9 has been the source of some confusion in the literature. Rutledge et al [54] first proposed σAOAM could be estimated using the interaction term in (1), which more recently was also advocated as an appropriate way of estimation in Lynch and Walsh (1998). However, this method of estimation was subsequently shown to be incorrect [55,56]. Riska et al [42] outlined that component 9 could be estimated in another two equivalent ways, but we did not find the exact numerical agreement between these two methods suggested by these authors (unpublished results). We therefore used an established alternative method for estimating component 9 described by [55,57], which uses the difference between the covariance between full sibs raised by different nurses, and the covariance between unrelated sibs raised by the same nurse. We note that a limitation of the Riska et al [42] approach is that using the same mean squares for the estimation of different causal components generates covariance between the estimates which is not accounted for in the model as implemented either by Riska et al [42,55] or here (i.e. the off-diagonal elements of the V matrix are set to zero, see below). By using the estimation method of component 9 employed here, this potential problem is likely to be exacerbated as the estimate of component 9 is a linear combination of the mean squares used in other observational components (5 and 6). Nevertheless, component 9 as estimated here has an established interpretation, and facilitated the isolation of the important σAOAM causal component.
Variances of the observational components were used as the diagonal elements of the square matrix V, with off-diagonal elements all zero. To obtain variances for each observational component [49] for components 1–4:
VAR(σ2) = (VAR(A) VAR(B) + COV(A, B)2) / (N) (2)
where A and B represented the two kinds of individuals whose covariance is being estimated and N is the number of bivariate observations. The variance of components 5–10 were estimated as weighted sums of the variances of the appropriate mean squares, where the variance of a mean square is given by (Ref [49], equ. A1.10c):
VAR(σ2) = (2MS2) / (N + 2) (3)
in which MS represents the mean square of the term of interest and N is the number of blocks. The causal components of variance were estimated as elements of the vector:
b = (X'V-1X)-1X'V-1y (4)
with covariance matrix:
S = (X'V-1X)-1. (5)
where the square root of the corresponding diagonal element of S was used to estimate the standard error of b. Phenotypic variance was estimated as the sum of the elements of b and its corresponding standard error approximated by the square root of the summed diagonal elements of S.
Estimation of the genetic correlation between our indices of body condition and cell-mediated immune response level required all observational components to be estimated as cross-covariances [58]. Cross-covariances were estimated as the product of the values of our indices of body condition and immune response for each individual and the sums of products partitioned according to the source of variation (Falconer 1981). The variances of these cross-covariances (used as the diagonal elements of the square matrix V) were determined by calculating separate variances of cross-covariances for (i) our index of body condition in parents and our index of immune response in offspring, and (ii) our index of immune response in parents and our index of body condition in offspring. The mean was then taken of these two variances of cross-covariances. The additive genetic correlation (rA) and an estimate of its standard deviation were calculated using equations 19.2 and 19.4 in Falconer [58], respectively. The proportion of additive genetic variance in our index of body condition explained by genetic variation in our index of immune response was then calculated as the square of the genetic correlation [43].
Authors' contributions
DJG helped to design the project, collected animals from the wild, conducted all crosses and measurements, performed statistical analyses, and helped to write the paper. MWB helped with the statistical analyses and the writing of the paper. IPFO helped to design the project, collect animals, and write the paper. All authors read and commented on drafts of the manuscript, and approved the final manuscript.
Acknowledgements
We thank R. Brooks, C. Godfray, F. Hausmann, L. Kruuk, T. Price, S. Scott, R. Whitney and three anonymous referees for their help; the Australian Research Council for funding; the University of Queensland for ethical permission; and the Northern Territory and Queensland governments for collection permits.
==== Refs
Roff DA The Evolution of Life Histories 2002 New York: Chapman and Hall
Stearns SC The Evolution of Life Histories 1992 New York: Oxford University Press
Andersson M Sexual selection 1994 Princeton, New Jersey: Princeton University Press
Johnstone RA Sexual selection, honest advertisement and the handicap principle: reviewing the evidence Biol Rev 1995 70 1 65 7718697 10.1086/418864
Cotton S Fowler K Pomiankowski A Do sexual oranaments demonstrate heightened condition-dependent expression as predicted by the handicap hypothesis? Proc Yoy Soc Lond B 2004 271 771 783 10.1098/rspb.2004.2688
Andersson M Evolution of condition-dependent sex ornaments and mating preferences: sexual selection based on viability differences Evolution 1986 4 804 816
Sheldon BC Verhulst S Ecological immunology: costly parasite defences and trade-offs in evolutionary ecology Trends Ecology & Evolution 1996 11 317 321 10.1016/0169-5347(96)10039-2
Rowe L Houle D The lek paradox and the capture of genetic variance by condition dependent traits Proc R Soc Lond B 1996 263 1415 1421
Hamilton WD Zuk M Heritable true fitness and bright birds: a role for parasites Science 1982 218 384 387 7123238
Loye JE Zuk M Bird-parasite interaction: ecology, evolution and behaviour 1991 Oxford University Press
Folstad I Karter AJ Parasites, bright males, and the immunocompetence handicap Am Nat 1992 139 603 622 10.1086/285346
Gustafsson L Nordling D Andersson MS Sheldon BC Qvarnstrom A Infectious diseases, reproductive effort and the cost of reproduction in birds Phil Trans R Soc Lond B 1994 346 323 331 7708827
Moller AP Hormones, handicaps and bright birds Trends Ecology & Evolution 1995 10 121 10.1016/S0169-5347(00)89008-4
Westneat DF Birkhead TR Alternative hypotheses linking the immune system and mate choice for good genes Proc R Soc Lond B 1998 265 1065 1073 10.1098/rspb.1998.0400
Lochmiller RL Vestey MR Boren JC Relationship between protein nutritional status and immunocompetence in northern bobwhite chicks The Auk 1993 110 503 510
Deerenberg C Apanius V Daan S Bos N Reproductive effort decreases antibody responsiveness Proc R Soc Lond B 1997 264 1021 1029 10.1098/rspb.1997.0141
Lochmiller RL Deerenberg C Trade-offs in evolutionary immunology: just what is the cost of immunity? Oikos 2000 88 87 98 10.1034/j.1600-0706.2000.880110.x
Norris K Evans MR Ecological immunology: life history trade-offs and immune defense in birds Behav Ecol 2000 11 19 26 10.1093/beheco/11.1.19
Zuk M Stoehr AM Immune defense and host life history Am Nat 2002 160 S9 S22 10.1086/342131
Gonzalez G Sorci G Moller AP Ninni P Haussy C De LF Immunocompetence and condition-dependent sexual advertisement in male house sparrows (Passer domesticus) J Anim Ecol 1999 68 1225 1234 10.1046/j.1365-2656.1999.00364.x
Alonso-Alvarez C Tella JL Effects of experimental food restriction and body-mass changes on avian T-cell-mediated immune response Can J Zool 2001 79 101 105 10.1139/cjz-79-1-101
Moller AP Petrie M Condition dependence, multiple sexual signals, and immunocompetence in peacocks Behav Ecol 2002 13 248 253 10.1093/beheco/13.2.248
Merila J Genetic variation in offspring condition – an experiment Funct Ecol 1996 10 465 474
Gosler AG Harper DGC Assessing the heritability of body condition in birds: a challenge exemplified by the great tit Parus major L. (Aves) Biol J Linn Soc 2000 71 103 117 10.1006/bijl.1999.0432
Saino N Calza S Moller AP Immunocompetence of nestling barn swallows in relation to brood size and parental effort J Anim Ecol 1997 66 827
Wakelin D Apianus V Clayton DH, Moore J Immune defences: genetic control Host-parasite evolution General principles and avian models 1997 Oxford University Press 30 58
Brinkhof MWG Heeb P Kolliker M Richner H Immunocompetence of nestling great tits in relation to rearing environment and parentage Proc R Soc Lond B 1999 266 2315 2322 10.1098/rspb.1999.0925
Kurtz J Sauer KP The immunocompetence handicap hypothesis: testing the genetic predictions Proc R Soc Lond B 1999 266 2515 2522 10.1098/rspb.1999.0954
Roulin A Jungi TW Pfister H Dijkstra C Female barn owls (Tyto alba) advertise good genes Proc R Soc Lond B 2000 267 937 941 10.1098/rspb.2000.1022
Christe P Moller AP de Lope F Immunocompetence and nestling survival in the house martin: the tasty chick hypothesis Oikos 1998 83 175 179
Tella JL Bortolotti GR Forero MG Dawson RD Environmental and genetic variation in T-cell-mediated immune response of fledgling American kestrels Oecologia 2000 123 453 459 10.1007/s004420000331
Soler JJ Moreno J Potti J Environmental, genetic and maternal components of immunocompetence of nestling pied flycatchers from a cross-fostering study Evolutionary Ecology Research 2003 5 259 272
Ryder JJ Siva-Jothy MT Quantitative genetics of immune function and body size in the house cricket, Acheta domesticus J Evol Biol 2001 14 646 653 10.1046/j.1420-9101.2001.00302.x
Verhulst S Dieleman SJ Parmentier HK A tradeoff between immunocompetence and sexual ornamentation in domestic fowl Proc Natl Acad Sci USA 1999 96 4478 4481 10200287 10.1073/pnas.96.8.4478
Kraaijeveld AR Godfray HCJ Trade-off between parasitoid resistance and larval competitive ability in Drosophila melanogaster Nature 1997 389 278 280 9305840 10.1038/38483
Kotiaho JS Simmons LW Tomkins JL Towards a resolution of the lek paradox Nature 2001 410 684 686 11287953 10.1038/35070557
Burley N Coopersmith CB Bill colour preferences for zebra finches Ethology 1987 76 133 151
Collins SA ten Cate C Does beak colour affect female preference in zebra finches? Anim Behav 1996 52 105 112 10.1006/anbe.1996.0156
Birkhead TR Fletcher F Pellatt EJ Sexual selection in the zebra finch Taeniopygia guttata : condition, sex traits and immune capacity Behav Ecol Sociobiol 1998 44 179 191 10.1007/s002650050530
Blount JD Metcalfe NB Birkhead TR Surai PF Carotenoid modulation of immune function and sexual attractiveness in zebra finches Science 2003 300 125 127 12677066 10.1126/science.1082142
Ewenson E Zann R Flannery G PHA immune response assay in captive zebra finches is modulated by activity prior to testing Anim Behav 2003 66 797 800 10.1006/anbe.2003.2251
Riska B Rutledge JJ Atchley WR Genetic analysis of crossfostering data with sire and dam records The Journal of Heredity 1985 76 247 250 4031461
Lande R The dynamics of peak shifts and the pattern of morphological evolution Paleobiol 1986 12 343 354
Pease CM Bull JJ A critique of methods for measuring life history trade-offs J Evol Biol 1988 1 293 303 10.1046/j.1420-9101.1988.1040293.x
Green AJ Mass/length residuals: measures of body condition or generators of spurious results? Ecology 2001 82 1473 1483
Brown ME Assessing body condition in birds Current Ornithol 1996 13 67 135
Owens IPF Wilson K Immunocompetence: a neglected life history trait or conspicuous red herring? Trends Ecol Evol 1999 14 170 172 10.1016/S0169-5347(98)01580-8
Hoi-Leitner M Romero-Pujante M Hoi H Pavlova A Food availability and immune capacity in serin (Serinus serinus) nestlings Behav Ecol Sociobiol 2001 49 333 339 10.1007/s002650000310
Lynch M Walsh B Genetics and analysis of quantitative traits 1998 Sunderland, Mass: Sinauer
Atchley WR Rutledge JJ Genetic components of size and shape. I. Dynamics of components of phenotypic variability and covariability during ontogeny in the laboratory rat Evolution 1980 34 1161 1173
Price T Environmental and genotype-by-environment influences on chick size in the Yellow-browed leaf warbler Phylloscopus inornatus Oecologia 1991 86 535 541 10.1007/BF00318320
Birkhead TR Burke T Zann R Hunter FM Krupa AP Extra-pair paternity and intraspecific brood parasitism in wild Zebra Finches Taeniopygia guttata, revealed by DNA fingerprinting Behav Ecol Sociobiol 1990 27 315 324 10.1007/BF00164002
Merila J Sheldon BC Avian Quantitative Genetics Current Ornithol 2001 16 179 255
Rutledge JJ Robinson OW Eisen EJ Legets JE Dynamics of genetic and maternal effects in mice J Anim Sci 1972 35 911 918 5085299
Riska B Rutledge JJ Atchley WR Covariance between direct and maternal genetic effects in mice, with a model of persistent environmental-influences Genet Res 1985 45 287 297 4029615
Newman S Rutledge JJ Riska B Estimation of prenatal maternal genetic-effects J Anim Breed Gen 1989 106 30 38
McAdam AG Boutin S Effects of food abundance on genetic and maternal variation in the growth rate of juvenile red squirrels J Evol Biol 2003 16 1249 1256 14640416 10.1046/j.1420-9101.2003.00630.x
Falconer DS Introduction to Quantitative Genetics 1989 USA Longman Scientific & Technical
|
16274488
|
PMC1298291
|
CC BY
|
2021-01-04 16:29:16
|
no
|
BMC Evol Biol. 2005 Nov 7; 5:61
|
utf-8
|
BMC Evol Biol
| 2,005 |
10.1186/1471-2148-5-61
|
oa_comm
|
==== Front
BMC GenetBMC Genetics1471-2156BioMed Central London 1471-2156-6-531626909110.1186/1471-2156-6-53Research ArticleRB1 gene mutation up-date, a meta-analysis based on 932 reported mutations available in a searchable database Valverde José R [email protected] Javier [email protected] Itziar [email protected]ña Ángel [email protected] Servicio de Informática. Centro Nacional de Biotecnología, CSIC. Campus de Cantoblanco. 28049-Madrid, Spain2 Oncolab. Deparatamento de Biología Molecular y Celular del Cáncer. Instituto de Investigaciones Biomédicas "A. Sols", CSIC-UAM. 28029 Madrid, Spain2005 4 11 2005 6 53 53 25 4 2005 4 11 2005 Copyright © 2005 Valverde et al; licensee BioMed Central Ltd.2005Valverde 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
Retinoblastoma, a prototype of hereditary cancer, is the most common intraocular tumour in children and potential cause of blindness from therapeutic eye ablation, second tumours in germ line carrier's survivors, and even death when left untreated. The molecular scanning of RB1 in search of germ line mutations lead to the publication of more than 900 mutations whose knowledge is important for genetic counselling and the characterization of phenotypic-genotypic relationships.
Results
A searchable database (RBGMdb) has been constructed with 932 published RB1 mutations. The spectrum of these mutations has been analyzed with the following results: 1) the retinoblastoma protein is frequently inactivated by deletions and nonsense mutations while missense mutations are the main inactivating event in most genetic diseases. 2) Near 40% of RB1 gene mutations are recurrent and gather in sixteen hot points, including twelve nonsense, two missense and three splicing mutations. The remainder mutations are scattered along RB1, being most frequent in exons 9, 10, 14, 17, 18, 20, and 23. 3) The analysis of RB1 mutations by country of origin of the patients identifies two groups in which the incidence of nonsense and splicing mutations show differences extremely significant, and suggest the involvement of predisposing ethnic backgrounds. 4) A significant association between late age at diagnosis and splicing mutations in bilateral retinoblastoma patients suggests the occurrence of a delayed-onset genotype. 5) Most of the reported mutations in low-penetrance families fall in three groups: a) Mutations in regulatory sequences at the promoter resulting in low expression of a normal Rb; b) Missense and in-frame deletions affecting non-essential sequence motifs which result in a partial inactivation of Rb functions; c) Splicing mutations leading to the reduction of normal mRNA splicing or to alternative splicing involving either true oncogenic or defective (weak) alleles.
Conclusion
The analysis of RB1 gene mutations logged in the RBGMdb has shown relevant phenotype-genotype relationships and provided working hypothesis to ascertain mechanisms linking certain mutations to ethnicity, delayed onset of the disease and low-penetrance. Gene profiling of tumors will help to clarify the genetic background linked to ethnicity and variable expressivity or delayed onset phenotypes.
==== Body
Background
Retinoblastoma (MIM# 180200), a rare embryonic neoplasm of retinal origin, is the most common intraocular tumor in children, with a relative incidence of 3% of all pediatric tumors. Although current therapeutic strategies have led to dramatic improvement of individual prognosis, retinoblastoma is still life-threatening when leaved untreated or in cases of late diagnosis, a condition of concern in developing countries [1]. The frequency estimates of retinoblastoma in different populations range between 1:34.000 and 1:10.000 live-born, with the most reliable figures between 1:28.000 and 1:15.000. An increasing incidence observed in recent studies can result from more complete ascertainment and also from population-genetic reasons, due to the increased survival of retinoblastoma patients [2]. Most of the clinical phenotypes can be explained by the double mutational inactivation of the retinoblastoma susceptibility gene [3], the prototype tumor suppressor gene that controls cell cycle progression [4]. However, additional mutations in apoptosis signaling may well be involved in tumor development [5], a hypothesis that has been in the cell-of-origin studies in mice [6]. In addition, a detailed analysis of the relations between genotype and phenotypic expression suggest that the hereditary retinoblastoma has features of a complex trait [7]. In the hereditary form of the disease, a germ line mutation is transmitted as a high penetrance (90%) autonomic dominant trait, resulting in a 45% risk in offspring of patients with hereditary retinoblastoma; the second inactivating mutation occurs in retinal cell precursors [8]. Most of these patients have bilateral retinoblastoma and a mean age at diagnosis of 12 months. In the non-hereditary form of the disease, both inactivating events occur during somatic development of retinal cells and result in the relatively late onset of a single tumor in one eye [9]. However, nearly 15% of the unilaterally affected patients have germ line RB1 mutations, representing a 45% risk for their offspring, and these patients cannot be clinically distinguished from patients with true somatic unilateral retinoblastoma, who present a negligible risk for siblings and offspring. Taking these situations together, the hereditary form represents nearly 50% of all the retinoblastoma patients, according to recent epidemiological figures [10]. The presence of RB1 germ line mutations confers an increased risk for development of second primary tumors in the survivors of hereditary retinoblastoma, with a cumulative incidence of 22% at the age of 25 years. Most of these second tumors were osteosarcomas (37.0%), other sarcomas (16.8%) and melanomas (7.4%), while brain tumors (4.5%), leukemia (2.4%) and non-Hodkin lymphomas (1.6%) were less frequent [11]. In addition, hereditary retinoblastoma survivors have a lifetime risk of developing common epithelial cancers [12].
The development of sensitive and reliable genetic tests to detect RB1 mutations has greatly improved the identification of carriers, facilitating accurate genetic counseling. In addition, the detection of children at risk among siblings would obviate the need for many routine examinations and potentially decrease the economic impact of the disease [13]. Attempts by several groups to define the mutations resulting in the inactivation of the RB1 gene in retinoblastoma have led to the identification of a broad spectrum of mutations. To date, the most comprehensive report of RB1 mutations corresponds to Lohman's database [14] which contains 228 different mutations and 130 recurrences. In this article we describe an up-dated, searchable database containing 500 distinct somatic or germ line RB1 mutations and more than 400 recurrences that we have retrieved from publications. In addition, we analyze the spectrum of RB1 mutations, with emphasis in molecular epidemiology and phenotype-genotype relationships. This information is important for the development of rapid procedures to detect mutations in patients and also to understanding the molecular mechanisms leading to tumors with different degrees of penetrance or expressivity.
Results and discussion
The scope of the database of RB1 gene mutations (RBGMdb) is to retrieve and arrange data from the literature in a flexible and standardized electronic format as described in methods. In its present version, it contains 932 entries extracted from 68 articles referred in Additional File 1, together with the number of mutations they contribute and the country of origin of the reporting group. Out of these entries, 753 correspond to germ line mutations, 155 are somatic mutations in retinoblastoma patients and 24 correspond to RB1 somatic mutations found in other tumors. The distribution of these mutations in different retinoblastoma patient is shown in Table 1. These figures cannot be considered representative of the true incidence of retinoblastoma phenotypes, since most germ line studies were carried out in bilateral retinoblastoma patients and mutation analysis were performed in a limited amount of unilateral tumor samples. The database also gives information about the sex of the patient (140 entries) and the age at diagnosis or treatment (258 entries).
Table 1 Distribution of germ line and somatic RB1 mutations by phenotype of the patients.
Germ line Somatic Total
Bilateral sporadic 424 24 448
Bilateral familiar 99 99
Low penetrance 27 27
Unilateral sporadic 52 131 183
Unilateral familiar 6 6
Not reported 145 145
Total 753 155 908
Type of mutation
The RBGMdb contains 500 distinct mutations and 433 recurrences (see Table 2). Most of the 932 entries (42 %) correspond to nonsense (NS) point mutations. This figure is reduced to 18% of the sample when the 302 recurrent NS are omitted. In this case, the proportion of NS mutations is closer, although significantly different, from the data logged in the Human Gene Mutation Database [15], which gives information of more than 46000 mutations in human disease. In contrast to the high recurrence of NS mutations in retinoblastoma (70% of the total recurrences), small insertions, deletions or complex ins_del (ins/del) show a low recurrence and represent a high proportion (48%) of the 500 distinct RB1 mutations; this figure is much higher than the data in HGMD. Splicing mutations in RB1 have a moderate (19%) recurrence and represent twice as much of the proportion of the distinct SP mutations found in HGMD. On the contrary, MS RB1 mutations are very low (10% of the total) as compared with the 50% given in HGMD.
Table 2 Number of entries by mutation type as compared with HGMD data. Statistical analysis of distinct RBGMdbDB vs. HGMD using χ2 test.
All mutations Distinct mutations Distinct mutat.
RBGMdb % RBGMdb % HGMD % χ2 test
Nonsense 395 42.4 93 18.6 5840 12.6 P < 0.0001
Missense 81 8.7 50 10.0 22940 49.6 P < 0.0001
Splicing 194 20.8 111 22.2 4771 10.3 P < 0.0001
Regulatory 7 0.7 5 1.0 626 1.4 P = 0.6267
Small ins/del 255 27.3 241 48.2 12076 26.1 P < 0.0001
Total 932 500 46253
Hot spots of RB1 gene mutations
As shown in Figure 1, RB1 mutations are scattered along the genomic sequence, but also accumulate in discrete spots of high recurrence, comprising twelve nonsense mutations, three affecting splicing sites and two missense. These results confirm and extend previous observations [13,14] with new hot spots and higher recurrence figures. Most of the recurrences (270 out of 341, 79%) correspond to C to T transitions in eleven CGA-arginine codons, in exons 8, 10, 11, 14, 15, 17, 18 and 23 (see Additional File 2), but no mutations were found in three other CGA codons, located in exons 1 and 27. It is generally admitted that the hyper mutability of CGA codons depends on the methylated status of CpG dinucleotides and the spontaneous deamination of 5 methyl cytosine to thymidine [16]. In four of the mutated CGA codons (R251 and R255 in exon 8, R451 and R455 in exon 14) a high frequency of constitutive methylation (see Additional File 2) has been demonstrated and it is assumed that constitutive hyper methylation would also be present in the other CGA hot spots [17]. The absence of mutations at the CGA codon (R7) in exon 1, whose predicted consequence would be a short and inactive peptide, fits in this model since this codon is part of an unmethylated CpG island in the RB1 promoter [18]. On the other hand, the absence of mutations in two highly methylated CGA codons (R908 and R910) in exon 27 is expected, in view of the fact that mutations in the last RB1 exon are not oncogenic [18]. The same argument can apply to the absence of C>T mutations in the highly methylated CpG codons in position 451 (CGC), 855 (AGC), 857 (CGT) and 876 (CGC) in RB1 gene [17], whose predicted outcome would be non oncogenic missense or silent amino acid substitutions (R451R, S855S, R857C and R876R).
Figure 1 Mutational hot spots in RB1. The number of entries found for each mutation is represented against the modified genomic nucleotide. Description of high recurrent mutations is shown.
In this sense it is worth to mention the different fate of three CGG-codons in RB1 gene: two of unknown methylated status in exon 20, comprising the highly recurrent R661W (20 mutations as shown in Additional File 2) and R656W (1 mutation); the third, in exon 8 (R262) is frequently methylated although no mutations have been so far observed. In these cases, the differences in mutability cannot be explained by the methylation status not by differences in tumorogenity (oncogenity), since both R656W and R661W lie in the folded and hydrogen-bond rich structural domain of the A-B interface, together with other missense mutations [19]. Alternatively, the hyper mutability found in R661 can be explained by neighboring-nucleotide effects (TCGG in R661 versus ACGG in R656 and R 262) and the known differences in mutability of these tetranucleotides [20]. In the case of other non-CpG hot spots in RB1 gene, such as the nonsense E137X and the three splicing mutations affecting the first invariant nucleotide in introns 6, 12 and 19, the presence of short quasi-repeat sequences could be documented (see Additional file 2). These sequence motifs would favor replication errors such as misinsertions or misalignments leading to base substitution or single-base frameshift with the mismatch repair machinery [21]. Although this DNA environment offers an attractive explanation for the presence of hot spots in non CpG sites, no preponderance of direct or inverted repeats has been observed in the spectrum of single-base-pair substitutions logged in the HGMD [22]. However, the HGDM only includes distinct mutations and therefore an association of repeated sequence motifs with recurrent mutations cannot be excluded.
Mutational spectrum of RB1 by exon
In addition to hot spots, frameshift and point mutations leading to amino acid substitutions or splicing are scattered along the retinoblastoma cDNA and non-coding adjacent splicing sites, giving the spectrum of mutations shown in Figure 2. With the exception of exons 5, 14, 15, 24, 25 and the non-mutated exons 26 and 27, frameshift mutations are randomly distributed through the RB1 coding sequence. Splicing mutations are also evenly distributed, but show preference for intronic sequences adjacent to exons 6, 12, 16, 17, 19 and 24; three of them are associated to the above described recurrences (see Figure 1). The exonic distribution of point mutations correspond to the hot spots already described in Figure 1. It is worth to mention that most missense substitutions (60 %) are located in cyclin box B, underlined by exons 19 to 21. The spectrum of RB1 mutations by exon has important implications in the mutational screening of retinoblastoma patients, which might benefit from the sequential analysis employing quantitative multiplex PCR (QM-PCR) methodology to detect frameshift mutations[23,24] and PCR-sequencing of highly mutated exons 9, 10, 14, 17, 18, 20 and 23.
Figure 2 Spectrum of RB1 mutations by exon. The number of entries for each type of mutation is distributed by exon and adjacent 5' and 3' intronic sequences.
Spectrum of RB1 gene mutations by country of origin
The distribution of mutations logged in the RBDB by type and country of origin (shown in Additional File 3) allows establishing the two different spectra of RB1 gene mutations shown in Figure 3. In certain South American countries (Argentina, Brazil, Colombia, Cuba and Ecuador) as well as in Russia, United Kingdom and Germany, amounting to a total of 392 mutations (group A in Figure 3), the incidence of NS and SP mutations is respectively higher (p = 0.017) and lower (P = 0.003) than in the grand total of 925 mutations. On the contrary, in United States, France and Spain (group B in Figure 3) the incidence of NS is lower (P = 0.022) and that of SP is higher (P = 0.023) than the average found for all mutations. The differences in incidence of NS (50.8 and 35.4%, P = 0.0002) and SP (13.7 and 27.9%, P < 0.0001) mutations between groups A and B respectively, are extremely significant and suggest the presence of predisposing ethnic backgrounds. Since most NS mutations in RB1 (80% in RBGMdb) correspond to C>T transitions, origin of a G: T mispair, and the eukaryotic mismatch (MMR) complex MSH2–MSH6 (MIM#120435 and 600678, respectively) seems to be more efficient in G: T mismatch repair [25], it is suggested that susceptibility to NS RB1 mutations can be increased by an imbalance between DNA methylation vs. mismatch repair (MSH2–MSH6) activities [26].
Figure 3 Spectrum of RB1 mutations by country of origin. Distribution of mutations in Canada and two groups of nations and statistical comparison with all mutations in RBGMdb. (*) and (**) stand for P < 0.05 and 0.01, respectively).
Age and type of mutations
In correspondence with larger statistical studies [2,9] the mean age at diagnosis for bilateral and unilateral retinoblastoma patients in DBRB is of 12.5 and 24.8 months respectively, and this difference was extremely significant (P = 0.00006). Similar differences are observed in all but splicing mutations (Table 3), whose mean age at diagnostic in bilateral and unilateral patients are statistically indistinguishable. This result, which confirms previous observations [27], suggests that splicing mutations can be associated to a delayed onset phenotype. The molecular basis of this phenotype could be related to mechanisms considered in low-penetrance splicing mutations.
Table 3 Age at diagnosis in retinoblastoma patients according to phenotype and type of mutation. Statistical analysis using the Welch's unpaired t test.
Phenotype Frameshift Nonsense Missense Splicing SP vs. NS-MS-FS
Unilateral 21.1 26.0 23.7 26.7 P = 0.7237
Bilateral 9.5 11.8 9.4 19.3 P = 0.0025
All 12.4 15.6 16.0 21.8 P = 0.0044
Unilateral vs. Bilateral P = 0.0073 P = 0.0147 P = 0.3321
Low penetrance retinoblastoma
In 27 out of the 133 (20%) familial retinoblastoma entries in RBGMdb, the presence of unaffected carriers (reduced penetrance) or unilateral retinoblastoma or benign retinoma (reduced expressivity) were documented (see Additional file 4 for a complete description of the mutations and references). This figure probably represents an overestimation of the true incidence of the low penetrance (LP) phenotype, biased by its notorious scientific interest. As shown in Table 3, most of the reported mutations associated to LP families (23/28 = 82%) correspond to MS and SP mutations, with hot spots in g.156713 C>T (R661W), g.160757T>C (C712R) and g.45867G>T (IVS6+1G>T). Different mechanisms have been proposed to explain this rare phenotype, including epigenetic events, delayed mutation, involvement of a second retinoblastoma locus ("three-hit hypothesis") or host resistant factors. However, most low-penetrance retinoblastoma can be explained by mutations at the RB1 locus [28]. Mutations affecting regulatory sequences in the RB1 promoter are known to reduce the expression of normal Rb protein below a threshold level, necessary for tumor suppression functions [29,30]. Missense single-nucleotide substitutions can, under certain circumstances, partially inactivate the retinoblastoma function. In the case of the highly recurrent R661W mutant allele, Otterson et al. have shown that the mutant retinoblastoma protein has a temperature-sensitive pocket activity whose reversible fluctuations may result in the low penetrance phenotype. Temperature-sensitive Rb pocket activity may also explain the low penetrance of C712R and delN480 [31]. In the case of the large in frame deletions outside the pocket domain, such as Del: E04 [32] and Del: E24–25 [33], unessential functions of Rb protein seems to be affected.
After the report of a splicing mutation (c.2211G>A) affecting the last nucleotide of exon 21 [34], a new category of low-penetrance genotype has been proposed to occur through alternative splicing mechanisms. In the case reported by these authors, the RNA analysis showed skipping of exon 21 and a low amount (10%) of normally spliced RNA (E732E) which can explain the variable expressivity observed in that family. A similar mechanism can explain the low-penetrance of other splicing mutations affecting the last exonic nucleotide, such as c.1331A>G and c.1960G>C. In these cases, the splicing machinery could alternate between defective missense splicing (Q443P and V654L, respectively) and inactivating skipping of exons 13 or 21, both in the pocket box domain of Rb. In the low-penetrance family studied by Lefevre et al., a T>C substitution in the polypyrimidine tract of intron 8 (IVS9-10T>C) was shown to be at the origin of the in frame deletion of exon 9 giving a defective or inactive Rb protein lacking 26 amino acids from the N-terminal region. It is also possible that this mutation only partially affected spliceosome assembly and that the mutated allele could be in part correctly spliced. While this alternative splicing mechanism would better explain the low-penetrance phenotype, no supporting evidence is available [35]. Similar alternative mechanism could also explain the low-penetrance of c.2325+5G>A, causing in frame skipping of exon 22 [36]. In this case, the G>C transversion makes a slight reduction of the consensus value of the 5' splice site [37] from 88 to 75 compatible with the presence of a fraction of correctly spliced mutant allele.
Alternative splicing mechanisms might also be involved in the recurrent c.607+1G>T transversion, firstly described in a low-expressivity and delayed onset phenotype in one Spanish family [27], and thoroughly analyzed by Klutz et al [38] in two low-penetrance German pedigrees. These authors made the interesting observation of a posttranscriptional mechanism that reduces the level of the mutant transcripts (skipping of exon 6 giving a stop codon in exon 7) when the mutant allele was received from the father. However this parent-of-origin effect does not explain the low-penetrance phenotype, since all but one unaffected carriers have maternally inherited mutant alleles and high level of expression, while most of the tumor bearing carriers have paternally inherited mutant alleles and low level of expression of the mutant allele. In a search for alternative splicing mechanisms, using the electronic exon-search facilities at HGMP, we have observed the presence of a cryptic exon whose usage in mutant alleles (see Additional File 5) could result in a defective Rb protein lacking 12 amino acids in the amino terminal region. Since no evidence of in-frame restoring mechanisms in lymphoblastoid cell line derived from unaffected carriers was observed [38], the hypothetical alternative splicing mechanism should be explored in retinoblastoma derived cell lines. In the only nonsense substitution (Q675X) observed in a low-penetrance phenotype, the G>T transversion in c.2023 could also activate a cryptic splicing site involving the stop codon, with the result of a defective Rb lacking 22 amino acids [39].
Conclusion
The analysis of RB1 gene mutations logged in the RBGMdb has shown relevant phenotype-genotype relationships and provided working hypothesis to ascertain mechanisms linking certain mutations to ethnicity, delayed onset of the disease and low-penetrance. In considering the variable phenotypes associated to low-penetrance genotypes (see frequencies in Table 4) Richter et al. have proposed that unilateral sporadic carriers of these mutations should be considered the founders of low-penetrance families [23]. The same observation has led to Lohmann et Galli to suggest that the hereditary retinoblastoma has features of a complex trait [7]. In order to clarify these alternatives, functional studies should be carried out in order to provide better insights into the proposed mechanisms for low-penetrance mutations. In addition, gene expression profiling of tumors will help to clarify the genetic background linked to ethnicity and variable expressivity or delayed onset phenotypes. It will also be desirable to build up an international retinoblastoma study group in order to gather high quality information relevant to molecular studies, prognosis, therapy response and long-range follow-up of carriers of low-penetrance mutations.
Table 4 Molecular basis of low penetrance retinoblastoma
Type of mutation Number of LP families Description of mutations and frequency a Functional consequences
Regulatory 3 -198G>A (1/2)
-188G>T (1/2)
-149G>C (1/1) Low expression of normal Rb protein
MS point mutations 12 R661W (8/20)
C712R (2/5)
W563L (1/1)
R787Q (1/1) Partial inactivation of Rb protein
Inframe deletion 3 Del:N480 (1/1)
Del:E04
Del: E24–25 Partial inactivation of Rb protein
Splicing 10 607+1G>T (4/11)
862-10T>C (1/1)
539+1delG, del E05 (1/1)
2325+5G>A, del E:22 (1/1)
Q443P/del:E13 (1/1)
V654L/splice (1/2)
E732E/del E:21(1/2) Alternative splicing and/or unessential exon skipping resulting in low expression or partial inactivation of Rb protein
NS point mutation 1 Q675X (1/1) Alternative splicing involving the stop codon
aRatio of mutations found in LP families vs. all mutations in the database is shown in brackets
Materials and methods
Data mining
Primary bibliographic resources were retrieved from Entrez-PubMed, searching for human retinoblastoma (RB1) mutations. Reprints of all these articles were obtained and additional articles describing RB1 mutations were picked up from its reference lists. In addition, all the articles from the same authors or research group were thoroughly scrutinized in order to avoid repetitions of mutations present in one patient. In all, data from 68 research articles have been compiled in the data base. These articles, together with the number of mutation they contributed and the country of origin of the main research group are supplied in the Additional file 1.
Description of the flat-file format
All the mutations were thoroughly revised according the recommended nomenclature for sequence variations [40] using the genomic sequence GenBank: L11910.1, the cDNA sequence NCBI: NM_000321.1, and the protein sequence NCBI: NP_000312.1. The mutations were annotated in a Microsoft Excel data sheet containing 15 columns (Shown in Additional File 6).
Structure and management of the database
To facilitate access to the collected data, we have created an SQL database and developed an easy to use web interface that provides for simple and complex queries to the database, allowing to sort results by any field and to produce both, HTML and PDF reports. The database engine is based on MySQL [41], a popular, open source DBMS. The user interface is based on HTML forms and PHP [42] scripts, with some bits of public Javascript code [43] for data validation and online help. PDF output is generated by using the FPDF library [44]. The whole code of the web-based user interface and SQL schema is publicly available from the site.
The public interface of RBGMdb is located at EMBnet Spanish node [45]. As we have already mentioned, it provides resources for issuing simple queries for records containing a search term, as well as for complex queries where up to four search terms may be looked up on user-specified fields each combined with its logical operator. Searches produce their results as an HTML table with online help that is displayed whenever the cursor moves over any field. Results may be further sorted by any field by just clicking on the field name in the table header. The user is offered the possibility of generating PDF reports for download and/or printing at every step of the process. In addition to the search forms, we have created a submission form to facilitate the public addition of new data. This form sends an e-mail to the database coordinator who effectively acts as supervisor for all new additions. Restricted access database update forms are also available for the database coordinator(s) to actually modify the data in the database.
Computer analysis and statistics
When indicated, normal and mutated RB1 sequences were analyzed with the programs for exon identification available at the Computing Services of UK HUMAN GENOME MAPPING PROJECT [46]. Statistical significance tests used in this study included Chi square and Fisher exact test for 2 × 2 tables, and the Welch's unpaired t test and nonparametric Mann-Whitney test for the analysis of means. Statistical summaries from the Human Gene Mutation Database (HGMD) were retrieved from the web.
List of abbreviations used
RBGMdb (Retinoblastoma gene mutation database), LP (low-penetrance), NS (nonsense), MS (missense), FS (frameshift), SP (splicing),
Authors' contributions
JRV designed and constructed the database; JA and IP helped in the acquisition of data and the revision of the description of mutations; AP designed and coordinated the study, and drafted the article
Supplementary Material
Additional File 1
List of contributing authors to RBGMdb with indication of country of origin of the main research group, number of mutations and reference.
Click here for file
Additional File 2
Table of recurrent mutations with indication of sequence motifs and methylation status.
Click here for file
Additional File 3
Table with the distribution of mutations by country of origin of the patients. Unless otherwise stated in the original publication, it is assumed that the country of origin of the patients correspond to the country of the main authoring group.
Click here for file
Additional File 4
Table listing mutations found in low-penetrance families
Click here for file
Additional File 5
Possible alternative splicing in the low-penetrance mutation c.607+1G>T
Click here for file
Additional File 6
Description of the flat-file format of RBGMdb.
Click here for file
==== Refs
Zucker JM Desjardins L Doz F Retinoblastoma Eur J Cancer 1998 34 1045 1049 9849453 10.1016/S0959-8049(98)00181-6
Vogt F Genetics of retinoblastoma Hum Genet 1979 52 1 54 393614
Cavenee WK Dryja TP Phillips RA Benedict WF Godbout R Gallie BL Murphree AL Strong LC White RL Expression of recessive alleles by chromosomal mechanism in retinoblastoma Nature 1983 305 779 784 6633649 10.1038/305779a0
Kaelin WG Functions of the retinoblastoma protein Bioessays 1999 21 950 958 10517868 10.1002/(SICI)1521-1878(199911)21:11<950::AID-BIES7>3.0.CO;2-D
Gallie BL Campbell CH Devlin H Duckett A Squire JA Developmental basis of Retinal-specific induction of cancer by RB mutation Cancer Res 1999 59 1731s 1735s 10197588
Chen D Livne-bar T Vanderluit JL Slack RS Agochiya M Bremner R Cell-specific effects of Rb or RB/p107 loss on retinal development implicate an intrinsically death-resistant cell-of-origin in retinoblastoma Cancer Cell 2004 5 539 551 15193257 10.1016/j.ccr.2004.05.025
Lohmann DR Gallie BL Retinoblastoma: revisiting the model prototype of inherited cancer Am J Med Gene Part C 2004 129C 23 28 10.1002/ajmg.c.30024
Knudson AG Mutation and cancer. Statistical study of retinoblastoma Proc Natl Acad Sci USA 1971 68 820 823 5279523
Draper GJ Sanders PA Brownbill Hawkins MM Patterns of risk of hereditary retinoblastoma and applications to genetic counseling Br J Cancer 1992 66 211 219 1637670
Murphree AL Molecular genetics of retinoblastoma Ophtalmol Clin North America 1995 8 155 166
Moll AC Imhof SM Schouten-Van Meeteren AYN Kuik DJ Hofman P Boers M Second primary tumors in hereditary retinoblastoma: A register-based study, 1945–1997 Ophthalmology 2001 108 1109 1114 11382638 10.1016/S0161-6420(01)00562-0
Fletcher O Easton D Anderson K Gilham C Jay M Peto J Lifetime risks of common cancers among retinoblastoma survivors J Natl Cancer Inst 2004 96 357 363 14996857
Harbour JW Overview of RB gene mutations in patients with retinoblastoma Ophthalmology 1998 105 1442 1447 9709755 10.1016/S0161-6420(98)98025-3
Lohmann DR RB1 gene mutations in retinoblastoma Hum Mutat 1999 14 283 288 Link: RB1 gene mutation database 10502774 10.1002/(SICI)1098-1004(199910)14:4<283::AID-HUMU2>3.0.CO;2-J
Stenson PD Ball EV Mort M Phillips AD Shield JA Thomas NS Abeysinghe S Krawczak M Cooper DN Human Gene Mutation Database (HGMD): 2003 update Human Mut 2003 21 577 578 Link: Human Gene Mutation Database-Statistics 10.1002/humu.10212
Cooper DN Krawczak M Cytosine methylation and the fate of CpG dinucleotides in vertebrate genomes Hum Genet 1989 78 151 155 3338800 10.1007/BF00278187
Mancini D Singh S Ainsworth P Rodenhiser D Constitutively methylated CpG nucleotides as mutation hot spots in the retinoblastoma gene Am J Hum Genet 1997 61 80 87 9245987
Lohmann DR Brandt B Hopping W Passarge E Horsthemke B The spectrum of RB1 germ-line mutations in hereditary retinoblastoma Am J Hum Genet 1996 58 940 9 8651278
Lee J-O Russo AA Pavletich NP Structure of the retinoblastoma tumour-suppressor pocket domain bound to a peptide from HPV E7 Nature 1998 391 859 865 9495340 10.1038/36038
Ollila J Lappalainen I Vihinen M Sequence specificity in CpG mutation hotspots FEBS Lett 1996 396 119 122 8914970 10.1016/0014-5793(96)01075-7
Maki H Origins of spontaneous mutations: Specificity and directionality of base-substitutions, frameshift, and sequence-substitution mutagenesis Annu Rev Genet 2002 36 279 303 12429694 10.1146/annurev.genet.36.042602.094806
Krawczak M Ball EV Cooper DN Neighboring-nucleotide effects on the rates of germ-line single-base-pair substitutions in human genes Am J Hum Genet 1998 63 474 488 9683596 10.1086/301965
Richter S Vandezande K Chen N Zhang K Sutherland J Anderson J Han L Panton R Branco P Gallie B Sensitive and efficient detection of RB1 gene mutations enhances care for families with retinoblastoma Am J Hum Genet 2003 72 253 69 12541220 10.1086/345651
Houdayer C Gauthier-Villars M Lauge A Pages-Berhouet S Dehainault C Caux-Moncoutier V Karczynski P Tosi M Doz F Desjardins L Couturier J Stoppa-Lyonnet D Comprehensive screening for constitutional RB1 mutations by DHPLC and QMPSF Hum Mutat 2004 23 193 202 14722923 10.1002/humu.10303
Kolodner RD Marsischky GT Eukaryotic DNA mismatch repair Curr Opin Dev Biol 1999 9 89 96 10.1016/S0959-437X(99)80013-6
Lengauer C Kinzler KW Vogelstein B Genetic instabilities in human cancers Nature 1998 396 643 649 9872311 10.1038/25292
Alonso J Mendiola M García-Miguel P Abelairas J Sarret E Vendrell T Navajas A Pestaña A Spectrum of germline RB1 gene mutations in Spanish retinoblastoma patients. Phenotypic and molecular epidemiological implications Hum Mutat 2001 17 412 422 11317357 10.1002/humu.1117
Harbour JW Molecular basis of low-penetrance retinoblastoma Arch Ophthalmol 2001 119 1699 1704 11709023
Sakai T Ohtani N McGee TL Robbins PD Dryja TP Oncogenic germ-line mutations in Sp1 and ATF sites in the human retinoblastoma gene Nature 1991 353 83 6 1881452 10.1038/353083a0
Cowell JK Bia B Akoulitchev A A novel mutation in the promotor region in a family with a mild form of retinoblastoma indicates the location of a new regulatory domain for the RB1 gene Oncogene 1996 12 431 436 8570221
Otterson GA Modi S Nguyen K Coxon AB Kaye FJ Temperature-sensitive RB mutations linked to incomplete penetrance of familial retinoblastoma in 12 families Am J Hum Genet 1999 65 1040 1046 10486322 10.1086/302581
Dryja TP Rapaport J McGee TL Nork TM Schwartz TL Molecular etiology of low-penetrance retinoblastoma in two pedigrees Am J Hum Genet 1993 52 1122 8 8099255
Bremner R Du DC Connolly-Wilson MJ Bridge P Ahmad KF Mostachfi H Rushlow D Dunn JM Gallie BL Deletion of RB exons 24 and 25 causes low-penetrance retinoblastoma Am J Hum Genet 1997 61 556 70 9326321
Schubert EL Strong LC A splicing mutation in RB1 in low penetrance retinoblastoma Hum Genet 1997 100 557 563 9341870 10.1007/s004390050551
Lefevre SH Chauveinc L Stoppa-Lyonnet D Michon J Lumbroso L Berthet P Frappaz D Dutrillaux B Chevillard S Malfoy B A T to C mutation in the polypyrimidine tract of the exon 9 splicing site of the RB1 gene responsible for low penetrance hereditary retinoblastoma J Med Genet 2002 39 E21 12011162 10.1136/jmg.39.5.e21
Weir-Thompson E Condie A Leonard RC Prosser J A familial RB1 mutation detected by the HOT technique is homozygous in a second primary neoplasm Oncogene 1991 6 2353 2356 1662795
Shapiro MB Senepathy P RNA splice junctions of different classes of eukaryotes: sequence statistics and functional implications in gene expression Nucl Acid Res 1987 15 7155 7177
Klutz M Brockmann D Lohmann DR A parent-of-origin effect in two families with retinoblastoma is associated with a distinct splice mutation in the RB1 gene Am J Hum Genet 2002 71 174 179 12016586 10.1086/341284
Onadim Z Hogg A Baird PN Cowell JK Oncogenic point mutations in exon 20 of the RB1 gene in families showing incomplete penetrance and mild expression of the retinoblastoma phenotype Proc Natl Acad Sci U S A 1992 89 6177 6181 1352883
Dunnen JT Antonarakis SE Mutation nomenclature estensions and suggestions to describe complex mutations: a discussion Hum Mutat 2000 15 7 12 10612815 10.1002/(SICI)1098-1004(200001)15:1<7::AID-HUMU4>3.0.CO;2-N
MySQL
PHP
JavaScript Kit
FPDF Library
Retinoblastoma gene mutation database (RBGMdb)
UK Human Genome Mapping Project
Watakabe A Tanaka K Shimura Y The role of exon sequences in splice site selection Genes Dev 1993 7 407 18 8449402
|
16269091
|
PMC1298292
|
CC BY
|
2021-01-04 16:30:39
|
no
|
BMC Genet. 2005 Nov 4; 6:53
|
utf-8
|
BMC Genet
| 2,005 |
10.1186/1471-2156-6-53
|
oa_comm
|
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-1581628392810.1186/1471-2164-6-158Research ArticleLarge-scale genetic variation of the symbiosis-required megaplasmid pSymA revealed by comparative genomic analysis of Sinorhizobium meliloti natural strains Giuntini Elisa [email protected] Alessio [email protected] Filippo Carlotta [email protected] Duccio [email protected] Nadia [email protected] Christian R [email protected] Anke [email protected] Marco [email protected] Dipartimento di Biologia Animale e Genetica, Università di Firenze, via Romana 17, I-50125 Firenze, Italy2 Dipartimento di Farmacologia, Università di Firenze, Viale Pieraccini 6, 50139 Firenze, Italy3 Bauer Center for Genomics Research, Harvard University, 7 Divinity Avenue, Cambridge, Massachusetts, 02138, USA4 Department of Organismic and Evolutionary Biology, Harvard University, 16 Divinity Avenue, Cambridge, Massachusetts, 02138, USA5 Lehrstuhl fur Genetik, Universitat Bielefeld, 33594 Bielefeld, Germany2005 10 11 2005 6 158 158 13 6 2005 10 11 2005 Copyright © 2005 Giuntini et al; licensee BioMed Central Ltd.2005Giuntini 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
Sinorhizobium meliloti is a soil bacterium that forms nitrogen-fixing nodules on the roots of leguminous plants such as alfalfa (Medicago sativa). This species occupies different ecological niches, being present as a free-living soil bacterium and as a symbiont of plant root nodules. The genome of the type strain Rm 1021 contains one chromosome and two megaplasmids for a total genome size of 6 Mb. We applied comparative genomic hybridisation (CGH) on an oligonucleotide microarrays to estimate genetic variation at the genomic level in four natural strains, two isolated from Italian agricultural soil and two from desert soil in the Aral Sea region.
Results
From 4.6 to 5.7 percent of the genes showed a pattern of hybridisation concordant with deletion, nucleotide divergence or ORF duplication when compared to the type strain Rm 1021. A large number of these polymorphisms were confirmed by sequencing and Southern blot. A statistically significant fraction of these variable genes was found on the pSymA megaplasmid and grouped in clusters. These variable genes were found to be mainly transposases or genes with unknown function.
Conclusion
The obtained results allow to conclude that the symbiosis-required megaplasmid pSymA can be considered the major hot-spot for intra-specific differentiation in S. meliloti.
==== Body
Background
Environmental bacteria are free-living bacteria colonising soil and water. Most of these species are involved in key steps of the biogeochemical cycles of elements such as nitrogen, sulphur, iron, phosphorus and carbon [1]. One of the genomic features of environmental bacteria, and particularly of those belonging to the α-proteobacteria subdivision, is the presence of large genomes of several megabases, consisting of many replicons of similar size, whereas pathogenic and parasitic bacterial genomes often consist of a single replicon. In particular, many of the symbiotic nitrogen-fixing bacteria are characterised by the presence of multiple megaplasmids [2]. In an evolutionary perspective, plasmids have been shown to contribute to symbiosis, pathogenesis and colonisation of new environments, providing resistance to antibiotics or the ability to use specific carbon sources [3-5]. Because megaplasmids can be as large as bacterial genomes and are often not conjugative, their evolutionary dynamics may be closer to that of a real chromosome [2]. Therefore, the role of such megaplasmids in adaptation and consequently their genomic dynamics in the bacterial species is particularly intriguing in the perspective of complex, multi-replicon genome evolution.
Comparative genomic hybridisation (CGH) is a powerful methodology which relies on microarray genome-wide comparison of DNA from different organisms or cells [6-9]. In the field of microbiology, where the number of sequenced species is over 200, CGH has been applied to investigate genomic variation in a certain number of bacterial strains, mainly human pathogens, in order to relate genomic feature to virulence and host adaptation [10-24]. These studies showed that the main sources of variation within bacterial genomes were often duplications or deletions of large DNA fragments. Up to now, most of these studies were performed on species whose genome consist of one replicon and therefore very limited information is available about the genome-scale polymorphism in bacterial species with complex multi-replicon genomes [23]. Here we address this issue in the bacterium Sinorhizobium meliloti.
Sinorhizobium meliloti is a soil bacterium that forms nitrogen-fixing nodules on the roots of leguminous plants such alfalfa (Medicago sativa). It belongs to the Rhizobiales group of the α-Proteobacteria subdivision, together with important human pathogens such as Bartonella and Brucella, and with several plant-associated bacteria of major agricultural importance, such as Agrobacterium, Ochrobactrum, Bradyrhizobium, Mesorhizobium and Rhizobium [2]. S. meliloti is distributed world-wide and is present in many soil types, both in association with legumes or in a free-living form [25]. This species is a model species to study plant-bacteria interactions and in particular legume-rhizobia symbiosis and symbiotic nitrogen-fixation. Its genome contains 6206 ORFs distributed in three replicons, one chromosome of 3.6 Mbp and two megaplasmids, 1.3 Mbp and 1.7 Mbp in size [26-30]. The smallest of the megaplasmids, called either pSymA, pNod-Nif, or pRmeSU47a, contains 1293 ORFs, including many of the genes involved in root nodule formation (nod) and nitrogen fixation (nif) [28,31,32]. The other megaplasmid, pSymB, contains 1570 ORFs and carries genes encoding solute uptake systems, genes involved in polysaccharide biosynthesis and in catabolic activities [29]. Finally, most of 3342 predicted ORFs of the chromosome code for proteins involved in transport and degradation of amino-acids and peptides, as well as sugar metabolism [30].
Previous studies using molecular markers showed that natural populations of rhizobia, and in particular of S. meliloti, exhibit high levels of genetic polymorphism [33-38]. These natural strains also harbour a high number of different mobile genetic elements such as insertion sequences (IS), transposons and bacterial mobile introns [39-41]. However, which functional genes are variable in natural populations contributing to ecological adaptations remains to be fully investigated. Moreover, how the evolutionary dynamics of the diverse replicons differ is still unknown.
To address these questions, genomic DNA of four strains of S. meliloti, previously isolated from agricultural Italian soil and from soil around the Aral Sea region, were compared with the sequenced laboratory strain Rm1021 on a full-genome S. meliloti microarray [42].
Results
Overall results
Four strains, two isolates from soils in the Aral Sea region and two from Northern Italy soil, were compared by whole genome hybridisation with type strain Rm 1021. Four slides with three copies of each ORF were used for each comparison and the results were analysed as described in Methods. Genes were considered to be variable if a statistically significant difference (P < 0.001) in hybridisation intensity was detected between the type and the strain under comparison. The fraction of variable genes detected with comparative genomic hybridisation (CGH) on the microarray containing oligonucleotide probes for all currently predicted protein-coding genes of strain Rm1021, ranged from 4.6 to 6.5% (Table 2). In particular, strain BL225C showed the highest number of variable genes (401), while strain AK58 displayed the lowest number (287). The majority of variable genes (77–94%) showed decreased hybridisation intensity (Log2-ratio > 0) of the natural isolate versus the Rm 1021 strain, suggesting deletion or nucleotide divergence in the region covered by the oligonucleotide. The remaining fraction of variable genes, (6–23%, with a Log2-ratio < 0), showed an increased hybridisation signal of the natural strains compared to Rm 1021, suggestive of gene duplication (Table 2).
Table 2 Genes variable in each strain compared to strain Rm 1021
Strain Log2-ratio > 0 Log2-ratio < 0 Total genes variable Total genes analysed* % of variable genes
AK58 237 50 287 6192 4.6%
AK83 273 80 353 6199 5.7%
BL225C 379 22 401 6181 6.5%
BO21CC 292 56 348 5670 6.1%
*The total number of genes analysed varies according to the number of spots discarded due to technical defects. The list of variable genes is reported as Additional Material '[see Additional file 1]'.
In order to corroborate the results of the microarray hybridisation analysis, we randomly selected 116 ORFs, with 66 of these being included in the variable ones (P < 0.001) and 52 found to show no significant difference compared to the type strain (table 3).
Table 3 Experimental analysis of 118 genes from microarray hybridisation
P-value classes N° of ORFs Total ORFs amplified by PCR* N° of ORFs positive to PCR amplification N° of ORFs analysed by Southern blotting N° of ORFs with duplication after Southern hybridisation** N° of sequenced ORFs N° of ORFs with nucleotide variation in the 70-mer oligo sequence
p < 0.001 66 log2-ratio < 0 19 19 7 7 - -
log2-ratio > 0 47 8 - - 8 8
p > 0.001 52 log2-ratio < 0 21 21 4 0 - -
log2-ratio > 0 31 31 - - 7 0
* PCR amplification was carried out with primer anchored to the flanking regions of the gene.
** Determined as higher number of observed fragments after Southern-blot analysis of Rm1021 and of the wild strain.
The 66 genes showing differential hybridisation (Table 3) were PCR amplified from genomic DNA of both strain Rm1021 and the natural isolate showing the difference. The 19 ORFs with a Log2-ratio < 0 selected (suggesting gene duplication) showed positive amplification. For 7 of these 19 ORFs, Southern hybridisations were carried out on restricted DNA of both tested and reference strains. All 7 ORFs showed more than one band in the DNA of wild strain compared to the single band of strain Rm 1021, confirming that the higher intensity of the microarray hybridisation of the wild strain was indeed due to a duplication of the ORF. Of the 47 ORFs with a Log2-ratio > 0 (indicating gene deletion or divergence), 39 gave no amplification in the wild strain, confirming the microarray result that suggested that the ORF was deleted in this strain. Eight ORFs, on the contrary, were amplified both in the wild strain and in strain Rm1021. These ORFs were sequenced and showed the occurrence of nucleotide variations in the DNA of the wild strain within the region covered by the 70-mer oligonucleotide microarray probe. In that latter case, the lower level of microarray hybridisation of the wild strains was attributed to the mismatches between the genome sequence of the natural isolate and the 70-mer probe sequence. These data confirmed the assumption made for the interpretation of the results.
We also amplified 52 ORFs randomly selected from those with a low level of probability to be variable (Table 3). All of them showed amplification from DNA of strain Rm 1021 and from DNA of the wild strain. Four of these were further analysed by Southern hybridisation showing no sign of copy number variation. Seven from the 31 ORFs with a Log2-ratio < 0 but with P > 0.001 were sequenced and no nucleotide polymorphism was observed.
Variable genes are mainly localised on pSymA
The genes that were found to be variable in the comparison between the type strain Rm 1021 and the four natural isolates were not randomly distributed among the three replicons. A highly significant enrichment (probability of observing this proportion <0.0001) for pSymA was found in all the strains, both for duplicated and deleted/mutated genes (Figure 1). pSymB was also found to be enriched for duplicated ORFs, though not as significantly as pSymA, except in strain AK58 where pSymB was significantly enriched for duplicated ORFs (probability of observing this proportion <0.0001).
Figure 1 Number and location of variable ORFs on the three replicons. Genes considered were significantly different in hybridisation from strain Rm1021 at p < 0.001. A), Genes with Log2-ratio > 0; B) genes with Log2-ratio < 0. Asterisks over the columns indicate significant enrichment at p < 0.0001.
Within the replicons, the variable ORFs had a significant tendency to be spatially clustered (runs-test). In particular, in the pSymA plasmid, one region appeared to be duplicated in all natural strains. This region of at least 1000 bp includes two genes located upstream of the nodD2 gene. These genes encode the transcription factors SMa0748 and SMa0750 of the putative MucR/LysR-families. This duplication was confirmed by Southern-blot analysis on DNA extracted from strain AK83 (not shown). Among the putative deleted/mutated genes, several clusters were also identified (Figure 2)
Figure 2 Location of variable ORFs along the replicons. Up and down bars indicate ORFs with Log2-ratio > 0 (duplication) or Log2-ratio < 0 (divergence or deletion), respectively. Thickness of bars indicates clusters of variable genes. A, AK58 strain; B, AK83 strain; C, BL225C strain; D, BO21CC strain. Replicon lengths are not in scale.
Functional groups of variable genes
Figure 3 reports the proportion of functional groups among the variable ORFs using the biological classification as defined by the S. meliloti consortium. The most frequently affected functional categories in all strains were, within the Elements of External Origin, Transposases (V.A) and, within Miscellaneous, Unknown Function ORFs (VI.D). These categories were found to be statistically significantly enriched with variable genes (see Methods).
Figure 3 Functional groups of variable ORFs. A, strain AK58; B, strain AK83; C, strain BL225C; D, strain BO21CC. Classification is as defined by the S. meliloti consortium, subgroups are not reported: I, Small molecule metabolism; II, Macromolecule metabolism; III, Structural elements; IV, Cell processes; V, Elements of external origin; VI, Miscellaneous/unknown function. Groups V and VI were statistically significantly enriched for all strains.
Discussion
The alpha-proteobacteria display diverse life-styles. In particular, they keep close relationships with the eukaryotic cell, a trait that is possibly linked to the presence in their genomes of multiple replicons [2]. In the case of the symbiotic species Sinorhizobium meliloti, there are three replicons, a 3.6 Mbp circular chromosome and two megaplasmids 1.3 Mbp and 1.7 Mbp in size [31,32,29].
The four strains of S. meliloti, whose genome have been compared in the present work with that of the type strain Rm1021, exhibited similar proportions of genes that differ in presence, nucleotide polymorphism or copy number from the type strain. The Italian strain BL225C showed the highest number of altered ORFs, while the Aral Sea strain AK58 displayed the lowest one. The overall results indicate that in the multi-replicon genome of S. meliloti, a fraction accounting for 4.6–6.5% of all ORFs were variable in the natural strains compared to the sequenced laboratory strain Rm1021. In particular, most of the variation was due to gene losses or to nucleotide divergence (Log2-ratio > 0), while a smaller fraction of the variation could be attributed to gene duplication (Log2-ratio < 0). These values are similar to those obtained from other studies using DNA microarrays for CGH on Camplylobacter jejuni, and Staphylococcus aureus [11,21], and are lower than those observed in human pathogens such as Helicobacter pylori, in which ~22% of ORFs were found to be variable [8]. However, in other Rhizobiales, such as Brucella, a similar value of gene diversity (around 4%) was found in an inter-specific analysis [23]. Of course this is a minimum amount of variation because of the unsurveyed parts of each ORF, the variation in intergenic region or new genes that are not present in the lab strain, and we therefore present a conservative estimate of genetic variation in natural strains. The variable ORFs were found to be unevenly distributed in the three replicons. The megasplasmid pSymA carried most of the variable genes. This replicon harbours nod genes, which are required for establishing the symbiotic relationship with host plants; nif genes, for nitrogen-fixation, and genes putatively involved in nitrogen and carbon metabolism and transport, as well as in stress and resistance responses, all functions intimately related to S. meliloti's ecological niche [28]. Actually, the detected variable genes were found to be mainly distributed in clusters along the replicons of Rm1021. In particular for pSymA, a duplication region common to all the wild strains was found just near the nodD2 gene, the transcriptional activator of nodulation cascade (SMa0748-SMa0752). Moreover, pSymA contains the highest percentage of mobile genetic elements among the three S. meliloti replicons (3.6% for pSymA, 0.9% pSymB, 2.6% chromosome). Transposases and other related functions were particularly frequent among the variable genes. Transposable elements tend to accumulate in chromosomal regions where they do not disrupt essential cellular functions [43]. The largest amount of genetic polymorphism observed in pSymA is therefore consistent with the observation that pSymA is not essential for cell survival. Indeed, pSymA can be cured from some S. meliloti strains, such as Rm2011, without affecting growth in either rich or minimal-succinate media, but the cured strain is defective in the utilisation of certain carbon sources [44]. Furthermore, the analysis of the complete genome sequence of S. meliloti suggests that pSymA could be of foreign origin because of its lower G+C content (60.4%) and its distinct codon usage [45] compared to the other replicons. An enrichment for variable genes was also found for the pSymB megaplasmid, but only in the strain AK58. From the genomic point of view pSymB shows many features of a typical chromosome [45], carries several genes for carbohydrate metabolism and is thought to be of high adaptive value for the colonisation of soil and rhizosphere environments [2]. Since pSymA has not be mapped on the tested strains, some of the genes that hybridize on the microarray (derived from strain Rm1021) could be actually located on other replicons in the natural strains.
All functional categories of genes were represented within the variable ones (e.g., Small molecule metabolism, Macromolecule metabolism, Structural elements, Cell processes, Elements of external origin and, Miscellaneous/unknown function) with some gene found to be variable in more than one strain. However, among the different categories, the variable genes appeared to be distributed as theoretically expected from the numerical consistency of all but the last two categories. Actually, "Elements of external origin" and "Miscellaneous/unknown function" were significantly enriched in variable genes because of the large number of transposases and unknown function ORFs found to be variable. The presence of such a large proportion of unknown function genes among the polymorphic ones in natural isolates raises interesting hypotheses regarding the diversification of S. meliloti strains. Barnett and collaborators [28] using transcriptional profiling showed that a certain number of unknown function genes were found to be expressed below the detection threshold in both free-living culture and nodulation conditions. Several of these genes (21%, data not shown) were found in our analysis to be among the deleted ones, suggesting that they may represent pseudogenes, non-coding sequences or more interestingly, genes expressed only in very specific conditions.
Conclusion
Using DNA microarray technology, we assessed genetic variation of the coding regions of 4 natural strains of S. meliloti. We found that most of the genetic differences accumulate on the symbiosis-required megaplasmid pSymA, which consequently can be considered the major hot-spot for intra-specific differentiation in S. meliloti.
Methods
Bacterial strains, microbiological media and DNA extraction
S. meliloti AK58 and AK83 (Table 1) are a part of alfalfa nodulating rhizobia collected by RIAM (St. Petersburg, Russia) and were trapped from soil samples collected in the Northern Aral Sea Region during May 2001 by M. falcata. Isolates BO21CC and BL225C, from Lodi, Italy, were trapped on M. sativa [34]. Rhizobia were cultured at 30°C in liquid TY medium (Tryptone 5 g/l, Yeast extract 3 g/l, CaCl2 0.4 g/l). DNA was extracted with the FastDNA Kit (Bio 101, Inc.) according to the manufacturer's instructions. Extracted DNA was quantified by spectrophotometric reading (Biophotometer, Eppendorf).
Table 1 Bacterial strains used in this study
Strain Species Geographical origin Host plant of isolation
Rm 1021 S. meliloti Galibert et al. 2001 Sequenced strain
AK83 S. meliloti North Aral Sea, Kazakhstan Medicago falcata
AK58 S. meliloti North Aral Sea, Kazakhstan Medicago falcata
BL225C S. meliloti Lodi, Italy Medicago sativa
BO21CC S. meliloti Lodi, Italy Medicago sativa
PCR, Southern blot analysis and sequencing
PCR amplification reactions were performed with a Primus 96 Thermal Cycler (MWG-AG Biotech) in a 50 μl total volume with 30 ng of extracted DNA as template and contained 5 μl of 10× reaction buffer (Polytaq, Polymed, Italy), 1.5 mM MgCl2, 0.2 mM of each dNTP, 1 U of Taq DNA polymerase (Polytaq, Polymed, Italy), 10 pmols of each primer. The cycling conditions were as follows: after incubation at 95°C for 2 min, samples were cycled for 35 cycles through the following temperature profile: denaturation at 94°C for 30 sec, annealing at 57°C for 30 sec, extension at 72°C for 2 min. Finally, the mixtures were incubated at 72°C for 5 min. Then, 5 μl of each amplification mixture were analysed by agarose gel (1.2% w/v) electrophoresis in TAE buffer containing 1 μg/ml (w/v) of ethidium bromide. Southern blot analysis was performed with 1 microgram of total DNA, digested overnight at 37°C with the restriction enzymes XhoI, EcoRI or PvuII, and electrophoresed for 3 h on a 0.7% agarose gel in TAE buffer with a DIG-labelled DNA marker II (Roche). DNA was blotted on a nylon membrane (Amersham). The cDNA probe preparation, the hybridisation and detection conditions were as described previously in Biondi et al. [40]. Automated DNA sequencing was performed directly from the primers used for the amplification on the purified PCR products using the BigDye Terminator v.1.1 chemistry and an ABI310 sequencer (PE-Applied Biosystems) according to the manufacturer's recommendations.
Hybridisation and microarray scanning
Microarray slides were printed by the Center for Biotechnology, University of Bielefeld [42]. Microarrays contained 6208 70 mer oligonucleotides directed against protein-coding ORFs of S. meliloti 1021, four 70 mer oligonucleotides directed against transgenes (gusA, lacZ, nptII, aacC1), two 70 mer stringency control oligonucleotides (80% identity), 12 alien 70 mer oligonucleotides and three alien DNA fragments (Stratagene) that can be used as spiking controls. Each microarray slide contained 6.229 triplicate spots in 48 grids of 20 rows and 21 columns. The 48 grids were arrayed in a 4 × 12 pattern of 4 metacolumns and 12 metarows. Alien oligonucleotides and 12 "housekeeping" genes were arrayed in 13 additional replicates. Oligonucleotides directed against the S. meliloti 1021 genome and the alien oligonucleotide controls were taken from the Sinorhizobium meliloti Array Ready Oligo Set Version 1.0 (Qiagen).
Genomic DNA was labelled with FluoroLink Cy3- or Cy5-dCTP (Amersham Biosciences, Milano, Italy) by using the method described by Pollack et al. [46] and the components of the BioPrime DNA labeling system (Invitrogen, Milano, Italy). Two micrograms of each restriction enzyme (TaqI and MspI) digested genomic DNA was labelled by using 20 μl of the 2.5X Random Primer, 40 U of the Klenow fragment, and 3 μl of the Cy5-dCTP or Cy3-dCTP (1 mM stocks) at 37°C for 2 h. Unincorporated fluorescent nucleotides were removed by using Microcon 30 filter columns (Millipore, Milano, Italy). The appropriate Cy5 and Cy3 labelled probes were combined and mixed with 30 μl Cot-1 DNA (1 mg/ml), 20 μl Yeast t-RNA (5 mg/ml), 450 μl TE to concentrate the samples until about 40 μl using Microcon 30 filter columns (Millipore, Milano, Italy). To each combined sample 8.5 μl of 20 × SSC and 0.74 μl of 10% SDS were added. The sample was denatured to 100°C for 1.5 min, and then incubated for 37°C for 30 min. The hybridisation probe was added to the microarray under a coverslip, and hybridisation was performed at 65°C for 16 h. Slides were washed at 60°C with 2 × SSC for 5 min and then at 60°C with 0.2 × SSC containing 0.1% SDS for 5 min and finally at room temperature with 0.2 × SSC for 2 min. The last step was conducted twice. The slides were immediately dried and scanned for fluorescence intensity by using a GenePix 4000B microarray scanner (Axon Instruments, Union City, CA), and the results were recorded in 16-bit multi-image TIFF files. Competitive hybridisation was done twice for one strain. In the first experiment, the Rm1021 reference DNA and the sample DNA from natural strain were labelled with Cy3 and Cy5, respectively. In the second hybridisation, the dyes for labelling were swapped.
For each sample a total of four slides were hybridised (after dye swapping of the two different restriction enzyme DNA preparations); considering that one slide carries three replicas of each ORF, any sample was hybridized twelve times at each ORF.
Normalisation and significant hybridisation differences
Raw data from Genepix was imported into R (1.9) [47] and analysed using the LIMMA library (Linear Models for Microarray Data version 1.7, [48]). Spots showing hybridisation intensity two standard deviations above background intensity and that were not flagged as bad were used in normalisation and model fitting. For unknown reasons, strain BO21CC showed a lower number of analysable ORFs (see Table 2) as the quality of slides was apparently comparable. A within-array loess normalisation of intensities was applied. A gene was considered to have a statistically significant differences in hybridisation (moderated t-statistics using empirical Bayes shrinkage of the standard errors) when 2 of the 3 spots on the array representing that gene had a p-value lower than p < 0.001. This stringent cut-off allows preventing false positive. This analysis was designed such that positive log2 fold change occurred when hybridisation was higher in the Rm1021 strain. Such a result is indicative of sequence divergence/gene loss in other strain compared to Rm1021. Negative Log2 fold change occurred when more hybridisation was detected in the other strain competitively hybridised with strain Rm1021. Such a fold change pattern is indicative of gene duplication in the other strain tested compared to Rm1021.
Physical genome location
We estimated if deleted and duplicated genes in each strain were found significantly more frequently on a given replicon. We calculated the proportion of genes associated with chromosome, megaplasmid pSymA and megaplasmid pSymB in the whole genome and then the same proportion in the significantly divergent and duplicated gene lists (p < 0.001). The hypergeometric distribution was used to calculate the probability of observing this proportion of variable genes for each replicon in comparison to their total number of genes. A Bonferroni correction was applied to adjust the cut-off probability at which a replicon is considered significantly enriched for variable genes. We multiplied the p-value by the number of tests performed and considered a replicon to be significantly enriched if this adjusted probability was below 5%.
Spatial clustering within a replicon
Genes were binned as 1 or 0 respectively if they were differentially hybridising or not. They were then ordered according to their position along the replicons and the distribution of 1 and 0 was analysed using a runs-test [49]. This analysis tests the null hypothesis that successes in a series of binomial trials are randomly distributed. The alternative hypotheses of this test are that successes are spatially clustered or they are more evenly spaced than by chance. Genes identified as being duplicated or diverged were analysed separately.
Functional enrichment analysis
Genes found to have a significant difference in hybridisation at a level of p < 0.001, hereafter referred to as variable genes, were used in a functional enrichment analysis. Each gene has been attributed a biological classification by the "S. meliloti strain Rm 1021 genome project" consortium [45]. We calculated the proportion of genes associated with each biological process in the whole genome and then in the variable gene list for each strain. The hypergeometric distribution was used to calculate the probability of observing this proportion of variable genes for each biological process in a particular strain compared to the representation in the whole genome. A Bonferroni correction for multiple testing was applied to adjust the cut-off probability at which a gene list is considered significantly enriched for a given biological classification. We multiplied the probability of observing the proportion of variable gene in a category by the number of tests performed (dependant on number of functional categories represented) and considered a gene list to be significantly enriched if this adjusted probability was below 0.05.
Authors' contributions
EG carried out the microarray hybridizations, participated in the conceiving and in the design of the experiment and drafted the manuscript. AM carried out most of the Southern-blots and PCR verifications, participated in the conceiving and design of the experiment and drafted the manuscript. CDF and DC contributed in setting the microarray experiment protocol. NA-H and CRL performed the statistical analysis. AB contributed in providing the microarray slides and helped discussing the results. MB conceived the study, participated in its design and coordination and drafted the manuscript.
Supplementary Material
Additional file 1
List of variable ORFs. Each column reports the list ORF's names (with p-value < 0.001) for the four different S. meliloti strains with Log2-ratio > 0 or <0.
Click here for file
Acknowledgements
This work has been funded by grants M.I.U.R. (F.I.R.B., "Post Genomica di Leguminose Foraggere")
==== Refs
Varnam AH Evans M Environmental Microbiology 2000 New York, Blackwell-Publishing
Teyssier C Marchandin H Jumas-Bilak E Related Articles, The genome of alpha-proteobacteria: complexity, reduction, diversity and fluidity Can J Microbiol 2004 50 383 96 15284884 10.1139/w04-033
Ziebuhr W Ohlsen K Karch H Korhonen T Hacker J Evolution of bacterial pathogenesis Cell Mol Life Sci 1999 56 719 28 11212331 10.1007/s000180050018
Edwards RA Olsen GJ Maloy SR Comparative genomics of closely related salmonellae Trends Microbiol 2002 10 94 9 11827811 10.1016/S0966-842X(01)02293-4
Leahy JG Colwell RR Microbial degradation of hydrocarbons in the environment Microbiol Rev 1990 54 305 15 2215423
Kallioniemi A Kallioniemi OP Sudar D Rutovitz D Gray JW Waldman F Pinkel D Comparative genomic hybridization for molecular cytogenetic analysis of solid tumors Science 1992 258 818 21 1359641
Behr MA Wilson MA Gill WP Salamon H Schoolnik GK Rane S Small PM Comparative genomics of BCG vaccines by whole-genome DNA microarray Science 1999 284 1520 3 10348738 10.1126/science.284.5419.1520
Salama N Guillemin K McDaniel TK Sherlock G Tompkins L Falkow S A whole-genome microarray reveals genetic diversity among Helicobacter pylori strains Proc Natl Acad Sci U S A 2000 97 14668 73 11121067 10.1073/pnas.97.26.14668
Murray AE Lies D Li G Nealson K Zhou J Tiedje JM DNA/DNA hybridization to microarrays reveals gene-specific differences between closely related microbial genomes Proc Natl Acad Sci U S A 2001 98 9853 8 11493693 10.1073/pnas.171178898
Beres SB Sylva GL Sturdevant DE Granville CN Liu M Ricklefs SM Whitney AR Parkins LD Hoe NP Adams GJ Low DE DeLeo FR McGeer A Musser JM Genome-wide molecular dissection of serotype M3 group A Streptococcus strains causing two epidemics of invasive infections Proc Natl Acad Sci U S A 2004 101 11833 8 15282372 10.1073/pnas.0404163101
Cassat JE Dunman PM McAleese F Murphy E Projan SJ Smeltzer MS Comparative genomics of Staphylococcus aureus musculoskeletal isolates J Bacteriol 2005 187 576 92 15629929 10.1128/JB.187.2.576-592.2005
Broekhuijsen M Larsson P Johansson A Bystrom M Eriksson U Larsson E Prior RG Sjostedt A Titball RW Forsman M Genome-wide DNA microarray analysis of Francisella tularensis strains demonstrates extensive genetic conservation within the species but identifies regions that are unique to the highly virulent F. tularensis subsp. tularensis J Clin Microbiol 2003 41 2924 31 12843022 10.1128/JCM.41.7.2924-2931.2003
Kato-Maeda M Rhee JT Gingeras TR Salamon H Drenkow J Smittipat N Small PM Comparing genomes within the species Mycobacterium tuberculosis Genome Res 2001 11 547 54 Erratum in: Genome Res 2001, 11:1796 11282970 10.1101/gr.166401
Ge H Chuang YY Zhao S Tong M Tsai MH Temenak JJ Richards AL Ching WM Comparative genomics of Rickettsia prowazekii Madrid E and Breinl strains J Bacteriol 2004 186 556 65 14702324 10.1128/JB.186.2.556-565.2004
Fukiya S Mizoguchi H Tobe T Mori H Extensive genomic diversity in pathogenic Escherichia coli and Shigella Strains revealed by comparative genomic hybridization microarray J Bacteriol 2004 186 3911 21 15175305 10.1128/JB.186.12.3911-3921.2004
Zhou D Han Y Dai E Pei D Song Y Zhai J Du Z Wang J Guo Z Yang R Identification of signature genes for rapid and specific characterization of Yersinia pestis Microbiol Immunol 2004 48 263 9 15107536
Cummings CA Brinig MM Lepp PW van de Pas S Relman DA Bordetella species are distinguished by patterns of substantial gene loss and host adaptation J Bacteriol 2004 186 1484 92 14973121 10.1128/JB.186.5.1484-1492.2004
Brunelle BW Nicholson TL Stephens RS Microarray-based genomic surveying of gene polymorphisms in Chlamydia trachomatis Genome Biol 2004 5 R42 15186493 10.1186/gb-2004-5-6-r42
Boyd EF Porwollik S Blackmer F McClelland M Differences in gene content among Salmonella enterica serovar typhi isolates J Clin Microbiol 2003 41 3823 8 12904395 10.1128/JCM.41.8.3823-3828.2003
Taboada EN Acedillo RR Carrillo CD Findlay WA Medeiros DT Mykytczuk OL Roberts MJ Valencia CA Farber JM Nash JH Large-scale comparative genomics meta-analysis of Campylobacter jejuni isolates reveals low level of genome plasticity J Clin Microbiol 2004 42 4566 76 15472310 10.1128/JCM.42.10.4566-4576.2004
Leonard EE 2ndTompkins LS Falkow S Nachamkin I Comparison of Campylobacter jejuni isolates implicated in Guillain-Barre syndrome and strains that cause enteritis by a DNA microarray Infect Immun 2004 72 1199 203 14742576 10.1128/IAI.72.2.1199-1203.2004
Alvarez J Porwollik S Laconcha I Gisakis V Vivanco AB Gonzalez I Echenagusia S Zabala N Blackmer F McClelland M Rementeria A Garaizar J Detection of a Salmonella enterica serovar California strain spreading in spanish feed mills and genetic characterization with DNA microarrays Appl Environ Microbiol 2003 69 7531 4 14660409 10.1128/AEM.69.12.7531-7534.2003
Rajashekara G Glasner JD Glover DA Splitter GA Comparative whole-genome hybridization reveals genomic islands in Brucella species J Bacteriol 2004 186 5040 51 15262941 10.1128/JB.186.15.5040-5051.2004
Dorrell N Mangan JA Laing KG Hinds J Linton D Al-Ghusein H Barrell BG Parkhill J Stoker NG Karlyshev AV Butcher PD Wren BW Whole genome comparison of Campylobacter jejuni human isolates using a low-cost microarray reveals extensive genetic diversity Genome Res 2001 11 1706 15 11591647 10.1101/gr.185801
Sadowsky MJ Graham PH Spaink HP, Kondorosi A, Hooykaas PJJ Soil Biology of the Rhizobiaceae The Rhizobiaceae Molecular Biology of Plant Associated Bacteria 1998 Dordrecht, The Netherlands: Kluwer Academic Publishers 155 172
Honeycutt RJ McClelland M Sobral BW Physical map of the genome of Rhizobium meliloti 1021 J Bacteriol 1993 175 6945 52 8226638
Banfalvi Z Kondorosi E Kondorosi A Rhizobium meliloti carries two megaplasmids Plasmid 1985 13 129 38 2987992 10.1016/0147-619X(85)90065-4
Barnett MJ Fisher RF Jones T Komp C Abola AP Barloy-Hubler F Bowser L Capela D Galibert F Gouzy J Gurjal M Hong A Huizar L Hyman RW Kahn D Kahn ML Kalman S Keating DH Palm C Peck MC Surzycki R Wells DH Yeh KC Davis RW Federspiel NA Long SR Nucleotide sequence and predicted functions of the entire Sinorhizobium meliloti pSymA megaplasmid Proc Natl Acad Sci U S A 2001 98 9883 8 11481432 10.1073/pnas.161294798
Finan TM Weidner S Wong K Buhrmester J Chain P Vorholter FJ Hernandez-Lucas I Becker A Cowie A Gouzy J Golding B Puhler A The complete sequence of the 1,683-kb pSymB megaplasmid from the N2-fixing endosymbiont Sinorhizobium meliloti Proc Natl Acad Sci U S A 2001 98 9889 94 11481431 10.1073/pnas.161294698
Capela D Barloy-Hubler F Gouzy J Bothe G Ampe F Batut J Boistard P Becker A Boutry M Cadieu E Dreano S Gloux S Godrie T Goffeau A Kahn D Kiss E Lelaure V Masuy D Pohl T Portetelle D Puhler A Purnelle B Ramsperger U Renard C Thebault P Vandenbol M Weidner S Galibert F Analysis of the chromosome sequence of the legume symbiont Sinorhizobium meliloti strain 1021 Proc Natl Acad Sci U S A 2001 98 9877 82 11481430 10.1073/pnas.161294398
Banfalvi Z Sakanyan V Koncz C Kiss A Dusha I Kondorosi A Location of nodulation and nitrogen fixation genes on a high molecular weight plasmid of R. meliloti Mol Gen Genet 1981 184 318 25 6949000
Rosenberg C Boistard P Denarie J Casse-Delbart F Genes controlling early and late functions in symbiosis are located on a megaplasmid in Rhizobium meliloti Mol Gen Genet 1981 184 326 33 6949001
Biondi EG Pilli E Giuntini E Roumiantseva ML Andronov EE Onichtchouk OP Kurchak ON Simarov BV Dzyubenko NI Mengoni A Bazzicalupo M Genetic relationship of Sinorhizobium meliloti and Sinorhizobium medicae strains isolated from Caucasian region FEMS Microbiol Lett 2003 220 207 13 12670682 10.1016/S0378-1097(03)00098-3
Carelli M Gnocchi S Fancelli S Mengoni A Paffetti D Scotti C Bazzicalupo M Genetic diversity and dynamics of Sinorhizobium meliloti populations nodulating different alfalfa cultivars in Italian soils Appl Environ Microbiol 2000 66 4785 9 11055924 10.1128/AEM.66.11.4785-4789.2000
Roumiantseva ML Andronov EE Sharypova LA Dammann-Kalinowski T Keller M Young JP Simarov BV Diversity of Sinorhizobium meliloti from the Central Asian Alfalfa Gene Center Appl Environ Microbiol 2002 68 4694 7 12200335 10.1128/AEM.68.9.4694-4697.2002
Jebara M Mhamdi R Aouani ME Ghrir R Mars M Genetic diversity of Sinorhizobium populations recovered from different medicago varieties cultivated in Tunisian soils Can J Microbiol 2001 47 139 47 11261493 10.1139/cjm-47-2-139
Paffetti D Scotti C Gnocchi S Fancelli S Bazzicalupo M Genetic diversity of an Italian Rhizobium meliloti population from different Medicago sativa varieties Appl Environ Microbiol 1996 62 2279 85 8779566
Souza V Nguyen TT Hudson RR Pinero D Lenski RE Hierarchical analysis of linkage disequilibrium in Rhizobium populations: evidence for sex? Proc Natl Acad Sci U S A 1992 89 8389 93 1518873
Selbitschka W Arnold W Jording D Kosier B Toro N Puhler A The insertion sequence element ISRm2011-2 belongs to the IS630-Tc1 family of transposable elements and is abundant in Rhizobium meliloti Gene 1995 163 59 64 7557479 10.1016/0378-1119(95)00371-C
Biondi EG Fancelli S Bazzicalupo M ISRm10: a new insertion sequence of Sinorhizobium meliloti: nucleotide sequence and geographic distribution FEMS Microbiol Lett 1999 181 171 6 10564804 10.1016/S0378-1097(99)00534-0
Biondi EG Femia AP Favilli F Bazzicalupo M IS Rm31, a new insertion sequence of the IS 66 family in Sinorhizobium meliloti Arch Microbiol 2003 180 118 26 12819859 10.1007/s00203-003-0568-x
Krol E Becker A Global transcriptional analysis of the phosphate starvation response in Sinorhizobium meliloti strains 1021 and 2011 Mol Genet Genomics 2004 272 1 17 15221452 10.1007/s00438-004-1030-8
Burrus V Waldor MK Shaping bacterial genomes with integrative and conjugative elements Res Microbiol 2004 155 376 86 15207870 10.1016/j.resmic.2004.01.012
Oresnik IJ Liu SL Yost CK Hynes MF Megaplasmid pRme2011a of Sinorhizobium meliloti is not required for viability J Bacteriol 2000 182 3582 6 10852892 10.1128/JB.182.12.3582-3586.2000
Galibert F Finan TM Long SR Puhler A Abola P Ampe F Barloy-Hubler F Barnett MJ Becker A Boistard P Bothe G Boutry M Bowser L Buhrmester J Cadieu E Capela D Chain P Cowie A Davis RW Dreano S Federspiel NA Fisher RF Gloux S Godrie T Goffeau A Golding B Gouzy J Gurjal M Hernandez-Lucas I Hong A Huizar L Hyman RW Jones T Kahn D Kahn ML Kalman S Keating DH Kiss E Komp C Lelaure V Masuy D Palm C Peck MC Pohl TM Portetelle D Purnelle B Ramsperger U Surzycki R Thebault P Vandenbol M Vorholter FJ Weidner S Wells DH Wong K Yeh KC Batut J The composite genome of the legume symbiont Sinorhizobium meliloti Science 2001 293 668 72 11474104
Pollack JR Perou CM Alizadeh AA Eisen MB Pergamenschikov A Williams CF Jeffrey SS Botstein D Brown PO Genome-wide analysis of DNA copy-number changes using cDNA microarrays Nat Genet 1999 23 41 6 10471496 10.1038/14385
R Development Core Team R: A language and environment for statistical computing 2004 R Foundation for Statistical Computing, Vienna, Austria ISBN 3-900051-00-3
Smyth GK Linear models and empirical Bayes methods for assessing differential expression in microarray experiments Statistical Applications in Genetics and Molecular Biology 2004 3 Article 3
Bradley JV Distribution-Free Statistical Tests 1968 New Jersey: Prentice Hall
|
16283928
|
PMC1298293
|
CC BY
|
2021-01-04 16:32:47
|
no
|
BMC Genomics. 2005 Nov 10; 6:158
|
utf-8
|
BMC Genomics
| 2,005 |
10.1186/1471-2164-6-158
|
oa_comm
|
==== Front
BMC GeriatrBMC Geriatrics1471-2318BioMed Central London 1471-2318-5-121627766610.1186/1471-2318-5-12Research ArticleA descriptive study of older adults with persistent pain: Use and perceived effectiveness of pain management strategies [ISRCTN11899548] Kemp Carol A [email protected] Mary [email protected] Judith A [email protected] Pain and Palliative Care Research Department, Swedish Medical Center, 500 17th Ave, Providence Professional Building Suite 405, Seattle, WA 98122-5711, USA2 Department of Biobehavioral Nursing and Health Systems, University of Washington School of Nursing, Box 357266, Seattle, WA 98195-1406, USA3 Department of Psychiatry and Behavioral Sciences, University of Washington School of Medicine, Box 356560, Seattle, WA 98195-6560, USA4 Department of Rehabilitation Medicine, University of Washington School of Medicine, Box 356560, Seattle, WA 98195-6560, USA2005 8 11 2005 5 12 12 8 7 2005 8 11 2005 Copyright © 2005 Kemp et al; licensee BioMed Central Ltd.2005Kemp 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
Persistent pain is a common, often debilitating, problem in older adults; however, few studies have focused on the experiences of older adults in managing their pain. The objective of this study was to describe the use and perceived effectiveness of pain management strategies in a sample of older adults and to explore the associations of these variables with demographic and psychosocial characteristics.
Methods
Adults ≥ 65 years old and living in retirement facilities who reported persistent pain (N = 235, mean age = 82 years, 84% female, 94% white) completed measures of demographics, pain, depression, self-efficacy for managing pain, and a Pain Management Strategies Survey. Participants identified current and previous-year use of 42 pain management strategies and rated helpfulness of each on a 5-point scale.
Results
Acetaminophen, regular exercise, prayer, and heat and cold were the most frequently used pain management strategies (61%, 58%, 53%, and 48%, respectively). Strategies used by >25% of the sample that were rated moderately or more helpful (i.e., >2 on a 0 to 4 scale) were prayer [mean (SD) = 2.9 (0.9)], opioids [2.6 (0.8)], regular exercise [2.5 (1.0)], heat/cold [2.5 (1.0)], nonsteroidal anti-inflammatory drugs [2.4 (1.0)], and acetaminophen [2.3 (1.0)]. Young-old (65–74 years) study participants reported use of more strategies than did old-old (85+ years) participants (p = .03). Perceived helpfulness of strategy use was significantly associated with pain intensity (r = -.14, p < .0001), self-efficacy (r = .28, p < .0001), and depression (r = -.20, p = .003).
Conclusion
On average, older adults view the strategies they use for persistent pain as only moderately helpful. The associations between perceived helpfulness and self-efficacy and depression suggest avenues of pain management that are focused less on specific treatments and more on how persons with persistent pain think about their pain.
==== Body
Background
Persistent pain is common among adults age 65 years and older [1,2], affecting 58–70% of community-dwelling older adults [3,4]. It is often associated with significant physical and psychosocial disability [5]. The most common types of persistent pain in this age group are neuropathic and musculoskeletal (e.g., low back pain, osteoarthritis pain, and pain in previous fracture sites) [2,5].
Despite the prevalence and importance of persistent pain among older adults, little research has systematically examined the pain management strategies used in this population. A few studies have examined the use of complementary and alternative therapies among people of various ages with persistent pain or diseases associated with pain, such as arthritis [6-9]. Among the studies that have examined the use of many types of conventional and complementary therapies for pain [10-14], few have focused on older adults and only three have assessed older adults' evaluations of the effectiveness of the treatments that they tried [11,12,15].
Blomqvist and Edberg [12] examined pain management strategies in a sample of 90 Swedes aged 75 years and older. Participants lived either in their own homes or in sheltered accommodations; all required assistance in activities of daily living from paid providers. The investigators used a structured interview to assess participants' use and perceived effectiveness of various pain management techniques. The most commonly used strategies were medication (used by 86% of the participants), distraction (e.g., watching television, reading, praying, meeting friends; 68%), rest (67%), and mobility (e.g., physical therapy, walking, housework; 66%). Only mobility was perceived as effective and without side effects. Medications and rest were considered effective, but associated with negative side effects, and distraction was categorized as being harmless, but having uncertain effectiveness.
Barry and colleagues [11] asked 245 predominantly male community-dwelling Veterans Affairs primary care patients aged 65–90 to describe their pain management strategies and to rate their perceived effectiveness. Analgesic medication use was the most commonly reported strategy (78%), followed by exercise (35%), cognitive coping (27%), religious activities (21%), and activity restriction (20%). Only four of the strategies were rated as quite or extremely effective by more than half of users; these strategies were "seek care of a physician" (80%), physical therapies (56%), complementary therapies (55%), and perseverance (52%). Little is known about the influence of older adults' sociodemographic and clinical characteristics on pain management strategy use and perceived effectiveness. Barry et al. [11] found that women were more likely than men to use cognitive coping methods and religious activities, and patients with persistent pain due to a musculoskeletal cause were more likely to use analgesic medications than were those with pain due to other causes. They did not find an association between age and any commonly used coping strategy. However, more studies are needed to establish whether, among adults aged 65 and older, age, gender, and other characteristics are associated with the use of specific pain management strategies or effectiveness of strategies used.
Kung et al. [15] examined the use and perceived effectiveness of pain management strategies among 230 community-dwelling Australians over age 55. The most frequently used strategies were heat (83%), distraction (82%), prescription medications (81%), rest (81%), physical exercise (79%), social activities (75%), and positive thinking (72%). The strategies perceived as most helpful among this sample were community support services, such as disability parking, home help, taxi transportation assistance, appropriate housing, and home modifications.
Depression and self-efficacy for managing pain are two characteristics that might well be associated with pain management strategy use and effectiveness. We previously reported that self-efficacy for managing pain (perceptions of personal capability to exercise control over pain or associated problems) was associated significantly and positively with use of task persistence, exercise/stretch, coping self-statements, and activity pacing to cope with pain in a sample of 140 retirement facility residents with persistent pain [16]. However, we did not examine whether self-efficacy was associated with use or perceived effectiveness of medical and complementary therapies for pain, and we could identify no studies of older adults that have addressed this question. Likewise, we could identify no studies of older adults that examined the association of depression with use or perceived effectiveness of pain management strategies. Studies of young and middle-aged adults with persistent pain have established the positive association between depression and passive pain coping (e.g., resting and guarding painful parts of the body) and the negative association between depression and active pain coping (e.g., task persistence, cognitive coping) [17-19]. Thus, it might prove fruitful to examine whether depression is associated with pain management strategy use among older adults. Furthermore, the tendency of depressed individuals to appraise experiences more negatively might result in such individuals viewing pain management strategies tried as ineffective.
Knowledge of pain management practices and perceptions of benefit is important for understanding how to support older adults in managing persistent pain. Increased understanding in this area could help clinicians better advise older patients with persistent pain as to pain management strategies most likely to be considered beneficial by the patient, and could also help identify strategies to study prospectively in more rigorous research. Given the dearth of information about persistent pain management strategies and complementary therapies used by older adults, the primary purpose of this study was to describe the use and perceived effectiveness of pain treatments, including complementary therapies, in a sample of older adults with persistent pain. In addition, we explored the associations of age, gender, pain characteristics, depression, and self-efficacy with number and perceived effectiveness of strategies used. Given the lack of previous research, we made no a priori hypotheses regarding these associations.
Methods
Study participants and procedures
Data for this study were collected in the baseline assessment of participants in a randomized controlled trial (RCT) evaluating the effectiveness of a pain self-management group intervention [20]. The study was approved by the institutional review board of Swedish Medical Center (Seattle, WA). All study participants provided written informed consent.
The study sample was recruited from adults residing in 43 for-profit and not-for-profit senior housing or retirement communities in the greater Seattle, Washington, area. Participants were recruited through newsletter announcements, flyers, and presentations at the communities. Although most facilities offered exclusively or predominantly independent living, some also provided assisted living. Fourteen communities offered continuing care ranging from independent living apartments to skilled nursing facilities, and eight provided subsidized housing. The majority of study participants lived independently.
Study inclusion criteria were age 65 years or older, pain of more than three months' duration that interfered with daily activities, average pain in the past week greater than 2 on a 0–10 numerical rating scale, ability to complete study questionnaires, and ability to attend seven weekly sessions at the participant's retirement facility (due to the nature of the larger RCT). Exclusion criteria were current cancer and surgery within the past six months or planned in the next six months.
Among the 362 individuals screened for the study, 44 (12%) were ineligible (Figure 1). Of the 318 eligible individuals, 256 (80.5%) enrolled and completed the baseline assessment, and 62 (19.5%) declined to participate or did not complete the baseline assessment. Among the 256 study participants, 21 (8%) were not included in analyses for this report due to missing data. The 235 individuals included in the current report did not differ significantly from the 83 individuals who were eligible but declined to enroll or did not complete all of the baseline measures in age, race, income, proportion living alone, average pain in the past week, average pain interference with activity in the past week, or average pain interference with enjoyment of life in the past week. However, the sample included in the current report, as compared with those eligible but not included, had more males (16% vs. 6%, chi-square test, p = .02) and a trend toward more participants with education past high school (77% vs. 66%, chi-square test, p = .05).
Figure 1 Study flow.
Measures
The study measures are described below. The Charlson Index of Comorbidity and the Mini-Mental State Exam were administered by study staff at the time of participant enrollment. Questions concerning pain and sociodemographic characteristics and the other measures were completed by study participants at home and were returned by mail. Because the measure used to assess pain management strategy use was developed after enrollment for the RCT began, it was completed at home and was returned by mail after the baseline assessment for 71 individuals included in this report.
Charlson Index of Comorbidity (CI)
The Charlson Index (CI) is an extensively used, valid, and reliable measure of comorbid conditions [21]. The CI uses 19 categories of comorbidity, which are primarily defined using International Classification of Diseases-9-Clinical Modifications (ICD-9-CM) diagnostic codes; each category is weighted and scored according to an algorithm. Katz et al. [22] adapted the CI for use as an interview or mailed questionnaire and found the adapted version to be reliable and valid in a group of older adults. Scores range from 0 to 35, with higher scores indicating greater health burden from comorbid causes.
Mini-Mental State Examination (MMSE)
The MMSE [23] is widely used to screen for cognitive impairment in older adults. It consists of 30 items and requires 5–10 minutes to administer. Items assess orientation, memory, attention, and calculation. The score is the total number of correct answers out of 30 possible; scores of 24–30 indicate no cognitive impairment, scores of 18–23 suggest mild cognitive impairment, and scores of less than 18 suggest severe cognitive impairment. The MMSE has been demonstrated to be valid and to have good test-retest reliability [24].
Brief Pain Inventory (BPI)
The BPI is a widely used, reliable, valid measure that assesses pain history, location, intensity, and interference with activities [25,26]. Ratings of average, current, least, and worst pain during the past week on a scale of 0 ("No pain") to 10 ("Pain as bad as you can imagine") were averaged to create a single pain intensity score [27]. Pain-related interference was calculated as the mean of 0 ("does not interfere") to 10 ("completely interferes") ratings of pain interference with general activity, mood, walking, work (including housework), relations with others, sleep, and enjoyment of life.
Geriatric Depression Scale (GDS)
The Geriatric Depression Scale (GDS) [28-30] is a 30-item self-report measure designed to assess depressive symptoms in older persons. Scores of 11 or higher are considered indicative of depression in older adults. Good sensitivity (84–100%) and specificity (73–96%) for detecting depression in geriatric psychiatric and medical outpatients have been demonstrated [31,32]. The GDS was selected over other available depression measures because of its screening efficiency with geriatric outpatient populations, its focus on affective rather than physical symptoms, and its true/false scoring format, which studies have found to be simpler for older adults to complete [32].
Self-Efficacy Scale (SES)
To assess self-efficacy for managing pain, we used the eight-item Arthritis Self-Efficacy Scale, modified by replacing the word "arthritis" with "pain" [33]. The Arthritis Self-Efficacy Scale has been demonstrated to have high internal consistency, adequate test-retest reliability, and validity [34,35]. Study participants rated on a scale from 1 = "very uncertain" to 10 = "very certain" their confidence that they can decrease their pain, keep pain from interfering with sleep, keep pain from interfering with the things they want to do, regulate activity to remain active, keep fatigue from interfering with activities, do something to feel better if they are feeling blue, manage pain during daily activities, and deal with the frustration of pain. Scores for the scale are reported as the mean of the eight ratings.
Pain Management Strategies Survey (PMSS)
We developed a Pain Management Strategies Survey (PMSS) to assess the use and perceived effectiveness of 42 medical, complementary, and self-care strategies used by older adults to manage persistent pain (see Additional file 1). The instrument was adapted from the work of Warms, Turner, Marshall, and Cardenas [13]. Several items were added to capture complementary therapies that were not included in the Warms et al. instrument [36].
Space was provided to allow respondents to add up to four therapies beyond those listed. Study participants were asked to indicate whether they were using each strategy currently or had used it in the past year, and if so, to rate the strategy's helpfulness on a scale of 0 = "not at all helpful" to 4 = "extremely helpful." A value of 2 was labeled as "moderately helpful."
Statistical analysis
Each pain management strategy was analyzed as a dichotomous variable ("used" versus "not used" currently and/or in the past year). We used descriptive statistics to summarize the demographic characteristics of the sample, the strategies used, and their perceived effectiveness. To examine age differences, we categorized age into young-old (65–74 years), mid-old (75–84 years), and old-old (85 or more years). We inspected the distributions of number of pain sites; scores on the SES, GDS, pain intensity, and pain interference measures; strategy helpfulness scores for strategies used by 25% or more participants (to ensure an adequate subgroup size); and number of treatments used; none were substantially skewed (i.e., skewness > 1.0). For each participant, ratings of helpfulness of strategies used currently or in the past year were averaged to create a single mean strategy helpfulness score. We used t-tests to examine gender differences in number of treatments used and analyses of variance (ANOVA) to examine whether the age groups differed in number of treatments used and perceived helpfulness. We used Pearson's correlation and chi-square analyses to examine associations between participant strategy use/perceived effectiveness and age, gender, pain characteristics, depression, and self-efficacy. Finally, we conducted sensitivity analyses to determine if there were significant differences in results for the 71 participants who completed the PMSS after the baseline assessment versus the other study participants. We considered the most important comparison to be in terms of proportions of participants who endorsed the use of the strategies taught in the self-management classes, because it is possible that if the survey was completed after beginning the intervention, survey responses could be affected by intervention content. We conducted chi-square analyses on the use of relaxation, exercise, heat/cold, opioids, NSAIDS, acetaminophen, anti-seizure medications, and antidepressants between the two groups. Analyses were conducted using SPSS for Windows software, version 11.5 (Chicago, IL).
Results
Sample characteristics
The sample (N = 235) was 84% female and 94% white. Seventy-seven percent were educated beyond high school and 72% lived alone (Table 1). The mean (SD) age was 82 (6.3) years (range 65–99 years). The mean CI score was 1.2 (SD = 1.4, range 0–30) and 91% of the sample scored 3 or less, indicating a relatively healthy sample. On average, the sample reported moderate pain intensity [mean (SD) = 5.4 (1.8)] and pain-related interference [mean (SD) = 4.3 (2.0)] on the BPI. The mean (SD) scores for the Geriatric Depression Scale and the Self-Efficacy Scale were 8.4 (5.7) and 5.6 (1.9), respectively. Three participants (1%) had an MMSE score under 24 (MMSE data were not available for 10 participants). Eighty-five percent of the study participants reported pain in more than one location [mean (SD) = 3 (1.5)] (Table 2). Seventy-four percent reported pain in the lower extremities, 57% reported pain in the back, and 55% reported pain in the buttocks/hips.
Table 1 Sample demographic characteristics (n = 235)
Characteristic Age 65–74 Age 75–84 Age 85+ Total n (%) of Sample
Gender
Male 5 24 9 38 (16)
Female 30 90 77 197 (84)
Living arrangement
Lives alone 24 78 67 169 (72)
Lives with someone 11 35 19 65 (28)
Not reported 1 (<1)
Education
High school or less 10 22 21 53 (23)
Post-secondary education 25 91 64 180 (77)
Not reported 2 (<1)
Income (annual)
<$45,000 30 72 56 158 (67)
$45,000 or more 2 30 10 42 (18)
Not reported 35 (15)
Ethnicity
Hispanic or Latino 1 3 3 7 (3)
Not Hispanic or Latino 34 109 83 226 (96)
Not reported 2 (1)
Race
White 29 106 85 220 (94)
Non-white 6 8 1 15 (6)
Table 2 Pain locations by age group (n = 235)
Pain Locations Age 65–74 (%) Age 75–84 (%) Age 85+ (%) Total n (%) of Sample
Legs or feet 29 (17) 84 (49) 60 (35) 173 (73.6)
Back 24 (18) 63 (47) 47 (35) 134 (57.0)
Buttocks or hip 21 (16) 64 (50) 44 (34) 129 (54.9)
Shoulders 16 (16) 44 (44) 41 (41) 101 (43.0)
Arms or hands 14 (17) 39 (48) 28 (35) 81 (34.5)
Neck 11 (22) 23 (46) 16 (32) 50 (21.3)
Head 4 (22) 9 (50) 5 (28) 18 (7.7)
Chest 3 (20) 8 (53) 4 (27) 15 (6.4)
Abdomen 1 (9) 7 (64) 3 (27) 11 (4.7)
Pain management strategy use and perceived helpfulness
The pain management strategies used most frequently were acetaminophen (used by 61% of the sample), regular exercise (58%), prayer (53%), and heat or cold (48%) (Table 3). Eighty percent were currently using or had used in the previous year at least one analgesic or adjuvant medication; 36% used two or more. The mean (SD) number of strategies reported was 5.6 (3.2) (range = 0–20). The 71 participants who completed the PMSS after the baseline assessment were significantly more likely to use relaxation (48% vs. 10%, χ2 = 41.05, p = .0001) or antidepressants (18% vs. 9%, χ2 = 4.65, p = .03) than exercise, opioids, acetaminophen, NSAIDS, heat/cold, antiseizure medications or antidepressants as compared with the participants who completed the PMSS as part of the baseline assessment.
Table 3 Treatments used by study participants (n = 235) and perceived helpfulness
Strategy Reported use* Helpfulness** Rated strategy > moderately helpful %***
n % Mean (SD)
Acetaminophen (e.g., Tylenol®) 143 61 2.3 (1.0) 40
Regular exercise program (e.g., walking, swimming, lifting weights) 136 58 2.5 (1.0) 43
Prayer or spiritual practice 124 53 2.9 (0.9) 59
Heat or cold 112 48 2.5 (1.0) 45
Glucosamine &/or chondroiton 96 41 1.7 (1.2) 20
Physical therapy 88 37 2.0 (1.2) 30
Creams or ointments (e.g., Icy Hot®, Tiger Balm®, capsaicin) 73 31 1.9 (0.9) 21
NSAIDS (e.g., Motrin®, Celebrex®) 60 26 2.4 (1.0) 48
Opioids (e.g., Vicodin®, Tylenol® #3, morphine) 59 25 2.6 (0.8) 52
Relaxation techniques (e.g., meditation, relaxation response, progressive muscle relaxation) 51 22 2.1 (0.9) 24
Injection of medication directly into joint (e.g., knee, hip) 34 15 2.2 (1.7) 58
Massage therapies (e.g., Rolfing, Swedish, shiatsu) 31 13 2.2 (1.1) 36
Antidepressants (e.g., nortriptyline, desipramine) 27 12 2.2 (1.1) 42
Chiropractic care 26 11 2.3 (1.6) 44
Anti-seizure medications (e.g., Neurontin®) 25 11 2.1 (1.5) 36
High-dose or mega-vitamin therapies, not including a daily vitamin or vitamins prescribed by your physician 20 9 1.4 (1.3) 18
Splints or braces 19 8 2.3 (0.9) 42
Special diet programs (or losing or gaining weight, like the kind you have to pay for, but not including trying to lose or gain weight on your own) 19 8 1.7 (1.5) 29
Spiritual or religious healing by others 17 7 3.0 (0.8) 69
Acupuncture 17 7 1.1 (1.2) 7
Energy healing (e.g., magnets, energy machines, the laying of hands, Reiki, Therapeutic Touch) 16 7 1.5 (1.0) 13
Movement therapy (e.g., yoga, tai chi, feldenkrais) 15 6 2.4 (1.5) 53
Foot reflexology 14 6 1.7 (1.2) 23
Chronic illness or arthritis education classes 10 4 2.0 (0.8) 25
Herbal therapies (e.g., arnica, evening primrose) 10 4 2.0 (1.5) 33
TENS unit 9 4 1.4 (1.6) 22
A lifestyle diet like vegetarianism or macrobiotics 8 3 2.5 (1.2) 63
Special jewelry (e.g., copper bracelet) 8 3 0.4 (0.8) 0
A self-help group, other than this study 7 3 2.4 (1.1) 57
Imagery techniques (e.g., guided imagery) 7 3 1.6 (1.1) 14
Homeopathy 6 3 1.5 (1.8) 33
Lidoderm patch 6 3 2.4 (1.8) 60
Infusion of pain medication directly into spine using a pump 4 2 2.3 (2.1) 50
Spinal cord stimulator 4 2 2.5 (1.3) 50
Folk remedy 3 1 3.0 (1.0) 67
Naturopathy 3 1 1.3 (1.2) 0
Osteopathy 3 1 4.0 (0.0) 100
Psychotherapy/counselling 3 1 2.3 (1.5) 33
Aromatherapy 3 1 1.5 (0.7) 0
Biofeedback 2 1 1.0 (1.4) 0
Hypnosis 2 1 3.0 (1.4) 50
Nerve blocks 2 1 3.5 (0.7) 100
NSAIDS – non-steroidal anti-inflammatory medications; TENS – transcutaneous electrical nerve stimulation
*Currently or in past year
**Scale = 0–4
*** Among participants who reported use of the strategy, percent who rated it as more than moderately helpful (3 or 4 on 0–4 scale).
Participants wrote in 46 responses in the "other" category. Of these, 27 were redundant with strategies listed on the survey. If the respondent did not indicate the use of that strategy as listed on the survey, we considered the write-in response to be the same a checking that strategy on the PMSS; these responses were included in the results shown in Table 3. The remaining 19 strategies not listed on the survey, but written in by participants, were rest, reading, music, elevate feet, elastic stockings (each written in by two participants), and "shoe inserts," "walk with walker," "considering surgery," "therapeutic mattress," "foot soaks," "brain/mind," "stop reading in bed," "nighttime snacks," and "alcohol" (each written in by one participant).
We limited our examination of ratings of helpfulness of strategies to the strategies endorsed by 25% or more of the sample, in order to ensure a sufficient size in subgroup analyses. Mean helpfulness ratings ranged from 1.7 (glucosamine) to 2.9 (prayer). Table 3 also shows the percentages of participants who rated the helpfulness of a treatment used as a 3 or 4 (extremely helpful). Seventy-four percent of the sample rated at least one strategy as a 3 or 4. Among the strategies, only prayer, opioid medication, and joint injections were rated 3 or 4 by more than 50% of participants who used the strategy.
Gender and age differences in strategy use and perceived helpfulness
Among the comparisons of men versus women in the use of each of the strategies reported by 25% more of the sample only one statistically significant difference emerged. Women were more likely than men to report use of heat or cold (51% vs. 29%, chi-square test, p = .01). Men and women did not differ significantly in mean treatment helpfulness ratings or number of treatments tried.
The three age groups differed significantly in number of treatments used (ANOVA, p = .03). Post-hoc contrasts (Tukey HSD) revealed that those aged 65–74 used more strategies on average [mean (SD) = 6.9 (3.5)] than did those aged 85 or older [mean (SD) = 5.3 (2.9); p = .03]. There was a trend toward the use of more strategies by those aged 65–74 than by those aged 75–84 [mean (SD) = 5.5 (3.3); p = .05]. There were no statistically significant differences by age group in the use of each of the 10 strategies reported most frequently, although there was a trend toward a significant difference in the use of relaxation (reported by 34% of those aged 65–74 years, 16% of those aged 75–84 years, and 24% of those aged 85 years or older; chi-square test, p = .05). The three age groups did not differ significantly in mean treatment helpfulness ratings.
Association of pain, depression, and self-efficacy with number of pain management strategies used and helpfulness
Number of strategies used was not associated significantly with pain intensity, pain interference, depression, or self-efficacy scores, but was associated positively with number of body pain locations (r = .26, p < .0001). Mean treatment helpfulness scores were associated negatively with pain intensity (r = -.14, p = .02), pain interference (r = -.18, p = .006) and depression (r = -.20, p = .003), and positively with self-efficacy scores (r = .28, p < .0001). There was no significant association between mean helpfulness and number of painful body locations.
Discussion
In this sample of older adults with persistent pain, most participants reported use of multiple pain management strategies that were perceived as only moderately effective on average. However, 74% rated at least one strategy as a 3 or 4 on the 0–4 (4 = "extremely helpful") helpfulness scale, indicating that the majority of the sample found at least one strategy that was more than moderately effective for their pain. The most commonly used strategies that were assessed as being the most effective by users included prayer or spiritual practice, opioids, nonsteroidal anti-inflammatory drugs, heat and cold, and physical exercise. We found little difference between genders or age groups within our sample of older adults in pain management strategies used.
In this sample, "prayer or spiritual practice" was the third most commonly reported strategy (with only acetaminophen and regular exercise reported by more participants). Among strategies endorsed by at least 25% of participants, this strategy was rated as most helpful on average. Barry et al. [11] also found that religious activities were one of the most commonly reported pain coping strategies in a sample of older adults, and almost half of their sample rated this strategy as quite or extremely effective. Dunn and Horgas [37] found that among older adults with religious affiliations, women and racial minorities were more likely to report using religious coping strategies to manage pain. These findings suggest the need for further study of prayer and religious practices as used by older adults to cope with persistent pain, and the potential value for clinicians to inquire about this in understanding how their patients manage pain.
Older adults appear to rely in large part on medications to manage persistent pain. In this sample, 80% currently used or had used in the past year at least one analgesic or adjuvant medication; 36% used two or more. The high reported use of analgesics among older adults with pain is consistent with previous studies [11,12,15,38]. Moreover, the perceived effectiveness for some analgesics was relatively high. These findings indicate that analgesic therapy can be helpful in this group; however, additional studies are necessary to explore the effectiveness of medications while taking into account the risk of adverse effects.
Only slightly more than half of the sample reported use in the past year of a regular exercise program to manage pain. This percentage is considerably higher than that reported by Barry et al. [11] (35% reported using the strategy in the past month), but lower than reported by Kung et al. [15] (79% of their sample used exercise as a pain management strategy). In all three studies, exercise was rated as quite/extremely helpful/effective by 26–43% of older adults who used the strategy to manage pain. The reason for the variation in perceived effectiveness is unclear, but may indicate the need for more structured exercise programs that target pain and inclusion of activity pacing into patient teaching about persistent pain. Several studies have documented the benefits of structured exercise programs in decreasing pain [39-42] in older adults; our findings suggest the need for more research on interventions to increase regular exercise among older adults with persistent pain.
Significantly more participants in the group who completed the PMSS after baseline data collection used relaxation and antidepressant medications as pain management strategies. Relaxation strategies were covered in weeks two and three of the seven-week self-management program (for participants randomized to the program), which may have influenced the reported use of relaxation by participants who completed the PMSS after beginning the self-management program. Antidepressants were covered in week six of the classes, making the intervention influence less likely for this strategy; this difference may be due to chance.
In general, the characteristics of study participants that we examined were not associated with number or type of strategies used, although women, the young-old, and those with pain in multiple locations tended to report use of more strategies. Keefe et al. reported few differences in coping strategies between men and women with osteoarthritis pain, although women used more problem-focused coping than men [43]. Barry et al. [11] found that women were more likely than men to use prayer to cope with their pain; in our study women were more likely to use hot and cold than men, but there were no significant differences in the use of prayer. It is unclear how much the differences in samples (i.e., younger age in Keefe et al.'s sample, high percentage of men in Barry et al.'s study, high percentage of women in the current study) influenced the findings related to gender. To our knowledge, this is the first study to observe a difference among different age groups over age 65 in number of pain management strategies used; further research is needed to confirm this finding in other samples and to determine reasons for differential use in this group. A possible explanation for the association between number of pain sites and number of strategies used is the use of different pain therapies for different pain problems.
Not surprisingly, study participants who reported greater pain intensity and depressive symptom severity viewed pain treatments tried as less effective, and study participants with greater self-efficacy for managing their pain reported treatments as more effective. In this exploratory study, we did not construct multivariate models to examine how pain intensity, depression, and self-efficacy interacted in explaining variance in treatment effectiveness ratings. The bivariate findings could help guide future studies constructed to test hypotheses concerning relative contributions of these variables to perceptions of treatment effectiveness. Despite the limitations of bivariate analyses, the findings suggest the potential value of interventions to treat depression and increase sense of self-efficacy for managing pain for older adults with persistent pain. It is possible that medical and complementary pain treatments might be more beneficial when patients are less depressed and have more confidence in their ability to manage their pain.
Several study limitations need to be acknowledged. First, given the multiple comparisons conducted, some significant associations may have been found by chance. We elected not to adjust for multiple comparisons given the exploratory nature of this descriptive study; further research is needed to replicate the associations found in other samples. Second, recall biases and inaccuracies may have affected the reports of treatments used and their helpfulness. Participants may have misinterpreted items on the PMSS, resulting in inaccurate reporting. For example, participants may not have known whether they used treatments such as homeopathy, glucosamine, "folk remedy," and imagery. The perception of a treatment as helpful may be due to reasons other than active ingredients of the therapy, such as placebo effects and natural history. Third, several factors (such as treatment intensity, duration, and adherence) that may have affected outcomes were not assessed. Fourth, participants in this study chose to enroll in an RCT of a self-management program for persistent pain and thus may have differed from older adults with pain not interested in participating in such a study, resulting in sample bias. The generalizability of our findings to older adults with different sociodemographic characteristics is unknown. Finally, the study results should not be interpreted as evidence for or against the effectiveness of specific treatments or pain self-management strategies. Better evidence for or against efficacy will come from high-quality RCTs.
Conclusion
Despite these limitations, the study findings indicate that as a group, older adults appear willing to try a variety of strategies to help manage persistent pain. Gender and age do not appear to influence which strategies are tried. The findings point to the need for further research in several areas: (1) to learn more about the use of prayer and spiritual practices by older adults to manage persistent pain, (2) to develop interventions effective in increasing the use of regular exercise among older adults with persistent pain, and (3) to explore further the relationships among depression, pain intensity, self-efficacy, and pain management strategy use and perceived effectiveness. Interventions to increase self-efficacy for managing pain and decrease depression in this population may be helpful in improving pain and response to medical and complementary therapies.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
All authors contributed to the design of the study, specification of study questions addressed, and decisions regarding statistical analyses to address these questions. All authors 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
PMSS questionnaire.
Click here for file
Acknowledgements
Preparation of this article was supported by a grant from the National Institute of Nursing Research, National Institutes of Health (R01NR07787). Data used in this study were collected as part of an ongoing RCT [ISRCTN-CCT-NAPN-11230].
==== Refs
Helme RD Gibson SJ The epidemiology of pain in elderly people Clinics in Geriatric Medicine 2001 17 417 431 11459713 10.1016/S0749-0690(05)70078-1
Gagliese L Melzack R Chronic pain in elderly people Pain 1997 70 3 14 9106804 10.1016/S0304-3959(96)03266-6
Mobily PR Herr KA Clark MK Wallace RB An epidemiologic analysis of pain in the elderly: the Iowa 65+ Rural Health Study J Aging Health 1994 6 139 145
Farrell MJ Gibson SJ Helme RD Ferrell BR, Ferrell BA Chronic nonmalignant pain in older people Pain in the Elderly 1996 Seattle, WA: IASP Press
American Geriatrics Society The management of persistent pain in older persons J Am Geriatr Soc 2002 50 S205 224 12067390
Rao JK Mihaliak K Kroenke K Bradley J Tierney WM Weinberger M Use of complementary therapies for arthritis among patients of rheumatologists Ann Intern Med 1999 131 409 416 10498556
Fautrel B Adam V St-Pierre Y Joseph L Clarke AE Penrod JR Use of complementary and alternative therapies by patients self-reporting arthritis or rheumatism: results from a nationwide Canadian survey J Rheumatol 2002 29 2435 2441 12415605
Haetzman M Elliott AM Smith BH Hannaford P Chambers WA Chronic pain and the use of conventional and alternative therapy Fam Pract 2003 20 147 154 12651788 10.1093/fampra/20.2.147
Sherman KJ Cherkin DC Connelly MT Erro J Savetsky JB Davis RB Eisenberg DM Complementary and alternative medical therapies for chronic low back pain: What treatments are patients willing to try? BMC Complement Altern Med 2004 4 9 15260884 10.1186/1472-6882-4-9
Turner JA Ciol MA Von Korff M Liu Y Berger R Men with pelvic pain: perceived helpfulness of medical and self-management strategies Clinical Journal of Pain
Barry LC Kerns RD Guo Z Duong BD Iannone LP Carrington Reid M Identification of strategies used to cope with chronic pain in older persons receiving primary care from a Veterans Affairs Medical Center J Am Geriatr Soc 2004 52 950 956 15161460 10.1111/j.1532-5415.2004.52263.x
Blomqvist K Edberg AK Living with persistent pain: experiences of older people receiving home care J Adv Nurs 2002 40 297 306 12383181 10.1046/j.1365-2648.2002.02371.x
Warms CA Turner JA Marshall HM Cardenas DD Treatments for chronic pain associated with spinal cord injuries: many are tried, few are helpful Clin J Pain 2002 18 154 163 12048417 10.1097/00002508-200205000-00004
Turunen JH Mantyselka PT Kumpusalo EA Ahonen RS How do people ease their pain? A population-based study J Pain 2004 5 498 504 15556828 10.1016/j.jpain.2004.08.003
Kung F Gibson SJ Helme RD Development of a pain management strategies survey questionnaire-preliminary findings The Pain Clinic 2000 12 299 313 10.1163/156856900750233794
Turner JA Ersek M Kemp CA Self-efficacy for managing pain is associated with disability, depression, and pain coping among retirement community residents with chronic pain J Pain 2005
Brown GK Nicassio PM Wallston KA Pain coping strategies and depression in rheumatoid arthritis J Consult Clin Psychol 1989 57 652 657 2794186 10.1037/0022-006X.57.5.652
Snow-Turek AL Norris MP Tan G Active and passive coping strategies in chronic pain patients Pain 1996 64 455 462 8783309 10.1016/0304-3959(95)00190-5
Tan G Jensen MP Robinson-Whelen S Thornby JI Monga TN Coping with chronic pain: a comparison of two measures Pain 2001 90 127 133 11166978 10.1016/S0304-3959(00)00395-X
Ersek M Turner JA Cain KC Kemp CA Chronic pain self-management for older adults: a randomized controlled trial [ISRCTN11899548] BMC Geriatr 2004 4 7 15285783 10.1186/1471-2318-4-7
Charlson ME Pompei P Ales KL MacKenzie CR A new method of classifying prognostic comorbidity in longitudinal studies: development and validation J Chron Dis 1987 40 373 383 3558716 10.1016/0021-9681(87)90171-8
Katz JN Chang LC Sangha O Fossel AH Bates DW Can comorbidity be measured by questionnaire rather than medical record review? Med Care 1996 34 73 84 8551813 10.1097/00005650-199601000-00006
Folstein MF Robins LN Helzer JE The Mini-Mental State Examination Arch Gen Psychiatry 1983 40 812 6860082
Tombaugh TN Mclntyre NJ The mini-mental state examination: a comprehensive review J Am Geriatr Soc 1992 40 922 935 1512391
Caraceni A Cherny N Fainsinger R Kaasa S Poulain P Radbruch L De Conno F Pain measurement tools and methods in clinical research in palliative care: recommendations of an Expert Working Group of the European Association of Palliative Care J Pain Symptom Manage 2002 23 239 255 11888722 10.1016/S0885-3924(01)00409-2
Cleeland CS Ryan KM Pain assessment: global use of the Brief Pain Inventory Ann Acad Med Singapore 1994 23 129 138 8080219
Shacham S Dar R Cleeland CS The relationship of mood state to the severity of clinical pain Pain 1984 18 187 197 6709386 10.1016/0304-3959(84)90886-8
Yesavage JA Geriatric Depression Scale Psychopharmacol Bull 1988 24 709 711 3249773
Yesavage JA Geriatric depression scale: consistency of depressive symptoms over time Percept Mot Skills 1991 73 1032 1792116
Yesavage JA Brink TL Rose TL Lum O Huang V Adey M Leirer VO Development and validation of a geriatric depression screening scale: a preliminary report J Psychiatr Res 1983 17 37 49 7183759 10.1016/0022-3956(82)90033-4
Logsdon RG Teri L Depression in Alzheimer's disease patients: caregivers as surrogate reporters J Am Geriatr Soc 1995 43 150 155 7836639
Olin JT Schneider LS Eaton EM Zemansky MF Pollock VE The Geriatric Depression Scale and the Beck Depression Inventory as screening instruments in an older adult outpatient population Psychol Assess 1992 4 190 192 10.1037/1040-3590.4.2.190
Lorig K Chastain RL Ung E Shoor S Holman HR Development and evaluation of a scale to measure perceived self-efficacy in people with arthritis Arthritis Rheum 1989 32 37 44 2912463
Gonzalez VM Stewart A Ritter PL Lorig K Translation and validation of arthritis outcome measures into Spanish Arthritis Rheum 1995 38 1429 1446 7575693
Lorig K Stewart A Ritter PL Gonzalez VM Laurent D Lynch J Outcome measures for health education and other health care interventions 1996 Thousand Oaks, CA: Sage Publications
What Is Complementary and Alternative Medicine (CAM)?
Dunn KS Horgas AL Religious and nonreligious coping in older adults experiencing chronic pain Pain Manag Nurs 2004 5 19 28 14999650 10.1016/S1524-9042(03)00070-5
Roberto KA Reynolds SG Older women's experiences with chronic pain: daily challenges and self-care practices J Women Aging 2002 14 5 23 12537073 10.1300/J074v14n03_02
Hughes SL Seymour RB Campbell R Pollak N Huber G Sharma L Impact of the fit and strong intervention on older adults with osteoarthritis Gerontologist 2004 44 217 228 15075418
Baker KR Nelson ME Felson DT Layne JE Sarno R Roubenoff R The efficacy of home based progressive strength training in older adults with knee osteoarthritis: a randomized controlled trial J Rheumatol 2001 28 1655 1665 11469475
Lin SY Davey RC Cochrane T Community rehabilitation for older adults with osteoarthritis of the lower limb: a controlled clinical trial Clin Rehabil 2004 18 92 101 14763724 10.1191/0269215504cr706oa
King AC Pruitt LA Phillips W Oka R Rodenburg A Haskell WL Comparative effects of two physical activity programs on measured and perceived physical functioning and other health-related quality of life outcomes in older adults J Gerontol A Biol Sci Med Sci 2000 55 M74 83 10737689
Keefe FJ Affleck G France CR Emery CF Waters S Caldwell DS Stainbrook D Hackshaw KV Fox LC Wilson K Gender differences in pain, coping, and mood in individuals having osteoarthritic knee pain: a within-day analysis Pain 2004 110 571 577 15288397 10.1016/j.pain.2004.03.028
|
16277666
|
PMC1298294
|
CC BY
|
2021-01-04 16:30:33
|
no
|
BMC Geriatr. 2005 Nov 8; 5:12
|
utf-8
|
BMC Geriatr
| 2,005 |
10.1186/1471-2318-5-12
|
oa_comm
|
==== Front
BMC Health Serv ResBMC Health Services Research1472-6963BioMed Central London 1472-6963-5-651620738210.1186/1472-6963-5-65Study ProtocolThe design of the Dutch EASYcare study: a randomised controlled trial on the effectiveness of a problem-based community intervention model for frail elderly people [NCT00105378] Melis René JF [email protected] Eijken Monique IJ [email protected] George F [email protected] Michel [email protected] Eddy [email protected] de Lisdonk Eloy H [email protected] Achterberg Theo [email protected] Rikkert Marcel GM [email protected] Department of Geriatric Medicine, Radboud University Nijmegen Medical Centre, internal postal code 318, PO box 9101, 6500 HB Nijmegen, The Netherlands2 Centre for Quality of Care Research, Radboud University Nijmegen Medical Centre, internal postal code 229, PO box 9101, 6500 HB Nijmegen, The Netherlands3 Department of Epidemiology and Biostatistics, Radboud University Nijmegen Medical Centre, internal postal code 252, PO box 9101, 6500 HB Nijmegen, The Netherlands4 Department of Medical Technology Assessment, Radboud University Nijmegen Medical Centre, internal postal code 253, PO box 9101, 6500 HB Nijmegen, The Netherlands5 Department of General Practice, Radboud University Nijmegen Medical Centre, internal postal code 229, PO box 9101, 6500 HB Nijmegen, The Netherlands2005 5 10 2005 5 65 65 17 8 2005 5 10 2005 Copyright © 2005 Melis et al; licensee BioMed Central Ltd.2005Melis 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
Because of their complex clinical presentations and needs frail elderly people require another approach than people who age without many complications. Several inpatient geriatric health services have proven effectiveness in frail persons. However, the wish to live independently and policies that promote independent living as an answer to population aging call for community intervention models for frail elderly people. Maybe models such as preventive home visits, comprehensive geriatric assessment, and intermediate care qualify, but their efficacy is controversial, especially in frail elderly persons living in the community. With the Dutch EASYcare Study Geriatric Intervention Programme (DGIP) we developed a model to study effectiveness of problem based community intervention models in frail elderly people.
Methods/Design
DGIP is a community intervention model for frail elderly persons where the GP refers elderly patients with a problem in cognition, mood, behaviour, mobility, and nutrition. A geriatric specialist nurse applies a guideline-based intervention with a limited number of follow up visits. The intervention starts with the application of the EASYcare instrument for geriatric screening. The EASYcare instrument assesses (instrumental) activities of daily life, cognition, mood, and includes a goal setting item. During the intervention the nurse regularly consults the referring GP and a geriatrician. Effects on functional performance (Groningen Activity Restriction Scale), health related quality of life (MOS-20), and carer burden (Zarit Burden Interview) are studied in an observer blinded randomised controlled trial. 151 participants were randomised over two treatment arms – DGIP and regular care – using pseudo cluster randomisation. We are currently performing the follow up visits. These visits are planned three and six months after inclusion. Process measures and cost measures will be recorded. Intention to treat analyses will focus on post intervention differences between treatment groups.
Discussion
The design of a trial evaluating the effects of a community intervention model for frail elderly people was presented. The problem-based participant selection procedure satisfied; few patients that the GP referred did not meet our eligibility criteria. The use of standard terminology makes detailed insight into the contents of our intervention possible using terminology others can understand well.
==== Body
Background
In frail elderly persons chronic conditions and loss of function challenge their autonomy. This harms their well-being, and often leads to institutionalisation and high health care costs.
There is much heterogeneity in the degree to which frailty affects older people. While some have many problems, others age successfully [1]. The introduction of the concept of successful aging voiced a change in our thinking about 'age-related' decline [2]. It marked the awareness that we cannot simply regard functional loss and dependency as consequences of the aging process itself when disease is absent. With this understanding these 'age-related' deficits became amenable to intervention. Of course, intervention should take the heterogeneity of the population into account; because of their complex clinical presentations and needs frail persons require another approach than people who age without many complications [3]. Although special services for frail older people have proven effectiveness in the form of inpatient geriatric health services [4], several societal developments ask for community equivalents. People prefer to stay at home, even with considerable disability [5]. Another drive behind the development of community intervention models comes from policies that promote independent living as an answer to the questions raised by population aging [6]. Possibly, models such as preventive home visits, in home comprehensive geriatric assessment, and intermediate care provide effective health services for frail older people in the community. Unfortunately, both the definition and efficacy of these community intervention models remain subject of a vivid debate [7-10]. The debate stems from the fact that the models gathered under these names show much heterogeneity as well as considerable overlap [11]. The lack of detailed insight into the content of these care models further complicates comparison [12,13]. One of the major issues is the effectiveness of these models in the expanding group of frail older people.
Despite the diversity, from literature we can distil certain elements that are used in many community intervention models. These are elements such as multidimensional and multidisciplinary working, person centred care, participant selection, and treatment adherence. Empirical evidence is available for some of these elements.
In this paper we will briefly summarise this knowledge on multidimensional assessment and management of elderly people in the community. This information grounds the choices we have made in designing a new community intervention model for frail elderly people living at home. Then, we will present the outlines of our intervention model and the design of the randomised trial in which we are currently evaluating the model. At this moment the recruitment period is already completed, and we are performing the follow up visits. Therefore, in addition to the details of the design, we will highlight some results of the conduct of the recruitment phase of our study.
Evidence on intermediate care models
Most research has been done on preventive home visits and comprehensive geriatric assessment, less scientific knowledge is available for intermediate care models.
The studies that have been evaluating intermediate care focused mainly on the evaluation of intermediate care alternatives (e.g. rapid response teams, hospital at home, early discharge schemes) in direct comparison with hospital care [11,14]. For most of the models that are not intended as direct alternatives to hospital care (e.g. residential rehabilitation, and community assessment and rehabilitation services) only descriptive data are available [15].
(Evidence-based) elements of community intervention models
Virtually all community intervention models for older people share a similar multidimensional nature covering a variety of medical, psychological, functional, and social domains. As multidimensional working is a ubiquitous feature of these models, it is in itself not thoroughly studied. There are some discussions on which domains are to be included [8].
Both in preventive home visits and comprehensive geriatric assessment it is suggested that models with a multidisciplinary team are more effective than models with a unidisciplinary approach [8,16]. Effectiveness is also claimed for longer follow up and more home visits, although a recent trial did not confirm this [16,17].
Many models provide person centred care. Some even argue that 'patient-centred, problem-driven, goal-oriented management' is a 'key minimum specification' [16].
Another element that might strengthen the effectiveness of comprehensive geriatric assessment is to secure control over the implementation of the recommendations done in the programme [4]. Models implemented in regular care often do not have complete clinical control over the enforcement of the recommendations following from the programme. In this scenario, it is very important to involve the primary care provider who will be responsible for the implementation of the proposed plan [8].
This is also important because providers' co-operation is a determinant of patient adherence to program recommendations [18]. It is difficult to change physicians' behaviour and this urges the use of high intensity programmes. Furthermore, programme effectiveness might benefit from stronger emphasis on direct recommendations to participants, and should not rely too much on the uptake of recommendations by the primary care provider [8].
Participant selection is a feature of community intervention models for elderly people that received much attention in literature. This discussion focuses on two matters: participant selection on the basis of age and on the basis of participants' needs. Age as a selection criterion is not discussed much, but causes controversy. Some authors state that home visits are more effective in persons aged 75 and over, compared to younger individuals [19]. One meta-analysis did not find an age effect, and another meta-analysis concluded most benefits are to be expected in the youngest old [13,20]. Frailty has received much more attention than age with respect to targeting these health services models to those who will benefit most. Most authors agree that too healthy elderly persons should be excluded, because both preventive home visits and comprehensive geriatric assessment are ineffective in these sprightly people [13,21]. There is more dispute about the effectiveness of these models in frail older persons. While some exclude the frailest participants, because in these persons there are too few possibilities for reversibility, other authors stress the importance of including the frailest [8,13,21,22]. Combining the evidence on the relevance of both age and frailty for participants selection Stuck concludes that health risk appraisal with individual reinforcement is beneficial to healthy persons aged 60 to 75, preventive home visits should focus on independent people aged 75 and over, and that other types of (institutional) services are needed for the frailest [23].
Unfortunately, considered this is true, this conclusion still disregards the population of frail elderly persons living in their own home.
Considerations on designing the Dutch EASYcare study
We wonder whether the effectiveness of community intervention models for frail elderly people can be enhanced using an alternative way of participant selection. In addition to selecting participants on the basis of age and frailty criteria, we ask the general practitioner (GP) to initiate the intervention when a problem requiring action emerges. This problem-based approach may enhance effectiveness because of better timing of the intervention. Others have shown this type of targeting can be effective, albeit in a non-randomised design [24]. General practitioner's and participant's compliance may also benefit, because both have discussed and agreed on the involvement of another health provider. The general practitioner is directly involved in the intervention model which realises more control over the clinical management. Direct involvement of the GP also provides feedback possibilities to better tailor the intervention and it safeguards continuity of care. We presume this continuity prevents the occurrence of negative effects that could result from discontinuation of the intervention. Hypothetically, the result is that the intensive involvement of health workers than other the general practitioner and regular home care is needed only temporarily.
If an informal carer was involved, we actively engaged this person in our intervention. We believe this involvement is a precondition for an effective community intervention model focussing on frail elderly people. However, to our knowledge, this caregiver involvement has not received much attention in the empirical studies of community intervention models.
Objectives
The objective of our study is to determine the effects of the Dutch EASYcare Study Geriatric Intervention Programme (DGIP) compared to regular medical care in improving health related quality of life in independently living elderly persons and in improving caregiver burden. Moreover, we want to determine the costs of the Dutch EASYcare Study Geriatric Intervention Programme.
Methods/Design
Study design and setting
The study is an observer blinded randomised controlled trial. Pseudo cluster randomisation was used to randomly allocate the participants to the DGIP or to a regular care group. Pseudo cluster randomisation is a randomisation method that aims to prevent both the occurrence of selection bias and contamination in a single design. We will discuss it in more detail below. The Ethical committee of the Radboud University Nijmegen Medical Centre approved of the study.
Study population
54 general practitioners from 36 GP practices in and around Nijmegen, the Netherlands, were willing to recruit subjects. We started with 38 GPs, but increased this number during the recruitment period because of disappointing inclusion rates. During the inclusion period of 21 months 155 eligible participants were randomised. We decided not to include in follow up and analysis those participants who experienced severe intercurrent disease necessitating hospital admittance, were admitted to a nursing home, died, or withdrew informed consent within one week after randomisation. The possibility of the study to have effect within one week after randomisation was judged as negligible, because it took about a week before nurses started the intervention, and the follow up visits were judged to be too strenuous for these seriously ill patients. Therefore 151 participants were included in follow up and analysis; 85 were included in the group that received the intervention model, and 66 were included in the regular care group.
Eligibility criteria
Subjects had to be eligible for participation in our intervention model (table 1). All participants had to be living in their own home or in a home for the aged and had to be 70 years or older.
Table 1 Eligibility criteria for Dutch EASYcare Study
Inclusion criteria
70 years of age and over
The patient lives independently or in a home for the aged
The patient has a health problem that was recently presented to the GP by the patient or informal caregiver
The request for help is related to the following problem fields: cognitive disorders, behavioural and psychological symptoms of dementia, mood disorders, mobility disorders and falling, or malnutrition
The patient/informal caregiver and GP have determined a goal they want to achieve
Fulfil one or more of these criteria: MMSE (Mini Mental State Examination) equal to or less than 26, GARS (Groningen Activity Restriction Scale) equal to or greater than 25 or MOS-20/subscale mental health equal to or less than 75
Exclusion criteria
The problem or request for help has an acute nature, urging for action (medical or otherwise) within less than one week
The problem or request for help is merely a medical diagnostic issue, urging for action only physicians (GP or specialist) can offer
MMSE < 20 or proven moderate to severe dementia (Clinical Dementia Rating scale [CDR] > 1, 0) and no informal caregiver (no informal caregiver is defined as: no informal caregiver who meets the patient for at least once a week on average)
The patient receives other forms of intermediate care or health care from a social worker or community-based geriatrician
The patient is already on the waiting list for a nursing home because of the problem the patient is presented with in our study
Life expectancy < 6 months because of terminal illness
When we started recruiting participants we applied an age criterion of 75 years or older. Unfortunately, seven months after the start of the recruitment the inclusion rates fell short of expectations. We decided we were able to broaden the age criterion, because the combination of frailty criteria and a problem driven approach safeguarded selection of eligible participants.
We restricted participant inclusion to those who scored below maximum (indicating good performance) on at least one of the following instruments: Mini Mental State Examination (MMSE), MOS-20 subscale mental health, or Groningen Activity Restriction Scale (GARS) [25-27]. For the MMSE the cut off was equal to or less than 26 out of 30, for MOS-20 mental health equal to or less than 75 out of 100, and for GARS the cut off was equal to or greater than 25. The GARS score ranges 18 to 54, where 18 indicates best functional performance.
We excluded participants with an MMSE of less than 20 or a proved moderate to severe dementia (Clinical Dementia Rating scale [CDR] > 1, 0) and no informal caregiver, because we expected serious problems in the acquisition of research data in these persons.
Persons already receiving forms of intermediate care or health care from a social worker or community-based geriatrician were also excluded, because this made it difficult to establish which effect was measured. Receiving home care, however, was not an exclusion criterion.
Persons already on the waiting list for a nursing home, or who had a life expectancy of less than six months, because of terminal illness, were excluded as well.
As a result of a mistake, in one case the age criterion was violated. However, the intervention team agreed that this younger case (age of this participant was 69 years) fitted well into the model. As exclusion was judged to be in disagreement with the ethical treatment of participant data, this participant was kept in follow up and analysis.
Treatment arms and randomisation
Participants were randomly allocated over two treatments arms: DGIP and regular care. No restrictions were imposed on the care participants were allowed to receive in the regular care group.
Given the nature of our intervention we considered the use of two different allocation procedures available in literature: cluster randomisation or individual randomisation [28]. The use of a cluster randomised design may have had an advantage over the use of an individual randomised design, because of the possible occurrence of contamination in our trial when individual randomisation was applied [29]. On the other hand a cluster randomised design had several disadvantages. The GP would have known the allocation outcome for his cluster after the first patient in a fully cluster randomised design. This might have caused selection bias resulting in incomparability of treatment arms [30,31]. At the same time we presumed it likely that the recruitment of subjects in the control clusters would progress slowly. Why should a GP bother to refer a patient to a study, when the GP knows already that the patient will enter the control group? There is also evidence for differential recruitment rates in cluster randomisation [32].
We therefore choose to use an innovative two-step pseudo cluster randomisation procedure [28,33]. First the GPs were randomised into two groups; group I and group C. The results of this randomisation were not revealed. Then within each of these groups randomisation at the patient level was carried out. This randomisation was stratified by GP and performed in such a way that in group I the majority (approximately 80%) of the participants received the intervention treatment, while the others received standard treatment. In group C the dysbalance was reversed: the majority received standard treatment and the others got the intervention treatment.
This approach had important advantages. The GP did not know in advance which treatment a patient was going to get, so this reduced the chance of selection bias. It also prevented the occurrence of negative recruitment effects that might have resulted from being randomised to a control cluster. Had the GPs known in advance the group they were assigned to (I or C), the predictability of an individual randomisation decision had been larger than in an individually randomised trial. However, the randomisation of GPs occurred blinded. In such a situation, the GP can only gain knowledge on the randomisation proportion through the recruitment of participants. As the number of enrolled patients per practice was expected to be no more than 10, the chances to correctly guess the odds for each individual treatment are limited.
We expect the contamination due to the intervention treatment to be negligible in group C, because there are only a limited number of participants in this group on the experimental treatment. As the majority of the patients is on intervention treatment, the contamination may be a problem in patients in group I who are on standard treatment, but then it probably affects only a small portion of the patients.
A randomisation procedure with adaptive weights (minimisation) was used to ensure a balanced distribution of high versus low percentage of elderly per GP-practice and of the availability of a nurse practitioner in GP practice in the two groups I and C [34]. The patients were randomised with adaptive weights to get evenly distributed numbers of sex, and presented health problem. A person not related to the study conduct performed the randomisation.
Intervention model: DGIP
GPs referred independently living older patients to our model when there was a problem in cognition, nutrition, behaviour, mood, or mobility. The problem had to urge for nursing assessment, co-ordination of care, or therapeutic monitoring and case management. Requests were rejected if they had an acute nature or if they were purely medical diagnostic requests.
A suitable case is for example a widow living on her own in a flat on the second floor with no elevator. The GP has doubts about her cognitive abilities and she has depressive symptoms as well. This seems to affect her daily functioning, although to what extent is unclear. She has only a daughter to look after her.
After negotiating a preliminary goal with the patient, the referring GP contacted the geriatrician involved in the study. Within two weeks a geriatric specialist nurse visited the patient at home. The instrument EASYcare was applied during this first visit [35]. EASYcare is an instrument for geriatric assessment that consists of items about (instrumental) activities of daily life, cognition, mood, and ends with a goal setting item. The goal initially negotiated by patient and GP was further elaborated in an operational objective. If an informal carer was present, the nurse provided this person a carer burden assessment and the results were implemented in the care plan.
During maximum three months up to five follow up visits for additional geriatric evaluation and management were planned. The nurse, geriatrician, and general practitioner frequently discussed the necessary nursing interventions, the effect of the interventions, the level of care that was needed, and the possibilities for reversibility. If necessary the nurse consulted and advised other involved health care workers, such as home care or physical therapist.
We had two nurses and two geriatricians available for the execution of our intervention. We developed guidelines based on best nursing practice for each health problem to structure activities, because literature has pointed at the possibility that the effects of home visiting programmes are related to the home visitor's performance in conducting the visits [36]. Therefore, we structured the intervention in order to diminish this effect, without harming the flexibility of the model. Our guidelines divided the nursing process into four phases: nursing diagnosis, definition of expected outcomes, nursing interventions and assessment of outcomes. Secondly, the guidelines used standardised NANDA (North American Nursing Diagnosis Association), NOC (Nursing outcomes classification) and NIC (Nursing interventions classification) terminology for nursing diagnosis, nursing outcomes and nursing interventions respectively [37-39].
We piloted our intervention model in a feasibility study [40]. With some minor changes, this model was judged to be applicable in the current study.
Data collection and outcome measures
Within one week after referral a researcher (RM, ME) interviewed patients at home to obtain written informed consent and to collect baseline demographic characteristics and data on general health conditions. If the participant was not able to give informed consent we asked a proxy to do so. The participants always gave verbal assent and did not reject the measurements. Before the interview the participant received a written confirmation of the appointment and a questionnaire. We asked the participant to fill out the questionnaire before the appointment. If the participant was unable to fill out the questionnaire independently, we allowed help from another person. In some cases the interviewer filled out the questionnaire during the interview. We recorded the amount of help the participant received in filling out the questionnaire.
The participants provided data on the following measures: age, gender, type of residence, and the use of home care. Also, data were collected on functional abilities, cognitive condition, mobility, health-related quality of life, and loneliness.
If an informal carer was available we collected data on informal carer characteristics using a questionnaire. We collected data on type and amount of care provided, time spent on caring, and carer burden.
These measurements are repeated three and six months after inclusion. The same researcher that performed the baseline visit carries out these interviews. This researcher is not involved in the intervention nor does the researcher know the allocation decision. After each follow up visit the researcher indicates whether blinding remained intact or not.
Primary outcome measures relating to participant characteristics are functional performance in (independent) activities of daily living measured using Groningen Activity Restriction Scale and mental health using subscale mental health MOS-20. Primary outcome measure in informal carers is carer burden using the Zarit Burden Interview (ZBI) [41]. An overview of secondary outcomes and a complete list of all measurements are provided in table 2.
Table 2 Outcome measures
Variable Instrument
Background variable
Secondary outcome
Primary outcome Measured at T0 T1 T2
Functional performance (ADL/IADL) GARS-3 [27]
• Mobility Timed up and go test [44]
Health Related quality of life MOS-20 [26]
Mood Subscale mental health MOS-20
Well-being Cantril self-anchoring ladder [45]
Dementia Quality of Life questionnaire [46]
question general life satisfaction
Cognition MMSE [25]
Social functioning Loneliness scale de Jong-Gierveld [47]
Mortality
Housing conditions/sort of residence Own questionnaire
Subjective treatment effects (participant, informal carer) Patient Enablement Instrument [48]
Burden informal carer Zarit Burden Interview [41]
Questions taken from 'Zorgkompas Mantelzorger' [49]
Time spend on care (informal carer) Own questionnaire
Age (participant, informal carer) Own questionnaire
Sex (participant, informal carer) Own questionnaire
Socio-economic status Own questionnaire, classify using ISEI-92 [50]
• (Former) occupation Own questionnaire, classify using SBC-92 [50]
Nativity Own questionnaire
Co-morbidity Cumulative Illness Rating Scale-Geriatrics (CIRS-G) [51] from medical history in GP Information System
Use of home care Own questionnaire
T0 is baseline measurement
T1 is first follow up measurement, after 3 months
T2 is second follow up measurement, after 6 months
Process evaluation
We collect data on the following set of process variables: the content of the intervention programme, the adherence of participants and informal carers in the intervention group to advices given during an intervention, experiences of participants and informal carers with the intervention model, and data on GP care and care of other involved professionals in both treatment arms.
We collect data on the content of the intervention process, because this may help to identify which programme characteristics are most beneficial. An abstract form is used to extract this information from the nursing records after completion of all individual interventions. We extract information on treatment goals, nursing diagnoses (NANDA) [37], nursing interventions (NIC) [38], nursing outcomes (NOC) [39], and the employed diagnostic instruments.
Compliance of participants and informal carers is an important determinant of carrying out a successful intervention. When an individual intervention is finished the nurse that executed the intervention indicates in an MS Access® form which of a number of pre-specified advices were given. Another nurse calls the participant or informal carer one month later to check compliance on these advices.
We score subjective treatment effects in treatment group using a questionnaire that participants and informal carers filled out after the first follow up visit.
Data on GP care will be collected in both treatment arms from the information that is routinely available from the General Practice's Information System (Huisartsen Informatie Systeem). We collect the following data: medical history using ICPC-2 (International Classification of Primary Care) [42], number and content of contacts during six months of follow up using ICPC-2, number and nature of referrals, and medication using ATC classification (Anatomical Therapeutic Chemical drug classification) [43]. Data on the use of home care are collected in the participant questionnaire. The data on GP care will be collected at the end of the follow up period. These data are collected in order to be able to clarify the observed intervention effect and to establish costs.
Costs
To be able to calculate costs, data will be collected on the following cost variables. Nurses will register the time spent on the intervention using the MS Outlook® agenda. They will register the number of visits per participant. They also register the time spent on consultation, phone calls, travelling, and administration.
Data on the workload of the GP and the geriatrician will be extrapolated from the workload of the nurses. The data we collect on the care provided were already described in the paragraph 'process evaluation'. Finally, we will derive salary costs, administrative costs, and costs for materials.
Sample size considerations
A change in the primary outcome measure of functional performance (GARS-3) of 4.5 points on a scale ranging from 18 (complete independence) to 54 (complete dependence) can be found with a power (1-β) of 0.80 and α (two sided) of 0.05 in comparing two groups of 77 subjects, when pseudo cluster randomisation is applied. We use a standard deviation of 8.5, which we calculated from a pilot study. This standard deviation is well in the range of the measures of spread other studies have found [27]. A mean increase of 4.5 points is chosen as clinically relevant, because a 4.5 point increase of the overall score indicates an improvement of 25% of all items by one functional class (each item's score is classified as follows: completely dependent 3 point, partly dependent 2 points and completely independent 1 point). Cluster size is estimated to be approximately 10 participants per GP. The exact calculations and considerations are extensively described in Teerenstra et al [33].
Statistical analysis
Descriptives will be used to assess comparability of both intervention and control group for background and confounding variables. Our primary analysis will focus on the treatment arms' differences in the primary outcome measures' changes from baseline (GARS, MOS-20 subscale mental health, and Zarit Burden Interview) at three months of follow up (T1). This will be done in intention-treat-analysis. We will use mixed linear model analysis (Proc Mixed in SAS® 8) to quantify these differences. We will account for clustering at the level of the GP through the addition of a random intercept for GP to the three models. The baseline measurements of GARS, MOS-20 subscale mental health, and Zarit Burden Interview will be added to the respective models as a covariate. The factors we stratified for in the randomisation (GP-characteristics, sex of participant, and participant's presented health problem) will also be added to the models as covariates. No further corrections will be made. A conditional analysis of the treatment arms' differences in changes from baseline at six months (T2) will be performed if there is a significant effect at T1. Apart from replacing the scores at three months with those at six months the same three models will be used.
The secondary analyses will be performed on the treatment arms' differences in time trend of the primary outcome measures GARS, MOS-20 subscale mental health, and Zarit Burden Interview during follow up. Secondary analysis will further focus on the differences between treatment arms of the secondary outcome measures at three and six months of follow up. Kaplan-Meier estimates and hazard ratios will be used to quantify the intervention's effect on living conditions and mortality. Subgroup analyses will be performed for the following subgroups: living in one's own home versus living in a home for the aged, and higher versus lower levels of cognitive function. All analyses will be performed in SAS® 8.
Discussion
In this paper we presented the design of a randomised controlled trial that evaluates the effects of a community intervention model for frail elderly people living on their own. The design of this study has shown to be very challenging.
Although the recruitment of the participants took much effort, we have included a number of subjects that should be large enough to provide reliable answers to our research questions.
Our participants were selected using a problem-based approach in which the GPs decided in co-operation with the geriatrician which patients were suitable for this intervention model. This participant selection procedure satisfied; only a minor number of the referred patients did not meet our eligibility criteria based on frailty and age. Probably, piloting our intervention model was important to achieve this.
As discussed earlier, there is a lack of insight into the content of most community intervention models studied. We decided to use standard terminology such as ICPC, NANDA, NIC, NOC and ATC codes to provide insight into our intervention when used in practice. This makes detailed insight possible using terminology others can understand well.
The selection of the best randomisation method was a final major issue we had to deal with and that took much of our time. We think this randomisation procedure satisfies. Nevertheless, we will closely monitor and report in future papers how the randomisation procedure performs in practice.
Dissemination of the results of this study is planned for 2006.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MO and RM were responsible for the research question. RM, ME, TA, and MO designed the study. RM and ME carried out the data acquisition. RM wrote the first draft of this manuscript and was responsible for the revisions. ME, TA and MO contributed to drafting of the manuscript. GB gave advices on the statistical analysis, EA on the economic analysis. MW, EL, GB and EA commented on the design and the manuscript. All authors read and approved the final manuscript.
Figure 1 flow chart Dutch EASYcare Study. This flow chart summarises the progress through the phases of the Dutch EASYcare Study until the allocation of participants to each treatment arm.
Pre-publication history
The pre-publication history for this paper can be accessed here:
==== Refs
von FM Bootsma-van der WA van EE Gussekloo J Lagaay AM van DE Knook DL van der GS Westendorp RG Successful aging in the oldest old: Who can be characterized as successfully aged? Arch Intern Med 2001 161 2694 2700 11732934 10.1001/archinte.161.22.2694
Rowe JW Kahn RL Human aging: usual and successful Science 1987 237 143 149 3299702
Rubenstein LZ An overview of comprehensive geriatric assessment: Rationale, History, Program Models, Basis Components Geriatric Assessment Technology: The State of the Art 1996 1 312
Stuck AE Siu AL Wieland GD Adams J Rubenstein LZ Comprehensive geriatric assessment: a meta-analysis of controlled trials Lancet 1993 342 1032 1036 8105269 10.1016/0140-6736(93)92884-V
Salvage AV Jones DA Vetter NJ Options of people aged over 75 years on private ad local authority residential care. Age Ageing 1989 18 380 386 2629485
Active ageing: a policy framework 2002 Geneva, World Health Organisation
Melis RJF Olde Rikkert MG Parker SG van Eijken MI What is intermediate care? BMJ 2004 329 360 361 15310588 10.1136/bmj.329.7462.360
Rubenstein LZ Stuck AE Preventive home visits for older people: defining criteria for success Age Ageing 2001 30 107 109 11395336 10.1093/ageing/30.2.107
Carpenter I Gladman J Parker S Potter J Clinical and research challenges of intermediate care Age Ageing 2002 31 97 100 11937472 10.1093/ageing/31.2.97
Black C Black D Alberti G Intermediate care: statement from the Royal College of Physicians of London 2000 London, The Royal College of Physicians
Parker G Bhakta P Katbamna S Lovett C Paisley S Parker S Phelps K Baker R Jagger C Lindesay J Shepperdson B Wilson A Best place of care for older people after acute and during subacute illness: a systematic review J Health Serv Res Policy 2000 5 176 189 11556369
Walsh B Steiner A Warr J Sheron L Pickering R Nurse-led inpatient care: opening the 'black box' Int J Nurs Stud 2003 40 307 319 12605953 10.1016/S0020-7489(02)00091-3
Elkan R Kendrick D Dewey M Hewitt M Robinson J Blair M Williams D Brummell K 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
Shepperd S Iliffe S Hospital-at-home versus in-patient hospital care Cochrane Database Syst Rev 2000 CD000356 10796533
Parker G Peet S Position paper: intermediate care (produced for the Department of Health 2001
Wieland D The effectiveness and costs of comprehensive geriatric evaluation and management Crit Rev Oncol Hematol 2003 48 227 237 14607385
Fletcher AE Price GM Ng ES Stirling SL Bulpitt CJ Breeze E Nunes M Jones DA Latif A Fasey NM Vickers MR Tulloch AJ Population-based multidimensional assessment of older people in UK general practice: a cluster-randomised factorial trial Lancet 2004 364 1667 1677 15530627 10.1016/S0140-6736(04)17353-4
Bula CJ Alessi CA Aronow HU Yubas K Gold M Nisenbaum R Beck JC Rubenstein LZ Community physicians' cooperation with a program of in-home comprehensive geriatric assessment J Am Geriatr Soc 1995 43 1016 1020 7657917
Newbury J Marley J Preventive home visits to elderly people in the community. Visits are most useful for people aged >/= 75 BMJ 2000 321 512 11012259 10.1136/bmj.321.7259.512
Stuck AE Egger M Hammer A Minder CE Beck JC Home visits to prevent nursing home admission and functional decline in elderly people: systematic review and meta-regression analysis JAMA 2002 287 1022 1028 11866651 10.1001/jama.287.8.1022
van Rossum E Frederiks CM Philipsen H Portengen K Wiskerke J Knipschild P Effects of preventive home visits to elderly people BMJ 1993 307 27 32 8343668
Egger M Commentary: When, where, and why do preventive home visits work? BMJ 2001 323 724 725 10.1136/bmj.323.7323.1253
Stuck AE Beck JC Egger M Preventing disability in elderly people Lancet 2004 364 1641 1642 15530604 10.1016/S0140-6736(04)17365-0
Landi F Gambassi G Pola R Tabaccanti S Cavinato T Carbonin PU Bernabei R Impact of integrated home care services on hospital use J Am Geriatr Soc 1999 47 1430 1434 10591237
Folstein MF Folstein SE McHugh PR "Mini-mental state". 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
Kempen GI Het meten van de gezondheidstoestand van ouderen. Een toepassing van een Nederlandse versie van de MOS-schaal. [Assessment of health status of the elderly. Application of a Dutch version of the MOS scale] Tijdschr Gerontol Geriatr 1992 23 132 140 1412616
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
Borm GF Melis RJF Teerenstra S Peer PG Pseudo cluster randomization: a treatment allocation method to minimize contamination and selection bias Stat Med 2005 in press
Donner A Birkett N Buck C Randomization by cluster. Sample size requirements and analysis Am J Epidemiol 1981 114 906 914 7315838
Torgerson DJ Contamination in trials: is cluster randomisation the answer? BMJ 2001 322 355 357 11159665 10.1136/bmj.322.7282.355
Jordhoy MS Fayers PM Ahlner-Elmqvist M Kaasa S Lack of concealment may lead to selection bias in cluster randomized trials of palliative care Palliat Med 2002 16 43 49 11963450 10.1191/0269216302pm523oa
Puffer S Torgerson D Watson J Evidence for risk of bias in cluster randomised trials: review of recent trials published in three general medical journals BMJ 2003 327 785 789 14525877 10.1136/bmj.327.7418.785
Teerenstra S Melis RJF Peer PG Borm GF Pseudo cluster randomization deals with selection bias and contamination in clinical trials J Clin Epidemiol 2005 in press
Altman DG Bland JM Treatment allocation by minimisation BMJ 2005 330 843 15817555 10.1136/bmj.330.7495.843
Richardson J The Easy-Care assessment system and its appropriateness for older people Nurs Older People 2001 13 17 19 12008559
Stuck AE Minder CE Peter-Wuest I Gillmann G Egli C Kesselring A Leu RE Beck JC A randomized trial of in-home visits for disability prevention in community-dwelling older people at low and high risk for nursing home admission Arch Intern Med 2000 160 977 986 10761963 10.1001/archinte.160.7.977
NANDA Nursing diagnoses: definitions and classification 2005 6 Philadelphia, NANDA
McCloskey JC and Bulechek GM Nursing Interventions Classification (NIC) 2000 3 St Louis, Mosby
Moorhead S, Johnson M and Maas M Nursing Outcomes Classification (NOC) 2003 3 St Louis, Mosby
van Eijken MIJ Melis RJF Wensing M Olde Rikkert MGM van Achterberg T Contents of a GP supportive geriatric intermediate care programme: a descriptive study submitted 2005
Zarit SH Reever KE Bach-Peterson J Relatives of the impaired elderly: correlates of feelings of burden Gerontologist 1980 20 649 655 7203086
lamberts H and Wood M ICPC International Classification of Primary Care 1987 Oxford, Oxford University Press
About the ATC/DDD system 2005 Oslo, WHO Collaborating Centre for Drug Statistics Methodology
Podsiadlo D Richardson S The timed "Up & Go": a test of basic functional mobility for frail elderly persons J Am Geriatr Soc 1991 39 142 148 1991946
Cantril H The pattern of human concerns 1965 New Brunswick, NJ, Rutgers University Press
Brod M Stewart AL Sands L Walton P Conceptualization and measurement of quality of life in dementia: the dementia quality of life instrument (DQoL) Gerontologist 1999 39 25 35 10028768 10.1159/000026556
de Jong Gierveld J Kamphuis FH The development of a Rasch-type loneliness scale Appl Psychol Measurement 1985 9 289 299
Howie JG Heaney DJ Maxwell M Walker JJ A comparison of a Patient Enablement Instrument (PEI) against two established satisfaction scales as an outcome measure of primary care consultations Fam Pract 1998 15 165 171 9613486 10.1093/fampra/15.2.165
Duijnstee M Guldemond H Hendriks L Zorgkompas voor mantelzorgers van ouderen en chronisch ziekenLeidraad voor het in kaart brengen van de belasting van zorgende familieleden [care compass for informal carers of elderly people and people with a chronic illness A guide for mapping the burden of caring family members] 2001 1 Utrecht, Nederlands Instituut voor Zorg en Welzijn/NIZW
Bakker B Sieben I Nieuwbeerta P Ganzeboom H Maten voor prestige, sociaal-economische status en sociale klasse voor de standaard beroepen classificatie 1992 [Scales for prestige, socio-economic status and social class for the Standard Occupational Classification 1992] Sociale Wetenschappen 1997 40 1 22
Miller MD Towers A A Manual of Guidelines for Scoring the Cumulative Illness Rating Scale for Geriatrics (CIRS-G) 1991 Pittsburg, University of Pittsburgh
|
16207382
|
PMC1298295
|
CC BY
|
2021-01-04 16:31:49
|
no
|
BMC Health Serv Res. 2005 Oct 5; 5:65
|
utf-8
|
BMC Health Serv Res
| 2,005 |
10.1186/1472-6963-5-65
|
oa_comm
|
==== Front
BMC Health Serv ResBMC Health Services Research1472-6963BioMed Central London 1472-6963-5-671625963410.1186/1472-6963-5-67Research ArticleCommunication of bed allocation decisions in a critical care unit and accountability for reasonableness Cooper Andrew B [email protected] Amit S [email protected] Jennifer [email protected] Alissa H [email protected] Douglas K [email protected] Department of Critical Care Medicine, Sunnybrook and Women's College Health Science Centre and University of Toronto, Toronto, Canada2 Department of Anaesthesia, University of Toronto, Toronto, Canada3 Joint Centre for Bioethics, University of Toronto, Toronto, Canada4 Department of Philosophy, University of North Florida and Blue Cross Blue Shield of Florida Center for Ethics, Public Policy, and the Professions, Jacksonville, USA5 Department of Health Policy, University of Toronto, Toronto, Canada2005 31 10 2005 5 67 67 18 4 2005 31 10 2005 Copyright © 2005 Cooper et al; licensee BioMed Central Ltd.2005Cooper 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
Communication may affect perceptions of fair process for intensive care unit bed allocation decisions through its impact on the publicity condition of accountability for reasonableness.
Methods
We performed a qualitative case study to describe participant perceptions of the communication of bed allocation decisions in an 18-bed university affiliated, medical-surgical critical care unit at Sunnybrook and Women's College Health Sciences Centre. Interviewed participants were 3 critical care physicians, 4 clinical fellows in critical care, 4 resource nurses, 4 "end-users" (physicians who commonly referred patients to the unit), and 3 members of the administrative staff. Median bed occupancy during the study period (Jan-April 2003) was 18/18; daily admissions and discharges (median) were 3. We evaluated our description using the ethical framework "accountability for reasonableness" (A4R) to identify opportunities for improvement.
Results
The critical care physician, resource nurse, critical care fellow and end-users (trauma team leader, surgeons, neurosurgeons, anesthesiologists) functioned independently in unofficial "parallel tracks" of bed allocation decision-making; this conflicted with the official designation of the critical care physician as the sole authority. Communication between key decision-makers was indirect and could exclude those affected by the decisions; notably, family members. Participants perceived a lack of publicity for bed allocation rationales.
Conclusion
The publicity condition should be improved for critical care bed allocation decisions. Decision-making in the "parallel tracks" we describe might be unavoidable within usual constraints of time, urgency and demand. Formal guidelines for direct communication between key participants in such circumstances would help to improve the fairness of these decisions.
==== Body
Background
While there are consensus documents describing how decisions about ICU admission should be made and epidemiological data about the characteristics and outcomes of patients referred but refused admission [1,2] there is little information about this complex decision making process in the context of resource scarcity and unrelenting demand. Fairness in limit setting for admission to the intensive care unit can be evaluated using "Accountability for Reasonableness" (A4R). In the A4R framework, a fair process supplements moral deliberation based on substantive criteria and moral principles to guide resource allocation [3]. Priority setting in health care institutions is considered fair and legitimate if it satisfies four conditions (Table 1). When A4R was used to evaluate fairness of priority setting in two previous studies of intensive care units, deficiencies in the publicity and appeals conditions were observed [4]. The role of communication in the process of limit setting and its relationship with the publicity condition was highlighted in both studies, however a commentator noted "an incomplete picture of the chain of evidence" leading to the conclusion that there should be greater transparency to improve fairness [5].
Table 1 The four conditions of accountability for reasonableness
Publicity The decisions and reasons behind priority-setting decisions must be publicly available.
Relevance These rationales must rest on evidence, reasons, and principles that fair-minded people can agree are relevant to deciding how to meet the diverse needs of patients in the context of limited resources.
Appeals There is a process for revision and dispute resolution regarding priority-setting decisions.
Enforcement There is a method of regulation in place to ensure that the first three criteria are met.
We sought to describe participants' perceptions of communication during the priority-setting process in the Critical Care Unit (CrCu) of Sunnybrook and Women's to gain an understanding of how the fairness of our bed allocation process might be improved.
Methods
Design
Qualitative case study research was used to describe priority setting at the microallocation level. This is the appropriate method for investigating a complex social phenomenon in its real life context [6].
Setting
Our study was conducted at Sunnybrook and Women's College Health Science Centre in its busy critical care unit (CrCu). Beds in CrCu are available to Sunnybrook hospital programs; Surgical Oncology, Trauma, and Community. From the unit census, 1100 medical, trauma and surgical patients were admitted in 2001–2002. The median number of occupied beds at daily census in the critical care unit during the interview period (January – April 2003) was 18/18 with a median of 3 admissions and discharges per day.
Sample
Theoretical sampling was used in choosing "key" participants and documents. We conducted 18 semi-structured interviews with people directly involved in microallocation in our critical care unit and we examined the content of three relevant hospital policies. We examined documents if interviewees made use of them to support decision-making. We included individuals who were recommended to us in previous interviews. For example, if an interviewee told us that resource nurses had a role in bed allocation, we interviewed resource nurses. The interviewees consisted of 3 critical care physicians, 4 clinical fellows in critical care, 4 resource nurses, 4 end-user physicians who commonly referred patients to the intensive care unit and 3 members of the administrative staff. We continued interviewing until we reached saturation; a point at which no new concepts or important actors in the process were identified in subsequent interviews.
Bed allocation in the critical care unit is specified by a policy "Critical Care Directorate [CrCu] Resource Prioritisation" which addresses the ICU needs of the following institutional programs; Trauma, Surgical Oncology and Community. This policy specifies priorities for intensive care unit admission as follows: 1.) Intramural patients; any program affiliation, including war veterans from an adjoining chronic care facility, 2.) Extramural trauma referrals [as long as one intensive care unit bed remains available] and surgical oncology patients (up to a maximum of 2 occupied intensive care unit beds per day), 3.)Extramural referrals including neurosurgical patients and other elective surgical patients requiring postoperative care in the intensive care unit. The critical care physician in charge is empowered by policy "VI-A-10 Admission of Acute Care Patients: Section 6 – Admission to a Critical Care Unit. Priority of Admission to a Critical Care Unit" as the sole participant making bed allocation decisions. Disputes or concerns about intensive care unit bed allocation decisions, such as their potential to compromise patient safety due to inadequate resources are arbitrated by the Medical Director on Call whose authority to over-rule decisions is established by policy "Core Patient Care Policy: Patient Flow".
Data collection
Interviews consisted of open-ended questions related to priority setting and fairness of the decision-making process (Table 2). The interviewer asked further specific questions to clarify answers and focus the subject on the question that was asked. The interview was audio taped and then transcribed verbatim. The transcripts were read and analysed by two individuals (AJ, AC). Three policy documents that pertained to admissions and discharges from the ICU were identified and analysed.
Table 2 Interview guide
How do you decide who gets an ICU bed?
Who is involved in making these decisions?
To whom are these decisions communicated?
Describe an example when this was a very difficult decision?
What happens if someone wants to appeal or challenge a decision?
How does making these decisions, in the way you have described, make you feel?
Do you think that the way decisions are made is fair?
What resources are available to help your decision-making?
Are priority-setting decisions consistent with the guidelines?
Who else should I talk to about this?
Data analysis
Modified thematic analysis of the interviews and documents involved reading through the transcribed data and identifying concepts that related to how priority-setting decisions were made. Conceptually associated text was organised in Microsoft Excel© spreadsheets labelled with descriptive categories. We grouped data with similar concepts into overarching themes related to allocation decisions, and then sorted it according to the identified key participants We then extracted data describing the process of communication involving these participants, focussing on content and the communication methods used. Finally we used writing to formalise our interpretation of concepts emerging from the data, and to make our interpretations explicit.
Ethics approval
The study was reviewed and approval was granted by the research ethics committees of Sunnybrook and Women's and the University of Toronto. No patients were interviewed.
Results
We will describe the perspectives of those involved in the bed allocation decisions, showing their perceptions of the lines of communication, its usual methods and content.
Critical care physicians
Critical Care physicians communicated their decisions to the Medical Director On call (an administrator empowered by policy to adjudicate controversial decisions), Trauma Team Leader, Resource Nurse, Operating Room, Surgical House staff, and Surgeons. Most critical care physicians in our sample did not identify a need to communicate their decisions to families of patients cared for by end-users.
Depending upon how full we are in the ICU I may need to communicate directly with the overhanging users of the ICU ... to say, sort of, "the ICU is getting full," or, "is full and we are really be hard [sic] to accept any more traumas," so they know not to accept if we can't care for patients. As well, the operating room, other than the surgeons to say, you know, "this is how we're doing, we are not going to be able to accommodate the patient that was scheduled to come here.
When communicating with end-users, Critical Care Physicians discussed reasons for being unable to agree to a request for an ICU bed. They included program-based prioritization, (e.g. surgical oncology would have precedence over surgical services not identified as high priority), lack of capacity, and anticipated or approaching lack of capacity. Some physicians communicated reasons for denying a bed directly to end-users, while others noted that they might communicate indirectly via house staff or by leaving a message with the receptionist at the operating room desk for the surgeon.
if a surgeon has a patient booked, the OR desk and the operating room that the surgeon would be working in that day would be advised, it might not necessarily be the surgeon directly, the OR desk may say put room 1 on hold because we are not sure about the bed; but the message gets to the surgeon who has booked the case that way
Critical care physicians required direct communication of requests for beds. In harmony with the policy VI-A-10 Admission of Acute Care Patients: Section 6 – Admission to a Critical Care Unit. Priority of Admission to a Critical Care Unit they viewed this communication as a necessary condition for any end-user to access an ICU bed. Although required by policy, direct communication was often omitted in practice. The indirectness of communication between the critical care physician and end-users resulted in unofficial "parallel tracks" of allocation decision-making (see below and Fig. 1). The lack of direct communication between the critical care physician and end-users could result in each being unaware of the other's bed allocation decisions and rationales.
Figure 1 Communication of ICU Bed Allocation Decisions. Communication involved in intensive care unit bed allocation decision-making is indirect. "End user physicians" speaking directly to the resource nurse, but not the critical care fellow or attending physician, constituted a "parallel track " of bed allocation decision-making. Arrows -direction of communication between participants Solid Lines- consistent communication of need for intensive care resource. Broken Lines- inconsistent communication of need for intensive care resource.
"End user" physicians
We use the term "end-user physician" to designate participants from any hospital program. Our end-user participants included physicians from medical and surgical services who sought access to the ICU resource on behalf of their patients. End-user physicians discussed ICU bed allocation decisions with critical care physicians, clinical fellows and house staff, the ICU resource nurse, the surgical oncology triage coordinator, bed care coordinators, extramural referring services, and patients or families. End-Users perceived the critical care fellow or resource nurse as good sources of information, but the availability of a bed could be better ascertained from a discussion with the resource nurse.
I've found some fellows when I talk to them, they've been happy to make the decision on their own and that may reflect a discussion that they've already had with the attending. And other times the fellow will say, "well this is something I need to clear with the attending. I think its ok but let me check and I'll get back to you." So at times they act as a go between other times they make the decision and I don't know if that decision is one that's already made beforehand or they're at a senior enough level or comfortable enough to, you know, so it's variable.
At Sunnybrook and Women's, clinical fellows communicate with critical care physicians about proposed admissions to the units, but they are not permitted to allocate beds independently. Reflecting this, some end-users bypassed the clinical fellows completely and based their possibilities for access to available beds on discussion with the ICU resource nurse. One respondent perceived that this information was enough to allow a decision to be made to accept an extramural referral.
On the other hand if there are beds, you're given the situation when there's 6 beds, and then I don't really have to talk to anybody I just have to tell the resource nurse and they're happy usually to "ok" the decision.
End users perceived that communication with the critical care physician would clarify rationales when patients were refused admission.
human nature being what it is, if the answer to you is "no you don't get a bed" than usually you're going to ask why and usually they try to give you a bit of a rationale as to why you didn't get the bed and "Sid Viscious " got the bed.
However, they perceived that communication with the critical care physician was less than ideal. Critical care physicians could be vague about the status of bed availability.
there's this frustration of, you don't get a straight answer, "well maybe we have a bed," "well when will you know?" "Well, give us an hour."
There was variability in the response of individual critical care physicians to families about the rationales for refusal of admission. Some end-users felt this was not inappropriate.
well hopefully, hopefully, they would communicate it to everyone involved. I think in practice its communicated with either the resident or the staff and may or may not discuss it with the family. the ICU doc has no previous relationship with the family member so it may be more appropriate for the primary care doc to go back to the family
When elective surgery requiring postoperative intensive care was affected by bed shortage, end-users identified a mediator; the Surgical Oncology Triage Co-ordinator who would intervene to decide which of two surgeries requesting the same bed would proceed.
on a given day, the ICU may say to us, "you know what, you're allowed 2 beds but we don't have 2 beds, we have one bed or we have zero beds," so then you have to fight with yourselves and what they do is say that the two surgeons who have beds will fight with themselves or there is a coordinator who has to decide and generally the ICU tries to keep itself out of that discussion.
Resource nurse
The resource nurse and critical care physician approve proposed admissions after considering their impact on physical and human resources in the intensive care unit.
It's more the resource nurse and the doctors along with the bed flow person, really, because I get them involved, you know, you've gotta get them involved, because you've gotta know how your bed situation goes. So right of the bat, in the morning I call them, "I've got three people to go out, how's the beds looking?"
The central role of the resource nurse in ICU bed allocation is seen in the diversity of stakeholders with whom she must communicate; the critical care physician, critical care fellows, trauma team leader, operating room, referring services and patient flow specialist. Combined with end -user perceptions that the resource nurse could make bed allocation decisions, this resulted in the resource nurse assuming an unofficial (contrary to policy) role as the primary decision maker.
the patient was gone to the OR and I took their bed and just hoped that they'd be in the OR for the same time as the other patient was going to organ procurement and this patient was coming back from the OR. That was, I didn't really go through anybody to make that decision.
we hear from an anaesthetist in the OR, "we're having trouble, it doesn't look like this patient is going to be extubatable, is there a bed?" If there isn't a bed we usually say "can they go to recovery room, and then we'll assess them later?" Or sometimes they say adamantly "no, they have to come." Then usually I just let the fellow or staff man know.
Communication of bed allocation decisions was verbal, informal and typically occurred over the telephone. Resource nurses played a key role in identifying breaks in the flow of information that might exclude participants, such as the critical care physician from involvement in the decision making process.
By phone and it starts with the resident or the fellow and it doesn't usually go straight to the staff. Usually. And sometimes they call me; they'll call the resource nurse looking to see if there is a bed or if there'll be bed.
For priority hospital programs with special entitlement to ICU resources, communication of bed requests might be made with very little direct communication between the ICU participants and the progam's end-user physicians. This was especially true for the trauma program. Communication between the trauma program and the resource nurse mirrored the indirectness already described in communication between the critical care physician and end-users.
traumas usually make their way through the resource nurse and the resource nurse usually tells the fellow that there's a trauma on its route and that's how that one is communicated.
if it's a trauma patient it could be the ward clerk down in emerg who'll just call and tell us there's a trauma coming, like eta or whatever and if they're tubed or not. Usually when they tell us that they're tubed we just automatically say, "well ok they're going to need to come to the ICU".
Clinical fellows
Clinical fellows communicated ICU bed allocation decisions with team-members, including the critical care physician, resource nurse, on-call resident and bedside nurses. They also discussed decisions with stakeholders external to the ICU; referring services, patients and their families.
I guess to start with, the people taking care of the patient wherever they are, so the nurse and the resident physician. I don't always talk to the staff directly. I don't often deliver it to the attending on the floor. The patient, the patient's family if they're available, if not maybe a little later on then the patient's family. The resource nurse in the ICU, the resident that I'm on call with, and eventually after the patient... the staff that's on call; once they're all monitored and everything.
Clinical fellows reviewed requests for ICU beds and their appropriateness (in relation to admission criteria) with referring services. Clinical fellows were the only participants who described discussing reasons for non-admission with patients or families when patients were deemed inappropriate for intensive care. However, not all clinical fellows we interviewed described doing this, and some respondents were uncomfortable with the practice.
when the ICU says, "there is no role for ICU care here", and the patient's family thinks that there is and in that situation first I would try talking to both the patient and the family.
Saying to someone that well "you're low priority as compared to another patient who's sick elsewhere," seems kind of wrong.
Discussion of referrals with critical care physicians was usually done directly. Clinical fellows described contacting referring service superiors if there was disagreement about their decisions not to admit a patient to the ICU. However, they commented on feeling left out of decisions made by other participants to allocate beds; fellows found that they needed to take the initiative to remain informed.
If my staff decides to bring somebody in the unit, they always tell the charge nurse and they usually tell me, they don't always tell the house staff. If the charge nurse decides to bring somebody into the unit they usually talk to the staff but they don't often inform the house staff or the fellow. So that's why, like when I first come in, in the morning, I go to the charge nurse and find out how many flow throughs we're having from the OR, cause otherwise you don't know how many beds you need to have empty and then I go to them, essentially, every 2 hours and say, "are there any other patients you've heard of, are there any traumas in the wings, are there any criticall patients coming, is anything going wrong in the OR, cause they're the person who will have that information but they don't necessarily tell all the house staff.
Clinical fellows observed that communication of bed allocation decisions to end-user physicians and families of patients affected by them were without formal guidance. Clinical fellows, like end-users observed that the reasons for bed allocations were not always accessible.
it happens when outside services really want somebody to come into the unit, I mean we've initially said no and their staff is more that welcome to call my staff and discuss it and come to some amicable agreement but usually at that stage the fellows and the house staff are usually out of the decision making process
Administrators: patient flow specialist
Patient flow specialists monitor resource availability (critical care and other beds) within the institution and communicate this to the critical care physician, end users and resource nurse.
sometimes it is definitely a collaborative effort between all the different physicians and sometimes nursing input, often patient flow will be called to see, where the pressure points are "can we do this, can we take another patient, can you get people out so that we can make this happen?"
When disagreements arise about interpretation of prioritization rules outlined by policy, the patient flow specialist is obligated by policy to communicate with the medical director on call. The role of the medical director was to direct physicians to decline transfers and to cancel or reprioritize surgery. This involvement was sought when the patient flow specialist was unable to convince those requesting a bed that the proposed ICU admission would be unsafe.
if there's a problem with someone doing something like... pulling someone into critical care or doing something extraneous to get people out of there which is going to impact the hospital or hurt someone. We can say "no" and we can pull a physician in to talk to them immediately.
Patient flow specialists' perceptions of the communication in bed allocation decision making included an appreciation of its complexity and urgency. Communication of these important decisions was ad hoc and lacked formal procedural guidance.
there isn't any sort of formal mechanism that we must make sure on the tick sheet that this person contacts this person, this person, it just happens form a what makes sense point of view for the situation.
Discussion
Our case study is the first to concentrate on the communication of critical care unit bed allocation decisions. We found a complex interaction between multiple participants in which information impacting on decisions to allocate beds was exchanged (Table 3, Figure 1). However, the methods of communication were often indirect, sometimes with complete breaks in the flow of information between key participants. Communication breaks often impacted those most directly affected by a decision not to allocate a bed, such as the end-user surgical oncologist or the families of patients requiring intensive care treatment. These results are important because they suggest ways to improve the publicity of limit setting decisions in critical care units.
Table 3 Communication of ICU Bed Allocation Decisions
Multiple Decision Makers Critical Care Physician Intensive Care Unit " Resource Nurse" End-User Physicians Critical Care Fellows Patient Flow Co-ordinator
Independent Functioning ''Parallel tracks'' of decision-making
Indirect Communication Telephone Intermediaries [housestaff, receptionists]
No Guidelines Affected parties left "out of the loop"
In a study of priority setting in a large university affiliated teaching hospital, Mielke et al found multiple participants in the decision making process [4]. Further to the key participants we identified, these investigators found members of the hospital medical advisory committee, legal counsel and bioethicists as participants. This finding may reflect differences in scope between our investigation and theirs; we did not extend our investigation to committee meetings. Arguments have previously been made for nurses to be advocates for care preferences and operational safety within the process of decision-making [7,8]. We found that key decision makers talked to the resource nurse during their evaluation of the availability of intensive care unit beds. The prominent involvement of some nursing personnel in critical care unit bed allocation is a unique finding of the present study. This implicit acknowledgement of the importance of the resource nurse in our intensive care unit contrasts with survey findings that nurses perceive their input is not valued [9]. The perception that a bed was available after discussion with the resource nurse was sufficient for some participants to make a commitment to admit their patients to the unit, even without further attempts to communicate with the official gatekeeper, the critical care physician.
Adding to the complexity of the process we studied were "parallel tracks" of decision-making. End-user physicians (neurosurgeons, operating room surgeons and anesthesiologists), the critical care unit resource nurse, and the most entitled service (trauma program) all identified circumstances in which they might make independent bed allocation decisions. This may explain why some end-user physicians experienced frustration in their attempts to access critical care unit resources. While official accountability for bed allocation decisions rested with the critical care physician, "parallel track" decision makers could allocate beds without this participant's knowledge. Previous investigators have demonstrated that parallel track decision makers may view the critical care unit physician as responsible for making decisions about the appropriateness of discharges, but may not consult with them about decisions to admit patients [10]. This exclusion from the decision making process may cause the official gatekeeper to be unable to explain why beds are not available in the critical care unit at a given time. When ICU beds are readily available, decision makers find it difficult to deny patient access to a bed- the " non triage mode". When there is a high census or full occupancy, decision makers must prioritize admissions to the ICU- "triage mode". Involvement of the ICU physician in triage mode involves reviewing and prioritizing proposed admissions and decision making about discharges of patients who can safely be discharged or transferred. Review of patients who are not responding or benefiting from continuing intensive care also occurs [11]. "Parallel-track" decision-making resulting in ICU bed allocation occurred in both modes in our case study. When ICU beds are readily available, parallel – track decision makers such as the resource nurse, clinical fellow or end user physicians might reasonably make bed allocation decisions based on ICU admission criteria, improving the responsiveness and flexibility of the resource. However, the importance of direct communication between parallel – track decision makers and the critical care physician – gatekeeper is not diminished because admissions in the "non -triage" mode increase the probability the unit will later enter " triage" mode conditions.
The fairness of limit setting decisions in our critical care unit may have been adversely affected by indirect communication. Until recently, very little has been known about communication in this milieu. Mielke et al found that ICU physicians communicated their bed allocation decisions and the reasons for them primarily to end user (referring) physicians. However, we found that communication of bed allocation decisions in our critical care unit was sometimes conducted through intermediaries, such as fellows, residents or receptionists. The effectiveness of communication conducted in this way, especially in regard to families is limited because family members often feel they have not been given enough time and require extra explanations if spoken to by junior physicians such as residents [12]. We observed that reasons for denial of an ICU bed were not directly communicated directly to families, who might learn of the decision from the referring service, as did those in Mielke's case study. Communication failures ("botch ups") were deemed the root cause of conflicts and appeals of decisions [4]. Bernstein et al also found that decisions and rationales were disseminated within the ICU and to end-users but not to patients, families or the public [10]. Families of critically ill patients may experience inadequate communication in close to half of their meetings with physicians, when these do occur. While there are many possible reasons, including the impact of anxiety and depression on family members' capacity to understand complex information about their loved one [13], physician related factors, such as meetings of short duration are also highlighted [14]. Even when surgical intensive care unit doctors do have meetings, they dominate the dialogue and utter more words per conversational turn than family members [15]. Within the critical care unit, the environment is noisy [16] and there are frequent disruptions by simultaneous events, which present problems not only for patients but also for researchers [17]. Our findings strengthen a chain of evidence showing how the publicity and appeals conditions of the leading fair process framework (accountability for reasonableness) may break down in critical care units [5]. They support recommendations that the publicity and appeals conditions can be improved by direct explanations of the reasons for admission to the intensive care unit and improved opportunities for debate about the appropriateness of bed allocation decisions.
Our case study methodology has some important limitations. Our choice of physician, nurse and administrative participants was guided by theoretical sampling. While this sampling method is systematic and non probabilistic, it identified initial participants comparable to those chose in previous investigations [4,10]. Although none of the participants we interviewed suggested that administrators should be included in the sample, their absence from our investigation limits our description because of the important mediation role some of them (such as the Medical Director on Call) played. We tried to ensure the validity of our analysis by giving interviewees transcripts of the analysis as it developed, but we did not do this systematically due to restrictions on health worker communication during the Toronto SARS epidemic [18]. Although this decreases the validity of our results, they are nonetheless congruent with other investigations in critical care units in ways we have already discussed. We continued to collect data until content "saturation" was encountered, but we did not interview family members of critically ill patients. This deliberate exclusion was felt to be justifiable because we were interested in how communication occurred between those who were officially responsible for decisions resulting in bed allocation in the critical care unit. However, the exclusion is unfortunate because it can be argued that family members represent an additional category of parallel track decision makers. For example, when disagreement over the appropriateness of life sustaining therapy occurs, our institutional policy mandates full treatment pending the outcome of a conflict resolution process. In effect this means that demands for critical care (even when judged medically inappropriate) from family members must be satisfied, even if for a short time by the allocation of a bed for the patient. Finally, our research setting in an urban, university affiliated teaching hospital may restrict the generalization of our results; communication in non-university teaching hospital contexts may involve participants who interact differently from those we observed.
Conclusion
A formal policy guideline for communication of 'parallel track' decision making to the resource nurse should be created to ensure that independent decisions can be safely integrated within the resource constraints of the institution. Next, communication of such important decisions should not be delegated to third party intermediaries (e.g. Receptionists, Junior Housestaff) lacking meaningful involvement in the process. To improve the fairness of process in our institution, direct communication between the critical care physician and end-users would begin to satisfy the publicity condition of accountability for reasonableness. This seems especially important in relation to communication with the families of incapable patients affected by critical care unit bed allocation decisions.
Competing interests
An earlier draft of this paper was awarded the 2003 K.J.R. Wightman Award for Research in Biomedical Ethics, Second Prize. The Royal College of Physicians & Surgeons of Canada has granted permission for publication of this material.
Authors' contributions
ABC co-ordinated research group, created and maintained database, performed thematic analysis, composed manuscript and revisions
ASJ conducted participant interviews, transcribed tapes, performed thematic analysis
JG reviewed manuscript and gave suggestions for analysis and revisions
AHS reviewed manuscript and gave suggestions for analysis and revisions
DKM oversight and planning of research methodology, reviewed manuscript revisions
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 all of the interviewees at Sunnybrook and Women's College Health Science Centre for sacrificing their valuable time during the Toronto outbreak of Severe Acute Respiratory Syndrome [SARS]. Douglas K Martin is supported by an Ontario Ministry of Health and Long-Term Care Career Scientist award. The University of Toronto Faculty of Medicine, Department of Anaesthesia, supports Dr. Jogeklar's training.
==== Refs
Anonymous Fair Allocation Of Intensive Care Unit Resources American Thoracic Society American Journal of Respiratory & Critical Care Medicine 1997 156 1282 1301
Society of Critical Care Medicine Ethics Committee Consensus Statement On The Triage Of Critically Ill Patients JAMA 2004 271 1200 1203
Daniels N Sabin JE Setting Limits Fairly: Can We Learn to Share Medical Resources? 2002 Oxford, Oxford University Press
Mielke J Martin DK Singer PA Priority Setting In A Hospital Critical Care Unit: Qualitative Case Study Crit Care Med 2003 31 2764 2768 14668612 10.1097/01.CCM.0000098440.74735.DE
Strosberg MA Teres D Intensive Care Unit Admissions Do Not Pass The Reasonableness Test Crit Care Med 2003 31 2809 2811 14668623 10.1097/01.CCM.0000098849.84086.FF
Yin RK Case Study Research: Design and Methods 1994 Second Thousand Oaks, California: Sage Publications Inc
Baggs JG Collaborative Interdisciplinary Bioethical Decision Making In Intensive Care Units Nurs Outlook 1993 41 108 112 8346049
Reigle J Resource Allocation Decisions In Critical Care Nursing Nurs Clin North Am 1989 24 1009 1015 2685769
Thomas EJ Sexton JB Helmreich RL Discrepant Attitudes About Teamwork Among Critical Care Nurses And Physicians Crit Care Med 2003 31 956 959 12627011 10.1097/01.CCM.0000056183.89175.76
Martin DK Singer PA Bernstein M Access To Intensive Care Unit Beds For Neurosurgery Patients: A Qualitative Case Study Journal of Neurology, Neurosurgery & Psychiatry 2003 74 1299 1303 10.1136/jnnp.74.9.1299
Strosberg MA The Organizational Context of Ethical Dilemmas: A Role Playing Simulation For The Intensive Care Unit J Health Admin Educ 2001 19 173 193
Moreau D Goldgran-Toledano D Alberti C Jourdain M Adrie C Annane D Garrouste-Orgeas M Lefrant JY Papazian L Quinio P Pochard F Azoulay E Junior Versus Senior Physicians For Informing Families Of Intensive Care Unit Patients American Journal of Respiratory & Critical Care Medicine 2004 169 512 517 14656750 10.1164/rccm.200305-645OC
Pochard F Azoulay E Chevret S Lemaire F Hubert P Canoui P Grassin M Zittoun R le Gall JR Dhainaut JF Schlemmer B French FAMIREA G Symptoms Of Anxiety And Depression In Family Members Of Intensive Care Unit Patients: Ethical Hypothesis Regarding Decision-Making Capacity Crit Care Med 2001 29 1893 1897 11588447 10.1097/00003246-200110000-00007
Azoulay E Chevret S Leleu G Pochard F Barboteu M Adrie C Canoui P le GJR Schlemmer B Half The Families Of Intensive Care Unit Patients Experience Inadequate Communication With Physicians Crit Care Med 2000 28 3044 3049 10966293 10.1097/00003246-200008000-00061
Gottschalk LA Bechtel RJ Buchman TG Ray SE Computerized Content Analysis Of Conversational Interactions 2003 21 CIN: Computers, Informatics, Nursing 249 258 14504601
Gabor JY Cooper AB Crombach SA Lee B Kadikar N Bettger HE Hanly PJ Contribution Of The Intensive Care Unit Environment To Sleep Disruption In Mechanically Ventilated Patients And Healthy Subjects American Journal of Respiratory & Critical Care Medicine 2003 167 708 715 12598213 10.1164/rccm.2201090
Happ MB Kagan SH Methodological Considerations For Grounded Theory Research In Critical Care Settings Nurs Res 2001 50 188 192 11393642 10.1097/00006199-200105000-00011
Booth CM Stewart TE Communication In The Toronto Critical Care Community: Important Lessons Learned During SARS Critical Care (London) 2003 7 405 406 10.1186/cc2389
|
16259634
|
PMC1298296
|
CC BY
|
2021-01-04 16:31:49
|
no
|
BMC Health Serv Res. 2005 Oct 31; 5:67
|
utf-8
|
BMC Health Serv Res
| 2,005 |
10.1186/1472-6963-5-67
|
oa_comm
|
==== Front
BMC Health Serv ResBMC Health Services Research1472-6963BioMed Central London 1472-6963-5-681626290210.1186/1472-6963-5-68Research ArticleTranslating research into policy and practice in developing countries: a case study of magnesium sulphate for pre-eclampsia Aaserud Morten [email protected] Simon [email protected] Simon [email protected] Elizabeth J [email protected] Astrid T [email protected] Mari [email protected] Lelia [email protected] Merrick [email protected] Andrew D [email protected] Norwegian Knowledge Centre for Health Services, Box 7004 St. Olavs Plass, N-0130 Oslo, Norway2 Department of Public Health and Policy, London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, UK3 Health Systems Research Unit, Medical Research Council of South Africa, South Africa4 Directorate for Health and Social Affairs, Postbox 7000 St. Olavs plass, N-0130 Oslo, Norway5 Department of Psychiatry and Behavioural Sciences, 15 Hyde Terrace, Leeds LS2 9JT, UK6 Institute for Clinical Evaluative Sciences, University of Toronto, G1 06, 2075, Bayview Avenue, Toronto, ON, Canada M4N 3M52005 1 11 2005 5 68 68 18 5 2005 1 11 2005 Copyright © 2005 Aaserud et al; licensee BioMed Central Ltd.2005Aaserud 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 evidence base for improving reproductive health continues to grow. However, concerns remain that the translation of this evidence into appropriate policies is partial and slow. Little is known about the factors affecting the use of evidence by policy makers and clinicians, particularly in developing countries. The objective of this study was to examine the factors that might affect the translation of randomised controlled trial (RCT) findings into policies and practice in developing countries.
Methods
The recent publication of an important RCT on the use of magnesium sulphate to treat pre-eclampsia provided an opportunity to explore how research findings might be translated into policy. A range of research methods, including a survey, group interview and observations with RCT collaborators and a survey of WHO drug information officers, regulatory officials and obstetricians in 12 countries, were undertaken to identify barriers and facilitators to knowledge translation.
Results
It proved difficult to obtain reliable data regarding the availability and use of commonly used drugs in many countries. The perceived barriers to implementing RCT findings regarding the use of magnesium sulphate for pre-eclampsia include drug licensing and availability; inadequate and poorly implemented clinical guidelines; and lack of political support for policy change. However, there were significant regional and national differences in the importance of specific barriers.
Conclusion
The policy changes needed to ensure widespread availability and use of magnesium sulphate are variable and complex. Difficulties in obtaining information on availability and use are combined with the wide range of barriers across settings, including a lack of support from policy makers. This makes it difficult to envisage any single intervention strategy that might be used to promote the uptake of research findings on magnesium sulphate into policy across the study settings. The publication of important trials may therefore not have the impacts on health care that researchers hope for.
==== Body
Background
Translating research evidence into policy is crucial to improving the evidence base of health care as well as to improving health care outcomes. Indeed, the recent Ministerial Summit on Health Research in Mexico City noted that, "Research has a crucial but under-recognised part to play in strengthening health systems, improving the equitable distribution of high quality health services, and advancing human development." [1]. Other have voiced similar views [2-4]. However, knowledge regarding both the factors affecting the use of evidence by policy makers and the effectiveness of interventions to improve the use of evidence by policy makers remains underdeveloped [5,6]. While some studies have been undertaken in developing countries [7,8], knowledge from these settings is particularly thin. For example, a recent systematic review of interview studies and surveys of health policy makers' perceptions of their use of evidence found 24 eligible studies, of which only four were conducted in low and middle income country settings [5]. Further studies of research-policy linkages in developing countries could therefore be helpful.
A number of models have been suggested to explain the role of research in policy making [9-13]. In summary, the models include so-called 'rational' approaches in which the uptake of research findings into policy making is envisaged as a linear process, with research-based knowledge promoting policy change and practice. Alternative approaches view the research-policy relationship as one of 'enlightenment' suggesting, in contrast to the linear models, that research findings 'percolate' through the policy environment, gradually influencing ideas and approaches. These approaches suggest a more complex and contested relationship between research and policy making, with research being one of many knowledge sources used by policy makers in taking decisions. Others have focused on the networks of influence and 'policy communities' – including civil servants, civil society organisations and others – that may form around, and influence, specific issues. It has also been suggested that research may be used strategically to delay or support policy decisions. We return to these models later in discussing the study results.
In this paper, we report the findings of a study that examined, firstly, the factors perceived as limiting the use of the results of a recently published trial of magnesium sulphate for the treatment of pre-eclampsia and, secondly, the extent to which it might be possible to make generalisable inferences regarding barriers to and facilitators of evidence-based pharmaceutical policies for maternal and child health.
Eclampsia is an important contributor to maternal morbidity and mortality in low-income countries [14]. Eclampsia is the occurrence of a convulsion (fit) in association with pre-eclampsia. Pre-eclampsia has been defined as "a multisystem disorder [of pregnancy] that is usually associated with raised blood pressure and proteinuria but, when severe, can involve the woman's liver, kidneys, clotting system, or brain. The placenta is also often involved, with an increased risk of poor growth and early delivery for the baby." [15] An estimated 50,000 women die annually following eclamptic convulsions and 99% of these deaths occur in low and middle income countries [16]. Strong evidence of the effectiveness of magnesium sulphate for women with eclampsia has been available since 1995 [15,17,18]. Until recently, however, there was little reliable evidence regarding the effectiveness of magnesium sulphate for preventing the onset of eclampsia for women with pre-eclampsia. The results of the Magpie (MAGnesium sulphate for Prevention of Eclampsia) Trial, published in 2002, provide convincing evidence that magnesium sulphate is also effective for prevention of eclampsia [19,20]. Given this evidence, there is concern that this effective, safe and inexpensive drug may still not be available in many countries for women with eclampsia or pre-eclampsia [21,22]. Hundreds of thousands of women could benefit from these research results, provided they are translated into appropriate policies and practice. Given these concerns, this study examined factors that might hinder or facilitate the translation of the results of the Magpie Trial into appropriate policies and practice. We also discuss what might be done at international level by researchers, the WHO and other international agencies to address these.
Methods
A case study approach, that involves examining a naturally occurring case or cases, was used to explore the research questions [23]. The selected case – in this instance how the findings of the Magpie Trial might be translated into policy in developing countries – was viewed as a whole, allowing processes to be compared and explored [24,25]. As such, we did not aim to investigate the fine detail of specific elements of the research translation process in each setting, but rather attempted to assemble a broad overview of the research-policy interface for the research finding of interest. For the purposes of this study, we understood 'policy' to include both service policies, which relate to resource allocation and the organisation of services, and practice policies, which are concerned with the use of resources by clinicians involved in delivering health care ([26]).
Data were collected in two phases using a range of complementary qualitative and quantitative methods. Firstly, we conducted observations, a semi-structured group interview and a survey of members of the Magpie Trial Collaborative Group – a research team representing 33 countries and involved in the Magpie Trial. Secondly, we conducted a survey of drug information officers in twelve developing countries. The details of these methods are described below.
Phase 1: perceptions of the Magpie Trial collaborative group
A meeting of the Magpie Trial collaborators in March 2002 to discuss the trial findings prior to publication provided an opportunity to explore the perceptions of this group, with a strong interest in developing the evidence based for obstetric care, regarding the barriers and facilitators to implementing the findings of their study. Limited time was available during this meeting for data collection. However, we saw this opportunity as important since most of the trial collaborating centres were represented. Several data collection methods were used, as described below.
Observations
Small group and plenary discussions on the dissemination and implementation of the trial results were observed during the Magpie Trial meeting. The collaborators planned these independently of our study. The groups were organised by geographical regions: Africa, Asia, Latin America, and a fourth group that included European, North American and Australian collaborators. Notes of the discussions were made.
Group interview
A brief semi-structured group discussion was conducted with six collaborators from Brazil, Egypt, Pakistan and Uganda, purposively selected to represent a variety of developing countries. This focused on the implications of the possible Magpie Trial results and strategies for dissemination and implementation. Detailed notes of the interview were taken and the data from the interview and the observations were coded and subjected to thematic analysis. The issues that emerged were used to help interpret the survey results.
Survey of the Magpie Trial collaborative group
A questionnaire was designed to explore barriers and facilitators to the uptake of the Magpie Trial findings into national, regional and hospital level policies in the respondent's country. A mix of open- and closed-ended questions addressed current policies regarding magnesium sulphate, specific barriers and facilitators and necessary changes in policies. It also attempted to identify key policy makers. The questionnaire was distributed after presentation of the Magpie Trial results. Spanish speaking participants were given the option of answering the open-ended questions in Spanish. These responses were later translated into English. MA and SI independently coded and qualitatively analysed the responses to the open ended questions. Other data were entered into an Excel database for analysis.
Phase 2: Survey of national drug information officers
Following analysis of Phase 1 data, a pilot survey of national drug information officers in twelve low and lower-middle income countries (Albania, Armenia, Bolivia, Cambodia, India, Indonesia, Iran, Nicaragua, the Philippines, Rwanda, South Africa, and Yemen) was conducted. The survey had the following objectives: to explore the generalisability of the perceptions of the Magpie Trial Collaborative Group; to collect data for countries that had not participated in the Magpie Trial (Armenia, Bolivia, Cambodia, Indonesia, Iran, Nicaragua, Philippines, Rwanda); to overcome some of the limitations of the information collected from the collaborators; to explore the feasibility of a survey approach for collecting data from a larger sample of countries; and to assess the reliability of the information gathered. The countries were purposively selected to represent geographic, cultural and political variety.
Drug information officers were identified through the World Health Organisation (WHO) Department of Essential Drugs and Medicines. We attempted to validate the information provided by drug information officers through surveying obstetricians in each country. These were identified by personal contacts. The questionnaire for drug information officers focused on the registration, supply and distribution of magnesium sulphate. Information was also sought on five other drugs: folic acid to prevent anaemia and the risk of fetal malformation, ergometrin to decrease post partum haemorrhage, oxytocin to induce labour, hydralazine to treat hypertension during pregnancy, and nevirapine to prevent mother-to-child transmission of HIV [27]. The first three of these are similar to magnesium sulphate in that they are inexpensive, have been used for many years, and indications for their use are common. Hydralazine was selected because it was being considered for removal from the model essential medicines list. Nevarapine was selected because it was expensive and relatively new. The questionnaire for obstetricians focused on the availability and use of magnesium sulphate. Both questionnaires consisted of closed-ended questions with space for comments. Data were entered into an Excel database and analysed using simple descriptive statistics.
Results
Group interview and observations – Magpie Trial collaborators
The group interview and observations identified a number of issues seen by Magpie collaborators as barriers to translating the Magpie Trial results into appropriate policies and practice. In presenting these, we have utilised the policy analysis framework of actors, context, content and process as an organising scheme [28]. With regard to context, a range of problems with the availability of magnesium sulphate across the trial settings was discussed. The drug was reported to be available in Latin America, Bangladesh and to some extent in India. In Pakistan and Uganda, however, magnesium sulphate was not registered as a pharmaceutical and respondents noted that there was controversy over whether it should be included in the essential medicines list. Access was also seen as a problem in Zimbabwe as the drug had not been registered even though it was listed on the essential medicines list. One participant commented that magnesium sulphate "is so cheap that the pharmaceutical companies don't bother to push for its registration", indicating one reason for the failure to register the drug in some settings. In contrast, access to magnesium sulphate was not considered to be a problem in high income settings and very frequent use was identified as a problem in parts of Latin America. There was a range of views regarding the potential for local manufacture of magnesium sulphate – a route seen as potentially useful in ensuring a low cost supply of the drug. Other problems raised included inequalities in availability between government and private hospitals and patient level barriers to improving attendance for antenatal care, including cost, transportation, traditions and beliefs. It was suggested that policy makers and politicians needed to be brought "on board" with health care professionals to address these issues.
A number of actors were identified as important to the research translation process. The collaborators suggested that few clinicians or policy makers in their settings were aware of the concept of evidence-based medicine, would read the Magpie Trial report or be able to interpret the findings. Even if clinicians were aware of the Magpie Trial results, few had opportunities to influence relevant policies. Collaborators from West Africa felt that there was a need, through medical associations, to involve politicians at the national level in discussing the trial findings. Such policy makers were seen to be distant from poor and under-resourced areas of the country, further exacerbating barriers to the development and application of appropriate policies for these areas. Collaborators from high income countries also thought that professional organisations would be important in drawing out the implications of the trial findings and making clear recommendations.
The collaborators held differing views about the importance of support from WHO and the United Nations Children's Fund (UNICEF) in influencing policies and practice. For example, collaborators from Nigeria considered it important to get support from the WHO region as this office had close relations with policy makers and politicians. One collaborator in the Asian group argued that if WHO provided magnesium sulphate to health facilities at no cost for one year, this would create pressure on governments and pharmaceutical companies to continue supplying it thereafter. In at least one of the countries represented in the discussion, UNICEF was attempting to implement protocols for emergency obstetrical care and was therefore seen as having as import role in policy making. The influence of international agencies, such as WHO and UNICEF, on the mass media was also highlighted.
A range of problems with the process of developing and implementating clinical guidelines for the treatment of hypertensive disorders of pregnancy was raised. Some countries had no national guidelines and institutions were not under any obligation to follow recommended procedures. Consequently, each hospital followed its own policies. Elsewhere national guidelines existed, but magnesium sulphate was used routinely only for eclampsia and not for pre-eclampsia. Furthermore, even where magnesium sulphate was widely available, appropriate hospital facilities, such as intensive care units, were seen as necessary to implementing the Magpie Trial results due to the perceived need for close monitoring of patients. Several participants suggested that an active guideline implementation process was required at each hospital. For example, in South Africa, where national policy recommended the use magnesium sulphate for eclampsia, the main problem was considered to be training for health professionals. Training was also seen as important in the Asian countries. However, it was acknowledged that this would require substantial resources.
The content and focus of the policy implementation process was also seen as important and to require tailoring to specific settings. In Latin America one of the important issues identified was the very frequent use, rather than the underuse, of magnesium sulphate. Here, it was suggested, training and implementation needed to focus on raising awareness of the lack of evidence to support giving magnesium sulphate to women with relatively mild disease. A lack of accountability of obstetricians was also considered to be a key problem in several settings. It was suggested that legislation was needed to specify standards of, and responsibilities for care.
A number of dissemination and implementation strategies for translating the Magpie Trial findings were suggested. The African group considered it important for the researchers themselves to publish articles in a variety of national and regional medical journals to gain support for developing and implementing appropriate guidelines. Some participants felt that the mass media could play a role in advocating for the registration of magnesium sulphate, but there were perceived problems with the quality of health care reporting and a need for health care professionals to work closely with journalists. It was also noted that many journalists were based in well resourced central urban areas and did not reflect the views of more poorly resourced peripheral regions.
In summary, significant regional and national differences in the policy making context were identified, including the importance of clinical guidelines and availability of and access to magnesium sulphate. The role and influence of different policy actors, including international agencies and the media, was also seen to vary between settings, as did the barriers to the process of developing and implementing evidence based policies.
Survey of the Magpie Trial collaborative group
A response rate of 81% (89/110 participants) was achieved. Some questions were not answered by some respondents and other responses were illegible. Table 1 shows the countries represented by the 89 respondents. The distribution reflects that of the Magpie Trial participants, with half of the respondents coming from just three countries – South Africa, Argentina and the United Kingdom – from where half of trial participants were recruited [19]. Sixty-five (73%) of the respondents were obstetricians. There were twelve "other" physicians, five midwives, two researchers, two health managers and three respondents who did not state their profession. Based on World Bank country classifications [29], 39 (44%) of the respondents represented 13 low or lower-middle income countries, 20 (22%) represented 3 upper-middle income countries, 24 (27%) represented 8 high-income countries and 6 (7%) did not state which country they represented.
Table 1 Geographical distribution of the respondents to the Magpie Trial Collaborative Group survey
Region Country Number of respondents
Africa 27
South Africa 13
Nigeria 6
Uganda 3
Egypt 1
Ghana 1
Malawi 1
Sierra Leone 1
Zimbabwe 1
Latin America 20
Argentina 16
Brazil 2
Mexico 2
Asia 13
India 5
Pakistan 3
Bangladesh 2
Singapore 1
UAE 1
Yemen 1
Europe 19
UK 16
Albania 1
Italy 1
Netherlands 1
North America 2
Canada 1
USA 1
Australia Australia 2
Not stated 6
Total 24 89
Table 2 shows the main barriers identified by respondents to the dissemination and implementation of the Magpie Trial results. Respondents from low and lower-middle income countries most frequently identified "political barriers" as a factor that might hinder the dissemination and implementation of the trial results (8/13 countries including 13/39 respondents). Issues cited by these respondents included lack of political will; that policy makers were poorly informed of, and insufficiently involved in, these matters; and that they did not see pre-eclampsia as a priority health problem. Public authorities were also identified as a barrier (4/13 countries including 6/39 respondents). In addition, lack of availability of the drug and of appropriate health personnel for its administration, as well as costs, were frequently raised. Respondents from two of the three upper-middle income countries (Argentina, Brazil and Mexico) highlighted political barriers (5/20 respondents) such as a lack of political engagement, lack of information or awareness (5/20 respondents), and costs (4/20 respondents) as important obstacles. A number of respondents from high-income countries indicated that applying the results of the Magpie Trial was not an important issue in their setting.
Table 2 Barriers to the dissemination and implementation of the results of the Magpie Trial results by country group*
Barrier Number of countries from which at least one representative gave a response
Low and lower-middle income countries (n = 13#) Upper-middle income countries (n = 3#) High-income countries (n = 8#) Total
Political barriers 8 2 2 12
Lack of information or awareness regarding magnesium sulphate 6 2 1 9
Costs of treatment 5 2 1 8
Lack of availability of personnel and hospitals 5 1 1 7
Lack of support from public authorities 4 1 2 7
Lack of availability of magnesium sulphate 5 0 1 6
Lack of clinical practice guidelines 1 1 3 5
Magnesium sulphate not registered for treatment of eclampsia/pre-eclampsia 1 1 0 2
* 17 respondents either did not reply or had illegible responses to this question.
# Total number of countries in this group.
Table 3 shows factors identified by respondents as having the potential to facilitate dissemination and implementation of the Magpie Trial results. In low and lower-middle income countries as well as upper middle-income countries, the most frequently mentioned facilitator was establishing channels to public authorities (6/13 and 2/3 countries respectively). In addition, clinical practice guidelines, resources and international organisations were seen as playing an important role in low and lower-middle income countries. Professional organisations were seen as important in upper middle-income countries (3/3 countries). The most frequently mentioned facilitator for the high-income countries was clinical practice guidelines (4/8 countries). In general, the facilitators identified addressed the barriers highlighted by the respondents.
Table 3 Facilitators to implementing the results of the Magpie Trial by country*
Facilitator Number of countries from which at least one representative gave a response
Low and lower-middle income countries
(n = 13#) Upper-middle income countries
(n = 3#) High-income countries
(n = 8#) Total
Channels to public authorities 6 2 2 10
Development/use of clinical practice guidelines 4 1 4 9
Publications in medical journals 5 0 2 7
Resources 4 1 1 6
International organisations 4 1 0 5
Professional organisations 2 3 3 8
Licensing and availability of magnesium sulphate 3 0 3 6
* 29 respondents provided illegible responses to this question.
# Total number of countries in this group.
Representatives from 7/13 low and lower-middle income countries and 2/8 high-income countries reported that magnesium sulphate was not licensed for the treatment of pre-eclampsia (not shown in tables). This barrier was not raised by respondents in upper-middle income countries. However, there were frequently conflicting responses to this question by respondents from same country. Reasons why magnesium sulphate was not licensed included failure to submit an application for licensing and "red tape". In each of 4/13 low and lower-middle income countries, one or more respondents said that magnesium sulphate was not available. Distribution of magnesium sulphate was reported to be a problem by at least one respondent from 8/13 low and lower-middle income countries and 1/3 upper-middle income countries.
Respondents indicated that there were no national clinical practice guidelines for pre-eclampsia in 8/13 low and lower-middle income countries and 2/8 high-income countries. Local (hospital) guidelines were available in most settings (82% of respondents). However, only half of respondents considered the recommendations in local guidelines to be appropriate in light of the Magpie Trial results.
Contextual health system factors were identified by respondents as important in determining the uptake of the trial findings into policy. At least one respondent in 13/16 low and middle income countries noted that fewer than 60% of pregnant women had access to services where pre-eclampsia was likely to be appropriately diagnosed. Furthermore, fewer than 40% of pregnant woman had access to professionals who could administer magnesium sulphate, according to at least one respondent in 10/13 low and lower-middle income countries. The costs of magnesium sulphate and hospital care were thought likely to hinder the dissemination and implementation of the Magpie Trial results in six of these countries. However, there were frequently different responses to this question from representatives of the same country. At least one respondent from ten of the thirteen low and lower-middle income countries thought that costs were not a barrier.
Most of the respondents from low and lower-middle income countries considered it important to change policies in their countries to improve the availability (median importance score of 5.0 on a scale from 1 [not important] to 5 [very important]) and distribution of magnesium sulphate (median importance score = 5.0). Improving access to hospitals offering treatment with magnesium sulphate and to professionals trained to manage pre-eclampsia were also seen as important (median importance scores = 5.0 and 5.0 respectively). Respondents from upper-middle income countries saw changes in policies to improve the availability of the drug as less important (median score = 2.5). The other three issues (drug distribution, access to hospitals and access to trained professionals) were seen as important by these respondents (median importance scores = 4.0, 4.0 and 5.0). For high income countries, however, all four of these issues were perceived to be less crucial (median importance scores = 1.0, 1.0, 1.0 and 2.0). Changes in clinical practice guidelines were seen as necessary in low and high-income countries (median importance 5.0 and 4.8 respectively), but less so in upper-middle income countries (median importance 3.5). Changes in policies regarding payment for magnesium sulphate and hospital costs were considered more important in low and lower-middle income countries (median importance 5.0 and 5.0 respectively) than in upper-middle income countries (median importance 2.0 and 4.0) and high-income countries (median importance 1.0 and 1.0). Responsibility for payment for magnesium sulphate and hospital care for pre-eclampsia varied across and within countries and included governments, patients and private insurance.
A range of professional and organisational actors was seen to be influential in changing relevant policies related to the use of magnesium sulphate (Table 4). A majority of respondents from low and middle-incomes countries considered central health authorities and hospital administrations to be influential. This contrasted with high-income countries where these groups were less frequently seen as having an important role in policy change. There were important differences between low and lower-middle income countries and upper middle-income and high income countries regarding the perceived influence of the WHO, drug licensing agencies, the pharmaceutical industry, politicians and patient organisations. Notably, only a minority of respondents considered patient organisations to be influential (18% overall).
Table 4 Organisations and individuals noted by respondents as having an important influence on changing policies related to the use of magnesium sulphate in their countries (%)
Organisation or individuals Low and lower-middle income countries
(n = 39#) Upper-middle income countries
(n = 20#) High-income countries
(n = 24#)
Medical or obstetrical association 92% 85% 88%
Hospital department of obstetrics 87% 85% 83%
Central health authorities 82% 70% 38%
Hospital administration 79% 60% 17%
World Health Organisation 79% 45% 8%
Nurse or midwife association 72% 35% 58%
Drug licensing agency 72% 10% 46%
Pharmaceutical industry 69% 25% 33%
Regional or local health authorities 67% 90% 33%
Mass media 56% 20% 46%
Individual influential professionals 51% 55% 75%
Politicians 51% 55% 8%
Public health insurance program 26% 65% 13%
Private health insurers 26% 35% 8%
Non-governmental organisations 26% 15% 4%
Other international organisations 26% 15% 4%
Other professional associations 21% 15% 17%
Patient organisations 13% 10% 33%
Others 13% 25% 21%
# Number of respondents.
Survey of drug information officers
Following repeated efforts, responses were obtained from national drug information officers or drug regulatory officials for only nine of the twelve countries. Their responses regarding the licensing, supply and distribution of magnesium sulphate are shown in Table 5. In contrast to the survey of Magpie collaborators, these respondents did not report problems with licensing, supply or distribution in their countries, and noted that magnesium sulphate was on the essential medicines list for seven of the nine sampled.
Table 5 Licensing, supply and distribution of magnesium sulphate as reported by drug information officers or drug regulatory officials – survey
Country Licensed for eclampsia Licensed for pre-eclampsia Imported or produced locally Problems with supply or distribution MgSO4 on EML Date included on EML
Armenia Yes Yes Imported No Yes 1994
Bolivia Yes Yes Both No Yes 1985
Cambodia Yes Yes Imported No Yes 2000
India NR NR Produced locally No No NA
Indonesia Yes Yes Produced locally No Yes 1983
Iran Yes Yes Produced locally No Yes 1981
Philippines Yes NR Produced locally No Yes 1989
Rwanda Yes Yes Both No Yes NR
Yemen Yes Yes Imported No No NA
MgSO4 = magnesium sulphate
EML = essential medicines list
NR = not reported
NA = not applicable
Obstetricians in five of these nine countries (Armenia, India, Indonesia, the Philippines and South Africa) suggested that magnesium sulphate was widely available and widely used for both eclampsia and pre-eclampsia (Table 6). However, there was some uncertainty regarding the importance of limitations to availability and differences in availability across geographical areas and between public and private hospitals.
Table 6 Availability of magnesium sulphate and use for eclampsia and pre-eclampsia as reported by obstetricians – survey
Eclampsia Pre-eclampsia
Country Available
in hospitals Geographic differences
in availability Public vs. Private
hospitals Used for
women with eclampsia Reasons for MgSO4 not
being used for all women Used for women with
pre-eclampsia? Reasons for MgSO4 not
being used for all women
Armenia
Obstetrician 1 All No No All All
Obstetrician 2 Most No No All All
Obstetrician 3 Most No No All All
India
Obstetrician 1 All No No All All
Obstetrician 2 Some Yes No Most Few D, E
Obstetrician 3 Some Yes No Some A Some E
Obstetrician 4 Some Yes No Most Some D, E
Indonesia
Obstetrician 1 Most Yes No Most A, B, C Most D, E
Obstetrician 2 Some Yes Yes Most A Most D, E
Philippines
Obstetrician 1 Most Yes Yes All All
Obstetrician 2 All Don't know Don't know All All
South Africa
Obstetrician 1 All No No All Most
Obstetrician 2 All No No All Some F
A = Problems with availability
B = Different drug used
C = Lack of awareness among clinicians
D = Lack of awareness among clinicians
E = Problems with availability
F = Not a priority
MgSO4 = magnesium sulphate
No problems were reported by drug information officers regarding the registration, supply or distribution of folic acid in any of the nine countries, with the exception of Bolivia. The same was true for ergometrin, with the exception of Cambodia, where frequent shortages were reported due to its short shelf life. No problems were reported for oxytocin (Table 7). This contrasts with magnesium sulphate for which numerous problems with supply and distribution were noted.
Table 7 Licensing, supply and distribution of five other obstetrical drugs as reported by drug information officers or drug regulatory officials – survey
Country/drug Licensed Problems with supply or distribution Comments
Armenia
Folic acid Yes No
Ergometrin Yes No The registered form is methylergometrine.
Oxytocin Yes No
Hydralazine No Yes There is a demand for the drug, but no interest from drug companies.
Nevirapine Yes Yes There is a demand for the drug, but no interest from drug companies.
Bolivia
Folic acid Yes & No Yes Availability problems within the public health services (logistical).
Ergometrin No No
Oxytocin No No
Hydralazine No Yes Not available in the national market. Only imported by two suppliers for use in public health facilities.
Nevirapine Yes Yes Not on national EML & not registered so not available on the national market. AZT is available but expensive.
Cambodia
Folic acid Yes No
Ergometrin Yes Yes There are often shortages due to short shelf-life.
Oxytocin Yes No
Hydralazine Yes Yes There are often shortages due to short shelf-life.
Nevirapine No DN Not available in Cambodia.
India
Folic acid NR No
Ergometrin NR No
Oxytocin NR No
Hydralazine NR Yes Low demand as a better therapeutic alternative is available. Produced by one manufacturer only.
Nevirapine NR Yes & No
Indonesia
Folic acid No No
Ergometrin No No
Oxytocin No No
Hydralazine No No
Nevarapine DN DN Only been registered in 2002 so too early to know if there are any problems with supply.
Iran
Folic acid Yes No
Ergometrin Yes No
Oxytocin Yes No
Hydralazine Yes No
Nevirapine No Yes This drug is not being used in Iran.
Philippines
Folic acid Yes No
Ergometrin Yes No
Oxytocin Yes No
Hydralazine Yes No
Nevirapine No NR Not registered by the Bureau of Food & Drugs.
Rwanda
Folic acid Yes No
Ergometrin DN No
Oxytocin Yes No
Hydralazine Yes Yes Many health facilities of out of stock.
Nevirapine Yes No
Yemen
Folic acid Yes DN
Ergometrin Yes DN
Oxytocin Yes DN
Hydralazine DN DN
Nevirapine DN DN No cases.
DN = do not know
NR = not reported
For hydralazine, drug information officers in three countries reported that it was not licensed and in five countries that there were problems with its supply. Problems with the registration, supply and distribution of nevirapine were reported in four of the nine countries, with Rwanda the only country not reporting a problem with the supply or distribution of this drug. Nevirapine is an expensive drug compared to magnesium sulphate and indications for its use vary across countries.
Given the difficulties in accessing appropriate respondents for this study, and respondents' limited knowledge of the information required by the questionnaire, we decided that a larger survey of drug information officers in low and lower-middle income countries would not likely provide reliable and useful information.
Discussion
This multi-country study used a range of methods to explore the wide variety of factors that hinder and facilitate the translation of research findings, in this case from the Magpie Trial, into policies and practice across a range of settings. It highlights two important issues: firstly, the barriers to translating the findings of studies such as the Magpie Trial into policy are complex, multifactoral and context specific, as has been suggested elsewhere [2,11]. For example, if one considers barriers within the health system, such as drug availability and licensing, this study demonstrates that the availability of magnesium sulphate remains an issue in some settings, although it was difficult to ascertain the extent of this problem and there is likely to be local variation. Furthermore, it proved difficult to ascertain the number of countries in which magnesium sulphate is not licensed for the treatment of eclampsia. Even if the availability of magnesium sulphate was assured, clinical practice guidelines and effective strategies to implement these would still be needed to address barriers at the practitioner level, and would need to be tailored to particular settings. Although there is evidence from some settings that practitioners may rapidly and extensively change practice in response to new evidence from RCTs and systematic reviews [30-33], this has not been commonly demonstrated in resource poor settings. This may reflect differences in facilitators of change between low and high income settings, as discussed earlier.
Another complex issue is that of interaction with and support from policy makers – the barrier most frequently cited by respondents. Although we had not anticipated this barrier, and had not asked about it in the closed-ended questions, this finding mirrors that of a recent systematic review of the use of research in health policy-making in which 'absence of personal contact' between researchers and policy makers was identified as the most common problem [5]. Although magnesium sulphate is relatively inexpensive in terms of purchase price – less than US$5 per patient [21] – political support and national policies may be needed for a number of reasons. These include registering the drug where it is not licensed, ensuring that it is distributed and available in hospitals, ensuring that health care professionals are appropriately trained, and facilitating the access of women with pre-eclampsia to hospital care. Given the complex nature of the barriers to translating the Magpie Trial findings into policy, the 'rational' model of the research-policy relationship, in which new knowledge results in policy change, therefore seems a less appropriate description of this research translation process than the 'enlightenment' model, which describes a more diffuse, non-linear relationship between research and policy and recognises that a single research study is unlikely to have a direct impact on policy [10].
The second key finding is the difficulties experienced in obtaining detailed information on drug availability and use at country level. The pilot survey focused on low and lower-middle income countries, where there appeared to be both the greatest need and the largest problems. Information was sought from the WHO's network of designated national drug information officers because we believed they would be able to reliably answer questions about the licensing, supply and distribution of magnesium sulphate. However, it was frequently difficult to contact these officers. When they could be reached, they had reliable but limited information about licensing, importation and local production, but no information on actual use. Beyond this there were little, if any, reliable data from WHO or other sources regarding the availability of magnesium sulphate or other drugs at country level. This suggests that a large investment of resources would be required both to ascertain the magnitude and precise nature of problems regarding the distribution and use of magnesium sulphate internationally and to explore how best to address these. The difficulties in obtaining information, combined with the wide and differing range of barriers between settings, makes it difficult to design any single, widely generalisable intervention package that might be used to promote the uptake of the Magpie Trial findings into policy and practice across the study settings. This is also likely to be the case for other drug and maternal and child health interventions that research has shown to be effective and that researchers and others wish to translate into policy and practice.
This study has several limitations. Firstly, the findings from the Magpie Trial Collaborative Group reflect the nature of that group, comprised primarily of self-selected obstetricians with a professional interest in pre-eclampsia. However, there were several reasons why this study was initiated with this Group, despite these limitations. It was their concerns about failures to translate results from their earlier trial [17] into practice in developing countries [21,22] that prompted this study. They were therefore motivated to participate in it. Furthermore, the Group had expert knowledge of both the specific topic and obstetrical care in their settings. Also, our survey of drug information officers, conducted to complement the data provided by the Group, demonstrates that it is difficult to access reliable information, reflecting in part the paucity of rigorous data regarding the availability and use of magnesium sulphate in low and middle-income countries. The data from the Collaborative Group therefore provide useful insights into the potential and limitations of senior researchers across a wide range of countries to influence the translation of their findings into appropriate health care policies.
This study attempted a broad overview of research translation across a range of countries. This approach had the advantage of allowing us to identify commonalities and differences between countries, thereby improving the generalisability of the findings, but also limited the depth of enquiry in any one setting. Although we have taken the approach of grouping country findings using the World Bank classification of level of income, we acknowledge that this approach might mask differences both within these groups and at sub-national levels. Further in-depth studies, currently in progress, are exploring the use of research evidence in policy making in several of the countries included in the Magpie Trial (G Woelk, personal communication). These studies are exploring the viewpoints of a wider group of policy actors, including senior policy makers within ministries and departments of health and representatives of international organisations.
Conclusion
Despite robust evidence from a landmark trial and systematic review of the effectiveness of magnesium sulphate for the treatment of pre-eclampsia, the drug is still not available in some countries and its availability varies in many others. Licensing, importation and production are probably not the most important barriers in most settings to translating this research evidence into practice. Rather, a complex and multifaceted group of issues, differing across contexts, inhibits the uptake into policy of research findings on health issues such as the treatment of eclampsia and pre-eclampsia [2]. The changes that are needed to ensure widespread and equitable availability and use of magnesium sulphate are therefore also variable and complex. This helps to explain why publication of the findings of important studies, such as the Magpie Trial, are important but generally not sufficient to change policy and practice in the health services, particularly in low and middle income countries. In the meantime, many women die each year from complications of pregnancy associated with pre-eclampsia [16]. What then can be done by researchers and other actors and where does responsibility lie?
Firstly, there is a need to identify credible national advocates or "knowledge brokers"[4,6]. The Magpie Trial collaborators are willing advocates but frequently may not have the contacts, skills or resources to influence relevant policies [34]. Moreover, they only represent thirteen low and lower-middle income countries. Secondly, once appropriate advocates are identified, they may need help in identifying the key target audiences for knowledge transfer and in identifying channels to overcome political barriers and influence those who are able to act [6]. In some countries, policy actors such as the WHO, UNICEF and other international agencies may play an important role in opening doors to national level policy makers and in promoting evidence-based policies and practice. Given this influence, the WHO and other international agencies should consider whether to raise the standards of evidence that they use in their advisory work to national governments and in their own choices of policy and programme recommendations. Not all international agencies, however, see themselves as having an advocacy role in changing clinical practice. For example, efforts made by the Magpie Trial investigators and others to persuade the President of the International Federation of Obstetricians and Gynaecologists that he and the organisation had potentially important roles in promoting the uptake of magnesium sulphate met with reluctance for himself or his organisation to interfere in the clinical freedom of individual clinicians (Chalmers, personal communication 2005) [21].
Our findings highlight the importance placed by respondents on interactions between policy makers, researchers and other stakeholders in facilitating the uptake of research findings into policies. Other studies examining the process of transferring research findings to key policy actors have also suggested that this process should be interactive in order to maximise its effectiveness [6]. These interactions can be initiated by individual researchers or policy makers but systems level mechanisms, such as observatories that bring together the producers and users of research, may also be useful [3,4,34,35] and should be considered for low and middle income countries, as part of strengthening national health research systems [1]. The sharing of experiences between such units needs to be encouraged so as to identify generalisable lessons for research translation. Financial support to ensure that magnesium sulphate is licensed and imported or produced locally may also be necessary. Moreover, financial support may be important to eliminate out of pocket payments that are acting as barriers in some settings to accessing essential medical care such as magnesium sulphate for eclampsia and severe pre-eclampsia.
At the implementation level, training health care professionals in developing countries to provide appropriate care presents challenges that are similar to those encountered in high-income countries [36-38]. These challenges are even more important to address in low-income countries because of the more severe consequences of not doing so and a greater need to use scarce resources efficiently. Evidence-based, international guidelines that can be locally adapted, such as those provided in the WHO Reproductive Health Library [39], could provide a valuable basis for both developing and implementing appropriate national clinical policies and practice guidelines. Other interventions to improve the use of medicines in developing countries have been outlined elsewhere [40], and may assist in addressing delays in responding to new evidence on the effectiveness of drugs.
More broadly, it would seem sensible to organise a wide programme of support rather than multiple one-off efforts for essential medicines in maternal and child health. Such a programme could provide an ongoing framework and support for ensuring that important research findings, such as those of the Magpie Trial, are translated into appropriate policies and practice.
Competing interests
MA has previously carried out short-term pharmacoeconomic projects for the National Insurance Service and the Norwegian Medicines Agency. In 1997–99 he worked for a private company, Brevreklame, doing market research for pharmaceutical firms in Norway. LD was the clinical coordinator of the Collaborative Eclampsia and Magpie trials.
Authors' contributions
MA, SI, MT, SL and ADO planned the group interview, observation and survey of the Magpie Collaborative Group with advice from LD. MA, SI, MT and SL undertook the interview, observations and survey. MA and SI coded and analysed the results of the survey. MA, EJP, ATD and ADO planned the survey of drug information officers with advice from SL, LD and MZ. EJP and ATD undertook the pilot survey. MA and SI prepared the first draft of the report. All of the authors commented on successive drafts.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The study was partially funded by the Practihc (Pragmatic randomised controlled trials in healthcare) network, a project funded by the European Commission's 5th Framework international collaboration with Developing Countries, Research Contract ICA4-CT-2001-10019. We are grateful to the Magpie Trial Collaborative Group for their cooperation and enthusiasm. This study would, of course, not have been possible if they had not first undertaken the Magpie Trial. Our thanks also to Barbara Farrell for her helpful comments at various stages of the project and her assistance in organising the phase 1 of the study. We would also like to thank the drug information officers and obstetricians who responded to our pilot survey. We are grateful to Shalini Jayasekar and Richard Guidotti for providing us with unpublished data and to the following people for helpful comments and advice: Metin Gulmezoglu and José Villar, WHO Department of Reproductive Health; Hans Hogerzeil and Shalini Jayasekar, WHO Department of Essential Drugs and Medicines; Iain Chalmers; Stephen Hanney; Mathieu Ouimet; and several anonymous referees.
==== Refs
The Mexico Statement on Health Research Knowledge for better health: strengthening health systems Mexico City, November 16–20: The Ministerial Summit on Health Research 2004
Haines A Kuruvilla S Borchert M Bridging the implementation gap between knowledge and action for health Bull World Health Organ 2004 82 724 731 15643791
Lavis JN Posada FB Haines A Osei E Use of research to inform public policymaking Lancet 2004 364 1615 1621 15519634 10.1016/S0140-6736(04)17317-0
World Report on Knowledge for Better Health: Strengthening Health Systems 2004 Geneva: WHO
Innvaer S Vist G Trommald M Oxman A Health policy-makers' perceptions of their use of evidence: a systematic review J Health Serv Res Policy 2002 7 239 244 12425783 10.1258/135581902320432778
Lavis JN Robertson D Woodside JM McLeod CB Abelson J Knowledge Transfer Study Group How can research organizations more effectively transfer research knowledge to decision makers? Milbank Q 2003 81 221 248 12841049 10.1111/1468-0009.t01-1-00052
Garner P Kale R Dickson R Dans T Salinas R Getting research findings into practice: implementing research findings in developing countries BMJ 1998 317 531 535 9712608
Mills A Bennett S Bloom G Gonzalez-Block M Pathmanathan I Strengthening health systems: the role and promise of policy and systems research Geneva, Alliance for Health Policy and Systems Research 2004
Caplan N Weiss CH A minimal set of conditions necessary for the utilization of social science knowledge in policy formation at the national level Using social research in public policy making 1997 Toronto: Lexington Books 183 198
Weiss C The many meanings of research utilisation Public Administration Review 1979 39 426 431
Walt G Health policy: an introduction to process and power 1994 London: Zed Books
Berridge V Thom B Research and policy; what determines the relationship? Policy studies 1996 17 23 34
Berridge V Stanton J Science and policy: historical insights Soc Sci Med 1999 49 1133 1138 10501636 10.1016/S0277-9536(99)00159-8
WHO International Collaborative Study of Hypertensive Disorders of Pregnancy Geographic variation in the incidence of hypertension in pregnancy Am J Obstet Gynecol 1988 158 80 83 2962500
Duley L Henderson-Smart DJ Magnesium sulphate versus diazepam for eclampsia The Cochrane Database of Systematic Reviews 2003 Art. No.: CD000127. DOI: 10.1002/14651858.CD000127.
Duley L Maternal mortality and hypertensive disorders of pregnancy in Africa, Asia, Latin America and the Caribbean Br J Obstet Gynaecol 1992 99 547 553 1525093
The Eclampsia Trial Collaborative Group Which anticonvulsant for women with eclampsia? Evidence from the Collaborative Eclampsia Trial Lancet 1995 345 1455 1463 7769899
Duley L Henderson-Smart DJ Magnesium sulphate versus phenytoin for eclampsia The Cochrane Database of Systematic Reviews 2003 Art. No.: CD000128. DOI: 10.1002/14651858.CD000128.
The Magpie Trial Collaborative Group Do women with pre-eclampsia, and their babies, benefit from magnesium sulphate? The Magpie Trial: a randomised placebo-controlled trial Lancet 2002 359 1877 1890 12057549 10.1016/S0140-6736(02)08778-0
Duley L Gulmezoglu AM Henderson-Smart DJ Magnesium sulphate and other anticonvulsants for women with pre-eclampsia The Cochrane Database of Systematic Reviews 2003 Art. No.: CD000025. DOI: 10.1002/14651858.CD000025.
Sheth SS Chalmers I Magnesium for preventing and treating eclampsia: time for international action Lancet 2002 359 1872 1873 12057545 10.1016/S0140-6736(02)08783-4
Mahomed K Garner P Duley L Tocolytic magnesium sulphate and paediatric mortality Lancet 1998 351 293 9457124 10.1016/S0140-6736(05)78237-4
Hammersley M What's wrong with ethnography? 1992 London: Routledge
Mitchell JC Case and situation analysis Sociological Review 1983 31 184 211
Patton MQ Qualitative research and evaluation methods 2002 London: Sage
Black N Evidence based policy: proceed with care BMJ 2001 323 275 279 11485961 10.1136/bmj.323.7307.275
WHO Essential drugs for managing complications in pregnancy and childbirth Managing Complications in Pregnancy and Childbirth: A guide for midwives and doctors 2000 Geneve: WHO, RHR
Walt G Gilson L Reforming the health sector in developing countries: the central role of policy analysis Health Policy Plan 1994 9 353 370 10139469
World Bank Country Classification Accessed 11 May, 2005.
Gulmezoglu AM Duley L Use of anticonvulsants in eclampsia and pre-eclampsia: survey of obstetricians in the United Kingdom and Republic of Ireland BMJ 1998 316 975 976 9550956
Hogle KL Kilburn L Hewson S Gafni A Wall R Hannah ME Impact of the international term breech trial on clinical practice and concerns: a survey of centre collaborators J Obstet Gynaecol Can 2003 25 14 16 12548320
Austin PC Mamdani MM Tu K Zwarenstein M Changes in prescription patterns following publication of the ALLHAT trial JAMA 2004 291 44 45 14709573 10.1001/jama.291.1.44-b
Rietberg CC Elferink-Stinkens PM Visser GH The effect of the Term Breech Trial on medical intervention behaviour and neonatal outcome in The Netherlands: an analysis of 35,453 term breech infants BJOG 2005 112 205 209 15663585
Stephenson R Hennink M Moving beyond research to influence policy: barriers and strategies for developing countries 2002 University of Southampton (Opportunities and Choices Working Paper No. 2002/05).
Hanney SR Gonzalez-Block MA Buxton MJ Kogan M The utilisation of health research in policy-making: concepts, examples and methods of assessment Health Res Policy Syst 2003 1 2 12646071 10.1186/1478-4505-1-2
Oxman AD Thomson MA Davis DA Haynes RB No magic bullets: a systematic review of 102 trials of interventions to improve professional practice CMAJ 1995 153 1423 1431 7585368
Grimshaw JM Shirran L Thomas R Mowatt G Fraser C Bero L Changing provider behavior: an overview of systematic reviews of interventions Med Care 2001 39 II2 45 11583120 10.1097/00005650-200108002-00002
Lewin S Dick J Zwarenstein M Lombard C Staff training and ambulatory TB treatment outcomes: a cluster randomised controlled trial in South Africa Bulletin of the WHO 2005 83 250 259
WHO Reproductive Health Library 2003 Oxford: Update Software
Laing RO Hogerzeil HV Ross-Degnan D Ten recommendations to improve use of medicines in developing countries Health Policy Plan 2001 16 13 20 11238425 10.1093/heapol/16.1.13
|
16262902
|
PMC1298297
|
CC BY
|
2021-01-04 16:31:49
|
no
|
BMC Health Serv Res. 2005 Nov 1; 5:68
|
utf-8
|
BMC Health Serv Res
| 2,005 |
10.1186/1472-6963-5-68
|
oa_comm
|
==== Front
BMC Health Serv ResBMC Health Services Research1472-6963BioMed Central London 1472-6963-5-711628007710.1186/1472-6963-5-71Research ArticleThe Swiss cheese model of safety incidents: are there holes in the metaphor? Perneger Thomas V [email protected] Institute of Social and Preventive Medicine, University of Geneva, Geneva, Switzerland2 Quality of Care Service, University Hospitals of Geneva, CH-1211 Geneva 14, Switzerland2005 9 11 2005 5 71 71 2 8 2005 9 11 2005 Copyright © 2005 Perneger; licensee BioMed Central Ltd.2005Perneger; 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
Reason's Swiss cheese model has become the dominant paradigm for analysing medical errors and patient safety incidents. The aim of this study was to determine if the components of the model are understood in the same way by quality and safety professionals.
Methods
Survey of a volunteer sample of persons who claimed familiarity with the model, recruited at a conference on quality in health care, and on the internet through quality-related websites. The questionnaire proposed several interpretations of components of the Swiss cheese model: a) slice of cheese, b) hole, c) arrow, d) active error, e) how to make the system safer. Eleven interpretations were compatible with this author's interpretation of the model, 12 were not.
Results
Eighty five respondents stated that they were very or quite familiar with the model. They gave on average 15.3 (SD 2.3, range 10 to 21) "correct" answers out of 23 (66.5%) – significantly more than 11.5 "correct" answers that would expected by chance (p < 0.001). Respondents gave on average 2.4 "correct" answers regarding the slice of cheese (out of 4), 2.7 "correct" answers about holes (out of 5), 2.8 "correct" answers about the arrow (out of 4), 3.3 "correct" answers about the active error (out of 5), and 4.1 "correct" answers about improving safety (out of 5).
Conclusion
The interpretations of specific features of the Swiss cheese model varied considerably among quality and safety professionals. Reaching consensus about concepts of patient safety requires further work.
==== Body
Background
James Reason proposed the image of "Swiss cheese" to explain the occurrence of system failures, such as medical mishaps [1-5]. According to this metaphor, in a complex system, hazards are prevented from causing human losses by a series of barriers. Each barrier has unintended weaknesses, or holes – hence the similarity with Swiss cheese. These weaknesses are inconstant – i.e., the holes open and close at random. When by chance all holes are aligned, the hazard reaches the patient and causes harm (Figure 1). This model draws attention to the health care system, as opposed to the individual, and to randomness, as opposed to deliberate action, in the occurrence of medical errors.
Figure 1 Swiss cheese model by James Reason published in 2000 (1). Depicted here is a more fully labelled black and white version published in 2001 (5). On the survey questionnaire, all labels and comments were hidden.
The Swiss cheese model is frequently referred to and widely accepted by patient safety professionals. This was summarised by safety expert Ronald Westrum in a testimony before a United States Advisory Committee on Blood Safety and Availability on April 25, 2000 [6]:
"Reason's model has become the common language through which complex accidents can be understood. I remember being at one conference where six speakers in a row got up and showed Swiss cheese diagrams as a kind of academic overkill. The popularity of this model obviously comes from its wide application. It's generally felt, as I said, this provides a common ground for discussing system safety."
There is no clear evidence, however, that the Swiss cheese metaphor is understood in the same way by all concerned. In this study, I explored the understanding of the Swiss cheese model by professionals who work in healthcare quality improvement.
Methods
Samples and data collection
The data for this cross-sectional survey came from two sources: paper questionnaires filled by conference delegates, and online questionnaires. A self-completed questionnaire ("the Swiss cheese quiz") was handed out to attendees of the 20th conference of the International Society for Quality in Health Care (Amsterdam, October 19–22, 2004), at the booth of the International Journal for Quality in Health Care. Completed questionnaires were collected in a ballot box. The same questionnaire was also posted on the internet [7]. Links to this site were placed on the access page of the International Journal for Quality in Health Care and the home page of the International Society for Quality in Health Care between November 2004 and January 2005.
Questionnaire
The questionnaire displayed the picture of the Swiss cheese model, as published in the BMJ [1], but with the words "hazards" and "losses" hidden. The figure was followed by this statement: "As with many metaphors, there are several ways of interpreting this model. We would like to know your own interpretation. There are no right or wrong answers." The first questions probed the familiarity of the respondent with the model and its perceived usefulness. Further questions addressed the interpretation of various aspects of the Swiss cheese model: what is represented by a slice of cheese, by a hole, and by the arrow, how is an active error represented, and how would one make the system safer (Table 1). Each question was followed by 5 statements, and respondents were asked to check all that applied. Answer statements were classified a priori by the author as being compatible or incompatible with the model (Table 1). There were 11 compatible statements and 12 incompatible statements. Two additional statements were initially rated as incompatible with the model, but this was revised to "ambiguous" following advice from other experts: a) a health care professional could indeed be considered as a barrier if her or his role was primarily to prevent the occurrence of errors or of patient harm, and b) the arrow would represent the series of events leading to a medical error if the error was concomitant with patient harm. The questionnaire ended with questions about descriptive characteristics of the respondents.
Table 1 Interpretation of the Swiss cheese model of medical error by 85 professionals who claimed to be fairly or very familiar with the model.
Compatibility with Swiss cheese model N (%) endorsing statement Percent "correct" answers
In your opinion, what does a slice of cheese represent?
A health care professional Sometimes3 14 (16.5) -
A barrier that protects patients from harm yes 61 (71.8) 71.8
A root cause of an error no 9 (10.6) 89.4
A procedure that alleviates the consequences of an error yes 14 (16.5) 16.5
A defence that prevents the occurrence of an error yes 52 (61.2) 61.2
In your opinion, what does a hole represent?
A latent error1 yes 28 (32.9) 32.9
A loss (in terms of health or money) due to an error no 5 (5.9) 94.1
An opportunity for error yes 53 (62.4) 62.4
A weakness in defences against error yes 54 (63.5) 63.5
An unsafe act yes 17 (20.0) 20.0
What does the arrow represent?
The patient's trajectory through the health care system no 29 (34.1) 65.9
A transfer of energy that injures a patient no 2 (2.4) 97.6
The transformation of a latent error1 into an active error2 no 24 (28.2) 71.8
The series of events leading to a medical error Sometimes4 51 (60.0) -
The path from hazard to patient harm yes 41 (48.2) 48.2
How or where is an active error represented on this figure?
At the base (origin) of the arrow no 10 (11.8) 88.2
At the tip of the arrow no 24 (28.2) 71.8
As one of the holes yes 26 (30.6) 30.6
As the arrow itself no 24 (28.2) 71.8
As the alignment of holes no 28 (32.9) 67.1
How can we make the health care system safer, using the "Swiss cheese" metaphor?
By adding a slice of cheese yes 27 (31.8) 31.8
By removing a slice of cheese no 6 (7.1) 92.9
By plugging a hole yes 76 (89.4) 89.4
By adding a hole no 1 (1.2) 98.8
By making all slices thinner no 6 (7.1) 92.9
1 Latent error: Failure of system design that increases the probability of harmful events. Loosely equivalent to causal factor or contributing factor.
2 Active error: Error (of commission or omission) committed at the interface between a human and a complex system.
3 A professional whose role is to make the process of care safer may be thought of as a protective barrier
4 This would be true if the error equates with patient harm, as in the case of wrong site surgery
Analysis
Because the goal was to explore the understanding of respondents who believed that they knew the model, only respondents who said they were "very much" or "quite a bit" familiar with the model were analysed. There was no statistically significant difference between those who filled the questionnaire at the conference and those who filled it online, thus the samples were pooled for the analysis.
The analysis consisted of simple frequencies of endorsement for each proposed answer. Endorsement of an item that was compatible with Reason's model according to the author, and non-endorsement of an incompatible item, were treated as "correct" answers. Sums of endorsed compatible and incompatible items were computed, as well as the number of "correct" answers out of 23. The two ambiguous items were left out of this analysis.
Results
Sample characteristics
Forty-eight usable questionnaires were collected at the conference (4 others were incomplete), and 111 on the internet (11 others were empty, duplicated, only partially filled, included only "I have no idea" answers, or had all answer options checked). Eighty-five respondents (53.5%) stated that they were "very" (N = 45) or "quite" (N = 40) familiar with the Swiss cheese model. Only these respondents are reported on in the analyses.
Participants were 44 years old on average (SD 9, range 25 to 70, 5 missing), and comprised 42 women and 42 men (1 missing), from 31 countries representing all continents. Most job titles were in health care policy and quality management, but respondents included also health care professionals (doctors, nurses and pharmacists) and academic researchers. Sixty-three (74.1%) respondents had worked in quality management in the past 5 years, 31 (36.9%) in risk management, and 19 (22.6%) in safety science (several answers were allowed; one person did not answer).
Most respondents thought that the Swiss cheese model was "very useful" (44, 51.8%) or "quite useful" (32, 37.6%). At the end of the questionnaire, 9 (10.7%) rated themselves as very knowledgeable about patient safety, 63 (75.0%) as quite knowledgeable, and 12 (14.3) as only a little knowledgeable.
Interpretation of the model
Respondents endorsed on average 5.3 of 11 statements that were compatible with the Swiss cheese model (SD 2.2, range 1 to 10), and 2.0 of the 12 statements that were incompatible (SD 1.7, range 0 to 9). The mean number of "correct" answers was 15.3 (SD 2.3, range 10 to 21) out of 23 (66.5%). This was significantly more than 11.5 "correct" answers that would expected if answers were given at random (p < 0.001). Respondents gave on average 2.4 "correct" answers regarding the slice of cheese (out of 4), 2.7 "correct" answers about the holes (out of 5), 2.8 "correct" answers about the arrow (out of 4), 3.3 "correct" answers about the active error (out of 5), and 4.1 "correct" answers about improving safety (out of 5). None of the following variables were associated with the mean number of "correct" answers: sex, age, previous experience in quality management, risk management or safety science, familiarity with the model (very familiar versus quite familiar), perceived usefulness of the model, and knowledge of patient safety.
Specific items
Most respondents interpreted the slice of cheese as intended by J. Reason (barrier that protects patients from harm), and inferred correctly that this would include a defence that prevents the occurrence of an error (Table 1). However, only few recognised that procedures that alleviate consequences of an error may also appear as barriers. Majorities interpreted a hole as suggested by Reason – a weakness in defences, but only few respondents understood that a hole is either a latent error or an unsafe act. The most obvious interpretation of the arrow (path from hazard to harm) was chosen by only half of the respondents. The majority choice (series of events leading to an error) is not entirely correct, as it is patient harm, not an error, that is represented by Reason at the tip of the arrow (however, the error may be equivalent to patient harm, as in wrong site surgery). Only three out of ten respondents identified correctly an active error as one of the holes. Making the system safer by plugging a hole was correctly selected by most respondents, but the solution of adding a barrier (a slice of cheese) was not.
Discussion
This survey shows that among quality improvement professionals, the meaning of the Swiss cheese model of medical error is far from univocal. On average, respondents gave answers that were compatible with the model to about two thirds of the proposed statements. This is better than half – the proportion that would be expected by chance – but far from a general consensus. This suggests that invoking the Swiss cheese model will not necessarily lead to effective communication, even among quality and safety professionals.
There was substantial variability among respondents as to what the various features of the model represent. The murkiest notion appeared to be the representation of the medical error itself. Few of the respondents recognised that an active error is a type of weakness in defences against patient harm within the health care system, represented by a hole in the Swiss cheese model (a "hole" is either an active or a latent error). The model is almost too successful in placing emphasis on systemic causes of patient harm, as opposed to an individual's failure.
The variability in interpretations revealed in this survey is more understandable if one considers the evolution of Reason's model between 1990 and 2000. In the first rendition of the model (Figure 2), what was predicted was an accident, latent errors were placed as antecedents of the accident trajectory at the far left, and unsafe acts (i.e., active errors) were represented by a separate "slice" [2]. Two subsequent models place the set of barriers between harm and the patient (the "slices of Swiss cheese") in a more global context (Figure 3 and 4). In particular, these models attempt to show causal chains that lead up to patient harm. For the sake of clarity, it should be noted that these more complete models have not been dubbed "Swiss cheese models." The model of 1995 (Figure 3) shows a sequence of conditions and events leading to an accident, and defences and barriers are represented as intervening only after the occurrence of an error or a violation [3]. The model published in 1997 (Figure 4) depicts the Swiss cheese model as leading to human losses, not accidents [4]. This supports the view that patient safety interventions should focus on patient harm, rather than errors [8,9]. Importantly, the 1997 model also displays unsafe acts and workplace factors as orthogonal to the arrow leading from hazards to losses – presumably, each weakness in the system has its own set of causal or contributing factors. The current version of the Swiss cheese model (Figure 1), published in 2000, appears to be a simplification of the previous model, from which the causal pathways have been removed.
Figure 2 Reason's model published in 1990 (2).
Figure 3 Reason's model published in 1995 (3), as adapted by Vincent et al (10).
Figure 4 Reason's model published in 1997 (4).
While no model ever claims to represent fairly a complex reality, it is possible that the latest rendition of the Swiss cheese model has become too simplified to remain effective in promoting patient safety. More realistic alternatives include the model of 1995 (Figure 3), which clearly separates the event, or accident, from patient harm, and has remained in use [10], and that of 1997 (Figure 4), which suggests that the occurrence of a system failure cannot be easily represented by a simple linear sequence. In this Reason's model rejoins Haddon's matrix, a successful epidemiologic model for investigating injuries [11]. The relevance of Haddon's matrix for investigating medical mishaps has been recognised by others [8]. The integration of Reason's and Haddon's models may be a worthwhile next step toward a comprehensive model of patient safety.
More generally, the diversity of views documented in this study raises questions about the current status of a "culture of safety" among quality and safety professionals. A culture is a set of values, concepts and beliefs that are shared by a social group [12,13]. The Swiss cheese model is the leading candidate for a common understanding of how harmful events occur and how they can be prevented. Until most or all actors agree on what the model means, the emergence and dissemination of a shared culture of safety may prove difficult. The danger is that people today use the label "Swiss cheese model" without realising that its meaning varies from one person to the next.
This study has several limitations. The main concern is that the sample of respondents was self-selected, and may not represent fairly the broader community of patient safety and healthcare quality professionals. It is likely that those who were most interested and most knowledgeable about patient safety are over-represented among participants. Secondly, it is possible that respondents misrepresented their level of familiarity with the model, and that results would have been better among true experts. Nevertheless, the discrepancy between self-perceived familiarity with the model and variable interpretations of the model features is striking. Finally, the questionnaire was developed ad hoc, and its reliability and validity are untested.
In summary, this study has shown that quality and safety professionals vary considerably in their interpretation of various components of the Swiss cheese model applied to medical mishaps. This finding echoes the variability in interpretation that exists even for basic terms of patient safety, such as "incident," "error," "mishap," etc. [14]. Recent proposals of a comprehensive taxonomy of patient safety illustrate the necessity of a global conceptual model [15]. Good models and clear concepts are required for a common terminology, and a common terminology is a pre-requisite for effective communication and progress in the field of patient safety.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
The author conceived the study, pre-tested the questionnaire, collected and analysed the data, and wrote the paper.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
Maria-Julia Stonborough and Paul Kidd helped with data collection. Vincent Baujard created the online version of the questionnaire. No specific funding was obtained for this study.
==== Refs
Reason J Human error: models and management BMJ 2000 320 768 70 10720363 10.1136/bmj.320.7237.768
Reason J Human Error 1990 Cambridge: Cambridge University Press
Reason J Understanding adverse events: human factors Qual Health Care 1995 4 80 89 10151618
Reason J Managing the Risks of Organizational Accidents 1997 Aldershot, UK: Ashgate
Reason JT Carthey J de Leval MR Diagnosing "vulnerable system syndrome": an essential prerequisite to effective risk management Qual Health Care 2001 10 ii21 5 11700375
United States Department of Health and Human Services. Blood Safety Transcripts. Advisory Committee on Blood Safety and Availability, 11th meeting Accessed on October 22, 2005
Swiss Cheese Quiz Accessed on October 22, 2005
Layde PM Cortes LM Teret SP Brasel KJ Kuhn EM Mercy JA Hargarten SW Maas LA Patient safety efforts should focus on medical injuries JAMA 2002 287 1993 7 11960544 10.1001/jama.287.15.1993
McNutt RA Abrams R Arons DC Patient Safety Committee Patient safety efforts should focus on medical errors JAMA 2002 287 1997 2001 11960545 10.1001/jama.287.15.1997
Vincent C Taylor-Adams S Chapman EJ Hewett D Prior S Strange P Tizzard A How to investigate and analyse clinical incidents: Clinical Risk Unit and Association of Litigation and Risk Management protocol BMJ 2000 320 777 81 10720366 10.1136/bmj.320.7237.777
Haddon W A logical framework for categorizing highway safety phenomena and activity J Trauma 1972 12 193 207 5012817
Hudelson PM Culture and quality: an anthropological perspective Int J Qual Health Care 2004 16 345 6 15375093 10.1093/intqhc/mzh076
Westrum R A typology of organisational cultures Qual Saf Health Care 2004 13 ii22 ii27 15576687 10.1136/qshc.2003.009522
Weingart SN Beyond Babel: prospects for a universal patient safety taxonomy Int J Qual Health Care 2005 17 93 4 15772256 10.1093/intqhc/mzi029
Chang A Schyve PM Croteau RJ O'Leary DS Loeb JM The JCAHO patient safety event taxonomy: a standardized terminology and classification schema for near misses and adverse events Int J Qual Health Care 2005 17 95 105 15723817 10.1093/intqhc/mzi021
|
16280077
|
PMC1298298
|
CC BY
|
2021-01-04 16:31:50
|
no
|
BMC Health Serv Res. 2005 Nov 9; 5:71
|
utf-8
|
BMC Health Serv Res
| 2,005 |
10.1186/1472-6963-5-71
|
oa_comm
|
==== Front
BMC Infect DisBMC Infectious Diseases1471-2334BioMed Central London 1471-2334-5-991626643810.1186/1471-2334-5-99Research ArticleThe trend of susceptibilities to amphotericin B and fluconazole of Candida species from 1999 to 2002 in Taiwan Yang Yun-Liang [email protected] Shu-Ying [email protected] Hsiao-Hsu [email protected] Hsiu-Jung [email protected] Hospitals [email protected] Department of Biological Science and Technology, National Chiao Tung University, Hsinchu, Taiwan, China2 Laboratory for Mycopathogen, Chlamydia and Mycoplasma, Division of Laboratory Research and Development, Center for Disease Control, Taipei, Taiwan, China3 Division of Clinical Research, National Health Research Institutes, Miaoli, Taiwan, China2005 3 11 2005 5 99 99 8 3 2005 3 11 2005 Copyright © 2005 Yang et al; licensee BioMed Central Ltd.2005Yang 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
Candida species have various degrees of susceptibility to common antifungal drugs. The extent of resistance to amphotericin B and fluconazole of Candida glabrata isolates causing candidemia has been reported. Active surveillance may help us to monitor the trend of susceptibility to antifungal drugs and to determine if there is an emerging co-resistance to both drugs of Candida species, specifically, of C. glabrata in Taiwan.
Methods
The susceptibilities to amphotericin B and fluconazole of Candida species collected in 1999 and 2002 of the Taiwan Surveillance of Antimicrobial Resistance of Yeasts (TSARY) were determined by the microdilution method.
Results
The antifungal susceptibilities of 342 and 456 isolates collected from 11 hospitals participating in both TSARY 1999 and TSARY 2002, respectively, have been determined. The resistance rate to amphotericin B has increased from 0.3% in the TSARY1999 to 2.2% in the TSARY 2002. In contrast, the resistance rate to fluconazole has decreased from 8.8% to 2.2%. Nevertheless, significantly more C. glabrata isolates were not susceptible to fluconazole in the TSARY 2002 (47.4%) than that in the TSARY 1999 (20.8%). There were 9.8% and 11% of C. glabrata isolates having susceptible-dose dependent and resistant phenotype to fluconazole in the TSARY 1999, verse 45.3% and 2.1% in the TSARY 2002.
Conclusion
There was an increase of resistance rate to amphotericin B in C. glabrata. On the other hand, although the resistance rate to fluconazole has decreased, almost half of C. glabrata isolates were not susceptible to this drug. Hence, continuous monitoring the emerging of co-resistance to both amphotericin B and fluconazole of Candida species, specifically, of C. glabrata, will be an important early-warning system.
==== Body
Background
In the past decade, nosocomial yeast infections have increased globally. In Taiwan, the prevalence of nosocomial candidemia increased 16-fold from 1981 through 1993 [1,2]. In the United States, yeast infections rank as the fourth most common cause of nosocomial bloodstream infection [3,4]. Furthermore, candidemia contribute considerable mortality (31% to 38%), extend the length of hospital stay [5,6], and increase social cost due to lost productivity and disabling complications [7]. Consequently, the Taiwan Surveillance of Antimicrobial Resistance of Yeasts (TSARY) was initiated in 1999 for epidemiological study of yeast infections in Taiwan [8,9]
Candida species have various degrees of susceptibility to common antifungal agents. Candida lusitaniae is less susceptible to amphotericin B [10] while Candida krusei and Candida glabrata are less susceptible to fluconazole than other Candida species [11-14]. The extent of fluconazole resistance of C. glabrata isolates causing candidemia has been reported throughout the United States [15]. Furthermore, C. glabrata exhibits variable cross-resistance to the other triazoles, such as voriconazole and posaconazole [13,16-18] and amphotericin B became the next choice. The aim of this study is to investigate the trend of susceptibility to amphotericin B and fluconazole of Candida species in Taiwan from 1999 to 2002. Especially, we would like to determine if there is an emerging co-resistance to amphotericin B and fluconazole of Candida species, specifically, of C. glabrata, in Taiwan.
Methods
Organisms and media
Yeast isolates were collected from 11 hospitals participating in both TSARY 1999 and TSARY 2002 [9,19]. Isolates were stored frozen at -70°C in bead containing Microbank cryovials (PRO-LAB Diagnostics, Austin, TX, USA). At the end of the collection period, isolates were kept frozen and transported by an express delivery company to the laboratory at National Health Research Institutes (NHRI) within 24 hours. After their arrival, the isolates were first sub-cultured on to sabouraud dextrose agar (SDA, BBL, Becton Dickinson Cockeysville, MD, USA) to check for purity and identifications. Pure isolates were labeled and stored in vials containing 50% glycerol at -70°C for subsequent analyses.
Identification
The identification procedure of yeast isolates in the NHRI laboratory was performed as described previously [8]. In general, isolates identified as C. albicans by hospitals were first subjected to the germ tube assay in brain heart infusion (BHI, BBL) medium containing 10% fetal bovine serum (JR12003, JRH Biosciences, Australia) at 37°C for 2–3 hours [20]. Isolates positive in germ tube assay were checked for growth at 42°C to differentiate C. albicans from C. dubliniensis [21]. The VITEK Yeast Biochemical Card (YBC, bioMerieux, St. Louis, MI, USA) was then used to analyze isolates appearing to be negative by the germ tube assay in the NHRI laboratory and isolates identified as non-albicans Candida species by the hospitals. API-32C (bioMerieux) was used to assess the NHRI result when the VITEK-YBC showed less than 90% confidence.
Antifungal susceptibility testing
The minimum inhibitory concentration (MIC) to amphotericin B or fluconazole of each yeast isolate was determined by in vitro antifungal susceptibility testing according to the guidelines by the Clinical and Laboratory Standards Institute (CLSI, formerly NCCLS) [22]. The RPMI medium 1640 (31800-022, Invitrogen Corporation, Carlsbad, CA, USA) was used for dilution. Several strains from American Type Culture Collection, namely, ATCC 14053 C. albicans, ATCC 9003 C. glabrata, ATCC 6258 C. krusei, and ATCC20019 Candida parapsilosis were used as controls. The growth of each isolate was measured by a Spectra MAX Plus (Molecular Devices Cop. Sunnyvale, California, USA) after 48-hour incubation at 35°C. We also measured the MICs of some randomly-sampled isolates by Etest (AB Biodisk Solna, Sweden) to confirm our results by microdilution.
The interpretation of MICs was conducted according to the guidelines of the CLSI. The MICs to amphotericin B and fluconazole were defined as the lowest concentration of amphotericin B and fluconazole to reduce the turbidity of cells to greater than 95% and 50%, respectively. For amphotericin B, isolates with MIC ≧ 2 μg/ml were considered to be resistant, whereas those with MIC ≦ 1 μg/ml were susceptible. For fluconazole, isolates with MIC ≧ 64 μg/ml were considered resistant, while those with MIC ≦ 8 μg/ml were susceptible. Isolates with MICs between 16 and 32 μg/ml were susceptible-dose dependent. The MICs of 50% and 90% of the total population were defined as MIC50 and MIC90. For any species with less than ten, the MIC50 and MIC90 were not showed.
Database and analysis
The database for this study contained the following characteristic information of each submitted isolate: hospital origin, location and type of the hospital, identification and source of the isolate. The statistic significance of the differences in frequencies and proportions was determined by the chi-square test with Yates' correction. A p value of ≦ 0.05 was considered statistically significant.
Results
Distribution of Candida species
The distribution of Candida species was similar in both surveys. Candida albicans was the most common species consisting 37.7% of the total isolates in the TSARY 1999 and 41.2% in the TSARY 2002. Candida tropicalis (28.7% in 1999 vs. 28.3% in 2002) and C. glabrata (24% in 1999 vs. 20.8% in 2002) were the two most common non-albicans Candida species, followed by C. parapsilosis (6.4% in 1999 vs. 7.9% in 2002), C. krusei (1.2% in 1999 vs. 1.1% in 2002), and others (2% in 1999 vs. 0.7% in 2002). When classified according to the sources, isolates from urine, sputum, blood, wound, and others were 143 (41.8%), 101 (29.5%), 30 (8.8%), 26 (7.6%), and 42 (12.3%), respectively, in the TSARY 1999 verse 186 (40.8%), 111 (24.3%), 50 (11%), 20 (4.4%), and 89 (19.5%), respectively, in the TSARY 2002.
Susceptibilities to amphotericin B
The susceptibilities to amphotericin B are shown in Table 1. A total of 10 isolates (2.2%) were resistant to amphotericin B in the TSARY 2002, whereas only one (0.3%) in the TSARY 1999 (p < 0.05). Of these 11 amphotericin B resistant isolates, 9 were non-albicans Candida species, including 5 C. krusei, 3 C. glabrata, and 1 C. tropicalis. In general, C. krusei was less susceptible to amphotericin B than other species.
Table 1 The Susceptibilities of Candida Species to Amphotericin B
TSARY 1999 MIC μg/ml cal ctr cgl cpa ckr Others Total
≦ 0.25 19 (14.7)a 5 (5.1) 5 (6.1) 5 (22.7) 0 3 (42.8) 37 (10.8)
0.5 81 (62.8) 57 (58.2) 52 (63.4) 8 (36.4) 1 (25) 2 (28.6) 201 (58.8)
1 29 (22.5) 36 (36.7) 25 (30.5) 9 (40.9) 2 (50) 2 (28.6) 103 (30.1)
2 0 0 0 0 1 (25) 0 1 (0.3)
Total 129 98 82 22 4 7 342
MIC50 μg/ml 0.5 0.5 0.5 0.5 1.0 0.5 0.5
MIC90 μg/ml 1.0 1.0 1.0 1.0 2.0 1.0 1.0
TSARY 2002 MIC μg/ml cal ctr cgl cpa ckr Others Total
≦ 0.25 8 (4.2) 1 (0.8) 0 3 (8.3) 0 0 12 (2.6)
0.5 122 (64.9) 70 (54.2) 17 (17.9) 17 (47.2) 1 (20) 1 (33.3) 228 (50)
1 56 (29.8) 57 (44.2) 75 (78.9) 16 (44.5) 0 2 (66.7) 206 (45.2)
2 2 (1.1) 1 (0.8) 3 (3.2) 0 4 (80) 0 10 (2.2)
Total 188 129 95 36 5 3 456
MIC50 μg/ml 0.5 0.5 1 0.5 2 1 0.5
MIC90 μg/ml 1 1 1 1 2 1 1
cal, C. albicans; ctr, C. tropicalis; cgl, C. glabrata; cpa, C. parapsilosis; ckr, C. krusei
anumber of isolates (%)
Susceptibilities to fluconazole
The susceptibilities to fluconazole of Candida species are shown in Table 2. In the TSARY 1999, a total of 289 (84.5%), 23 (6.7%), and 30 (8.8%) isolates were susceptible, susceptible-dose dependent, and resistant to fluconazole, respectively, whereas in the TSARY 2002, there were 391 (85.5%), 55 (12%), and 10 (2.2%). The MIC50 and MIC90 of these isolates in the TSARY1999 were 2 μg/ml and 16 μg/ml, respectively, and in the TSARY 2002, they were 1 μg/ml and 16 μg/ml. In the TSARY 1999, 12 (12.2%) C. tropicalis, 9 (11%) C. glabrata, 5 (3.9%) C. albicans, and 4 (100%) C. krusei, while in the TSARY 2002, 4 (2.1%) C. albicans, 3 (60%) C. krusei, and 2 (2.1%) C. glabrata were resistant to fluconazole. Fewer isolates in the TSARY 2002 were resistant to fluconazole than that in the TSARY 1999 (p < 0.05). In contrast, more isolates from the TSARY 2002 were susceptible-dose dependent than that in the TSARY 1999 (p < 0.05). Consequently, there were similar portions of isolates susceptible to fluconazole in both surveys. Nevertheless, there were less isolates with MICs ≦ 2 μg/ml to fluconazole in the TSARY 1999 (71.6%, 207/289) than in the TSARY 2002 (81.8%, 320/391) (p < 0.05). Finally, in the TSARY 1999, 82 (24%) of isolates had MICs between 4 and 8 μg/ml to fluconazole. It was down to 71 (15.6%) in the TSARY 2002.
Table 2 The Susceptibilities of Candida Species to Fluconazole
TSARY 1999 MIC μg/ml cal ctr cgl cpa ckr Others Total
S 121 (93.8)a 77 (78.6) 65 (79.2) 21 (95.5) 0 5 (71.4) 289 (84.5)
SDD 3 (2.3) 9 (9.2) 8 (9.8) 1 (4.5) 0 2 (28.6) 23 (6.7)
R 5 (3.9) 12 (12.2) 9 (11) 0 4 (100) 0 30 (8.8)
Total 129 98 82 22 4 7 342
MIC50 μg/ml 0.25 2 4 1 ND ND 2
MIC90 μg/ml 4 64 64 4 ND ND 16
TSARY 2002 MIC μg/ml cal ctr cgl cpa ckr Others Total
S 178 (94.7) 124 (96.1) 50 (52.6) 36 (100) 1 (20) 2 (66.7) 391 (85.8)
SDD 6 (3.2) 5 (3.9) 43 (45.3) 0 1 (20) 0 55 (12)
R 4 (2.1) 0 2 (2.1) 0 3 (60) 1 (33.3) 10 (2.2)
Toal 188 129 95 36 5 3 456
MIC50 μg/ml 0.25 1 8 1 ND ND 1
MIC90 μg/ml 1 4 32 2 ND ND 16
cal, C. albicans; ctr, C. tropicalis; cgl, C. glabrata; cpa, C. parapsilosis; ckr, C. krusei
anumber of isolates (%), S, susceptible; SDD, susceptible-dose dependent; R, resistant ND, not showed due to small number of isolates
Discussion
The trend of susceptibilities to antifungal drugs of Candida species from 1999 to 2002 has been determined in this study. As expected, C. krusei had the highest resistance rate to fluconazole among Candida species tested, which is consistent with previous reports [9,11]. In contrast, all C. parapsilosis isolates were susceptible to fluconazole, which is also consistent with previous reports that C. parapsilosis is the most susceptible species to fluconazole [9,18,23,24]. Though the overall resistance rate to fluconazole has decreased from 8.8% to 2.2%, there were significantly more C. glabrata isolates not susceptible to fluconazole in the TSARY 2002 than that in the TSARY 1999. Overexpression of CgCDR1, CgCDR2, and CgSNQ2-encoded efflux pumps has been shown to be a major mechanism contributing to the drug resistance [25-27]. It would be interesting to investigate the molecular mechanisms of drug resistance of those clinical resistant isolates.
Recently, triazoles have been developed as the new savior to the issue of drug resistance in Candida infection. Nevertheless, C. glabrata exhibits variable cross-resistance among triazoles [9,18,23]. Thus, amphotericin B appears to be the choice for treating systemic infections caused by this species. However, along with the increased use of amphotericin B, 20% and 36% of C. glabrata isolates from North America and Latin America, respectively, were reported to be resistant [23]. These data suggest that co-resistance to amphotericin B and fluconazole of C. glabrata species may become a problem for clinical therapy worldwide. In our study, we found only three C. glabrata isolates resistant to amphotericin B, which is lower than what has been reported. In that study, 20% of C. glabrata causing candidemia collected in Taiwan in 2003 were resistant to amphotericin B [16]. Coincidently, more C. glabrata isolates in the TSARY 2002 (78.9%) had the MICs of amphotericin B at 1 μg/ml than that in the TSARY 1999 (30.5%). Hence, periodic surveillance is needed to closely monitor the trends of susceptibility to antifungal drugs and for early detection of the newly emerging co-resistance to amphotericin B and fluconazole of Candida species, especially, of C. glabrata.
Abbreviations used
TSARY, Taiwan Surveillance of Antimicrobial Resistance of Yeasts; NHRI, National Health Research Institutes; SDA, sabouraud dextrose agar; BHI, brain heart infusion; YBC, Yeast Biochemical Card; MIC, minimum inhibitory concentration; NCCLS, National Committee of Clinical Laboratory Standards; CLSI, Clinical and Laboratory Standards Institute.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
YLY and HJL design the study and drafted the manuscript. HHC conduct the experiments with contribution with SYL. TSARY Hospitals provided isolates.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We would like to thank Bristol Myers Squibb and Pfizer for supplying the amphotericin B and fluconazole, respectively. We also wish to thank the 11 participating hospitals for providing clinical isolates and information regarding to those isolates. They are Buddhist Tzu-Chi General Hospital, Hua-Lien Hospital, DOH, the Executive Yuan, Kaohsiung Military Hospital, Kaohsiung Medical College Chung-Ho Memorial Hospital, Kuan-Tien General Hospital, Lo-Hsu Foundation Inc. Lo-Tung Poh Ai Hospital, St. Mary Hospital, Tri Service General Hospital, Veterans General Hospital-Taichung, Veterans General Hospital-Kaohsiung, Zen Ai General Hospital. This work was in part supported by the grants DOH93-DC-1101, and DOH94-DC-1102 from Center for Disease Control and CL-93-PP-06 from National Health Research Institutes.
==== Refs
Hung CC Chen YC Chang SC Luh KT Hsieh WC Nosocomial candidemia in a university hospital in Taiwan J Formos Med Assoc 1996 95 19 28 8640090
Chen YC Chang SC Sun CC Yang LS Hsieh WC Luh KT Secular trends in the epidemiology of nosocomial fungal infections at a teaching hospital in Taiwan, 1981 to 1993 Infect Control Hosp Epidemiol 1997 18 369 375 9154483
Pfaller MA Jones RN Messer SA Edmond MB Wenzel RP National surveillance of nosocomial blood stream infection due to species of Candida other than Candida albicans: frequency of occurrence and antifungal susceptibility in the SCOPE Program. SCOPE Participant Group. Surveillance and Control of Pathogens of Epidemiologic Diagn Microbiol Infect Dis 1998 30 121 129 9554180 10.1016/S0732-8893(97)00192-2
Beck-Sague C Jarvis WR Secular trends in the epidemiology of nosocomial fungal infections in the United States, 1980–1990. National Nosocomial Infections Surveillance System J Infect Dis 1993 167 1247 1251 8486965
Leleu G Aegerter P Guidet B Systemic candidiasis in intensive care units: a multicenter, matched-cohort study J Crit Care 2002 17 168 175 12297992 10.1053/jcrc.2002.35815
Wey SB Mori M Pfaller MA Woolson RF Wenzel RP Hospital-acquired candidemia. The attributable mortality and excess length of stay Arch Intern Med 1988 148 2642 2645 3196127 10.1001/archinte.148.12.2642
Tortorano AM Caspani L Rigoni AL Biraghi E Sicignano A Viviani MA Candidosis in the intensive care unit: a 20-year survey J Hosp Infect 2004 57 8 13 15142710 10.1016/j.jhin.2004.01.017
Lo HJ Ho AH Ho M Factors accounting for mis-identification of Candida species J Microbiol Immunol Infect 2001 34 171 177 11605807
Yang YL Ho YA Cheng HH Ho M Lo HJ Susceptibilities of Candida species to amphotericin B and fluconazole: the emergence of fluconazole resistance in Candida tropicalis Infect Control Hosp Epidemiol 2004 25 60 64 14756222
Hadfield TL Smith MB Winn RE Rinaldi MG Guerra C Mycoses caused by Candida lusitaniae Rev Infect Dis 1987 9 1006 1012 3317728
Akova M Akalin HE Uzun O Gur D Emergence of Candida krusei infections after therapy of oropharyngeal candidiasis with fluconazole Eur J Clin Microbiol Infect Dis 1991 10 598 599 1915408 10.1007/BF01967286
Orozco AS Higginbotham LM Hitchcock CA Parkinson T Falconer D Ibrahim AS Mechanism of fluconazole resistance in Candida krusei Antimicrob Agents Chemother 1998 42 2645 2649 9756770
Yang YL Cheng HH Lo HJ In vitro activity of voriconazole against Candida species isolated in Taiwan Int J Antimicrob Agents 2004 24 294 296 15325436 10.1016/j.ijantimicag.2004.01.014
Piemonte P Conte G Flores C Barahona O Araos D Alfaro J Emergency of fluconazole-resistant infections by Candida krusei and Candida glabrata in neutropenic patients Rev Med Chil 1996 124 1149 9197032
Pfaller MA Diekema DJ Messer SA Boyken L Hollis RJ Activities of fluconazole and voriconazole against 1,586 recent clinical isolates of Candida species determined by Broth microdilution, disk diffusion, and Etest methods: report from the ARTEMIS Global Antifungal Susceptibility Program, 2001 J Clin Microbiol 2003 41 1440 1446 12682127 10.1128/JCM.41.4.1440-1446.2003
Hsueh PR Lau YJ Chuang YC Wan JH Huang WK Shyr JM Antifungal susceptibilities of clinical isolates of Candida species, Cryptococcus neoformans, and Aspergillus species from Taiwan: surveillance of multicenter antimicrobial resistance in Taiwan program data from 2003 Antimicrob Agents Chemother 2005 49 512 517 15673726 10.1128/AAC.49.2.512-517.2005
Pfaller MA Messer SA Boyken L Hollis RJ Rice C Tendolkar S In vitro activities of voriconazole, posaconazole, and fluconazole against 4,169 clinical isolates of Candida spp. and Cryptococcus neoformans collected during 2001 and 2002 in the ARTEMIS global antifungal surveillance program Diagn Microbiol Infect Dis 2004 48 201 205 15023430 10.1016/j.diagmicrobio.2003.09.008
Pfaller MA Diekema DJ Jones RN Sader HS Fluit AC Hollis RJ International surveillance of bloodstream infections due to Candida species: frequency of occurrence and in vitro susceptibilities to fluconazole, ravuconazole, and voriconazole of isolates collected from 1997 through 1999 in the SENTRY antimicrobial surveillance program J Clin Microbiol 2001 39 3254 3259 11526159 10.1128/JCM.39.9.3254-3259.2001
Yang YL Li SY Cheng HH Lo HJ Susceptibilities to amphotericin B and fluconazole of Candida species in TSARY 2002 Diagn Microbiol Infect Dis 2005 51 179 183 15766603 10.1016/j.diagmicrobio.2004.11.004
Larone DH Laboratory Procedures Medically Important Fungi: A Guide to Identification 1995 New York: ASM Press 209 224
Sullivan D Coleman D Candida dubliniensis: characteristics and identification J Clin Microbiol 1998 36 329 334 9466736
Clinical and Laboratory Standards Institute (formly NCCLS) Reference method for broth dilution antifungal susceptibility testing of yeasts; approved standard M27A, Wayne, PA 1997
Pfaller MA Espinel-Ingroff A Jones RN Clinical evaluation of the Sensititre YeastOne colorimetric antifungal plate for antifungal susceptibility testing of the new triazoles voriconazole, posaconazole, and ravuconazole J Clin Microbiol 2004 42 4577 4580 15472311 10.1128/JCM.42.10.4577-4580.2004
Yang YL Cheng HH Ho YA Hsiao CF Lo HJ Fluconazole resistance rate of Candida species from different regions and hospital types in Taiwan J Microbiol Immunol Infect 2003 36 187 191 14582563
Bennett JE Izumikawa K Marr KA Mechanism of increased fluconazole resistance in Candida glabrata during prophylaxis Antimicrob Agents Chemother 2004 48 1773 1777 15105134 10.1128/AAC.48.5.1773-1777.2004
Sanglard D Ischer F Calabrese D Majcherczyk PA Bille J The ATP binding cassette transporter gene CgCDR1 from Candida glabrata is involved in the resistance of clinical isolates to azole antifungal agents Antimicrob Agents Chemother 1999 43 2753 2765 10543759
Sanguinetti M Posteraro B Fiori B Ranno S Torelli R Fadda G Mechanisms of azole resistance in clinical isolates of Candida glabrata collected during a hospital survey of antifungal resistance Antimicrob Agents Chemother 2005 49 668 679 15673750 10.1128/AAC.49.2.668-679.2005
|
16266438
|
PMC1298299
|
CC BY
|
2021-01-04 16:28:16
|
no
|
BMC Infect Dis. 2005 Nov 3; 5:99
|
utf-8
|
BMC Infect Dis
| 2,005 |
10.1186/1471-2334-5-99
|
oa_comm
|
==== Front
BMC Med EthicsBMC Medical Ethics1472-6939BioMed Central London 1472-6939-6-111627766310.1186/1472-6939-6-11DebateEvidence-based ethics? On evidence-based practice and the "empirical turn" from normative bioethics Goldenberg Maya J [email protected] Department of Philosophy, Michigan State University, 503 South Kedzie Hall, East Lansing, Michigan, USA2005 8 11 2005 6 11 11 3 8 2005 8 11 2005 Copyright © 2005 Goldenberg; licensee BioMed Central Ltd.2005Goldenberg; 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 increase in empirical methods of research in bioethics over the last two decades is typically perceived as a welcomed broadening of the discipline, with increased integration of social and life scientists into the field and ethics consultants into the clinical setting, however it also represents a loss of confidence in the typical normative and analytic methods of bioethics.
Discussion
The recent incipiency of "Evidence-Based Ethics" attests to this phenomenon and should be rejected as a solution to the current ambivalence toward the normative resolution of moral problems in a pluralistic society. While "evidence-based" is typically read in medicine and other life and social sciences as the empirically-adequate standard of reasonable practice and a means for increasing certainty, I propose that the evidence-based movement in fact gains consensus by displacing normative discourse with aggregate or statistically-derived empirical evidence as the "bottom line". Therefore, along with wavering on the fact/value distinction, evidence-based ethics threatens bioethics' normative mandate. The appeal of the evidence-based approach is that it offers a means of negotiating the demands of moral pluralism. Rather than appealing to explicit values that are likely not shared by all, "the evidence" is proposed to adjudicate between competing claims. Quantified measures are notably more "neutral" and democratic than liberal markers like "species normal functioning". Yet the positivist notion that claims stand or fall in light of the evidence is untenable; furthermore, the legacy of positivism entails the quieting of empirically non-verifiable (or at least non-falsifiable) considerations like moral claims and judgments. As a result, evidence-based ethics proposes to operate with the implicit normativity that accompanies the production and presentation of all biomedical and scientific facts unchecked.
Summary
The "empirical turn" in bioethics signals a need for reconsideration of the methods used for moral evaluation and resolution, however the options should not include obscuring normative content by seemingly neutral technical measure.
==== Body
Background
The increase in empirical methods of research in bioethics (or "empirical ethics") over the last two decades is typically perceived as a welcomed broadening of the discipline, with increased integration of social and life scientists into the field and ethics consultants into the clinical setting [1]. Evidence-based ethics is the newest empirical approach to bioethics inquiry, and while this method is still underdeveloped and the growing body of literature still small, there is good reason to expect this approach to really "take off" given the currency of "evidence based" approaches in so many professional disciplines. This paper is an effort to temper the momentum of the evidence-based movement and to reject its proliferation into bioethics. By examining the norms and implications of evidence-based practice in medicine, I aim to demonstrate that an evidence-based approach is incompatible with bioethics' normative mandate and therefore evidence-based ethics should not be pursued.
Empirical research in bioethics
Empirical research in bioethics (or "empirical ethics") is "the application of research methods in the social sciences (such as anthropology, epidemiology, psychology, and sociology) to the direct examination of issues in [bioethics]" [2]. Empirical approaches describe (rather than prescribe) "particular state[s] of affairs that [have] some moral or ethical relevance" [2] and are thought to enrich bioethics by calling attention to the social, cultural, and cross-cultural aspects of morality accessed via the opinions, interests and beliefs of patients, families, physicians, nurses and others involved in care-giving [3]. For example, empirical research can help describe cultural beliefs about the appropriateness of informing the patient of a diagnosed life-threatening illness, which will inform deliberation about the extent to which it is morally important for clinicians to provide comprehensive information to patients in different cultural contexts [4]. Similarly, empirical research can delineate popular attitudes and experiences related to contentious issues such as abortion, cloning, stem-cell research, and physician-assisted suicide for consideration in discussions and policy formulations [4]. Empirical work can also map the effects of particular interventions aimed at improving how clinicians or policy makers attempt to meet ethical obligations, such as whether a particular method of presenting health related information to a patient actually improves the patient's understanding of her circumstances and the quality of informed consent [4].
It is likely because proponents of empirical approaches to bioethics focus on differentiating this area of bioethics from normative ethics that the literature tends to use such loose overarching descriptors as "an amalgam of empirical contributions" [1] and "methodological roots in social sciences...to gather quantitative and qualitative data about ethical issues" [1] when characterising empirical ethics. The ease at which empirical ethics is generalised, and the differences between the various social scientific disciplines represented under this heading glossed over, may also be assisted by the common presumption that empirical research presents "only the facts". This understanding leads to underappreciation of the values typical to each discipline as well as the beliefs of the individual practitioner that influence the data gathering and interpretation [5]. While presumably no one would deny the different orientations, research agendas, and methods, of, say, psychology and sociology, the emphasis placed on similarities or shared features in order explain the novelty of empirical ethics and distinguish it from philosophical approaches to bioethics takes important attention away from the relevant differences between the social scientific disciplines that may render some empirical approaches incompatible with the goals of ethical inquiry. This paper focuses on one method of empirical ethics – evidence-based bioethics, which is grounded in clinical epidemiology and supported by the discipline's most distinctive application, evidence-based medicine.
Discussion
Evidence based medicine and the evidence-based movement
While evidence-based ethics arises within the momentum of what has been called "the empirical turn in bioethics" [1] – the increased interest in empirical research in bioethics – it draws unique content from the evidence-based movement that began in medicine only a decade and a half ago in the form of "evidence-based medicine" and then exploded into other professional disciplines. The term "evidence based medicine" was first introduced in a ubiquitous 1992 publication of the Journal of the American Medical Association as a "new paradigm" in medical education and practice by a group of professors of clinical epidemiology, medical informatics, and biostatistics at McMaster University calling themselves "The Evidence Based Medicine Working Group" [6]. The evidence-based medicine movement diagnosed the problems of medical error and wasteful healthcare spending as stemming from the prevalent use of unestablished medical interventions and proposed to remedy these difficulties by way of a decision-making technology that would eschew unsystematic and so-called "intuitive" methods of individual clinical experience in favour of a more scientifically rigorous approach. According to the dictates of evidence-based decision-making, clinical decisions should be based on the best available scientific evidence and "identifying the best evidence means using epidemiological and biostatistical ways of thinking" [7]. The methodological privileging of outcomes measures, statistical analysis, and indexes of aggregate behaviour that characterises clinical epidemiology serves to distinguish evidence-based medicine from traditional medicine, as the latter is charged with relying on unsystematic observations, medical intuition, pathophysiologic principles, and clinical experience [8].
While numerous techniques have been put in place to facilitate the systematic management, evaluation, and application of clinical data into evidence-based medical practice, the most distinctive technology is the hierarchy of evidence, a pre-graded ranking of experimental methodologies. Evidence-based medicine proponents strongly hold that the trustworthiness or validity of evidence is a function of the design of the study from which the evidence is obtained [9,10], and so the desire to use only the "best evidence from clinical research" in the management of individual patients [11] has resulted in elaborate classificatory schemes for ranking the value of different types of studies. Among the numerous published formulations [12,13], there is a consistent placement of randomised controlled trials or the systematic review of them at the top, retrospective studies well down the list, and clinical anecdotes are seen as providing little if any evidence for the value of intervention.
Evidence as accumulated data has been made widely and easily available to clinicians, educators, actuaries, and medical funding bodies by evolving information technologies such as electronic databases and systematic reviews of clinical trials. The political and professional capital of evidence-based medicine cannot be overstated, as this evidence-based practice is supposed to increase professional responsibility and accountability, improve patient care, and make managed care and medical research more cost effective by ensuring that only the most promising technologies are funded. The combined picture of evidence-based medicine as ethically driven to improve patient care, fiscally responsible, and technologically up-to-date likely drove the rapid integration of the movement into medicine, where fifteen years after the Evidence Based Medicine Working Group first formed, evidence-based medicine is now common parlance within health care. As a burgeoning institution, academic centres and journals dedicated to evidence-based medicine's advancement have been established with much fanfare, and the evidence-based movement has moved into numerous other fields, including nursing, public health, education and social work. The promise of the evidence-based movement to provide a systematic method for determining best practices has been so enthusiastically adopted in these fields that it is hardly surprising to find it generating attention as a promising new approach to bioethics in the form of "evidence-based ethics".
Bioethics
Since its evolution into a distinct discipline, bioethics has typically employed the analytic methods of Anglo-American philosophy to answer the ethical questions that arise in health care. While the methods have diversified considerably from the deductive approaches of "applied ethics" that still typify the field to include casuist, contextualist, and reflexive methods, the field maintains a normative mandate. It is this self-understanding, and the worry that empirical approaches to bioethics waver on the fact/value distinction, that encouraged a significant antagonism toward empirical research in bioethics that is only beginning to subside [1]. Despite numerous contestations of this bifurcation [14], the descriptive and prescriptive sciences are still regarded as quite distinct entities.
Some commentators on the "empirical turn in bioethics" have regarded the interest in empirical research as representing a loss of confidence in the typical normative and analytic methods of bioethics [15]. While many describe those "typical" methods inaccurately – offering a "straw man" account of applied ethics where absolutely no empirical considerations are included in the deductive process of ethical deliberation, for example, [16] – they are at least correct in recognising a felt ambivalence regarding the possibility of negotiating competing values in a pluralist society that respects difference. The technique of "evidence-based decision-making" offers what seems like a solution to this so-called "postmodern" problem, as it proposes to ground decisions in something concrete and universal, namely the evidence. The allure of evidence should not be underappreciated, as it is thought to be able to assist us in seeing past our habits, biases, and mistakes to decipher "best practices". The rapid ascendancy of the evidence-based movement, which started in medicine and quickly spread to other professional disciplines, speaks to the movement's enormous appeal. Even the popularity of the CSI television series – which depicts "evidence-based" police work par excellence – demonstrates how the stability, fairness, and truth of "the evidence" have captured our imagination. There are considerable difficulties with an "evidence-based" approach to bioethics, however, that require consideration.
Evidence-based ethics
Evidence-based ethics has been defined in the literature as follows:
As in medical decisions based on evidence-based medicine, ethical decisions based on evidence-based ethics would involve conscientious and judicious use of the best evidence relevant to the care and prognosis of the patient to promote better informed and better justified ethical decision making [17].
What this actually entails in practice is somewhat vague, as there are numerous ways in which empirical research can inform ethical decision making, numerous types of evidence that are relevant to the care and prognosis of patients, and numerous measures of best evidence. What is clear, however, is evidence-based ethics' close methodological proximity to evidence-based medicine, as the language of "conscientious and judicious use of best evidence" is recognisably lifted from the early programmatic literature on evidence-based medicine [8]. The implications of this relationship are the focus of this paper, as evidence-based medicine offers a distinct accounting of the nature of evidence, what evidence counts, and what role the evidence plays in the decision-making process. Understanding evidence-based ethics requires comprehension of the evidence-based approach to medicine.
Jeremy Sugarman, a known supporter of evidence-base ethics, argues that in both medical and ethical investigations, "it is important to 'raise the bar' on what evidence is acceptable to determine the most effective approaches" [18]. In both cases, he argues, evidence derived from randomised trials has the most utility [18]. Sugarman's appeal for rigorous methods and his attention to experimental methodology recall the hierarchy of evidence's consistent privileging of randomised controlled trials and systematic review of these trials over less objective measures such as surveys or qualitative research. The founding of evidence based medicine by clinical epidemiologists and biostatisticians should explain this methodological privileging, as randomised controlled trials produce the clinical data required for health outcomes research.
Keeping in mind evidence-based ethics' subscription to the "evidence based" doctrine, it becomes apparent that the term "evidence-based ethics" has been misunderstood and misused by some of its alleged proponents. In Robert Jansen's paper, "Evidence-Based Ethics and the Regulation of Reproduction" [19], the author uses the term to mean the testing of ethical arguments, statements, and the background assumptions informing those arguments, by means of empirical research. Jansen argues that Canada's prohibitions on sex selection for human reproduction relies on the untested empirical claim that sex selection often leads to some index of family dysfunction. He finds it ironic that Canada insists on evidence-based approaches for medical services but not for the social restrictions on reproductive medicine proposed by the Report of the Royal Commission on New Reproductive Technologies, which, he claims, made determinations of "women's true interests" without properly surveying the relevant attitudes and behaviours exhibited by the public. Against what Jansen perceived as the Commission's "hijacking" of ethical questions and their treatment of empirically verifiable hypotheses about the social consequences of permissive policies as "self-evident moral truths", he recommends a publicly accountable empirical approach that encourages debate and the determination of facts.
Jansen's understanding of "evidence based ethics" seems to be no different from the empirical ethics already in circulation insofar as it serves to inform moral deliberation (and therefore does not introduce a new empirical/ethics relation) [20]. It is worth noting that even prior to the incipience of empirical ethics, empirical content always informed ethical deliberation, whether to determine the actual or probable consequences of actions for consequentialist reasoning or to specify the norms of deontological consideration. In bioethics, surveys or in-depth interviews that gauge patients' or clinicians' attitudes or behaviours often serve as the data that philosophically-trained bioethicists reflect on in order to draw moral conclusions [21].
The evidence-based medicine hierarchy of evidence's maligning of the very techniques that empirical ethics so often employs suggests dissimilarity between empirical ethics and evidence-based bioethics. The surveys and in-depth interviews that are commonly used to determine the attitudes and behaviours of patients, clinicians, or the general public regarding bioethical issues are less valued and are ranked lower than the carefully controlled and quantified evidence that is derived from randomised controlled trials and other more objective methods. This suggests evidence-based ethics to be a distinct moment within the "empirical turn in bioethics" rather than, as Pascall Borry and colleagues' historical account seems to suggests, more of the same [1].
The second sense in which evidence-based ethics is used is as "the necessary grounding of ethical decisions in the best available scientific evidence" [1]. Jon Tyson's [22] and Terri Major-Kincade and colleagues' [17] work on clinical determinations of whether or not to treat severely disabled premature newborns enlist this use of the term "evidence-based bioethics", which I read to be a more accurate interpretation of the term because of its consistency with the methods of evidence-based medicine. It has already been discussed that evidence-based medicine is typified by the systematic introduction of scientific proof in healthcare interventions. Health care practices are thought to surely improve by means of decision-making based on a careful appraisal of the best available scientific evidence [23]. Tyson's and Major-Kincade et al.'s work offers decision-making techniques for determining whether or not to treat the patient that rely almost exclusively on the projected survival and disability outcomes of these infants. Major-Kincade et al. even employ a controlled trial to demonstrate the efficacy of their educational curriculum for teaching evidence-based ethics to NICU residents.
Tyson describes evidence-based ethics as involving multiple considerations in its determination of what constitutes "reasonable care" that include: (i) the quality of evidence available; (ii) the identified benefits, hazards, and costs of treatment; and (iii) the values and preferences of the parent or surrogate. In Major-Kincade et al's complementary paper detailing the implementation of an evidence-based ethics educational intervention, the "evidence" was specified to mean mortality and disability outcomes for infants that receive intensive care.
Given that Tyson claims to appreciate that treatment decisions for extremely premature infants involve highly complex ethical issues and multiple considerations, it comes as a surprise when he proposes, in the end, an algorithm [24] for instances of "mandatory", "unreasonable", and "optional" treatment based entirely on the projected outcomes (that is, survival rates and disability-free years) for neonates of particular birth weights, gestational ages, and health conditions. Even the professed importance of considering the parents or surrogates' values and preferences is limited to situations where the infant's clinical indicators fit her into the category of "optional" treatment. While the description of the multiple considerations that go into evidence-based ethical decision-making sounded reasonably comprehensive at first glance, certain limitations on how evidence is understood, what constitutes a "benefit" or a "harm" and who determines and measures them, and even when the parents' values play in, all narrow the deliberative process to a decision based on projected outcomes and an imposed cost per value calculation of Quality Adjusted Life Years and Disability Adjusted Life Years relative to financial cost of treatment. Mandatory treatment, for example, occurs when there is "credible evidence that benefits outweigh burdens" [22], with is no mention of who determines these criteria and how they are measured. These determinations were formulated against the backdrop of standardised clinical protocols being simply assumed to be preferable, more transparent, and fairer than case-by-case decision-making. These assumptions will soon be demonstrated to be consistent with the "epidemiological and biostatistical ways of thinking" that the founders of evidence-based medicine so strongly promoted.
The feature of Tyson's and Major-Kincade et al.'s methods that truly exemplify an evidence-based approach is that rather than having a wide range of empirical evidence inform ethical decision making (as is seen in typical accounts of empirical ethics), their techniques use scientific evidence (narrowly construed) to determine right action. Against Sugarman's claim that "empirical research [into bioethics] will not answer the ought question of bioethics" [4], evidence-based ethics seems to do just that. The slide from "is" to "ought" has already been noticed in evidence-based medicine. While the is of evidence-based medicine is that science is producing new and better ways of predicting, detecting and treating disease than were once even imaginable, the ought is that its advocates believe that clinicians ought to be responsible for keeping up to date with these advances and ought to be prepared to offer them to patients. Brian Haynes, one of the founders of the movement who has noticeably tempered his proclamations in recent years about the transformative ability of evidence-based medicine, has noted that "evidence-based medicine has taken on the tones of a moral imperative [even though] it is premature to get very preachy about the ought of evidence-based medicine" [25]. Another similarity between evidence-based medicine and evidence-based ethics is that the scope of "scientific evidence" is narrowed to exclude most forms of social scientific and qualitative evidence and is limited instead almost exclusively to medical outcomes, which is the evidence of choice according to the methodological hierarchies of evidence-based medicine.
The evidence-based doctrine problematically assumes that the presence of reliable evidence ensures that better decisions will be made. Medical decision-making, however, draws upon a broad spectrum of knowledge (or multiple dimensions of evidence), including scientific evidence, personal experience, personal values, economic and political considerations, and philosophical principles. It is not always clear how practitioners integrate these factors into a final decision, but what is clear is that medicine can never be entirely free of value judgments [26]. Normative content seems to enter at all levels of decision-making, even in the production and presentation of the scientific evidence that is supposed to univocally inform evidence-based decisions [27]. The very notion of evidence and the boundaries of what counts as evidence is a social construct, as evidence is always the product of a socially produced question. Even "evidence-based" is a normative concept.
In Tyson's attention to systematic measures and formulaic approaches, he glosses over the value judgments that go into the evaluation of "reasonable" and "unreasonable" actions. He similarly takes as "given" the implicit normativity in his "medical cost relative to value" formula for deciding how to use limited health care resources. In his accounting, the cost utility of neonatal intensive care is expressed as the cost per quality-adjusted life-year (QALY) gained as a result of neonatal intensive care. The life-years gained are then reduced according to the number of disabled survivors and the severity of those disabilities. While Tyson seems to think that deferral to measurement is transparent and fair – presumably because the life circumstances of individual families do not bias the assessment – the values implicit in these measures go unchecked. Even his recognition of "the fact that it is difficult to know how to adjust appropriately for disability and disease, in part because quality of life in the presence of handicaps and chronic illnesses may be rated higher by those affected than by other persons" does not seem to deter him from formulating an evidence-based decision-making algorithm and assuming the justice of measurement in general.
By this account of evidence-based ethics, one might ask how evidence-based ethics differs from evidence-based medicine, as both involve making health care decisions based on the best evidence, where evidence is narrowly defined as having to do with systematic observations from certain types of scientific research. Alternatively, one might question whether evidence-based ethics represents a misappropriation of the word "ethics" [28].
"Evidence based" approaches and practices
While "evidence-based" is typically read in medicine and other life and social sciences as the empirically-adequate standard of reasonable practice and as a means for increasing certainty, the evidence-based movement in fact gains consensus by displacing normative discourse with aggregate or statistically-derived empirical evidence as the "bottom line". The techniques invoked in the name of "evidence-based" decision-making require a positivistic reliance on "the evidence" in its epistemological promise to ascertain truth or certainty by examination of the evidence. These techniques act to obscure the multiple and complex considerations that unavoidably go into health care decisions at both the micro- and macro- level and allows for the promotion of particular political agendas and interests under the guise of "better science" [29].
Despite the promise to revolutionise medicine and the language of "new paradigms", the term "evidence based medicine" has a ring of obviousness to it that makes it difficult to argue against. Few physicians, one suspects, would be willing to assert that they do not attempt to base their clinical decision-making on available evidence. Scientific progress, in fact, is popularly understood to have been motivated by the evidence-based practices of innovative scientists. Rejecting the dogma and superstition that pervaded their historical moment, these innovators let the evidence, as gathered through unbiased and careful experimentation, dictate their scientific practices, beliefs, and theories.
Yet the seeming obviousness of evidence-based medicine is suspect. Post-positivist philosophies of science over the past half century have contested the popular understanding of observational evidence as value-free, pretheoretical, self-apparent, and therefore sufficient to verify or at least falsify scientific hypotheses [30]. The social nature of science is thought to involve considerable normative content in its knowledge producing activities, and these values are not excised in the context of justification. Feminist epistemologists of science, such as Helen Longino [31] and Lynne Hankinson Nelson [32], have called for explicit recognition and critical appraisal of these values, as techniques that presume the value neutrality of science in fact distort scientific practice. The same concern arises in medicine, where evidence-based medicine only seems common sense because it has been stripped of the social context of medical practice in its professed deferral to only "the evidence". In an age where the institutional power of medicine is suspect, a model that represents biomedicine's power as disinterested (or merely "scientific") should give pause. Keith Denny reads evidence-based medicine as a discourse that resists contemporary challenges to established medical authority [33]. While evidence-based medicine appears to question the authority of individual physicians, it works instead to reinforce that authority through its regulation. Furthermore, evidence-based medicine does not question the institutional authority of medicine within society, the way healthcare dollars are allocated for the necessary clinical research, and what role the pharmaceutical industry plays in setting the research agenda.
Evidence-based practices maintain the distinct ability to sidestep value differences and political disputes by appealing to the evidence as the bottom line. This move is positivist in its elimination of culture, contexts, and the subjects of knowledge production from consideration. It is also attractive in an age of moral pluralism. This conceptual linking of methods of abstraction to ascertain truth and progressive politics is reminiscent of the radical politics of early logical positivists like Otto Neurath. In post-war Germany, an epistemological system that avoided the pitfalls of fascism and successfully unified systems of science and thought was perceived to be a progressive step. Only later did "unity of science" theses come to be seen as imperialist and assimilationist and rejected by innovative thinkers in favour of "disunity" and "pluralist" post-modern positions [34].
In health care justice and policy, we see appeals by liberal thinkers to allegedly neutral markers like "species normal functioning". We know, of course, that these measures are not neutral, as people with disabilities and chronic illnesses and elderly people consistently fare poorly in this political calculus. Popular thinking holds, however, that if it is neutrality that is desired, numbers are the pinnacle. In this age of the ascendancy of health outcomes research, where statistical analysis dominates health policy decisionmaking, "evidence" is tantamount to measure and not meaning.
Statistical inference is pursued precisely for its superficiality and its ability to measure broad rather than individual experience. It was its ability to isolate more general variables and phenomena that permit more open and egalitarian debate about social questions that caught the attention of liberals and would-be reformers such as Neurath and Auguste Comte. Yet generalisations and standards contain implicit socially framed and mediated values with a range of implications that can order and enhance, but also tyrannise, aspects of our lives. The success of generalisation is achieved at the expense of contingent and contextual knowledge that needs to be filtered out [35].
Evidence-based decision-making faces inherent limitations insofar as only certain kinds of experience can be quantified and only certain questions explored [26]. Data-driven approaches to patient care have been argued to narrow our ability to effect actions in clinical encounters, as statements of averaged probability become unquestioned laws of possibilities [36]. They limit appreciation of the subtleties and exceptions that characterise all efforts to diagnose and treat illness and displace the critical and vast source of information for treatment, diagnosis and meaningful management of illness that is found in human interaction. Furthermore, evidence-based decision-making ignores the contingency of medical knowledge. While efforts to capture such encounters in aggregate terms have become increasingly sophisticated and thorough, the limitations just mentioned are part and parcel of epidemiological methods and simply cannot be overcome.
We ought to be attentive to the historical moment in which we find ourselves, when the challenges and questions raised by the fragmentation of the medical subject and its recreation as an average probability seem "obvious". In his historical account of the rise of managed care, Gary Belkin situates managed care not as an inevitable response to cost control and economic inefficiency, but within a history of appeals to standardised and ostensible objective measures and models of human behaviour to resolve contentious issues in complex and modern capitalist democratic societies [37]. Evidence-based medicine and evidence-based ethics similarly resolve normative questions and policy issues, such as whether or not to treat a newborn with severe disabilities, by transforming them into problems of measurement. Interestingly, the appeals to technical formulas and standardised information satisfies the strong wish that we have for open, reliable, and presumably objective methods for the resolution of controversy.
Because evidence-based medicine is largely an effort to manage the unruly social world (in which medicine is practiced) via objective scientific procedure, the movement appears to be the latest expression of "scientism", modernity's rationalist dream that science can produce the knowledge required to emancipate us from scarcity, ignorance, and error. However, such efforts tend to disguise political interests in the authority of so-called "scientific evidence". The configuration of policy considerations and clinical standards into questions of evidence conveniently transforms normative questions into technical ones. Political issues are not resolved, but merely disguised in technocratic consideration and language. Thus the goals of medicine and other normative considerations lie just below the surface of these evidentiary questions, and evidence becomes an instrument of, rather than a substitute for, politics.
To illustrate this, consider that the outcomes movement has been invoked to support political efforts to increasingly privatise American health care [38], as the valuing of standardised measures, aggregate behaviour, and radically fragmented medical knowledge supports the logic of the medical marketplace. Because evidence-based medicine legitimates the distillation of medical truth outside of the clinical encounter, where statistical information is privileged over the physician's clinical judgment in clinical decisionmaking, the rationale of a healthcare marketplace populated by independent and rational buyers and sellers is validated. It is in the name of "better science" that particular economic interests can be furthered. The enthusiasm for standardised measures in clinical practice, consistency among professionals in therapeutic interventions, and gold standards of clinical science reflect a medical logic that prefers abstracted measure over individualised history and pathology. Medical authority is, therefore, no longer framed in scientific discourse but in late 20th/early-21st century capitalist discourse with its ideological extolling of the importance of "information" – a move that co-opts demands for democracy and holism in medicine [33]. This is done by appeal to "the evidence" – the unbiased bottom line. Evidence serves as a tool to maintain power by attempting to ignore the conflict of norms at play in politically contentious issues. Habermas has argued that the separation of the technical and the political is an instructive mark of modernity [39]. This removal of normative content from the ideological apparatuses has the dangerous effect of depoliticising the organization of social life and therefore justifying its institutions by rendering them functional within a system of supposedly technically necessary activity.
Much like positivism threatened ethics by rendering it "senseless", an "evidence based" approach proposes to make moral deliberation redundant as it offers a method to resolve ethical and political questions about healthcare spending and practice by appeal to technical measure. The normative issues therefore get co-opted by supposedly neutral technique. An "evidence based" ethics would therefore threaten bioethics' normative mandate.
Summary
In medicine, bioethicists are typically attuned to the multiple dimensions of the illness experience that eludes quantification and measurement. Science, according to medical humanists, is just one layer of description of the phenomenological world. Evidence-based medicine's reliance on scientific evidence has been criticised for mischaracterising modern health care's constitution by diverse academic traditions and knowledges – including the humanities, social sciences, and the pure and applied sciences – that rely on equally diverse notions of evidence [40]. While bioethicists attend to the normative features of medical decision-making, evidence-based ethics suggests a moment of inattentiveness to the normativity of moral decision-making. Recognition of the plurality of values and meanings in operation complicates our use of moral and ethical terms and categories; however, the quick turn to various truth-producing strategies labelled "empirical" that has taken place warrants careful consideration. While the "empirical turn" in bioethics signals a need for reconsideration of the methods used for moral evaluation and resolution, the options should not include obscuring normative content by seemingly neutral technical measure.
Competing interests
The author(s) declare that they have no competing interests.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This paper is part of my doctoral research. Thanks to my dissertation committee members Jim Lindemann Nelson (chair), Marilyn Frye, Lisa Schwartzman, Judy Andre, and Fred Gifford for guidance and support. Thanks also to the organisers and audience at the 2005 Canadian Society for the Study of Practical Ethics meeting in London, Ontario, where an earlier draft of this paper was presented.
==== Refs
Borry P Schotsmans P Dierickx K The birth of the empirical turn in bioethics Bioethics 2005 19 49 71 15812972 10.1111/j.1467-8519.2005.00424.x
Sugarman J Faden R Weinstein J Sugarman J, Sulmasy DR A decade of empirical research in bioethics Methods in Medical Ethics 2001 Washington: Georgetown University Press 19 33
Hoffmaster B Can ethnography save the life of medical ethics? Soc Sci Med 1992 35 1421 1431 1485190 10.1016/0277-9536(92)90045-R
Sugarman J The future of empirical research in bioethics J Law Med Ethics 2004 32 226 231 15301187
Nelson JL Moral teachings from unexpected quarters Hastings Cent Rep 2000 30 12 17 10742941
The Evidence-Based Medicine Working Group Evidence-based medicine JAMA 1992 268 2420 2425 1404801 10.1001/jama.268.17.2420
Davidoff F Haynes B Sackett D Smith R Evidence based medicine BMJ 1995 310 1085 1086 7742666
Sackett D Rosenberg W Gray JAM Haynes B Richardson WS Evidence-based medicine: What it is and what it isn't Br Med J 1996 312 71 72 8555924
Sackett DL A science for the art of consensus J Natl Cancer Inst 1997 89 1003 1005 9230881 10.1093/jnci/89.14.1003
Sackett DL Richardson WS Rosenberg WMC Haynes RB Evidence-Based Medicine: How to Practice and Teach EBM 1997 London: Chirchill Livingstone
Sackett DL Hanes RB On the need for evidence-based medicine Evidence-Based Medicine 1995 1 5 6
Levels of evidence and grades of recommendations Oxford University
Canadian Medical Association Canadian task force on the periodic health examination Can Med Assoc J 1994 150
Zussman R The contributions of sociology to medical ethics Hastings Cent Rep 2000 30 7 11 10742940
Ashcroft RE Teaching patient centred ethics Med Health Care Philos 2000 3 287 295 11200029 10.1023/A:1026473829939
Hedgecoe A For example, Critical bioethics: Beyond the social science critique of applied ethics Bioethics 2004 18 120 143 15146853 10.1111/j.1467-8519.2004.00385.x
Major-Kincade T Tyson J Kennedy K Training pediatric house staff in evidence-based ethics: An explanatory controlled trial J Perinatology 2001 21 161 166 10.1038/sj.jp.7200570
Sugarman J Determining the appropriateness of including children in clinical research: How thick is the ice? JAMA 2004 291 494 496 14747509 10.1001/jama.291.4.494
Jansen R Evidence-based ethics and the regulation of reproduction Hum Reprod 1997 12 2068 2075 9363732 10.1093/humrep/12.9.2068
This is not to say, however, that this "informational" model, where empirical data informs moral deliberation, is without certain difficulties. While the failings of the qualitative tendencies in empirical ethics are beyond the scope of this paper, I refer readers to Richard Ashcroft's thoughtful analysis of how social research evidence plays a part in the construction of the social order. See Ashcroft RE: Constructing empirical bioethics: Foucauldian reflections on the empirical turn in bioethics research Health Care Anal 2003 11 3 13 14510305 10.1023/A:1025329811812
An exhaustive account of the difficulties with this division of labour along "fact" and "value" or "empirical" and "normative" lines is beyond the scope of this paper
Tyson J Evidence-based ethics and the care of premature infants Future Child 1995 161 166 8835519
Gray JAM Evidence-Based Health Care 1997 New York: Churchill Livingstone
Despite defenders of evidence-based medicine's considerable attempts to deflect accusations of offering "cook book medicine", that is, treating patients strictly according to a formula or algorithm derived from a research study, even Brian Haynes, one of the movement's founders, admits that much of the programmatic literature regarding evidence-based medicine does not "clearly emphasise that evidence from research can be no more than one component of any clinical decision" Op cit 25
Haynes RB What kind of evidence is it that evidence-based medicine advocates want health care providers and consumers to pay attention to? BMC Health Services Research 2002 2 3 11882257 10.1186/1472-6963-2-3
Kerridge I Lowe M Henry D Ethics and evidence based medicine British Medical Journal 1998 316 1151 1153 9552959
Molewijk AC Stiggelbout AM Otten W Dupuis HM Kievit J Implicit normativity in evidence-based medicine: A plea for integrated empirical ethics research Health Care Anal 2003 11 69 92 14510310 10.1023/A:1025390030467
Thanks to an audience member at a presentation of an earlier draft of this paper at the Canadian Society for the Study of Practical Ethics for this suggestion
Goldenberg MJ On evidence and evidence-based medicine: Lessons from the philosophy of science Soc Sci Med 5688227
Quine WVO Word and Object 1960 Cambridge: MIT Press
Longino H Science as Social Knowledge: Values and Objectivity in Scientific Inquiry 1990 Princeton: Princeton University Press
Nelson LH Alcoff L, Potter E Epistemological communities Feminist Epistemologies 1993 New York: Routledge 121 160
Denny K Evidence-based medicine and medical authority J Med Humanities 1999 20 247 263 10.1023/A:1022924404779
Galison P Galison P, Stump D The context of disunity The Disunity of Science: Boundaries, Contexts, and Power 1996 Stanford: Stanford University Press 1 33
Belkin G The technocratic wish: Making sense and finding power in the "managed" medical marketplace J Health Politics Policy Law 1997 22 509 532
Frankford DM Scientism and economism: Regulating health care J Health Politics Policy Law 1994 19 773 799
Belkin G Nelson JL Shifting the focus: The historical meaning of managed care and the search for ethics in mental health Rationing Sanity: Ethical Issues in Managed Mental Health Care 2003 Washington: Georgetown University Press 9 27
Tanenbaum SJ Getting from there to here: Evidentiary quandaries of the US outcomes movement J Eval Clin Prac 1995 1 97 103
Habermas J Seidman S Technology and science as ideology Jurgen Habermas on Society and Politics: A Reader 1989 Boston: Beacon Press 237 265
Upshur REG VanDenKerkhof EG Goel V Meaning and measurement: An inclusive model of evidence in health care J Eval Clin Prac 2001 7 91 96 10.1046/j.1365-2753.2001.00279.x
|
16277663
|
PMC1298300
|
CC BY
|
2021-01-04 16:31:58
|
no
|
BMC Med Ethics. 2005 Nov 8; 6:11
|
utf-8
|
BMC Med Ethics
| 2,005 |
10.1186/1472-6939-6-11
|
oa_comm
|
==== Front
BMC MicrobiolBMC Microbiology1471-2180BioMed Central London 1471-2180-5-641626643910.1186/1471-2180-5-64Research ArticleThe gene for a lectin-like protein is transcriptionally activated during sexual development, but is not essential for fruiting body formation in the filamentous fungus Sordaria macrospora Nowrousian Minou [email protected] Patricia [email protected] Lehrstuhl für Allgemeine und Molekulare Botanik, Ruhr-Universität Bochum, 44801 Bochum, Germany2005 3 11 2005 5 64 64 10 9 2005 3 11 2005 Copyright © 2005 Nowrousian and Cebula; 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 filamentous fungus Sordaria macrospora forms complex three-dimensional fruiting bodies called perithecia that protect the developing ascospores and ensure their proper discharge. In previous microarray analyses, several genes have been identified that are downregulated in sterile mutants compared to the wild type. Among these genes was tap1 (transcript associated with perithecial development), a gene encoding a putative lectin homolog.
Results
Analysis of tap1 transcript levels in the wild type under conditions allowing only vegetative growth compared to conditions that lead to fruiting body development showed that tap1 is not only downregulated in developmental mutants but is also upregulated in the wild type during fruiting body development. We have cloned and sequenced a 3.2 kb fragment of genomic DNA containing the tap1 open reading frame and adjoining sequences. The genomic region comprising tap1 is syntenic to its homologous region in the closely related filamentous fungus Neurospora crassa. To determine whether tap1 is involved in fruiting body development in S. macrospora, a knockout construct was generated in which the tap1 open reading frame was replaced by the hygromycin B resistance gene hph under the control of fungal regulatory regions. Transformation of the S. macrospora wild type with this construct resulted in a tap1 deletion strain where tap1 had been replaced by the hph cassette. The knockout strain displayed no phenotypic differences under conditions of vegetative growth and sexual development when compared to the wild type. Double mutants carrying the Δtap1 allele in several developmental mutant backgrounds were phenotypically similar to the corresponding developmental mutant strains.
Conclusion
The tap1 transcript is strongly upregulated during sexual development in S. macrospora; however, analysis of a tap1 knockout strain shows that tap1 is not essential for fruiting body formation in S. macrospora.
==== Body
Background
Fruiting body formation in ascomycetes is a complex process leading to the formation of a number of specialized cell types from a comparatively undifferentiated vegetative mycelium [1]. Recently, the molecular basis of this process has been investigated by forward and reverse genetics approaches, and a number of genes that are essential for fruiting body development have been identified. However, a coherent picture of fungal multicellular development has yet to emerge [2].
One avenue towards a deeper understanding of developmental processes is by functional genomics analyses, e.g. microarray studies. Such approaches can help to identify genes that are regulated differentially during fruiting body development and are therefore candidates for further functional analysis. In a previous study, we have performed microarray analyses of fruiting body development in S. macrospora [3]. This ascomycete is homothallic and produces fruiting bodies called perithecia within seven days under laboratory conditions. S. macrospora is a close relative of N. crassa, but in contrast to N. crassa, it does not produce any asexual spores. Therefore, changes of gene expression patterns during sexual development are not superimposed by changes related to asexual sporulation. We have previously analyzed gene expression in three developmental mutants of S. macrospora, and have identified a number of genes that are downregulated in the sterile mutants when compared with the wild type [3]. One of these genes is tap1 (transcript associated with perithecial development, formerly known as SMU5651 [3]). tap1 transcript levels are downregulated in the three developmental mutants pro1, pro11 and pro22 as well as in all three double mutants which led us to speculate that the gene might be involved in sexual development in S. macrospora. In addition to this intriguing expression pattern, the derived TAP1 amino acid sequence shows homology to lectins from other filamentous fungi with the highest similarity to lectins isolated from fruiting bodies of several basidiomycetes [3]. It has long been speculated that lectins play a role in fungal development; however, as no mutants in lectin-encoding genes from fruiting body-producing fungi have been analyzed to date, definite proof for this hypothesis is lacking [4,5]. The only known fungal lectin mutant is a strain of Arthrobotrys oligospora in which the lectin gene aol was deleted [6]. This mutant does not exhibit any phenotypical differences from the wild type under all conditions investigated, but as no sexual cycle is known for A. oligospora, the question whether aol might be involved in fruiting body development could not be addressed. Here, we present data on the expression of the S. macrospora tap1 gene as well as the characterization of a tap1 knockout strain to address the role of a putative lectin in fungal sexual development.
Results
Expression of tap1 during sexual development in S. macrospora
Previously, we reported that tap1 is downregulated in several developmental mutants when compared with the wild type [3]. In that analysis, all strains were grown under conditions that allowed sexual development, i.e. in floating culture. To investigate whether tap1 expression in the wild type is associated with sexual development, we compared transcript levels of tap1 in sexually developing mycelia with vegetative mycelia. For this purpose, the wild type was grown in floating or in submerged culture where it develops either fruiting bodies or vegetative mycelium, respectively. Analysis of tap1 expression by quantitative real time PCR revealed that tap1 transcript levels are upregulated nearly 60-fold in mycelia undergoing sexual development (Figure 1). These data further confirm that tap1 expression is linked with fruiting body development in S. macrospora.
Sequence analysis of the S. macrospora tap1 gene
For previous expression analyses, part of the tap1 open reading frame was amplified from S. macrospora genomic DNA by PCR [3]. For further characterization of tap1, we have isolated a tap1-containing cosmid clone from an indexed cosmid library of S. macrospora [7]. A 3.2 kb restriction fragment carrying the tap1 open reading frame and adjacent regions was subcloned and sequenced. The fragment carries the uninterrupted tap1 open reading frame of 429 bp encoding a predicted polypeptide of 143 amino acids. In addition, the fragment contains part of another open reading frame downstream of tap1 that was named ibl1 (importin beta like 1) due to its similarity to an importin beta subunit. The sequenced region is syntenic to a region within the N. crassa genome that contains the open reading frames NCU05651.2 and NCU05650.2 that are orthologs of tap1 and ibl1 respectively (Figure 2). In a previous comparison of 85 protein-coding genes of S. macrospora and N. crassa, the average nucleic acid identity within coding regions was found to be close to 90 %, and even non-coding regions of these two closely related Pyrenomycetes can be readily aligned [8]. These findings are further supported by our analysis of the genomic region containing tap1 (Figure 2).
The predicted TAP1 polypeptide was used for BLASTP [9] searches of the public databases, and a multiple alignment was constructed of TAP1 and several of its closest homologs from other fungi (Figure 3). The closest homolog is the predicted protein NCU05651.2 from N. crassa; however, interestingly, the second-best sequence identity is found in two lectins from the basidiomycetes Xerocomus chrysenteron and Paxillus involutus. Surprisingly, lectins and predicted lectin-like proteins from the ascomycetes Podospora anserina, Fusarium graminearum, and A. oligospora are less similar to the S. macrospora TAP1 even though these fungi are much more closely related to S. macrospora than the basidiomycetes (Figure 3). However, the (predicted) lectins from P. anserina, F. graminearum, and A. oligospora have a higher similarity compared to the basidiomycete lectins than TAP1 does (Figure 3). These findings might indicate that TAP1 and its corresponding N. crassa ortholog, even though being clearly members of this fungal lectin family, have evolved rather fast compared to other ascomycete homologs.
Construction of a Δtap1 strain
A construct for generating a tap1 deletion strain was generated. It consists of a hygromycin resistance cassette flanked by about 1 kb of sequences upstream and downstream of the tap1 open reading frame (Figure 4). The construct was used to transform the S. macrospora wild type, and transformants were screened for homologous recombination by PCR with oligonucleotide pairs d1 and d2 as well as d3 and d4 (Figure 4). These primer combinations yield amplicons only in the case of homologous recombination. Among 50 primary transformants, one transformant was found that displayed the expected PCR fragments (data not shown). Ascospores from this transformant were isolated to purify the putative knockout strain, and Southern blot analysis of nine ascospore isolates was performed with probes for tap1 and the Hygromycin resistance gene hph. For two of the ascospore isolates, T2.35 and T2.41, results are shown in Figure 5. As expected, the tap1 probe hybridized to a 3.2 kb BglII fragment in genomic DNA of the wild type, whereas no hybridization signal was obtained with genomic DNA from the knockout strains T2.35 and T2.41. Conversely, the hph probe marked a 4.2 kb BglII fragment in the knockout strains and produced no signal with wild type DNA. The 4.2 kb fragment is of the size expected in a strain where homologous recombination has taken place (Figure 4). The Δtap1 strains T2.35 and T2.41 were used for further analysis.
Phenotypic characterization of the Δtap1 strain in different genetic backgrounds
The tap1 deletion strains T2.35 and T2.41 were analyzed with respect to their phenotype during the sexual cycle. Surprisingly, both were completely wild type-like in all aspects of fruiting body development, i.e. both tap1 deletion strains produced mature perithecia in the same numbers and in the same amount of time as the wild type. Moreover, neither perithecial nor spore morphology were altered in any recognizable way (Figure 6). Fruiting body formation took place both in constant light as well as in constant darkness both on complete as well as minimal media, similar to that of the wild type. Ascospores from the knockout strains germinated readily and were black like wild type spores (Figure 6).
We then investigated whether there were any phenotypes unrelated to sexual development present in the mutant strains. Mycelial growth rates were determined as 24.7 (± 1.7) and 24.8 (± 1.4) mm per day for Δtap1 and wild type respectively; thus, indicating that mycelial growth is not altered in the knockout strains. Also, renewal of growth and fruiting body development after incubation at 4°C or 37°C, conditions that prevent growth and fruiting body formation, respectively, was not different in the mutants when compared to the wild type (data not shown). Also, wettability of mycelium and fruiting bodies was similar in wild type and mutants indicating that the hydrophobic coating of the mycelium was not altered in the knockout strains. As the lack of tap1 did not cause any discernable phenotype, we wondered whether tap1 might have a redundant function or whether its absence might be masked by the presence of other genes. We therefore crossed the Δtap1 allele into strains bearing mutations in other developmental genes, namely pro1, pro11, pro22, and pro41. The pro1 mutant is defective in a gene encoding a transcription factor [10], the pro11 mutant lacks a functional WD40 repeat protein [11], and pro22 and pro41 are mutants non-allelic to tap1 or the other pro genes (Kück et al., unpublished data). tap1 transcript levels are downregulated in mutants pro1, pro11, pro22 [3], and pro41 (Nowrousian, unpublished data). Thus, we speculated whether a complete lack of tap1 would lead to a more pronounced developmental phenotype, especially as several other developmental genes are also downregulated in the pro mutants [3]. We obtained double mutants with strains pro1, pro11, pro22, and pro41 and verified the presence of the Δtap1 allele in the double mutants by Southern blot analysis (Figure 5). All double mutant strains were phenotypically similar to the respective single pro mutants in that they produced only protoperithecia and were therefore sterile. There were no differences in the number of protoperithecia produced by the double mutants when compared with the single mutants. Also, growth rates of the vegetative mycelium as well as overall morphological appearance were the same (data not shown). For the crosses of the tap1 knockout with the pro mutant strains and subsequent back-crosses, we analyzed a total of 17 full tetrads and 37 partial tetrads from crosses with different mutant strains and in all cases found the expected 1:1 segregation pattern for each of the single markers (hygromycin resistance and pro mutant phenotype, respectively). This is a further indication that deletion of tap1 does not interfere with sexual development, and also shows that the process of generating the Δtap1 allele did not introduce further mutations into the strains that would cause any different segregation patterns. Overall, no phenotype for Δtap1 was found in any of the genetic backgrounds investigated.
Discussion
The tap1-encoded polypeptide from S. macrospora has significant homology to lectins and lectin-like protein from other fungi (Figure 3). The highest degree of amino acid identity is found in comparison with lectins that were isolated from fruiting bodies of several basidiomycetes [12-14]. Lectins are carbohydrate-binding proteins that are found in a variety of organisms [4,5]. On the basis of sequence homology, the S. macrospora TAP1 polypeptide can be included into a class of fungal lectins; however, whether it has lectin activity, i.e. whether it specifically binds carbohydrates, remains to be elucidated. Interestingly, TAP1 displays a greater sequence identity towards basidiomycete lectins than to lectins or putative lectins from ascomycetes with the (notable) exception of its closest homolog, the N. crassa protein NCU05651.2 (Figure 3). However, BLAST searches in the N. crassa genome [15] with the sequences of TAP1 as well as the lectin sequences from the other fungi used in our sequence comparisons yielded only NCU05651.2 as a significant result (data not shown). This finding indicates that the gene is present as a single copy in N. crassa, and that there is no other member of this lectin gene family present in the N. crassa genome. As N. crassa and S. macrospora are close relatives with highly syntenic genomes [8], it is likely that this is the case in S. macrospora as well. This observation is supported by the fact that only a single band in S. macrospora genomic DNA hybridizes with a tap1 probe (Figure 5). Thus, it seems that this particular class of fungal lectins has evolved faster in S. macrospora and N. crassa compared to other ascomycetes, for which it still retains more similarity with its basidiomycete relatives (Figure 3). Another class of fungal lectins has been found in the basidiomycete Coprinus cinereus. These lectins bind galactose and are therefore called galectins, and two galectins from C. cinereus are specifically expressed during different stages of fruiting body formation [16,17]. However, BLAST searches for galectin homologs in the N. crassa genome yielded no significant results (data not shown) indicating that no galectin-like proteins exist in N. crassa or that their sequences are too dissimilar to the C. cinereus sequences to be detected by sequence comparisons alone. Thus, the N. crassa NCU05651.2 gene and its S. macrospora ortholog tap1 are so far the only genes encoding putative lectins that have been identified by sequence analysis in these two ascomycetes.
In fungi, most lectins have been isolated from basidiomycetes, especially from mushroom fruiting bodies, and it has been speculated that they play a role in fruiting body development [5,18,19]. However, as no lectin mutant in a fruiting body-producing fungus has been characterized to date, this hypothesis has not been verified experimentally. Previous investigations and the results presented here show that tap1 transcript levels are closely correlated with fruiting body development in S. macrospora; therefore, we decided to construct a tap1 knockout strain to analyze whether this putative lectin plays a role in fruiting body formation. A Δtap1 strain was generated by gene replacement, but the knockout strain has no discernable phenotype under all conditions investigated. This might indicate that tap1 has indeed no function in vegetative growth or sexual development of S. macrospora; however, the possibility that tap1 is needed under environmental conditions not tested in our experiments or that its function is redundant or that in the absence of tap1, another gene product can take its place, cannot be excluded. The latter effect is well known in other organisms, and it was, for example, tested in a large-scale analysis of yeast synthetic lethal interactions where it was found that genes involved in similar biological processes, but not necessarily in the same regulatory pathway, can buffer one another in single mutant backgrounds but show a phenotype in the double mutant strain [20]. To test whether any of the known developmental genes of S. macrospora show this kind of genetic interaction with tap1, we obtained double mutants of Δtap1 with strains bearing mutations in the developmental genes pro1, pro11, pro22, or pro41. However, all double mutant strains were phenotypically similar to the respective pro mutant strains. tap1 transcript levels are downregulated in the pro mutants, and our analysis demonstrates that even the complete loss of tap1 does not worsen the condition of the mutants. This leaves open the possibility that tap1 is necessary in a different genetic background or under different environmental conditions.
With respect to fruiting body formation, our results show that the putative lectin-encoding gene tap1 is not an essential gene for this developmental process. As mentioned previously, the only other known fungal lectin mutant is the aol mutant of the nematode-trapping fungus A. oligospora [6]. Similar to our findings, the aol mutant also has no phenotype under all conditions investigated. Since no sexual stages from A. oligospora are known, these observations do not include fruiting body formation; however, vegetative growth, conidiation, and nematode-trapping were unchanged in the aol mutant strain [6]. Thus, possible functions of this class of lectins and lectin-like proteins in filamentous fungi remain enigmatic.
Conclusion
tap1 expression is strongly associated with sexual development in S. macrospora. An analysis of the tap1 gene and its surrounding genomic region revealed a high degree of sequence identity and overall synteny with the corresponding region in the genome of N. crassa. Sequence comparisons of TAP1 with lectins and lectin-like proteins from other fungi indicate that it is most closely related to lectins isolated from basidiomycete fruiting bodies. However, analysis of a tap1 knockout strain shows that tap1 is not essential for fruiting body formation nor vegetative growth in S. macrospora. This is the case for a Δtap1 allele in an otherwise wild type genetic background as well as in combination with mutations in several developmental genes. Whether tap1 has any function under growth conditions not investigated here, e.g. in a more natural setting, remains to be elucidated.
Methods
Strains and growth conditions
Details of the S. macrospora strains used in this study are provided in Table 1. Double mutants were obtained from crosses of single mutant strains. Double mutant genotypes were verified by crosses against single mutants. Unless stated otherwise, standard growth conditions and DNA-mediated transformation were as previously described [10,21]. For analysis of growth velocity, 30 cm long race tubes were filled with 15 ml of medium, inoculated at one end and the growth front was marked every 24 h for 7 consecutive days. For RNA extraction from cultures developing fruiting bodies, S. macrospora was grown at 25°C in constant light in floating culture as described [3]. For RNA extraction from vegetative mycelium, a mycelial plug of 0.7 cm in diameter from a Petri dish with liquid medium [3] was inoculated into an Erlenmeyer flask with 100 ml of liquid medium and shaken at 130 rpm.
RNA extraction and quantitative real time PCR
Extraction of total RNA and quantitative real time PCR were performed as described previously [3] with the following modifications: reverse transcription was performed with 400 U Superscript II (Invitrogen) and 0.33 mM dNTPs, and real time PCR was carried out in a DNA Engine Opticon 2 (MJ Research).
Identification of a cosmid clone carrying tap1 and analysis of the tap1 gene
An indexed S. macrospora cosmid library [7] was screened for tap1 by PCR with oligonucleotides SMU5651-1 (5' CATCAACGACACCTCCGACACCC) and SMU5651-2 (5' CATCGGCCTGATAGAACTTGATCC). For a first round of screening, pooled DNA from 48 cosmid clones was used as a template, DNA from clones from positive pools was then subpooled and used for the next round of screening. This led to the isolation of cosmid D3 from pool VI518-614 that contains the tap1 gene. A 3.2 kb BglII restriction fragment carrying tap1 was subcloned from cosmid D3 into pBluescript II/KS+ (Stratagene). The insert of the resulting vector pPC24 was sequenced at MWG Biotech or GATC Biotech AG [emb:AJ781427.2].
Construction of a Δtap1 strain
To create a tap1 knockout construct for homologous recombination in S. macrospora, flanking regions upstream and downstream of the tap1 open reading frame were amplified by PCR from S. macrospora genomic DNA using oligonucleotides SMU5651-BamHI (5' AGGATCCGTGATTCTCATGCTGTGGAAGGAAGC) and SMU5651-NheI (5' AGCTAGCTTTGGCGGTTTGGTTGGGGGGTTGGT) for the upstream region and oligonucleotides SMU5651-ApaI (5' AGGGCCCGTACTCGTCAGTGGGAAAGTGGGTGG) and SMU5651-SacI (5' AGAGCTCTATGCACTTGCTCCTCAAGCGTCTC) for the downstream region, introducing restriction sites as indicated in the oligonucleotide names. Additionally, the hygromycin-resistance cassette consisting of the hph gene from Escherichia coli and the trpC promoter from Aspergillus nidulans was amplified from plasmid pCB1004 [22] using oligonucleotides Hph-ApaI (5' AGGGCCCTCAACGGAACCCTATTCCTTTGCCC) and Hph-NheI (5' AGCTAGCAACTGATATTGAAGGAGCATTTTTGG). All PCR fragments were subcloned in pDrive (Qiagen) resulting in pDriveA (upstream region), pDriveB (downstream region) and pDrivehph (hygromycin-resistance cassette). Sequences of inserts and orientation within the vector was verified by restriction analysis and sequencing (MWG Biotech AG). The PCR fragment containing the downstream region was obtained by ApaI/SacI digestion of pDriveB and cloned into ApaI/SacI digested vector pDriveA resulting in plasmid pDriveAB. The hygromycin-resistance cassette was obtained by digesting plasmid pDrivehph with NheI and ApaI and was cloned into plasmid pDriveAB hydrolyzed with NheI and ApaI resulting in the knockout plasmid pABXY. For transformation of S. macrospora, plasmid pABXY was digested with BamHI and SacI and the knockout cassette was obtained by gel elution. The knockout cassette was transformed into the S. macrospora wild type and primary transformants were screened for homologous integration by PCR. For this purpose, total DNA was prepared from the transformants according to [23], and PCR was performed with oligonucleotides d1 (5' CGATGGCTGTGTAGAAGTACTCGC) and d2 (5' TGCCTCCTCCGAGGCTGATAACCT) for the downstream region or d3 (5' CGGTGGGTAAGGTATCTCTGATG) and d4 (5' CACCGCCTGGACGACTAAACCAA) for the upstream region (Figure 4).
Authors' contributions
MN designed the study, performed expression analyses and isolation of the cosmid clone as well as part of the characterization of the knockout strain and wrote the manuscript. PC cloned and sequenced the tap1 gene, made the knockout construct and the tap1 knockout strain and carried out part of the characterization of the knockout strain.
Acknowledgements
The authors would like to thank Swenja Ellßel and Susanne Schlewinski for excellent technical assistance and Prof. Dr. Ulrich Kück for his generous support that enabled us to carry out this study as part of project A1 of research initiative SFB 480 funded by the Deutsche Forschungsgemeinschaft (DFG).
Figures and Tables
Figure 1 Transcript levels of tap1 are upregulated during sexual development. tap1 transcript levels in the wild type under conditions of vegetative and sexual development were analyzed by real time PCR. For normalization, transcript levels of the SSUrRNAs were used as described previously [3]. Expression is given as fold induction (mean of two independent experiments, error bars indicate standard deviation) with transcript levels during vegetative growth set to 1. Real time PCR results were tested for significance of differential expression at a level of p = 0.001 using REST [24].
Figure 2 Synteny between S. macrospora (S.m.) and N. crassa (N.c.) in the genomic region containing tap1. Coding regions are given as dark gray and intergenic regions as light gray boxes. Nucleic acid identities between the S. macrospora open reading frames and their N. crassa orthologues as well as between the intergenic sequences are indicated.
Figure 3 Multiple alignment of TAP1 and lectins from filamentous fungi. The multiple alignment was created using CLUSTALX [25] with the following sequences: S.m., Sordaria. macrospora TAP1 [this work, emb:CAH03681.2]; N.c., Neurospora crassa NCU05651.2 [ref:XP_325506.1]; X.c., Xerocomus chrysenteron lectin [gb:AAL73236.1]; P.i., Paxillus involutus LECA [gb:AAT91249.1]; P.c., Pleurotus cornucopiae PCL-F2 [dbj:BAB63923.1]; A.b., Agaricus bisporus ABL [sp:Q00022]; P.a., Podospora anserina Pa5D0092 [emb:CAD60779.1]; F.g., Fusarium graminearum FG07558.1 [gb:EAA76455.1]; A.o., Arthrobotrys oligospora lectin [emb:CAA65781.1]. Jalview was used to visualize the alignment [26]. Amino acid residues conserved in at least eight of the nine sequences are given in dark blue, residues conserved in at least six sequences in medium blue, and residues present in at least four sequences in light blue. At the end of the alignment, amino acid identity in % is given for all sequences in pair-wise comparisons.
Figure 4 Strategy for the generation of a tap1 knockout strain. The tap1 open reading frame is shown in dark gray, flanking non-coding regions that are present in the knockout construct are given as white boxes, and adjoining regions not present in the knockout construct are shown in light gray. The Hygromycin B resistance cassette comprising the hph gene and the trpC promoter and 5' UTR is given in black. Positions of oligonucleotide primers d1 to d4 are indicated. For further information see text.
Figure 5 Southern blot analysis of Δtap1 strains. Genomic DNA from the wild type, from two different Δtap1 single spore isolates from the original knockout strain, and from double mutants of Δtap1 with pro1, pro41, pro11, and pro11 was hydrolyzed with BglII, and after gel electrophoresis, the Southern blot was probed with radio-labeled DNA fragments containing the open reading frames of tap1 (A) and hph (B), respectively. Marker sizes in kb are provided on the right. Numbers in brackets are strain numbers for single spore isolates.
Figure 6 Δtap1 strains have a wild type-like phenotype. The wild type and Δtap1 single spore isolates T2.35 and T2.41 (as indicated above each column) were grown in Petri dishes on BMM solid medium [10, 21] for 6d (A) or 7d (B, C) at 25°C in constant light. A) Segments of petri dishes, black dots are individual fruiting bodies. B) Segments of ascus rosettes with mature (black-spored) and immature asci. Scale bar 100 μm. C) Mature asci. Scale bar 20 μm.
Table 1 S. macrospora strains used in this study. All strains are single spore isolates and are kept in our laboratory collection. The fus allele that is present in some of the strains is a spore color marker (brown instead of black spores) but has no influence on growth or fertility.
strain number genotype description
S48977 wild type wild type
T2.35 Δtap1 tap1 deletion strain
T2.41 Δtap1 tap1 deletion strain
M8871 pro1 sterile mutant [10]
S24117 pro11 sterile mutant [11]
S22528 pro22 sterile mutant
S46357 pro41 sterile mutant
S62222 Δtap1, pro1 sterile mutant
S62355 Δtap1, pro1 sterile mutant
S62497 Δtap1, pro41 sterile mutant
S63491 Δtap1, pro11, fus sterile mutant
S63433 Δtap1, pro11, fus sterile mutant
S62292 Δtap1, pro22, fus sterile mutant
S62456 Δtap1, pro22, fus sterile mutant
==== Refs
Bistis GN Perkins DD Read ND Different cell types in Neurospora crassa Fungal Genet Biol 2003 50 17 19
Pöggeler S Nowrousian M Kück U Kües U, Fischer R Fruiting body development in ascomycetes The Mycota I 2005 Berlin, Heidelberg: Springer
Nowrousian M Ringelberg C Dunlap JC Loros JJ Kück U Cross-species microarray hybridization to identify developmentally regulated genes in the filamentous fungus Sordaria macrospora Mol Genet Genomics 2005 273 137 149 15778868 10.1007/s00438-005-1118-9
Wang H Ng TB Ooi VEC Lectins from mushrooms Mycol Res 1998 102 897 906 10.1017/S0953756298006200
Guillot J Konska G Lectins in higher fungi Biochem Syst Ecol 1997 25 203 230 10.1016/S0305-1978(96)00110-X
Balogh J Tunlid A Rosen S Deletion of a lectin gene does not affect the phenotype of the nematode-trapping fungus Arthrobotrys oligospora Fungal Genet Biol 2003 39 128 135 12781671 10.1016/S1087-1845(03)00023-9
Pöggeler S Nowrousian M Jacobsen S Kück U An efficient procedure to isolate fungal genes from an indexed cosmid library J Microbiol Meth 1997 29 49 61 10.1016/S0167-7012(97)00018-3
Nowrousian M Würtz C Pöggeler S Kück U Comparative sequence analysis of Sordaria macrospora and Neurospora crassa as a means to improve genome annotation Fungal Genet Biol 2004 41 285 292 14761789 10.1016/j.fgb.2003.10.005
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
Masloff S Pöggeler S Kück U The pro1+ gene from Sordaria macrospora encodes a C6 zinc finger transcription factor required for fruiting body development Genetics 1999 152 191 199 10224253
Pöggeler S Kück U A WD40 repeat protein regulates fungal cell differentiation and can be replaced functionally by the mammalian homologue striatin Eukaryot Cell 2004 3 232 240 14871953 10.1128/EC.3.1.232-240.2004
Iijima N Yoshino H Ten LC Ando A Watanabe K Nagata Y Two genes encoding fruit body lectins of Pleurotus cornucopiae: sequence similarity with the lectin of a nematode-trapping fungus Biosci Biotechnol Biochem 2002 66 2083 2089 12450118 10.1271/bbb.66.2083
Crenshaw R Harper SN Moyer M Privalle LS Isolation and characterization of a cDNA clone encoding a lectin gene from Agaricus bisporus Plant Physiol 1995 107 1465 1466 7770537 10.1104/pp.107.4.1465
Trigueros V Lougarre A Ali-Ahmed D Rahbé Y Guillot J Chavant L Fournier D Paquereau L Xerocomus chrysenteron lectin: identification of a new pesticidal protein Biochim Biophys Acta 2003 1621 2992 2298
Galagan J Calvo S Borkovich K Selker E Read N Jaffe D FitzHugh W Ma L Smirnov S Purcell S The genome sequence of the filamentous fungus Neurospora crassa Nature 2003 422 859 868 12712197 10.1038/nature01554
Boulianne RP Liu Y Aebi M Lu BC Kües U Fruiting body development in Coprinus cinereus: regulated expression of two galectins secreted by a non-classical pathway Microbiology 2000 146 1841 1853 10931889
Cooper DN Boulianne RP Charlton S Farrell EM Sucher A Lu BC Fungal galectins, sequence and specificity of two isolectins from Coprinus cinereus J Biol Chem 1997 272 1514 1521 8999822 10.1074/jbc.272.3.1514
Oguri S Ando A Nagata Y A novel developmental stage-specific lectin of the basidiomycete Pleurotus cornucopiae J Bacteriol 1996 178 5692 5698 8824614
Kaneko T Oguri S Kato S Nagata Y Developmental appearance of lectin during fruit body formation in Pleurotus cornucopiae J Gen Appl Microbiol 1993 39 93 100
Tong AH Lesage G Bader GD Ding H Xu H Xin X Young J Berriz GF Brost RL Chang M Global mapping of the yeast genetic interaction network Science 2004 303 808 813 14764870 10.1126/science.1091317
Nowrousian M Masloff S Pöggeler S Kück U Cell differentiation during sexual development of the fungus Sordaria macrospora requires ATP citrate lyase activity Mol Cell Biol 1999 19 450 460 9858569
Carroll AM Sweigard JA Valent B Improved vectors for selecting resistance to Hygromycin Fungal Genet Newsl 1994 41 22
Lecellier G Silar P Rapid method for nucleic acid extraction from petri dish-grown mycelia Curr Genet 1994 25 122 123 8087879 10.1007/BF00309536
Pfaffl MW Horgan GW Dempfle L Relative expression software tool (REST) for group-wise comparison and statistical analysis of relative expression results in real-time PCR Nucleic Acids Res 2002 30 e36 11972351 10.1093/nar/30.9.e36
Thompson JD Gibson TJ Plewniak F Jeanmougin F Higgins DG The ClustalX windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools Nucleic Acids Res 1997 24 4876 4882 10.1093/nar/25.24.4876
Clamp M Cuff J Searle SM Barton GJ The Jalview Java alignment editor Bioinformatics 2004 12 426 427 10.1093/bioinformatics/btg430
|
16266439
|
PMC1298301
|
CC BY
|
2021-01-04 16:03:40
|
no
|
BMC Microbiol. 2005 Nov 3; 5:64
|
utf-8
|
BMC Microbiol
| 2,005 |
10.1186/1471-2180-5-64
|
oa_comm
|
==== Front
BMC MicrobiolBMC Microbiology1471-2180BioMed Central London 1471-2180-5-651628008210.1186/1471-2180-5-65Methodology ArticleSpecific and sensitive detection of the conifer pathogen Gremmeniella abietina by nested PCR Zeng Qing-Yin [email protected] Per [email protected] Xiao-Ru [email protected] National Institute for Working Life, SE-90713 Umeå, Sweden2 Department of Molecular Biology, Umeå University, SE-90187 Umeå, Sweden3 Department of Silviculture, the Swedish University of Agricultural Sciences, SE-90183 Umeå, Sweden2005 9 11 2005 5 65 65 28 6 2005 9 11 2005 Copyright © 2005 Zeng 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
Gremmeniella abietina (Lagerb.) Morelet is an ascomycete fungus that causes stem canker and shoot dieback in many conifer species. The fungus is widespread and causes severe damage to forest plantations in Europe, North America and Asia. To facilitate early diagnosis and improve measures to control the spread of the disease, rapid, specific and sensitive detection methods for G. abietina in conifer hosts are needed.
Results
We designed two pairs of specific primers for G. abietina based on the 18S rDNA sequence variation pattern. These primers were validated against a wide range of fungi and 14 potential conifer hosts. Based on these specific primers, two nested PCR systems were developed. The first system employed universal fungal primers to enrich the fungal DNA targets in the first round, followed by a second round selective amplification of the pathogen. The other system employed G. abietina-specific primers in both PCR steps. Both approaches can detect the presence of G. abietina in composite samples with high sensitivity, as little as 7.5 fg G. abietina DNA in the host genomic background.
Conclusion
The methods described here are rapid and can be applied directly to a wide range of conifer species, without the need for fungal isolation and cultivation. Therefore, it represents a promising alternative to disease inspection in forest nurseries, plantations and quarantine control facilities.
==== Body
Background
Gremmeniella abietina (Lagerb.) Morelet is among the most destructive conifer forest pathogens in the northern hemisphere. This ascomycete fungus has a broad host range and causes stem canker and shoot dieback in many conifer species of the genera Pinus, Abies, Picea, Larix, Tsuga and Pseudotsuga [1-3]. In Sweden, G. abietina is found on the two native conifers Picea abies and Pinus sylvestris, as well as the introduced species, Pinus contorta [4-6].
Under favorable conditions, the life cycle of G. abietina takes two years to complete [5]. The fungus may grow in the host as an endophyte for more than one year [7,8], and infected trees can remain undetected for several years before manifesting visible symptoms. This poses difficulties for diagnosing disease at an early stage, using asymptomatic materials. Nursery seedling inspection and quarantine control require sensitive detection methods to limit the spread of the pathogen. Various morphological, physiological, pathogenic and biochemical characters and molecular markers have been employed to distinguish and characterize different races and types of the fungus, such as the North American race (NA), European (EU) race, and large tree type (LTT) and small tree type (STT) [9-15]. Most of these methods require isolation of the fungus in culture. This process is time consuming and is not appropriate for the detection of the fungus directly in infected but asymptomatic tissues.
Specific polymerase chain reaction (PCR) based detection methods are sensitive and robust techniques when used in plant disease diagnostic research. By employing specific PCR primers it is possible to selectively amplify the pathogen from infected tissues without the need for isolation. Among the different PCR techniques, nested PCR (a two-step PCR system in which the first round PCR products are subjected to a second round PCR amplification with more specific primers) is extremely sensitive and allows the detection of a fungal pathogen in minute amounts of infected material [16-18]. Recently, a nested PCR procedure was developed for the detection of G. abietina [19]. This method is based on the polymorphic sites in the ribosomal DNA (rDNA) internal transcribed spacer (ITS). The ITS evolves rapidly and significant variations within a species or even within a genome have been reported for fungi and plants [20-23]. The specific markers from the ITS region are, therefore, potentially unstable because of the high mutation rate, and would need to be validated by extensive sample testing. The 18S rRNA gene is much more conservative compared to the ITS. Specific markers developed from this DNA region are less likely to be invalid due to intraspecific variations producing false negative detections.
The aim of the present study was to develop a stable, specific and sensitive method for detecting G. abietina infection in a broad range of hosts. There was no intention to differentiate between the races/types of the fungus, which would be difficult using this conserved DNA segment. The 18S rRNA gene was sequenced from G. abietina isolated from four host species. Highly specific primers were designed for G. abietina, based on extensive sequence homology analysis. Two nested PCR systems were developed for sensitive detection of the pathogen in host tissues. The first system employed fungal universal primers in the first round PCR, to enrich the fungal rDNA in the plant genomic background, followed by the specific amplification of G. abietina rDNA in the second round PCR. The second system employed G. abietina specific primers in both PCR rounds. Both approaches can detect the presence of G. abietina in composite samples with high sensitivity. This procedure is rapid and can be used directly on plant materials without the need for fungal isolation and subsequent cultivation. The methods described here represent a promising alternative to disease inspection in forest nurseries, plantations and quarantine control facilities.
Results
Specific amplification of G. abietina
The fungal universal primer pair NS1/8 gave an amplification product of about 1700 bp for all of the 58 fungal strains tested, except Stachybotrys bisbyi which produced a larger PCR fragment (Fig. 2A). None of the 14 conifer species produced any amplicon with this primer pair, demonstrating the incompatibility of the primers to conifer 18S rDNA. The two pairs of specific primers, NS.Grem3/4 and NS.Grem5/6, amplified fragments of 1080 bp and 837 bp, respectively, only in G. abietina (Fig. 2B, C). No amplified products were generated from samples of any of the other fungi and plants with these two primer pairs. When used as inner primers in nested PCR, in combination with NS1/8 as outer primers, NS.Grem3/4 and NS.Grem5/6 gave amplification patterns identical to those in corresponding single-step PCR assays. The nested PCR using NS.Grem3/4 as outer primers and NS.Grem5/6 as inner primers gave an amplification pattern identical to that of NS.Grem5/6. No amplifications were observed for any of the plants and fungi except for G. abietina by any of the nested PCR (amplification pattern identical to Fig. 2B, C, thus not shown). These tests were performed in 3 – 5 replicates and the same specific amplification pattern was observed. The only difference between the nested PCR and the single-step specific PCR was that the amplification signals were stronger for the nested PCR. Thus, the prime pairs NS.Grem3/4 and NS.Grem5/6 are highly specific to G. abietina at annealing temperatures of 60°C and 56°C, respectively, both in single-step specific PCR and in two-step nested PCR. The lack of cross-amplification from any of the conifer species in either of the PCR systems demonstrates the scope for specific detection of the pathogen in all these hosts.
Evaluation of detection sensitivity
If the method is to be used for the early detection of infection in bulked samples in forest practice, a high level of sensitivity is required. Three different tests using a dilution series of G. abietina genomic DNA, with and without other background DNA, were conducted to compare the detection limits of the different PCR setups. The lowest DNA dilution that could provide a reproducible, unambiguous visible signal (in 3 μl PCR product) on ethidium bromide stained gels after electrophoresis was defined as the PCR detection limit.
Both the universal primers and the specific primers were tested first in single-step PCR assays. As shown in Fig. 3, all three pairs of primers can detect 0.15 ng DNA template. Among the three primer pairs, NS.Grem3/4 gave the strongest amplification signal indicating greater PCR efficiency than the other two. Nevertheless, the magnitude of detection sensitivity was the same for all three single-step PCR systems. By combining NS1/8 amplification with a second round amplification using either of the specific primer pairs in a nested PCR, the detection sensitivity dramatically increased. As little as 15 fg G. abietina DNA can be detected, approximately equivalent to a single fungal genome (Fig. 4a, b Test 1). The nested PCR using NS.Grem3/4 followed by NS.Grem5/6 gave a detection sensitivity similar to the two NS1/8-based methods though with even stronger amplification signal (Fig. 4c Test 1). Thus, the nested PCR is about 10,000 times more sensitive than the single step PCR assays.
To simulate the detection of G. abietina in a host's genomic background, the dilution series of G. abietina DNA was mixed with P. contorta DNA in equal volume and subjected to nested PCR analysis (Test 2, Table 2). The amplification resulted in a detection limit of 7.5 fg G. abietina DNA in the background of 6 ng of P. contorta DNA (i.e. a millionth of the quantity of host DNA, Fig. 4 Test 2). This test was performed in 3 – 5 replicates and the same magnitude of detection sensitivity was observed. This result is very similar to the detection limit achieved with G. abietina DNA alone (Fig. 4 Test 1). Thus, the presence of P. contorta DNA, even at very high relative concentrations, did not affect the efficiency of the nested PCR for detecting G. abietina.
To simulate the detection of G. abietina in a composite fungal background, the dilution series of G. abietina DNA was mixed with equal volumes of DNA from seven other fungi (Test 3, Table 2). In contrast to the previous two tests, this composite fungal DNA produced a visible amplification product (ca. 1.7 Kb in size) in all seven DNA dilution mixtures after the first round PCR with universal fungal primers (data not shown). This was due to the presence of background fungal DNA (ca. 9 ng) in all PCR mixes, no matter how little G. abietina DNA was present: the NS1/8 primers were compatible with all of the fungi in the mixture. Following the nested PCR, the detection limit was 0.075 ng G. abietina DNA in the background of 9 ng of other fungal DNA (i.e. about a hundredth of the fungal background DNA, Fig. 4a, b Test 3). This is a pronounced decrease in detection sensitivity compared to Tests 1 and 2. In contrast, the detection sensitivity of the nested PCR using the specific primers in both PCR steps was not affected by the presence of other fungal DNA, and gave the same detection sensitivity as in Test 1 and 2 (Fig. 4c Test 3).
Detection of G. abietina infection in Pinus contorta by nested PCR
To test the ability of the nested PCR systems to detect directly G. abietina infection in conifer trees, both brown and green needles from the infected trees of P. contorta were used for DNA isolation and PCR amplifications. All the nested PCR assays detected G. abietina in the needle samples from the infected twigs, which gave clear products of 1080 bp and 837 bp long with the inner primer pairs NS.Grem3/4 and NS.Grem5/6, respectively (Fig. 5). Neither the healthy P. contorta sample, nor the negative control produced any amplification product (Fig. 5). This verifies that the nested PCR product from the needle samples was amplified from G. abietina.
Discussion
Gremmeniella abietina is widespread and causes severe damage to several conifer species. Large-scale forest epidemics have been reported from several continents [1,6,24-26]. The disease can spread through infected seedling nurseries. Intercontinental migration of the pathogen has also been reported as a result of international transportation of infected forest materials [27,28]. Rapid detection methods that can be applied directly to asymptomatic tissues would be valuable for forest disease management. Previously reported methods for the morphological, biochemical, pathogenic or genetic characterization of this fungus require its isolation in culture [14,28-31]. Such characterization allows the species to be subdivided into different races and types. However, host differentiation of the fungus is very limited and different races/types can coexist in the same geographic region and infect the same host species [27,29,30,32,33]. Therefore, for forest management, a general detection method for G. abietina, regardless of race/type, would be highly desirable. The identical 18S rDNA sequences from G. abietina of NA, EU, LTT and STT race/type isolated from different hosts indicate that, in this fungus, the sequence is conservative. Therefore, markers based on it exhibit general intraspecific applicability. A high specificity of the detection system is a prerequisite for its application in pathogen diagnosis. The specific primers developed in this study successfully detected G. abietina at species level and, thus, can function as rapid molecular markers for its identification and detection in composite fungal or plant samples without the need for isolation and cultivation. By verifying these DNA markers in a wide range of conifer species, the present study indicates that this detection system can be applied to all potential hosts, so it should be a valuable forest management tool across broad geographic regions.
Apart from the specificity, the sensitivity of a detection system is also important for early infection diagnosis, particularly in bulk samples. The sensitivity of a PCR assay depends on several factors, most importantly on the primer composition, structure and homology to the target molecule. In this study, when the three pairs of primers, NS1/8, NS.Grem3/4 and NS.Grem5/6, were tested on G. abietina DNA in single-step PCR with the same number of cycles, NS.Grem3/4 was found to be ca. 10 times more efficient than the other two pairs. The higher amplification efficiency of NS.Grem3/4 was consistent across all PCR assays (Fig. 4). Thus, careful design and selection of the primers can significantly improve the sensitivity of a PCR assay.
Nested PCR was employed in this study to improve the detection sensitivity of the pathogen in the hosts. The use of universal fungal primers in the first round PCR enriches the fungal rDNA in the plant genomic background, then, in the second round, there is selective amplification of the target pathogen. This approach is particularly attractive when screening for multiple fungal pathogens in minor amount of plant tissue. The nested PCR developed in this study can detect as little as a single fungal genome even in high background levels of pine DNA. The template DNA concentration and composition can influence the efficiency of nested PCR. In the presence of a high proportion of other fungal DNA the detection limit of nested PCR with outer primers being the fungal universal primers NS1/8 was significantly decreased, mainly due to primer competition in the first round PCR. Since the universal primers are compatible with all the fungi in the mixture, the very small proportion of G. abietina present (<0.1%) would have little chance to compete for the primers. The target template was, therefore, not enriched in the first round PCR, which in turn affected the nested PCR sensitivity. This problem can be avoided by employing G. abietina-specific primers in both PCR rounds. The nested PCR based on this approach showed high detection sensitivity for G. abietina even in a high background of other fungal DNA. In real situations, the detection systems would usually be used on either suspected G. abietina infections or asymptomatic tissues. DNA isolated from these materials would mostly comprise the host DNA and DNA from endophytic fungi. G. abietina may or may not be the major component among the endophytic fungi. Thus, the nested PCR system using specific primers in both PCR steps would satisfy both the specificity and sensitivity requirements for diagnostic applications.
In the analysis of plant samples, extraction of sufficient fungal genomic DNA is also important since the fungal tissue is usually present at low levels relative to the amount of host tissue. If this is not achieved, the detection sensitivity may be inadequate and could result in a false negative. For large conifer trees, the stage and degree of the disease development as well as tissue sampling position would also affect the pathogen's detection. In the early stage of infection the amount and the spread of fungal mycelia is limited. Tissues from parts of the tree other than the close vicinity of the infection site may give negative detection. Thus, for large conifers multiple samples should be collected from the suspected tree for examination.
Conclusion
This study developed rapid, specific and sensitive detection systems for the conifer pathogen G. abietina. The specific markers were validated for a broad range of conifer hosts and fungi. Thus, the detection methods described here could have broad applications in forest protection and disease management programs. It should be also recognized that different race/types of G. abietina have distinct epidemiological and aetiological attributes and the ideal molecular assay should allow the user to identify not only the species but also the races and biotypes. The assay reported here could be used in combination with other race or biotype-specific assays [12,19], either in a multiplex or in a sequential fashion, to better understand the distribution and disease development of different G. abietina infections.
Methods
Fungal strains, plant species and genomic DNA extraction
Gremmeniella abietina was isolated from Pinus sylvestris, P. contorta, P. resinosa and P. banksiana (Table 1). Eleven isolates representing NA, EU, LTT and STT race/type were sequenced for the 18S rRNA gene (Table 1). Thirty-one other fungal strains, from 15 Ascomycota genera, were included in this study to determine the specificity of the markers developed for G. abietina (Table 2). These fungi were selected taking into account their 18S rDNA-based phylogenetic relationships to G. abietina [34], so both closely related and divergent groups were included. Pure cultures of these fungi were used for DNA isolation. Phylogenetic study has shown that species in the Helotiales and Rhytismatales are closely related to Gremmeniella [35] and many of them are plant pathogenies. We could not include them in the tests due to the lack of fungal material. However, 18S rDNA sequences of 40 species of Helotiales and Rhytismatales were downloaded from GenBank for sequence analysis.
Fourteen conifer species from three families (Pinaceae, Cupressaceae and Taxodiaceae) were selected to represent the potential range of hosts (Table 2). Since pines are most susceptible to this pathogen, eight pine species native to Asia, Europe and North America were selected, representing the two Pinus subgenera: Pinus and Strobus. Three other reported hosts of G. abietina were also included: Picea, Abies and Larix. Seeds of each species were germinated on sterilized Petri dish for 2 – 3 weeks and used for DNA isolation. Twigs from six infected trees of P. contorta were collected in the forest of northern Sweden. From these, both brown and green needles were collected for DNA isolation.
The fungal genomic DNAs were isolated from a pure culture of each strain following the procedure described by Wu et al [36]. The genomic DNAs of the conifer species were isolated using a DNeasy® Plant Mini Kit (Qiagen, Germany). The pine needles were thoroughly homogenized, as follows. Two ceramic beads, 4 mm in diameter (Iuchi, Japan) and 350 mg of 0.5 mm zirconia-silica beads (Biospec Products, Inc., Bartlesville, OK, USA), were placed in a 2-ml microtube containing 100 mg pine needles. The tubes were placed in a Mini-Bead Beater (Biospec Products, Inc.) and homogenized for 2 min at the maximum speed. The rest of the isolation procedure followed that suggested by the manufacturer of the DNeasy Plant Mini Kit (Qiagen, Germany).
Specific primers
The 11 G. abietina isolates sequenced for the 18S rRNA gene produced identical sequences, with one exception of a single substitution among them (GenBank accession numbers see Table 1). This sequence was aligned with the 18S rRNA sequences of the 31 other fungal strains listed in Table 2 as well as 40 Helotiales and Rhytismatales fungi accessed from GenBank (data not shown). Unique sequence patterns to G. abietina were utilized to design specific primers. Two pairs of primers, NS.Grem3/4 and NS.Grem5/6 (Fig. 1), were designed for G. abietina. To ensure the specificity of the PCR assay, these primers were first screened against sequences in GenBank using the BLAST function to examine their possible homology to other fungi. The "Search for short, nearly exact matches" program was used. These primers showed >20% mismatches to any other fungal sequences in GenBank including the 40 18S rDNA sequences of Helotiales and Rhytismatales. Under stringent PCR conditions, a >20% mismatch between the target molecule and the primer would not result in specific amplification.
Nested PCR
To increase the detection sensitivity, two nested PCR systems were developed. One approach used the 18S rDNA-based universal fungal primers NS1 and NS8 [37] as outer primers in the first round PCR. Amplification was performed in a volume of 25 μl containing 1 – 5 ng of template DNA, 10 pmol of each primer, 0.75 U of Taq DNA polymerase (Invitrogen Life Technologies, USA), 200 μM of each dNTP (Amersham Pharmacia Biotech, USA), and 1.5 mM MgCl2. PCR conditions were optimized to comprise an initial denaturation of 3 min at 95°C, followed by 36 cycles of 94°C for 30 s, 45°C for 45 s and 72°C for 90 s, followed by a final extension of 10 min at 72°C. A 1 μl aliquot of the first round PCR product was used as the template in the second round, using the G. abietina-specific primers NS.Grem3/4 or NS.Grem5/6. The PCR conditions for these two specific primer pairs were similar to the NS1/8 amplification, except that the PCR cycles were decreased to 25 in the second round PCR and the annealing temperatures were 60°C and 56°C for NS.Grem3/4 and NS.Grem5/6, respectively. In another approach, G. abietina-specific primer pair NS.Grem3/4 was used in the first round PCR and NS.Grem5/6 in the second round. The PCR procedure is the same as that described above. A negative control was included in all PCR runs. PCR products (3 μl) were analyzed by electrophoresis on 1.4% agarose gels in 1× TAE buffer. The gels were stained with ethidium bromide and visualized under UV light using a Gel Doc 2000 fluorescent gel documentation system (Bio-Rad, USA).
Sensitivity evaluation
To determine the detection limit of the nested PCR, three DNA dilution series were created and subjected to the PCR analysis (Table 3). First, a 10-fold dilution series of G. abietina genomic DNA was tested (Test 1). The initial DNA concentration of 5 ng/μl was quantified using a GeneQuant Pro RNA/DNA Calculator spectrophotometer (Amersham Biosciences, Sweden). To simulate the detection of infection in a host, this dilution series of G. abietina DNA was mixed with genomic DNA (4 ng/μl) of P. contorta in equal volume (Test 2). A 3 μl aliquot of this mixture at each dilution (equivalent to 1.5 μl of G. abietina DNA solution plus 1.5 μl of P. contorta genomic DNA) was used in each PCR (Test 2). A third test was conducted to simulate the detection of G. abietina in a mixed fungal background. For this, genomic DNAs of seven fungi Aspergillus ochraceus, Penicillium brevicompactum, Trichoderma viride, Eurotium herbariorum, Fusarium culmorum, Ulocladium botrytis and Phacidium infestans were mixed in equal quantities. The final concentration of this composite DNA was 6 ng/μl. The 10-fold dilution series of G. abietina DNA was mixed with this composite fungal DNA in equal volume for PCR analyses (Test 3). A 3 μl aliquot of this mixture at each dilution was used in each PCR (Test 3). To compare the sensitivity and specificity of single-step PCR and nested PCR, the 10-fold dilution series of G. abietina genomic DNA was also analyzed directly with the inner specific primers NS.Grem3/4 and NS.Grem5/6 in a single-step PCR. All the experiments conducted in this study were repeated 3 – 5 times.
Authors' contributions
XRW and QYZ designed and conducted the experiments. PH, provided the Swedish G. abietina isolates, and collected and identified the infected P. contorta twigs. All authors read and approved the final manuscript.
Acknowledgements
We thank Prof Antti Uotila, Helsinki University, Finland, and Dr Gaston Laflamme, Laurentian Forestry Centre, Canadian Forest Service, for providing the Gremmeniella isolates from Finland and US and Canada. We also thank Prof Margareta Karlman, Dr Jesper Witzell and Andreas Bernhold, the Swedish University of Agricultural Sciences, for collecting field samples of Gremmeniella in the northernmost of Sweden. This study was supported by grants from the Swedish Research Council for Environment, Agricultural Sciences and Spatial Planning (Formas 24.0372/99), the Swedish Council for Working Life and Social Research (FAS) and the Swedish University of Agricultural Sciences.
Figures and Tables
Figure 1 Location and sequences of the primers used in this study.
Figure 2 Specificity test of different PCR assays. A, amplification using fungal universal primer pair NS1/8; B and C, amplification using specific primer pairs NS.Grem3/4 and NS.Grem5/6, respectively. G. abietina isolates 1–19 were not shown on this gel due to space limitation. Nested PCR using NS.Grem3/4 and NS.Grem5/6 as inner primers gave amplification patterns identical to B and C, respectively. Fungal strains in each lane are in the order given in Table 1 and 2. M: 1 kb Plus DNA Ladder (Invitrogen).
Figure 3 Comparison of detection sensitivity between single-step PCR assays. A, universal primers NS1/8; B, specific primers NS.Grem3/4; C, specific primers NS.Grem5/6. A 10-fold dilution series of G. abietina was used as the template.
Figure 4 Comparison of the detection sensitivity between different nested PCR setups. Test 1, G. abietina; Test 2, G. abietina mixed with P. contorta; Test 3, G. abietina mixed with seven other fungi, in amplifications with NS1/8-NS.Grem3/4 (a), NS1/8-NS.Grem5/6 (b) and NS.Grem3/4-NS.Grem5/6 (c). See Table 3 for DNA contents in each lane.
Figure 5 Detection of G. abietina in P. contorta trees. Lanes 1–6: needles from six infected twigs. Lanes 7–9: positive controls using G. abietina DNA. Lane 10: healthy P. contorta. Lane 11: negative control. M: 1 kb Plus DNA Ladder (Invitrogen). Panels A, B and C show the nested PCR results of NS1/8-NS.Grem3/4, NS1/8-NS.Grem5/6 and NS.Grem3/4-NS.Grem5/6, respectively.
Table 1 Gremmeniella abietina isolates examined in this study.
Isolate Sequenced Strain type Pine host Origin
1 G. abietina ALI 531 EU, LTT P. contorta Vindeln, Sweden
2 G. abietina ALI 532 EU, LTT P. contorta Vindeln, Sweden
3 G. abietina ALI 533 EU, LTT P. contorta Vindeln, Sweden
4 G. abietina ALI 534 EU, LTT P. contorta Vindeln, Sweden
5 G. abietina ALI 569 DQ084550 EU, STT P. sylvestris Arctic circle, Finland
6 G. abietina ALI 570 DQ084551 EU, STT P. sylvestris Finland
7 G. abietina ALI 571 DQ084552 EU, STT P. sylvestris Finland
8 G. abietina ALI 572 DQ084553 EU, LTT P. sylvestris Finland
9 G. abietina ALI 573 DQ084554 EU, LTT P. sylvestris Finland
10 G. abietina ALI 574 DQ084555 EU, LTT P. sylvestris Finland
11 G. abietina US 810105 DQ084556 EU P. resinosa USA
12 G. abietina US 790048 EU P. resinosa USA
13 G. abietina CF 910032 DQ084557 NA P. banksiana Canada
14 G. abietina ALI G90 AF548076 EU, STT P. contorta Åsele, Sweden
15 G. abietina ALI G148 AF548075 EU, STT P. contorta Åsele, Sweden
16 G. abietina ALI G139 AF548074 EU, STT P. contorta Åsele, Sweden
17 G. abietina ALI G3 EU, STT P. sylvestris Nattavaara, Sweden
18 G. abietina ALI G4 EU, STT P. sylvestris Nattavaara, Sweden
19 G. abietina ALI G6 EU, STT P. sylvestris Nattavaara, Sweden
20 G. abietina ALI G15 EU, STT P. sylvestris Nattavaara, Sweden
21 G. abietina ALI G16 EU, STT P. sylvestris Nattavaara, Sweden
22 G. abietina ALI G17 EU, STT P. sylvestris Nattavaara, Sweden
23 G. abietina ALI F113 EU, STT P. contorta Hede, Sweden
24 G. abietina ALI F114 EU, STT P. contorta Hede, Sweden
25 G. abietina ALI F116 EU, STT P. contorta Ramsele, Sweden
26 G. abietina ALI F129 EU, STT P. sylvestris Östersund, Sweden
27 G. abietina ALI F174 EU, STT P. sylvestris Östersund, Sweden
Table 2 Other fungal strains and conifer species included in this study.
Fungal strains Conifer species
28 Aspergillus niger UPSC 1769 59 Pinus sylvestris
29 Aspergillus ochraceus UPSC 1983 60 Pinus contorta
30 Aspergillus flavus UPSC 1768 61 Pinus massoniana
31 Aspergillus penicilloides ALI 231 62 Pinus banksiana
32 Aspergillus versicolor UPSC 2027 63 Pinus jeffreyi
33 Aspergillus silvaticus ALI 234 64 Pinus strobus
34 Cladosporium cladosporioides ALI 50 65 Pinus koraiensis
35 Chrysonilia sitophila ALI 346 66 Pinus yunnanensis
36 Eurotium herbariorum ALI 216 67 Picea abies
37 Fusarium culmorum UPSC 1981 68 Larix deciduas
38 Microdochium nivale UPSC 3273 69 Abies procera
39 Mucor plumbeus UPSC 1492 70 Chamaecyparis nootkatensis
40 Paecilomyces variotii UPSC 1651 71 Tsuga canadensis
41 Penicillium commune CBS 343.51 72 Taxodium distichum
42 Penicillium italicum UPSC 1577
43 Penicillium chrysogenum UPSC 2020
44 Penicillium brevicompactum ALI 319
45 Penicillium frequentans ALI 218
46 Rhizopus microsporus UPSC 1758
47 Stachybotrys bisbyi CBS 317.72
48 Stachybotrys chartarum CBS 330.37
49 Stachybotrys dichroa CBS 182.80
50 Stachybotrys oenanthes CBS 252.76
51 Stachybotrys kampalensis CBS 388.73
52 Stachybotrys microspora CBS 186.79
53 Trichoderma reesei QM 9414
54 Trichoderma viride ALI 210
55 Ulocladium botrytis CBS 173.82
56 Wallemia sebi UPSC 2502
57 Phacidium infestans A387
58 Phacidium infestans A391
Table 3 DNA dilutions and mixtures used in the sensitivity test. The relative abundance of G. abietina DNA to the other genomic background DNA is indicated in parentheses.
Test 1 G. abietina Test 2 G. abietina + P. contorta Test 3 G. abietina + 7 fungi mix
Sample
1 5 × 10-1 ng/μl 5 × 10-1 ng/μl + 4 ng/μl (1:8) 5 × 10-1 ng/μl + 6 ng/μl (1:12)
2 5 × 10-2 ng/μl 5 × 10-2 ng/μl + 4 ng/μl (1:80) 5 × 10-2 ng/μl + 6 ng/μl (1:120)
3 5 × 10-3 ng/μl 5 × 10-3 ng/μl + 4 ng/μl (1:800) 5 × 10-3 ng/μl + 6 ng/μl (1:1200)
4 5 × 10-4 ng/μl 5 × 10-4 ng/μl + 4 ng/μl (1:8000) 5 × 10-4 ng/μl + 6 ng/μl (1:12000)
5 5 × 10-5 ng/μl 5 × 10-5 ng/μl + 4 ng/μl (1:80000) 5 × 10-5 ng/μl + 6 ng/μl (1:120000)
6 5 × 10-6 ng/μl 5 × 10-6 ng/μl + 4 ng/μl (1:800000) 5 × 10-6 ng/μl + 6 ng/μl (1:1200000)
7 5 × 10-7 ng/μl 5 × 10-7 ng/μl + 4 ng/μl (1:8000000) 5 × 10-7 ng/μl + 6 ng/μl (1:12000000)
3 μl in PCR 1:1 vol. mix, 3 μl in PCR 1:1 vol. mix, 3 μl in PCR
==== Refs
Donaubauer E Distribution and hosts of Scleroderris lagerbergii in Europe and North America Eur J For Pathol 1972 2 6 11
Hellgren M Gremmeniella abietina - disease biology and genetic variation within Fennoscandia Department of Forest Mycology and Pathology 1995 Uppsala, Swedish University of Agricultural Sciences
Skilling DD Riemenschneider DE Manion PD Screening conifers for resistance to Gremmeniella abietina. Proc Int Symposium Scleroderris Canker of Conifers 1983 Syracuse, USA, The Hague, Martinus Nijhoff - Dr. W. Junk Publisher 189 196
Barklund P Rowe J Gremmeniella abietina (Scleroderris lagerbergii), a primary parasite in a Norway spruce dieback Eur J For Path 1981 11 97 108
Hellgren M Barklund P Studies of the life-cycle of Gremmeniella abietina on scots pine in southern Sweden Eur J For Path 1992 22 300 311
Karlman M Hansson P Witzell J Scleroderris canker on lodgepole pine introduced in northern Sweden Can J For Res 1994 24 1948 1959
Lang KJ Schütt P Anatomische untersuchungen zur infektionsbiologie von Scleroderris lagerbergii Gr (Brunchorstia pinea (Karst.) von Höhn.) Eur J For Pathol 1974 4 166 174
Marosy M Patton RF Upper CD A Conducive Day Concept to Explain the Effect of Low-Temperature on the Development of Scleroderris Shoot Blight Phytopathology 1989 79 1293 1301
Dorworth CE Krywienczyk J Comparisons among isolates of Gremmeniella abietina by means of growth rate, conidia measurement, and immunogenic reaction Can J Bot 1975 53 2506 2525
Uotila A Physiological and morphological variation among Finnish Gremmeniella abietina isolates Commun Inst For Fenn 1983 119 1 12
Petrini O Toti L Petrini LE Heiniger U Gremmeniella abietina and G. laricina in Europe: characterization and identification of isolates and laboratory strains by soluble protein electrophoresis Can J Bot 1990 68 2629 2635
Hamelin RC Ouellette GB Bernier L Identification of Gremmeniella abietina races with random amplified polymorphic DNA markers Appl Environ Microbiol 1993 59 1752 1755 16348951
Lecours N Toti L Sieber TN Petrini O Pectic Enzyme Patterns as a Taxonomic Tool for the Characterization of Gremmeniella Spp Isolates Can J Bot 1994 72 891 896
Müller MM Uotila A The diversity of Gremmeniella abietina var. abietina FAST-profiles Mycol Res 1997 101 557 564 10.1017/S0953756296002894
Dusabenyagasani M Lecours N Hamelin RC Sequence-tagged sites (STS) for studies of molecular epidemiology of scleroderris canker of conifers Theor Appl Genet 1998 97 789 796 10.1007/s001220050957
McCartney HA Foster SJ Fraaije BA Ward E Molecular diagnostics for fungal plant pathogens Pest Manag Sci 2003 59 129 142 12587866 10.1002/ps.575
Grote D Olmos A Kofoet A Tuset JJ Bertolini E Cambra M Specific and sensitive detection of Phytophthora nicotianae by simple and nested-PCR Eur J Plant Path 2002 108 197 207 10.1023/A:1015139410793
Renker C Heinrichs J Kaldorf M Buscot F Combining nested PCR and restriction digest of the internal transcribed spacer region to characterize arbuscular mycorrhizal fungi on roots from the field Mycorrhiza 2003 13 191 198 12938031 10.1007/s00572-002-0214-5
Hamelin RC Bourassa M Rail J Dusabenyagasani M Jacobi V Laflamme G PCR detection of Gremmeniella abietina, the causal agent of Scleroderris canker of pine Mycol Res 2000 104 527 532 10.1017/S0953756299002026
O'Donnell K Ribosomal DNA internal transcribed spacers are highly divergent in the phytopathogenic ascomycete Fusarium sambucinum (Gibberella pulicaris) Curr Genet 1992 22 213 220 1525873 10.1007/BF00351728
Gernandt DS Liston A Pinero D Variation in the nrDNA ITS of Pinus Subsection Cembroides: Implications for Molecular Systematic Studies of Pine Species Complexes Mol Phylogenet Evol 2001 21 449 467 11741386 10.1006/mpev.2001.1026
James TY Moncalvo JM Li S Vilgalys R Polymorphism at the ribosomal DNA spacers and its relation to breeding structure of the widespread mushroom Schizophyllum commune Genetics 2001 157 149 161 11139499
Ko KS Jung HS Three nonorthologous ITS1 types are present in a polypore fungus Trichaptum abietinum Mol Phylogenet Evol 2002 23 112 122 12069544 10.1016/S1055-7903(02)00009-X
Dorworth CE Epidemiology of Scleroderris lagerbergii in central Ontario Can J Bot 1972 50 751 765
Skilling DD Schneider B Dasking D Biology and control of scleroderris canker in North America 1986 , USDA Forest Service, North Central Experimental Station
Nevalainen S Gremmeniella abietina in Finnish Pinus sylvestris stands in 1986-1992: a study based on the National Forest Inventory Scand J Forest Res 1999 14 111 120 10.1080/02827589950152836
Dorworth CE Krywienczyk J Skilling DD New York isolates of Gremmeniella abietina (Scleroderris lagerbergii) identical in immunogenic reaction to European isolates Plant Dis Rep 1977 61 887 890
Hamelin RC Lecours N Laflamme G Molecular evidence of distinct introductions of the European race of Gremmeniella abietina into North America Phytopathology 1998 88 582 588
Hansson P Wang XR Szmidt AE Karlman M RAPD variation in Gremmeniella abietina attacking Pinus sylvestris and Pinus contorta in northern Sweden Eur J For Path 1996 26 45 55
Hantula J Muller MM Variation within Gremmeniella abietina in Finland and other countries as determined by Random Amplified Microsatellites (RAMS) Mycol Res 1997 101 169 175 10.1017/S0953756296002225
Wang XR Genetic variability in the canker pathogen fungus, Gremmeniella abietina. contribution of sexual compared with asexual reproduction Mycol Res 1997 101 1195 1201 10.1017/S0953756297003791
Hansson P Susceptibility of different provenances of Pinus sylvestris, Pinus contorta and Picea abies to Gremmeniella abietina Eur J For Path 1998 28 21 32
Kaitera J Seitamaki L Jalkanen R Morphological and ecological variation of Gremmeniella abietina var. abietina in Pinus sylvestris, Pinus contorta and Picea abies sapling stands in northern Finland and the Kola Peninsula Scand J Forest Res 2000 15 13 19 10.1080/02827580050160420
Wu Z Tsumura Y Blomquist G Wang XR 18S rRNA Gene Variation among Common Airborne Fungi, and Development of Specific Oligonucleotide Probes for the Detection of Fungal Isolates Appl Environ Microbiol 2003 69 5389 5397 12957927 10.1128/AEM.69.9.5389-5397.2003
Gernandt DS Platt JL Stone JK Spatafora JW Holst-Jensen A Hamelin RC Kohn LM Phylogenetics of Helotiales and Rhytismatales based on partial small subunit nuclear ribosomal DNA sequences Mycologia 2001 93 915 933
Wu Z Wang XR Blomquist G Evaluation of PCR primers and PCR conditions for specific detection of common airborne fungi J Environ Monit 2002 4 377 382 12094931 10.1039/b200490a
White TJ Bruns T Lee S Taylor J Innis MA, Gelfand DH, Sninsky JJ and White TJ Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics PCR protocols: A guide to methods and applications 1990 San Diego, Academic Press 315 322
|
16280082
|
PMC1298302
|
CC BY
|
2021-01-04 16:03:41
|
no
|
BMC Microbiol. 2005 Nov 9; 5:65
|
utf-8
|
BMC Microbiol
| 2,005 |
10.1186/1471-2180-5-65
|
oa_comm
|
==== Front
BMC Med Res MethodolBMC Medical Research Methodology1471-2288BioMed Central London 1471-2288-5-361627447310.1186/1471-2288-5-36Research ArticleA non-parametric framework for estimating threshold limit values Salanti Georgia [email protected] Kurt [email protected] MRC Biostatistics Unit, Cambridge, UK2 Institute for Medical Statistics and Epidemiology, Technical University of Munich, Germany2005 7 11 2005 5 36 36 25 2 2005 7 11 2005 Copyright © 2005 Salanti and Ulm; licensee BioMed Central Ltd.2005Salanti and Ulm; 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
To estimate a threshold limit value for a compound known to have harmful health effects, an 'elbow' threshold model is usually applied. We are interested on non-parametric flexible alternatives.
Methods
We describe how a step function model fitted by isotonic regression can be used to estimate threshold limit values. This method returns a set of candidate locations, and we discuss two algorithms to select the threshold among them: the reduced isotonic regression and an algorithm considering the closed family of hypotheses. We assess the performance of these two alternative approaches under different scenarios in a simulation study. We illustrate the framework by analysing the data from a study conducted by the German Research Foundation aiming to set a threshold limit value in the exposure to total dust at workplace, as a causal agent for developing chronic bronchitis.
Results
In the paper we demonstrate the use and the properties of the proposed methodology along with the results from an application. The method appears to detect the threshold with satisfactory success. However, its performance can be compromised by the low power to reject the constant risk assumption when the true dose-response relationship is weak.
Conclusion
The estimation of thresholds based on isotonic framework is conceptually simple and sufficiently powerful. Given that in threshold value estimation context there is not a gold standard method, the proposed model provides a useful non-parametric alternative to the standard approaches and can corroborate or challenge their findings.
==== Body
Background
Estimation of a threshold limit value (TLV) is an important task in many medical areas, where risk factors are often scrutinized for values beyond which important medical or political decisions need to be taken – e.g. beyond which blood pressure value one should prescribe antihypertensive. The classical approach suggests several steps; a dose-response relationship needs to be established by applying a test for trend and then a set of candidate threshold values xi is identified. Finally, a threshold model is fitted at each of these candidate values. Considering an exposure variable in N doses x1,x2,...,xN, a widely applied threshold model for a binary outcome p(xi) is
where f is the identity function for the 'elbow' shape or the logit function for the logistic version, and l is the baseline risk. Then, the threshold is estimated as the exposure xi associated with 'better model' in terms of a goodness of fit criterion, e.g. the likelihood ratio statistics or the Akaike's information criterion. Usually only a set of plausible threshold locations is considered. To determine the set of candidate threshold locations, one has to screen visually the dose-response regression line and take a neighbourhood around a point where a 'jump' in the risk seems to occur. This practice, although widely applied in practice, is prone to bias and does not control efficiently the type I error.
A justified approach is to consider all doses xi as possible thresholds. The model
f(p(xi) = l + b log (xi/t)) if xi >t (2)
has been previously proposed together with a likelihood maximisation method for estimate the threshold and its confidence interval [1]. Modifications and further developments on such threshold value models can be found in [2,3].
A useful alternative to these parametric approaches is provided by isotonic regression. The main advantage compared to the models developed so far is that no specific assumption is made regarding the shape of the regression and a flexible step function is fitted. Moreover, the fitting algorithm automatically selects a small set of candidate threshold values without any a priori information about their location.
An underlying assumption throughout this paper is that a threshold value exists in every dose-response relationship. Evaluating this assumption is rather controversial. We align ourselves with those arguing that the question whether a threshold exists or not cannot be answered by means of statistical analysis [1,4]. Assuming a threshold is plausible in many toxicological and clinical studies, and even in cases where there is no biological justification, the threshold assumption can have practical implications.
Under this scope, we constrain this paper on presenting a framework for estimating the TLV given that its existence has been established as plausible. We explain the use of isotonic regression and we discuss how to select the actual threshold among the candidate locations suggested by the isotonic transformation.
Methods
Threshold estimation and isotonic regression
Isotonic regression is a maximum likelihood estimator under the assumption of a monotone dose response relationship. Whereas several algorithms are possible to fit the data, we use the Pooled Adjacent Violators Algorithm (PAVA) as it is the most efficient and comprehensive (see appendix) [5]. The result is a step function that summarizes the exposure in L constant risk groups (level sets) that are automatically selected by the algorithm without any a priori information about the location of the changepoints (here called isotonic cutpoints). The isotonic model can be thought of as a categorisation's procedure, where a set of cutpoints for the exposure variable is estimated with respect to the monotonicity in the associated risk [6].
Based on isotonic regression, we propose a two-stages approach, where all possible xi doses are evaluated. In the first stage, we take advantage of the fact that monotonicity is a pre-requisite condition for the threshold hypothesis. At this step, PAVA is screening all xi for compliance with monotonicity and re-estimates the regression shape under this condition. This screening is very powerful, given that isotonic regression provides a powerful and robust test for trend. Variations of the isotonic test and description of its advantages have been outlined in several papers [7-10]. The result is a step function with few cutpoints.
Once isotonic regression is fitted, the second step needs to be taken; the actual threshold needs to be selected among the isotonic cutpoints. On this purpose, we propose two methods described in the following sections.
The framework based on isotonic regression is in a sense equivalent to the methods that check all the dose values; however in isotonic regression most of the dose values are 'rejected' during the first stage on the basis of their compliance to the hypothesis of monotonicity, and only few are tested under the threshold value location hypothesis. An assumption underlying the proposed methodology is that the observations can be grouped in constant-risk exposure intervals. Whereas we acknowledge that this assumption may be questionable depending on the nature of the exposure, it is true for many environmental and toxicological factors.
Selecting the threshold
Reduced isotonic regression
The cutpoints resulting from PAVA are estimated so that monotonicity is achieved and they do not necessarily correspond with a significant increase in the risk. The model can become more parsimonious if these 'non significant' level sets are eliminated. This parsimonious version is called reduced isotonic regression (RIR) and has been described and studied elsewhere [6,11]. When the outcome is binary, the reduction is accomplished by a sequence of Fisher tests for the adjacent 2 × 2 tables. The correction for multiple comparisons is made by a-priori estimation of the elimination significance level ε* in a permutation procedure so that the actual type I error is kept constant at a (e.g. a = 5%). The result is a step function with less level sets than the original isotonic model. Moreover, each cutpoint defines now a significant increase in the associated risk. Consequently the first 'step' in an RIR line will indicate the TLV. The model can be thought of as an extension of the 'elbow' threshold model in equation (1) where the first level set estimates the background risk l and the first cutpoint defines the threshold t.
Closed testing procedure
An alternative for selecting the threshold out of several candidate locations suggested by PAVA is to consider the closed family of hypotheses. The classical closed testing approach suggests that a hypothesis H is tested only if all hypotheses that contain H have been rejected at some fraction of the significance level a. The family-wise error is controlled, but power is usually low.
An option to increase the power is to make one part of the regression line conditional to the other [12]. This concept suits the TLV estimation context where the beginning of the dose-response is more important than its end and testing for threshold among levels of higher exposure is conditional on the rejection for the lower exposure levels.
The diagram in figure 1 presents a closed family of hypotheses. Every 'vertical' hypothesis about testing an isotonic cutpoint xi is conditional on the rejection of the hypothesis above (vertical conditioning) and testing every hypothesis on the right hand side is conditional on the retention of the hypothesis on its left (horizontal conditioning).
Figure 1 Closed testing procedure. The arrows show the direction of the conditional testing.
1. Every hypothesis H(l,k):' the risk is constant between two isotonic cutpoints xl and xk' and its nested hypotheses are conditional on the rejection of every hypothesis H(l',k') with l< l' and k< k'
2. Retain every hypothesis implied by any other hypothesis that has not been rejected
Consider for example that isotonic regression suggests 4 isotonic cutpoints as candidate thresholds defining 4 level sets with associated risks p1,...,p4 (figure 1). The direction of the arrows shows the conditioning in testing. The procedure starts by evaluating H(1,3) (between x1 and x3 the risk is constant, i.e. p1 = p2 = p3). If we retain, then we test H(2,4) (rejection: threshold = x3, no-rejection: no dose-response relationship). If we reject H(1,3) then we test H(1,2) (rejection: threshold = x1, no-rejection: continue by testing H(2,3)). In every step, the exact isotonic test for trend is used [10].
When the conditioning occurs 'horizontally' in addition to the vertical restriction as in figure 1 (dotted arrows), then the power increases. Consider that H(1,4) is not true due to H(1,2) being not true. Then with only vertical restriction the probability to correctly reject H(1,2) using α = 5% is 0.955+0.9530.052+0.9520.05+0.9540.05 = 0.862 whereas with the horizontal conditioning it is 0.952+0.9520.05 = 0.947.
Compared to RIR, this approach is easier to apply, but provides no information about the shape of the dose-response relationship after the threshold. Closed testing elimination concentrates on increases in the risk whereas RIR achieves a complete re-estimation of the regression line with overall improvement in the fit. For both methods, bootstrap can be used to calculate the confidence intervals.
Extension: The isotonic surfaces model
There are situations where thresholds need to be identified for multiple factors that interact. One of the main advantages of the presented methods is that they can be easily combined with multivariate isotonic models; either the isotonic-surfaces model where the level sets correspond to combinations of the predictor variables or the additive isotonic model [6,13,14].
Consider two continuous predictors x and y, a binary outcome and their three-dimensional scatter plot. The isotonic surfaces model is simply a surface fitted in the scatter plot that is monotone along both x- and y-axes. Two-dimensional blocks of constant response are built, and a reducing procedure similar to the one followed in univariate regression is applied to improve parsimony. Details on fitting and elimination algorithms can be found in [6]. This model can be used to estimate two-dimensional thresholds, as we exemplify in the application.
Simulation study
The performance of the two methods described above as threshold detectors depends upon their power to establish a dose-response relationship. Consequently, in this simulation study we first concentrate on evaluating the power to reject the constant risk assumption. Higher power is associated here with greater proportion of non-constant estimated regression lines. Subsequently we evaluate the ability to detect the isotonic level set that is associated with a threshold, conditional to the rejection of the constant risk assumption.
We simulate under 5 isotonic level sets, with equal number of observations per group that varies from 50 to 250. Four shapes for the isotonic groups are studied. For each combination of slope (for the increasing part of the regression) and sample size, 1000 simulations are analyzed (figure 2).
Figure 2 Regression shapes studied in the simulation study.
Shape A assumes no dose-response and the type I error of the methods as 'tests for trend' can be accessed. Shape B corresponds to a linearly increasing relationship starting with 20% risk and a threshold is assumed at the first dose group. Shape C represents a segmented regression line assuming a threshold at the second dose level. The baseline constant risk is 20% and afterwards the risk increases linearly. The last type of regression (Shape D) is also a segmented line but assumes a threshold at the first cutpoint. Between the first two dose groups the risk increases linearly and is flat between the second and the fifth. This dose-response relationship is close to the horizontal line (figure 2). In every shape that includes linear regression part we studied different slopes of 2%, 5%, 10% and 15%.
Application to MAK study
The study "Maximale Arbeitsplatz-Konzentration" was conducted by the German Research Foundation and comprises among others a cohort of 920 smoking workers exposed to a mixture of dust, mainly from iron, steel, foundry and engineering [15,16]. The endpoint of the study was chronic bronchitic reaction (CBR). Dust concentration and the length of exposure have been previously established as important risk factors. The goal in this present application is first to estimate a TLV for the cumulative exposure in dust over time in a univariate regression and subsequently to assess a two-dimensional threshold for both dust concentration and duration of exposure.
Results
Results from simulations
Establishing dose-response relationship
When RIR is applied, a dose-response relationship is established for shape A in 5% of the simulated samples. This was somewhat expected, since the elimination procedure is designed to keep type I error fixed at the nominal level [6]. The closed testing approach yields an error rate of 2%.
Table 1 presents the estimated power to establish a dose-response relationship in shapes B-D for sample size 100 and the two lower slopes. For comparability between the two methods, the calculations were carried out by calibrating ε* so that the error rate is 2% for RIR. The RIR performs better for shapes B and C, and although for flat regression lines is not very powerful, its power increases very fast with higher slopes or sample size. The power lies between 78% and 99% for slopes higher than 5% for every sample size. For shape D, both methods are largely underpowered with closed testing procedure having double the power of RIR. In this regression line, the contrast in the risk is between a single level set versus four levels having higher but equal risk, and both elimination procedures are likely to miss it and pool all level sets together. For RIR sample size as high as 200 observations per level set and a slope higher than 10% are required to achieve power of at least 65%.
Table 1 The average power for reduced isotonic regression and closed testing procedure in establishing a dose-response relationship for the two lower slopes and sample size 100 observations per level set.
Shape Slope Power to establish dose-response relationship
Closed testing RIR
B 2% 19% 27%
5% 29% 88%
C 2% 10% 13%
5% 32% 40%
D 2% 7% 2%
5% 19% 10%
Assigning thresholds to the level sets
The assignment of thresholds was studied among the datasets where a dose-response relationship was established. Closed testing procedure was less successful than RIR for slopes lower than 5% and sample size less than 100 observations per isotonic level set. For these values, the estimated threshold in shapes B and D was higher than the first level set – and thus overestimated- in 99% of the cases. It is only after sample size 150 that closed testing procedure starts getting a bit more successful, following similar patterns as the ones observed for RIR.
Table 2 presents the threshold assignment for RIR. The estimated thresholds in shape B do not seem to follow any specific pattern for slope 2% where every cutpoint has more or less the same probabilities to be selected as threshold. However, for slopes higher than 5% or greater sample size, the first group is most likely to contain the threshold.
Table 2 Average probability (for sample size and slope) to select a threshold at a given isotonic cutpoint applying RIR, average for sample size and slope. In italic appears the number that is considered to be the pertinent estimation in each shape.
Probability to select an isotonic cutpoint as threshold among the non-constant lines
Shape 1st 2nd 3rd 4th
B 47% 37% 14% 3%
C 4% 68% 24% 5%
D 96% 3% 1% 0.5%
Shape C is of particular interest regarding detection of the threshold location. The probability to assess it correctly, averaged for sample size and slope was 68%. Figure 3 presents the power in greater details for every slope and sample size. The ability to detect the threshold increases sharply with the slope and sample size in an almost linear way. It is remarkable that the probability to assess a threshold at the first level set is very low (4%) and decreases with sample size. Both methods present an important tendency to assign thresholds to the adjacent group that corresponds to a higher dose is observed (average probability 24%). The estimated background risk is slightly biased and lies within 20–20.6%. When the true threshold value was assumed to be at the third dose level, we did not observe any important differences in the performance of both methods apart from a slight drop in the already low probability to assign a threshold to the first cutpoint.
Figure 3 Power of reduced isotonic regression to detect the correct threshold in shape C for different slopes. The size of the circle is proportional to the sample size per dose group (N = 50, 100, 150, 200, 250).
In shape D, the first group has the greatest probability to be selected as threshold, which is in agreement with the underlying regression shape.
Application to CBR study
Figure 4 depicts the fitted isotonic regression with 11 dose groups along with its reduced version (4 dose-groups) truncated for clarity at 250 mg/m3year. A smoothing spline with 6 degrees of freedom shows that the level sets defined by RIR are reasonable. The threshold was estimated at 7 mg/m3year with 95% bootstrap confidence intervals (4.9, 10) mg/m3year and background risk of 7.6%.
Figure 4 Full isotonic and reduced isotonic regression fitted in the sample from Munich. The 95% confidence bands correspond to the reduced regression.
The isotonic level sets were analyzed with the closed testing algorithm. The hypothesis H(1,10) for equal risk between the lowest exposure and the tenth cutpoint (exposure at 390 mg/m3year) was rejected with exact p-value < 0.001 and so did all nested hypotheses up to H(1,4). Hypothesis H(1,3) was retained (p-value = 0.33). This means that the first three level sets are lumped together, and the third isotonic cutpoint (7 (5–9.9) mg/m3year) defines the threshold. The logistic regression (equation (2)) estimated in a threshold of 7 (5–8) mg/m3 with a likelihood ratio statistics of 7.3 compared to the model with no threshold (p-value < 0.01).
In figure 5 we present the two dimensional isotonic regression, after the reducing procedure. Three major blocks are built for dust exposure and duration, and the first important step sets the threshold for dust at 4.5 (3.1, 7.2) mg/m3 and more than 18 years of exposure. This estimation is compatible with the threshold from the extended logistic model in equation (2) using time and ln(dust concentration) as covariates (3.8 [1.4, 4.6] mg/m3). However, the isotonic surfaces model gives more detailed information according to time subgroups. For less than 10 years exposure for example, no increase in the risk is observed.
Figure 5 Two-dimensional reduced isotonic regression modelling the dust concentration and the duration of exposure. The labels in the bars are the upper limit for the time and concentration intervals.
Discussion
In this paper we presented a method for estimating thresholds that does not rely on any parametric assumption. Isotonic regression is flexible and easy to apply, and detects thresholds with relative good power. It can be applied both with continuous exposure variables and categorical, where every dose level is actually a range of values. In both cases the same framework is followed, since isotonic regression groups the continuous predictor in constant risk level sets, as we exemplified in the application section. At this point some limitations of the methods should be discussed. Whereas we believe that the bootstrap confidence intervals offer good coverage for the true threshold, the point estimate may be biased. Some observations right after the threshold may have lower risk due to sampling error, misclassification or different individual response, and subsequently PAVA will pool them all together shifting the isotonic cutpoint towards higher exposures. This means, the threshold may be overestimated, as the simulation study revealed. Assigning a threshold to the whole level set rather than its upper cutpoint may be a reasonable compromise. When the exposure is in continuous form, the researcher has the advantage of more detailed investigation; the level set that appears to have the threshold can be re-analysed using non-isotonic models to reveal any particularities of the data. However the practice will be suggested by the working definition of the threshold i.e. whether a 'No Observed Adverse Effect Level' or a 'Lowest Observed Adverse Effect Level' is of interest.
Modifications of the proposed models are possible. Hothorn suggests a procedure based on odds ratios [17]. He argues that when one wants to detect an important increase in the risk, the use of confidence intervals is more accurate than comparing p-values. This idea can lead to a modification of backward elimination; instead of using p-values, the confidence intervals for the odd ratios could indicate a significant increase in the risk.
To our knowledge, there is no statistical model for threshold estimation that claims high power. Thus, it is important that more than one approach should be applied to confirm a TLV's location. External validation for the applied models can also provide useful information since there are cases where the data can be fitted by a variety of models (that consider a threshold or not) and all of them may fit well yielding however discrepant estimates or contradictory conclusions [1,18]. Modelling the data using smoothing splines or fractional polynomials would be useful in revealing the true shape of the relationship and avoid misinterpretations. In practice however the final decision about establishing a threshold is often taken on an ethical, political and economical basis. In threshold value estimation context statistical methods are tools that can eventually inform and direct the decision making process.
Conclusion
If the dose-response relationship is flat (slope less than 5%), the closed testing procedure fails to reject the constant risk assumption and thus has little power to detect the correct threshold. The method based on RIR is preferable as more powerful in most of the studied cases. When the increase in the risk is sufficiently high (at least 5%) both elimination approaches will detect the threshold successfully with RIR presenting the best results. In such situations, the threshold value estimation based on isotonic framework is conceptually simple and powerful. Given that no threshold value estimation method has been proven to have high power, isotonic regression provides at least a useful non-parametric alternative to the standard approaches and can corroborate or challenge their findings.
Appendix
The pooled adjacent violators algorithm
Consider a set x1,x2,...,xN of dose groups in increasing order, the observed outcome g(xi) for each dose group and the weights wi. To estimate g*(xi) the isotonic regression of g(xi), the pooled adjacent violators algorithm outlined below is the most popular approach. Note that the isotonic estimator is a maximum likelihood estimator under the monotonicity assumption. For simplicity assume only non-decreasing trend.
If g(xi) is in non-decreasing order then g*(xi) = g(xi).
Otherwise there is somewhere a violator such that g(xi) > g(xi+1) for some xi. Replace these two values by their weighted average
Av(g(xi),g(xi+1)) = (wig(xi) + wi+1g(xi+1))/(wi + wi+1).
Now the elements xi, xi+1 form a block called level set (LS) or solution block. If the new set of N-1 values is isotonic, then g*(xi) = g*(xi+1) = Av(g(xi),g(xi+1)) for the violator and g*(xi) = g*(xi+1) for all other observations.
If the set is not isotonic repeat the procedure using the new set of values.
The algorithm assuming decreasing trend is similar. Starting from the end of the shape and proceeding backwards (reversing the monotonicity) would give the same results.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
The first author performed the analysis and the simulation study while both authors contributed to the writing.
Pre-publication history
The pre-publication history for this paper can be accessed here:
==== Refs
Ulm K On the estimation of threshold values Biometrics 1989 45 1324 1326
Pastor R Guallar E Use of two-segmented logistic regression to estimate change-points in epidemiologic studies Am J Epidemiol 1998 148 631 642 9778169
Ulm K A statistical method for assessing a threshold in epidemiological studies Stat Med 1991 10 341 349 2028118
Cox C Threshold dose-response models in toxicology Biometrics 1987 43 511 523 3663815
Robertson T, Wright FT and Dykstra RL Order Restricted Statistical Inference 2005 New York, John Wiley and Sons Ltd
Salanti G Kurt U A nonparametric changepoint model for stratifying continuous variables under order restrictions and binary outcome Stat Methods Med Res 2003 12 351 367 12939101
Leuraud K Benichou J A comparison of several methods to test for the existence of a monotonic dose-response relationship in clinical and epidemiological studies Stat Med 2001 20 3335 3351 11746322 10.1002/sim.959
Mancuso JY Ahn H Chen JJ Order-restricted dose-related trend tests Stat Med 2001 20 2305 2318 11468765 10.1002/sim.849
Chuang-Stein G Agresti A A review of tests for detecting a monotone dose-response relationship with ordinal response data Stat Med 1997 16 2599 2618 9403959 10.1002/(SICI)1097-0258(19971130)16:22<2599::AID-SIM734>3.3.CO;2-0
Salanti G Ulm K Tests for Trend with Binary response Biomet Journal 2003 45 277 291 10.1002/bimj.200390012
Schell M Singh B The reduced isotonic regression JASA 1994 13 353
Rom DM Costello RJ Connell LT On closed test procedures for dose-response analysis Stat Med 1994 13 1583 1596 7973236
Ulm K Salanti G Estimation of the general threshold limit values for dust Int Arch Occup Environ Health 2003 76 233 240 12690498
Morton-Jones T Diggle P Parker L Dickinson HO Binks K Additive isotonic regression models in epidemiology Stat Med 2000 19 849 859 10734288 10.1002/(SICI)1097-0258(20000330)19:6<849::AID-SIM410>3.0.CO;2-D
Greim H Willey-VCH Occupational toxicants. Critical data evaluation for MAK values and classification of carcinogens. 1999 12 Weinheim, Commission for the investigation of the health hazards of chemical compounds in work area 10207125
Forschungsgemeinschaft D Willey-VCH List of MAK and BAT values 2001 2001 Weinheim
Hothorn L Trend tests in epidemiology: P-values or confidence intervals? Biometrical Journal 1999 41 817 825 10.1002/(SICI)1521-4036(199911)41:7<817::AID-BIMJ817>3.3.CO;2-3
Silvapulle M On the estimation of threshold values Biometrics 1991 45 1629
|
16274473
|
PMC1298303
|
CC BY
|
2021-01-04 16:32:51
|
no
|
BMC Med Res Methodol. 2005 Nov 7; 5:36
|
utf-8
|
BMC Med Res Methodol
| 2,005 |
10.1186/1471-2288-5-36
|
oa_comm
|
==== Front
BMC Musculoskelet DisordBMC Musculoskeletal Disorders1471-2474BioMed Central London 1471-2474-6-531626907510.1186/1471-2474-6-53Research ArticleThe use and diagnostic yield of radiology in subjects with longstanding musculoskeletal pain – an eight year follow up Lindgren Hans [email protected] Stefan [email protected] Helsingborg County Hospital, Helsingborg, Sweden2 Spenshult Hospital for Rheumatic Diseases, Halmstad, Sweden2005 3 11 2005 6 53 53 19 5 2005 3 11 2005 Copyright © 2005 Lindgren and Bergman; licensee BioMed Central Ltd.2005Lindgren and Bergman; 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
Longstanding musculoskeletal pain is common in the general population and associated with frequent use of health care. Plain radiography is a common diagnostic approach in these patients despite knowledge that the use in the investigation of musculoskeletal pain is associated with low diagnostic yield, substantial costs and high radiation exposure. The aim of this study was to assess the use of diagnostic imaging and the proportion of pathological findings with regard to duration and distribution of pain in a cohort from the general population.
Methods
An eight-year longitudinal study based on questionnaires at three occasions and medical records on radiological examinations done in medical care. Thirty subjects were selected from an established population based cohort of 2425 subjects that in 1995 answered a postal survey on pain experience. At baseline there were ten subjects from each of three pain groups; No chronic pain, Chronic regional pain, and Chronic widespread pain (CWP). Those who presented with CWP at two or all three occasions were considered to have a longstanding or re-occurring CWP. In total the thirty subjects underwent 102 radiological examinations during the eight year follow up.
Results
There was a non-significant (p = 0.10) finding indicating that subjects with chronic pain at baseline (regional or widespread) were examined three times more often than those with no chronic pain. When the indication for the examination was pain, there was a low proportion of positive findings in subjects with longstanding CWP, compared to all others (5.3% vs 28.9%; p = 0.045). On the other hand, in examinations on other indications than pain the proportion of positive findings was high in the CWP group (62.5% vs 14.8%; p = 0.001).
Conclusion
Radiological examinations had a low diagnostic yield in evaluation of pain in subjects with longstanding/reoccurring CWP. These subjects had on the other hand more often positive findings when examined on other indications than pain. This may indicate that subjects with longstanding/reoccurring CWP are more prone to other diseases. It is a challenge for caregivers, often primary care physicians, to use radiological examinations to the best for their patients.
==== Body
Background
Several studies from Western Europe and the USA have shown that chronic (most often defined as a duration of three months or more) musculoskeletal pain is common in the general population, with a prevalence (point or one year period) of 30–50% [1-3]. With regard to distribution of musculoskeletal pain in the body, a distinction is often made in epidemiological studies between chronic regional pain (CRP) and chronic widespread pain (CWP) [1,3,4]. In postal surveys approximately 11% of subjects in a general population report CWP [1-3]. Chronic musculoskeletal pain, and foremost CWP, has been shown to have other background factors than acute and more localized pain [5]. CWP is to a higher degree associated to adverse psychosocial and sociodemographic factors. Besides being common in the population, musculoskeletal pain is also associated with frequent use of health care. That is especially true for subjects with low-back and widespread pain [6].
Plain radiography is a common diagnostic approach in these patients despite knowledge that the use in the investigation of musculoskeletal pain could be associated with low diagnostic yield, substantial costs and high radiation exposure [7,8]. This is well studied in plain radiography for low back pain, were studies have shown that 25–50% may be unnecessary according to clinical recommendations [9,10]. There are internationally recognised guidelines concerning the use of radiography in the management of patients with low-back pain [11,12]. These guidelines have not substantially changed the pattern of doctors' referrals [13]. Patient expectations can also be an explanation. It has for example been reported that the majority (72%) of patients referred to radiological examination because of low back pain rate this as very important [14].
The use of radiography of the lumbar spine in primary care patients with low back pain has not only implications on diagnostic yield and costs, but has also been reported to be associated with a worse patient outcome and an increased doctors workload [15]. Furthermore the radiation exposure could be high, especially in radiography of the lumbar spine, where the gonadal dose for one investigation in women is equivalent to a daily chest radiograph over six years [16].
The aim of this study was to assess if subjects with chronic musculoskeletal pain (CRP and CWP) were submitted to more radiological investigations than those with no chronic pain. A second aim was to study the proportion of pathological findings with regard to duration and distribution (no chronic pain, CRP, and CWP), and the indication for referral to the examination (pain or other causes).
Methods
A sample of 30 subjects was derived from a cohort of 2755 subjects that in 1995 responded to a postal survey on experience of pain. This survey included initially a representative sample of 3928 subjects aged 20–74 years, living in two municipalities on the west coast of Sweden. According to pain history and pain distribution on a pain drawing the respondents were categorised into three groups: No chronic pain (NCP), Chronic regional pain (CRP), and Chronic widespread pain (CWP). Chronic pain was defined as pain more than three months during the last year. Widespread pain was defined according to ACR 1990 criteria for fibromyalgia. Details can be found in previous work [1]. Ten subjects from each of the three groups (NCP, CRP and CWP) were randomly selected and included in this study.
From the selection of 30 subjects, 28 had also answered follow up postal surveys in 1998 and 2003. These 28 subjects were for sub-analyses further divided into two groups; (1) those belonging to the CWP-group in at least two of the postal surveys (n = 5), and (2) all others (n = 23).
The radiology records between 1994 and 2002 were scrutinised and all radiographs were double-checked by the authors. The results were in total agreement with the original readers of the radiology examinations. A total of 102 examinations were included in the study. It was also noted if the referral was due to "pain" or "other cause", such as suspected infection or malignancy, and if the examination resulted in a positive finding or not.
Non-parametrical statistical tests in the statistical package SPSS 12.0 were used for the analyses. Mann-Whitney U-test was used when comparing number of radiological examinations for each individual in the different pain-groups. Chi-squared test was used for group comparisons. The study was approved by the Ethics Research Committee, Faculty of Medicine, University of Lund, Sweden.
Results
Out of the total sample of 30 subjects, there were 14 who had been examined by radiology, and together they had been submitted to 102 examinations. In 61% of the cases the referral was from a general practitioner and in 88% the examination was done in outpatient care. A majority of the examinations (66%) were scheduled and not acute. An overview of the examinations and their outcome show positive findings in 25.5% of the cases (Table 1).
Table 1 Radiological examinations and positive findings. The number of different radiological examinations from 1994 to 2002, and the number of positive findings.
Examination Number of examinations Positive findings (%)
Peripheral skeleton 30 5 (16.7)
Axial skeleton 18 5 (27.8)
Pulmonary 24 10 (41.7)
Mammography 9 0 (0)
CT/MR 7 2 (28.6)
Other 14 4 (28.6)
Total 102 26 (25.5)
The results from the examinations were further analysed with regard to pain group (NCP, CRP or CWP) belonging in 1995, and information regarding the indication for referral (pain or other cause). The mean age varied from 36 in the NCP-group, over 47 in the CRP-group, to 53 in the CWP-group. There was two-times more women than men in the CWP-group, but no such difference in the NCP- or CRP-groups. There was a non-significant (p = 0.10) finding indicating that subjects with chronic pain (regional or widespread) were examined more often than those with no chronic pain (Table 2). The outcome of the radiological examinations (positive or negative findings) varied nearly significantly (p = 0.08) with the indication for the examination ("pain" or "other cause"), with overall less positive findings in examinations done in the pain group. Most prominent result is the high proportion of positive findings (62.5%; p = 0.013) in the CWP-group examined on other indication than pain (Table 3).
Table 2 Number of examinations with respect to pain distribution. The number of radiological examinations in each subject from 1994 to 2002, for each of the three paingroups; chronic widespread pain (CWP), chronic regional pain (CRP, and those with no chronic pain (NCP).
Pain group Number of subjects Number of examinations Exam's/subject
CWP 10 48 4.8
CRP 10 40 4.0
NCP 10 14 1.4
Total 30 102 3.4
Table 3 Positive findings and the indication for radiological examination. The number of positive findings on radiological examination with regard to the indication for the examination (pain or other cause).
Examinations due to pain Examinations due to other causes
Pain group Number of examinations Number of positive findings (%) Number of examinations Number of positive findings (%)
CWP 32 8 (25.0) 16 10 (62.5)
CRP 22 4 (18.2) 18 3 (16.7)
NCP 5 0 (0.0) 9 1 (11.1)
Total 59 12 (20,3) 43 14 (32,6)
CWP = chronic widespread pain, CRP = chronic regional pain, NCP = no chronic pain
The radiological examinations were done during an eight-year period. It was decided to also take changes in pain group belonging under the time period into account in the further analyses. The results from the analysis above concerning the outcome in those with CWP at baseline motivated that subjects with longstanding/reoccurring CWP (two or more occasions during the eight year period) were compared to all others. The two groups did not differ in mean age, but the group with longstanding/reoccurring CWP had an over-representation of women. There was a non-significant finding (p = 0.24) that those with longstanding/reoccurring CWP were examined more than twice as often than all others (Table 4). If the indication for the examination was pain, there was a low proportion of positive findings in those with longstanding/reoccurring CWP compared to all others (5.3% vs. 28.9%; p = 0.045; Table 5). On the other hand, if the indication for the examination was other than pain, the proportion of positive findings in subjects with longstanding/reoccurring CWP was as high as 62,5%, compared to 14.8% amongst all others (p = 0.001; Table 5).
Table 4 Number of examinations and longstanding/reoccurring widespread pain. The number of radiological examinations in each subject from 1994 to 2002 for those with longstanding/reoccurring (reported in at least two of the three surveys) chronic widespread pain (CWP) compared to all others.
Pain group Number of subjects Number of examinations Exam/subjects
Two times CWP 5 35 7.0
All others 23 65 2.8
Total 28 100 3.6
Table 5 Positive findings and the indication for radiological examination. The number of positive findings on radiological examination with regard to the indication for the examination (pain or other cause), for those with longstanding/reoccurring (reported in at least two of the three surveys) chronic widespread pain (CWP) compared to all others.
Examinations due to pain Examinations due to other causes
Pain group Number of examinations Number of positive findings (%) Number of examinations Number of positive findings (%)
Two times CWP 19 1 (5.3) 16 10 (62.5)
All others 38 11 (28.9) 27 4 (14.8)
Total 57 12 (21.1) 43 14 (32.6)
Discussion
A main finding in this study was that subjects with chronic regional or widespread pain were more frequently submitted to radiological examinations than those with no chronic pain. The diagnostic yield was low if the indication for the examination was pain. On the other hand there was a high proportion of positive findings if the indication for the examination was other than pain. This was especially true for subjects with longstanding/reoccurring CWP during the eight-year follow up period.
During scrutinising the radiographs and when reviewing the literature it was found questionable how to define a significant positive finding with regard to degenerative changes in the spine. In the analyses it was decided to count five cases of degenerative spinal changes as negative results. Such findings are very common in lumbar spine radiology. The prevalence of degenerative changes in the lumbar spine rises with age with an occurrence of as much as 71% in the age group 65–74 years. Their relation to pain has been considered speculative, and the therapeutic consequences of these findings are minor [13].
We found no cases with disc herniation, by many clinicians considered a major positive finding, though this has been reported in as much as 20–70 % in pain free populations [17-19]. This was not surprising since plain radiography is considered relatively insensitive for disc herniation or more serious spinal conditions [20]. Magnetic resonance imaging (MRI) has been proposed as an alternative investigation for patients with low back pain [21]. MRI has however not been found to give a better outcome to primary care patients with low back pain and may increase the cost of care [22]. It has been stated that the majority of patients with low back pain should be assessed clinically and that radiological imaging seldom is required. [23].
Subjects with longstanding/reoccurring CWP have often a multifactorial background to their pain problem. The patophysiology in the musculoskeletal system could be of minor importance when pain is maintained in an interaction between neurophysiological processes and psychosocial factors [5]. It is thus understandable that the diagnostic yield is low when radiological examinations are done to find the cause of pain in the musculoskeletal structures.
There were more positive findings in the CWP group when radiological examinations were done on other indications than pain. This can be an expression of a higher incidence of other diseases amongst subjects with CWP. One important clinical implication of this is, that if an individual with chronic widespread pain seeks his/her doctor for a reason other than pain – there is good reason for alert and to go further on with diagnostic tests if necessary.
It has been reported that inappropriately referred patients tended to rate their radiography referrals more important than appropriately referred patients [14]. The patient's view could thus be a barrier to a more justified use of radiography. This is a challenge to the caregivers but a more appropriate use of radiography could be supported by guidelines [15].
Some argue that the usage of radiographs is justified to rule out serious disease and to reassure the patient. However in low back pain the prevalence of possible serious conditions (fracture, infection or tumour) is very low, which implies radiation exposure in many patients with no significant lesion [13]. The radiation dose from lumbar radiographs in a given patient is 40 times the dose received from chest radiography, with gonadal doses for women equivalent to daily chest radiograph over six years [16]. The risk from any investigation must be justified by being less harmful than the potential risk when neglecting to do it. In radiological investigations, due to the risk of radiation-induced malignancy, the radiation exposures must be justified and optimised [24].
The strength of the present study lies in that it is based on a cohort from the general population living in a well defined area, that have been followed over an eight year period. Pain experience have been recorded on three occasions during this eight year period, and there is only one hospital performing the radiological examinations in the area. The major drawback is the relatively small material that also gives power problems in some of the analyses. The age distribution between groups and the higher proportion of women with CWP is expected from previous studies [1], but could introduce a bias in the first analysis (Tables 2 and 3) with a presumed higher referral rate with age. The results were however consistent in the second analysis (Tables 4 and 5), where there was no difference in age between the two studied groups. The external validity is strengthened by the design with a sample from the general population but weakened by the small size of the study. The results must thus be interpreted by this in mind.
Conclusion
Radiological examinations had a low diagnostic yield in evaluation of pain in subjects with longstanding/reoccurring CWP. These subjects had on the other hand often positive findings when examined on other indications than pain. This may indicate that subjects with longstanding/reoccurring CWP are more prone to other diseases. It is a challenge for caregivers, often primary care physicians, to use radiological examinations to the best for their patients.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
SB was responsible for the Epipain project that established the pain cohorts. Both authors were equally involved in all aspects of the present study. Both authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
Professor Lennart Jacobsson, associate professors Ingemar Petersson and Per Herrström for their work with the Epipain project.
==== Refs
Bergman S Herrstrom P Hogstrom K Petersson IF Svensson B Jacobsson LT Chronic musculoskeletal pain, prevalence rates, and sociodemographic associations in a Swedish population study J Rheumatol 2001 28 1369 77 11409133
Andersson HI Ejlertsson G Leden I Rosenberg C Chronic pain in a geographically defined general population: studies of differences in age, gender, social class, and pain localization Clin J Pain 1993 9 174 82 8219517
Croft P Rigby AS Boswell R Schollum J Silman A The prevalence of chronic widespread pain in the general population J Rheumatol 1993 20 710 3 8496870
Wolfe F Smythe HA Yunus MB Bennett RM Bombardier C Goldenberg DL Tugwell P Campbell SM Abeles M Clark P Fam AG Farber SJ Fiechtner JJ Franklin CM Gatter RA Hamaty D Lessard J Lichtbroun AS Masi AT McCain GA Reynolds WJ Romano TJ Russel IJ Sheon RP The American College of Rheumatology 1990 Criteria for the Classification of Fibromyalgia. Report of the Multicenter Criteria Committee Arthritis Rheum 1990 33 160 72 2306288
Bergman S Herrstrom P Jacobsson LT Petersson IF Chronic widespread pain: a three year followup of pain distribution and risk factors J Rheumatol 2002 29 818 25 11950027
Andersson HI Ejlertsson G Leden I Schersten B Musculoskeletal chronic pain in general practice. Studies of health care utilisation in comparison with pain prevalence Scand J Prim Health Care 1999 17 87 92 10439491 10.1080/028134399750002700
Deyo RA Early diagnostic evaluation of low back pain J Gen Intern Med 1986 1 328 38 2945917
Waddell G Low back pain: a twentieth century health care enigma Spine 1996 21 2820 5 9112705 10.1097/00007632-199612150-00002
Halpin SF Yeoman L Dundas DD Radiographic examination of the lumbar spine in a community hospital: an audit of current practice Bmj 1991 303 813 5 1932970
Espeland A Albrektsen G Larsen JL Plain radiography of the lumbosacral spine. An audit of referrals from general practitioners Acta Radiol 1999 40 52 9 9973903
Gutierrez B Bloschichak A Kurlantzick V Clinical guidelines for lumbar radiographs for patients with low back pain Jama 1997 278 1741 2 9388145 10.1001/jama.278.21.1741
Burton AK Waddell G Clinical guidelines in the management of low back pain Baillieres Clin Rheumatol 1998 12 17 35 9668955
van den Bosch MA Hollingworth W Kinmonth AL Dixon AK Evidence against the use of lumbar spine radiography for low back pain Clin Radiol 2004 59 69 76 14697378 10.1016/j.crad.2003.08.012
Espeland A Baerheim A Albrektsen G Korsbrekke K Larsen JL Patients' views on importance and usefulness of plain radiography for low back pain Spine 2001 26 1356 63 11426152 10.1097/00007632-200106150-00020
Kendrick D Fielding K Bentley E Kerslake R Miller P Pringle M Radiography of the lumbar spine in primary care patients with low back pain: randomised controlled trial Bmj 2001 322 400 5 11179160 10.1136/bmj.322.7283.400
Hall FM Overutilization of radiological examinations Radiology 1976 120 443 8 778913
Boos N Rieder R Schade V Spratt KF Semmer N Aebi M 1995 Volvo Award in clinical sciences. The diagnostic accuracy of magnetic resonance imaging, work perception, and psychosocial factors in identifying symptomatic disc herniations Spine 1995 20 2613 25 8747239
Boden SD McCowin PR Davis DO Dina TS Mark AS Wiesel S Abnormal magnetic-resonance scans of the cervical spine in asymptomatic subjects. A prospective investigation J Bone Joint Surg Am 1990 72 1178 84 2398088
Jensen MC Brant-Zawadzki MN Obuchowski N Modic MT Malkasian D Ross JS Magnetic resonance imaging of the lumbar spine in people without back pain N Engl J Med 1994 331 69 73 8208267 10.1056/NEJM199407143310201
van Tulder MW Assendelft WJ Koes BW Bouter LM Spinal radiographic findings and nonspecific low back pain. A systematic review of observational studies Spine 1997 22 427 34 9055372 10.1097/00007632-199702150-00015
Reinus WR Strome G Zwemer FL Jr Use of lumbosacral spine radiographs in a level II emergency department AJR Am J Roentgenol 1998 170 443 7 9456961
Jarvik JG Hollingworth W Martin B Emerson SS Gray DT Overman S Robinson D Staiger T Wessbecher F Sullivan SD Kreuter W Deyo RA Rapid magnetic resonance imaging vs radiographs for patients with low back pain: a randomized controlled trial Jama 2003 289 2810 8 12783911 10.1001/jama.289.21.2810
McNally EG Wilson DJ Ostlere SJ Limited magnetic resonance imaging in low back pain instead of plain radiographs: experience with first 1000 cases Clin Radiol 2001 56 922 5 11603896 10.1053/crad.2001.0718
ICPR 1990 Recommendations of the International Commission on Radiological Protection (ICPR 60) Annals of the ICPR 1991 21 42 43
|
16269075
|
PMC1298304
|
CC BY
|
2021-01-04 16:32:05
|
no
|
BMC Musculoskelet Disord. 2005 Nov 3; 6:53
|
utf-8
|
BMC Musculoskelet Disord
| 2,005 |
10.1186/1471-2474-6-53
|
oa_comm
|
==== Front
BMC Musculoskelet DisordBMC Musculoskeletal Disorders1471-2474BioMed Central London 1471-2474-6-571628008910.1186/1471-2474-6-57Study ProtocolManipulative therapy and/or NSAIDs for acute low back pain: design of a randomized controlled trial [ACTRN012605000036617] Hancock Mark J [email protected] Christopher G [email protected] Jane [email protected] Andrew J [email protected] Chris W [email protected] Richard O [email protected] Megan F [email protected] James H [email protected] Back Pain Research Group, University of Sydney, PO Box 170, Lidcombe, NSW, 1825, Australia2 Faculty of Pharmacy, University of Sydney, NSW, 2006, Australia3 Discipline of General Practice, Balmain Hospital, 37A Booth St, Balmain, 2041, NSW, Australia4 Clinical Pharmacology, UNSW & St Vincent's Hospital, Darlinghurst 2010, NSW, Australia2005 10 11 2005 6 57 57 22 9 2005 10 11 2005 Copyright © 2005 Hancock et al; licensee BioMed Central Ltd.2005Hancock 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
Acute low back pain is a common condition resulting in pain and disability. Current national and international guidelines advocate general practitioner care including advice and paracetamol (4 g daily in otherwise well adults) as the first line of care for people with acute low back pain. Non-steroidal anti-inflammatory drugs (NSAIDs) and spinal manipulative therapy (SMT) are advocated in many guidelines as second line management options for patients with acute low back pain who are not recovering. No studies have explored the role of NSAIDs and/or SMT in addition to first line management for acute low back pain. The primary aim of this study is to investigate if NSAIDs and/or SMT in addition to general practitioner advice and paracetamol results in shorter recovery times for patients with acute low back pain. The secondary aims of the study are to evaluate whether the addition of SMT and/or NSAIDs influences pain, disability and global perceived effect at 1, 2, 4 and 12 weeks after onset of therapy for patients with significant acute low back pain.
Methods/design
This paper presents the rationale and design of a randomised controlled trial examining the addition of NSAIDs and/or SMT in 240 people who present to their general practitioner with significant acute low back pain.
==== Body
Background
Low back pain is a common condition with lifetime prevalence rates reported between 59 and 84% [1]. On any given day 12–33% of people report some back pain [1]. A recent review on the health of Australians placed back problems as one of the three health conditions responsible for the greatest health system expenditure [2]. Not only is back pain common and costly to the health system but an episode of acute low back pain can be seriously disabling and distressing for the patient. In many cases the patient cannot undertake, or is severely restricted in, normal work and home duties. While it is commonly believed that the majority of people with acute low back pain recover spontaneously within 4–6 weeks a recent systematic review of the prognosis of acute low back pain [3] did not find evidence of this. On average, people with acute back pain experience substantial improvements in the first month: pain and disability are reduced by 58% of initial values and 82% of patients have returned to work. Further but smaller improvements occur up to three months, after which pain and disability levels remain nearly constant.
A recent review of international guidelines [4] for the treatment of acute non-specific low back pain showed clear agreement on the appropriate first line of care. Each guideline recommended that the first line of care should be provision of advice and simple analgesic medicines. Advice includes reassuring the patient of a favourable prognosis, encouraging the patient to stay active and discouraging bed rest. It is recommended that medication should be taken in a time contingent manner and paracetamol (4 g daily) is suggested as first line choice due to its low risk of gastrointestinal side effects in otherwise well adults.
Each of the guidelines also recommended that if first line management options provide insufficient pain relief then additional therapies could be considered. However, there is no clear agreement on what the second line management options should include or at what time it/they should begin. All the guidelines endorsed the appropriate use of NSAIDs but only some guidelines (8 of 11) endorsed spinal manipulative therapy (SMT).
Research to date has primarily addressed the issue of whether SMT and/or NSAIDs are more efficacious than placebo, however, this question does not reflect contemporary evidence-based practice in the management of low back pain. For the practitioner following the available guidelines the salient question is whether either of these two treatments are effective when delivered in addition to first line care universally endorsed in the low back pain treatment guidelines.
The recent Australian National Health and Medical Research Council (NHMRC) acute low back pain guidelines 2003 [5] conclude that there is conflicting evidence in this patient group that spinal manipulation provides greater short term pain relief when compared to placebo. Subsequent to the completion of these guidelines, three systematic reviews evaluating the role of SMT in back pain management have been published [6-8]. Each review concludes that spinal manipulation provides a greater reduction in pain and disability than inactive or ineffective therapies, with all trials providing similar estimates for the size and precision of the treatment effect (approximately 18 points of pain relief measured on a 0–100 point scale;95%CI 13–24) [6].
The issue of whether spinal manipulation is more effective than placebo in patients who receive general practitioner (GP) advice and paracetamol as first-line care was not directly investigated in these systematic reviews. However, in the study by Curtis et al [9] all patients received guideline-based care (advice and paracetamol) with half randomised to also receive a course of spinal manipulative therapy. Furthermore, Curtis et al [9] reported that a greater proportion of the manipulative therapy group completely recovered after the first visit compared with the control group: 14% versus 6% (P = 0.01). Patients who received more intense manipulative therapy (four or more treatments) had more rapid return to functional recovery (7.8 days) compared with those who received less treatments (11.1 days; P = 0.02). However this study did not include an inactive manipulation intervention and thus it is possible that the results are influenced by intervention bias as a result of the more frequent patient-practitioner contact in the group who received spinal manipulation.
The Cochrane Review on the role of NSAIDs in the management of low back pain [10] located nine studies comparing NSAIDs to placebo and concluded that NSAIDs were effective for short-term symptomatic relief in patients with acute low back pain. Furthermore, the review concluded that there was strong evidence that various types of NSAIDs are equally effective [10]. Subsequent to this review one study has shown that aceclofenac provides similar results to diclofenac [11] and another study has suggested temporal differences in analgesia produced by diclofenac pain relief when compared to ibuprofen or placebo [12]. However, no studies have examined whether the addition of NSAIDs to standard first-line care produced greater benefit than standard first-line care alone or whether there is an additive effect of manipulative therapy and NSAIDs.
In summary, there is high quality level I evidence that both SMT and NSAIDs are more efficacious than placebo treatments in patients with acute low back pain who receive no additional care. This evidence is of limited use to GPs managing patients with acute low back pain according to the widely accepted first line care. The most important clinical question is whether NSAIDs and/or spinal manipulation are effective when delivered in addition to the first line of care. At present there is no high quality evidence for this. In 2003, the NHMRC clinical practice guidelines for the evidence-based management of acute low back pain identified the need for research into the use of NSAIDs and spinal manipulative therapy and recommended testing these interventions in well-designed randomised controlled trials (RCTs) with 'advice to avoid bed rest and maintain usual activities' as the appropriate comparator [5].
The primary aim of this study is to investigate whether the addition of SMT and/or NSAIDs to GP advice and paracetamol results in shorter recovery times for patients with significant acute low back pain. The secondary aims of this study are to evaluate whether the addition of SMT and/or NSAIDs to GP advice and paracetamol influences pain, disability and global perceived effect at 1, 2, 4 and 12 weeks for patients with significant acute low back pain.
Methods
This randomised controlled trial will be conducted at approximately 20 general practice clinics and 10 private physiotherapy clinics within Sydney, Australia. Ethics approval has been gained from the University of Sydney Human Research Ethics Committee.
Study population
Two hundred and forty participants with a new episode of significant acute non specific low back pain who present to GPs will be recruited. GPs will screen potential participants to ensure they satisfy the inclusion and exclusion criteria and will provide participants with an information sheet. The GP will then contact one of the researchers by phone and pass on the patient's contact details. The researchers will then organise an appointment within two days (excluding Sundays) to meet with the participant, formally enrol them in the trial if eligibility is confirmed at which time they will be randomised to a treatment arm of the study.
Inclusion criteria
To be eligible for the trial participants must meet all of the following criteria as assessed by the GP:
• Primary complaint of pain extending in an area between the 12th rib and buttock crease. This may or may not be accompanied by leg pain.
• New episode of low back pain. This is defined as an episode which was preceded by a period of at least one month without low back pain where the participant was not consulting a health care practitioner or continuing with medication for their low back pain [13].
• Pain of less than six weeks duration.
• Low back pain severe enough to cause moderate pain and moderate interference with normal work including work outside the home and housework (as measured by adaptations of items 7 and 8 of the SF-36).
• No known or suspected serious spinal pathology (metastatic, inflammatory or infective diseases of the spine, cauda equina syndrome, spinal fracture).
• No nerve root compromise evidenced by at least two of the following (i) myotomal weakness, (ii) dermatomal or widespread sensory loss, (iii) hypo or hyper-reflexia of the lower limb reflexes.
• Not currently taking NSAIDs.
• Not currently receiving SMT.
• No spinal surgery within the preceding six months.
• No history of peptic ulcer.
• No allergy to aspirin.
• Not currently receiving anticoagulant therapy.
• No serious co-morbidities preventing prescription of NSAIDs or paracetamol eg: cardiac, liver or renal failure.
• No contraindications to SMT or NSAIDs
Enrolment and baseline measures
At the first meeting with the researcher, baseline data will be collected from the participant. This will include contact details, personal details, outcome measures, and variables that will be assessed as predictors of response to treatment. Contact details for both the participant and a friend or relative who does not live with them will be collected to optimise follow up rates. Personal details recorded will include age, length of time symptoms have been present and the number of previous episodes of low back pain. The following baseline measures of outcome will be recorded:
• Numerical pain rating scale on a 0–10 scale [14];
• A back specific disability scale (Roland Morris Disability Scale) [15];
• A patient specific measure of disability (Patient Specific Functional Scale)[16];
Treatment allocation
Immediately after completing baseline measures participants will be allocated into treatment groups. Prior to the start of the study a researcher not involved in data collection or analysis will develop a randomisation schedule and produce consecutively numbered sealed opaque envelopes containing each participant's allocation. Randomisation will be performed using randomly permuted blocks of 4, 8 and 12. The researcher will select the next numerical randomisation envelope and open it. This envelope will contain a plastic bottle with the active or placebo NSAIDs which will be given to the participant along with a consumer medicine information sheet for diclofenac. The participant's name will be fixed to the medication. Placebo tablets with an identical shape and colour (including dose form excipients and coating) are dispensed making it impossible for the researchers or participant to differentiate between the active and placebo NSAIDs. The randomisation envelope will also contain a second smaller envelope containing the participant's allocation into active or placebo SMT. The researcher will give this envelope to the treating physiotherapist to open after the researcher has left. The researcher will therefore remain blinded to allocation for the NSAIDs and SMT arms. Treating physiotherapists can obviously not remain blinded to SMT allocation however they will be trained to respond identically to all patients regardless of treatment group except for the treatment provided.
Patients will be allocated to one of four treatment groups as follows:
• Control group (placebo NSAIDs and placebo SMT)
• NSAIDs group (active NSAIDs and placebo SMT)
• SMT group (placebo NSAIDs and active SMT)
• SMT and NSAIDs group (active NSAIDs and active SMT)
Treatments
All participants in the study will receive standard care from their GP before baseline and allocation into a treatment group. Standard care in this study will involve advice and paracetamol (1 g four times daily), this being the first line of care advocated in both national and international clinical practice guidelines. Advice will include reassurance of a favourable prognosis and encouragement to avoid bed rest and stay active. Paracetamol will be prescribed at 4 gm per day (Two 500 mg tablets every 6 h). Paracetamol is to be continued at this dosage for a maximum of four weeks. If subjects recover before four weeks (zero or one out of ten pain for seven consecutive days) paracetamol will be stopped. All participants will also receive two follow up visits with their GP, one week and two weeks after their initial visit. At follow up visits GPs will reinforce the initial advice and ensure participants have no adverse reactions to the treatments. GPs will remain blinded to group allocation and instructed not to ask about the physiotherapy management.
Participants allocated to receive active or placebo NSAIDs (diclofenac) will be instructed to take them according to an identical schedule. Dosage will be 50 mg bd taken with food for a maximum of four weeks or until the participant has recovered (zero or one out of ten pain for seven consecutive days).
Participants allocated to receive SMT will receive treatment two or three times per week (at the therapist's discretion) for a maximum of 12 treatments over four weeks. If the subject recovers (zero or one out of ten pain for seven consecutive days) before four weeks the SMT will be stopped. Patients will receive spinal manipulative therapy according to a treatment algorithm developed by the researchers based on the views of expert clinicians and researchers in the field (Additional file 1) [17-20]. The algorithm permits and excludes certain physiotherapy treatments. Consistent with contemporary best clinical practice, the physiotherapist will adjust the treatment to the clinical presentation of the patient rather than apply the same treatment to all patients (as per normal clinical practice). The algorithm however is sufficiently prescriptive to allow replication and accurate description of the trial treatment. All participants will be examined by the physiotherapist who will take a standard history and perform a physical examination that will include assessment of active range of lumbar spinal motion, routine tension tests (straight leg raise, passive neck flexion and prone knee bend), neurological examination (reflexes, muscle strength, sensation) when indicated and the application of manually applied postero-anteriorly directed forces to all levels of the lumbar spine. Based on findings from the examination the physiotherapist will initiate what they consider to be an optimal program of SMT within the guidelines of the study (Additional file 1).
SMT will be delivered by physiotherapists who have postgraduate training in manipulative therapy and who regularly use manipulative therapy in their clinical practice. Participating physiotherapists will be supplied with documentation defining their role in the trial and will undergo one hour of training from a researcher who is a physiotherapist. These physiotherapists will have as minimum training a Graduate Diploma in Manipulative Physiotherapy and two years clinical experience using manipulative therapy techniques.
Therapists will be asked to keep a record of the number of times the participant attended for SMT and details of the treatment including the techniques and dosage. These will be used to ensure compliance with the protocol and to help describe the treatment given in this study.
Placebo SMT will be prescribed using exactly the same schedule as active SMT. The placebo therapy used will be detuned ultrasound (US). The detuned US will be performed in a manner that mimics real US around the area involved for 5–12 minutes. Treatment sessions will last for the same time as for active SMT (30–40 minutes for the initial assessment and treatment and approximately 20 min for follow up sessions). Follow ups will include reassessing the participant's history and physical examination findings however no palpation of the spinal joints will be performed after the initial assessment. A re-evaluation of pain or range of motion will be done after the placebo treatment. In this way we will achieve close matching of the active and inactive interventions in terms of treatment duration and patient/therapist contact.
Patients will be asked not to seek other treatments for their low back pain during the treatment period. In cases where this is unavoidable a record of additional treatments will be kept (and these patients may be excluded from the per protocol but not intention to treat analysis depending on the nature of the treatment). Several mechanisms will be used to ensure that the trial protocol is consistently applied. Protocol manuals will be developed and all involved researchers (general practitioners, manipulative physiotherapists, trial manager and outcome assessors) will be trained to ensure that screening, assessment, random allocation and treatment procedures are conducted according to protocol. An independent researcher will monitor adherence to assessment, randomisation and treatment procedures in a random group of participants.
After four weeks all interventions will cease and participants will be asked not to seek other treatment before the three month follow-up if possible. Participants who do have further treatment will be asked to record the type and amount received.
Outcome measures
Outcome measures will be recorded by an assessor blinded to group allocation. Outcome measures will be collected using both a daily pain diary completed by the participants and weekly phone follow-ups for the first four weeks and at three months.
The primary outcome is the number of days to recovery, with recovery defined in two ways. Firstly recovery is defined as a pain score of 0 or 1 on a 0–10 pain scale (numerical pain rating scale) that is maintained for seven consecutive days. Secondly recovery is defined as the first day that the patient has a pain score of 0–1 on a 0–10 pain scale. To ensure a precise estimate of the time to recovery, subjects will complete a daily pain diary (completed either first thing in the morning or last thing at night determined by the participant) until recovery. To minimize potential for lost data pain scores from the diaries will be read to the researcher at each of the phone follow ups. The researcher will then transcribe these into a second participant record. Diaries will be kept until the patient has scored 0 or 1 out of 10 for seven consecutive days or for a maximum of three months.
The secondary outcomes are pain (numerical pain rating scale) [21], disability (Patient Specific Functional Scale[16] and Roland Morris Disability Questionnaire [15]), global perceived effect and satisfaction/beliefs about treatment [22]. Secondary outcomes will be recorded at baseline, 1 week, 2 weeks, 4 weeks and 3 months.
Compliance with physiotherapy will be recorded by the treating physiotherapists. Compliance with the medications will be assessed by collecting unused medications at the end of the treatment period (4 weeks).
Data analysis
Data will be analysed by a statistician who is blinded to group status. The primary analyses will be by intention-to-treat and we will restrict the number of analyses in order to reduce the possibility of Type I errors. For primary outcomes, a p value of <0.05 will be considered statistically significant. For the secondary outcomes a p value of <0.01 will be considered significant.
For the primary outcome of days to reach recovery we will use survival curves with a log-rank statistic to assess differences between groups [23] and, if required, Cox's regression to assess the effects of treatment (group) status on hazard rates for time to recovery. In the primary analysis, number needed to treat (NNT) and the 95% confidence intervals to reach recovery in three months will also be calculated.
For secondary outcomes we will use a mixed model with group as a fixed factor. In these analyses, if there is a significant difference in secondary outcomes between treatment groups, we will conduct post-hoc analyses to inspect differences in secondary outcome variables at 1, 2, and 4 weeks and 3 months. We will also test for any additive or multiplicative effects between treatments on the outcome variables
Sample size
Sample size was calculated using equations for survival data. Two hundred and forty participants was determined to provide 80% power to detect a 20% difference in recovery rates between the control and intervention groups with an alpha level of 0.05. These calculations were based on a 50% recovery rate in the control group by three months. These numbers are probably conservative and based on results from our recent prognostic study of over 1000 subjects with acute low back pain. Higher rates of recovery will increase the statistical power. We allowed for 10% loss to follow up.
Conclusion
We have presented the rationale and design for an RCT examining the effects of SMT and/or NSAIDs on patients with significant acute low back pain. The primary outcome will be days to recovery and secondary outcomes include pain, disability and global perceived effect at 1 week, 2 weeks, 4 weeks and 3 months. The results of this trial will be available in 2007.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MH, CM, JL, AM, CC, RD, MS and JM were responsible for the design of the study. All authors 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
Description of active SMT.
Click here for file
Acknowledgements
This project is funded by the National Health and Medical Research Council of Australia. The authors wish to thanks the participating GPs and physiotherapists for their participation in this project.
==== Refs
Walker BF The prevalence of low back pain: a systematic review of the literature from 1966 to 1998 Journal of Spinal Disorders 2000 13 205 217 10872758 10.1097/00002517-200006000-00003
Australian Institute of Health and Welfare Australia's health 2000: the seventh biennial health report of the Australian Institute of Health and Welfare 2000 Canberra, AIHW
Pengel LHM Herbert RD Maher CG Refshauge KM Acute low back pain: systematic review of its prognosis. BMJ 2003 327 323 12907487 10.1136/bmj.327.7410.323
Koes BW van Tulder MW Ostelo R Kim Burton A Waddell G Clinical guidelines for the management of low back pain in primary care: an international comparison. Spine 2001 26 2504 13; discussion 2513-4 11707719 10.1097/00007632-200111150-00022
Australian Acute Musculoskeletal Pain Guidelines Group Evidence-Based Management of Acute Musculoskeletal Pain - A guide for clinicians 2003 Bowen Hills, Australian Academic Press
Ferreira ML Ferreira PH Latimer J Herbert R Maher CG Efficacy of spinal manipulative therapy for low back pain of less than three months' duration J Manipulative Physiol Ther 2003 26 593 601 14673408 10.1016/j.jmpt.2003.08.010
Assendelft WJ Morton S Yu E Suttorp M Shekelle P Spinal manipulative therapy for low back pain. A meta-analysis of effectiveness relative to other therapies. Annals of Internal Medicine 2003 138 871 881 12779297
Cherkin D Sherman KJ Deyo R Shekelle P A review of the evidence for the effectiveness, safety and cost of acupuncture, massage therapy and spinal manipulation for back pain Annals of Internal Medicine 2003 138 898 906 12779300
Curtis P Carey T Evans P Rowane M Garrett J Jackman A Training primary care practitioners to give limited manual therapy for low back pain Spine 2000 25 2954 2961 11074684 10.1097/00007632-200011150-00018
van Tulder MW Scholten RJ Koes BW Deyo RA Non-steroidal anti-inflammatory drugs for low back pain Cochrane Database Syst Rev 2000 CD000396 10796356
Schattenkirchner M Milachowski KA A double-blind, multicentre, randomised clinical trial comparing the efficacy and tolerability of aceclofenac with diclofenac resinate in patients with acute low back pain Clin Rheumatol 2003 22 127 135 12740678 10.1007/s10067-003-0710-9
Dreiser RL Marty M Ionescu E Gold M Liu JH Relief of acute low back pain with diclofenac-K 12.5 mg tablets: a flexible dose, ibuprofen 200 mg and placebo-controlled clinical trial Int J Clin Pharmacol Ther 2003 41 375 385 14518597
de Vet HC Heymans MW Dunn KM Pope DP van der Beek AJ Macfarlane GJ Bouter LM Croft PR Episodes of low back pain: a proposal for uniform definitions to be used in research Spine 2002 27 2409 2416 12438991 10.1097/00007632-200211010-00016
Pengel LHM Refshauge KM Maher CG Responsiveness of pain, disability, and physical impairment outcomes in patients with low back pain. Spine 2004 29 879 883 15082988 10.1097/00007632-200404150-00011
Roland M Morris R A study of the natural history of back pain. Part I: development of a reliable and sensitive measure of disability in low-back pain Spine 1983 8 141 144 6222486
Stratford P Gill C Westaway M Binkley J Assessing disability and change on individual patients: a report of a patient specific measure Physiotherapy Canada 1995 47 258 263
Maitland GD Vertebral manipulation 1986 5th London, Butterworths
Grieve G Mobilisation of the spine. Notes on examination, assessment and clinical method 1984 4th Edinburgh, Churchill Livingston
Jull G Treleaven J Versace G Manual examination: is pain provocation a major cue for spinal dysfunction? Australian Journal of Physiotherapy 1994 40 159 165
Jull G Treleaven J Versace G Boyling J PN Examination of the articular system Grieve's modern manual therapy The Vertebral Column 1994 2nd Edinburgh, Churchill Livingston
Farrar JT Young JPJ LaMoreaux L Werth JL Poole RM Clinical importance of changes in chronic pain intensity measured on an 11-point numerical pain rating scale.[see comment] Pain 2001 94 149 158 11690728 10.1016/S0304-3959(01)00349-9
Long A Donelson R Fung T Does it Matter Which Exercise? A Randomised Control Trial of Exercise for Low Back Pain Spine 2004 29 2593 2602 15564907 10.1097/01.brs.0000146464.23007.2a
Pocock SJ Clayton TC Altman DG Survival plots of time-to-event outcomes in clinical trials: good practice and pitfalls. Lancet 2002 359 1686 1689 12020548 10.1016/S0140-6736(02)08594-X
|
16280089
|
PMC1298305
|
CC BY
|
2021-01-04 16:32:05
|
no
|
BMC Musculoskelet Disord. 2005 Nov 10; 6:57
|
utf-8
|
BMC Musculoskelet Disord
| 2,005 |
10.1186/1471-2474-6-57
|
oa_comm
|
==== Front
BMC NeurosciBMC Neuroscience1471-2202BioMed Central London 1471-2202-6-631625963610.1186/1471-2202-6-63Research ArticleThe neuroblast and angioblast chemotaxic factor SDF-1 (CXCL12) expression is briefly up regulated by reactive astrocytes in brain following neonatal hypoxic-ischemic injury Miller Jason T [email protected] John H [email protected] Hereward JC [email protected] Aisha L [email protected] David C [email protected] William D [email protected] James E [email protected] Department of Neurology, Medical College of Georgia, Augusta, GA 30912, USA2 Department of Pediatrics, Medical College of Georgia, Augusta, GA 30912, USA3 Veteran's Affairs Medical Center, 1 Freedom Way, Augusta, GA 30904, USA4 Department of Cellular Biology & Anatomy, Medical College of Georgia, Augusta, GA 30912, USA2005 31 10 2005 6 63 63 10 5 2005 31 10 2005 Copyright © 2005 Miller et al; licensee BioMed Central Ltd.2005Miller 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
Stromal cell-derived factor 1 (SDF-1 or CXCL12) is chemotaxic for CXCR4 expressing bone marrow-derived cells. It functions in brain embryonic development and in response to ischemic injury in helping guide neuroblast migration and vasculogenesis. In experimental adult stroke models SDF-1 is expressed perivascularly in the injured region up to 30 days after the injury, suggesting it could be a therapeutic target for tissue repair strategies. We hypothesized that SDF-1 would be expressed in similar temporal and spatial patterns following hypoxic-ischemic (HI) injury in neonatal brain.
Results
Twenty-five 7-day-old C57BL/J mice underwent HI injury. SDF-1 expression was up regulated up to 7 days after the injury but not at the later time points. The chief sites of SDF-1 up regulation were astrocytes, their foot processes along blood vessels and endothelial cells.
Conclusion
The localization of SDF-1 along blood vessels in the HI injury zone suggests that these perivascular areas are where chemotaxic signaling for cellular recruitment originates and that reactive astrocytes are major mediators of this process. The associated endothelium is likely to be the site for vascular attachment and diapedesis of CXCR4 receptor expressing cells to enter the injured tissue. Here we show that, relative to adults, neonates have a significantly smaller window of opportunity for SDF-1 based vascular chemotaxic recruitment of bone marrow-derived cells. Therefore, without modification, following neonatal HI injury there is only a narrow period of time for endogenous SDF-1 mediated chemotaxis and recruitment of reparative cells, including exogenously administered stem/progenitor cells.
==== Body
Background
Hypoxic-ischemic (HI) injury to the brain is a significant cause of neurological morbidity in infants. The resultant severe and longstanding disability to infants has provided impetus for exploring aggressive therapies that give promise of injury reversal. Transplantation of stem cells has emerged as a possible modality to help improve function in damaged tissues [1].
In organ-specific injury models, bone marrow cells have been shown to accumulate at the site of injury and to differentiate into tissue specific cells [2-5]. Chen and colleagues [6] reported that intravenously administered marrow stromal cells accumulate in the site of ischemic brain injury, with a small percentage expressing neuronal markers. Additionally, Hess et al. [7,8] have demonstrated that, following temporary ischemia, endogenous murine bone marrow cells give rise to perivascular/mural cells, microglial cells, and a limited number of cells positive for the neuronal marker, NeuN. However, it is unclear if the NeuN positive cells develop into mature neurons.
Among the milieu of chemicals expressed in the injured microenvironment, the chemotactic chemokine, stromal cell derived factor-1 (SDF-1 or CXCL12), has emerged as a major attractant of reparative cells. SDF-1 and its receptor, CXCR4, play a critical role in both engraftment of hematopoietic stem cells into bone marrow and the mobilization of stem/progenitor cells from bone marrow niches into the peripheral circulation where these cells are then available for delivery to sites of peripheral injury [9-11]. In addition to functioning in the hematopoietic system, CXCR4 positive cells respond to SDF-1 signals in non-hematopoietic injured organs [12]. That exogenous bone marrow-derived cells home to and migrate within the site of injury suggests that the brain's response to injury produces signals that attract these cells [7,8,13]. Functionally, SDF-1 is a powerful chemoattractant for CXCR4-expressing bone marrow derived cells, including putative bone marrow stem/progenitor cells such as CD34+ cells both in vitro and in vivo [14].
In addition to targeting hematopoietic cells, the SDF-1/CXCR4 system is also known to play a critical role in embryonic brain development and adult injury repair including neurogenesis, neuroblast migration, and neuronal organization as well as endothelial progenitor cell recruitment, endothelial cell migration and vasculogenesis [15-25].
Therefore, it is important to determine the spatial and temporal expression of SDF-1 relative to injury; SDF-1 expression patterns may establish not only the location and duration of the window of opportunity for endogenous injury-mediated chemotaxis of reparative cells, but also the timeframe for therapeutic transplantation and recruitment of various stem cell populations. Recently, we showed in an adult mouse model of stroke that SDF-1 is expressed perivascularly in the injured region up to 30 days after middle cerebral artery occlusion [13], suggesting that there may be a long post-injury window for homing of reparative cells. Anticipating that SDF-1 might be a critical factor in attracting transplanted cells to the injured area of the neonatal brain, we hypothesized that SDF-1 would be expressed in neonatal hypoxic-ischemic injury in a temporal and spatial pattern similar to that seen in adult models of ischemic brain injury.
Results
Spatial expression pattern of SDF-1 in uninjured brain
The SDF-1 antibody used for immunohistochemical localization of SDF-1 recognizes full length recombinant murine SDF-1 (Figure 1) and preabsorption with SDF-1 blocks its recognition of antigen on tissue sections (Figure 2). In uninjured (control) neonatal mice (P8 through P17) endogenous protein expression of SDF-1 in brain is limited and changes over time (Figures 3, 4, 5). At P8 (Figure 3) the most prominent SDF-1 expression is within the hippocampus associated with the hippocampal fissure and the neighbouring molecular layer of the dentate gyrus, the hippocampal lacunosum molecular layer, the dentate gyrus polymorph layer, the stratum radiatum and the oriens layer. Strong expression is also associated with cells in the choroid plexus, the neurogenic subventricular zone, and large cortical neurons in the cingulate and neocortex. Additionally, large white matter tracks such as the corpus callosum and external capsule, as well as cells of the meninges, also demonstrate SDF-1 expression. Within the white matter tracks and the hippocampal fissure numerous blood vessels are associated with SDF-1 labelling (Figure 3). However, unexpectedly, the granular cell layer of the dentate gyrus, CA and layer V neurocortical neurons do not show strong immunohistochemical staining for SDF-1 (Figure 3). There is some shift in SDF-1 protein expression with time. By P14 the hippocampal expression changes slightly (Figure 4). Expression associated with the hippocampal fissure is reduced while there is increased expression in hippocampal areas associated with dendritic arborization and axonal fields, including the dentate gyrus polymorph layer, the molecular layer of the dentate gyrus, the hippocampal lacunosum molecular layer and the stratum radiatum (Figure 4).
Figure 1 SDF-1 Western blot. Recombinant murine SDF-1α (25 or 100 ng per lane) was run on a one-D gel, transferred to PVDF membrane and probed with the SDF-1 antibody used in the immunohistochemical assays. The western blot demonstrates that the SDF-1 antibody recognizes murine SDF-1α.
Figure 2 SDF-1 Antibody Specificity. The polyclonal SDF-1 antibody was tested for non-SDF-1 antigenic interactions and non-specific tissue binding. Panel A demonstrates the SDF-1 antibody's recognition of SDF-1 antigen on mouse hippocampus. Panel B demonstrates that preabsorption of the SDF-1 antibody with recombinant murine SDF-1α completely blocks specific binding of the antibody to the tissue with only very low (diffuse background) non-specific binding. Panel C shows that leaving the primary (SDF-1) antibody incubation step out of the staining protocol does not result in non-specific binding of the secondary antibody or the detection reagents. Scale bar equals 200 μm.
Figure 3 Constitutive SDF-1 protein expression at P8. The left column shows SDF-1 antibody staining of P8 mouse brain (Cy3 – red). The middle column shows the corresponding nuclear labelling (bis-benzimide – blue) of the same sections. The right hand column shows the merged images. Panels A-C show the distribution of constitutive SDF-1 expression at low power. The arrows denote diffuse labelling over the neocortex. Panels C-E demonstrate the hippocampal SDF-1 expression pattern. Panels G-I show a high magnification of the hippocampal fissure down through the upper blade of the dentate gyrus granular cell layer. The arrowheads point out SDF-1 positive blood vessels. Abbreviations: ca – cornus ammons (CA1, 2, 3), cp – choroid plexus, dhc – dorsal hippocampal commissure, ec – external capsule, fi – fimbria hippocampus, fmj – forceps major corpus callosum, GrDG – granular layer dentate gyrus, hif – hippocampal fissure, LMol – lacunosum moleculare layer hippocampus, Mol – molecular layer dentate gyrus, Or – oriens layer, PoDG – polymorph layer dentate gyrus, Rad – stratum radiatum. All Scale bars equal 200 μm.
Figure 4 Constitutive SDF-1 protein expression at P14. Panel A shows the pattern of SDF-1 expression (Cy3 – red) at P14 (Scale bar equals 200 um). Panel B demonstrates the hippocampal expression of SDF-1 (Cy3 – red) (Scale bar equals 200 μm). Panel C shows bis-benzimide nuclear labelling (blue) from the same field as Panel B. Abbreviations: ca – cornus ammons (CA1, 2, 3), cp – choroid plexus, GrDG – granular layer dentate gyrus, hif – hippocampal fissure, LMol – lacunosum moleculare layer hippocampus, Mol – molecular layer dentate gyrus, Or – oriens layer, PoDG – polymorph layer dentate gyrus, Py – pyramidal cell layer, Rad – stratum radiatum. S – subiculum, scc – splenium corpus callosum.
Figure 5 Constitutive and Injury Induced Expression of SDF-1. All panels are from P10 murine neonates. Panels A-C are from a non-Hypoxia-Ischemia (HI) control animal; D-I are from an animal that underwent HI injury three days previously at P7. Panels D-F are from the HI injured hemisphere and G-I from the corresponding contralateral (non-ischemic) hemisphere. Panels A, D, and G are of the lateral ventricle including the CP, SVZ (arrow) and adjacent striatum (*); arrowheads indicate SDF-1 expression (Cy3 – red) in the CP, scale bar equals 100 μm. Panels B, E, & H are of the cingulate cortex, arrowheads indicate SDF-1 expression in neurons, scale bar equals 50 μm. Panels C, F, and I show cortex and meninges, arrowheads indicate SDF-1 expression in mesothelial cells of the pia, scale bar equals 100 microns. Cell nuclei were identified by bis-benzimide counterstaining (blue). Abbreviations: choroid plexus (CP) and subventricular zone (SVZ).
Changes in spatial expression following HI injury
Following HI injury the endogenous constitutive expression in the large neurons of the cingulate and neocortex is lost. However, at the same time SDF-1 expression is dramatically up regulated in not only areas directly affected by HI, but also in the choroid plexus and by mesothelial cells of the pial meninges (Figures 5 and 6). The expression of SDF-1 in areas of HI injury principally includes the hippocampus and striatum (figures 6 and 7). Additionally, there is increased and extensive expression by cells within the HI side hemisphere white matter, particularly in the corpus callosum, external capsule and hippocampal fimbria (Figure 6). Interestingly, the choroid plexus and meninges expression of SDF-1 increases bilaterally with HI injury, although the degree of expression in the contralateral choroid plexus appears to be less robust than the HI side choroid plexus (Figures 5 and 6). This suggests a generalized injury response that extends beyond the area of ischemia but potentially within the area affected by hypoxia.
Figure 6 SDF-1 Expression 3 days Post Hypoxia-Ischemia. A coronal section from a P10 mouse that underwent hypoxia-ischemia (with right side ischemia) three days earlier, on postnatal day 7, was probed with a Cy3 (red) antibody to demonstrate SDF-1 expression. Single arrowheads indicate choroid plexus, double arrowheads the hippocampus, the large arrow indicates the fimbria of the hippocampus, small arrows the corpus callosum and external capsule, the asterisk (*) denotes the striatum. All five animals at this time point showed equivalent levels of SDF-1.
Figure 7 Temporal Expression of SDF-1 in Hippocampus Post-HI. Shows examples of hippocampal expression of SDF-1 at different time points post hypoxia-ischemia (HI) demonstrated by immunohistochemistry (Cy3 label – red), this includes 1 (P8), 3 (P10), 5 (P12), 7 (P14) and 10 (P17) days following HI. Panels A-E are from the ischemic hemisphere and a-e are from the contralateral hemisphere. M and L denote the medial and lateral aspects of the hippocampi respectively. Hippocampal regions are indicated: cornu ammonis (CA 1, 2 & 3), corpus callosum (CC), and dentate gyrus (DG). Cell nuclei were identified by bis-benzimide counterstaining (blue). Scale bar equals 200 μm.
Temporal expression pattern of SDF-1 following HI injury
The first day after HI injury (P8) patchy SDF-1 expression was demonstrated around blood vessels and within the parenchyma in the hippocampus and striatum within the ischemic hemisphere. By 3 days after HI injury (P10) there was robust, uniform hippocampal expression of SDF-1 in the CA1, CA2, CA3 and dentate regions as well as within the striatum and to a limited degree in the cingulate cortex (Figures 5, 6, 7). By day 5 after the injury (P12) SDF-1 continued to be expressed but was significantly reduced in the hippocampus and striatum. However, its expression was still increased in the choroid plexus (bilaterally) and was associated with blood vessels in the white matter of the corpus callosum and external capsule as well as pial mesothelial cells (bilaterally). By day 7 (P14), SDF-1 expression continued to wane, and by day 10 post-HI injury (P17) and beyond, expression was reduced below observable levels in the hippocampus, striatum and cortical areas. The only expression of SDF-1 at these late times was found along white matter tracts of the corpus callosum and the alveus and in some sites where normal constitutive expression is observed, specifically the subventricular zone, choroid plexus and a subset of large cingulate cortex and neocortical neurons. However, expression appeared to be depressed in hippocampal areas that normally expressed SDF-1.
Cellular expression of SDF-1
For all time points demonstrating injury responsive SDF-1 expression, cellular expression was in large part localized to GFAP+ reactive astrocytes (Figures 8 and 9). Astrocyte associated expression of SDF-1 is frequently found around small blood vessels, suggesting that this perivascular expression is related to astrocytic foot processes (Figures 8 and 9). The lectin RCA-1 labels cell surface carbohydrate moieties associated with both microglial and endothelial cells. The punctate RCA+/SDF-1+ labelling resembles cellular processes, however, the cell type demonstrating this co-labeling is uncertain, morphologically it does not resemble GFAP+/SDF-1+ astrocytic cell processes or endothelial cell profiles (Figure 10). Some of the SDF-1+/RCA-1+ profiles do appear to be ramified microglial cells. However, while a subset of microglia in the HI injury area co-label with SDF-1 the majority do not (data not shown). Further, SDF-1 did not co-label with markers for other process bearing cell types including oligodendrocytes (data not shown). Also, as described earlier, choroid plexus and pial epithelial cells demonstrate SDF-1 expression (Figures 3, 5 and 6).
Figure 8 SDF-1 and GFAP Co-localization. Panels A-C are from a P10 hippocampus, three days post-HI, from the ischemic hemisphere, a-c are from the corresponding contralateral hippocampus. Panels A & a demonstrate the astrocyte marker GFAP (FITC – green). Panels B & b show SDF-1 expression (Cy3 – red). Panels C & c indicate the overlap between GFAP and SDF-1 labeling. Scale bar equals 100 μm.
Figure 9 Astrocyte Expression of SDF-1. GFAP and SDF-1 dual labeling is demonstrated by confocal microscopy in P10 mice, three days post-HI. Panels A-C, from the striatum, show perivascular (arrowheads) astrocyte (GFAP – green) expression of SDF-1 (red). Panels D-F, from the hippocampus, show (arrowhead) similar co-labeling of GFAP (green) astrocyte cell bodies and processes for SDF-1 (red). Panels C and F show GFAP and SDF-1 co-labeling in merged orthogonal views of confocal z stacks. GFAP staining is shown in green and SDF-1 staining in Red. Panels A, B, D, and E are single z plane field images captured with a 63× objective. Scale bar equals 50 μm.
Figure 10 SDF-1 and RCA-1 Co-localization. Figure shows RCA and SDF-1 dual labeling in hippocampus from P10 mice three days post-HI; RCA-1 labeling in hippocampus (A & D), SDF-1 expression in hippocampus (B & E), merged RCA-1 and SDF-1 labeling (C & F). Panels A-C represent 20× objective fields. Panels D-E are single z plane fields and panel F is a merged orthogonal view of a confocal z stack. The arrowhead points out an example of an SDF-1 negative RCA-1 positive cell. Scale bar equals 50 μm.
Discussion
The regulation of SDF-1 may be critical to the incorporation of a variety of new cells into brain. In contrast to the prolonged up regulation of SDF-1 following stroke in adult mice [13], its up regulation in neonatal mice after HI injury was brief. This may reflect a difference between a mature, relatively stable, adult brain structure and a more plastic, still developing, neonatal CNS where there may be a more active regulation of SDF-1 expression. The timing of post-HI injury SDF-1 expression suggests that SDF-1 mediated cell homing may be most effective from 3 to 5 days after HI injury in neonatal animals, peaking around 3 days. During this time SDF-1 expression is high around vascular elements, which may provide a portal of access for circulating reparative cells to the areas of ischemic injury.
Another possible explanation for the temporal differences of SDF-1 expression between injured adult and neonatal brains may be related to the type of injury. HI injury produces a more diffuse and graded tissue injury, while middle cerebral artery occlusion is more focal and discrete.
Constitutive SDF-1 expression
Interestingly, the pattern of peak expression for SDF-1 we show at postnatal day 10, 3 days after HI injury (Figure 6), is virtually identical to that recently published by Felszeghy et al for I125 SDF-1α binding to CXCR4 in a rat HI model, 2 days after HI injury, at postnatal day 9 [26]. In Felszeghy et al's study the localization of CXCR4 receptors suggests the distribution of SDF-1 expression. Protein expression observed in our study, with SDF-1 antibody probes, is consistent with mRNA expression patterns assayed by in situ hybridization in previous reports with some exceptions [26-28]. The SDF-1 mRNA expression patterns Tham et al and Stumm et al [22,23] show at days P7 and P9 in rats match protein expression patterns we detected here at P8 and P10 in mice (Figures 3, 5 and 6). This includes SDF-1 expression in the meninges, the hippocampal fissure, choroid plexus and the proliferative subventricular zone. SDF-1 mRNA expression in both in situ hybridization studies [22,23] was also associated, to a limited degree, with blood vessels, we saw the same constitutive pattern for SDF-1 immunostaining particularly with vessels associated with the hippocampal fissure and with white matter tracks (Figure 3). In the in situ hybridization studies [22,23] there is a decline in hippocampal fissure SDF-1 mRNA expression between the early time points (P7 & P9) and later time points (P14 and P12 – P21). A similar decrease in hippocampal fissure SDF-1 immunostaining was observed in our study between P8 and P14 (Figures 3 and 4). Of interest, in the in situ hybridization studies [22,23], there are consistent increases in SDF-1 mRNA in the dentate gyrus granular layers, in CA3 (and CA4) and in layer V neocortical neurons between the early time points (P7 & 9) and later time points (P14 and P12 – P21). However, we did not detect significant SDF-1 immunolabelling over the cell bodies of the granular cells at equivalent time points in the mice (P8 – 14). In contrast we did detect immunolabelling in the molecular and polymorph layers, regions of granular cell dendritic arborizations and axon fields [27]. This labelling increased between P8 and P14 (Figures 3 and 4). Similarly we did not detect strong SDF-1 immunoreactivity over CA 3, 2 or 1 neuronal somas, but we found immunoreactivity that increased between P8 and P14 in the hippocampal lacunosum molecular layer, the stratum radiatum and the oriens layer (Figures 3 and 4), regions that contain the dendritic and axonal fields for the CA region [27]. Additionally, the layer V neocortical neurons are reported to show strong mRNA labelling [22,23], but we only detected weak diffuse antibody labelling in the area corresponding to layers V and VI in the neocortex (Figure 3).
The discrepancy between the reported neuronal somal labelling for mRNA [22-24] and the neuropil immunolabelling for SDF-1 protein we observed may be due to a lack of sensitivity of the antibody, although we can clearly detect SDF-1 expressing structures that are reported to express lower levels of SDF-1 mRNA than dentate gyrus granular cells, CA and layer V neocortical neurons [22,23]. There may also be differences in SDF-1 transcription and translation rates as well as message and protein turnover in different cell populations. Alternatively, this may reflect that SDF-1 mRNA is synthesized in, and restricted to, the neuronal somas while in some neurons the protein product is mainly exported to cellular processes with little remaining in the somas. For example, while the reported dentate gyrus granular cell mRNA expression [22-24] is more intense than any other location, including the hippocampal fissure, we show that this is at best poorly labelled immunohistochemically (Figures 3 and 4). In contrast we demonstrate that regions that reflect the synaptic fields (dendritic and axonal) of the granular cells (molecular and polymorph layers) have strong immunoreactive expression for SDF-1 (Figures 3 and 4) while these same regions show little mRNA expression for SDF-1 [22-24]. Further, structures lacking projecting processes, which could distance mRNA form protein localization, like the hippocampal fissure, choroid plexus, and meninges, are reported to be strongly labelled for SDF-1 mRNA by in situ hybridization [22-24] and we demonstrate also by immunohistochemistry for the SDF-1 protein. There is precedent for this type of pattern in neonatal hippocampus. Czarnecki et al have recently shown that synaptopodin mRNA was restricted to dentate granular layer, and CA3, cell somas while immunostaining for synaptopodin protein labeled dendritic layers but not the somas [28].
Injury induced SDF-1 expression
Recently Ceradini et al. demonstrated that SDF-1 gene expression is regulated by the hypoxia responsive transcription factor HIF-1 [21]. The level of SDF-1 expression is directly related to the degree of hypoxia such that hypoxic gradients within tissues correlate with gradients of SDF-1 gene expression [21]. We have previously shown in an adult middle cerebral artery occlusion model of stroke that SDF-1 expression follows a gradient pattern consistent with the hypoxia gradient in the penumbra [13]. In our neonatal HI model the hippocampus did not appear to show a gradient of SDF-1 labeling during peak expression, rather the labeling was uniform and delimited by anatomical boundaries. In contrast the striatal expression did appear to show a gradient of labeling intensity (Figure 6). Specific cell types may be more sensitive to hypoxic changes in inducing SDF-1 expression as illustrated by changes in constitutive expression of SDF-1 as well as induced expression in cells that did not show significant constitutive expression. For example, subsets of neocortical neurons responded by decreasing SDF-1 expression in both the hypoxic and HI affected hemispheres while choroid plexus and pial labeling increased on both sides and astrocytes dramatically expressed SDF-1 only in the HI hemisphere. This suggests that HIF-1 expression may be differentially induced or that other transcription factors may be involved as well.
Potential role of SDF-1 in injury repair
SDF-1 and its receptor CXCR4 are emerging as a common axis for directing the migration of stem cells associated with injury repair in many systems. Ji et al recently demonstrated that SDF-1 mediates the migration of mesenchymal stem cells to the site of injury [29]. Neural progenitor cells and neuroblasts express CXCR4 and are attracted by SDF-1α developmentally and following brain ischemia [25,30-32]. Recently SDF-1 has been shown to promote neovascularization in somatic tissues and to recruit bone marrow-derived (BMD) cells to areas of ischemic injury. In mouse models of peripheral tissue ischemia, SDF-1 treated animals had increased capillary density and blood flow compared to controls and recruited injected endothelial precursor cells (EPC) to integrate into blood vessels demonstrating vasculogenesis [16,20,21]. Importantly this SDF-1 induced neovascularization has been demonstrated to improve perfusion and function in peripheral tissue ischemic injury models [16,21,33]. Work from our laboratory shows SDF-1 is up regulated in the penumbral, or peri-infarct, zone within hours of acute ischemia in adult mice. This elevated expression persists in the peri-infarct zone for at least 30 days post-ischemia. Accompanying this we observe bone marrow-derived cell infiltration, pericyte and microglial engraftment and neovascularization [13]. Given the apparent function of SDF-1 to serve as an attractant for bone marrow-derived cells, including cells associated with vascular repair and vasculogenesis, as well as for other stem/progenitor cells, including neural stem cells, this brief time span of SDF-1 expression, peaking at 3 days with only limited expression through 7 days after the injury in neonatal animals, may be critical for the acquisition of new brain or reparative cells.
The cell type expressing SDF-1 and its localization is highly related to its function. Previously, we found that the principal localization of SDF-1 expression following injury was astrocytic, often strongly associated with small blood vessels [13]. This was the first demonstration that activated, or reactive astrocytes, express SDF-1 in vivo following ischemia injury [13]. The current study confirms and extends our original finding and demonstrates that SDF-1 expression is massively up regulated in reactive astrocytes in neonatal HI injury, but for a shorter period of time relative to adult middle cerebral artery occlusion ischemia. Recently Imitola [25] also reported SDF-1 expression in reactive astrocytes in an adult HI model. Earlier in vitro studies showed that cultured astrocytes up regulate SDF-1 expression following exposure to various cytotoxic or inflammation-inducing agents [34,35]. Additionally, we have recently shown that there is only minimal proliferation and maturation of astrocytes following HI, suggesting that the astrocytic expression of SDF-1 following HI originates from existing astrocytes [36].
The influence of SDF-1 upon ischemic recovery in neonates appears to be restricted to a few days under circumstances of "normal" endogenous repair. Such a brief up regulation of SDF-1 in neonatal animals may disadvantage their recovery relative to adults. However, it may still may provide a defined time frame in which therapeutically transplanted cells, or autologous stem cells mobilized from the bone marrow, may be able to home, engraft and differentiate into needed cell types. This may also provide an opportunity to target engineered cells that express specific factors to the site of injury. Using the SDF-1/CXCR4 axis as a vectoral targeting system may also permit local application of cell expressed therapeutic factors in higher doses than systemic delivery would safely allow. In adults it may be possible to use the natural up regulation of the SDF-1 for up to 30 days after the injury, while in neonates such a procedure would have to be carried out much sooner. If transplantation protocols are carried out in humans, it may be necessary to find ways of either up regulating the SDF-1 or engineering the expression of the chemokine to desired temporal and spatial locals.
Conclusion
SDF-1 up regulation following neonatal HI injury persists only a few days in comparison to the much longer up regulation in adult models of stroke. The up regulation in neonatal HI injury is quite prominent in astrocytes abutting small blood vessels. This process may be critical in attracting CXCR4 expressing cells to the brain. Awareness of these events could be helpful in planning stem cell transplantation therapy delivered via the vascular route.
Methods
Animals
This study was performed in accordance with the guidelines provided by the Laboratory Animal Studies Committee of the Veterans Administration Medical Center (Augusta, GA) and the Medical College of Georgia. The mice were housed in individual cages under standard conditions. Offspring were reared with their dams until the time of surgery, and then until weaning at three weeks of life.
Surgical procedure and treatment
Twenty-five C57 BL/6 mouse pups underwent permanent ligation of the left common carotid artery at post-natal day seven (P7) [37]. A midline cervical incision was made to expose the left common carotid artery, and a single suture ligated the artery. The incision was closed with interrupted 6-0 silk sutures. During the procedure, the mice were anesthetized with 2% isoflurane. The pups were then placed with the dams for two hours prior to placement in an 8% oxygen chamber partially immersed in a water bath at 37° for 75 minutes.
Animal sacrifice and tissue processing
The pups were sacrificed at P8, P10, P12, P14, P17, P21, or P35 (respectively, 1, 3, 5, 7, 10, 14 or 28 days after the injury). At least three animals were sacrificed at each time point. At the time of sacrifice, the animals were anesthetized with 70 mg/kg of ketamine and 15 mg/kg xylazine. Tissue was fixed by transcardiac perfusion with 0.9% saline, followed by 4% paraformaldehyde in PBS. Brains were removed and post-fixed in 4% paraformaldehyde for 4 hours, cut into 3–5 mm sections, and continued in 4% paraformaldehyde for an additional 20 hours. Tissue was then dehydrated in an ethanol series, cleared in xylene, and infiltrated and embedded in PolyFin Embedding and Infiltration Wax (TBS Biomedical Sciences, Durham, NC). Paraffin embedded tissue was sectioned 5 μ thick and mounted on Superfrost Plus slides (Fisher Scientific).
Immunohistochemistry
Section processing, antigen retrieval and immunostaining methods have been previously described [13,36] Briefly, slide mounted sections were deparaffinized through 3 xylene changes followed by rehydration in an alcohol dilution series. Sections were permeablized in 0.1% Triton-X 100 in PBS for ten minutes, followed by 3 washes in 1× PBS. Antigen retrieval was performed using a microwave defixation method. Slides were microwaved at a slow boil for 10 minutes in citrate buffer (0.01 M sodium citrate pH 6.5) then incubated at rest for twenty minutes, then washed in 3 changes of 1× PBS and blocked using 2% normal calf serum in PBS for 20 minutes at room temperature (RT).
Sections were labeled with antibodies to SDF-1 (Santa Cruz cat # sc-6193, goat, recognizes both SDF-1α & β splice variants). SDF-1 signal was detected by utilizing either a biotin/avidin (Vector Elite ABC kit) or peroxidase anti-peroxidase (PAP; Sternberger) amplification system and visualized with Cy3 conjugated antibody (Jackson ImmunoResearch Laboratories). In some instances dual labeling for SDF-1 and a second antigen was utilized to identify the phenotype of SDF-1 expressing cells. For co-labeling SDF-1 and the astrocyte marker glial fibrillary acidic protein (GFAP) a rabbit anti-GFAP antibody (DAKO cat# Z0334) was used, for oligodendricytes anti-CNPase (Sigma cat# C5922, mouse) and the biotinylated lectin Ricinus Communis Agglutinin I (RCA 1, Vector cat# B-1085, detects unblocked N-Acetyl galactosamine and galatose) for the visualization of endothelial cells and microglial cells [13,36]. All antibodies (and RCA-1) were allowed to incubate at room temperature for one hour and washed four times (1× PBS 5 minutes each). All primary antibodies were detected using indirect staining with a secondary, fluorescent antibody. Cy3 anti-Goat (Jackson # 705-165-147), diluted 1:400, was used to visualize SDF-1 in all cases. FITC anti-rabbit (Jackson # 711-095-152), diluted 1:100, was used to visualize GFAP. Strepavidin FITC (Jackson #016-010-084), diluted 1:100, was used to visualize biotinylated RCA. FITC anti-mouse (Jackson #715-095-151), diluted 1:100, was used to visualize CNPase. Slides were rinsed in 4 changes of PBS for 5 minutes each. Negative controls consisted of preabsorbing 660 ng/ml, (4.3 pM/ml) the SDF-1 antibody (Santa Cruz cat # sc-6193, goat) with 1.5 ng/ml (43 pM/ml) of recombinant murine SDF-1α (Peprotech # 250-20A) at 37°C for 1 hour prior to use or leaving off the primary antibody. Cell nuclei were counterstained with bis-benzimide (Sigma, 10 mg/ml stock diluted 1:12,000 in phosphate buffered saline) for 8 minutes then given a final wash prior to coverslipping with Vectashield mounting media (Vector Labs #H-1000).
Western blotting
Recombinant murine SDF-1α (25 & 100 ng) was run on a Novex NuPage 10% Bis-Tris Gel (Invitrogen NP0302BOX) with MES buffer at 75 volts then transferred to a 0.2 um PVDF membrane (Millipore Immobilon PSQ #ISEQ00010). The blot was blocked with 4% cold water fish gelatin (Sigma #G-7765) in PBS at room temperature for 1 hour. The SDF-1 antibody (Santa Cruz cat # sc-6193, goat) was incubated shaking (1:100, 10 ug/ml) overnight at 4°C. The blot was rinsed, then washed 2 times for 10 minutes each in PBS with 0.5% Tween-20 (Fisher #BP337-500). The anti-SDF-1 antibody was detected with an anti-goat secondary antibody conjugated with HRP (Jackson ImmunoResearch #705-035-003, 1:50,000) for 1 hour at room temperature. The blot was washed again (3 times for 10 minutes each in PBS with 0.5% Tween-20 and 1 time with PBS only), developed with ECL detection reagent (Amersham # RPN2109D1) and exposed to X-Ray film (Fujifilm Super RX #100NIF).
Confocal imaging
Confocal imaging was used to confirm dual labeling with a Zeiss 510 laser scanning confocal microscope (Zeiss LTM software) as previously described [13,36]. The objective used was Zeiss 63× C-Apochromat 1.2NA. To excite the FITC fluorochrome (green), a 488-nm laser line generated by an argon laser was used, and for the Cy3 fluorochrome (red), a 543-nm laser from a HeNe laser was used. Filter sets used were a bandpass 500- to 600-filter ("green" channel) and a long-pass 585- to 650-nm filter ("red" channel). We identified cells that were labeled for SDF-1 and either GFAP or RCA with 1-μm step Z series.
Authors' contributions
JM, HW, and AW performed the tissue cutting, staining and section assessments. JM also assisted in writing the manuscript. DH participated in the conception of the project and in evaluation of the results. WDH participated in the conception of the project, directed the immunohistochemical studies, assisted in evaluation of the results and in drafting the manuscript. JC developed and directed the overall project, evaluated the results, and drafted the manuscript.
==== Refs
Bartley J Carroll JE Stem cell therapy for cerebral palsy Expert Opin Biol Ther 2003 3 541 549 12831360
Ferrari G Cusella-DeAngelis G Coletta M Paolucci E Stornaiuolo A Cossu G Mavilio F Muscle regeneration by bone marrow-derived myogenic progenitors Science 1998 279 1528 1530 9488650 10.1126/science.279.5356.1528
Gussoni E Soneoka Y Strickland CD Buzney EA Khan MK Flint AF Kunkel LM Mulligan RC Dystrophin expression in the mdx mouse restored by stem cell transplantation Nature 1999 401 390 394 10517639 10.1038/43922
Petersen BE Bowen WC Patrene KD Mars WM Sullivan AK Murase N Boggs SS Greenberger JS Goff JP Bone marrow as a potential source of hepatic oval cells Science 1999 284 1168 1170 10325227 10.1126/science.284.5417.1168
Priller J Persons DA Klett FF Kempermann G Kreutzberg GW Dirnagl U Neurogenesis of cerebellar purkinje neurons from gene-marked bone marrow cells in vivo J Cell Biol 2001 155 733 738 11724815 10.1083/jcb.200105103
Chen J Li Y Wang L Lu M Zhang X Chopp M Therapeutic benefit of intracerebral transplantation of bone marrow stromal cells after cerebral ischemia in rats J Neurol Sci 1997 189 49 57 11535233 10.1016/S0022-510X(01)00557-3
Hess DC Hill WD Martin-Studdard A Carroll J Brailer J Carothers J Bone marrow as a source of endothelial cells and NeuN-expressing cells after stroke Stroke 2002 33 1362 1368 11988616 10.1161/01.STR.0000014925.09415.C3
Hess DC Abe T Hill WD Studdard AM Carothers J Masuya M Fleming PA Drake CJ Ogawa M Hematopoietic origin of microglial and perivascular cells in brain Exp Neurol 2004 186 134 44 15026252 10.1016/j.expneurol.2003.11.005
Peled A Grabovsky V Habler L Sandbank J Arenzana-Seisdedos F Petit I Ben-Hur H Lapidot T Alon R The chemokine SDF-1 stimulates integrin-mediated arrest of cd34(+) cells on vascular endothelium under shear flow J Clin Invest 1999 104 1199 1211 10545519
Hattorri K Heissig B Rafii S The regulation of hematopoietic stem cell and progenitor mobilization by chemokine SDF-1 Leukemia & Lymphoma 2003 44 575 82 12769333 10.1080/1042819021000037985
Liles WC Broxmeyer HE Rodger E Wood B Hubel K Cooper S Hangoc G Bridger GJ Henson GW Calandra G Dale DC Mobilization of hematopoietic progenitor cells in healthy volunteers by AMD a CXCR4 antagonist Blood 3100 102 2728 30 12855591 10.1182/blood-2003-02-0663
Kollet O Shivtiel S Chen Y-Q Suriawinata J Thung SN Dabeva MD Kahn J Speigel A Dar A Samir S Goichberg P Kalinkovich A Arenzana-Seisdedos F Nagler A Izhar I Revel M Shafritz DA Lapidot TH HGH, SDF-1, and MMP-9 are involved in stress-induced human CD34+ stem cell recruitment to the liver J Clin Invest 2003 112 160 169 12865405 10.1172/JCI200317902
Hill WD Hess DC Martin-Studdard A Carothers JJ Zheng J Hale D Maeda M Fagan SC Carroll JE Conway SJ SDF-1 (CXCL 12) is up regulated in the ischemic penumbra following stroke: association with bone marrow cell homing to injury J Neuropath Exp Neurology 2004 63 84 96
Auiti AT Webb IJ Bleul C Springer T Gutierrez JC The chemokine SDF-1 is achemoattracttant for human CD34+ hematopoietic progenitor cells and provides a new mechanism to explain the mobilization of CD34+ progenitors for peripheral blood J Exp Med 1997 185 111 120 8996247 10.1084/jem.185.1.111
Lazarini F Tham TN Casanova P Arenzana-Seisdedos F Dubois-Dalcq M Role of the α-chemokine stromal cell-derived factor (SDF-1) in the developing and mature central nervous system GLIA 2003 42 139 148 12655598 10.1002/glia.10139
Yamaguchi J-I Kusano KF Masuo O Kawamoto A Silver M Murasawa S Bosch-Marce M Masuda H Losordo DW Isner JM Asahara T Stromal cell-derived factor-1 effects on ex-vivo expanded endothelial progenitor cell recruitment for ischemic neovascularization Circulation 2003 107 1322 1328 12628955 10.1161/01.CIR.0000055313.77510.22
Salcedo R Wasserman K Young HA Vascular endothelial growth factor and basic fibroblast growth factor induce expression of CXCR4 on human endothelial cells: In vivo neovascularization induced by stromal-derived factor-1alpha Am J Pathol 1999 154 1125 35 10233851
Mirshahi F Pourtau J Li H SDF-1 activity on microvascular endothelial cells: Consequences on angiogenesis in vitro and in vivo models Thromb Res 2000 99 587 94 10974345 10.1016/S0049-3848(00)00292-9
Salvucci O Yao L Villalba S Sajewicz A Pittaluga S Tosato G Regulation of endothelial cell branching morphogenesis by endogenous chemokine stromal-derived factor-1 Blood 2002 99 2703 11 11929756 10.1182/blood.V99.8.2703
Askari AT Unzek S Popovic ZB Goldman CK Forudi F Kiedrowski M Rovner A Ellis SG Thomas JD DiCorleto PE Topol EJ Penn MS Effect of stromal-cell-derived factor 1 on stem-cell homing and tissue regeneration in ischaemic cardiomyopathy Lancet 2003 362 697 703 12957092 10.1016/S0140-6736(03)14232-8
Ceradini DJ Kulkarni AR Callaghan MJ Tepper OM Bastidas N Kleinman ME Capla JM Galiano RD Levine JP Gurtner GC Progenitor cell trafficking is regulated by hypoxic gradients through HIF-1 induction of SDF-1 Nat Med 2004 10 858 64 15235597 10.1038/nm1075
Tham TN Lazarini F Franceschini IA Lachapelle F Amara A Dubois-Dalcq M Developmental pattern of expression of the alpha chemokine stromal cell-derived factor 1 in the rat central nervous system Eur J Neurosci 2001 13 845 56 11264658 10.1046/j.0953-816x.2000.01451.x
Stumm RK Zhou C Ara T Lazarini F Dubois-Dalcq M Nagasawa T Hollt V Schulz S CXCR4 Regulates Interneuron Migration in the Developing Neocortex J Neurosci 2005 23 5123 5130 12832536
Lu M Grove EA Miller RJ Abnormal development of the hippocampal dentate gyrus in mice lacking the CXCR4 chemokine receptor Proc Natl Acad Sci U S A 2002 99 7090 95 11983855 10.1073/pnas.092013799
Imitola J Raddassi K Park KI Mueller F-J Nieto M Teng YD Frenkel D Li J Sidman RL Walsh CA Snyder EY Khoury SJ Directed migration of neural stem cells to sites of CNS injury by the stromal cell-derived factor 1a/CXC chemokine receptor 4 pathway PNAS 2004 101 18117 18122 15608062 10.1073/pnas.0408258102
Felszeghy K Banisadr G Rostene W Nyakas C Haour F Dexamethasone downregulates chemokine receptor CXCR4 and exerts neuroprotection against hypoxia/ischemia-induced brain injury in neonatal rats Neuroimmunomodulation 2004 11 404 13 15467356 10.1159/000080151
Traub RD Neuronal Networks of the Hippocampus 1991 Cambridge, Cambridge: University Press
Czarnecki K Haas CA Orth CB Deller T Frotscher M Postnatal Development of Synaptopodin Expression in the Rodent Hippocampus J Comp Neurol 2005 490 133 144 16052494 10.1002/cne.20651
Ji JF He BP Dheen TS Tay SST Interactions of chemokine receptors mediate the migration of mesenchymal stem cells to the impaired site in the brain after hypoglossal nerve injury Stem Cells 2004 22 415 427 15153618 10.1634/stemcells.22-3-415
Ni HT Hu S Sheng WS Olson JM Cheeran MC-J Chan ASH Lokensgard JR Peterson PK High-level expression of functional chemokine receptor CXCR4 on human neural precursor cells Dev Brain Res 2004 152 159 169 15351504 10.1016/j.devbrainres.2004.06.015
Peng H Huang Y Rose J Erichsen D Herek S Fujii N Tamamura H Zheng J Stromal cell-derived factor 1-mediated CXCR4 signaling in rat and human cortical neural progenitor cells J Neurosci Research 2004 76 35 50 15048928 10.1002/jnr.20045
Tran PB Ren D Veldhouse TJ Miller RJ Chemokine receptors are expressed widely by embryonic and adult neural progenitor cells J Neurosci Res 2004 76 20 34 15048927 10.1002/jnr.20001
Hiasa K Ishibashi M Ohtani K Inoue S Zhao Q Kitamoto S Sata M Ichiki T Takeshita A Egashira K Gene transfer of stromal cell-derived factor-1alpha enhances ischemic vasculogenesis and angiogenesis via vascular endothelial growth factor/endothelial nitric oxide synthase-related pathway: next-generation chemokine therapy for therapeutic neovascularization Circulation 2004 109 2454 61 15148275 10.1161/01.CIR.0000128213.96779.61
Bajetto A Bonavia R Barbero S Glial and neuronal cells express functional chemokine receptor CXCR4 and its natural ligand stromal cell-derived factor 1 J Neurochem 1999 73 2348 57 10582593 10.1046/j.1471-4159.1999.0732348.x
Guillemin GJ Croitoru-Lamoury J Dormont D Armati PJ Brew BJ Quinolinic acid upregulates chemokine production and chemokine receptor expression in astrocytes Glia 2003 41 371 81 12555204 10.1002/glia.10175
Bartley JH Soltau T Wimborne H Kim S Martin-Studdard A Hess DC Hill WD Waller JL Carroll JE BrdU-Positive Cells in the Neonatal Mouse Hippocampus Following Hypoxic-Ischemic Brain Injury BMC Neurology 2005 6 15
Rice JE Vannucci RC Brierly JB The influence of immaturity on hypoxic-ischemic brain damage in the rat Ann Neurol 1981 9 131 141 7235629 10.1002/ana.410090206
|
16259636
|
PMC1298306
|
CC BY
|
2021-01-04 16:39:09
|
no
|
BMC Neurosci. 2005 Oct 31; 6:63
|
utf-8
|
BMC Neurosci
| 2,005 |
10.1186/1471-2202-6-63
|
oa_comm
|
==== Front
BMC NursBMC Nursing1472-6955BioMed Central London 1472-6955-4-61628008710.1186/1472-6955-4-6Research ArticleA qualitative study of nursing student experiences of clinical practice Sharif Farkhondeh [email protected] Sara [email protected] Psychiatric Nursing Department, Fatemeh (P.B.U.H) College of Nursing and Midwifery Shiraz University of Medical Sciences, Zand BlvD, Shiraz, Iran2 English Department, Shiraz University, Shiraz, Iran2005 9 11 2005 4 6 6 10 6 2005 9 11 2005 Copyright © 2005 Sharif and Masoumi; licensee BioMed Central Ltd.2005Sharif and Masoumi; 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
Nursing student's experiences of their clinical practice provide greater insight to develop an effective clinical teaching strategy in nursing education. The main objective of this study was to investigate student nurses' experience about their clinical practice.
Methods
Focus groups were used to obtain students' opinion and experiences about their clinical practice. 90 baccalaureate nursing students at Shiraz University of Medical Sciences (Faculty of Nursing and Midwifery) were selected randomly from two hundred students and were arranged in 9 groups of ten students. To analyze the data the method used to code and categories focus group data were adapted from approaches to qualitative data analysis.
Results
Four themes emerged from the focus group data. From the students' point of view," initial clinical anxiety", "theory-practice gap"," clinical supervision", professional role", were considered as important factors in clinical experience.
Conclusion
The result of this study showed that nursing students were not satisfied with the clinical component of their education. They experienced anxiety as a result of feeling incompetent and lack of professional nursing skills and knowledge to take care of various patients in the clinical setting.
==== Body
Background
Clinical experience has been always an integral part of nursing education. It prepares student nurses to be able of "doing" as well as "knowing" the clinical principles in practice. The clinical practice stimulates students to use their critical thinking skills for problem solving [1]
Awareness of the existence of stress in nursing students by nurse educators and responding to it will help to diminish student nurses experience of stress. [2]
Clinical experience is one of the most anxiety producing components of the nursing program which has been identified by nursing students. In a descriptive correlational study by Beck and Srivastava 94 second, third and fourth year nursing students reported that clinical experience was the most stressful part of the nursing program[3]. Lack of clinical experience, unfamiliar areas, difficult patients, fear of making mistakes and being evaluated by faculty members were expressed by the students as anxiety-producing situations in their initial clinical experience. In study done by Hart and Rotem stressful events for nursing students during clinical practice have been studied. They found that the initial clinical experience was the most anxiety producing part of their clinical experience [4]. The sources of stress during clinical practice have been studied by many researchers [5-10] and [11].
The researcher came to realize that nursing students have a great deal of anxiety when they begin their clinical practice in the second year. It is hoped that an investigation of the student's view on their clinical experience can help to develop an effective clinical teaching strategy in nursing education.
Methods
A focus group design was used to investigate the nursing student's view about the clinical practice. Focus group involves organized discussion with a selected group of individuals to gain information about their views and experiences of a topic and is particularly suited for obtaining several perspectives about the same topic. Focus groups are widely used as a data collection technique. The purpose of using focus group is to obtain information of a qualitative nature from a predetermined and limited number of people [12,13].
Using focus group in qualitative research concentrates on words and observations to express reality and attempts to describe people in natural situations [14].
The group interview is essentially a qualitative data gathering technique [13]. It can be used at any point in a research program and one of the common uses of it is to obtain general background information about a topic of interest [14].
Focus groups interviews are essential in the evaluation process as part of a need assessment, during a program, at the end of the program or months after the completion of a program to gather perceptions on the outcome of that program [15,16]. Kruegger (1988) stated focus group data can be used before, during and after programs in order to provide valuable data for decision making [12].
The participants from which the sample was drawn consisted of 90 baccalaureate nursing students from two hundred nursing students (30 students from the second year and 30 from the third and 30 from the fourth year) at Shiraz University of Medical Sciences (Faculty of Nursing and Midwifery). The second year nursing students already started their clinical experience. They were arranged in nine groups of ten students. Initially, the topics developed included 9 open-ended questions that were related to their nursing clinical experience. The topics were used to stimulate discussion.
The following topics were used to stimulate discussion regarding clinical experience in the focus groups.
1. How do you feel about being a student in nursing education?
2. How do you feel about nursing in general?
3. Is there any thing about the clinical field that might cause you to feel anxious about it?
4. Would you like to talk about those clinical experiences which you found most anxiety producing?
5. Which clinical experiences did you find enjoyable?
6. What are the best and worst things do you think can happen during the clinical experience?
7. What do nursing students worry about regarding clinical experiences?
8. How do you think clinical experiences can be improved?
9. What is your expectation of clinical experiences?
The first two questions were general questions which were used as ice breakers to stimulate discussion and put participants at ease encouraging them to interact in a normal manner with the facilitator.
Data analysis
The following steps were undertaken in the focus group data analysis.
1. Immediate debriefing after each focus group with the observer and debriefing notes were made. Debriefing notes included comments about the focus group process and the significance of data
2. Listening to the tape and transcribing the content of the tape
3. Checking the content of the tape with the observer noting and considering any non-verbal behavior. The benefit of transcription and checking the contents with the observer was in picking up the following:
a. Parts of words
b. Non-verbal communication, gestures and behavior...
The researcher facilitated the groups. The observer was a public health graduate who attended all focus groups and helped the researcher by taking notes and observing students' on non-verbal behavior during the focus group sessions. Observer was not known to students and researcher
The methods used to code and categorise focus group data were adapted from approaches to qualitative content analysis discussed by Graneheim and Lundman [17] and focus group data analysis by Stewart and Shamdasani [14] For coding the transcript it was necessary to go through the transcripts line by line and paragraph by paragraph, looking for significant statements and codes according to the topics addressed. The researcher compared the various codes based on differences and similarities and sorted into categories and finally the categories was formulated into a 4 themes.
The researcher was guided to use and three levels of coding [17,18]. Three levels of coding selected as appropriate for coding the data.
Level 1 coding examined the data line by line and making codes which were taken from the language of the subjects who attended the focus groups.
Level 2 coding which is a comparing of coded data with other data and the creation of categories. Categories are simply coded data that seem to cluster together and may result from condensing of level 1 code [17,19].
Level 3 coding which describes the Basic Social Psychological Process which is the title given to the central themes that emerge from the categories.
Table 1 shows the three level codes for one of the theme
Table 1 Examples of 3 levels of coding
Level 1 codes (Meaning unit) Level 2 codes (categories) Level 3 codes (theme)
• Lack of confidence * Fear of failure
• Lack of knowledge *Feeling incompetent Initial clinical
• Lack of confidence in the first day *Feeling under pressure Anxiety
*Fear of facing the procedure
• Being anxious about Starting clinical practice
• Fear of hospital environment
• First week anxiety
• Fear of unknown in the first day
The documents were submitted to two assessors for validation. This action provides an opportunity to determine the reliability of the coding [14,15]. Following a review of the codes and categories there was agreement on the classification.
Ethical considerations
The study was conducted after approval has been obtained from Shiraz university vice-chancellor for research and in addition permission to conduct the study was obtained from Dean of the Faculty of Nursing and Midwifery. All participants were informed of the objective and design of the study and a written consent received from the participants for interviews and they were free to leave focus group if they wish.
Results
Most of the students were females (%94) and single (% 86) with age between 18–25.
The qualitative analysis led to the emergence of the four themes from the focus group data. From the students' point of view," initial clinical anxiety", "theory-practice gap", clinical supervision"," professional role", was considered as important factors in clinical experience.
Initial clinical anxiety
This theme emerged from all focus group discussion where students described the difficulties experienced at the beginning of placement. Almost all of the students had identified feeling anxious in their initial clinical placement. Worrying about giving the wrong information to the patient was one of the issues brought up by students.
One of the students said:
On the first day I was so anxious about giving the wrong information to the patient. I remember one of the patients asked me what my diagnosis is. ' I said 'I do not know', she said 'you do not know? How can you look after me if you do not know what my diagnosis is?'
From all the focus group sessions, the students stated that the first month of their training in clinical placement was anxiety producing for them.
One of the students expressed:
The most stressful situation is when we make the next step. I mean...clinical placement and we don't have enough clinical experience to accomplish the task, and do our nursing duties.
Almost all of the fourth year students in the focus group sessions felt that their stress reduced as their training and experience progressed.
Another cause of student's anxiety in initial clinical experience was the students' concern about the possibility of harming a patient through their lack of knowledge in the second year.
One of the students reported:
In the first day of clinical placement two patients were assigned to me. One of them had IV fluid. When I introduced myself to her, I noticed her IV was running out. I was really scared and I did not know what to do and I called my instructor.
Fear of failure and making mistakes concerning nursing procedures was expressed by another student. She said:
I was so anxious when I had to change the colostomy dressing of my 24 years old patient. It took me 45 minutes to change the dressing. I went ten times to the clinic to bring the stuff. My heart rate was increasing and my hand was shaking. I was very embarrassed in front of my patient and instructor. I will never forget that day.
Sellek researched anxiety-creating incidents for nursing students. He suggested that the ward is the best place to learn but very few of the learner's needs are met in this setting. Incidents such as evaluation by others on initial clinical experience and total patient care, as well as interpersonal relations with staff, quality of care and procedures are anxiety producing [11].
Theory-practice gap
The category theory-practice gap emerged from all focus discussion where almost every student in the focus group sessions described in some way the lack of integration of theory into clinical practice.
I have learnt so many things in the class, but there is not much more chance to do them in actual settings.
Another student mentioned:
When I just learned theory for example about a disease such as diabetic mellitus and then I go on the ward and see the real patient with diabetic mellitus, I relate it back to what I learned in class and that way it will remain in my mind. It is not happen sometimes.
The literature suggests that there is a gap between theory and practice. It has been identified by Allmark and Tolly [20,21]. The development of practice theory, theory which is developed from practice, for practice, is one way of reducing the theory-practice gap [21]. Rolfe suggests that by reconsidering the relationship between theory and practise the gap can be closed. He suggests facilitating reflection on the realities of clinical life by nursing theorists will reduce the theory-practice gap. The theory- practice gap is felt most acutely by student nurses. They find themselves torn between the demands of their tutor and practising nurses in real clinical situations. They were faced with different real clinical situations and are unable to generalise from what they learnt in theory [22].
Clinical supervision
Clinical supervision is recognised as a developmental opportunity to develop clinical leadership. Working with the practitioners through the milieu of clinical supervision is a powerful way of enabling them to realize desirable practice [23]. Clinical nursing supervision is an ongoing systematic process that encourages and supports improved professional practice. According to Berggren and Severinsson the clinical nurse supervisors' ethical value system is involved in her/his process of decision making. [24,25]
Clinical Supervision by Head Nurse (Nursing Unit Manager) and Staff Nurses was another issue discussed by the students in the focus group sessions. One of the students said:
Sometimes we are taught mostly by the Head Nurse or other Nursing staff. The ward staff are not concerned about what students learn, they are busy with their duties and they are unable to have both an educational and a service role
Another student added:
Some of the nursing staff have good interaction with nursing students and they are interested in helping students in the clinical placement but they are not aware of the skills and strategies which are necessary in clinical education and are not prepared for their role to act as an instructor in the clinical placement
The students mostly mentioned their instructor's role as an evaluative person. The majority of students had the perception that their instructors have a more evaluative role than a teaching role.
The literature suggests that the clinical nurse supervisors should expressed their existence as a role model for the supervisees [24]
Professional role
One view that was frequently expressed by student nurses in the focus group sessions was that students often thought that their work was 'not really professional nursing' they were confused by what they had learned in the faculty and what in reality was expected of them in practice.
We just do basic nursing care, very basic. ...You know...giving bed baths, keeping patients clean and making their beds. Anyone can do it. We spend four years studying nursing but we do not feel we are doing a professional job.
The role of the professional nurse and nursing auxiliaries was another issue discussed by one of the students:
The role of auxiliaries such as registered practical nurse and Nurses Aids are the same as the role of the professional nurse. We spend four years and we have learned that nursing is a professional job and it requires training and skills and knowledge, but when we see that Nurses Aids are doing the same things, it can not be considered a professional job.
Discussion
The result of student's views toward clinical experience showed that they were not satisfied with the clinical component of their education. Four themes of concern for students were 'initial clinical anxiety', 'theory-practice gap', 'clinical supervision', and 'professional role'.
The nursing students clearly identified that the initial clinical experience is very stressful for them. Students in the second year experienced more anxiety compared with third and fourth year students. This was similar to the finding of Bell and Ruth who found that nursing students have a higher level of anxiety in second year [26,27]. Neary identified three main categories of concern for students which are the fear of doing harm to patients, the sense of not belonging to the nursing team and of not being fully competent on registration [28] which are similar to what our students mentioned in the focus group discussions. Jinks and Patmon also found that students felt they had an insufficiency in clinical skills upon completion of pre-registration program [29].
Initial clinical experience was the most anxiety producing part of student clinical experience. In this study fear of making mistake (fear of failure) and being evaluated by faculty members were expressed by the students as anxiety-producing situations in their initial clinical experience. This finding is supported by Hart and Rotem [4] and Stephens [30]. Developing confidence is an important component of clinical nursing practice [31]. Development of confidence should be facilitated by the process of nursing education; as a result students become competent and confident. Differences between actual and expected behaviour in the clinical placement creates conflicts in nursing students. Nursing students receive instructions which are different to what they have been taught in the classroom. Students feel anxious and this anxiety has effect on their performance [32]. The existence of theory-practice gap in nursing has been an issue of concern for many years as it has been shown to delay student learning. All the students in this study clearly demonstrated that there is a gap between theory and practice. This finding is supported by other studies such as Ferguson and Jinks [33] and Hewison and Wildman [34] and Bjork [35]. Discrepancy between theory and practice has long been a source of concern to teachers, practitioners and learners. It deeply rooted in the history of nurse education. Theory-practice gap has been recognised for over 50 years in nursing. This issue is said to have caused the movement of nurse education into higher education sector [34].
Clinical supervision was one of the main themes in this study. According to participant, instructor role in assisting student nurses to reach professional excellence is very important. In this study, the majority of students had the perception that their instructors have a more evaluative role than a teaching role. About half of the students mentioned that some of the head Nurse (Nursing Unit Manager) and Staff Nurses are very good in supervising us in the clinical area. The clinical instructor or mentors can play an important role in student nurses' self-confidence, promote role socialization, and encourage independence which leads to clinical competency [36]. A supportive and socialising role was identified by the students as the mentor's function. This finding is similar to the finding of Earnshaw [37]. According to Begat and Severinsson supporting nurses by clinical nurse specialist reported that they may have a positive effect on their perceptions of well-being and less anxiety and physical symptoms [25].
The students identified factors that influence their professional socialisation. Professional role and hierarchy of occupation were factors which were frequently expressed by the students. Self-evaluation of professional knowledge, values and skills contribute to the professional's self-concept [38]. The professional role encompasses skills, knowledge and behaviour learned through professional socialisation [39]. The acquisition of career attitudes, values and motives which are held by society are important stages in the socialisation process [40]. According to Corwin autonomy, independence, decision-making and innovation are achieved through professional self-concept 41. Lengacher (1994) discussed the importance of faculty staff in the socialisation process of students and in preparing them for reality in practice. Maintenance and/or nurturance of the student's self-esteem play an important role for facilitation of socialisation process 42.
One view that was expressed by second and third year student nurses in the focus group sessions was that students often thought that their work was 'not really professional nursing' they were confused by what they had learned in the faculty and what in reality was expected of them in practice.
Conclusion
The finding of this study and the literature support the need to rethink about the clinical skills training in nursing education. It is clear that all themes mentioned by the students play an important role in student learning and nursing education in general. There were some similarities between the results of this study with other reported studies and confirmed that some of the factors are universal in nursing education. Nursing students expressed their views and mentioned their worry about the initial clinical anxiety, theory-practice gap, professional role and clinical supervision. They mentioned that integration of both theory and practice with good clinical supervision enabling them to feel that they are enough competent to take care of the patients. The result of this study would help us as educators to design strategies for more effective clinical teaching. The results of this study should be considered by nursing education and nursing practice professionals. Faculties of nursing need to be concerned about solving student problems in education and clinical practice. The findings support the need for Faculty of Nursing to plan nursing curriculum in a way that nursing students be involved actively in their education.
Competing interests
The author(s) declare that they no competing interests.
Authors' contributions
FSH: Initiation and design of the research, focus groups conduction, data collection, analysis and writing the paper, SM: Editorial revision of paper
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The author would like to thank the student nurses who participated in this study for their valuable contribution
==== Refs
Dunn SV Burnett P The development of a clinical learning environment scale Journal of Advanced Nursing 1995 22 1166 1173 8675872
Lindop E Factors associated with student and pupil nurse wastage Journal of Advanced Nursing 1987 12 751 756 3693733
Beck D Srivastava R Perceived level and source of stress in baccalaureate nursing students Journal of Nursing Education 1991 30 127 132 1850792
Hart G Rotem A The best and the worst: Students' experience of clinical education The Australian Journal of Advanced Nursing 1994 11 26 33 7980880
Sheila Sh Huey-Shyon L Shiowli H Perceived stress and physio-psycho-social status of nursing students during their initial period of clinical practice International Journal of Nursing Studies 2002 39 165 175 11755447 10.1016/S0020-7489(01)00016-5
Johnson J Reducing distress in first level and student nurses Journal of Advanced Nursing 2000 32 66 74 10886436 10.1046/j.1365-2648.2000.01421.x
Admi H Nursing students' stress during the initial clinical experience Journal of Nursing Education 1997 36 323 327 9309567
Blainey GC Anxiety in the undergraduate medical-surgical clinical student Journal of Nursing Education 1980 19 33 36 6253420
Wong J Wong S Towards effective clinical teaching in nursing Journal of Advanced Nursing 1987 12 505 513 3655138
Windsor A Nursing students' perceptions of clinical experience Journal of Nursing Education 1987 26 150 154 3035128
Sellek T Satisfying and anxiety creating incidents for nursing students Nursing Times 1982 78 137 140 6925180
Krueger RA Focus Groups: A Practical Guide for Applied Research. Sage Publications: California 1988
Denzin NK The Research Act 1989 3 Prentice Hall: Englewood Cliffs, New Jersey
Stewart DW Shamdasani PN Shamdasani PN Analysing focus group data Focus Groups: Theory and Practice 1990 Sage Publications: Newbury Park
Barbour RS Kitzinger J Developing focus group research : politics, theory and practice Sage 1999
Patton MQ Qualitative Evaluation and Research Methods 1990 2 Sage publications
Graneheim UH Lundman B Qualitative content analysis in nursing research: concepts, procedures and measures to achieve trustworthiness Nurse Education Today 2004 24 105 112 14769454 10.1016/j.nedt.2003.10.001
Streubert HJ Carpenter DR Qualitative Research in Nursing Advancing the Humanistic Imperative 1995 J.B. Lippincott Company: Philadelphia
Polit DF Hungler BP Nursing research: Principles and Methods Philadelphia newyork 1999
Allmark PA classical view of the theory-practice gap in nursing Journal of Advanced Nursing 1995 22 18 23 7560527 10.1046/j.1365-2648.1995.22010018.x
Tolley KA Theory from practice for practice: Is this a reality? Journal of Advanced Nursing 1995 21 184 190 7897072 10.1046/j.1365-2648.1995.21010184.x
Rolfe G Listening to students: Course evaluation as action research Nurse Education Today 1994 14 223 227 7968967 10.1016/0260-6917(94)90085-X
Johns Ch clinical supervision as a model for clinical leadership Journal of Nursing Management 2003 11 25 34 12472865 10.1046/j.1365-2834.2002.00288.x
Berggren I Severinsson E Nurses supervisors'action in relation to their decision-making style and ethical approach to clinical supervision Journal of Advanced Nursing 2003 41 615 622 12622870 10.1046/j.1365-2648.2003.02573.x
Begat I Severinsson E Nurses' satisfaction with their work environment and the outcomes of clinical nursing supervision on nurses' experiences of well-being Journal of Nursing Management 2005 13 221 230 15819834 10.1111/j.1365-2834.2004.00527.x
Bell P Anxiety in mature age and higher school certificate entry student nurses – A comparison of effects on performance Journal of Australian Congress of Mental Health Nurses 1984 4/5 13 21
Ruth L Experiencing before and throughout the nursing career Journal of Advanced Nursing 2002 39 119 12100655 10.1046/j.1365-2648.2000.02251.x
Neary M Project 2000 students' survival kit: a return to the practical room Nurse Education Today 1997 17 46 52 9087021 10.1016/S0260-6917(97)80078-0
Jinks A Pateman B Nither this nor that: The stigma of being an undergraduate nurse Nursing Times 1998 2 12 13
Stephen RL Imagery: A treatment for nursing student anxiety Journal of Nursing Education 1992 31 314 319 1326031
Grundy SE The confidence scale Nurse Educator 1993 18 6 9 8487987
Copeland L Developing student confidence Nurse Educator 1990 15 7 2308705
Ferguson K Jinks A Integrating what is taught with what is practised in the nursing curriculum: A multi-dimensional model Journal of Advanced Nursing 1994 20 687 695 7822604 10.1046/j.1365-2648.1994.20040687.x
Hewison A Wildman S The theory-practice gap in nursing: A new dimension Journal of Advanced Nursing 1996 24 754 761 8894893 10.1046/j.1365-2648.1996.25214.x
Bjork T Neglected conflicts in the discipline of nursing: Perceptions of the importance and value of practical skill Journal of Advanced Nursing 1995 22 6 12 7560537 10.1046/j.1365-2648.1995.22010006.x
Busen N Mentoring in advanced practice nursing Journal of Advanced Nursing Practice 1999 2 2
Earnshaw GP Mentorship: The students' view Nurse Education Today 1995 15 274 279 7565519 10.1016/S0260-6917(95)80130-8
Kelly B The professional self-concepts of nursing undergraduates and their perceptions of influential forces Journal of Nursing Education 1992 31 121 125 1313087
Lynn MR McCain NL Boss BJ Socialization of R.N. to B.S.N Image: Journal of Nursing Scholarship 1989 21 232 237
Klein SM Ritti RR Understanding Organisational Behaviour 1980 Kent: Boston
Corwin RG The professional employee: A study of conflict in nursing roles The American Journal of Sociology 1961 66 604 615 10.1086/223010
Lengacher CA Effects of professional development seminars on role conception, role deprivation, and self-esteem of generic baccalaureate students Nursing Connections 1994 7 21 34 8202168
|
16280087
|
PMC1298307
|
CC BY
|
2021-01-04 16:30:31
|
no
|
BMC Nurs. 2005 Nov 9; 4:6
|
utf-8
|
BMC Nurs
| 2,005 |
10.1186/1472-6955-4-6
|
oa_comm
|
==== Front
BMC Pregnancy ChildbirthBMC Pregnancy and Childbirth1471-2393BioMed Central London 1471-2393-5-141625964110.1186/1471-2393-5-14Research ArticleDetermination of Interleukin-6 and Tumor Necrosis Factor-alpha concentrations in Iranian-Khorasanian patients with preeclampsia Afshari J Tavakkol [email protected] N [email protected] A [email protected] MT [email protected] MA [email protected] E [email protected] R [email protected] M [email protected] Immunogenetics department, Immunology Research Group, Bu-Ali Research Institute, Bu-Ali Sq., Mashhad University of Medical Sciences (MUMS), Mashhad, Iran2 Obstetrics & Gynecology department, Imam Reza Hospital, Imam Reza Sq., Mashhad University of Medical Sciences (MUMS), Mashhad, Iran3 Community medicine department, Medical school, Daneshgah St., Mashhad University of Medical Science (MUMS), Mashhad, Iran4 Department of Cardiology, Ghaem Hospital, Ahmad Abad St., Mashhad University of Medical Sciences (MUMS), Mashhad, Iran2005 1 11 2005 5 14 14 25 2 2005 1 11 2005 Copyright © 2005 Afshari et al; licensee BioMed Central Ltd.2005Afshari 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 objective was to determine the role of Interleukin-6 (IL-6) and Tumor Necrosis Factor-alpha (TNF-alpha), markers of immune activation and endothelial dysfunction, in patients with preeclampsia.
Methods
Twenty four women with preeclampsia and eighteen antepartum normotensive pregnant women were recruited as controls. Serum levels of IL-6 and TNF-alpha were measured by enzyme-linked immunosorbent assay. We used independent-samples t test to assess the differences in the concentration of cytokines in preeclamptic patients and control subjects.
Results
IL-6 levels [mean (S.D.)] were significantly higher in preeclamptic women [5.8 (4.85) pg/ml] compared to normal pregnant women [3.01 (2.45) pg/ml] (p = 0.02). There was no significant change in concentration of TNF-alpha in preeclamptic women [53.8 (30.0) pg/ml] compared to normal pregnant women [51.9 (33.8) pg/ml] (p > 0.1).
Conclusion
The results of this study show that IL-6 as a pro-inflammatory cytokine is present in higher concentration in women with preeclampsia. The study was undertaken in women with established preeclampsia and it is not possible to determine whether the increased concentration of IL-6 is a cause or consequence of the disease. Furthermore, these findings suggest that serum TNF-alpha level is not associated with preeclampsia.
==== Body
Background
Preeclampsia is a critically important disease of pregnancy, one of the major causes of fetal and maternal morbidity and mortality throughout the world. In spite of its importance for public health, the etiology of preeclampsia has not yet been fully elucidated. Although the pathophysiology of preeclampsia is not understood completely, there is an interest in a possible link between inflammation and endothelial dysfunction [1] and between endothelial dysfunction and preeclampsia [2]. Cytokines have been described to play a major role in the pathogenesis of preeclampsia. Recent studies have demonstrated that cytokines – mediators of inflammatory response-may cause endothelial dysfunction through different mechanisms such as oxidative stress [3] and endothelial cell damage [4].
IL-6 is a proinflammatory cytokine produced by mononuclear phagocytes, endothelial cells, fibroblasts and T cells and has many functions and effects [5]. IL-6 is involved in immune activation, vascular wall function and modulation of TNF- production.
Recent studies have shown that amniotic fluid IL-6 is decreased in pregnancies complicated by preeclampsia and placental IL-6 production is decreased in these patients. These findings suggest a role for this cytokine in the pathophysiology of this disease. TNF-α is a potent modulator of immune and inflammatory responses that are produced by macrophages, lymphocytes and trophoblasts and contribute to the trophoblast growth and invasion. Its role in different pathologic conditions of pregnancy has been shown in recent studies. Increased amniotic fluid concentrations of TNF-α in patients with severe preeclampsia suggest a role for this cytokine in the pathophysiology of this disease. There are some studies in different populations that demonstrated the possible role of these cytokines in the pathophysiology of preeclampsia [6-10]. Therefore in the present study we measured their concentrations in serum of Khorasanian women (northeast of Iran) with preeclampsia, and compared them with concentrations found in normotensive control women.
Both IL-6 and TNF-α are expressed in adipose tissue [11,12] and in vitro release of TNF-α by adipocytes has been reported [13]. Among the known effects of these cytokines are inhibition of insulin signaling [14] and induction of both hypertriglyceridemia [15] and endothelial activation [16].
Methods
This study was approved by the ethical committee of the Mashhad University of Medical Sciences (MUMS) and was conducted at Imam-Reza Hospital and Bu-Ali Research Institute. It was a cross-sectional study and preeclamptic patients were selected from those admitted to Imam-Reza Hospital between February 2004 and December 2004. Written informed consent was obtained from participating women. Twenty four patients with preeclampsia at 26–41 weeks' gestation were enrolled in this study. Preeclampsia was diagnosed according to the clinical criteria defined by the "International Society for the study of Hypertension in Pregnancy" [17]. All preeclamptic patients had blood pressure of at least 140/90 mmHg and proteinuria greater than (+) as assessed by a dipstick or 300 mg or greater in a 24-hour urine sample. None of the preeclamptic patients had underlying diseases such as chronic hypertension, diabetes mellitus, renal disease or urinary infection. None of these patients were in labor at the time of sampling. Some patients were prescribed hydralazine, dexamethazone or magnesium sulphate before sampling.
Eighteen antepartum normotensive healthy women at 26–39 weeks' gestation were included as control group. Maternal venous blood samples (5 ml) were collected from patient and control groups and were centrifuged at 3500 RPM for 10 minutes. Serum specimens were stored at -70°C in aliquots to avoid possible interference with assay results due to repeated freeze-thaw cycles.
Serum IL-6 and TNF-α concentrations were determined by commercially available high sensitivity indirect sandwich enzyme-linked immunosorbent assay (Bender MedSystems, Austria). Briefly, IL-6 or TNF-α present in the samples or standard binds to anti-IL-6 or anti-TNF-α monoclonal antibody adsorbed to the microwells. A biotin-conjugated monoclonal anti-IL-6 or anti-TNF-α antibody was added and binds to IL-6 or TNF-α captured by the first antibody. Following incubation unbound biotin-conjugated anti-IL-6 or anti-TNF-α is removed during a wash step. Streptavidin-HRP was added and binds to the biotin-conjugated anti-IL-6 or anti-TNF-α; following incubation unbound Streptavidin-HRP was removed during a wash step, and substrate solution reactive with HRP was added to the wells. A colored product was formed in proportion to the amount of IL-6 or TNF-α present in the sample. The reaction was terminated by addition of acid and absorbance was measured at 450 nm. A standard curve was prepared from seven IL-6 and TNF-α standard dilutions and IL-6 or TNF-α sample concentrations determined. The detection limit for IL-6 and TNF-α were 1.4 pg/ml and 3.83 pg/ml respectively. The overall inter-assay coefficient of variation for IL-6 and TNF-α were 5.2% and 6.9% respectively. All serum IL-6 and TNF-α analyses were performed at the same time, in the same batch, and in duplicate according to manufacturer's instructions. IL-6 and TNF-α serum levels are given as normal distribution. Independent-samples t test was used to compare the mean levels between patient and control groups. P-value < 0.05 were considered statistically significant.
Results
A total of 24 women with preeclampsia and 18 women as controls were examined. The characteristics of women included in the study are presented in Table 1. There is no significance difference between two groups. Therefore the two groups were match for age, gestational age and gravity. IL-6 serum concentrations are depicted in Figure 1. IL-6 serum levels [mean (S.D.)] were significantly higher (p = 0.02) in preeclamptic women [5.8 (4.85)] compared with normal pregnant women [3.01 (2.45)].
Figure 1 Serum IL-6 concentration (pg/ml) measured by ELISA in normal pregnant (n-18) and preeclamptic (n = 24) women.
TNF-α serum concentrations are depicted in figure 2. There was no significant Change in concentration of TNF-α serum levels in preeclamptic women [53.8 (30.0) pg/ml] compared to normal pregnant women [51.9 (33.8) pg/ml] (p > 0.1).
Figure 2 Serum TNF-alpha concentration (pg/ml) measured by ELISA in normal pregnant (n-18) and preeclamptic (n = 24) women.
Table 1 Normal pregnant (n = 18) Preeclampsia (n = 24) t and p
Age (years) 27.2 (5.8) 28.4 (4.9) 0.76, p > 0.1
Gestational age (weeks) 33.3 (3.7) 34.8 (4.3) 1.18, p > 0.1
Gravidity 2.3 (1.4) 2.2 (1.7) 0.65, p > 0.1
Data are presented as mean (S.D.).
Discussion
Although the pathogenesis of preeclampsia is still unknown, immunologic and inflammatory causes may play an important role. IL-6 and other cytokines are important components of immune response, and therefore can participate in the immune aspects of the pathophysiology of this disease. Proinflammatory cytokines appear to be involved in cellular events that establish and maintain pregnancy [18]; however, their role has not yet been well defined.
Our data indicate that IL-6 concentrations were increased in the circulation of Iranian-Khorasanian preeclamptic patients compared with control women. This finding is consistent with reports by Greer et al in 1994 [6], Vince et al in 1995 [7], Kupferminc et al in 1996 [8], Munno et al in 1999 [9] and Teran et al in 2001 [10]. Endothelial dysfunction and increased endothelial permeability are the characteristics of the pathophysiology of preeclampsia. IL-6 may increase the permeability of endothelial cells [5] by changing the cell shape and rearrangement of intracellular actin fibers [19]. IL-6 can also reduce prostacyclin (PG I2) synthesis by inhibiting the cyclooxygenase enzyme [20]. It can increase the tromboxane A2 to prostacyclin ratio, an abnormality that happens in preeclampsia. IL-6 can also stimulate platelet-derived growth factor, a process seen in preeclampsia [5]. Oxygen free radicals can induce synthesis of IL-6 by endothelium [21]. Oxygen free radicals are implicated in the pathogenesis of preeclampsia, because they can cause endothelial damage, which leads to reduction in nitric oxide synthesis and prostaglandin balance disturbance.
TNF-α is another pro-inflammatory cytokine that its contribution in the pathogenesis of preeclampsia has been suggested in recent studies [8,10]. In healthy pregnant women, TNF-α is thought to modulate the growth and invasion of tropoblasts in maternal spiral arteries [22]. TNF-α may contributes to abnormal placental invasion [23], endothelial cell damage [4] and oxidative stress [3]. TNF-α can stimulates IL-6 production [24], since IL-6 inhibits TNF-α release [25]. In contrast to IL-6, no increased serum concentration of TNF-α was found in preeclamptic patients compared to control pregnant women. Several investigators [26-28] have reported that serum concentrations of TNF-α were significantly higher in the first and second trimester among pregnant women who subsequently developed preeclampsia compared to those in the control group. This finding is consistent with reports by Greer et al in1994 [6], Heyl et al in 1999 [28] and Ellis et al in 2001 [29]. Furthermore, the levels of IL-6 and TNF-α that we detected in the Khorsanian women studied were significantly higher than those reported for European and North American women [7,26,27]. Whether these differences are related to genetic (inflammatory response and L-arginine; NO pathway) and/or environmental factors (e.g., infection) remains to be determined. Several evidence links infection and inflammatory processes with preeclampsia [30,31]. The role of infection in the pathogenesis of preeclampsia is particularly relevant in developing countries, where the high incidence of chronic subclinical infection may contribute to the high incidence of preeclampsia.
Our findings support the hypothesis that immune activation is involved in preeclampsia and that IL-6 may participate in the abnormal immune response. This study was undertaken in women with established preeclampsia therefore, it cannot be determined whether the increase IL-6 was a cause or a consequence of the disease [7,10,26,27]. We cannot determine whether IL-6 is an active mediator in preeclampsia or a marker of immune activation, and it seems necessary to perform further studies, including longitudinal studies before the onset of preeclampsia, to elucidate the role of this cytokine in the pathogenesis of preeclampsia.
Conclusion
The results of this study show that IL-6 as a pro-inflammatory cytokine is present in higher concentration in women with preeclampsia. The study was undertaken in women with established preeclampsia and it is not possible to determine whether the increased concentration of IL-6 is a cause or consequence of the disease. Furthermore, these findings suggest that serum TNF-α level is not associated with preeclampsia.
Authors' contributions
JTA has participated in preparation and submission of the manuscript and given final approval of the version to be published. Also, conceived of the study, and participated in its design and coordination. NG participated in the diagnosis of patients with preeclampsia and in coordination of the study. AS participated in the diagnosis and selection of patients with preeclampsia. MTS participated in the design of the study and performed the statistical analysis. MAF carried out the immunoassays. MM participated in preparation of samples and performing ELISA for IL-6. RK: participated in drafting (English correction). ME participated in revising the manuscript critically for important intellectual content. All authors read and approved the final manuscript.
Note
Table 1: Characteristics of women included in the study
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This study was supported by the Vice Chancellor for Research, Mashhad University of Medical Sciences (MUMS), Mashhad, Iran.
==== Refs
Vallance P Collier J Bhagat K Infection, inflammation and infarction: does acute endothelial dysfunction provide a link? Lancet 1997 349 1391 1392 9149715 10.1016/S0140-6736(96)09424-X
Roberts JM Taylor RN Musci TJ Rodgers GM Hubel CA McLaughlin MK Preeclampsia: an endothelial cell disorder Am J Obstet Gynecol 1989 161 1200 1204 2589440
Stark JM Preeclampsia and cytokine induced oxidative stress Br J Obstet Gynecol 1993 100 105 109
Nawroth PP Stern DM Modulation of endothelial cell hemostatic properties by tumor necrosis factor J Exp Med 1986 163 740 745 3753996 10.1084/jem.163.3.740
Akira S Taga T Kishimoto T Interleukin-6 in biology and medicine Adv Immunol 1993 54 1 78 8379461
Greer IA Lyall F Perera T Boswell F Macara LM Increased concentrations of cytokines interleukin-6 and interleukin-1 receptor antagonist in plasma of women with preeclampsia: a mechanism for endothelial dysfunction? Obstet Gynecol 1994 84 937 940 7526306
Vince GS Starkey PM Austgulen R Kwiatkowski D Redman CWG Interleukin-6, tumor necrosis factor and soluble tumor necrosis factor receptors in women with preeclampsia Br J Obstet Gynecol 1995 102 20 25
Kupferminc MJ Peaceman AM Aderka D Wallach D Socol ML Soluble tumor necrosis factor receptors and interleukin-6 levels in patients with severe preeclampsia Obstet Gynecol 1996 88 420 427 8752252 10.1016/0029-7844(96)00179-2
Munno I Chiechi LM Lacedra G Berardesca C Patimo C Marcuccio L Nardelli P Loizzi P Evaluation of nonspecific immunity and plasma levels of interferon-gamma, interleukin-6 and tumor necrosis factor-alpha in preeclampsia Immunopharmacol Immunotoxicol 1999 21 551 564 10466079
Teran E Escudero C Moya W Flores M Vallance P Lopez-Jaramillo P Elevated C-reactive protein and pro-inflammatory cytokines in Andean women with preeclampsia Int J Gynecol Obstet 2001 75 243 9 10.1016/S0020-7292(01)00499-4
Hotamisligil GS Arner P Caro JF Atkinson RL Spiegelman BM Increased adipose tissue expression of tumor necrosis factor-α in human obesity and insulin resistance J Clin Invest 1995 95 2409 2415 7738205
Purohit A Ghilchik MW Duncan L Wang DY Singh A Walker MM Reed MJ Aromatase activity and interleukin-6 production by normal and malignant breast tissue J Clin Endocrinol 1995 9 61 72
Kern PA Saghizadeh M Ong JM Bosch RJ Deem R Simsolo RB The expression of tumor necrosis factor in human adipose tissue; Regulation by obesity, weight loss, and relationship to lipoprotein lipase J Clin Invest 1995 95 2111 2119 7738178
Hotamisligil GS Budavari A Murray DL Spiegelman BM Reduced tyrosine kinase activity of the insulin receptor in obesity-diabetes: central role of tumor necrosis factor-α J Clin Invest 1994 94 1543 1549 7523453
Hardardóttir I Grünfeld C Feingold KR Effects of endotoxin and cytokines on lipid metabolism Curr Opin Lipidol 1994 5 207 215 7952915
van der Poll T van Deventer SJH Pasterkamp G van Mourik JA Büller HR ten Cate JW Tumour necrosis factor induces von Willebrand factor release in healthy humans Thromb Haemost 1992 67 623 626 1509401
Davey DA MacGillivray I The classification and definition of the hypertensive disorders of pregnancy Am J Obstel Gynecol 1988 158 892 898
Terranova PF Hunter VJ Roby KF Hunt J Tumor necrosis factor-alpha in female reproductive tract Proc Society Exp Biol Med 1995 209 325 342
Maruo N Morita I Shirao M Murota S IL-6 increases endothelial permeability in vitro Endocrinology 1992 131 710 714 1639018 10.1210/en.131.2.710
Maruo N Morita I Ishizaki Y Murota S Inhibitory effects of interleukin-6 on prostaglandin I2 production in cultured bovine vascular endothelial cells Arch Biochem Biophys 1992 292 600 604 1731622 10.1016/0003-9861(92)90037-W
Ala Y Palluy O Favero J Bonne C Modat G Dornand J Hypoxia/reoxygenation stimulates endothelial cells to promote interleukin-1 and interleukin-6 production. Effects of free radical scavengers Agents Actions 1992 37 134 9 1456174 10.1007/BF01987902
Kupferminc MJ Peaceman AM Wington TR Tamura RK Rehnberg KA Socol ML Immunoreactive tumor necrosis factor-α is elevated in maternal plasma but undetected in amniotic fluid in the second trimester Am J Obstel and Gynecol 1994 171 976 979
Hunt JS Soares MJ Lei MG Smith RN Wheaton D Atheron RA Products of lipopolysaccharide-activated macrophages (tumor necrosis factor, transforming growth factor) but not lipopolysaccharide modify DNA synthesis by rat trophoblast cells exhibiting the 80-kDa lipopolysaccharide-binding protein J Immunol 1989 143 1606 13 2474606
Shalaby MR Waage A Aarden L Espevik T Endotoxin, tumor necrosis factor-alpha/cachectin and interleukin-1 induce interleukin-6 production in vivo Clin Immunol Immunopathol 1989 53 488 98 2805453 10.1016/0090-1229(89)90010-X
Aderka D Le JM Vilcek J IL-6 inhibits lipopolysaccharide-induced tumor necrosis factor production in cultured human monocytes, U937 cells, and in mice J Immunol 1989 143 3517 23 2584704
Williams M Farrand A Mittendorf R Sorensen T Zingheim R Reily CO King I Zebelman A Luthy D Maternal second trimester serum tumor necrosis factor-alpha-soluble receptor p55 (sTNFp55) and subsequent risk of preeclampsia Am J Epidemiol 1999 149 323 329 10025474
Sacks GP Studena K Sargent IL Redman CWG Normal pregnancy and preeclampsia both produce inflammatory changes in peripheral blood leukocytes akin those of sepsis Am J Obstel Gynecol 1998 179 80 86
Heyl W Handt S Reister F Gehlen J Elevated soluble adhesion molecules in women with preeclampsia. Do cytokines like tumor necrosis factor-alpha and interleukin-1beta cause endothelial activation Eur J Obstet Gynecol Repord Biol 1999 86 35 41 10.1016/S0301-2115(99)00042-1
Ellis J Wennerholm U Bengtsson A Lilja H Levels of dimethylarginines and cytokines in mild and severe preeclampsia Acta Obstetricia Et Gynecologica Scandinavica 2001 80 602 608 11437716 10.1034/j.1600-0412.2001.800703.x
Hill JA Devoe LD Bryans CI Jr Frequency of asymptomatic bacteriuria in preeclampsia Obstel Gynecol 1986 67 529 532
Hsu CD Witter FR Urogenital infection in preeclampsia Int J Gynecol Obstel 1995 49 271 275 10.1016/0020-7292(95)02373-K
|
16259641
|
PMC1298308
|
CC BY
|
2021-01-04 16:32:05
|
no
|
BMC Pregnancy Childbirth. 2005 Nov 1; 5:14
|
utf-8
|
BMC Pregnancy Childbirth
| 2,005 |
10.1186/1471-2393-5-14
|
oa_comm
|
==== Front
BMC Public HealthBMC Public Health1471-2458BioMed Central London 1471-2458-5-1151626644010.1186/1471-2458-5-115Research ArticleHomeopathic medical practice: Long-term results of a cohort study with 3981 patients Witt Claudia M [email protected]üdtke Rainer [email protected] Roland [email protected] Stefan N [email protected] Institute for Social Medicine, Epidemiology and Health Economics, Charité University Medical Center, D-10098 Berlin, Germany2 Karl und Veronica Carstens Foundation, Am Deimelsberg 36, D-45276 Essen, Germany2005 3 11 2005 5 115 115 9 5 2005 3 11 2005 Copyright © 2005 Witt et al; licensee BioMed Central Ltd.2005Witt 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
On the range of diagnoses, course of treatment, and long-term outcome in patients who chose to receive homeopathic medical treatment very little is known. We investigated homeopathic practice in an industrialized country under everyday conditions.
Methods
In a prospective, multicentre cohort study with 103 primary care practices with additional specialisation in homeopathy in Germany and Switzerland, data from all patients (age >1 year) consulting the physician for the first time were observed. The main outcome measures were: Patient and physician assessments (numeric rating scales from 0 to 10) and quality of life at baseline, and after 3, 12, and 24 months.
Results
A total of 3,981 patients were studied including 2,851 adults (29% men, mean age 42.5 ± 13.1 years; 71% women, 39.9 ± 12.4 years) and 1,130 children (52% boys, 6.5 ± 3.9 years; 48% girls, 7.0 ± 4.3 years). Ninety-seven percent of all diagnoses were chronic with an average duration of 8.8 ± 8 years. The most frequent diagnoses were allergic rhinitis in men, headache in women, and atopic dermatitis in children. Disease severity decreased significantly (p < 0.001) between baseline and 24 months (adults from 6.2 ± 1.7 to 3.0 ± 2.2; children from 6.1 ± 1.8 to 2.2 ± 1.9). Physicians' assessments yielded similar results. For adults and young children, major improvements were observed for quality of life, whereas no changes were seen in adolescents. Younger age and more severe disease at baseline were factors predictive of better therapeutic success.
Conclusion
Disease severity and quality of life demonstrated marked and sustained improvements following homeopathic treatment period. Our findings indicate that homeopathic medical therapy may play a beneficial role in the long-term care of patients with chronic diseases.
==== Body
Background
Homeopathy is one of the most frequently used and controversial systems of complementary and alternative medicine. It is based on the 'principle of similars', whereby highly diluted preparations of substances that cause symptoms in healthy individuals are used to stimulate healing in patients who have similar symptoms when ill [1]. When a single homeopathic remedy is selected based on a patient's total symptom picture, it is called 'classical' homeopathy [2]. According to a survey in the US [3], the proportion of patients obtaining homeopathic care has quadrupled in the last seven years. A survey in Britain [4] estimated that annual expenditures reached £34.04 million (out-of-pocket £30.74 million, NHS £3.3 million). For Germany, the country in which classical homeopathy originated, a recent survey demonstrated that approximately 10% of men and 20% of women in the general population used homeopathic medicines during the previous year [5]. General trends show a rise in the number of individuals utilising naturopathic and homeopathic therapeutic methods [6].
The General Medical Council in Germany grants an official certification in homeopathy to physicians upon successful completion of a three-year-long training programme. Approximately 4,500 physicians in Germany hold this additional certification [6]. However, with the exception of some randomised, controlled trials including patients with selected diagnoses [2,7] there is no data on the health care offered by classical homeopathic medical practices. Therefore, it is impossible to assess the state of homeopathic health care and its effectiveness. We designed this project with the goal of systematically collecting data in the area of homeopathic health care for the first time in Germany. The aim of the present study was to determine the spectrum of diagnoses and treatments, as well as the course of disease over time among patients who chose to receive homeopathic treatment.
Methods
Patients were included consecutively in this prospective, multi-centre observational study upon their first consultation with a participating physician and were followed up for a total of 24 months. Evaluations were made using standardised questionnaires. In order to provide as representative a picture of homeopathic health care as possible, patients were included in the study regardless of their diagnosis. Patients were eligible for the study if they were consulting the participating physician for the first time and were at least 1 year of age. In order to participate in the study, physicians were required to have passed certified training in classical homeopathy and at least three years of experience in its practice. A total of 187 physicians belonging to four different working groups were contacted either by post or telephone and informed about the study. Of these, 103 physicians chose to participate. Each participating physician was trained in study procedures and was subject to at least one monitoring visit during the study period. All study participants provided written, informed consent, and the study protocol was approved by the appropriate ethics review boards.
Outcome measures
For patients, we developed different questionnaires for three different age groups: 1–6 years of age, 7–16 years of age, adults (>16 years of age). All questionnaires were designed to document sociodemographic data, as well as information on prior medical history, patient symptoms and complaints, quality of life, and the use of any treatment other than homeopathy. At baseline, patients recorded the complaints that led them to consider homeopathic treatment. Independently of their physicians, patients rated the severity of their complaints on a numeric rating scale (0 = no complaints, 10 = maximum severity) [8]. All complaints listed by patients in their baseline questionnaire were transferred to their follow-up questionnaires by the study office personnel. This ensured that each baseline complaint was assessed at each subsequent follow-up. For children between 1 and 6 years of age, the KITA questionnaire [9] was used to assess general health-related quality of life. It was completed by the children's parents. Patients between 7 and 16 years of age completed the KINDL questionnaire [10,11]. In additional, parents were asked to provide the required medical information. For the adults, general health-related quality of life was assessed using the MOS SF-36 questionnaire [12]. The results of the SF-36 are presented in normalised scores, the results being scaled in such a way that the normal German population has a mean score of 0 and a standard deviation of 1.
The first questionnaire was distributed to the patients by the study physician and completed prior to the start of therapy (baseline). Patients sent their completed questionnaires to the study office in sealed envelopes. Follow-up questionnaires were sent to all patients by the study office at 3, 12, and 24 months.
For physicians, we developed a standardised questionnaire that allowed for continuous documentation during the treatment/follow-up period (24 months), as well as standardised points of assessment at 0, 3, 12 and 24 months. At each of these time points, the severity of a maximum of 4 diagnoses and maximum of 8 symptoms was rated by participating physicians using a numerical rating scale [8]. This information was then forwarded to the study office. The type of homeopathic treatment, the use of any conventional therapy, as well as any referrals to other physicians were recorded on a continuous basis.
Statistics
Data was double entered manually into an ACCESS© database and subsequently compared using the SAS© system followed by plausibility data checks if necessary. The diagnoses, documented by study physicians, were encoded in ICD-9 format and recorded by two specially trained study staff members using DIACOS.© Statistical analysis was performed using SAS/STAT© software (Version 8.2).
Data for adults (>16 years) and for children/adolescents were analysed separately. In order to calculate the average severity of the physicians' diagnoses, we took the four diagnoses named first for each patient during the baseline examination. For each of the follow-up assessment points (i.e. at 3, 12, and 24 months) we ascertained the respective severity ratings made by study physicians.
All results reported here are based on the intention-to-treat approach, i.e. each included patient entered the final analyses. If patients dropped out or withdrew from the study we replaced the respective missing values: baseline complaints that had been cured were given a severity rating of 0 in all following examinations. For patients who died during the study, we inserted the maximum severity rating of 10. Other missing values were multiply imputed following the suggestions of Rubin [13]. Instead of filling in a single value as a substitute for a missing value, multiple imputation is a strategy by which each missing value is replaced simultaneously by a set of plausible values that represents the uncertainty about the right value to impute. Thus, the missing values are filled in several times generating several distinct data tables, each with a complete set of data without any missing value. These complete data tables are analyzed separately using appropriate statistical models. Afterwards, the results from all statistical analyses are pooled to generate treatment effects and p-values. In our study we used the MCMC (Marcov chain Monte Carlo) replacement method and created 5 multiple imputed data tables.
For each imputed data set, treatment effects were estimated on the basis of generalized linear regression models. Generalized linear regression models are flexible and powerful tools to describe data from cohort studies [14]. They are generalizations of the well known and often applied multiple regression models which often appear to be too simple to describe longitudinal data adequately. A generalized linear model is best described by two components. First, the mean course of the outcome, and second, the correlation structure for measurements taken at the same individual at different times. In our study we divided the 2-year follow-up period into two parts. During the first part (months 0–3) we assumed that mean outcome increases (or decreases) linearly. For the second part (months 3–21) we assumed that the mean outcome increases (or decreases) according to a quadratic term. Moreover, we assumed that the correlation between two measurements can be described by a simple exponential function. This essentially means, that the correlation only depends from the time span between the two measurements, and it decreases the bigger this time span is. This approach is completely analogous to the recommendations given by Diggle, Liang, and Zeeger in their standard text book on the analysis of longitudinal data [14].
Subgroup analyses are based on essentially the same statistical approach adding the respective factors as a fixed covariate into the models. For subgroup analyses adults' and children's data were pooled.
Usually, patients for clinical studies are not selected randomly from a target population but according to some selection criteria that sample patients according to extreme measurements (high blood pressure, severe pain, low quality of life, ...). This inevitably leads to regression-to-the-mean, a statistical phenomenon that makes natural variation look like real changes [15]. Separating regression-to-the-mean effects from true treatment effects can be difficult but is at least feasible when the mean and the standard deviation of the target population are known. In this situation it is possible to calculate the expected outcome for each patient when regression-to-the-mean occurs [16]. In our study we made a rather conservative assumption on the target population (chronically ill patients seeking homeopathic care): to have the same quality of life as the general German population (i.e. a mean SF-36 score of 0 and a standard deviation of 1). From this we calculated the expected regression-to-the mean effect and compared it to the actually observed change of the SF-36 scores.
Results
A total of 103 physicians participated in the study (51 male, 45 ± 7 years of age; 52 female, 45 ± 7 years of age). Twenty-six percent of the participating physicians were specialists (10% internists, 9% paediatricians, 7% other) and 74% were general practitioners. The average duration of overall medical practice was 17.4 ± 8.4 years with 9.0 ± 4.4 years of practice in homeopathy (range 3–20 years). Forty percent of the physicians were certified to work in the public health care system, and 60% were in private practice.
Patients were recruited for the study between September 1997 and December 1999. Of the patients who met the inclusion criteria, 3981 (68%) chose to participate and were included in the study (for patient selection see Figure 1). Of these, 2851 were adults (71% women) and 1130 were children (48% girls). The baseline characteristics are listed in Table 1.
Figure 1 Patient selection.
Table 1 Baseline characteristics of study population
Adults Children
Gender (% female) 70.8 48.3
Age (years, mean ± std) 40.7 ± 12.7 6.7 ± 4.1
Marital status (% living in partnership) 84.0 /
Education (% attending school >10 years) 85.0 /
Belief in homeopathy (%) 65.7 68.6†
Duration of disease (years, mean ± std) 10.3 ± 9.8 4.3 ± 3.7
Intake of conventional drugs (%) 50.2 31.7
† Parents' perspective
On average, the homeopathic physicians made 2.6 ± 1.2 diagnoses per patient (2.8 ± 1.1 in adults, 2.3 ± 1.1 in children). Ninety-seven percent of all diagnoses were classified by these physicians as chronic with a median duration of 4.3 ± 2.7 years in children and 10.3 ± 9.8 years in adults. Almost all patients had received conventional treatment (95%) or had already contacted another physician (95%) prior to the start of this study. The most common diagnosis in women was migraine (9.7%), in men allergic rhinitis (10.3%), and in children of both genders atopic dermatitis (20%), for details see [17]. For the most common disease groups see Figure 2.
Figure 2 Most common medical complaints as reported by the homeopathy physicians (in % of documented complaints).
All patients underwent an initial homeopathic anamnesis, lasting an average of 2.0 ± 0.7 hours. Following enrolment in the study, patients had to wait an average of 57 ± 84 days before undergoing the initial anamnesis. During the 24-month observation period, patients consulted their physicians an average of 7.8 ± 8.4 times. During the study period, half of the patients (50.3%, adults: 50.8%, children 48.9%) noted additional visits to non-study physicians (gynaecologists and dentists excluded). The intake of conventional medication decreased from 45.0% at baseline (adults: 50.2%, children 31.7%) to 26.8% after 24 months (adults: 31.8%, children 14.2%).
According to patient assessments, disease severity decreased significantly between baseline and 12 months, as well as between 12 months and 24 months (see Table 2). According to physician assessments, 25.7% (adults: 21.9%, children: 37.6%) of the diagnoses were no longer present at 24 months, whereas patients judged 23.0% (adults 19.7%, children 32.8%) of the medical complaints to have resolved by this point. Thirteen percent of the patients documented that they had no complaints whatsoever at 24 months.
Table 2 Course of outcome parameters and estimated mean changes of outcome parameters
Estimated changes compared to baseline†
Baseline 3 months 12 months 24 month Δ 3 months Δ 12 months Δ 24 months
mean ± SD mean ± SD mean ± SD mean ± SD mean (95%CI) mean (95%CI) mean (95%CI)
Adults
Patients assessments (NRS) ‡ 6.2 ± 1.7 3.8 ± 2.2* 3.3 ± 2.1* 2.9 ± 2.2* -2.4 (-2.5 to -2.3) -2.8 (-2.9 to -2.7) -3.1 (-3.2 to -3.0)
Physicians assessments (NRS) ‡ 6.0 ± 1.6 3.9 ± 2.1* 2.8 ± 2.1* 2.1 ± 2.0* -2.1 (-2.2 to -2.0) -3.1 (-3.2 to -3.0) -3.7 (-3.8 to -3.6)
SF-36 QoL physical scale 46.5 ± 10.1 49.1 ± 9.5* 50.1 ± 9.6* 50.7 ± 9.5* 2.6 (2.3 to 2.9) 3.5 (3.0 to 3.9) 4.1 (3.5 to 4.6)
SF-36 QoL mental scale 39.3 ± 11.8 44.6 ± 10.8* 45.5 ± 10.8* 46.4 ± 10.6* 5.6 (5.2 to 6.0) 6.2 (5.7 to 6.7) 6.9 (6.3 to 7.6)
Children
Patients assessments (NRS) ‡ 6.1 ± 1.8 3.2 ± 2.3* 2.5 ± 2.1* 2.2 ± 2.0* -3.1 (-3.3 to -2.9) -3.5 (-3.7 to -3.4) -3.9 (-4.0 to -3.7)
Physicians assessments (NRS) ‡ 5.9 ± 1.7 3.2 ± 2.2* 2.0 ± 1.5* 1.5 ± 1.8* -2.7 (-2.8 to -2.6) -3.8 (-4.0 to -3.7) -4.4 (-4.6 to -4.3)
KINDL QoL 69.3 ± 13.3 72.1 ± 12.6 68.0 ± 9.2 67.3 ± 9.9* 2.7 (1.7 to 3.7) -0.4 (-1.5 to 0.8) -2.2 (-3.6 to -0.8)
KITA QoL mental/physical dimension 67.6 ± 16.9 75.4 ± 14.6* 77.0 ± 14.1* 77.5 ± 14.3* 8.3 (6.6 to 10.0) 9.3 (7.7 to 10.8) 10.0 (8.3 to 11.6)
KITA QoL aspects of daily living 58.6 ± 18.3 66.9 ± 15.9* 69.1 ± 16.7* 70.6 ± 16.0* 8.5 (7.2 to 9.8) 10.4 (8.8 to 12.0) 11.6 (9.7 to 13.5)
† estimations are based on generalised linear models, see text; ‡ = lower values indicate better status and negative Δ indicates improvement
QoL = quality of life; NRS = numeric rating scale, * p < 0.001 versus baseline
In adults, large improvements in quality of life were observed on both component scales (mental and physical) during the first three months of treatment, and continued to improve during the course of the study (see Table 2). Even with the pessimistic assumption that the test-retest correlation of the SF-36 is only 0.7 and that the study population is no more ill than a random sample of the general population, one could expect an improvement of only 3.8 (1.2) score points on the mental (physical) component scale, attributable to regression-to-the-mean [16], markedly lower than the 5.6 (2.6) score points observed in our study (Table 2). Statistically, the baseline quality of life of non-completers was not significantly lower than in other patients (p-values: MCS: p = 0.37; PCS: p = 0.48, Wilcoxon-tests).
Quality of life in young children (age 1–6 years) also improved markedly during the observation period (Table 2), having already risen during the first three months of study therapy as measured on both scales of the KITA questionnaire (mental-physical dimension and aspects of daily living, each p < 0.001, see Table 2). These improvements continued over the course of treatment (p < 0.001, see Table 2). In school children and adolescents, however, an improvement in quality of life was only visible during the first three months of study therapy (p < 0.001, see Table 2).
The diagnosis had no relevant influence on the changes in patient complaints or quality of life as measured in this investigation.
In patient and physician assessments, younger patients showed greater improvements than did older patients and more severe disease at baseline was followed by greater improvements compared to less severe disease (see Table 3). Gender, duration of disease and belief in homeopathy had only a minor influence on improvements.
Table 3 Subgroup analyses for patients and physicians assessments (mean changes of outcome parameters after 24 months compared to baseline, negative Δ indicates improvement)
Patients assessments (NRS) Physicians assessments (NRS)
Mean† 95%-CI p value* Mean† 95%-CI p value*
Total (n = 3981) -3.3 -3.4 to -3.2 -3.9 -4.0 to -3.8 0,060
Gender
Female (n = 2560) -3.4 -3.5 to -3.2 -3.9 -4.0 to -3.8
Male (n = 1412) -3.3 -3.4 to -3.1 0.387 -3.9 -4.0 to -3.8 0,060
Age groups (years)
<10 (n = 839) -4.0 -4.2 to -3.8 -4.4 -4.6 to -4.2
10–19 (n = 355) -3.5 -3.7 to -3.2 <0.001 -4.3 -4.5 to -4.0 0.149
20–39 (n = 1456) -3.4 -3.6 to -3.3 <0.001 -3.7 -3.8 to -3.6 <0.001
40–59 (n = 1041) -2.8 -3.8 to -2.0 <0.001 -3.6 -3.8 to -3.5 <0.001
≥ 60 (n = 281) -2.6 -2.9 to -2.2 <0.001 -3.5 -3.8 to -3.2 <0.001
Baseline severity of disease
NRS < 6.0 (n = 1660) -2.1 -2.3 to -2.0 -3.1 -3.2 to -3.0
NRS ≥ 6.0 (n = 2310) -4.1 -4.2 to -4.0 <0.001 -4.6 -4.7 to -4.5 <0.001
Duration of disease in adults (years)
< 10 (n = 1878) -3.2 -3.4 to -3.1 -3.7 -3.8 to -3.6
≥ 10 (n = 927) -2.9 -3.1 to -2.7 <0.001 -3.6 -3.7 to -3.4 0.043
Intake of conventional drugs at baseline
Yes (n = 1788) -3.3 -3.5 to -3.2 -3.8 -3.9 to -3.7
No (n = 2188) -3.3 -3.5 to -3.2 0.157 -3.9 -4.0 to -3.9 0.029
Belief in homeopathy
Strong (n = 2656) -3.4 -3.5 to -3.1 -3.9 -4.0 to -3.8
Weak (n = 1316) -3.1 -3.3 to -3.0 <0.001 -3.8 -3.9 to -3.7 0.563
† estimations are based on generalised linear models, NRS = numeric rating scale; * per item each subgroup compared to the first listed subgroup
Discussion
Patient and physician assessments of disease severity and quality of life consistently demonstrated substantial improvements following homeopathic treatment, which were maintained through 24 months' follow up. Improvements were more pronounced in younger patients and in those with greater disease severity compared to older patients and those with less severe disease at baseline.
To our knowledge, the present study is the first to evaluate systematically the range of diagnoses and therapies in classical homeopathic medical practices in Germany and Switzerland. In addition, the study provided information on the course of illness in patients receiving homeopathic treatment, as assessed by patients and physicians.
The methodological strengths of our study include consecutive enrolment of a large sample size, the participation of approximately 2% of all physicians certified to practice homeopathy in Germany and 28% of all members of the Hahnemann Association (an organisation for physicians practicing only 'classical' homeopathy) and the use of standardised outcome instruments also used in studies on conventional therapy.
One limitation of our study is that the observed effects cannot be categorized with respect to specificity, i.e. we cannot draw conclusions as to the beneficial mechanisms. Furthermore patients were allowed to use conventional therapies during the study period in addition to homeopathic treatment. Thus, the observed improvement cannot be attributed to homeopathic treatment alone. The aim of the investigation, however, was not to test the effectiveness of homeopathic treatment alone, but rather provide systematic and detailed information about the current status of homeopathic medical care in routine practice and its effectiveness. These data may also be helpful in the planning of further research projects on homeopathy.
The effects observed by patient and the physician assessment, as well as those seen with regard to quality of life, deserve additional comments. The average severity of the chronic diseases was reduced by approximately 50% after only 3 months of homeopathic treatment, and remained around this level during the follow-up period. Physician assessments tended to be more positive than patient assessments.
The improvements we observed in our patients cannot be attributed solely to regression-to-the-mean, because the improvements were greater than could be expected even under conservative model assumptions. This is supported by the fact that patients did not visit the study physicians when they were feeling the worst, but rather after a long waiting period.
A strength of this study is that patients with all diagnoses were included. Therefore, no disease-specific measurement instruments could be used. To assess the severity of different medical complaints, there is no other generally accepted measuring instrument available. Instead numerical rating scales [8] were applied, which would allow for the determination of illness severity in a diagnosis-independent manner.
Compared to the other quality two of life questionnaires used in our study, the KINDL questionnaire for the age group 7 to 16 years was not sensitive to change, as has been shown in other studies [18,19]. Other explanations might be that children adapt easier to perceived quality of life and that the dimensions of Quality of life used for adults are not transferable to children. However, there is no other generally accepted measuring instrument available in German-speaking countries.
In the range of baseline diagnoses, chronic illnesses clearly predominated (>95% of diagnoses). Among these, headache and atopic disease (allergic rhinitis, asthma and atopic dermatitis) were the most common diagnoses. As the clinical histories of our patients showed, most of our patients decided to consult a homeopathic physician only after having received conventional treatment. This, together with the extensive initial case taking and the reputation of homeopathy as a "medicine designed to treat the individual as a whole"' causes a selection for chronic illnesses.
We were unable to confirm the common notion that homeopathy is frequently used for trivial complaints or diseases. The duration of disease in study patients was very long and their symptoms were, on average, of moderate severity.
In this study we were not able to evaluate different types of homeopathic strategies. For quality assurance purposes, we avoided selecting a random sample of homeopathic physicians for the study, choosing instead to recruit physicians schooled and certified in 'classical' homeopathy. The results of our study are, therefore, representative only for the classical type of homeopathy that was practised by participating physicians. Compared to conventional medical practices, headache and atopic disease (allergic rhinitis, asthma and atopic dermatitis) were the most common diagnoses in homeopathic practices (as opposed to hypertension, hyperlipidemia and low back pain in 70,000 patients treated conventionally) [9]. An American study [20] found asthma, depression, otitis media, and allergic rhinitis to be the most common diagnoses treated in homeopathic practices, compared to hypertension, upper respiratory tract infection, otitis media and diabetes mellitus, which were treated most commonly in conventional practices.
A health insurance company project that included about 900 patients treated with homeopathy in routine care [21] showed an improvement in quality of life and in physician assessment. In Güthlin's study [21], however, only physicians certified to work in the public health care system were able participate. Homeopaths working in private practices (i.e. the great majority in Germany) were excluded. The advantage of the present study is that doctors in private practice were also included, thus providing a more detailed and broader basis for describing the current status of homeopathic health care. Another controlled study in cooperation with a German health insurance company [22], indicated similar overall effectiveness of homeopathically versus conventionally treated patients for selected diagnoses and in some groups, superiority of homeopathic treatment.
Conclusion
We evaluated for the first time the range of diagnoses and therapies at medical practices offering classical homeopathic treatment in Germany and Switzerland. The findings of our study demonstrate that patients who seek homeopathic treatment are primarily those suffering from long-standing, chronic disease. Both according to physician and patient assessments, the severity of complaints decreased markedly over the 24-month observation period. Younger patients and those with more severe disease appear to benefit most from homeopathic treatment. Among adults and children, we observed an increase in quality of life. Our findings indicate that homeopathic medical therapy may play a beneficial role in the long-term care of patients with chronic diseases.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
CW participated in the design of the study, coordination and statistical analysis. RL participated in its design and performed the statistical analysis. RB participated in the design of the study and data acquisition. SNW conceived of the study, and participated in its design and statistical analysis and had the overall scientific responsibility. All authors helped to draft the manuscript, read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This study was supported by a grant from the Karl and Veronica Carstens Foundation. We thank all participating physicians and patients.
==== Refs
Jonas W Jacobs J Healing With Homeopathy 1996 New York, Warner Books
Linde K Clausius N Ramirez G Melchart D Eitel F Hedges LV Jonas WB Are the clinical effects of homeopathy placebo effects? A meta-analysis of placebo-controlled trials Lancet 1997 350 834 843 9310601 10.1016/S0140-6736(97)02293-9
Eisenberg DM Davis RB Ettner SL Appel S Wilkey S Van Rompay M Kessler RC Trends in alternative medicine use in the United States, 1990-1997: results of a follow-up national survey JAMA 1998 280 1569 1575 9820257 10.1001/jama.280.18.1569
Thomas KJ Nicholl JP Coleman P Use and expenditure on complementary medicine in England: a population based survey Complement Ther Med 2001 9 2 11 11264963 10.1054/ctim.2000.0407
Härtel U Volger E Inanspruchnahme und Akzeptanz von klassischen Naturheilverfahren und alternativen Heilmethoden in Deutschland. Ergebnisse einer repräsentativen Bevölkerungsstudie Das Gesundheitswesen 2003 65 A35
G M S M T Z Gesundheitsberichtserstattung des Bundes - Inanspruchnahme alternativer Methoden in der Medizin 2002 Berlin, Robert Koch Institut
Kleijnen J Knipschild P ter Riet G Clinical trials of homoeopathy BMJ 1991 302 316 323 1825800
Huskisson EC Scott J Westhoff G VAS Visuelle Analog-Skalen; auch VAPS Visual Analogue Pain Scales, NRS Numerische Rating-Skalen; Mod. Kategorialskalen Handbuch psychosozialer Meßinstrumente - ein Kompendium für epidemiologische und klinische Forschung zu chronischer Krankheit 1993 Göttingen, Hogrefe 881 885
Wittorff M KITA Fragebogen zur Erfassung der gesundheitsbezogenen Lebensqualität von Kleinkindern zwischen 1-6 Jahren 2002
Bullinger M Mackensen S Kirchberger I KINDL - ein Fragebogen zur Erfassung der gesundheitsbezogenen Lebensqualität von Kindern Zeitschrift für Gesundheitspsychologie 1994 II 64 77
U. RS Bullinger M Assessing health-related quality of life in chronically ill children with the German KINDL: first psychometric and content analytical results Qual Life Res 1998 7 399 407 9691720 10.1023/A:1008853819715
Bullinger M Kirchberger I SF-36 Fragebogen zum Gesundheitszustand 1998 Göttingen, Hogrefe
Rubin DB Multiple Imputation for Nonresponsense in Surveys 1987 New York, John Wiley & Sons
Diggle PJ Liang KY Zeger SL Analysis of Longitudinal Data 1994 Oxford, Clarendon,Press
Barnett AG van der Pols JC Dobson AJ Regression to the mean: what it is and how to deal with it Int J Epidemiol 2005 34 215 220 15333621 10.1093/ije/dyh299
Hannan PJ Jacobs DRJ McGovern P Klepp KI Elmer P Estimating the effect of regression toward the mean under stochastic censoring Am J Epidemiol 1994 139 422 431 8109577
Becker-Witt C Lüdtke R Weisshuhn TE Willich SN Diagnoses and treatment in homeopathic medical practice Forsch Komplementarmed Klass Naturheilkd 2004 11 98 103 15138374 10.1159/000078231
Becker M Duft-Qigongunterricht für Schulkinder - Eine kontrollierte Pilotstudie zur Evaluation von Durchführbarkeit und Nutzen 2002 Humboldt-Universität zu Berlin
Keil T Becker-Witt C Vance W Roll S Wegscheider K Willich SN Vergleich von homöopathischer und konventioneller Therapie bei Kindern mit Neurodermitis Informatik, Biometrie und Epidemiologie in Medizin und Biologie 2002 33 283
Jacobs J Chapman EH Crothers D Patient characteristics and practice patterns of physicians using homeopathy Arch Fam Med 1998 7 537 540 9821827 10.1001/archfami.7.6.537
Güthlin C Lange O Walach H Measuring the effects of acupuncture and homoeopathy in general practice: An uncontrolled prospective documentation approach BMC Public Health 2004 4 6 15113434 10.1186/1471-2458-4-6
Becker-Witt C Keil T Roll S Menke D Vance W Wegscheider K Willich SN Hospital TRLH Effectiveness and costs of homeopathy compared to conventional medicine-a prospective multicenter cohort study Improving the success of homeopathy 2003 4 London, The Royal London Homeopathic Hospital 57 60
|
16266440
|
PMC1298309
|
CC BY
|
2021-01-04 16:28:58
|
no
|
BMC Public Health. 2005 Nov 3; 5:115
|
utf-8
|
BMC Public Health
| 2,005 |
10.1186/1471-2458-5-115
|
oa_comm
|
==== Front
BMC Public HealthBMC Public Health1471-2458BioMed Central London 1471-2458-5-1161626908410.1186/1471-2458-5-116Research ArticleThe contribution of leading diseases and risk factors to excess losses of healthy life in eastern Europe: burden of disease study Powles John W [email protected] Witold [email protected] Hoorn Stephen [email protected] Majid [email protected] Department of Public Health and Primary Care, Institute of Public Health, Robinson Way, Cambridge CB2 2SR, UK2 Cancer Epidemiology and Prevention Division, M Slodowska-Curie Memorial Cancer Centre, ul W.K.Roentgena 5, 02-781 Warsaw, Poland3 Clinical Trials Research Unit, University of Auckland, Auckland, New Zealand4 Department of Population and International Health, Harvard School of Public Health, 665 Huntington Avenue, Boston MA 02115, USA2005 3 11 2005 5 116 116 30 6 2005 3 11 2005 Copyright © 2005 Powles et al; licensee BioMed Central Ltd.2005Powles 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 East/West gradient in health across Europe has been described often, but not using metrics as comprehensive and comparable as those of the Global Burden of Disease 2000 and Comparative Risk Assessment studies.
Methods
Comparisons are made across 3 epidemiological subregions of the WHO region for Europe – A (very low child and adult mortality), B (low child and low adult mortality) and C (low child and high adult mortality) – with populations in 2000 of 412, 218 and 243 millions respectively, and using the following measures: 1. Probabilities of death by sex and causal group across 7 age intervals; 2. Loss of healthy life (DALYs) to diseases and injuries per thousand population; 3. Loss of healthy life (DALYs) attributable to selected risk factors across 3 age ranges.
Results
Absolute differences in mortality are most marked in males and in younger adults, and for deaths from vascular diseases and from injuries. Dominant contributions to east-west differences come from the nutritional/physiological group of risk factors (blood pressure, cholesterol concentration, body mass index, low fruit and vegetable consumption and inactivity) contributing to vascular disease and from the legal drugs – tobacco and alcohol.
Conclusion
The main requirements for reducing excess health losses in the east of Europe are: 1) favorable shifts in all amenable vascular risk factors (irrespective of their current levels) by population-wide and personal measures; 2) intensified tobacco control; 3) reduced alcohol consumption and injury control strategies (for example, for road traffic injuries). Cost effective strategies are broadly known but local institutional support for them needs strengthening.
==== Body
Background
The evolving picture of East-West disparities in health indicators across Europe have been documented in terms of life expectancy [1] and premature mortality [2] and in terms of death rates for major contributing causes [3,4]. In addition to aggregate or partial mortality measures, a comparative analysis of the level and distribution of diseases and injuries, and their risk factors, is a valuable guide to strategies for improving health and for reducing cross-population health differentials. An important aspect of such comparative analyses is the use of a consistent and comparable metric of lost healthy life and the attribution of such losses either to diseases or injuries or to the risk factors for those diseases and injuries. The Global Burden of Disease project for the year 2000 (GBD2000) [5] and the associated Comparative Risk Assessment (CRA) Project [6], use common metrics and comparable methodology to address the burden of disease and injuries and their risk factors. We use the databases (some only recently available [6]) and published results from this study, to evaluate the nature and reasons for the health disparities across Europe. The results are presented for 3 epidemiologically-defined subregions of the WHO region for Europe.
In analyzing the causes of the marked differences in health levels across Europe, we have deliberately restricted our scope to the more proximal determinants ('risk factors'), because knowledge of their role is more secure and lends itself more readily to quantitative analysis.
Methods
Populations
Following the Burden of Disease protocols, the WHO region for Europe (which extends to Israel, Turkey and the former Soviet republics of central Asia) is divided into 3 'sub-regions' on the basis of child and adult mortality levels – Europe A (very low child; very low adult mortality) with a population in 2000 of 412 millions, Europe B (low child, low adult mortality) population 218 millions and Europe C (low child and high adult mortality) with a population of 243 millions. These 'subregions' are neither contiguous nor culturally homogeneous and correspond only approximately to western Europe, Eastern Europe and the successor states to the Soviet Union. Table 1 lists the countries within each subregion.
Table 1 Global Burden of Disease sub-regions in Europe
Mortality stratum Countries
A Very low child; very low adult Andorra, Austria, Belgium, Croatia, Czech Republic, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Israel, Italy, Luxembourg, Malta, Monaco, Netherlands, Norway, Portugal, San Marino, Slovenia, Spain, Sweden, Switzerland, United Kingdom
B Low child, low adult Albania, Armenia, Azerbaijan, Bosnia and Herzegovina, Bulgaria, Georgia, Kyrgyzstan, Poland, Romania, Slovakia, Tajikistan, The Former Yugoslav Republic of Macedonia, Turkey, Turkmenistan, Uzbekistan, Yugoslavia
C Low child, high adult Belarus, Estonia, Hungary, Kazakhstan, Latvia, Lithuania, Republic of Moldova, Russian Federation, Ukraine
Mortality and burden of disease
Detailed methods for the Global Burden of Disease (GBD) 2000 project are described elsewhere [5]. In summary, for 14 epidemiological regions of the world, including 3 in Europe, GBD 2000 provides estimates of mortality and burden of disease for over 130 diseases and injuries. Mortality data are from national vital registration systems, reported annually to the World Health Organization. For countries in the European region with incomplete mortality data, or, more commonly, with substantial proportions of deaths allocated to non-specific codes [7], demographic techniques and epidemiological models are used to estimate mortality by age, sex, and cause. We calculated probabilites of death within age intervals from groups of causes using life table methods [8].
Burden of disease is expressed in disability-adjusted life years (DALYs), an aggregate measure of loss of healthy life to either premature mortality or to non-fatal illness or injury [9]. Inputs used to estimate losses of healthy life include estimates of disease incidence and/or prevalence, severity, and duration, generally from systematic reviews of disease-specific epidemiological literature or disease-specific registries (e.g. cancer registries). Flows of lost healthy life extending to future years are discounted and weighted by the age at which the lost healthy life would have been lived. The conceptual issues and sensitivity of results to these methodological details on the estimates of DALYs lost are described elsewhere [10].
Risk factors
The methods and data sources for the Comparative Risk Assessment project are described elsewhere [6,11]. In summary, the contribution of a risk factor to disease or mortality relative to some alternative exposure scenario (i.e. population attributable fraction, PAF, defined as the proportional reduction in population disease or mortality that would occur if exposure to the risk factor were reduced to an alternative exposure scenario [12-14]) is given by the generalized "potential impact fraction" in Equation 1. For each risk factor – disease pair, the population attributable fraction is then multiplied by total deaths or burden of disease to estimate risk factor attributable mortality or burden of disease.
where
RR(x): relative risk at exposure level x
P(x): population distribution of exposure
P'(x): counterfactual distribution of exposure
m: maximum exposure level
The estimates of burden of disease and injuries due to risk factors in the CRA project are based on a counterfactual exposure distribution that would (within the limits of current knowledge and data) result in the lowest population risk, irrespective of whether attainable using current interventions or policies. This is referred to as the theoretical-minimum-risk exposure distribution [15,16]. The theoretical-minimum-risk exposure distribution was zero for risk factors for which zero exposure could be defined and reflected minimum risk (e.g. no smoking). For some risk factors, zero exposure was an inappropriate choice because of physical lower limits to exposure reduction (e.g. particles in ambient air). For physiological risk factors such as blood pressure, where lower values are associated with lower risk, the lowest values reliably associated with favourable health outcomes were used to define the theoretical-minimum-risk exposure distribution. Alcohol has benefits as well as harms depending on the disease under consideration and the patterns of alcohol consumption. A counterfactual exposure distribution of zero was still the default choice for alcohol, however, in some regions (e.g. EUR-A) effects on vascular disease were estimated as negative (protective) and combined with hazardous effects on other diseases to derive an estimated net effect [17]. For details of the risk factors considered see Table 2.
Table 2 10 leading risk factors for the European region, exposure variables, theoretical minima, and contributions to total disease burden in the European region (source: Table 1 and Figure 1 in Ezzati et al. 1). See Table 1 in Ezzati et al. 1 for disease outcomes and data sources.
Risk Factor Exposure Variable Theoretical Minimum Contribution to European disease burden (%GBD)
High blood pressure Level of systolic blood pressure 115 SD 6 mmHg 12.8%
Tobacco Current levels of smoking impact ratio (indirect indicator of accumulated smoking risk based on excess lung cancer mortality); oral tobacco use prevalence No tobacco use 12.3%
Alcohol Current alcohol consumption volumes and patterns No alcohol use b 10.1%
High cholesterol Level of total blood cholesterol 3.8 SD 1 mmol/l (147 SD 39 mg/dl) 8.7%
High body mass index (BMI) Body mass index, BMI (height over weight squared) 21 SD 1 kg/m2 7.8%
Low fruit and vegetable intake Fruit and vegetable intake per day 600 g (SD 50 g) intake per day for adults 4.4%
Physical inactivity Three categories of inactive, insufficiently active (<2.5 hours per week of moderate-intensity activity, or less than 4000 KJ/week), and sufficiently active. Activity in discretionary-time, work, and transport considered All having at least 2.5 hours per week of moderate-intensity activity or equivalent (400 KJ/week) 3.5%
Illicit drugs Use of amphetamine, cocaine, heroin or other opioids and intravenous drug use No illicit drug use 1.6%
Lead Current blood lead levels 0.016 μg/dl blood lead levels c 0.8%
Unsafe sex Sex with an infected partner without any measures to prevent infection (represented as parameters of an HIV model) No unsafe sex 0.7%
a The resulting haemoglobin levels vary across regions and age-sex groups (from 11.66 g/dl in under-5 children in SEAR-D to >14.5 g/dl in adult males in developed countries) because the other risks for anaemia (e.g. malaria) vary.
b Theoretical minimum for alcohol is zero, the global theoretical minimum. Specific sub-groups may have a non-zero theoretical minimum.
c Theoretical minimum for lead is the blood lead levels expected at background exposure levels. Health effects were quantified for blood lead levels above 5 μg/dl where epidemiological studies have quantified hazards.
Results
The increase in mortality risks, by age group, from Europe A to B to C is illustrated in Figure 1. The most striking differences across the three subregions occur between ages 15 and 59. For example, for 30 year old males, the risk of death before reaching 45 in Europe B is twice that in Europe A, and in Europe C, nearly 5 times that in Europe A. For females, the relative mortality differences between Europe B and A are broadly similar, across all age intervals, to those for males, but mortality levels in Europe C are closer to those in Europe B than is the case for males. For example at ages 30 to 45 the risk of death in Europe C relative to Europe A is 2.8 fold for females compared to 5 fold for males. Beyond age 60, mortality risks in Europe B females are much closer to those in Europe C than to those in Europe A.
Figure 1 Probability of death within age intervals from 6 groups of causes by sex: Europe A, B and C, 2000.
In males below age 60, gradients are present for all diseases but the biggest contributors to subregional differences are injuries (up to 8 fold differences between Europe C and A) and vascular diseases. Vascular diseases generally make the biggest absolute contributions to differences in mortality risks. In the age range 45 to 59, for example, risks of vascular death are 3 to 4 fold higher in B relative to A and 4 to 5 fold higher in C relative to A. Interestingly, relative differences are slightly greater in females (see Figure 1). Beyond age 60 in males, cancer and communicable diseases are not important contributors to mortality differences between subregions.
Figure 2 shows the estimated losses of healthy life from the 15 leading causes of disease burden in Europe as a whole. For several causes – unipolar depressive disorders, adult onset hearing loss, osteoarthritis and for two conditions primarily caused by smoking: chronic obstructive pulmonary disease and lung cancer – the magnitude of health loss (DALYs) is comparable across the 3 sub-regions. Age-specific mortality data show, however, that the rough equality of male burdens from lung cancer results from a balancing of higher risks in Europe B and C under the age of 60 with higher risks in A at ages over 60, a consequence of the historic lag of the smoking epidemic in Europe B and C compared to A.
Figure 2 Burden of disease due to 15 leading diseases or disease clusters in 2000: crude rates of DALYs per thousand population for Europe A, B and C, ordered by ranking for the combined European region. Other cardiac diseases are those not classified as rheumatic, hypertensive, ischaemic or inflammatory. Other digestive diseases are those not classified as peptic ulcer, cirrhosis of the liver or appendicitis. Other unintentional injuries are those not classified as motor vehicle accidents, poisonings, falls, fires or drownings.
The burdens experienced in Euro B and C in excess of those in Euro A come predominantly from 2 clusters of diseases: from vascular diseases (a gradient from 42 to 68 to 125 DALYs/thousand people/year from A to B to C when ischaemic heart disease, cerebrovascular disease and other cardiac diseases are combined for both sexes) and injuries (a gradient from 16 to 26 to 50 DALYs/thousand people/year from A to B to C when road traffic accidents, self-inflicted injuries and other unintentional injuries are combined for both sexes).
The burden of disease attributable to leading risk factors in the 3 regions is shown in Figure 3. There is a substantial step down from the health losses attributed to the 7th ranking cause, physical inactivity, to those attributed to the 8th, illicit drugs. The 7 leading risk factors can be divided into 2 clusters: a nutritional/physiological group contributing especially to risk of vascular disease – blood pressure, blood cholesterol concentration, body mass index, BMI (serving as an operational measure for excess adiposity), low fruit and vegetable consumption and physical inactivity. The second group comprises the major legal drugs – tobacco and alcohol.
Figure 3 Burden of disease due to 10 leading risk factors: DALYs per thousand total population for Europe A, B and C, by sex and age group*. * DALYs are assigned to the age of death or of incidence (and not to the age at which the lost healthy life would have been lived)
The contributions of the legal drugs to sub-regional differences vary more by age and sex than do the contributions of the nutritional/physiological group of risk factors.
Much larger burdens are attributed to alcohol in males in Europe C, most strongly between ages 15 and 44, than in the other 2 subregions. Alcohol's contribution in Europe B is likely reduced by the proportion living in predominantly muslim states (Turkey and Uzbekistan alone account for more than a third of the population of Europe B and both are low alcohol consumers [18]).
Smoking contributes powerfully to the increases in disease burdens in males (but not in females) as one moves from Europe A to B to C. Further, the differences in male disease burdens attributed to smoking are relatively greater for streams of lost healthy life (DALYs) that begin in middle age and earlier: In the age range 45 to 59 the ratio of burdens attributable to smoking in Europe B to those in Europe A is around 2 fold and for C relative to A is around 3.5 fold. By contrast for males aged 60+ burdens attributable to smoking are comparable in B and A and less than 2 fold higher in C. Two factors are likely contributing to the greater differences in male burdens from smoking in middle rather than old age: i) very high tobacco consumption among younger men in in many countries in Europe B and most in Europe C, because of the historical lag in the decline in male smoking in Europe B and C relative to Europe A and; ii) the steeper gradient in vascular risks at younger ages, during which smoking acts as a more powerful multiplier than at higher ages.
The nutritional/physiological group of risk factors all show strong gradients in their attributed burdens across the 3 sub-regions, and assumedly account for much of the East-West gradient in premature vascular disease in both sexes (see also Discussion).
About half the DALYs attributed to smoking in Europe C are from vascular disease (because its multiplier effect is acting on very high background risks for vascular diseases), compared to just over a quarter in Europe A (where the background risks for vascular diseases have generally fallen markedly since the 1960s) [19]. Therefore, the magnitude of health losses from smoking, dominated by vascular diseases in the more disadvantaged European subregions, is strongly affected by the nutritional and physiological risk factors, much more so than vice versa. This multiplicative combination of multiple causes explains why such a large gradient exists in total burdens attributed to smoking whilst the gradient for lung cancer is more modest (Figure 2). Whilst the differences in burdens attributed to alcohol are much greater in younger adult males, the large differences in burdens attributed to the nutritional/physiological group of risk factors extend across all age-sex groups – making them responsible for a larger share of East-West health differences overall. At the same time, given the large hazards of alcohol on injuries and neuropsychiatric diseases, which are not affected by the other risks considered, measures to reduce health losses from alcohol will be important components of reducing East-West health differentials in Europe.
Discussion
We have identified three risk factor/disease clusters as leading modifiable causes of excess health losses in Europe B and C: first, in order of importance, is the 'nutritional/physiological' group of risk factors (blood pressure, blood cholesterol concentration, body mass index, low fruit and vegetable consumption and physical inactivity) contributing primarily to very large absolute differences in vascular disease burdens; second, and largely because of its important multiplier effect on vascular risks, is tobacco; and third is the role of alcohol and other contributors to injuries as major sources of differences in health experiences of adult males.
This is the first report summarising differences in health levels and health determinants across Europe using results of the GBD2000 and CRA. The unique and central strength of the methods employed here is their ability to compare and rank the burdens of ill-health caused by different diseases and injuries and to similarly compare and rank the established causes of these diseases and injuries.
The metrics we have used have been developed to favour comprehensiveness and comparability – not ease of precise measurement. In particular, estimates of the 'years lived with disability' (YLD) component of the DALY are subject to significant uncertainty. This is because data for calculating time spent in non-fatal health states are less available than death registrations and because non-fatal health states require valuation before they can be incorporated into summary measures. In Europe as a whole, the 'years lived with disability' (YLD) component accounts for about 45% of DALYs lost. Findings as to the relative contributions of different diseases and injuries to subregional differences in levels of health are, however, robust to the uncertainty intrinsic to YLD estimates, because differentials are dominated by vascular diseases and injuries with large YLL component.
Despite 4 years of systematic data collection and analysis, leading sources of uncertainty for our findings on risk factors based on the CRA project are likely to be the estimated effects of alcohol on injury burden, neuropsychiatric conditions and vascular disease (mainly due to the heterogeneity of hazards across populations), and the estimated contributions of risk factors such as low fruit and vegetable consumption and physical inactivity to vascular risks (due to difficulties in defining and measuring exposure). Estimates for the hazards attributable to risk factors such as blood pressure, blood cholesterol concentration, body mass index and tobacco are likely to be more secure, drawing, as they do, on very large bodies of knowledge.
There is continuing scientific uncertainty about the ability of the classic risk factors such as those included in the CRA to adequately account for the high level [20] and temporal variation [3] in vascular disease rates, especially in Europe C. Several candidate risk factors, including neuro-humorally mediated exposures ('stress'), have been advanced to fill this explanatory gap [21,22]. CRA risk factors have, however, been deliberately limited to those 'for which there was good potential for satisfactory quantification of population exposure distributions and health effects using existing scientific evidence and available data ...' [6] (p xx) – criteria which excluded such candidates, as well as other nutritional risks that require valid data on dietary composition. Despite the potentially important role of these other factors, vascular risk in individuals is related similarly to the classic risk factors across cultures with widely differing risk factor levels [23,24] making them appropriate, albeit not exclusive, targets for public health policy cross-nationally, even if other, currently less well understood, influences are also contributing to differences in national levels and trends.
Because of multicausality and because multiple causes magnify each other's hazards (in a multiplicative way in standard epidemiological models), the contribution of any given risk factor to (absolute) disease burdens depends heavily on the other risks with which it is combining. For example, the burden of disease attributable to suboptimal cholesterol concentrations is over 3 times higher (in relation to population) in Europe C compared to Europe A (Figure 3) even though mean cholesterol concentrations are not higher in Europe C [25]. This combination of multiple risks implies that it is the absolute level of disease risks rather than the levels of individual risk factors that should determine the intensity of the public health response. Where absolute risks of vascular disease are high, as in Europe B and C, the leading public health priority must be to reduce all amenable vascular risk factors irrespective of their starting levels using complementary individual-level and population-wide strategies. An indication of the potential for such interventions is the decline over the past decade and a half in premature vascular deaths, most notably in females, in some former communist countries [26]. Recent rates of decline in countries such as Slovenia, Poland and the Czech Republic have been greater than in Europe A as a whole (unpublished observations).
The appropriate mix of 'high risk' and 'population' strategies for each country will depend on the resources available for medical care and on institutional capacities. Reducing risk in high risk individuals may be achieved by preventive counselling and changes in lifestyle and by 'chemoprevention' (eg long term medication to lower blood pressure and blood cholesterol concentrations). 'High risk' strategies relying on 'chemoprevention' will however, have much less effect on population disease burdens than 'population' strategies, unless large proportions of the adult population are placed on preventive medications [27].
Given the greater potential effectiveness of population approaches and the constraints on implementing high risk approaches, it will generally be preferable to start with population wide measures such as public education on the known causes of heart attack and stroke, and economic and regulatory approaches to help lower salt consumption and saturated fat consumption and to increase fruit and vegetable consumption, along with increased physical activity levels [28].
The second requirement, and potential, for health convergence across Europe is a decline in tobacco consumption in Europe B and C. Trends in some parts of Europe B are again encouraging, with all 8 new EU member states in Eastern Europe showing falling male lung cancer mortality in early middle age, since at least the mid 1990s [29]. The rise in smoking in young women has also been reversed in Poland [30]. Further East, in Russia, Belarus and the Ukraine there is sustained high smoking prevalence in males and rising prevalence in young females [31] – pointing to an urgent need to intensify control measures in those countries. The role of tobacco in levels and distributions of health across Europe is particularly important because economic (e.g. taxes) and regulatory (comprehensive advertising restrictions and bans on smoking in public places) measures have been shown to be highly effective in smoking reduction in many countries [32].
The third cluster to be addressed to reduce health inequalities across Europe is that of alcohol and injuries. The alcohol – injury cluster imposes particularly heavy burdens on Europe C males, especially under the age or 44 (Figure 3). Injuries account for over half the health loss atttributed to alcohol in Europe C males; and 43% of the health loss from injuries in this group is attributable to alcohol [17]. Recent economic analyses indicate (assuming generalisability of findings from studies done elsewhere) that an optimally cost effective strategy for averting overall harms to health from hazardous alcohol use in Europe C would be the combination of a 50% increase in tax, a ban on advertising and brief advice from physicians [33]. In addition to programmes to reduce alcohol consumption, effective means of reducing road traffic injuries are known, including the control of driving under the influence of alcohol which in Europe C males is estimated to account for 64% of traffic deaths.
The former communist countries dealt well with the public health challenges they faced in the period immediately following the second world war [34]. They failed however, to respond effectively to the more complex challenges posed by chronic disease and injury. These failures contributed to one of the gravest tragedies at the latter 20th century – the loss of 2.5 – 3.0 million lives in Russia alone during the 1990s in excess of the losses expected at 1991 mortality levels [4]. The failure to develop public health infrastructures appropriate to the challenges faced, will have contributed substantially to these tragedies. Low rates of scientific publication in priority fields such as cardiovascular disease suggest serious under-investment in research of strategic importance to public health efforts: Medline indexed publication rates on cardiovascular diseases (per million population) are 7 fold lower in Europe B compared to Europe A and 7 fold lower again in Europe C (calculated from data in [35]). Low levels of relevant scientific activity will have provided little stimulus for the mass media to help raise public knowledge of chronic disease risks. Knowledge of established risk factors for vascular disease has been found to be low in Bulgarian [36] and Polish [37] populations. Remedial action is needed to address these infrastructural weaknesses throughout much of Europe B and all of Europe C if future public health endeavours are to be commensurate with the challenges faced.
The diseases and risk factors identified in this analysis also point to important data gaps and future research needs for using combinations of preventive and therapeutic interventions to reduce health disparities across Europe. These include: local analyses in the countries of Europe B and C to better define country-specific hazards for alcohol (including, for example, reasons for the very high mortality from liver cirrhosis in countries such as Hungary); better indicators and data on physical activity, diet and nutritional risks that can be used for identifying interventions; data on multi-risk correlation within countries (which is important for better quantification of hazards, for designing intervention packages for related risks and for within country equity); to develop scenarios for intervention packages and delivery options (including cost-effectiveness); to characterise local institutional strengths and weaknesses; and to assess risk reversibility.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
The scope and purpose of the paper was identified by JP, WZ and ME. The text was mainly written by JP and ME. SH provided analyses from the Global Burden of Disease databases. All reviewed and commented on the text.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
WZ leads an EU public health action (agreement number 2003121) funded for 3 years from December 2004 and aimed at helping to close the east-west health gap in Europe. JP and ME are also investigators in this action. All declare that the specific content of the work reported here has not been influenced by this funding source.
==== Refs
Chenet L McKee M Fulop N Bojan F Brand H Hort A Kalbarczyk P Changing life expectancy in central Europe: is there a single reason? Journal of public health medicine 1996 18 329 336 8887845
Cornia GA Paniccia R eds The mortality crisis in transitional economies 2000 Oxford, Oxford University Press
Leon DA Chenet L Shkolnikov VM Zakarov S Shapiro J Rakhmanova G Vassin G McKee M Huge variation in Russian mortality rates 1984-94: artefact, alcohol or what? Lancet 1997 350 383 388 9259651 10.1016/S0140-6736(97)03360-6
Men T Brennan P Boffetta P Zaridze D Russian mortality trends for 1991-2001: analysis by cause and region BMJ 2003 327 964 0 14576248 10.1136/bmj.327.7421.964
Mathers CD Stein C Ma Fat D Rao C Inoue M Tomijima N Bernard C Lopez AD Murray CJL Global Burden of Disease 2000: Version 2 methods and results Global Programme on Evidence for Health Policy Discussion Paper No. 50 2002 Geneva, World Health Organization
Ezzati M Lopez AD Rodgers A Murray CJL Comparative quantification of health risks: global and regional burden of diseases attributable to selected major risk factors 2004 Geneva, World Health Organization
Mathers CD Ma Fat D Inoue M Rao C Lopez AD Counting the dead and what they died from: an assessment of the global status of cause of death data Bull World Health Organ 2005 83 171 177 15798840
Preston SH Heuveline P Guillot M Demography: measuring and modeling population processes 2001 Oxford, Blackwell
Murray CJL Lopez AD The global burden of disease: a comprehensive assessment of mortality and disability from diseases, injuries and risk factors in 1990 and projected to 2020 Global burden of disease and injury, vol 1 1996 Boston, Harvard School of Public Health on behalf of WHO and the World Bank (distributed by Harvard University Press)
Murray CJL Lopez AD Murray CJL and Lopez AD The global burden of disease in 1990: Final results and their sensitivity to alternative epidemiological perspectives, discount rates, age-weights and disability weights The global burden of disease: a comprehensive assessment of mortality and disability from diseases, injuries and risk factors in 1990 and projected to 2020 Global burden of disease and injury, vol 1 1996 Boston, Harvard School of Public Health on behalf of WHO and the World Bank (distributed by Harvard University Press) 247 293
Ezzati M Lopez AD Rodgers A Van der Hoorn S Murray CJ Selected major risk factors and global and regional burden of disease Lancet 2002 360 1347 1360 12423980 10.1016/S0140-6736(02)11403-6
Miettinen OS Proportion of disease caused or prevented by a given exposure, trait or intervention Am J Epidemiol 1974 99 325 332 4825599
Walter SD Prevention for multifactorial diseases Am J Epidemiol 1980 112 409 416 7424889
Eide GE Heuch I Attributable fractions: fundamental concepts and their visualization Stat Methods Med Res 2001 10 159 193 11446147 10.1191/096228001680195148
Murray CJ Lopez AD On the comparable quantification of health risks: lessons from the Global Burden of Disease Study Epidemiology 1999 10 594 605 10468439 10.1097/00001648-199909000-00017
Murray CJ Ezzati M Lopez AD Rodgers A Van der Hoorn S Comparative quantification of health risks conceptual framework and methodological issues Popul Health Metr 2003 1 1 12780936 10.1186/1478-7954-1-1
Rehm J Room R Monteiro M Gmel G Graham K Rehn N Sempos CT Frick U Jernigan D Ezzati M, Lopez AD, Rodgers A and Murray CJL Alcohol use Comparative quantification of health risks: global and regional burden of diseases attributable to selected major risk factors 2004 Geneva, World Health Organization 959 1108
Organization WH Global status report on alcohol WHO/HSC/SAB/99.11 1999 Geneva, World Health Organization
Ezzati M Lopez AD Ezzati M, Lopez AD, Rodgers A and Murray CJL Smoking and oral tobacco use Comparative quantification of health risks: global and regional burden of diseases attributable to selected major risk factors 2004 Geneva, World Health Organization 883 957
Bobak M Hertzman C, Kelly S and Bobak M Determinants of the East-West life-expectancy gap; contribution of lifestyle and behaviour. Evidence from the MONICA study Environmental and non-environmental determinants of the East- West life-expectancy gap 1996 Dordrecht, Kluwer 99 108
Marmot M Bobak M Cornia GA and Paniccia R Psychosocial and biological mechanisms behind the recent mortality crisis in central and eastern Europe The mortality crisis in transitional economies 2000 Oxford, Oxford University Press 127 148
McEwen BS Protective and damaging effects of stress mediators N Engl J Med 1998 338 171 179 9428819 10.1056/NEJM199801153380307
Chen Z Peto R Collins R MacMahon S Lu J Li W Serum cholesterol concentration and coronary heart disease in population with low cholesterol concentrations [see comments] BMJ 1991 303 276 282 1888927
MacMahon S Peto R Cutler J Collins R Sorlie P Neaton J Abbott R Godwin J Dyer A Stamler J Blood pressure, stroke, and coronary heart disease. Part 1, Prolonged differences in blood pressure: prospective observational studies corrected for the regression dilution bias [see comments] Lancet 1990 335 765 774 1969518 10.1016/0140-6736(90)90878-9
Lawes CMM Vander Hoorn S Law MR Rodgers A Ezzati M, Lopez AD, Rodgers A and Murray CJL High cholesterol Comparative quantification of health risks: global and regional burden of diseases attributable to selected major risk factors 2004 Geneva, World Health Organization 391 496
Zatonski WA McMichael AJ Powles JW Ecological study of reasons for sharp decline in mortality from ischaemic heart disease in Poland since 1991 BMJ 1998 316 1047 1051 9552904
Emberson J Whincup P Morris R Walker M Ebrahim S Evaluating the impact of population and high-risk strategies for the primary prevention of cardiovascular disease Eur Heart J 2004 25 484 491 15039128 10.1016/j.ehj.2003.11.012
Murray CJ Lauer JA Hutubessy RC Niessen L Tomijima N Rodgers A Lawes CM Evans DB Effectiveness and costs of interventions to lower systolic blood pressure and cholesterol: a global and regional analysis on reduction of cardiovascular-disease risk Lancet 2003 361 717 725 12620735 10.1016/S0140-6736(03)12655-4
Didkowska J Manczuk M McNeill A Powles J Zatonski W Lung cancer mortality at ages 35 to 54 in the European Union: an ecological study of evolving tobacco epidemics BMJ 2005 331 191 192 16037451 10.1136/bmj.331.7510.191
Zatonski W Boyle P, Gray N, Henningford J, Seffrin J and Zatonski W Tobacco smoking in Central European countries: Poland Tobacco: Science, Policy and Public Health 2004 Oxford University Press 235 252
Gilmore A Pomerleau J McKee M Rose R Haerpfer CW Rotman D Tumanov S Prevalence of smoking in 8 countries of the former Soviet Union: results from the living conditions, lifestyles and health study Am J Public Health 2004 94 2177 2187 15569971
Jha P Chaloupka F eds Jha P, Chaloupka F and eds Tobacco control in developing countries 2000 New York, Oxford University Press
Chisholm D Rehm J Van OM Monteiro M Reducing the global burden of hazardous alcohol use: a comparative cost-effectiveness analysis J Stud Alcohol 2004 65 782 793 15700517
Zatonski W Boyle P Health transformations in Poland after 1988 J Epidemiol Biostat 1996 4 183 197
Mackay J Mensah G eds The atlas of heart disease and stroke 2004 Geneva, World Health Organization
Dokova KG Stoeva KJ Kirov PI Feschieva NG Petrova SP Powles JW Public understanding of the causes of high stroke risk in northeast Bulgaria The European Journal of Public Health 2005 15 313 316 15941746 10.1093/eurpub/cki086
Wiszniewska M Swiderski W Fryze W Drozdowski W Hertmanowska H Wlodek A Czlonkowska A [Knowledge of stroke problems among adults in Poland] Neurol Neurochir Pol 2000 34 1129 1136 11317489
|
16269084
|
PMC1298310
|
CC BY
|
2021-01-04 16:28:59
|
no
|
BMC Public Health. 2005 Nov 3; 5:116
|
utf-8
|
BMC Public Health
| 2,005 |
10.1186/1471-2458-5-116
|
oa_comm
|
==== Front
BMC Plant BiolBMC Plant Biology1471-2229BioMed Central London 1471-2229-5-231626907610.1186/1471-2229-5-23Methodology ArticleA fully automatable enzymatic method for DNA extraction from plant tissues Manen Jean-François [email protected] Olga [email protected] Lorène [email protected] Alexander V [email protected] Arkady [email protected] University of Geneva, Conservatoire et Jardin Botaniques de la Ville de Genève, Impératrice 1, CH-1292 Chambésy/Genève, Switzerland2 Moscow State University, Chemistry Department, Vorobyevy Gory, 119899, Moscow, Russia2005 3 11 2005 5 23 23 12 7 2005 3 11 2005 Copyright © 2005 Manen 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
DNA extraction from plant tissues, unlike DNA isolation from mammalian tissues, remains difficult due to the presence of a rigid cell wall around the plant cells. Currently used methods inevitably require a laborious mechanical grinding step, necessary to disrupt the cell wall for the release of DNA.
Results
Using a cocktail of different carbohydrases, a method was developed that enables a complete digestion of the plant cell walls and subsequent DNA release. Optimized conditions for the digestion reaction minimize DNA shearing and digestion, and maximize DNA release from the plant cell. The method gave good results in 125 of the 156 tested species.
Conclusion
In combination with conventional DNA isolation techniques, the new enzymatic method allows to obtain high-yield, high-molecular weight DNA, which can be used for many applications, including genome characterization by AFLP, RAPD and SSR. Automation of the protocol (from leaf disks to DNA) is possible with existing workstations.
==== Body
Background
DNA extraction from plant tissues, unlike DNA isolation from mammalian tissues, remains difficult due to the presence of a rigid cell wall surrounding the plant cells. Currently used methods inevitably require a laborious mechanical grinding step, necessary to disrupt the cell wall for the release of DNA. The field of plant molecular biology is therefore at a disadvantage, especially when an automated high-throughput system for the isolation of PCR-ready genomic DNA is required in population genetics, species identification, biodiversity investigation, selection screening, food control and plant biotechnology. QIAGEN GmbH has developed a 96-well grinding method (MagAttract 96 Plant kit), but it requires a special mixer mill, a centrifugation step and consequently is not fully automatable.
Large scale automatable DNA mini-prep facilities were recently offered by several companies for animal tissues (e.g. DYNAL ASA, Oslo, Norway; AGOWA, Berlin, Germany; QIAGEN GmbH, Hilden, Germany; BILATEC AG, Mannheim, Germany; ROCHE Diagnostics, Rotkreuz, Switzerland; PROMEGA Corporation, Madison, WI, USA; SCIL Diagnostic GmbH, Martinsried, Germany). However, because of their cell wall, the automation of the isolation of DNA from plants needs improvements. Whereas animal tissues need only a lysis buffer containing detergents and proteinase K to release their DNA, plant tissues need in addition a mixture of carbohydrase enzymes able to digest the cell wall. Enzymatic digestion of the cell wall of leaf tissues is routinely used for the production of protoplasts but this approach was never adapted for routine isolation of DNA from plant tissue.
We describe here a new method for the lysis of plant tissues using a powerful cocktail of enzymes isolated from Trichoderma longibrachiatum, which digests the cell walls in order to liquefy the tissue without the need of grinding. The enzymatically released DNA is then isolated with commercially available magnetic beads.
Results
Leaf disks from 24 different species were digested by 5 μl of the enzymatic cocktail in 50 μl of digestion buffer. Thirty μl of liquid containing cell debris were drawn up and released DNA was isolated using Dynabeads® DNA DIRECT™ Universal kit (Dynal). Fig. 1A shows an agarose gel of 25% of the DNA isolated (10 μl). Most DNA are high-yield and of high-molecular weight. The amount of lambda DNA/Hind III loaded into the gel was 250 ng (or 500 ng, bottom half, right). The 23 kb band thus represented approximately 120 ng of DNA. The amount of plant genomic DNA obtained was variable from species to species. For some of them (Humulus lupulus, Vitis vinifera, Narcissus pseudonarcissus, Tilia sp., Lilium henryi and Helleborus dumetorum) the amount of loaded DNA was equal to or higher than 120 ng. As only 25% of the isolated DNA were loaded into the agarose gel, it can be estimated that the method permits the isolation of approximately 50 to 500 ng of genomic DNA from a leaf disk, depending on the species.
In the experiment described above, species on which the method was previously tested were selected. In order to empirically examine to what extent the method works on different species, simultaneous extraction of 48 randomly chosen species was carried out in a microtitration plate (as shown on Fig. 2) using the Wizard® Magnetic 96 Plant System kit (Promega) to isolate the released DNA (Fig. 1B). One fourth of the isolated DNA was loaded into the gel. In approximately 75% of the species, genomic DNA was visible. Several DNA extracts show partial degradation. No DNA is visible for Gladiolus palustris, Viburnum farreri, Weigela sp, Prunus padus, Ribes petraeum, Betula sp, Sorbaria sorbifolia, Pelargonium sp, Saintpaulia magungensis.
In experiments described above, the cell wall digestion was done overnight for convenience. In order to follow the release of DNA at different times of enzymatic digestion, three 5 mm leaf disks from dry leaves of Ilex aquifolium were digested for 0.5 to 5 hours and the released DNA was isolated with the Wizard® Magnetic 96 Plant System kit (Promega). Fig. 3 shows that some DNA is already released at 0.5 h and that 3 to 4 hours are sufficient to release most of the DNA from this species. A short digestion time (1 to 3 hours) is sufficient for soft leaves such as Arabidopsis, Begonia, Brassica, Beta, Alium, Nicotiana, Triticum, Piper ...etc. However, incubation times need to be increased for Fragaria, Ribes, Oryza, Soya, Zea ...etc. Thus, although 1 to 3 hours are generally enough, the incubation time for a given species is not foreseeable and the appropriate length of digestion has to be determined experimentally before undertaking large-scale DNA isolations.
The method is highly reproducible. Fig. 4 shows 16 DNA isolated from dry leaf disks of Aster amellus and Ilex aquifolium and from seeds of Allium porum (cut into 2 pieces, see Materials and Methods), using the Wizard® Magnetic 96 Plant System kit (Promega). Fig. 5A shows a comparison of the amount of DNA isolated from Ilex aquifolium by a CTAB-based extraction method (lines 1–5, the protocol includes a treatment with RNase) and by the enzymatic method described here (lines 6–10) using the Wizard® Magnetic 96 Plant System. In both cases the amount of loaded DNA is one fifth of the DNA corresponding to a leaf disk of approximately 3 mg (dry weight). The amount of isolated DNA is similar for both methods.
As indicated by Fig. 5B, lambda DNA/Hind III does not show degradation during incubation with the enzymatic mix, nor in the presence of an overnight digesting leaf disk of Ilex at 50°C. This indicates that the digestion mix does not contain active DNase in the condition used for digestion, and that in the case of Ilex endogenous plant DNase are inactivated by the digestion mix.
Fig. 5 also shows PCR amplification of a plastid sequence (Fig. 5C), a multi-copy nuclear sequence (Fig. 5D) and a single-copy nuclear sequence (Fig. 5E) from 1 μl of DNA isolated by the enzymatic method from Ilex aquifolium, Aster amellus and Solanum tuberosum. The primers have been designed for the genus Ilex. Thus the few negative PCR in other species probably result from primer mismatch and not from polymerase inhibition. RAPD amplifications are also shown (Fig. 5F).
The enzymatic cocktail is produced from Trichoderma longibrachiatum fermentation and could be contaminated with its DNA. Moreover, in the case of a long overnight enzymatic digestion, there is a risk of contamination from bacteria or fungi covering the surface of plant tissues. To examine if such contamination could be a problem, fungi and bacteria specific PCR markers were tested on DNA extracted from (1) a digestion mix alone, or (2) a digestion mix "contaminated" with a Ilex aquifolium leaf disk removed after 10 min and further incubated overnight at 50°C, or (3) a mix digesting a Ilex aquifolium leaf disk overnight at 50°C (Fig. 6). No fungus (particularly Trichoderma longibrachiatum) or bacterial template could be detected. Instead, Ilex aquifolium templates are detected. Indeed, the ITS PCR product found in the digestion mix "contaminated" with a Ilex aquifolium leaf disk removed after 10 min and further incubated overnight at 50°C (line 10) has the same size as ITS of Ilex (larger than ITS of Trichoderma). Further sequencing demonstrated that this PCR product was an ITS sequence of Ilex and not of Trichoderma (data not shown). Similarly, the prokaryotic 16S rDNA sequence obtained from the mix digesting an Ilex aquifolium leaf disk overnight at 50°C (line 19) represented the plastid (prokaryotic) 16S rDNA of Ilex (data not shown), and not a bacterial sequence.
Discussion
To the best of our knowledge, only a few non-grinding methods for isolation of DNA from plant tissue have been proposed, but only low amounts of DNA are generally obtained. Jhingan [1] followed by Williams and Ronald [2] proposed a chemical method using potassium ethyl xanthogenate that damages the cell wall, subsequently disrupts cells and releases the DNA. The method involves many steps and the amount of DNA released is generally ten times lower than traditional methods [1] and than our enzymatic method. A non-grinding method is proposed by SIGMA (Extract-N-Amp Plant PCR kit). It is not based on enzymatic digestion of the cell wall and the leaf tissue usually does not appear to be degraded after the treatment with the lysis buffer. The DNA extract is extremely crude, of low DNA content and often contains PCR inhibitors. Consequently, a 10-fold dilution of the extract is necessary to dilute inhibitors and the template concentration is at the limit of detection. Another method is based on the squashing of plant tissues on a nylon membrane [3] and subsequent elution of the little amount of DNA bound to the membrane for PCR amplification. An adaptation of this method is commercialized by WHATMAN (FTA® gene card). In conclusion, the advantage of our enzymatic non-grinding method of DNA extraction compared with the above-described methods is that a large amount of high-quality DNA is isolated and that it is fully automatable.
In a paper on the comparative analysis of different DNA extraction protocols from plant tissues, Csaikl et al. [4] wrote that "the problem of DNA extraction is still an important issue in the field of plant molecular biology" and that "a chemical tissue disruption method as used in mammalian cells might be the method of choice". Plant DNA purification is time-consuming and laborious. It is considered as the "bottleneck" of basic and applied research [5]. Thus there is a need for a quick, easy and automated method of plant DNA isolation. The method that we present here exactly fits this expectation.
For a few species (approximately 25%, based on our results, see Additional file 1 and Fig. 2B) the method is not effective, but simple modifications of the protocol (particularly the digestion buffer) is expected to resolve the problem in the future. As the chemistry of plant tissues (contrary to animal tissues) is highly variable depending of species, it is not surprising that variable results are obtained. It was exactly the same situation with traditional DNA extraction methods where "recalcitrant" species needed further adaptations [6,7]. There are two situations in which the described protocol does not work (see Additional file 1). In the first case, the leaf disk of some species is not digested by the enzymatic cocktail. This is because some particular chemical compounds inhibit the enzymatic cocktail. Quercus represents such a case and high level of polyphenols (tannin) is suspected. Modifications of the digestion buffer by the addition of polyvinyl pyrrolidone (PVP [8]), or polyvinyl polypyrrolidone (PVPP [9]) or polyethylene glycol (PEG [10]) in order to neutralize polyphenolic compounds could greatly improve the method for "recalcitrant" species. In the second case, the leaf disk is perfectly digested but DNA is not released or, most probably, is highly degraded. This could be due to the release of endogenous recalcitrant nucleases or oxidative polyphenols during the digestion. In other cases (see Betula sp. in Additional file 1) different results can be obtained according the season of leaf harvesting, as it can be expected because of the modification of the chemical composition of the cell wall during the year [11]. To deal with species-dependent variability, it is obviously necessary to determine the optimal digestion conditions for each plant sample. In fact, the duration of incubation is not a problem because the protocol is entirely automatable from solid leaf disks to the PCR-ready DNA. Even if, in some case, it could be longer than mechanical grinding in reaction tube or plate, any human intervention is needed.
Conclusion
In summary, the protocol is simple and reliable, does not require grinding, centrifuging, or the use of hazardous chemicals. A large number of samples can be processed simultaneously, and full automation of the protocol is possible with existing workstations. Many different species were successfully tested. The method can be adapted to each species by modification of the digestion buffer, of the amount of the enzymatic cocktail added during the digestion or of the digestion time.
The method is perfectly adapted to situations when high-throughput isolation of PCR-ready genomic DNA is required. Moreover, because of the high-yield and high-molecular weight DNA reliably obtained, sensitive PCR-based techniques could be applied: AFLP (Amplified Fragment Length Polymorphism), RAPD (Random Amplified Polymorphic DNA), SSR (Simple Sequence Repeat polymorphism).
Methods
Enzymatic cocktail
A mixture of cell wall degrading enzymes was isolated from Trichoderma longibrachiatum Rifai (strain TW-1, deposited in the Russian Collection of Microorganisms under the number VKMF-3934D). The fermenting medium (7 liters) consisted of wheat bran (25 g/L), solid corn steep (25 g/L), hydrolyzed starch (45 g/L), mineral salts and fed by lactose (25% solution at feeding rate of 50 ml/h) after the first 48 hours of fermentation. The fermentation was carried out at 32°C for 144 h. Extracellular secreted enzymes were then isolated by centrifugation (5000 g for 30 min.) and concentrated by ultrafiltration (molecular weight cut-off 10 kD) at 200–250 mg/ml of protein. The obtained enzymatic cocktail was used directly for DNA isolation from plant tissues. It contains, among others, cellulases, beta-glucanases, xylanases, mannanases, xyloglucanases, pectinases, glycosidases (such as beta-glucosidae, beta-xylosidase, alpha-L-arabinofuranosidase, alpha-galactosidase). Additional file 2 gives some enzymatic activities of the cocktail, as assayed according to Ghose [12]. Ribosomal DNA from Trichoderma longibrachiatum was not detected by PCR in the enzymatic cocktail, and cellulase from this organism is in the GRAS list (Generally Recognized As Safe) of the US Food and Drug Administration under the number §184.1250.
The enzymatic cocktail remains stable at least for two years at 4°C. Substantial aliquots of the enzymatic preparation can be obtained from the first author.
Plant tissues
One hundred and fifty six plant species from the Botanical Garden of Geneva were tested with the described enzymatic method of DNA isolation (Additional file 1). Leaf tissue was used in most cases and some seeds were also tested as indicated.
Protocols
Leaf disks (5 mm in diameter) were incubated in 50 μl of digestion buffer (175 mM EDTA [pH 8.0], 100 mM sodium acetate [pH 4.6], 1% triton X100) and 5 μl of the enzymatic cocktail. After digestion (50°C with constant agitation from 3 to 16 hours, depending of species), 30 μl of liquid containing cell debris were drawn up and 200 μl of Dynabeads® DNA DIRECT™ Universal (Dynal ASA, Oslo, Norway) was added. The protocol of DNA isolation was then conducted according the manufacturer's instructions in 1.5 ml microtubes.
Alternatively, leaf disks of 48 randomly chosen species were digested simultaneously overnight in the same conditions on a sealed flat bottom microtitration plate (as shown on Fig. 2). Genomic DNA was further isolated using the Wizard® Magnetic 96 Plant System (Promega Corporation, Madison, WI, USA) and the MagnaBot® 96 Magnetic Separation Device, according to the manufacturer's instructions. For both protocols, DNA was eluted in 40 μl of TE8 (10 mM Tris-HCl, 1 mM EDTA, pH 8.0.). Fresh leaves were generally used, but silica gel-dried leaf tissue can also be digested. As well as leaf tissues, seed tissues were tested. To allow the enzyme solution to penetrate the seed tissue, seeds were broken into pieces of 1–3 mm in side, and one piece was used for DNA isolation. To examine the amount of DNA isolated, 10 μl of the eluted DNA was loaded on a 1% agarose gel containing ethidium bromide and the DNA band was compared with a known amount of lambda DNA /Hind III loaded into the gel.
Stability of the DNA during the enzymatic digestion of plant tissue
One μg of lambda DNA/Hind III was added to the digestion mixture alone or in the presence of a leaf disk of Ilex aquifolium and incubated overnight at 50°C. DNA was then isolated with the Wizard® Magnetic 96 Plant System (Promega) and one fifth of this DNA was loaded for agarose gel electrophoresis.
Comparison with a conventional method of DNA extraction
The amount of DNA isolated by a method of DNA extraction based on CTAB (hexadecyltrimethylammonium bromide, [13]) was compared to the amount of DNA isolated by the enzymatic method described here for dry leaf tissue of Ilex aquifolium. Known amounts (from 14 to 53 mg) of liquid nitrogen-ground leaf tissue of Ilex were conventionally extracted and the isolated DNA was re-dissolved in the proportion of 50 μl of TE buffer per leaf disk (approximately 3 mg), the proportion used in the enzymatic method. The amounts of DNA were then compared by agarose gel electrophoresis.
Genomic DNA analysis
PCR amplifications of a plastid fragment (the atpB-rbcL spacer, [14]), a multi-copy nuclear sequence (ribosomal ITS/5.8S, [15]) and a single-copy nuclear sequence (nuclear encoded plastid glutamine synthetase, [16]) were tested on DNA isolated from a leaf disk of Ilex aquifolium, Aster amellus and Solanum tuberosum. One μl of isolated DNA were used in 25 μl of standard PCR reactions (annealing temperature of 55°C). RAPD amplifications were tested with primer 5' CGGCCCCTGT using 1 μl of the isolated DNA was added to 25 μl of standard PCR reaction (annealing temperature of 37°C).
Authors' contributions
JFM conceived of the method, carried out preliminary experiments and drafted the manuscript. OS and AVM checked and analyzed the biological and chemical activities of the enzymatic cocktail. LA accumulated and interpreted the data. AS designed the enzymatic cocktail.
Supplementary Material
Additional File 1
List of species investigated.
Click here for file
Additional File 2
Typical enzymatic activities and properties of the cocktail used for plant DNA isolation.
Click here for file
Acknowledgements
We would like to thank R. Mayor who routinely uses this method for microsatellite analysis of Aster amellus, providing the picture for Fig. 4, and Catalys AG, Switzerland (Promega corporation) who provided, under advantageous conditions, their Wizard® Magnetic 96 Plant System. We also thank Michelle Price for many English adjustments. This work was supported by the Swiss National Science Foundation (grant SCOPES 7SUPJ062282).
Figures and Tables
Figure 1 Electrophoretic aspect of enzymatically isolated DNA. A: Agarose gel electrophoresis of typical enzymatically isolated DNA from 24 different species (in the following order: Phlomis fructicosa, Humulus lupulus, Veratrum album, Scilla bifolia, Astragalus gummifer, Vitis vinifera, Centaurea macrocephala, Narcissus pseudonarcissus, Allium ampeloprassum, Salvia officinalis, Viburnum carlesii, Colchicum speciosum, Triticum turgidum, Polygonum chinensis, Lathyrus vernus, Tilia sp., Caragana sophorifolia, Urtica dioica, Lilium henryi, Polygonum multiflorus, Geranium sp., Lupinus sp., Crocus albiflorus, Helleborus dumetorum). After digestion, the DNA was isolated with Dynabeads® DNA DIRECT™ Universal magnetic beads. One fourth of the isolated DNA was loaded. The first and last lines were loaded with 250 ng of lambda/HindIII DNA (500 ng, bottom half, right). B: Agarose gel electrophoresis of DNA of 42 randomly chosen species (in the following order: Danae racemosa, Epimedium alpinum, Gladiolus palustris, Viburnum farreri, Euonymus bungeana, Weigela sp., Prunus padus, Rhodea japonica, Polygonatum multiflorum, Daphne japonica, Ribes petraeum, Asplenuim scolopendrium, Carex morrowii, Aruncus dioicus, Bletilla striata, Helleborus odoratus, Hedera helix, Brunnera macrophylla, Paeonia belladonna, Atropa belladonna, Solanum tuberosum, Beta vulgaris, Anethum graveolens, Allium fistulosum, Sison amomum, Uniola latifolia, Sinningia magnifica, Peperomia sp., Alnus sp., Tillia sp., Betula sp., Skimmia sp., Liriope spicata, Anthericum liliago, Inula ensifolia, Phlomis fruticosa, Lilium pumilum, Sorbaria sorbifolia, Dietes bicolor, Ilex aquifolium, Vitis vinifera, Setaria italica, Triticum aestivum, Nymphea sp., Pelargonium sp., Saintpaulia magungensis, Morinda sp., Zea mais). After enzymatic digestion in half of a 96 microtitration plate, DNA was isolated using Wizard® Magnetic 96 Plant System magnetic beads. One fourth of the isolated DNA was loaded.
Figure 2 Enzymatic disruption of leaf disks in a microtitration plate. A flat bottom microtitration plate filled with 50 μl of digestion buffer and leaf disks of different species before the adding of the enzymatic cocktail.
Figure 3 Time scale DNA release from digesting leaf disks. Triplicate essay of time scale DNA release from leaf disks of Ilex aquifolium at 0.5 to 5 hours of enzymatic digestion. Size marker: lambda/HindIII DNA.
Figure 4 Reproducibility of enzymatical isolation of DNA. (A) From leaf disks of 16 individuals of Aster amellus (one tenth of the isolated DNA was loaded), (B) From 16 leaf disks of Ilex aquifolium (one fifth of the isolated DNA was loaded) and (C) From 16 seeds of Allium porum (one fifth of the isolated DNA was loaded) using Wizard® Magnetic 96 Plant System magnetic beads.
Figure 5 Features and properties of enzymatically isolated DNA. A: comparison of the amount of DNA isolated from Ilex aquifolium leaves by a conventional extraction method (lines 1–5) and by the enzymatic method described here (lines 6–10). In both case the amount of loaded DNA is one fifth of the DNA corresponding to one leaf disk. B: Study of the stability of lambda DNA/Hind III during the digestion of leaf disks of Ilex. Line 1: one leaf disk alone; line 2: one leaf disk and lambda DNA/Hind III; line 3: lambda DNA/Hind III alone. C, D, E, and F: PCR amplification of enzymatically isolated DNA from Ilex aquifolium (line 1), Aster amellus (line 2), and Solanum tuberosum (line 3). C: PCR amplification of the plastid atpB-rbcL spacer. D: PCR amplification of ITS/5.8S. E: PCR amplification of the nuclear encoded plastid glutamine synthetase. F: RAPD amplification.
Figure 6 Contamination checking: PCR markers for fungi and bacteria in DNA isolated by the enzymatic method. Lines 1 to 12: Internal transcribed spacer (ITS) of ribosomal DNA amplified with eukaryotic specific universal primers ITS1 and ITS4 [17]. Lines 13 to 20: 16S ribosomal DNA amplified with prokaryotic specific universal primers 9f and 1429r [18]. Amplifications from respectively 10, 1, 0.1 and 0.01 pg of genomic DNA of Trichoderma longibrachiatum (lines 1 to 4), Ilex aquifolium (lines 5 to 8) and Artrospira sp. (lines 13 to 16). Amplifications of DNA isolated from a digestion mix alone (lines 9 and 17), a digestion mix "contaminated" with an Ilex aquifolium leaf disk removed after 10 min and further incubated overnight at 50°C (lines 10 and 18) and a mix digesting an Ilex aquifolium leaf disk overnight at 50°C (lines 11 and 19). Line 12 and 20: negative controls.
==== Refs
Jhingan A A novel technology for DNA isolation Methods Mol Cell Biol 1992 3 15 22
Williams CE Ronald PC PCR template-DNA isolated quickly from monocot and dicot leaves without tissue homogenization Nucl Acids Res 1994 22 1917 1918 8208619
Langridge U Schwall M Langridge P Squashes of plant tissue as substrate for PCR Nucl Acids Res 1991 19 6954 1762926
Csaikl UM Bastian H Brettschneider R Gauch S Meir A Schauerte M Scholz F Sperisen C Vornam B Ziegenhagen B Comparative analysis of different DNA extraction protocols: a fast, universal maxi-preparation of high quality plant DNA for genetic evaluation and phylogenetic studies Plant Mol Biol Rep 1998 16 69 86 10.1023/A:1007428009556
Mace ES Buhariwalla HK Crouch HJ A high-throughput DNA extraction protocol for tropical molecular breeding programs Plant Mol Biol Rep 2003 21 459a 459h
Weeb DM Knapp SJ DNA extraction from a previously recalcitrant plant genus Plant Mol Biol Rep 1990 8 180 185
Baker SS Rugh CL Kamalay J RNA and DNA isolation from recalcitrant plant tissues Biotechniques 1990 9 268 272 1699562
Porebski S Bailey LG Baum BR Modification of a CTAB DNA extraction protocol for plants containing high polysaccharide and polyphenol components Plant Mol Biol Rep 1997 15 8 15
Lodhi MA Ye G-N Weeden NF Reisch BI A simple and efficient method for DNA extractions from grapevine cultivars and Vitis species Plant Mol Biol Rep 1994 12 6 13
Carlson JE Tulsieram LK Glaubitz JC Luk VMK Kauffeldt C Rutledge R Segregation of random amplified DNA markers in F1 progeny of conifers Theor Appl Genet 1991 83 194 200 10.1007/BF00226251
Boudet A-M A new view of lignification Trends Plant Sci 1998 3 67 71 10.1016/S1360-1385(97)01176-X
Ghose TK Measurement of cellulase activities Pure Appl Chem 1987 59 257 268
Doyle JJ Doyle JL A rapid DNA isolation procedure for small quantities of leaf tissue Phytochem Bull 1987 19 11 15
Cuénoud P Del Pero Martinez MA Loizeau P-A Spichiger R Andrews S Manen J-F Molecular phylogeny and biogeography of the genus Ilex L. (Aquifoliaceae) Ann Bot 2000 85 111 112 10.1006/anbo.1999.1003
Manen J-F Boulter MC Naciri-Graven Y The complex history of the genus Ilex L. (Aquifoliaceae): evidence from the comparison of plastid and nuclear DNA sequences and from fossil data Plant Syst Evol 2002 235 79 98 10.1007/s00606-002-0225-x
Emshwiller E Doyles JJ Chloroplast-expressed glutamine synthetase (ncpGS): potential utility for phylogenetic studies with an example from Oxalis (Oxalidaceae) Mol Phyl Evol 1999 12 310 319 10.1006/mpev.1999.0613
White TJ Bruns T Lee S Taylor JW Innis MA, Gelfand DH, Sninsky JJ, White TJ Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics PCR Protocols: A Guide to Methods and Applications 1990 Academic Press, Inc., New York 315 322
Weisburg WG Barns S Pelletier DA Lane DJ 16S ribosomal DNA amplification for phylogenetic study J Bacteriol 1991 173 679 703
Chaplin MF Kennedy JF Carbohydrate analysis: A practical approach 1994 second Oxford University Press, Oxford, New York, Tokyo 4
|
16269076
|
PMC1298311
|
CC BY
|
2021-01-04 16:03:52
|
no
|
BMC Plant Biol. 2005 Nov 3; 5:23
|
utf-8
|
BMC Plant Biol
| 2,005 |
10.1186/1471-2229-5-23
|
oa_comm
|
==== Front
BMC PsychiatryBMC Psychiatry1471-244XBioMed Central London 1471-244X-5-411628392510.1186/1471-244X-5-41Research ArticleBright light treatment of depression for older adults [ISRCTN55452501] Loving Richard T [email protected] Daniel F [email protected] Jeffrey A [email protected] Nancy C [email protected] Michael A [email protected] Department of Psychiatry, University of California, San Diego, USA2005 9 11 2005 5 41 41 11 5 2005 9 11 2005 Copyright © 2005 Loving et al; licensee BioMed Central Ltd.2005Loving 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 incidence of insomnia and depression in the elder population is significant. It is hoped that use of light treatment for this group could provide safe, economic, and effective rapid recovery.
Methods
In this home-based trial we treated depressed elderly subjects with bright white (8,500 Lux) and dim red (<10 Lux) light for one hour a day at three different times (morning, mid-wake and evening). A placebo response washout was used for the first week. Wake treatment was conducted prior to the initiation of treatment, to explore antidepressant response and the interaction with light treatment. Urine and saliva samples were collected during a 24-hour period both before and after treatment and assayed for aMT6s and melatonin respectively to observe any change in circadian timing. Subjects wore a wrist monitor to record light exposure and wrist activity. Daily log sheets and weekly mood (GDS) and physical symptom (SAFTEE) scales were administered. Each subject was given a SCID interview and each completed a mood questionnaire (SIGH-SAD-SR) before and after treatment. Also, Hamilton Depression Rating (SIGH-SAD version) interviews were conducted by a researcher who was blind to the treatment condition. A control group of healthy, age-matched, volunteers was studied for one day to obtain baseline data for comparison of actigraphy and hormone levels.
Results
Eighty-one volunteers, between 60 and 79 years old, completed the study. Both treatment and placebo groups experienced mood improvement. Average GDS scores improved 5 points, the Hamilton Depression Rating Scale (HDRS) 17 scores (extracted from the self-rated SIGH-SAD-SR) improved 6 points. There were no significant treatment effects or time-by-treatment interactions. No significant adverse reactions were observed in either treatment group. The assays of urine and saliva showed no significant differences between the treatment and placebo groups. The healthy control group was active earlier and slept earlier but received less light than the depressed group at baseline.
Conclusion
Antidepressant response to bright light treatment in this age group was not statistically superior to placebo. Both treatment and placebo groups experienced a clinically significant overall improvement of 16%.
==== Body
Background
Several reviews have emphasized the enormous socioeconomic impact in elders of insomnia and depression which are often intertwined [1-5]. Reasons for depression in elders include: loss of body strength, health and autonomy, loss of loved ones and friends, loss of occupational status, and fear of impending death [6-8]. Some years ago, the suicide rate was thought to be low among the oldest adults, but this rate has been rising [9,10], despite the improved financial condition of the over-60 population. Antidepressant drugs are reasonably effective in older patients, but depression often is inadequately-treated or chronic. In elders, there is a greater risk of medication side effects such as falls, over-sedation, and anti-cholinergic disturbances [11,12]. These difficulties point to the potential usefulness of light treatment for depressed elders either to supplement or replace pharmacotherapy. Development of new light treatments could have widespread benefit for millions of aging Americans with insomnia and depression.
Light treatment for seasonal depression (SAD) has become accepted in the Clinical Practice Guidelines issued by the U.S. Department of Health and Human Services [13] and the American Psychiatric Association's Treatment of Psychiatric Disorders [14]. It is gratifying to see this safe, inexpensive, rapid and effective treatment spreading worldwide into clinical practice. Fewer light treatment studies have been conducted to investigate nonseasonal major depressions, but light is effective for nonseasonal depression [15]. In nonseasonal depression, light treatment may produce as much incremental benefit or more when antidepressant drugs are also administered [15]. There has been little formal study of light treatment in the elderly, partly because seasonal depression seems less common among women after menopause [16].
Several small studies have suggested promising results with bright-light treatment of elderly patients who have distinctive phase-advanced sleep disturbances [17], or with dementia [18-22], but there has been little critical examination of the value of bright light for the majority of elderly Americans with less specific mood or sleep complaints.
The mechanisms by which light produces antidepressant effects are of essential scientific interest, both for understanding the underlying etiologies of depressive illnesses and for guiding our treatment strategies. There is considerable evidence that morning light is better than evening light for SAD [15,23-26], but some studies have found little difference between timings [27,28], and the apparent advantage of morning light may be partly explained by an anomalous order effect in cross-over designs [29]. When morning light is effective, it might work by suppressing or phase-advancing an overly-late melatonin offset [30-32]. It seems possible (though unproven) that conditions in which sleep complaints are most prominent and conditions in which mood complaints are most prominent might have a common etiology in circadian phase malsynchronization which is characterized by abnormal entrainment of circadian rhythms to the solar day and/or abnormal relationships among rhythms in the body.
Phase-typing as a predictor of light-treatment response was proposed by Lewy and colleagues [33] and has been previously employed to select light treatment timing prospectively [34]. We used this method in the belief that without phase-typing and individualized treatment assignment, sub-optimal results would be obtained.
Our early studies [35,36] the Praskos' study [37], and especially the work of Neumeister et al. [38], Loving [39], and Bloching [40] suggested that partial sleep deprivation combined with bright light produces remarkable antidepressant responses, as demonstrated by dramatic contrasts between bright light and placebo. Considering the evidence that sleep deprivation may tend to accentuate the contrast of bright light and placebo, we expected partial sleep deprivation would add to the potential effectiveness of individual timed light therapy.
In this paper, we report on a clinical trial of bright light treatments in the home, with individualized treatment timing. The goals of the study were to determine if light resistance in the 60–79 year age range can be overcome with 4 weeks of bright light treatment and to gain information on the relative benefit of different times of light treatment.
Our aims were to compare the effectiveness of morning, mid-day, and evening light treatments, individualized according to clinical and actigraphic assessments, in correcting circadian disturbances. Also, the trial sought to demonstrate greater improvements in mood and sleep among those volunteers receiving 4 weeks bright light as contrasted to placebo.
Methods
Recruitment from July 2000 to December 2003 used advertising and community presentations. Written informed consent was obtained from each participant prior to the start of the study, in accordance with the guidelines set forth by the Declaration of Helsinki. The study protocol and consent form were approved by the UCSD Human Research Protection Program. In addition, those participants who were being treated for depression by either a physician or counselor were requested to obtain the written agreement of the therapist for the study, to assure that there was no interference with ongoing treatment and treatment responsibility. Patients were encouraged to continue ongoing treatment during the study, with the assumption that psychotherapy and medication effects over an interval of 4 weeks were likely to be small and randomized between groups.
After signing written informed consent, volunteers, ages 60–79 years, with significant depressive complaints were evaluated for the randomized clinical trial. An intake assessment questionnaire included questions concerning depressive symptoms, the Geriatric Depression Scale (GDS) [41], the Horne-Östberg Morningness-Eveningness Scale [42], questions concerning sleep, questions concerning time in daylight, and questions concerning medication use and current illnesses. The Structured Clinical Interview for DSM-IV Axis I Disorders (Non-patient Edition) [43] was administered during the baseline week, prior to randomized treatment. For enrollment in the study, a GDS score of 11 (indicating probable major depression) [41] was required. This cut-off has 81% sensitivity and 61% specificity for DSMIII-R major depression. Meeting full DSM-IV criteria for current major depressive disorder was not required, because many aging depressed people are significantly troubled by minor depressive disorders without meeting criteria for major depressive disorder [44]. Any lifetime history of mania required exclusion of the potential volunteer, as a history of mania appears to predict a greatly increased risk of a manic switch during bright light treatment [45]. It may be assumed that almost all aging depressed volunteers will offer sleep complaints. The purpose of the sleep questions was to identify symptoms suggestive of circadian sleep phase disorders, such as evening drowsiness or prolonged sleep latency, early awakening or tending to awaken late, which might suggest particular responsiveness to bright light treatment. These symptoms were evaluated to select the optimal treatment, and their presence suggested that the volunteer was a good candidate for study recruitment. In contrast, symptoms strongly suggestive of sleep apnea such as obesity, loud snoring, and choking and gasping during sleep were contra-indications to selection, as light is not known to be useful for depressive symptoms related to breathing disorders in sleep. The questions concerning daylight exposure enabled us to avoid volunteers who were outdoors so much that light treatment had little to add (e.g. outdoors for more than an hour during times of potential light triggered circadian rhythm shifts, that is morning or evening hours). Throughout the five weeks of study, subjects continuously wore an Actillume wrist monitor (Ambulatory Monitoring Inc., Ardsley, NY) to record movement and light exposure.
Automated sleep scores were obtained from these records using a previously validated algorithm [46,47]. Available Actillume sleep measures for the night intervals in the baseline week (an average of 6.8 nights) and the final treatment week (an average of 6.7 nights) were averaged. Sleep parameters at baseline were contrasted among the groups assigned to bright and dim light in the morning, midday, or evening, in two-way factorial ANOVA. Further, sleep parameters at the end of treatment were likewise contrasted, using the baseline values as a covariate.
Volunteers began placebo treatment during the initial baseline week of the study. The purposes of placebo treatment during the baseline week were to identify placebo-responders, to test the volunteer's compliance, and to collect baseline data. A member of the research staff visited the volunteer's home bringing a modified Sunbox light enclosure containing two red LED light sources (Enerlight Corp. Model ENS2000), producing less than 10 photopic lux of red light measured at 18" by a photometer pointed towards the center of the diffuser (See Figure 1). The volunteer was asked to sit in front of this dim-red-placebo light box for 60 minutes, at mid-wake, for the baseline week of the study. Mid-wake, determined from the questionnaires, was half-way between out-of-bed time and lights-out time.
Figure 1 Spectrophotometric measures of illumination are shown comparing daylight with the white and red treatment lights. Sunlight was measured with the photometer pointed towards the horizon (and shaded from direct sun) near noon on a clear sunny day (32.85 North latitude, 2/2/05). White light was measured at 18" with the photometer oriented towards the center of the box. The red light was measured with the photometer adjacent to the diffuser, because at 18" the illumination was too dim to be plotted on the same scale. The irradiance scale was arbitrary (uncalibrated) but identical for the three measures.
If a volunteer's GDS score dropped 20% or more from the first day to the last day of the baseline week, the volunteer was dropped prior to randomization. This aspect of the design followed the general principle of clinical trial design that better contrasts between active and placebo treatments can often be obtained if early placebo-responders are eliminated. As a courtesy to those subjects who were dropped, they were permitted to continue with the dim-red-placebo light box for 4 weeks if they wished.
Volunteers underwent baseline circadian assessment during the week of dim-red-light placebo treatment. The investigators predicted the best-choice timing for each volunteer, in advance of treatment randomization, using subjective questionnaire, sleep log data and the Actillume recordings. Evening bright light was selected as the best choice for volunteers who reported that they were falling asleep in the evening excessively before going to bed or who found themselves going to bed earlier than desired, who complained of early awakening, or whose baseline light-activity recordings indicated that they had these problems. Conversely, early morning bright light was selected for volunteers who reported that they had trouble falling asleep at their desired bedtime, had a long sleep latency, had trouble waking up in the morning or whose baseline light-activity recordings suggested a delayed sleep phase. Mid-day bright light was selected for volunteers who reported no symptoms suggestive of a circadian sleep-phase disorder, who reported mixed symptoms which did not segregate into a consistent pattern suggestive either of advance or delay, and whose light-activity recordings gave no persuasive indication of a circadian-sleep-phase disorder. Where choice of assignment seemed inconclusive, the investigators sought to assign volunteers to equalize the numbers receiving morning, mid-day, and evening treatment. Utilization of placebo groups with each timing was necessary to control for the different sleep times of those assigned to the different treatment timings and for the behavioral influences of sitting in front of light boxes at different times of day.
Having been assigned one of three treatment times in advance, placebo non-responders within each assigned treatment time were then randomized into one of 2 treatment groups: A) 8,500 lux bright white light (model: SunRay, Sunbox, Gaithersburg, MD) or B) 10 lux dim-red-light placebo. Randomized assignment within blocks was stratified for time-of-treatment, age below or ≥ 68 years, and baseline GDS score below or ≥16 using computerized randomization and sealed envelopes.
To test the benefits of partial sleep deprivation, on the final night of baseline, we asked volunteers to awaken themselves 4 hours after going to bed and to remain awake for the second half of the night. They were asked to call our telephone answering machine every half hour to confirm that they were awake during that time period. In a previous study, we found such home sleep deprivations work well without complication [39]. An additional GDS rating was completed at the usual time of awakening after this half-night sleep deprivation.
As reviewed by Eastman [48], the issue of placebo responses has been a serious problem in clinical bright light studies, though the placebo problem has been negligible in studies of the physiologic effects of light. We have employed dim red light as a placebo, reasoning that because the light was dim and because the red part of the spectrum is relatively inactive biologically [49], there would be no substantial effect. Many subjects consider bright white fluorescent light to be glaringly ordinary, whereas the red light may appear more special. The investigators find no controlled evidence that dim-red-light is anything but placebo. Fortunately, claims by others of red-light benefits allowed us to tell volunteers, without deception, that some people think that red light is active (even though we do not agree). In this way, we attempted to maintain the best possible subject blind to the treatment expectation.
To assess subject perception of treatments, an expectation rating was obtained at the beginning of the study and at the end of treatment. This consisted of a 100 mm visual analog scale for the expectation for both sleep and mood improvement. The initial rating was obtained after the subject was randomized and had seen the light they would be using but before the first actual treatment. A final assessment was obtained on the last day of the study.
During the baseline (placebo) week, the volunteer completed sleep-activity logs daily and continuously wore an Actillume to monitor baseline sleep-wake patterns and baseline illumination patterns. The GDS [41] was obtained on the first and last days of the baseline week. Fractional urine samples supplemented by limited saliva collections were collected during baseline and final weeks to characterize the circadian phase of the subject's melatonin rhythms.
Urinary excretion of the major melatonin metabolite, 6-OH-melatonin sulfate (6-sulphatoxymelatonin or aMT6s) was used as the primary phase marker of the endogenous circadian pacemaker. For a 24-hour period prior to treatment and another at the end of four weeks of treatment, volunteers collected each fractional urine specimen, measured and recorded the time and volume, and froze duplicate 2 cc vials for assay. On the same occasions, from 4 hours before until bedtime and again from wake-up until 4 hours after wake-up, the volunteer collected and recorded the time of hourly saliva samples for a total of 5 evening samples, and 5 morning samples. Saliva samples were collected to provide a potentially more accurate measure of the onset and offset of melatonin secretion than could be interpolated from aMT6s excretion.
Urinary aMT6s was assayed using 96 well ELISA kits (Bühlmann Labs, EK-M6S) purchased from ALPCO, Ltd. (Windham, NH), a competitive immunoassay that uses a highly specific rabbit anti-6-sulfatoxymelatonin antibody and a second antibody capture technique. Assay performance has been extensively validated by the manufacturer and results correlate well with the Arendt (Stockgrand, Ltd) RIA (r = 0.987). Saliva samples were collected using polyester-swab Salivettes (Sarstedt, Numbrecht, Germany), centrifuged, and stored at -70°C. until assay. Samples were pretreated and assayed using Bühlmann laboratories Direct Saliva Melatonin ELISA kits (EK-DSM, ALPCO, ltd., Windham, NH).
Reliable estimates of aMT6s acrophase, onset and offset require a clear circadian pattern that is free of major irregularities. To ensure reliability, we examined all excretion curves visually to record an overall quality score for each 24-hour profile. This evaluation, blind to all other information on participants, was based mainly on the shape and completeness of the ng/h curve, but agreement between ng/h and ng/ml temporal patterns, regularity of the baseline, and reliability of the patient log were also considered. Based on this evaluation, the present analyses used 48 profiles from week 1 and 43 profiles from week 5 excluding 24% for poor quality (15 and 14 respectively). Aside from the above considerations, data were unavailable (assays were not performed) for 26% of cases due to the poor quality of the home collection or the loss of critical samples.
The aMT6s excretion rate for each urine sample was computed and transformed into 5 min epoch data and the resulting time series data were imported into Action3 software (Ambulatory Monitoring Inc., Ardsley, NY), where they were aligned with activity, sleep and illumination data and further checked for accuracy. Then 24-hour least-squares cosine fits were computed for each aMT6s collection, yielding mesors and acrophases. To estimate the duration of nocturnal aMT6s excretion, the onset and the offset of the excretion were estimated by interpolation of times at which the excretion rate (ng/h) crossed the mesor level. The time of onset of aMT6s excretion was estimated as the upward crossing and offset as the downward crossing of the mesor level. The aMT6s duration was defined as the interval between onset and offset times.
Salivary melatonin data were rated for reliability following guidelines similar to those used for urinary aMT6s, and poor quality data were eliminated from further analyses. Concurrently, where possible, onset was defined by the first data point in the evening that was elevated above a subjectively viewed baseline and was preceded by at least one point at or below that value. Similarly, offset was defined by the last data point in the morning that was elevated above the baseline and was followed by at least one point at or below baseline. Values used were an average from two blind raters. Due to problems with the quality of home collected saliva data, and some failed assays, good quality saliva melatonin data were available from only 28 collections in week 1 and 35 collections from week 5.
The primary test of the light treatment in correcting circadian phases was a contrast of the phase dispersion of aMT6s acrophases in the bright-light-treated group vs. the placebo group after 4 weeks of treatment, to test the hypothesis that light treatment reduces dispersion of circadian phases. This contrast was computed by calculating the median acrophase of each group. Then, for each subject in the trial, the [absolute value] deviation of each acrophase from the group median was computed pre- and post-treatment (d1 and d2), and the change in deviations was calculated (D = d2 - d1).
In addition to daily log sheets used to record activity, sleep behaviors, and visual analog self-ratings of mood, the subjects completed a weekly GDS and a Systematic Assessment for Treatment Emergent Events (SAFTEE) symptom scale [50]. Further mood measurements were made at baseline and end-of-treatment using the SIGH_SAD_SR, a self-rating form of the Hamilton Depression Rating Scale (HDRS) which includes atypical items previously shown to be responsive to light treatment [51]. Additionally, when a graduate student, blind to treatment, was available, an HDRS interview was administered at baseline and near the last day of the study.
Four weeks of treatment were carried out with weekly symptom assessments and continuous wrist recordings of activity and illumination exposure. The investigators visited subjects weekly to assure their safety and their compliance with the study, to administer and collect rating forms, and to transfer data from the Actillume recorders. A final symptom and circadian assessment was completed in the last 48 hours of the 4-week randomized treatment. Two-week, 4-week and 3-month follow-up assessments were obtained.
Records from the Actillume monitor indicating total activity, sleep-wake, log10[lux], and total lux were fitted to cosine curves for each subject. The mesors or fitted cosine means were examined, as well as the acrophases which indicate the time of day of the fitted peak.
Healthy controls
Control subjects were recruited over the same interval as depressed subjects. Volunteers who did not have a history of depression were invited to participate as healthy controls. The informed consent and screening procedures were similar to those for depressed subjects. If qualified, the volunteers wore an Actillume monitor for 48 hours. During this period they collected urine and saliva samples over 24 hours in the same manner as the depressed subjects.
Results
Based on power analysis, it was our intent to study 150 subjects over a five year period. After three and one-half years, an interim data analysis did not detect any trend towards a benefit from bright light (though a minimal trend appeared in the final results below). We convened an informal data safety and monitoring board to determine if continuing to collect data would be justified. Because demonstration of a significant light treatment benefit had become very unlikely, the board recommended discontinuing the study. Results after early termination are presented below.
Signed consents for 191 participants were obtained to conduct screening. Of these, 119 subjects met criteria for enrollment, of which 81 completed the protocol. About one third of the 119 subjects responded to placebo light and were dropped during the first week of the study, prior to randomization. There were 7 dropouts for other reasons, before randomization. Among these reasons were: family emergencies, discomfort with the research protocol and equipment, and a decision not to continue for lack of motivation. There were 2 dropouts after randomization: both were due to medical health issues, unrelated to the research, which required hospitalization. One had received bright light and the other received dim light. There were 34 male and 47 female subjects who completed the study. Bright light treatment was completed by 41 subjects, and 40 received dim red light. The mean age for the completers was 67.7 years (SD = 5.45) and ranged from 60 to 79.
Of the 81 subjects who completed the study, 24 were being seen by a psychotherapist during research treatment. Of these 24, 12 received bright light and 12 received dim light. In all 24 cases, treatment was stable during the study. Medication usage for the sample was as follows; Antidepressants = 30 (15 bright light, 15 dim light), Antianxiety = 11, Cardiac = 16, Antihypertensive = 25, Analgesic = 20, Hypnotic = 8, Thyroid = 15, Hormone Replacement Therapy = 22, Diabetes Drugs = 8 cholesterol-lowering drugs = 14. Nineteen subjects received both psychotherapy and psychiatric medication: 9 received bright light and 10 received dim light. Seven of the bright light subjects and 7 of the dim light subjects remained stable on their treatment regimen during the study. Two subjects receiving bright light had medication changes and 3 dim light subjects had medication changes, without any consistent patterns. Among all 81 subjects, those patients who were not taking antidepressant medication experienced nonsignificantly greater mood improvement from light treatment as measured by the GDS (p = .124), HDRS17 (p = .150) and, HDRS21 (p = .146). Thus less than half of our sample subjects were receiving psychotherapy and/or antidepressant medication. They represented a broad spectrum of depressed patients which ranged from long term chronic conditions with multiple episodes to single events which both did and did not meet criteria for Major Depressive Disorder. We obtained this sample deliberately as we wished to test the general applicability of light therapy for elders with depressive symptoms.
Based on the SCID interviews, Axis I diagnoses for the sample were as follows; Major Depressive Disorder 57, Minor Depressive Disorder 14, Adjustment Disorder 2, Schizoaffective Disorder 1, and Mood Disorder due to a General Medical Condition with depressive features 1. None of the subjects met criteria for Seasonal Affective Disorder. Five subjects had GDS ≥11 but did not meet SCID criteria for any Axis I diagnosis.
For the 81 subjects who completed the protocol, light treatments were randomized and stratified as shown in Table 1. Groups assigned to active and placebo light treatment were balanced in age and severity overall and by time-of-day of treatment. Expectations for sleep and mood effects of light treatment are shown in Table 2. Repeated-measures MANOVA showed no difference in the sleep and mood expectations between the bright and dim treatment groups either before or after treatment, suggesting that the dim red placebo induced balanced expectations.
Table 1 Randomization of Light Treatment Time and Stratification by Age and Depression Severity
Time of Light Treatment Bright Dim Total
Morning 13 15 28
Mid-Wake 15 16 31
Evening 13 9 22
Total 41 40 81
Age-Depression Severity Group
Age<68, GDS<16 4 4 8
Age<68, GDS ≥16 18 16 34
Age ≥68, GDS<16 6 6 12
Age ≥68, GDS ≥16 13 14 27
Total 41 40 81
Table 2 Expectations for Improvement in Sleep and Mood
100 mm Visual Analog Scale, 0 = Worse 100 = Better
Measure Sleep Mood
Light Bright Dim Bright Dim
Time Initial Final Initial Final Initial Final Initial Final
N 40 40 38 38 40 40 38 38
Mean 71.4 71.2 68.7 63.5 72.2 73.4 72.3 68.3
SD 15.3 14.7 13.5 21.6 15.5 13.4 15 21
Sleep changes
As might be anticipated from the phase-typing, subjects assigned to morning, midday, and evening light went to bed at 00:03, 23:12, and 22:49, respectively (p < 0.05). Similarly, they got out of bed at 08:14, 07:21, and 07:09 respectively (p < 0.06). Further, those treated with bright light got out of bed an average of 37 min. earlier (p < 0.06), regardless of treatment timing and without significant interaction with treatment time. There was no significant effect of randomized treatment assignment on time of going to bed or getting out of bed, nor did total time in bed vary significantly by treatment. The data for times of sleep onset and sleep offset within the nocturnal periods had similar trends. The total amount of Actillume-estimated sleep was balanced at baseline and not significantly affected by treatment assignment. The baseline, initial week, estimate of total sleep during the nocturnal period was 327 minutes. During the final week the estimate of total sleep during the nocturnal period was 330 minutes. The sleep efficiencies for the initial week and the final week were 70.8 and 72.6 respectively. Wake After Sleep Onset (WASO) did not vary by treatment time or randomization at baseline, but controlled for the baseline value, there was a marginal interaction of randomization with time of treatment at the end of treatment (P < 0.08), perhaps indicating that morning bright light tended to decrease WASO. However, the number of awakenings during the night and the sleep latency did not vary significantly either at the beginning or end of treatment by randomization or treatment timing.
Mood improvement
Mean mood scores for the different groups at each measurement point are shown in Tables 3 and 4. There was little improvement in mood after 1/2 night of wake therapy (Table 3). Seventy-one subjects had some usable actigraphic data for week one and week two, but 10 were missing part of these data. Approximately 85% of the subjects phoned in every 30 minutes during the wake therapy period. Nine of these 71 subjects either did not attempt to wake early or were not able to remain awake for a significant portion of the planned wake therapy period. According to actigraphic scoring, twenty-four of 71 (34%) were able to remain 100% awake for the entire wake therapy period (4 hours); however, only 12 of these (17 % of the total) also remained 100% awake for the rest of the day. The mean percentage awake during wake therapy period for those attempting wake therapy was 87.5% (SD = 15.9, N = 62), whereas the mean percentage awake on a comparison night in the previous (baseline) week was 39.1% (SD = 20.8, N = 61). The mean percentage awake for the full wake therapy day (20 hours) was 88.2% (SD = 32.9, N = 61). There was no significant Spearman correlation of wake therapy compliance (percent awake during the wake therapy period or during the full wake therapy day) with the change in mood from baseline to the morning after wake therapy. Likewise, there was no significant correlation between wake therapy compliance and GDS mood score changes one week later, either for the entire sample or for bright or dim light-treated groups separately.
Table 3 GDS Scores by Week by Light Condition and Treatment Timing: Mean (SD) N
Time Light Morning Mid-day Evening Total
Bright Dim Bright Dim Bright Dim Bright Dim
Baseline start 21.23 (6.08) 13 21.00 (4.47) 15 18.73 (2.91) 15 19.50 (4.43) 16 18.54 (5.63) 13 17.00 (4.27) 16 19.46 (5.01) 41 19.50 (4.56) 40
End of baseline week 21.77 (5.85) 13 20.33 (4.53) 15 18.93 (3.41) 15 19.63 (4.83) 16 18.62 (5.61) 13 17.89 (5.09) 9 19.73 (5.07) 41 19.50 (4.74) 40
After wake therapy 21.67 (5.68) 9 19.43 (6.73) 14 18.00 (5.31) 12 18.00 (5.85) 13 16.60 (5.97) 10 17.14 (4.14) 7 18.61 (5.82) 31 18.41 (5.85) 34
Treatment Week 1 20.00 (5.55) 13 19.33 (7.69) 15 18.20 (4.06) 15 16.94 (3.99) 16 15.08 (8.41) 13 15.11 (5.11) 9 17.78 (6.35) 41 17.43 (5.95) 40
Treatment Week 2 20.08 (6.22) 13 18.60 (7.60) 15 16.21 (5.69) 14 14.38 (5.30) 16 14.75 (8.07) 12 12.33 (5.45) 9 17.05 (6.87) 39 15.50 (6.65) 40
Treatment Week 3 18.85 (6.50) 13 18.50 (6.44) 14 15.00 (5.41) 15 14.81 (5.97) 16 13.23 (6.66) 13 13.00 (7.05) 9 15.66 (6.45) 41 15.72 (6.61) 39
Treatment Week 4 17.38 (6.96) 13 18.67 (7.79) 15 14.07 (6.06) 15 13.81 (5.60) 16 11.77 (7.00) 13 12.22 (6.53) 9 14.39 (6.88) 41 15.28 (7.07) 40
Two-Week follow-up 16.25 (7.96) 12 16.00 (7.72) 11 13.69 (5.41) 13 11.85 (5.80) 14 11.80 (6.66) 10 12.13 (8.41) 8 14.03 (6.77) 35 13.30 (7.18) 33
Four-Week follow-up 17.25 (6.54) 8 14.00 (9.03) 9 12.77 (5.69) 13 13.08 (5.27) 13 13.55 (6.25) 11 10.25 (8.40) 8 14.16 (6.18) 32 12.60 (7.29) 30
3-Month follow-up 14.00 (9.47) 8 18.14 (9.58) 7 15.91 (5.26) 11 11.21 (5.74) 14 16.29 (6.97) 7 11.60 (9.63) 5 15.42 (6.99) 26 13.15 (7.97) 26
Table 4 HDRS Scores by Week by Light Condition and Treatment Timing: Mean (SD) N
Time Light Morning Mid-day Evening Total
Bright Dim Bright Dim Bright Dim Bright Dim
Self Report HDRS 17-Baseline 19.76 (8.46) 13 19.20 (6.05) 15 16.80 (6.12) 15 17.31 (5.53) 16 14.00 (6.22) 13 15.22 (3.38) 9 16.85 (7.18)41 17.55 (5.44) 40
Self Report HDRS 17-Final 15.46 (8.49) 13 13.27 (5.78) 15 10.87 (5.87) 15 10.38 (4.79) 16 6.92 (4.13) 13 9.55 (7.92) 9 11.07 (7.12) 41 11.27 (6.02) 40
Blind HDRS 17-Baseline 21.14 (7.86) 7 21.60 (6.35) 10 18.50 (5.40) 10 18.20 (7.30) 10 15.13 (8.18) 8 16.40 (3.78) 5 17.96 (6.95) 27 18.89 (6.35) 27
Blind HDRS 17-Final 16.00 (8.74) 7 11.80 (5.47) 10 10.70 (4.74) 10 12.30 (6.26) 10 6.62 (5.24) 8 9.20 (7.60) 5 11.15 (7.01) 26 11.38 (5.97) 26
Subjects' moods improved under both treatments at all times of day by treatment week 4. Combining the 3 times of day, GDS scores improved from 19.46 to 14.39 in those treated with bright light and from 19.50 to 15.28 in those treated with placebo dim light. However, there were no significant differences in treatment effects nor were there time-by-treatment interactions. There was no significant treatment effect for any treatment time. The average HDRS17 (extracted from the self-rated SIGH-SAD-SR) improved by 6 points. Combining the 3 times of day, HDRS17 scores improved from 16.85 to 11.07 in those treated with bright light and from 17.55 to 11.28 in those treated with dim light. Again there were no significant treatment effects or time-by-treatment interactions. Blind HDRS17 ratings, when available, were consistent with the self-rating (HDRS17) scores. A power analysis indicated approximately 81% power to detect an effect size of 0.32 in either the GDS or self-rated HDRS.
Adverse reactions
Participants experienced no psychiatric hospitalizations, suicide attempts, or deaths during the study. However, one participant who dropped out while receiving bright light treatment died in the hospital due to late stage emphysema, 3 months after leaving the study. There were no incidents of mania or hypomania during the light treatment.
The weekly SAFTEE physical symptom inventory was examined for adverse reactions to both light treatments using Wilcoxon's Signed Rank Test. To improve the stability of measurement, the 94 individual symptoms were grouped into 17 SAFTEE-defined categories. The results of these group tests are contained in Table 5. The symptom groups for Head, and Other improved with bright light while Urination worsened, and with dim light, Mouth and Teeth improved while Urination, and Genital/Sexual Functioning worsened.
Table 5 SAFTEE Symptoms, Mean Scores for Beginning and End of Light Treatment with Wilcoxon Signed Ranks Test
Light Condition Bright Dim
Symptom Category Baseline Post Treatment p Baseline Post Treatment p
Head 5.62 4.92 .012 5.35 5.14 .065
Eyes 8.53 8.61 .404 8.76 8.27 .168
Ears 5.15 5.22 .791 5.14 5.53 .502
Mouth/Teeth 8.48 7.82 .193 8.41 7.28 .001
Nose/Throat 6.71 7.03 .297 6.58 6.12 .312
Chest 8.17 7.74 .683 7.71 8.00 .315
Heart 2.28 2.19 .276 2.69 2.55 .106
Stomach/Abdomen 6.03 5.46 .080 5.37 5.00 .239
Bowel 8.64 8.40 .487 8.39 8.32 .790
Appetite 7.91 7.58 .394 7.69 7.66 .699
Urination 5.79 7.11 <.001 6.28 8.18 <.001
Gynecology 9.40 9.00 .854 8.91 9.75 .655
Genital/Sexual 8.31 9.64 .077 7.85 9.87 .004
Muscle/Bone 5.97 5.85 .573 5.25 4.94 .178
Walking/Moving 7.06 7.36 .271 7.21 7.74 .239
Scalp/Skin 5.76 5.70 .636 5.39 5.53 .120
Other 28.73 23.4 <.001 27.32 25.17 .091
Testing with the Mann-Whitney U-test showed there were no significant differences in change scores (baseline minus post treatment) of the 94 individual SAFTEE items between bright and dim treatment. Of the 17 symptom groups, only one showed a significant treatment difference, a greater improvement in "Other" in the bright light group than in the dim light controls, (N = 47, p = 0.04, uncorrected for multiple testing.) The "Other" category contains a large number of mood related items (see Table 6).
Table 6 Mean SAFTEE "Other" Scores Before and After Light Treatment
Other, Baseline average – Bright 28.73
Other, Baseline average – Dim 27.32
Other, End-of-treatment average – Bright 23.39
Other, End-of-treatment average – Dim 25.17
Mann-Whitney (N = 47, p = 0.04, uncorrected for multiple testing)
Measures of urinary aMT6s and saliva melatonin
The aMT6s mesors were not significantly different between patients and controls nor between genders, but a marginal interaction was observed in log10 [mesor] between gender and patient vs. control (DF = 1,88, F = 4.3, p = 0.043), with male patients having over 4 times the aMT6s mesor, whereas female controls had slightly higher mesors. Baseline salivary melatonin onsets and offsets and aMT6s acrophases, onsets and offsets tended to be later among patients than controls, but none of these differences were significant after adjustment for age and gender, nor were durations of secretion or excretion different. At baseline, aMT6s acrophases did not differ by treatment or time of treatment or their interaction, nor did acrophase shifts from baseline to the end of treatment. However, in females aMT6s acrophase was more phase advanced during randomized treatment (p < 0.05). Salivary melatonin offsets (not onsets) showed significant treatment, time of treatment, and interaction effects (greatest advance in morning-treated subjects receiving bright light), but the high-quality usable cases in each cell were as few as 2 (See Table 7). The Horne-Östberg Morningness-Eveningness scores for patients and healthy controls are reported in Table 7.
Table 7 Timing of Saliva Melatonin and Urinary aMT6s at Baseline Mean (SD) N
Times are in decimal hours
Saliva Melatonin Depressed Healthy Controls Sig. (T-test, p)
Onset 21.48 (1.67) 28 20.36 (1.98) 22 0.16
Offset 8.76 (2.14) 32 8.78 (1.22) 16 0.873* unequal variance
Duration 11.19 (1.64) 25 11.75 (1.75) 15 0.574
Urinary aMT6s
Onset 23.95 (3.15) 48 22.22 (2.44) 28 0.015
Offset 9.72 (2.82) 48 8.70 (2.02) 28 0.099
Duration 9.75 (2.48) 48 10.48 (2.21) 28 0.201
Acrophase 4.25 (2.56) 49 3.36 (1.75) 28 0.107
Mesor (ng/hour) 652.12 (911.31) 49 485.40 (377.80) 28 0.36
Horne-Östberg 54.75 (11.45) 81 67.65 (9.38) 23 < 0.001
Correcting circadian phases
Testing the hypothesis that light treatment reduces dispersion of circadian phases did not show any statistically significant difference in the pre- and post-treatment deviations (d1 and d2) from group median acrophases for aMT6s, sleep or activity. Light acrophase dispersion showed a significant effect of treatment time (p = .017). The difference in light acrophases was independent of the light condition (dim vs. bright). Therefore, it was more consistent with spontaneous regression than any treatment effect.
Sleep-wake acrophases
For the five-week study period, regression slopes were calculated for the acrophase changes in light, activity and sleep. These slopes were compared (ANOVA) to differences between the week 1 and week 5 acrophases. The regression slope of the 5 weeks was similar to the week 1 minus week 5 results (data not shown).
There was a significant (p = 0.001) advance in illumination acrophase across the three treatment times, from baseline to the end of treatment, for the bright light group and not for the dim light group. Overall, the bright and dim light conditions did not differ significantly in light acrophase. The illumination acrophase change, week 1 minus week 5, showed a significant interaction between treatment time and treatment condition (p = 0.001) in the expected directions: that is, the group treated with bright light in the morning advanced acrophases more than the placebo morning group, whereas this did not occur with evening treatment. This demonstrates that light acrophase shifted in the expected direction in response to the morning bright light treatment. This change in light acrophase was associated with significant differences in the activity (p = 0.003) and sleep (p = 0.001) acrophase shifts between the bright and dim treatment groups but no significant treatment versus time interaction.
Although an objective measure of compliance was not determined, observation of the light records from the Actillume data suggested a high degree of compliance with bright light treatment. Dim light treatment could not be detected from the Actillume data. The self-reports of treatment time and duration were almost entirely consistent with instructions. The low dropout rates and the compliance of participants with wearing Actillumes and completing logs and questionnaires also suggested that compliance with treatment was probably high.
Healthy controls
The healthy control group was active earlier, slept earlier, and reported significantly greater morningness but received less light than the depressed group at baseline (see Table 9).
Table 9 Baseline Activity, Light, and Sleep Comparing Depressed and Control Groups
Measure ACTIVITY LIGHT SLEEP
Baseline MESOR ACROPHASE DEGREES MESOR ACROPHASE DEGREES MESOR ACROPHASE DEGREES
Depressed Mean 12.85 217.63 0.99 211.83 0.27 54.53
N 78 78 78 78 78 78
S.D. 4.46 26.05 0.26 19.47 0.08 26.05
Healthy Mean 14.54 200.44 0.76 206 0.26 37.54
N 28 28 28 28 28 28
S.D. 3.94 15.66 0.28 9.52 0.05 13.65
ANOVA Sig. (p) 0.079 0.001 < 0.001 0.132 0.704 0.001
Discussion
Apart from an advantage on one scale of the SAFTEE side effects inventory (which would not be significant by Bonferroni criteria), bright light treatment had no advantage over dim placebo light. The beneficial effects found in the study might be attributed to several factors that were common to the treatment and control groups. The "placebo" effect, chiefly positive expectations, positive staff contacts, and spontaneous remission may have contributed to positive responses. Weekly visits, even with minimal social interaction, could have a positive effect. In addition, the social structure and regularized sleep provided by the protocol might be beneficial. An hour a day engaging in a treatment, thought to be helpful, may have induced a reduction in depressive symptoms. Participants with ongoing pharmacologic or psychotherapeutic treatment (possibly associated with greater severity or chronicity of illness) improved somewhat less than other participants, so such treatment was probably not a substantial confound.
As a result of the phase-typing treatment assignment, a group going to bed and arising later were assigned to morning light, and a group going to bed earlier were assigned to evening light, with the midday group intermediate. There was little evidence for the anticipated effects of the bright light treatment, which had been expected to alter sleep timing. Likewise, the effects of bright light treatment on melatonin phase were modest. Therefore, we would suspect either that these depressed elders were resistant to bright light effects or that we were less successful in obtaining compliance than we realized.
Even though this study is among the longest treatment trials investigating the effects of light on depression, it may be that the duration of treatment was not sufficient for this age group. A longer period may be needed for an elderly population with chronic depressive symptoms. Although the use of home based treatment may have given rise to less treatment compliance, these subjects did not require hospitalization clinically. Inpatient studies of light treatment have demonstrated antidepressant effects. To be cost effective for patients with milder depressions, light treatment must be feasible in the home. If compliance was a problem in our trial, it would probably be a similar problem in clinical application of this treatment for this age group.
The antidepressant response to placebo light treatment in this study was similar to that reported for placebo in drug studies [52]. However, placebo response is significant and needs to be considered as a confound in antidepressant light treatment studies. Removing placebo responders is common in pharmacology trials but has not commonly been practiced previously in light studies. Removing placebo responders from sample groups may decrease the placebo effects in the randomization portion of clinical trials, but even so the subsequent placebo effects were quite large. It is not evident that removal of placebo-responders was advantageous in this design. Exclusion of bipolar participants effectively prevented hypomanic adverse events, but possibly bipolar depressives are more light responsive.
Treatments were balanced by age and severity of depression, validating that the stratification randomization procedures were successful. The design was successful in balancing expectations between bright light and placebo. The similarity of blind-observer HDRS ratings with self-rating results tends to reduce concern with the blinding issue.
Only a small minority of elderly participants in this study were able to wake themselves up in the middle of the night and remain entirely awake until the next bedtime at home. Because study participants had a very low dropout rate and complied very well with other aspects of the protocol, it would appear that the poor compliance was due to the difficulty of the assignment rather than to poor motivation. The small number of subjects who were able to successfully remain awake for the entire prescribed time may explain why there was no detected antidepressant benefit of the wake therapy. The literature suggests that even a short nap may reduce the mood improvement obtained from wake therapy [53]. Unfortunately, in this age group, the tendency to fall asleep during the day was surprisingly strong [54,55]. Home wake therapy may be too difficult for elderly depressed patients. In addition, there is not much of evidence that sleep deprivation is useful in this age group [56,57]. Indeed, a recent study suggested that in this age group, sleep deprivation may actually interfere with antidepressant treatment [58], which is a possible explanation for the lack of success of this trial.
Tables 7 and 9 indicate that the rhythms of melatonin, sleep, and activity all peaked later (were delayed) as compared to the normal control group. Although this finding is not consistent with the older theory that aging depressives are phase-advanced, it is generally consistent with more modern results in post-menopausal women [32,59]. Table 9 indicates that mesor illumination (the daily amount of illumination) was significantly higher in the depressed participants than controls, which indicates that the relative baseline phase delay of depressives cannot be ascribed to below-average light exposure. However, since increased light exposure is generally associated with more advanced rhythms (as was observed comparing the bright light and placebo groups in this study), the results are consistent with the possibility that these depressed patients were subsensitive to circadian effects of light. The data in Table 8 show that the morning and evening bright light treatments tended to have the predicted phase-shifting effects, but only to a quite modest extent not significant in the interaction effects for activity and sleep. The lack of greater phase-shifting effects related to the experimental treatments provides an additional indication that either these depressives may have been subsensitive to bright light or their compliance may have been less than we realized.
Table 8 Acrophase Changes (Minutes) by Light Condition and Treatment Timing: Mean (SD) N
Acrophase Difference (minutes) Baseline minus Final Morning Evening
Bright Dim Bright Dim
Illumination 80.69 (69.14) 9 1.53 (39.72) 14 -41.99 (63.48) 11 -28.77 (56.96) 7
Activity 25.84 (40.35) 8 -32.42 (60.00) 13 -0.96 (34.46) 11 -40.88 (48.03) 7
Sleep 39.59 (34.81) 9 -25.90 (42.96) 12 -3.80 (34.74) 11 -24.86 (48.32) 7
ANOVA Sig. (p) Illumination Activity Sleep
Treatment Condition 0.485 0.003 0.001
Treatment Time 0.001 0.485 0.243
Condition by Time 0.001 0.128 0.054
Note: Positive change is earlier, negative change is later, at the end of treatment.
There was considerable overlap in baseline activity and aMT6s phases among the groups assigned to morning, mid-day, and evening bright light. Thus, our prospective phase-typing was imperfect. Nevertheless, examining the subgroups who were phase-typed correctly did not indicate any better responses to the bright light treatment.
Conclusion
Antidepressant response to bright light treatment in this age group was not statistically superior to placebo. No timing of bright light treatment was significantly better than placebo or consistently better in self-ratings and blind ratings. There was minimal evidence for a relative reduction of diverse SAFTEE complaints in the bright-light-treated group. Contrary to what has been suggested by previous light studies [15,27], the trend was for evening light to appear at least as successful as morning or mid-day light. Were it a statistically significant difference, an advantage of evening light would be unexpected in a group of depressed participants who are generally more phase delayed than healthy individuals. We do not know whether the lack of light response found in this study might be associated with the age of the subject group, the relative mildness and chronicity of their symptoms, inadequate duration of bright light treatment, perhaps undetected compliance problems occurring in home treatment, or some aspect of the trial design such as the sleep deprivations. Considering the general evidence that light treatment is useful for depression, including in this age group, further testing with altered designs might be fruitful.
Upon early termination of this study, a trial of green light was initiated with somewhat more promising results. These results are presented in a separate report 60.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
RTL coordinated and carried out the clinical trial, performed statistical analyses, and drafted the manuscript. DFK conceived and drafted the design, administered and participated in data collection, and participated in statistical analyses and manuscript preparation. JAE carried out the immunoassays and contributed to statistical analyses and manuscript preparation. NCK carried out subject recruitment and collection and scoring of Actillume recordings and subject questionnaires. MAG constructed a software database and administered blind HDRS ratings. All authors reviewed and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors acknowledge the Sunbox company of Gaithersburg, Maryland for donation of the light fixtures used in this research. Mary Ann Mowen worked with participants. Appreciation and acknowledgment is extended to the volunteers without whom this research would not have been possible. Supported by the NIH National Institute on aging (AG12364, MH68545, and HL07123).
==== Refs
Chilcott LA Shapiro CM The socioeconomic impact of insomnia PharmacoEconomics 1997 10 1 14 10163422
Ormel J VonKorff M Ustun B Pini S Korten A Oldehinkel T Common mental disorders and disability across cultures JAMA 1994 272 1741 1748 7966922 10.1001/jama.272.22.1741
Walsh JK Engelhardt CL Insomnia Wake Up America: A National Sleep Alert Working Group Reports 1994 2 Washington, DC: US Govt. Printing Office 114 149
Simon GE VonKorff M Prevalence, burden, and treatment of insomnia in primary care Am J Psychiatry 1997 154 1417 1423 9326825
Foley DJ Monjan AA Brown SL Simonsick EM Wallace RB Blazer DG Sleep complaints among elderly persons: an epidemiologic study of three communities Sleep 1995 18 425 432 7481413
McCurren CD Hall LA Rowles GD Community elders: prevalence and correlates of depressive symptoms Clin Nurs Res 2003 2 128 144 8481678
Turvey CL Carney C Arndt S Wallace RB Herzog R Conjugal loss and syndromal depression in a sample of elders aged 70 years or older Am J Psychiatry 1999 156 1596 1601 10518172
Blazer D Burchett B Service C George LK The association of age and depression among the elderly: an epidemiologic exploration J Gerontology 1991 46 M210 M215
Bennett AT Collins KA Elderly suicide: a 10-year retrospective study Am J Forensic Med Pathol 2001 2 169 172 11394752 10.1097/00000433-200106000-00011
Hyman SE Mental health in an aging population: The NIMH perspective Am J Geriatr Psychiatry 2005 9 330 339 11739060 10.1176/appi.ajgp.9.4.330
Ray WA Griffin MR Schaffner W Baugh DK Melton LJI Psychotropic drug use and the risk of hip fracture N Engl J Med 1987 316 363 9 2880292
Ruthazer R Lipsitz LA Antidepressants and falls among elderly people in long-term care Am J Public Health 1993 83 746 750 8484463
Depression Guideline Panel Depression in Primary Care: Treatment of Major Depression 1993 2 Washington, DC: Agency for Health Care Policy and Research, HHS, AHCPR Publication No. 93-0551 U.S. Government Printing Office
Rosenthal NE Gabbard GO Light Therapy Treatment of Psychiatric Disorders 1995 1 Washington, DC: American Psychiatric Press 1263 1273
Tuunainen A Kripke DF Endo T Light Therapy for Non-Seasonal Depression (Cochrane Review) The Cochrane Library 2004
Rosenthal NE Sack DA Gillin JC Lewy AJ Goodwin FK Davenport Y Mueller PS Newsome DA Wehr TA Seasonal affective disorder: a description of the syndrome and preliminary findings with light therapy Arch Gen Psychiatry 1984 41 72 80 6581756
Campbell SS Dawson D Anderson MW Alleviation of sleep maintenance insomnia with timed exposure to bright light J Am Geriatr Soc 1993 41 829 836 8340561
Mishima K Okawa M Hishikawa Y Hozumi S Hori H Takahashi K Morning bright light therapy for sleep and behavior disorders in elderly patients with dementia Acta Psychiat Scand 1994 89 1 7 8140901
van Someren E Kessler A Mirmiran M Swaab DF Indirect bright light improves circadian rest – Activity rhythm disturbances in demented patients Biol Psychiatry 1997 41 955 963 9110101 10.1016/S0006-3223(97)89928-3
Sumaya IC Rienzi BM Deegan JFI Moss DE Bright light treatment decreases depression in institutionalized older adults: A placebo-controlled crossover study Journal of Gerontology: Medical Sciences 2001 56A M356 M360
Moffit MT Bright light treatment of late-life depression PhD Thesis 1993 The Wright Institute Graduate School of Psychology
Tsai Y-F Wong TKS Juang Y-Y Tsai H-H The effects of light therapy on depressed elderly Int J Ger Psych 2004 19 545 548 10.1002/gps.1125
Lewy AJ Sack RL Singer CM White DM Hoban TM Winter depression and the phase-shift hypothesis for bright light's therapeutic effects: history, theory, and experimental evidence J Biol Rhythms 1988 3 121 134 2979635
Sack RL Lewy AJ White DM Singer CM Fireman MJ Vandiver R Morning vs evening light treatment for winter depression Arch Gen Psychiatry 1990 47 343 351 2322085
Avery DH Khan A Dager SR Cohen SA Cox GB Dunner DL Morning or evening bright light treatment of winter depression? The significance of hypersomnia Biol Psychiatry 1991 29 117 126 1995083 10.1016/0006-3223(91)90040-S
Terman JS Terman M Lo ES Cooper TB Circadian time of morning light administration and therapeutic response in winter depression Arch Gen Psychiatry 2001 58 69 75 11146760 10.1001/archpsyc.58.1.69
Terman M Terman JS Quitkin FM McGrath PJ Stewart JW Rafferty B Light therapy for seasonal affective disorder: A review of efficacy Neuropsychopharmacol 1989 2 1 22 10.1016/0893-133X(89)90002-X
Terman M Terman JS Randomized, controlled trial of bright light, dawn simulation and negative air ionization for winter depression [abstract] Soc Light Treatment Bio Rhythms Abst 2002 14 25
Terman M Wetterberg L Problems and prospects for use of bright light as a therapeutic intervention Light and Biological Rhythms in Man 1993 1 Stockholm: Pergamon Press 421 436
Tuunainen A Kripke DF Elliott JA Assmus JD Rex KM Klauber MR Langer RD Depression and endogenous melatonin in postmenopausal women J Affect Dis 2002 69 149 158 12103461 10.1016/S0165-0327(01)00303-2
Wehr TA Duncan WC Sher L Aeschbach D Schwartz PJ Turner EH Postolache TT Rosenthal NE A circadian signal of change of season in patients with seasonal affective disorder Arch Gen Psychiatry 2001 58 1108 1114 11735838 10.1001/archpsyc.58.12.1108
Sekula LK Lucke JF Heist K Czambel K Rubin RT Neuroendocrine aspects of primary endogenous depression XV: mathematical modeling of nocturnal melatonin secretion in major depressives and normal controls Psychiatry Res 1997 69 143 153 9109182 10.1016/S0165-1781(96)02937-X
Lewy AJ Sack RL Singer CM Treating phase typed chronobiologic sleep and mood disorders using appropriately timed bright artificial light Psychopharmacol Bull 1985 21 368 372 4034852
Thalen B-E Kjellman BF Morkrid L Wibom R Wetterberg L Light treatment in seasonal and nonseasonal depression Acta Psychiatr Scand 1995 91 352 360 7639093
Kripke DF Risch SC Janowsky DS Bright white light alleviates depression Psychiatry Res 1983 10 105 112 6581487 10.1016/0165-1781(83)90109-9
Kripke DF Risch SC Janowsky DS Lighting up depression Psychopharmacol Bull 1983 19 526 530
Prasko J Foldmann P Praskova H Zindr V Hastened onset of the effect of antidepressive drugs when using three types of timing of intensive white light Cs Psychiat 1988 84 373 383
Neumeister A Goessler R Lucht M Kapitany T Bamas C Kasper S Bright light therapy stabilizes the antidepressant effect of partial sleep deprivation Biol Psychiatry 1996 39 16 21 8719121 10.1016/0006-3223(95)00086-0
Loving RT Kripke DF Shuchter SR Bright light augments antidepressant effects of medication and wake therapy Depression and Anxiety 2002 16 1 3 12203667 10.1002/da.10036
Bloching B Dechene C Taschner KL Outlasting antidepressant effect of late partial sleep deprivation by bright light therapy J Sleep Res 2000 9 21
Yesavage JA Brink TL Rose TL Lum O Huang V Adey M Leirer VO Development and validation of a geriatric depression screening scale: a preliminary report J Psychiatr Res 1983 17 37 49 7183759 10.1016/0022-3956(82)90033-4
Horne JA Ostberg O A self-assessment questionnaire to determine morningness-eveningness in human circadian rhythms Int J Chronobiol 1976 4 97 110 1027738
First MB Spitzer RL Gibbon M Williams JBW Structured Clinical Interview for DSM-IV Axis I Disorders – Non-patient Edition (SCID-I/NP, Version 20) 1996 New York City: New York State Psychiatric Institute
Lyness JM King DA Cox C Yoediono Z Caine ED The importance of subsyndromal depression in older primary care patients: prevalence and associated functional disability J Am Geriatr Soc 1999 47 647 652 10366161
Kripke DF Timing of phototherapy and occurrence of mania Biol Psychiatry 1991 29 1156 1157 1873376 10.1016/0006-3223(91)90258-N
Jean-Louis G Kripke DF Cole RJ Assmus JD Langer RD Sleep detection with an accelerometer actigraph: comparisons with polysomnography Physiol Behav 2001 72 21 28 11239977 10.1016/S0031-9384(00)00355-3
Jean-Louis G Kripke DF Mason WJ Elliott JA Youngstedt SD Sleep estimation from wrist movement quantified by different actigraphic modalities J Neurosci Methods 2001 105 185 191 11275275 10.1016/S0165-0270(00)00364-2
Eastman CI What the placebo literature can tell us about light therapy for SAD Psychopharmacol Bull 1990 26 495 504 2087542
Brainard GC Hanifin JP Greeson JM Byrne B Glickman G Gerner E Rollag MD Action spectrum for melatonin regulation in humans: Evidence for a novel circadian photoreceptor J Neurosci 2001 21 6405 6412 11487664
Moynihan C Instruction Manual for Systematic Assessment for Treatment Emergent Events (SAFTEE) 1983 Rockville, MD: ADAMHA
Williams JBW Link MJ Rosenthal NE Amira L Terman M Structured Interview Guide for the Hamilton Depression Rating Scale: Seasonal Affective Disorder Version 2002 New York, New York State Psychiatric Institute 1 9
Khan A Warner HA Brown WA Symptom reduction and suicide risk in patients treated with placebo in antidepressant clinical trials Arch Gen Psychiatry 2000 57 311 328 10768687 10.1001/archpsyc.57.4.311
Hemmeter U Bischof R Hatzinger M Seifritz E HolsboerTrachsler E Microsleep during partial sleep deprivation in depression Biol Psychiatry 1998 43 829 839 9611673 10.1016/S0006-3223(97)00297-7
Yoon I Youngstedt SD Kripke DF Elliott JA Evening naps of older people in relation with early morning awakening [abstract] Sleep 2002 25 A58 A59
Yoon I Kripke DF Elliott JA Naps and circadian rhythms in postmenopausal women [abstract] Sleep 2003 26 A159
Grube M Sleep deprivation and aging Z Gerontol Geriatr 1999 32 364 367 10552175 10.1007/s003910050128
Cole MG Muller HF Sleep deprivation in the treatment of elderly depressed patients Journal of the American Geriatric Society 1976 24 308 313
Reynolds CF Smith GS Dew MA Mulsant BH Miller MD Schlernitzauer M Stack JA Houck PR Pollock BG Accelerating symptoms-reduction in late-life depression A double-blind, randomized, placebo-controlled trial of sleep deprivation Am J Geriatr Psychiatry 2005 13 353 358 15879583 10.1176/appi.ajgp.13.5.353
Tuunainen A Kripke DF Elliott JA Assmus JD Rex KM Klauber MR Langer RD Depression and endogenous melatonin in postmenopausal women J Affect Disord 2001 1 1 10
Loving RT Kripke DF Elliot JA Knickerbocker NC Grandner MA Bright light treatment of depression for older adults [ISCRTN69400161] BMC Psychiatry
|
16283925
|
PMC1298312
|
CC BY
|
2021-01-04 16:33:02
|
no
|
BMC Psychiatry. 2005 Nov 9; 5:41
|
utf-8
|
BMC Psychiatry
| 2,005 |
10.1186/1471-244X-5-41
|
oa_comm
|
==== Front
Biomed Eng OnlineBioMedical Engineering OnLine1475-925XBioMed Central London 1475-925X-4-641627114110.1186/1475-925X-4-64ResearchFluid-structure interaction in abdominal aortic aneurysms: effects of asymmetry and wall thickness Scotti Christine M [email protected] Alexander D [email protected] Satish C [email protected] Ender A [email protected] Biomedical Engineering Department, Carnegie Mellon University, Pittsburgh, Pennsylvania, USA2 Department of Mathematical Sciences, Carnegie Mellon University, Pittsburgh, Pennsylvania, USA3 Division of Vascular Surgery, Allegheny General Hospital, Pittsburgh, Pennsylvania, USA4 Institute for Complex Engineered Systems and Biomedical Engineering Department, Carnegie Mellon University, Pittsburgh, Pennsylvania, USA2005 4 11 2005 4 64 64 15 8 2005 4 11 2005 Copyright © 2005 Scotti et al; licensee BioMed Central Ltd.2005Scotti 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
Abdominal aortic aneurysm (AAA) is a prevalent disease which is of significant concern because of the morbidity associated with the continuing expansion of the abdominal aorta and its ultimate rupture. The transient interaction between blood flow and the wall contributes to wall stress which, if it exceeds the failure strength of the dilated arterial wall, will lead to aneurysm rupture. Utilizing a computational approach, the biomechanical environment of virtual AAAs can be evaluated to study the affects of asymmetry and wall thickness on this stress, two parameters that contribute to increased risk of aneurysm rupture.
Methods
Ten virtual aneurysm models were created with five different asymmetry parameters ranging from β = 0.2 to 1.0 and either a uniform or variable wall thickness to study the flow and wall dynamics by means of fully coupled fluid-structure interaction (FSI) analyses. The AAA wall was designed to have a (i) uniform 1.5 mm thickness or (ii) variable thickness ranging from 0.5 – 1.5 mm extruded normally from the boundary surface of the lumen. These models were meshed with linear hexahedral elements, imported into a commercial finite element code and analyzed under transient flow conditions. The method proposed was then compared with traditional computational solid stress techniques on the basis of peak wall stress predictions and cost of computational effort.
Results
The results provide quantitative predictions of flow patterns and wall mechanics as well as the effects of aneurysm asymmetry and wall thickness heterogeneity on the estimation of peak wall stress. These parameters affect the magnitude and distribution of Von Mises stresses; varying wall thickness increases the maximum Von Mises stress by 4 times its uniform thickness counterpart. A pre-peak systole retrograde flow was observed in the AAA sac for all models, which is due to the elastic energy stored in the compliant arterial wall and the expansion force of the artery during systole.
Conclusion
Both wall thickness and geometry asymmetry affect the stress exhibited by a virtual AAA. Our results suggest that an asymmetric AAA with regional variations in wall thickness would be exposed to higher mechanical stresses and an increased risk of rupture than a more fusiform AAA with uniform wall thickness. Therefore, it is important to accurately reproduce vessel geometry and wall thickness in computational predictions of AAA biomechanics.
==== Body
Background
Abdominal aortic aneurysms (AAA) are local enlargements of the aorta that occur preferentially below the renal bifurcation and they represent a socially relevant cardiovascular health disease. A recent study [1] reports that the prevalence of AAA disease is 8.8% in the population above 65 years of age and men are affected more often than women by a ratio of 4:1 [2]. Aneurysms are little-known among the lay public, but they are a significant cause of mortality; fifteen thousand people per year die from AAA rupture in the United States alone, making it the 13th leading cause of death in this country and affecting 1 in 250 individuals over 50 years of age. Since the likelihood of being diagnosed with an aneurysm increases with age, the incidence of aortic aneurysmal disease is expected to increase with the continuously aging population. Aneurysms can be treated surgically; however typical treatment is based on the surgeon's estimation of the risk of rupture and the patient's general fitness for surgery, along with his/her life expectancy. Despite significant improvements in surgical procedures and imaging techniques, the mortality and morbidity rates associated with untreated ruptured AAAs remain very high. AAA disease is a health risk of considerable importance since this kind of aneurysm is mostly asymptomatic until its rupture, which is frequently a lethal event with an overall mortality rate in the 80% to 90% range [3]. The optimal strategy is clear: prevention of aneurysm rupture is the primary goal in management of aneurysmal disease.
Deciding between elective aneurysm repair and conservative management of the disease is difficult due to the lack of a reliable predictor of rupture risk. A critical AAA transverse diameter of 5 to 6 cm is the most common threshold value used clinically to recommend surgical repair or endovascular intervention [4,5]. However, small aneurysms can also rupture and the overall mortality associated with these may exceed 50% [6]. Therefore, ideally, the decision to repair an aneurysm should not be guided by maximum transversal dimension alone, but rather by a more reliable criterion associated with the actual rupture potential of the patient-specific artery, such as peak AAA wall stress and strength [7]. Since aneurysm rupture is a phenomenon that occurs when the mechanical stress acting on the dilating inner wall exceeds its failure strength, a criterion for repair based upon quantifying aneurysm stress and strength could facilitate a better method to determine at-risk AAAs. Unfortunately, there is no current method of obtaining in vivo measurements of tissue stresses or strength. However, mathematical and computational models can be utilized to predict the fluid and solid mechanics environments within aneurysmal aortas.
AAA wall stress is the outcome of several factors, such as characterization of the wall material, the shape and size of the aneurysm sac, the presence of intraluminal thrombus (ILT), and the dynamic interaction of the wall with blood flow. Since the internal mechanical forces are maintained by the dynamic action of blood flowing in the aorta, the quantification of the hemodynamics of AAAs is essential for the characterization of their biomechanical environment. The justification of biomechanics research is based on the well-known fact that both fluid and wall mechanics play an important role in pathologic conditions of blood vessels.
Prior works have examined the computationally predicted and experimentally validated flow patterns within virtual AAA models [8-11], showing the effect of aneurysm asymmetry on the increase in flow-induced wall pressure and wall shear stress. This has led to the use of patient-specific models obtained from diagnostic images that allow the prediction of flow-induced stresses on a single patient basis [12]. Similarly, Di Martino and colleagues [13] provided the notion of interaction between solid and fluid domains as it contributes to aneurysm rupture potential. This interaction between the domains was recently compared with a peak systolic static prediction of wall stresses in the presence of ILT [14]. Fully-coupled fluid-structure interaction (FSI) of the domains allows computation of the flow and pressure fields in the aneurysm, simultaneously with the wall stresses [15]. This methodology provides a method for validating the computational results with clinical diagnostic data, such as Echo Doppler flow visualization. Therefore, it is important to include both the dynamics of blood flow as well as the wall motion response associated with the pulsatile nature of the flow to accurately model the aneurysm. With the continuous improvements in computer architecture processing times, patient-specific computational models in a clinical setting are likely to be used in the near future as a tool for effective decision making in AAA surgical and endovascular repair.
In this work we describe the complex interaction of blood flow and the compliant AAA wall by utilizing a time dependent, fully-coupled FSI methodology to determine the effects of aneurysm asymmetry and wall thickness heterogeneity on the mechanical stresses and vortex dynamics. Ten virtual models were utilized in the study, which aims to provide a non-invasive methodology for quantifying transient AAA wall mechanics. The latter can then be compared with mean AAA wall tissue strength to provide a predictor for rupture potential. Additionally, the FSI technique is compared with quasi-static and transient solid stress analyses as alternative approaches for the reduction of computational processing time.
Methods
AAA geometry
Ten virtual aneurysm models were generated with the CAD software ProEngineer Wildfire (Parametric Technology Corporation, Needham, MA). The models differ in degree of asymmetry and wall heterogeneity, and are comprised by a fluid domain, ΩF, representing the aortic lumen and a solid domain, ΩS, representing the AAA wall. The fluid domain is characterized by circular cross sections parallel to the x-y plane, with an undilated diameter, d = 2 cm, at the inlet and outlet sections and a maximum diameter, D = 3 d, at the midsection of the AAA sac. The asymmetry of the model is governed by the β parameter given by Eq. (1).
where r and R are the radii measured at the midsection of the AAA sac from the longitudinal z-axis to the posterior and anterior walls, respectively, as shown in the inset of Figure 1(a). Thus, β = 1.0 yields an axisymmetric aneurysm. The geometry of the fluid domain is given by Eq. (2), which defines the diameter of each cross section, φ(z), and the deviation of its centroid from the z-axis, δ(z):
Figure 1 CAD geometry for β = 0.6 model: (a) fluid and solid domains with uniform wall thickness (b) solid domain with variable wall thickness.
The geometry of the solid domain is given by an AAA wall with either a (i) uniform thickness (UW) or a (ii) variable thickness (VW). Both types of wall designs model the thickness of the aneurysm as material extruding normally from the surface enclosing the lumen. The UW model has a thickness given by λ = 1.5 mm, while the VW model is given by λ(z) = 1.5d/φ(z) (in mm) at each cross-section. It follows that the local wall thickness varies between 0.5 mm and 1.5 mm (with a mean of 1.0 mm for the AAA sac), inversely proportional to the local diameter of the cross-section. For a ruptured AAA, wall thickness can be as low as 0.23 mm at the rupture site with a surface-wide average of 1.45 mm [16]. The asymmetry of a virtual AAA is given by βε {1.0, 0.8, 0.6, 0.4, 0.2}, thus yielding a total of ten geometries, five of which have a uniform wall thickness (UW) and the other five variable wall thickness (VW). β = 1.0 corresponds to azymuthal symmetry, and β = 0.2 is an AAA for which only the anterior wall is dilated while the posterior wall is nearly flat. Figure 1(a) shows β = 0.6 with fluid domain ΩF entities in a uniform thickness model. Figure 1(b) illustrates qualitatively the variable thickness domain ΩS at three different wall locations for the same β = 0.6 model.
Governing equations and boundary conditions
The governing equations for the fluid domain are the continuity and Navier-Stokes equations with the assumptions of homogenous, incompressible, and Newtonian flow. Since the fluid domain is deformable in an FSI problem, an Arbitrary Lagrangian-Eulerian (ALE) formulation has been adopted. The ALE formulation [17] introduces a moving coordinate system to model the deformation of the fluid domain. The momentum and mass conservation equations governing the flow are given by Eq. (3) in ALE form:
ρfΔ·v = 0 (3b)
where ρf is the fluid density, τf is the fluid stress tensor, are the body forces per unit volume, v is the fluid velocity vector, and is the moving coordinate velocity, respectively. In the ALE formulation, is the relative velocity of the fluid with respect to the moving coordinate velocity. Blood is modeled to have a density ρf = 1.05 g/cm3 and a dynamic viscosity μ = 3.85 cP. The governing equation for the solid domain is the momentum conservation equation given by Eq. (4). In contrast to the ALE formulation of the fluid equations, a Lagrangian coordinate system is adopted:
where ρs is the AAA wall density, τs is the solid stress tensor, are the body forces per unit volume, and is the local acceleration of the solid. The AAA wall is assumed to be an isotropic, linear, elastic solid with a density ρs = 2.0 g/cm3, a Young's Modulus E = 2.7 MPa and a Poisson's ratio υ = 0.45. The wall material implemented in this work represents a tissue of average characteristics for the aneurysmal abdominal aorta, i.e. a linearization of the stress-strain curve as reported in [13]. Previous studies have shown that aneurysm tissue is a non-linear, isotropic, hyperelastic material [18]. Hence, the constitutive linearity of the AAA wall is a simplification of the FSI and stress analyses in this study. In the FSI formulation, the AAA wall is assumed to undergo large displacements and small strains.
The boundary of the fluid domain is divided into the following regions for the assignment of boundary conditions: inlet (), outlet (), and the fluid-structure interaction interface (), as shown in Figure 1(a). The applied boundary conditions on the non-FSI regions are (i) a time dependent fully developed velocity profile on and (ii) a time dependent normal traction (due to luminal pressure) on . These are presented by Eq. (5) as follows:
where τnn is the normal traction, u(t) and p(t) are the time dependent velocity and pressure waveforms shown in Figure 2, designates the normal of the respective boundary, and I is the standard identity matrix. Time dependency, as introduced by u(t) and p(t), is given by Fourier series representations of the waveforms generalized in Eq. (6):
Figure 2 In vivo luminal pulsatile velocity and pressure reproduced from [19]: (a) velocity waveform (b) pressure waveform. Inlet peak systolic flow occurs at t = 0.304 s and outlet peak systolic pressure at t = 0.4 s.
N is the number of harmonics used to reproduce the in vivo measurements of luminal velocity (N = 18), u(t), and pressure (N = 7), p(t), respectively. These waveforms are triphasic pulses appropriate for normal hemodynamics conditions in the infrarenal segment of the human abdominal aorta first reported by Mills et al [19]. The use of an input transient velocity based on normal physiology is justified by the fact that the inlet boundary condition is applied above the proximal neck of the aneurysm, an undilated segment of the abdominal aorta. For average resting conditions, blood flow in the abdominal aorta is generally laminar [20,21]; flow deceleration achieved after peak systole induces laminar disturbed flow conditions and vortex formation even under simulated exercise conditions [22-24]. Inlet peak systolic flow occurs at t = 0.304 seconds and outlet peak pressure at t = 0.4 seconds. The time-average Reynolds number is Rem = 410, which is characteristic of a patient in resting conditions [25]. Rem is calculated as , where is the time-averaged, mean inlet velocity and d is the inlet diameter. The Womersley number, , characterizes the flow frequency ω (ω = 2π/T and T = 1.0 seconds), the geometry and the fluid viscous properties, and is α = 13.1, a typical value for the human abdominal aorta under resting conditions [26]. The amplitude coefficient of the velocity waveform is defined as γ = Repeak/Rem = 5.25.
Figure 1(b) shows the boundary of the solid domain divided into inlet (), outlet () and the fluid-structure interface () regions. The FSI interfaces and are identical, coupling the fluid and solid domains. The boundary conditions on the non-FSI regions of the solid domain impose zero translation on the ends and as given by Eq. (7). This corresponds to completely fixing the ends of the domain, simulating the tethering of the aorta by the surrounding tissue and organs. Numerical experimentation with the objective of minimizing the stresses at the proximal and distal necks yielded placement of the inlet and outlet sections at a distance 1.5d apart from the aneurysm sac.
The boundary condition at the outer wall surface corresponds to a reference zero normal traction, as the peritoneum and surrounding tissues do not exert any significant pressure on the arterial wall. There are no published data on normal forces exerted by internal organs and tissue on the wall abdominal aorta. The final set of boundary conditions is applied to the FSI interfaces and as follows: (i) displacements of the fluid and solid domain FSI boundaries must be compatible, (ii) tractions at these boundaries must be at equilibrium and (iii) fluid obeys the no-slip condition. These conditions are given by Eq. (8):
where d, τ, and are displacement vectors, stress tensors, and boundary normals with the subscripts s and f indicating a property of the fluid and solid, respectively.
In addition to the FSI methodology, we investigated alternative computational solid stress (CSS) techniques for approximating the AAA wall stresses. In these analyses we disregard the blood flow and attempt to obtain comparatively accurate results by applying a spatially-uniform pressure function onto the inner wall. The CSST method (transient CSS) applies the transient function p(t) from Eq. 4(b) to simulate the effect of luminal pressure acting on the inner wall. Similarly, the CSSS method (static CSS) applies p(t = 0.4) in a quasi-static formulation to obtain the stresses at peak systolic pressure. In these two approaches we utilize only the solid domain ΩS as shown in Figure 1(b), with prescribed zero translation at the proximal and distal ends as given by Eq. (7), and with a pressure boundary condition as given by Eq.(9):
Numerical discretization
The software Adina (v8.0, ADINA R&D, Inc., Cambridge, MA) was utilized for the numerical simulation of fluid-structure interaction (FSI) between the wall and the lumen, as well as the alternative CSST and CSSS analyses involving only the aneurysmal wall, as described in [15]. The Finite Element Method (FEM) is used to solve the governing equations, which discretizes the computational domain into finite elements that are interconnected by nodal points. In this work we make use of linear hexahedral, eight-node elements to discretize the fluid and solid domains. The mesh generator Gridgen (Pointwise Inc., Fort Worth, TX) is used to develop the finite element grids. The ten aneurysm models are all composed of 17,280 hexahedral elements (19,093 nodes) for the fluid and 5,760 hexahedral elements (8,784 nodes) for the solid domains. Figure 3 illustrates the fluid and solid meshes for β = 0.6. Mesh sensitivity analyses were conducted with four additional mesh sizes ranging from 12,480 to 44,928 fluid elements and 3,840 to 13,824 solid elements. Independence in mesh size was obtained for the primary variables (velocity components, fluid pressure and structural displacements) within 5% relative error for the 4th mesh (32,256 fluid and 10,752 solid elements). However, the mesh used in the present study was chosen due to its adequate compromise between acceptable CPU simulation times (71 CPU-hours on average) and moderate relative errors of the primary variables at randomly selected nodal points (15% on average). In this regard, it is important to understand that this work is a baseline study for the development of a computational, pre-operative planning tool for physicians, and as such treatment decisions must be made within a reasonable turnaround time. Additionally, mass conservation in the fluid domain was met for a relative error (comparison between volume flow rate at the inlet and outlet sections, and rate of change of volume within the geometry) of 1%.
Figure 3 Computational domain for β = 0.6 models: (a) fluid mesh (b) solid mesh with variable wall thickness.
Equations (3–4) are reduced to weak form by following the standard Galerkin procedure [27]. In short, the fluid domain employs special Flow-Condition-Based-Interpolation (FCBI) hexahedral elements which use constant functions to interpolate velocity and bi-linear functions to interpolate pressure and displacements on each fluid element. The solid domain employs Mixed-Interpolation hexahedral elements, preferred in modeling nearly incompressible media, which use constant functions to interpolate pressure and bilinear functions to interpolate displacements on each solid element. The discretized equations for the fluid and solid elements are assembled into one system of equations, coupling the fluid and solid meshes. A sparse matrix solver based on Gaussian elimination is used for solving this system.
The FSI methodology utilizes an implicit time integration scheme, first applied on the fluid-structure interface of the fluid domain () where the coordinate system of the fluid and solid domains is Lagrangian. The results are then utilized to solve each domain entirely. Pulsatile flow is simulated over five to eight cycles with a time step size ΔtFSI = 5·10-3seconds until periodic convergence is achieved. Figure 4 shows convergence for the fluid domain (4a) and the solid domain (4b) for β = 0.2 in terms of five nodal point values of axial velocity and displacement magnitude, respectively. For the purpose of comparison, Figure 4(c) illustrates the convergence of displacement magnitude for the CSST analysis at the same five selected nodal points. The simulations were performed on a Tru64 Unix operating system using up to eight 1.15 GHz EV7 processors and in-memory computing. The CSS approaches only utilize the solid domain; hence, the final matrix assembly consists of only solid element equations and the computational times are significantly less when compared with the FSI computational times. For a consistent comparison, the CSST system of equations is solved with the same solution methods as in FSI but with a time step size seconds. The CSSS system of equations is solved with the sparse matrix solver for a steady state solution with a constant and uniform pressure p(t = 0.4) applied on the inner AAA wall. The computational times for the CSSS simulations are negligible in comparison with the CSST (3 CPU-hours on average) and FSI (71 CPU-hours on average) simulations. A fully-coupled FSI method is computationally more expensive due to the memory required to adapt the matrices for a moving mesh algorithm and, thus, must be weighed against the clinical benefit of obtaining a complete characterization of the flow dynamics and wall stresses of the aneurysm sac.
Figure 4 Time periodic convergence plot for β = 0.2: (a) velocity in the z-direction, (b) displacement for the FSI analysis and (c) displacement for the CSST analysis. The insets show a schematic of the location of the nodal points used for the time convergence studies.
Results and discussion
Asymmetry effect
The blood flow dynamics in aneurysm models is governed by the relative compliance of the vessel, determined by its non-homogenous shape (asymmetry and wall thickness) and material characterization of the wall. Figure 5 shows the flow patterns for β = 0.2 and β = 1.0 at t = 0.4 s (peak systolic pressure in reference to the outlet normal traction boundary condition) for uniform wall thickness. The velocity vectors illustrate a streamlined profile absent of vortices, a flow path customarily associated with a condition of systolic acceleration. This flow characterization is significantly different from that described in [11] for the most asymmetric rigid AAA model (β = 0.3), since the energy stored by the expanding compliant vessel during systole ejects the vortex downstream shortly after peak flow. The phase delay (0.096 seconds) between the inlet velocity and outlet pressure waveforms also accounts for this difference, i.e. the flow has begun to temporally decelerate at the time the wall is fully expanded due to systolic pressure.
Figure 5 Velocity vectors and fluid pressure at the wall at t = 0.4 s in y-z plane for (a) β = 1.0 and (b) β = 0.2 uniform wall (UW) thickness models.
The wall pressure distribution between the two models is nearly identical at the longitudinal cross-section given by the YZ-plane. As the β = 0.2 model contains the same volume of fluid in the aneurysm sac this result is not entirely unexpected. Moreover, the reversal of the pressure gradient, given by a lower pressure at the inlet rather than at the outlet, signifies the previously stated effect of a phase shift on the flow dynamics, as well as the influence of the opposing pressure waveform imposed at the outlet. With the onset of diastolic flow conditions, the inlet velocity begins to decelerate, reducing the overall pressure gradient. As the cardiac cycle continues its course this pressure drop is expected to decrease, yielding flow reversal and recirculation regions dominated by convective effects, i.e. local acceleration due to bulging of the sac and asymmetry.
An assessment of the effect of asymmetry on flow patterns downstream of the AAA sac shows that for uniform wall thickness, the vortex dynamics generated for β = 0.2 and β = 1.0 is almost identical at the same temporal stages of the flow cycle. The vortices that develop and dissipate in the aneurysm, however, remain in the midsection to distal end of the sac for β = 1.0, whereas β = 0.2 is subject to stronger recirculating flows and the vortices travel upstream towards the proximal end of the aneurysm, particularly in diastole (t ≥ 0.6 s). This vortex translation along the anterior wall is the effect of retrograde flow caused by the velocity-pressure waveform phase shift and the instantaneous flow reversal experienced during diastole. The asymmetric geometry of the β = 0.2 model magnifies this effect, allowing for vortices to remain longer along the bulging anterior wall due to local flow deceleration. Conversely, the symmetric model has a reduced curvature along the anterior wall and the vortices develop and dissipate more readily as convective effects are weaker.
Vortex growth inside the AAA sac creates favorable conditions for increased platelet deposition rates, thrombus formation, and an increased risk of rupture [28]. The effect of aneurysm flow dynamics on the wall mechanics can be determined by predictions of Von Mises stress, an energetic formulation adopted in lieu of representing the nine components of the second order stress tensor. This is a stress quantity used in the field of failure mechanics, which characterizes the distortion energy (α σ2/E) of a material subject to loading and deformation. Therefore, it can be used as a criterion for failure with respect to an experimental, permissible stress value. According to the Huber-Von Mises-Hencky theory, failure is predicted to occur if Eq. (10) is valid:
where the left-hand-side term represents the square of the Von Mises stress, which is a function of the local principal stresses σ1, σ2 and σ3 at a particular state of stress of the structure, and σf is the uniaxial failure strength of the material.
Geometry has been well established as a contributing factor to aneurysm expansion and rupture potential, independently of the heterogeneity of the wall. Figure 6 shows the displacement magnitude and stress distributions for the β = 0.2 and β = 1.0 uniform wall thickness models. In each case significant gradients occur at the inflection points of the aneurysm curvature. For β = 1.0, the changes in curvature result in higher displacements and increased stress, suggestive of the effect of flow through the gradual expansions and contractions of the geometry. For β = 0.2, only the anterior wall (left frame) displays a displacement gradient at the inflection points, while the Von Mises stress is maximum at the posterior wall (right frame). This is the outcome of blood flow and fluid pressure acting on a wall of decreasing curvature, as the posterior wall of this model is nearly flat.
Figure 6 Displacement and Von Mises wall stress distributions at t = 0.4 s for (a) β = 1.0 and (b) β = 0.2 uniform wall (UW) thickness models. The symbol Δ indicates the location of the maximum wall stress.
Comparing the magnitude of wall stress between β = 1.0 and β = 0.2 at t = 0.4 s reveals that a symmetric AAA is subject to a Von Mises stress 14% lower (23.8 N/cm2 compared to 27.7 N/cm2) than a highly asymmetric one. The effect of asymmetry is confirmed by Figure 7, with the stress for all five models scaled to the maximum of the β = 0.2 model. The change in location of the maximum stress is due largely to the changing shape of the aneurysm sac. In particular, between β = 0.6 and β = 0.4, the location shifts from the anterior wall at the distal end to the posterior wall at the midsection, a consequence of the decreasing curvature of the posterior wall, which allows for high speed flow along this surface.
Figure 7 Von Mises wall stress distribution at t = 0.4 s for models of increasing asymmetry and uniform wall (UW) thickness: (a) β = 1.0, (b) β = 0.8, (c) β = 0.6, (d) β = 0.4, and (e) β = 0.2. The symbol Δ indicates the location of the maximum wall stress.
Previous authors have reported on the effect of asymmetric shape and geometry changes in AAA wall mechanics. Elger et al [29] studied axisymmetric hypothetical AAA models, concluding that circumferential stresses are larger for aneurysm shapes of smaller curvature at constant maximum diameter. In quasi-static solid stress analyses (CSSS), Vorp and associates [30] evaluated Von Mises stress distributions in virtual AAA models similar to those presented in this work, for varying asymmetry and varying maximum diameter. They report on a concomitant increase in wall stress with increasing diameter and asymmetry, with stress values of the same order of magnitude as those reported in the present FSI studies for the UW models.
Wall thickness effect
A factor of increasing significance in AAA rupture risk prediction is the heterogeneity of the wall, in particular its thickness. It is difficult to accurately assess this dimension in patient-specific CT images due to calcification, thrombus, and the lack of clear image definition between the inner and outer wall surfaces. Therefore, a uniform thickness of 1.5 mm is typically assumed when modeling individual AAAs [13]. However, experimental sampling of wall specimens reveals that the wall is actually non-uniform, thinning in response to pulsatility and the progressive expansion of the aneurysm sac [31].
As evident from Figure 8, a variable wall thickness affects the flow dynamics as well as the stress distribution at t = 0.4 s, regardless of symmetry. For β = 1.0, a mostly attached flow pattern is evident with a uniform wall thickness, while a ring-shaped vortex is observed near the distal end for a variable wall thickness. Also of significance to the formation of vortices is the periodic nature of flow acceleration and retrograde flow found at the distal end in both models. In the variable wall thickness model the flow reverses direction more frequently, yielding negative flow rates at the distal end nearly twice as often compared to the uniform wall thickness model. This is mostly due to the increased compliance and larger deformation of the thinner wall that results from the momentum changes generated by the fluid throughout the cardiac cycle.
Figure 8 Effect of uniform (left) and variable (right) wall thickness: (a) y-z plane velocity vectors and (b) Von Mises stress distribution at t = 0.4 s for β = 1.0 model. The symbol Δ indicates the location of the maximum wall stress.
The patterns of Von Mises stress for the β = 1.0 model, shown in Figure 8 for t = 0.4 s, demonstrate the importance of wall thickness for a criterion of AAA rupture potential. The assumption of a uniform wall thickness translates into an underestimation of the maximum wall stress (23.8 N/cm2 for UW and 105.0 N/cm2 for VW) of nearly 77% when compared with a variable wall thickness model. The maximum stress occurs where the wall is thinnest, at the midsection of the sac; for the model with uniform wall thickness, the maximum stress occurs near the proximal and distal ends of the aneurysm, indicating the significance of geometry changes in aneurysm mechanics. This effect is illustrated in Figure 9 for both peak displacement and peak Von Mises stress, with the uniform wall scale represented by the left vertical axis and the variable wall scale according to the right vertical axis. For the FSI technique, peak wall stress and peak displacement are not achieved at t = 0.4 s as is the case for the CSSS and CSST techniques, but rather during 0.304 s < t < 0.4 s corresponding to the phase delay interval between peak inlet and outlet boundary conditions. Therefore, it follows that the effects of flow and traction boundary conditions in FSI are not transmitted instantaneously to the AAA wall, which yields maximum deformation after inlet peak flow. This is due to the elastic energy stored at the AAA wall from the previous pulsatile cycles of the simulation leading up to and during the last pulsatile cycle, which is used for postprocessing purposes. In Figure 9, the variable wall thickness model shows elevated displacements and stresses, decreasing nonlinearly with increasing symmetry. Given a fixed asymmetry, the nonlinear variation in maximum wall stress due solely to the heterogeneous nature of the wall thickness is more significant: a 4-fold increase for β = 1.0 and 4.7-fold increase for β = 0.2. These results indicate that, independently of location, variable AAA wall thickness as hypothesized in this work has a more significant effect on the peak wall stress than the asymmetric shape of the aneurysm sac itself, for a fixed maximum transverse diameter. Further verification of the relative weight of these two variables in FSI AAA rupture risk prediction is required for patient-specific models.
Figure 9 Effect of asymmetry on (a) peak displacement and (b) peak wall stress.
Comparison of computational methods
The use of computational modeling and numerical techniques for the assessment of AAA rupture risk has been traditionally limited to the simulation of wall mechanics under quasi-static (CSSS) conditions [7,18]. This approach has the advantage of being able to model the wall with two-dimensional solid shell elements under the assumption of a uniform wall thickness while ignoring the fluid dynamic events within the aneurysm sac [29-35], which translates into a quick turnaround for the computation of peak wall stresses. We have tested our virtual AAA models using the CSSS technique, as well as for CSST where the uniformly distributed luminal pressure is modeled as pulsatile and the maximum stresses evaluated at t = 0.4 s. Using our FSI-computed peak wall stresses as the baseline for comparison, Table 1 and Figure 10 (only for β = 0.6 and UW) show the difference in these stresses between the CSSS, CSST, and FSI methodologies. From Table 1, under the assumption of a uniform wall (UW) thickness, the quasi-static solid stress computations result in an underestimation of the peak stress of 9.4% on average; similarly, the pulsatile solid stress technique underestimates the peak stress by 9.2% on average. Under the assumption of a heterogeneous wall thickness, the CSSS and CSST techniques underestimate the peak stress predictions by an average of 29.5% and 29.4%, respectively. Given the importance of wall thickness heterogeneity in the accurate estimation of AAA rupture potential, these results indicate that fluid mechanics events should be taken into account in the modeling approach for the assessment of wall mechanics.
Table 1 Comparison of peak wall stress among the three numerical approaches.
β 0.2 0.4 0.6 0.8 1.0
UW FSI 32.3 29.1 28.3 28.0 27.7
UW CSST 29.0 (-10.2) 26.3 (-9.6) 26.0 (-8.1) 25.6 (-8.6) 25.1 (-9.4)
UW CSSS 29.2 (-9.6) 26.3 (-9.6) 25.8 (-8.9) 25.4 (-9.3) 25.1 (-9.4)
VW FSI 152.6 136.2 124.3 115.6 110.0
VW CSST 106.5 (-30.2) 95.7 (-29.7) 88.0 (-29.2) 82.3 (-28.8) 78.2 (-28.9)
VW CSSS 107.5 (-29.6) 95.6 (-29.8) 87.5 (-29.6) 82.0 (-29.1) 77.8 (-29.3)
Peak wall stress (in N/cm2) and comparison between the three numerical techniques. The parentheses show the % difference of the stress obtained with the CSST and CSSS methods with respect to the baseline FSI method.
Figure 10 Comparison of peak Von Mises wall stress distributions for β = 0.6 UW model evaluated for: (a) static computational solid stress -CSSS- approaches, (b) transient computational solid stress -CSST-, and (c) fluid-structure interaction -FSI. The symbol Δ indicates the location of the maximum wall stress.
The virtual AAA models presented in this work provide a fundamental baseline for application of the FSI methodology as a non-invasive tool for rupture risk prediction in individual patients, outlining the importance of aneurysm asymmetry and non-uniformity of the vessel wall. This approach takes into account blood flow dynamics, which is inherently transient, and its effect on the wall mechanics. Hence, the results of the FSI predictions demonstrate the relationship between the fluid velocity field and the flow-induced wall stresses, which previous studies have assessed indirectly only on the basis of a uniform and static fluid pressure distribution. During the cardiac cycle, the instantaneous fluid forces acting on the inner wall will deform and expand the artery. In turn, the wall motion alters the velocity field until equilibrium is reached; these events occur instantaneously with the pulsating flow and cannot be evaluated by utilizing a CSS technique.
As evident by Figure 11, the deformations of the AAA sac are not negligible, particularly for a heterogeneous wall. Regardless of asymmetry, the thin midsection of the wall where the diameter and stress are at their maximum, show a significant distortion of the original mesh. Therefore, while the modeling of a heterogeneous wall in this study represents a novel aspect of the research, it must be handled in the computational approach with considerable care so that a stable model can yield accurate and realistic results. The deformation of the geometry is further illustrated in Figure 12, which shows the volume of each virtual AAA's lumen as a function of time for the last cycle of the FSI physics. The initial volume of a non-deformed AAA for all the models used in this study is ∀o = 206.6 cm3. For the UW models, peak volume is obtained at 0.34 s, while for the VW models, it is at 0.38 s. This coincides with the instant of peak wall stress and peak displacement, at which the AAA wall achieves the greatest expansion. Table 2 shows the change in volume at peak deformation, demonstrating that the increase in volume is concomitant with asymmetry and heterogeneity of the wall. This pulsatile nature of AAA deformation cannot be assessed by utilizing a CSS technique, which takes into account volume changes of the AAA wall only.
Figure 11 Deformation of the AAA sac at t = 0.4 s for (a) β = 1.0 and (b) β = 0.2 models with variable wall (VW) thickness. The red mesh is the original, non-deformed artery, while the blue mesh is the deformed geometry at peak systole.
Figure 12 Volume of virtual AAA lumen during the last pulsatile cycle for (a) uniform wall (UW) thickness and (b) variable wall (VW) thickness models.
Table 2 Volume of AAA lumen at peak deformation.
β 0.2 0.4 0.6 0.8 1.0
UW 225.1 (+9.0) 222.9 (+7.9) 221.9 (+7.4) 221.5 (+7.2) 221.4 (+7.2)
VW 248.7 (+20.4) 246.4 (+19.3) 245.2 (+18.7) 244.7 (+18.4) 244.6 (+18.4)
Peak AAA volume (in cm3) and comparison between the UW and VW models. The parentheses show the % increase with respect to the non-deformed geometry volume (∀o = 206.6 cm3) at the start of the FSI simulation.
Limitations
Aneurysm rupture does not result solely from the stress exerted along the inner wall, but rather from the transmission of that stress onto the middle and outer wall layers, which causes the diseased arterial wall to fail. Furthermore, it is unlikely that the wall thickness of an actual AAA will decrease concomitantly towards the aneurysm's maximum transverse dimension and then increase by the same gradient towards the iliac bifurcation. The combination of virtual geometry, linearly elastic material properties, and a wall thickness that varies inversely proportional to vessel diameter, creates wall stresses in excess of 100 N/cm2, which may be physiologically unrealistic compared to average uniaxial tensile strength of AAA tissue [35]. Nonetheless, our mathematical description of wall thickness variation provides insight into the relative magnitudes of the stresses in UW and VW models, and the importance of wall thickness heterogeneity in the prediction of AAA wall mechanics.
Despite the more accurate predictions of AAA biomechanics utilizing an FSI methodology, there are additional limitations to the present study that restricts its application in a clinical environment. Among these are the assumption of a linear elastic modulus for modeling wall mechanical properties, the need for non-invasive predictors of wall thickness and strength, the lack of inclusion of thrombus and calcification in the geometric and material models, the anisotropic characterization of the tissue models, the absence of external forces induced by surrounding tissue and organs, the absence of iliac arteries, and the lack of assessment of biological activity. Several of these issues are contentious, such as the inclusion of intraluminal thrombus (ILT), as previous authors suggest that ILT may increase or decrease wall stress and aneurysm rupture risk [34,35]. While the investment in computational time proves to be a costly drawback of the FSI methodology with respect to the CSS techniques, the constant improvements in microprocessor technology will allow for practical applications in a clinical setting in the next few years. The ultimate multi-scale model for the non-invasive estimation of aneurysm rupture risk should incorporate biomechanical (fluid and solid dynamics), biological, and genetic aspects of AAA disease.
Conclusion
This work represents a numerical investigation of the fluid-structure interaction of ten virtual abdominal aortic aneurysm models for the prediction of wall stress as a means of assessing rupture potential non-invasively. The effects of asymmetric bulging of the anterior wall and non-uniformity of the wall thickness are studied in detail with respect to the peak wall stress, while maintaining the maximum transverse diameter constant at 6 cm. A comparison is made with traditional numerical techniques based on the quasi-static and transient computational solid stress analyses.
The fluid dynamics in a compliant asymmetric aneurysm model is characterized by the development of ring-shaped vortices during systole that are ejected from the sac shortly after peak pressure is achieved. The distortion energy stored in the vessel as it expands during the cardiac cycle contributes to the early formation of recirculation regions in the aneurysm that yield high velocity gradients at the distal end of the aneurysm. These flow patterns, in combination with the geometrical features of the model and the elastic characterization of the wall material, determine the distribution of flow-induced wall stresses. In a fusiform AAA for which the local thickness decreases inversely proportional to the local vessel diameter, the peak wall stress is 4 times greater than with a uniform wall thickness. Similarly, asymmetric bulging of the anterior wall as determined by β = 0.2 in a uniform wall thickness model results in a 17% increase in peak wall stress when compared to a fusiform AAA. For the same maximum diameter, wall thinning has a more significant effect on the concomitant rise in peak wall stress than the asymmetry of the aneurysm sac. The computational solid stress techniques underestimate wall stress calculations when compared to the fluid-structure interaction predictions. The trade-off for a better predictor tool is a 23-fold increase in computational costs. The use of this numerical technique as a rupture risk assessment tool based on an individual patient's status must incorporate non-invasive predictors of wall thickness and tissue strength.
Authors' contributions
Alexander Shkolnik created the virtual AAA model, conducted the computational simulations for this study and participated in the methods section of the manuscript. Christine Scotti performed the results post-processing and along with Ender Finol and Satish Muluk contributed to the preparation of the manuscript. Ender Finol also conceived of the study and participated in its design and coordination. All authors read and approved of the manuscript.
Acknowledgements
This research was supported in part by (i) the Pennsylvania Infrastructure Technology Alliance (PITA), a partnership of Carnegie Mellon, Lehigh University, and the Commonwealth of Pennsylvania's Department of Community and Economic Development (DCED), (ii) a pre-doctoral fellowship by the Dowd-ICES Fellowship Program, and (iii) a grant from the Pittsburgh Supercomputing Center through the NIH National Center for Research Resources cooperative agreement 2 p41 RR06009.
==== Refs
Newman AB Arnold AM Burke GL O'Leary DH Manolio TA Cardiovascular disease and mortality in older adults with small abdominal aortic aneurysms detected by ultrasonography: the cardiovascular health study Annals of Internal Medicine 2001 134 182 190 11177330
Karkos C Mukhopadhyay U Papakostas I Ghosh J Thomson G Hughes R Abdominal aortic aneurysm: the role of clinical examination and opportunistic detection European Journal of Vascular and Endovascular Surgery 2000 19 299 303 10753695 10.1053/ejvs.1999.1002
Katz DA Cronenwett JL The cost-effectiveness of early surgery versus watchful waiting in the management of small abdominal aortic aneurysms Journal of Vascular Surgery 1994 19 980 990 8201717
Galland RB Whiteley MS Magee TR The fate of patients undergoing surveillance of small abdominal aortic aneurysms European Journal of Vascular and Endovascular Surgery 1998 16 104 109 9728428 10.1016/S1078-5884(98)80150-0
Scott RA Ashton HA Lamparelli MJ Harris GJ Stevens JW A 14-year experience with 6 cm as a criterion for surgical treatment of abdominal aortic aneurysm British Journal of Surgery 1999 86 1317 1321 10540141 10.1046/j.1365-2168.1999.01227.x
Valentine RJ Decaprio JD Castillo JM Modrall JG Watchful waiting in cases of small abdominal aortic aneurysms – appropriate for all patients? Journal of Vascular Surgery 2000 32 441 448 10957650 10.1067/mva.2000.108635
Fillinger MF Marra SP Raghavan ML Kennedy FE Prediction of rupture risk in abdominal aortic aneurysm during observation: wall stress versus diameter Journal of Vascular Surgery 2003 37 724 732 12663969 10.1067/mva.2003.213
Peattie RA Riehle TJ Bluth EI Pulsatile flow in fusiform models of abdominal aortic aneurysms: flow fields, velocity patterns and flow-induced wall stresses Journal of Biomechanical Engineering 2004 126 438 446 15543861 10.1115/1.1784478
Walsh M McGloughlin T Liepsch DW O'Brien T Morris L Ansari AR On using experimentally estimated wall shear stresses to validate numerically predicted results Proceedings of the Institution of Mechanical Engineers Part H, Journal of Engineering in Medicine 2003 217 77 90 12666774 10.1243/09544110360579286
Salsac AV Sparks SR Lasheras JC Hemodynamics changes occurring during the progressive enlargement of abdominal aortic aneurysms Annals of Vascular Surgery 2004 18 14 21 14712379 10.1007/s10016-003-0101-3
Finol EA Keyhani K Amon CH The effect of asymmetry in abdominal aortic aneurysms under physiologically realistic pulsatile flow conditions Journal of Biomechanical Engineering 2003 125 207 217 12751282 10.1115/1.1543991
Finol EA Amon CH Flow dynamics in anatomical models of abdominal aortic aneurysms: Computational analysis of pulsatile flow Acta Cientifica Venezolana 2003 54 43 49 14515766
Di Martino ES Guadagni G Fumero A Ballerini G Spirito R Biglioli P Redaelli A Fluid-structure interaction within realistic three-dimensional models of the aneurysmatic aorta as a guidance to assess the risk of rupture of the aneurysm Medical Engineering and Physics 2001 23 647 655 11755809 10.1016/S1350-4533(01)00093-5
Finol EA Di Martino ES Vorp DA Amon CH Fluid-structure interaction and structural analyses of an aneurysm model Proceedings of the 2003 Summer Bioengineering Conference: 25–29 June 2003 2003 Key Biscayne, FL. American Society of Mechanical Engineers 75 76
Finol EA Shkolnik AD Scotti CM Amon CH Payan Y Computational modeling of abdominal aortic aneurysms: an assessment of rupture potential for presurgical planning Biomechanics Applied to Computer Assisted Surgery 2005 Kerala, India: Research Signpost Publisher
Raghavan M Kratzberg J da Silva ES Heterogeneous, variable wall-thickness modeling of a ruptured abdominal aortic aneurysm Proceedings of the 2004 International Mechanical Engineering Congress and R&D Expo: 13–19 November 2004 1 Anaheim, CA. American Society of Mechanical Engineers EMECE2004 60018
Donea J Huerta A Ponthot J-Ph Rodriguez-Ferran A Stein E, de Borst R, Hughes T Arbitrary lagrangian-eulerian methods Encyclopedia of Computational Mechanics 2004 1 New York: John Wiley & Sons 1 25
Raghavan ML Vorp DA Toward a biomechanical tool to evaluate rupture potential of abdominal aortic aneurysm: identification of a finite strain constitutive model and evaluation of its applicability Journal of Biomechanics 2000 33 475 482 10768396 10.1016/S0021-9290(99)00201-8
Mills C Gabe I Gault J Mason D Ross J JrBraunwald E Shillingford J Pressure-flow relationships and vascular impedance in man Cardiovascular Research 1970 4 405 417 5533085
Finol EA Amon CH Flow-induced wall shear stress in abdominal aortic aneurysms: Part II – pulsatile flow hemodynamics Computer Methods in Biomechanics and Biomedical Engineering 2002 5 319 328 12186711 10.1080/1025584021000009751
Finol EA Amon CH Flow-induced wall shear stress in abdominal aortic aneurysms: Part I – steady flow hemodynamics Computer Methods in Biomechanics and Biomedical Engineering 2002 5 309 318 12186710 10.1080/1025584021000009742
Moore JE JrKu DN Zarins CK Glagov S Pulsatile flow visualization in the abdominal aorta under differing physiologic conditions: implications for increased susceptibility to atherosclerosis ASME Journal of Biomechanical Engineering 1992 114 391 397
Moore JE JrKu DN Pulsatile velocity measurements in a model of the human abdominal aorta under resting conditions ASME Journal of Biomechanical Engineering 1994 116 337 346
Moore JE JrMaier SE Ku DN Boesiger P Hemodynamics in the abdominal aorta: a comparison of in vitro and in vivo measurements Journal of Applied Physiology 1994 76 1520 1527 8045828
Milnor W Hemodynamics 1989 2 Baltimore, Williams and Wilkins
Nichols WW O'Rourke MF McDonald's Blood Flow in Arteries – theoretical, experimental and clinical principles 1990 3 Philadelphia, Lea and Febiger
Adina R&D Inc ADINA Theory and Modeling Guide – Volume IV: ADINA-FSI 2002 Watertown, Adina R&D, Inc
Bluestein D Niu L Schoephoerster R Dewanjee M Steady flow in an aneurysm model: correlation between fluid dynamics and blood platelet deposition ASME Journal of Biomechanical Engineering 1996 118 280 286
Elger DF Blackketter DM Budwig RS Johansen KH The influence of shape on the stresses in model abdominal aortic aneurysms ASME Journal of Biomechanical Engineering 1996 118 326 332
Vorp DA Raghavan ML Webster MW Mechanical wall stress in abdominal aortic aneurysm: influence of diameter and asymmetry Journal of Vascular Surgery 1998 27 632 639 9576075
Venkatasubramaniam AK Fagan MJ Mehta T Mylankal KJ Ray B Kuhan G Chetter IC McCollum PT A comparative study of aortic wall stress using finite element analysis for ruptured and non-ruptured abdominal aortic aneurysms European Journal of Vascular and Endovascular Surgery 2004 28 168 176 15234698 10.1016/j.ejvs.2004.03.029
Fillinger MF Raghavan ML Marra SP Cronenwett JL Kennedy FE In vivo analysis of mechanical wall stress and abdominal aortic aneurysm rupture risk Journal of Vascular Surgery 2002 36 589 597 12218986 10.1067/mva.2002.125478
Di Martino E Mantero S Inzoli F Melissano G Astore D Chiesa R Fumero R Biomechanics of abdominal aortic aneurysm in the presence of endoluminal thrombus: experimental characterisation and structural static computational analysis European Journal of Vascular and Endovascular Surgery 1998 15 290 299 9610340 10.1016/S1078-5884(98)80031-2
Mower WR Quinones WJ Gambhir SS Effect of intraluminal thrombus on abdominal aortic aneurysm wall stress Journal of Vascular Surgery 1997 26 602 608 9357460 10.1016/S0741-5214(97)70058-2
Thubrikar MJ Labrosse M Robicsek F Al-Soudi J Fowler B Mechanical properties of abdominal aortic aneurysm wall Journal of Medical Engineering Technology 2001 25 133 142 11601439 10.1080/03091900110057806
|
16271141
|
PMC1298313
|
CC BY
|
2021-01-04 16:37:36
|
no
|
Biomed Eng Online. 2005 Nov 4; 4:64
|
utf-8
|
Biomed Eng Online
| 2,005 |
10.1186/1475-925X-4-64
|
oa_comm
|
==== Front
Cancer Cell IntCancer Cell International1475-2867BioMed Central London 1475-2867-5-311628196810.1186/1475-2867-5-31Primary ResearchSpecific distribution of overexpressed aurora B kinase in interphase normal epithelial cells Abdullah Ash-shafie [email protected] Charlene [email protected] Maki [email protected] Mammalian Cell Biology Group, Temasek Life Sciences Laboratory, 1 Research Link, National University of Singapore, 117604, Singapore 2 Temasek Junior College, 22 Bedok South Road, 469278, Singapore 2005 9 11 2005 5 31 31 31 8 2005 9 11 2005 Copyright © 2005 Abdullah et al; licensee BioMed Central Ltd.2005Abdullah 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 is known that aurora B, a chromosomal passenger protein responsible for the proper progression of mitosis and cytokinesis, is overexpressed throughout the cell cycle in cancer cells. Overexpression of aurora B produced multinuclearity and induced aggressive metastasis, suggesting that overexpressed aurora B has multiple functions in cancer development. However, the detailed dynamics and functions of overexpressed aurora B are poorly understood.
Results
We overexpressed GFP fused aurora B kinase in normal rat kidney epithelial cells. Using spinning disk confocal microscopy, we found that overexpressed aurora B-GFP was predominantly localized in the nucleus and along the cortex as a dot-like or short filamentous structure during interphase. Time-lapse imaging revealed that a cytoplasmic fraction of overexpressed aurora B-GFP was incorporated into the nucleus after cell division. Immunofluorescence studies showed that the nuclear fraction of overexpressed aurora B did not induce ectopic phosphorylation of histone H3 after cell division. The cytoplasmic fraction of overexpressed aurora B-GFP was mainly associated with cortical actin filaments but not stress fibers. Myosin II regulatory light chain, one of the possible targets for aurora B, did not colocalize with cortical aurora B-GFP, suggesting that overexpressed aurora B did not promote phosphorylation of myosin II regulatory light chain in interphase cells.
Conclusion
We conclude that overexpressed aurora B has a specific localization pattern in interphase cells. Based on our findings, we propose that overexpressed aurora B targets the nuclear and cortical proteins during interphase, which may contribute to cancer development and tumor metastasis.
==== Body
Background
Aurora B kinase is a chromosomal passenger protein responsible for maintaining of chromosomal integrity through the proper coordination of mitosis and cytokinesis [1,2]. Regulation of aurora B kinase is such that expression levels of the protein peak at G2-M phase, while its kinase activity is maximal during mitosis [3]. Previous studies showed that aurora B kinase is involved in targeting and regulating the activity of a number of substrates, which in turn drive mitotic progression [1,2]. This is coupled by the fact that the expression of aurora B kinase is minimal during interphase, and the activity of the protein reaches its maximum just after the deactivation of CDK1 kinase [4], suggesting that the kinase activity of aurora B is mainly required during late mitosis. However, it has been found that in a number of cancer cell lines, aurora-B is constantly overexpressed throughout the cell cycle [5]. It has been shown that the overexpression of aurora B kinase in normal cells produced multinuclearity, perhaps leading to genetic instability, possibly due to defects in mitotic progression [5]. Moreover, the cells stably overexpressing aurora B showed more aggressive and malignant cancerous growth over control tumours [6]. These suggest that overexpressed aurora B kinase has multiple functions in cancer development. However, the mechanism by which overexpressed aurora B kinase promotes cancer development is poorly understood. In addition, although biochemical evidences revealed that aurora B is overexpressed throughout the cell cycle [5], little is known about the dynamics and functions of overexpressed aurora B during interphase. To examine if overexpressed aurora B has specific distribution throughout the cell cycle, we overexpressed aurora B-GFP in normal rat kidney epithelial (NRK) cells and analyzed the subcellular distribution of overexpressed aurora B using modern microscopic imaging. We found that overexpressed aurora B was preferentially associated with the nucleus and the cortex. Overexpression of aurora B induced neither ectopic phosphorylation of histone H3 nor excessive phosphorylation of myosin II regulatory light chain in interphase NRK cells, suggesting that aurora B has other specific targets in these regions.
Results and Discussion
In order to investigate the distribution of overexpressed aurora B in interphase cells, NRK cells were transfected with GFP-tagged aurora B kinase [7] and were observed using spinning disk laser confocal microscope (Figure 1). Overexpressed aurora B-GFP was dispersedly localized in the nucleus (Figure 1a', arrowhead) but not nucleolus and associated with the cortex (Figure 1a', arrows), while GFP alone was diffusely localized in the cell and did not show any particular localization pattern (Figure 1b'). GFP tagged kinase-dead mutant of aurora B showed the similar localization pattern, indicating that overexpressed aurora B is distributed in the nuclei and along the cortex independently of its kinase activity (data not shown). It has been shown that aurora B was detected on the condensed chromosomes when the cells enter into prophase [7-10]. However, the localization pattern of overexpressed aurora B in the nucleus we observed here (Figure 1a', arrowhead) appeared to be different from that was observed in prophase cells [7]. Previous reports showed that aurora B localized to the centromeres as early as prophase and relocated to the spindle midzone during anaphase [7-9]. However, it is unknown if overexpressed aurora B has a unique localization pattern during cytokinesis. Thus, to understand the dynamics of overexpressed aurora B during cytokinesis, time-lapse imaging of dividing cells overexpressing aurora B-GFP was performed. A major fraction of aurora B-GFP was associated with the spindle midzone (Figure 2, arrowheads) [7]. Notably, as the chromosomes started decondensing, an increasing fluorescence signal of aurora B-GFP was observed around the chromosomes (Figure 2b', arrows). At late cytokinesis, the strong spherical mass of fluorescence signals was observed in the nuclear region (Figure 2c', arrows), suggesting that a cytoplasmic fraction of overexpressed aurora B enters the nucleus during nuclear envelope formation. Since aurora B phosphorylates histone H3 at Ser10 to catalyze chromosome condensation [8-10], we speculate that overexpressed aurora B may induce phosphorylation of histone H3 even after cell division. To test this possibility, we stained the cells overexpressing aurora B-GFP with antibodies that specifically recognize histone H3 phosphorylated at Ser10 (Figure 3). Phosphorylated histone H3 was observed in the neighboring mitotic cells (Figure 3c). However, we could not detect the phosphorylated histone H3 in the nucleus of G1 (late telophase) cells overexpressing aurora B (Figure 3, arrows), indicating that overexpressed aurora B did not induce ectopic phosphorylation of histone H3 in our experimental condition. A previous report showed that excessive phosphorylation of histone H3 by overexpressed aurora B appeared to be mainly induced during mitosis [6]. Based on these observations, we speculate that the nuclear fraction of overexpressed aurora B is likely to have other targets during interphase.
Figure 1 Localization of overexpressed aurora B-GFP in an interphase NRK cell. Spinning disk laser confocal microscopic images of living NRK cells overexpressing aurora B-GFP in interphase cells. The phase-contrast (a, b) and corresponding fluorescence images (a', b') show the distribution of overexpressed aurora B-GFP (a') or GFP alone (b'). Overexpressed aurora B-GFP is preferentially associated with nucleus (a', arrowhead) and the cortex (a', arrows), while GFP alone is diffusely distributed in the cell. Bar, 10 μm.
Figure 2 Overexpressed aurora B is incorporated into the nucleus after cell division. An NRK cell overexpressing aurora B-GFP was monitored by time-lapse microscopic imaging. Phase-contrast (a-c) and corresponding fluorescence images (a'-c') showed the dynamics of overexpressed aurora B-GFP during cytokinesis. Majority of aurora B is associated with the spindle midzone (a'-c', arrowheads) during cytokinesis. When the nuclear envelops started forming in daughter cells, a cytoplasmic fraction of overexpressed aurora B was incorporated into the nuclei (b' and c', arrows). Bar, 10 μm
Figure 3 Overexpression of aurora B kinase does not phosphorylate histone H3 in interphase NRK cells. NRK cells overexpressing aurora B-GFP was stained with antibodies that specifically recognised histone H3 phosphorylated at Ser10 and then examined the expression of aurora B-GFP (a) and phosphorylation of histone H3 (c) by confocal laser microscopy. Corresponding phase and merged images (green; aurora B-GFP, red; phosphorylated histone H3) are shown in panels b and d, respectively. Although a fraction of aurora B-GFP is accumulated in the nucleus in a late telophase cell overexpressing aurora B-GFP, (arrows), phosphorylated histone H3 was not detected in the cell (c, d). Phosphorylated histone H3 was observed in a neighbouring prophase cell (arrowheads). Bar, 10 μm.
The other fraction of overexpressed aurora B was preferentially distributed along the cortex with a dot-like or branched structure along the cortex in interphase cells, while GFP alone was uniformly localized in the cell (Figure 1). We tested if overexpressed aurora B colocalized with cortical actin filaments. The cells overexpressing aurora B-GFP were stained with rhodamine-labelled phalloidin and were then observed under the confocal microscope (Figure 4). Cortical aurora B-GFP was colocalized with short branched actin filaments along the membrane (Figure 4, arrows). Interestingly, overexpressed aurora B appeared not to be associated with the thick bundles of actin filaments (Figure 4, arrowheads). Since it has shown that aurora B kinase phosphorylates myosin II regulatory light chain at Ser19 [11], which stimulates actin-activated myosin II ATPase activity [12], we also stained the cells overexpressing aurora B-GFP with the antibodies that specifically recognize myosin II regulatory light chain phosphorylated at Ser19 (Figure 5). Consistently with the previous report [13], phosphorylated myosin II regulatory light chain was enriched around the cell periphery (Figure 5, arrowhead), while overexpressed aurora B-GFP was mainly associated with the cortex (Figure 5, arrow). Therefore, we hardly observed the colocalization of cortical aurora B-GFP with the phosphorylated myosin II regulatory light chain (Figure 5d). In addition, we did not detect an increase in the fluorescence signal of phosphorylated myosin II regulatory light chain in the cells overexpressing aurora B-GFP. We conclude that overexpressed aurora B does not promote phosphorylation of myosin II regulatory light chain in interphase NRK cells. During cytokinesis, aurora B is accumulated along not only midzone microtubules but also the lateral cortex through astral microtubules [14], indicating that aurora B may interact with cortical proteins which are associated with cortical actin filaments.
Figure 4 Overexpressed aurora B is colocalized with cortical actin filaments but not stress fibers in interphase NRK cells. An NRK cell overexpressing aurora B-GFP was stained with rhodamine-labelled phalloidin and then examined the subcelluar localization of aurora B-GFP (a) and actin filaments (c) by confocal laser microscopy. Corresponding phase and merged images (green; aurora B-GFP, red; actin filaments) are shown in panels b and d, respectively. Cortical aurora B-GFP is well colocalized with the short fragments of actin filaments around the cortex (a and d, arrows). However, overexpressed aurora B-GFP appeared not to colocalize with the thick stress actin filaments in the peripheral region of the cortical area (c and d, arrowheads). Bar, 10 μm.
Figure 5 Overexpressed aurora B is not colocalized with phosphorylated myosin II regulatory light chain in interphase NRK cells. An NRK cell overexpressing aurora B-GFP was stained with antibodies that specifically recognised myosin II regulatory light chain phosphorylated at Ser19 and then examined the subcelluar localization of aurora B-GFP (a) and phopshorylated myosin II regulatory light chain (c) by confocal laser microscopy. Corresponding phase and merged images (green; aurora B-GFP, red; phosphorylated myosin II regulatory light chain) are shown in panels b and d, respectively. Overexpressed aurora B is associated with the cortex (a and d, arrows), while phosphorylated myosin II regulatory light chain is enriched in the cell periphery (c, and d, arrowheads). Bar, 10 μm.
So far, several evidences have suggested that overexpression of aurora B promotes cancer development [5,6,15]. However, in our experimental condition, we did not observe any defects in cell division [7], cell morphology and cell growth. Since only a slight increase in the number of multinuclear cells was observe even when the normal fibroblast cells were transfected with a large amount of the plasmids encoding aurora B kinase [5], the expression level of aurora B in our experiments could be too weak to induce multinuclearity. Alternatively, overexpression of aurora B is not sufficient for the induction of carcinogenesis. Since the cells stably overexpressing aurora B were able to be isolated only when the p53 was mutated, it was suggested that overexpression of aurora B was induced after p53 defects in cancer development [6].
Our findings suggest that the targets for overexpressed aurora B are associated with the nucleus and the cortical actin filaments. The former target might be involved in the induction of polyploidy and/or signaling pathways in multi-step carcinogenesis, while the latter might be implicated in the development of metastasis. Our observations will help to search for the targets of overexpressed aurora B in mammalian cells.
Materials and methods
Cell Culture, Microscopy and Image Processing
Normal rat kidney epithelial cells (NRK-52E; American Type Culture Collection) were cultured in Kaighn's modified F12 medium supplemented with 10% fetal bovine serum, 100 U/ml Penicillin, and 100 μg/ml Streptomycin, on glass chamber dishes as previously described [16]. The cells were maintained at 37°C in an enclosed stage incubator built on top of an Axiovert 200 M inverted microscope (Carl Zeiss) and viewed with a 100×, numerical aperture 1.30, Oil Ph 3, Plan-NEOFLUAR lens, while another connected to a PerkinElmer RS-3 spinning disk confocal system. Live cell images were acquired with a cooled charge-coupled device camera (CoolSNAPHQ, Roper Scientific), processed with Metaview or a digital cooled ocra-ER camera (Hamamatsu). For fluorescence imaging using spinning disk confocal system, a CSU21 confocal optical scanner was used together with krypton-argon laser illumination source, with 488 nm excitation and emission filter (Chroma) HQ 525/50 M.
Fixed cells were viewed using inverted confocal Zeiss LSM 510 Meta microscope (Carl Zeiss) with a 100×, numerical aperture 1.25 Achroplan lens. Images were acquired using 488 nm Argon laser and 543 nm HeNe laser for excitation and signals were emitted through BP 505 – 530 nm filter and LP 560 nm filter.
Transfection and Immunofluorescence
NRK cells were plated on a coverslip chamber dish and incubated for 18–24 h. Immediately before transfection, the cells were rinsed once F12K supplemented with 1% FBS or Opti-MEM I medium (Life Technologies). The cells were transfected with the DNA construct (2 μg) using Superfect or Effectene transfection reagent according to manufacturer's instructions (Qiagen).
For phosphorylated myosin II regulatory light chain immunofluorescence, cells were rinsed with warm cytoskeleton buffer [17] and fixed with 4% paraformaldehyde (EM Science) in warm cytoskeleton buffer for 10 min. They were then rinsed thoroughly using cytoskeleton buffer and permeabilized with 0.5% Triton X-100 incubated for 5 min. Fixed cells were rinsed with cytoskeleton buffer, blocked with 1% bovine serum albumin (BSA) (Roche Diagnostics) in PBS. Following on, the fixed cells were incubated with phospho-myosin light chain 2 (Ser19) polyclonal antibodies (Cell Signalling Technology) at a dilution of 1:100 in PBS with 1% BSA (PBS/BSA) for 45 min at 37°C. After thorough washing with PBS/BSA, cells were incubated with Alexa 546-conjugated goat anti mouse antibodies (Molecular Probes) at a dilution of 1:100 in PBS/BSA for 30 min at 37°C. For actin staining, the fixed cells were incubated with rhodamine labeled phalloidin (Molecular Probes) at a dilution of 1:50 in PBS for 30 min at 37°C.
Acknowledgements
We would like to thank Ms. Shvetha Sankaran for the initial work. A.-S. A. joined M. M.-H. as an attachment student in Research Attachment Programme (REAP) conducted in Temasek Life Sciences Laboratory. This study was supported by intramural funds from the Temasek Life Sciences Laboratory to M. M.-H.
==== Refs
Carmena M Earnshaw WC The cellular geography of Aurora Kinases Nat Rev Mol Biol Cell 2003 4 842 854 14625535 10.1038/nrm1245
Andrews PD Knatko E Moore WJ Swedlow JR Mitotic mechanics: the auroras come into view Curr Opin Cell Biol 2003 15 672 683 14644191 10.1016/j.ceb.2003.10.013
Terada Y AIM-1: a mammalian midbody-associated protein required for cytokinesis The EMBO J 1998 17 667 676 9450992 10.1093/emboj/17.3.667
Bischoff JR Anderson L Zhu Y Mossie K Ng L Souza B Schryver B Flanagan P Clairvoyant F Ginther C Chan CSM Novotny M Slamon DJS Plowman GD A homologue of Drosophila aurora kinase is oncogenic and amplified in human colorectal cancers EMBO J 1998 17 3052 3065 9606188 10.1093/emboj/17.11.3052
Tatsuka M Multinuclearity and increased ploidy caused by overexpression of aurora- and Ipl1-like midbody-associated protein mitotic kinase in human cancer cells Cancer Res 1998 58 4811 4816 9809983
Ota T Increased mitotic phosphorylation of Histone H3 attributable to AIM-1/ Aurora-B overexpression contributes to chromosome number instability Cancer Res 2002 62 5168 5177 12234980
Murata-Hori M Tatsuka M Wang YL Probing the dynamics and functions of Aurora-B kinase in living cells during mitosis and cytokinesis Mol Biol Cell 2002 13 1099 1108 11950924 10.1091/mbc.01-09-0467
Adams RR Maiato H Earnshaw WC Carmera M Essential roles of Drosophila Inner Centromere Protein (INCENP) and Aurora-B in Histone H3 phosphorylation, metaphase chromosome alignment, Kinetochore disjunction, and chromosome segregation J Cell Biol 2001 153 865 879 11352945 10.1083/jcb.153.4.865
Giet R Glover DM Drosophila Aurora-B kinase is required for Histone H3 phosphorylation and Condesin recruitment during chromosome condensation and to organize central spindle during cytokinesis J Cell Biol 2001 152 669 681 11266459 10.1083/jcb.152.4.669
Crosio C Fimia GM Loury R Kimura M Okano Y Zhou H Sen S Allis D Sassone-Corsi P Mitotic phosphorylation of Histone H3: spatio-temporal regulation by mammalian Aurora kinases Mol Cell Biol 2002 22 874 885 11784863
Murata-Hori M Fumoto K Fukuta Y Kikuchi A Tatsuka M Hosoya H Myosin II regulatory light chain as a novel substrate for AIM-1, an aurora/Ipl1p-related kinase from rat J Biochem (Tokyo) 2002 128 903 907 11098131
Sellers JR Spudich JA Sheetz MP Light chain phosphorylation regulates the movement of smooth muscle myosin actin filaments J Cell Biol 1985 1897 1902 3840488 10.1083/jcb.101.5.1897
Matsumura F Ono S Yamakita Y Totsukawa G Yamashiro S Specific localization of serine 19 phosphorylated myosin II during cell locomotion and mitosis of cultured cells J Cell Biol 1998 140 119 129 9425160 10.1083/jcb.140.1.119
Murata-Hori M Wang YL Both midzone and astral microtubules are involved in the delivery of cytokinesis signals: Insights from the mobility of aurora B J Cell Biol 2002 159 45 53 12370248 10.1083/jcb.200207014
Katayama H Ota T Jisaki F Ueda Y Tanaka T Odashima S Suzuki F Teada Y Tatsuka M Mitotic kinase expression and colorectal cancer progression J Natl Cancer Inst 1999 91 1160 1162 10393726 10.1093/jnci/91.13.1160
MaKenna NM Wang YL Culturing cells on the microscope stage Methods Cell Biol 1989 29 195 205 2643760
Wheatley SP Wang YL Midzone microtubule bundles are continuously required for cytokinesis in cultured epithelial cells J Cell Biol 1996 135 981 989 8922381 10.1083/jcb.135.4.981
|
16281968
|
PMC1298314
|
CC BY
|
2021-01-04 16:40:10
|
no
|
Cancer Cell Int. 2005 Nov 9; 5:31
|
utf-8
|
Cancer Cell Int
| 2,005 |
10.1186/1475-2867-5-31
|
oa_comm
|
==== Front
Cardiovasc UltrasoundCardiovascular Ultrasound1476-7120BioMed Central London 1476-7120-3-351627765610.1186/1476-7120-3-35Case ReportCoronary artery to left ventricle fistula López-Candales Angel [email protected] Vivek [email protected] Cardiovascular Institute at the University of Pittsburgh Medical Center, Pittsburgh, PA, USA2005 8 11 2005 3 35 35 17 10 2005 8 11 2005 Copyright © 2005 López-Candales and Kumar; licensee BioMed Central Ltd.2005López-Candales and Kumar; 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
Coronary cameral fistulas are an uncommon entity, the etiology of which may be congenital or traumatic. They involve abnormal termination of a coronary artery, usually the right coronary, into a cardiac chamber, usually the right ventricle.
Case Presentation
We describe a case of female patient with severe aortic stenosis and interventricular septal hypertrophy that underwent bioprosthetic aortic valve replacement with concomitant septal myectomy. On subsequent follow-up an abnormal flow traversing the septum into the left ventricle was identified and Doppler interrogation demonstrated a continuous flow, with a predominantly diastolic component, consistent with coronary arterial flow.
Conclusion
The literature on coronary cameral fistulas is reviewed and the etiology of the diagnostic findings discussed. In our patient, a coronary artery to left ventricle fistula was the most likely explanation secondary to trauma to the septal perforator artery during myectomy. Since the patient was asymptomatic at the time of diagnosis no intervention was recommended and has done well on follow-up.
Aortic valve diseasecoronary fistulaeechocardiographyseptal hypertrophymyectomy
==== Body
Background
Communications between coronary arteries and cardiac chambers are likely congenital in origin. [1] However, in certain instances they might be acquired and is usually secondary to either trauma or after invasive cardiac procedures [2-8]. Physiologic derangements depend on the site of origin, size of the fistulae and on the receiving chamber [1,9-11]. It has been reported that the right coronary artery is the most likely site and the right ventricle the major receiving chamber [1,9]. We describe a case of female patient with severe aortic stenosis and interventricular septal hypertrophy that underwent bioprosthetic aortic valve replacement with concomitant septal myectomy. On subsequent follow-up and while asymptomatic, an abnormal continuous color flow signal with a predominant diastolic component, consistent with coronary arterial flow, traversing the septum into the left ventricle was identified.
Case Report
A 74-year-old female with a history of severe aortic stenosis and interventricular septal hypertrophy underwent bioprosthetic aortic valve replacement with concomitant septal myectomy. Two months after the surgical intervention she presented to another hospital with syncope. On presentation, it was described that this obese patient was bradycardic with a heart rate of 40 beats per minute with stable blood pressure readings. No jugular venous distention was noted and occasional cannon A waves were noted. Examination of the lungs revealed adequate aeration in all fields with no crackles or wheezing. Point of maximum impulse was not displaced. Regular heart sounds with variable intensity were noted with no atrial or ventricular gallops but an early systolic murmur grade II/VI was described noted at the left sternal border. Due to the symptomatic bradycardia a dual chamber pacemaker was recommended and placed without complications. The patient was subsequently discharged home 24 hours after the pacemaker implantation. Four months post pacemaker implantation she was seen in follow-up and she doing fine and reported no complaints. An echocardiogram was obtained and it was reported that normal left ventricular systolic function as well as prosthetic aortic valve function were noted, with no other abnormalities.
The patient then relocated and was seen a year later for the first time at our institution. An echocardiogram obtained at the time of her initial visit, while still asymptomatic, showed normal left ventricular chamber dimensions, systolic function, and bioprosthetic valve function. In addition, a pacer wire that was correctly positioned in the right ventricular apex was also seen. However, an abnormal color flow signal arising from the interventricular septum with a predominant flow away from the transducer into the left ventricular cavity was noted. Continuous and pulse wave Doppler interrogation demonstrated a continuous flow with a predominant diastolic component, as shown in Figures 1, 2 and 3, all these findings consistent with coronary arterial flow. This abnormal color flow signal, traversing the interventricular septum, was never identified in previous studies.
Figure 1 Parasternal long axis view showing an abnormal color flow signal arising from a thick interventricular septum with a predominant flow away from the transducer into the left ventricular cavity. A predominant diastolic component is shown. The position of the pacer wire and prosthetic aortic valve are also shown. (RV = right ventricle, LV = left ventricle, LA = left atria).
Figure 2 Similar parasternal image showing the predominantly diastolic color flow image component. (RV = right ventricle, LV = left ventricle, LA = left atria).
Figure 3 Short axis view showing the color flow signal with regards to the septum. (RV = right ventricle, LV = left ventricle, LA = left atria).
The patient presented in this case was asymptomatic at the time of diagnosis and consequently no intervention was recommended. The patient was seen in follow-up and was doing fine, reporting no complaints.
Discussion
Communications between coronary arteries and cardiac chambers are often congenital malformations [1]. However, in certain instances they are acquired usually as a result from either trauma or after invasive cardiac procedures [2-8]. Physiologic derangements depend on the site of origin, size of the fistulae and on the receiving chamber [1,9-11]. The right coronary artery is the most likely site of origin in 55% of the cases while the left coronary artery system is involved in 35%. The major receiving chamber is the right ventricle (45%), right atrium (25%), pulmonary artery (15–20%) and less commonly in the coronary sinus (7%) [1,9]. In all reports, coronary cameral fistula least often drains into the left atrium or left ventricle. The size of the fistulae and the difference between the systemic and receiving chamber resistances determine the volume of the shunt. Regardless of these variables, flow moves from the coronary arteries to the lower pressure chambers. Most coronary artery fistulae are small and consequently myocardial blood flow is not compromised and the patient is usually asymptomatic. In some cases, however, coronary artery steal does occur with consequent development of ischemia in myocardial segments perfused by the coronary artery distal to the fistula [1,9-11].
Spontaneous closure has been reported in children but is less frequently noted in adults [12,13]. Spontaneous closure may be a more common occurrence in biopsy-related coronary cameral fistula.
A loud continuous murmur usually located at the lower sternal border identifies many patients with coronary artery fistula. In the case presented, a grade II/VI systolic murmur was documented in this obese patient.
Significantly enlarged coronary arteries can be detected by two-dimensional echocardiography. The actual diagnosis of a coronary artery fistula can often be made with transthoracic two-dimensional and color Doppler echocardiography in children. However, in adults, transesophageal echocardiography may be more sensitive [14-17]. Nowadays, the anatomic course and localization of coronary artery fistula can be made with either contrast-enhanced computer tomography with three-dimensional reconstruction or magnetic resonance imaging [18-21].
Hemodynamically significant fistula with a left to right shunt may lead to congestive heart failure, pulmonary artery hypertension, and myocardial ischemia by steal phenomenon with or without cardiac arrhythmias. The hemodynamic consequence of the coronary cameral fistula depends on the size of the fistula and the communicating chamber. Uncommon sequelae associated with this clinical entity include endocarditis, embolization of thrombotic material from the aneurysmal fistula, and potential rupture of the aneurysm [22]. Hemodynamically insignificant fistulae are clinically silent and if not associated with other abnormal findings usually require no further treatment. The risk of endocarditis and the need for endocarditis prophylaxis in untreated patients remains controversial. In contrast, large and hemodynamically significant fistulae should be closed by ligation [22-24]. Smaller coronary fistulae tend to get larger with age. As a result, it is usually recommended that elective closure be performed early in patients who have symptoms or who are asymptomatic but have a continuous murmur or a systolic murmur with an early diastolic component [22-24].
Given the characteristics of this case that involved the insertion of a bioprosthetic valve, myectomy and pacemaker insertions; each variable needed individual consideration. First, traumatic formation of a ventricular septal defect secondary to either surgical myectomy or during aortic valve replacement was a strong consideration. A defect was never identified; the residual interventricular septum was still thick, and most importantly the characteristic spectral signal of a ventricular septal defect had a predominant systolic flow component. Secondly, a trauma to the ventricular septum after pacemaker implantation could also account for this. Given the adequate position of the pacemaker lead in the right ventricular apex and our own previous description of a case reporting pacemaker trauma [25], this possibility appears unlikely. Alternatively, a septal defect could have been created at the time of the pacemaker insertion. However, absence of the characteristic echocardiographic features of a ventricular septal defect, as previously explained, made this possibility also improbable. Therefore, we postulate that this abnormal continuous flow, with a predominant diastolic component, was most consistent with coronary arterial flow. Consequently, given the location of this abnormal flow the most likely explanation was trauma to the septal perforator artery during myectomy resulting in a fistula into the left ventricle. The delayed clinical presentation might have been the result of an initial aneurysmal dilatation of the involved coronary artery with subsequent rupture and formation of the fistula.
Conclusion
We describe a case of female patient with severe aortic stenosis and interventricular septal hypertrophy that underwent bioprosthetic aortic valve replacement with concomitant septal myectomy. On subsequent follow-up an abnormal flow traversing the septum into the left ventricle was identified and Doppler interrogation demonstrated a continuous flow, with a predominantly diastolic component, consistent with a coronary arterial flow. The literature on coronary cameral fistulas was reviewed and the etiology of the diagnostic findings discussed. In our patient, a coronary artery to left ventricle fistula was the most likely explanation secondary to a trauma to the septal perforator artery during myectomy. Since the patient was asymptomatic at the time of diagnosis no intervention was recommended and has done well on follow-up.
List of abbreviations
None
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
Dr. López-Candales interpreted the echocardiogram and Dr. Vivek and Dr. López-Candales prepared the manuscript and the literature reviewed. Both authors have approved the final review of the manuscript.
==== Refs
Said SA van der Werf T Acquired coronary cameral fistulas: are these collaterals losing their destination? Clin Cardiol 1999 22 297 302 10198740
Hauser M Congenital anomalies of the coronary arteries Heart 2005 91 1240 1245 16103577 10.1136/hrt.2004.057299
Misra P Tandon N Coronary artery fistulas: case report and review of literature Am Heart Hosp J 2003 1 252 254 15785202
Wieczorek M Braun P Schoels W Acquired arteriovenous fistula and chronic occlusion of the left anterior descending coronary artery in a patient with previous ablation of a left-sided accessory pathway Pacing Clin Electrophysiol 2005 28 874 876 16105019 10.1111/j.1540-8159.2005.00170.x
Cetin G Saltik L Ozkara A Babaoglu K Unusual presentation of myocardial ischemia after arterial switch operation Eur J Cardiothorac Surg 2005 27 518 519 15740969 10.1016/j.ejcts.2004.12.027
Chiu SN Wu MH Lin MT Wu ET Wang JK Lue HC Acquired coronary artery fistula after open heart surgery for congenital heart disease Int J Cardiol 2005 103 187 192 16080979 10.1016/j.ijcard.2004.09.005
Chen WH Lee PY Wang EP Left anterior descending artery-to-right ventricle fistula and left ventricular free wall perforation after rotational atherectomy and stent implantation J Invasive Cardiol 2005 17 450 451 16079454
Gasparovic H Novick W Anic D Belina D Gjergja Z Jelic I Iatrogenic coronary artery fistula in a patient with a single coronary artery Thorac Cardiovasc Surg 2002 50 109 111 11981716 10.1055/s-2002-26695
Hoffman JI Emmanoulides GC, Riemenschneider TA, Allen HD, Gutgesell HP Congenital anomalies of the coronary vessels and the aortic root Heart disease in infants, children and adolescents 1995 5 Williams and Wilkins, Baltimore 780
Gupta NC Beauvais J Physiologic assessment of coronary artery fistula Clin Nucl Med 1991 16 40 42 1999055
McClung JA Belkin RN Chaudhry SS Left circumflex coronary artery to left atrial fistula in a patient with mitral stenosis: invasive and noninvasive findings with pathophysiologic correlation Cathet Cardiovasc Diagn 1996 37 52 54 8770480 10.1002/(SICI)1097-0304(199601)37:1<52::AID-CCD13>3.0.CO;2-Y
Sunder KR Balakrishnan KG Tharakan JA Titus T Pillai VR Francis B Kumar A Bhat A Shankaran S Coronary artery fistula in children and adults: a review of 25 cases with long-term observations Int J Cardiol 1997 58 47 53 9021427 10.1016/S0167-5273(96)02792-1
Cotton JL Diagnosis of a left coronary artery to right ventricular fistula with progression to spontaneous closure J Am Soc Echocardiogr 2000 13 225 8 10708471
Vitarelli A De Curtis G Conde Y Colantonio M Di Benedetto G Pecce P De Nardo L Squillaci E Assessment of congenital coronary artery fistulas by transesophageal color Doppler echocardiography Am J Med 2002 113 127 33 12133751 10.1016/S0002-9343(02)01157-9
Goswami NJ Zabalgoitia M Localization of a coronary artery fistula using contrast transesophageal echocardiography J Am Soc Echocardiogr 2002 15 839 840 12174354 10.1067/mje.2002.119912
Krishnamoorthy KM Rao S Transesophageal echocardiography for the diagnosis of coronary arteriovenous fistula Int J Cardiol 2004 96 281 283 15262046 10.1016/j.ijcard.2003.03.031
Cox ID Heald SC Murday AJ Value of transesophageal echocardiography in surgical ligation of coronary artery fistulas Heart 1996 76 181 8795485
Allum C Knight C Mohiaddin R Poole-Wilson P Use of magnetic resonance imaging to demonstrate a fistula from the aorta to right atrium Circulation 1998 97 1024 9529272
Rathi VK Mikolich B Patel M Doyle M Yamrozik J Biederman RW Coronary artery fistula; non-invasive diagnosis by cardiovascular magnetic resonance imaging J Cardiovasc Magn Reson 2005 7 723 725 16136864 10.1081/JCMR-65643
Hara H Moroi M Araki T Kunimasa T Tsunoda T Suzuki M Sugi K Nakamura M Coronary artery fistula with an associated aneurysm detected by 16-slice multidetector row computed tomographic angiography Heart Vessels 2005 20 184 185 16025371 10.1007/s00380-004-0810-8
Ropers D Moshage W Daniel WG Jessl J Gottwik M Achenbach S Visualization of coronary artery anomalies and their anatomic course by contrast-enhanced electron beam tomography and three-dimensional reconstruction Am J Cardiol 2001 15 193 197 11152838 10.1016/S0002-9149(00)01315-1
Liberthson RR Sagar K Berkoben JP Weintraub RM Levine FH Congenital coronary arteriovenous fistula. Report of 13 patients, review of the literature and delineation of management Circ 1979 59 849 54
Kamiya H Yasuda T Nagamine H Sakakibara N Nishida S Kawasuji M Watanabe G Surgical treatment of congenital coronary artery fistulas: 27 years' experience and a review of the literature J Card Surg 2002 17 173 177 12220072
Wang S Wu Q Hu S Xu J Sun L Song Y Lu F Surgical treatment of 52 patients with congenital coronary artery fistulas Chin Med J 2001 114 752 755 11780343
López-Candales A Desai B Cardioversion induced pacemaker complication Geriatric Cardiology 1996 5 43
|
16277656
|
PMC1298315
|
CC BY
|
2021-01-04 16:38:31
|
no
|
Cardiovasc Ultrasound. 2005 Nov 8; 3:35
|
utf-8
|
Cardiovasc Ultrasound
| 2,005 |
10.1186/1476-7120-3-35
|
oa_comm
|
==== Front
Cell Commun SignalCell communication and signaling : CCS1478-811XBioMed Central London 1478-811X-3-131626643310.1186/1478-811X-3-13ResearchApical membrane P2Y4 purinergic receptor controls K+ secretion by strial marginal cell epithelium Marcus Daniel C [email protected] Jianzhong [email protected] Jun Ho [email protected] Elias Q [email protected] Margaret A [email protected] Philine [email protected] Cellular Biophysics Laboratory, Dept. Anatomy & Physiology, Kansas State University, Manhattan, KS 66506 USA2 Cell Physiology Laboratory, Dept. Anatomy & Physiology, Kansas State University, Manhattan, KS 66506 USA3 Molecular Pharmacology Laboratory, Dept. Pharmacology, Creighton School of Medicine, Omaha, NE 68178 USA2005 2 11 2005 3 13 13 9 6 2005 2 11 2005 Copyright © 2005 Marcus et al; licensee BioMed Central Ltd.2005Marcus 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 was previously shown that K+ secretion by strial marginal cell epithelium is under the control of G-protein coupled receptors of the P2Y family in the apical membrane. Receptor activation by uracil nucleotides (P2Y2, P2Y4 or P2Y6) leads to a decrease in the electrogenic K+ secretion. The present study was conducted to determine the subtype of the functional purinergic receptor in gerbil stria vascularis, to test if receptor activation leads to elevation of intracellular [Ca2+] and to test if the response to these receptors undergoes desensitization.
Results
The transepithelial short circuit current (Isc) represents electrogenic K+ secretion and was found to be decreased by uridine 5'-triphosphate (UTP), adenosine 5'-triphosphate (ATP) and diadenosine tetraphosphate (Ap4A) but not uridine 5'-diphosphate (UDP) at the apical membrane of marginal cells of the gerbil stria vascularis. The potencies of these agonists were consistent with rodent P2Y4 and P2Y2 but not P2Y6 receptors. Activation caused a biphasic increase in intracellular [Ca2+] that could be partially blocked by 2-aminoethoxy-diphenyl borate (2-APB), an inhibitor of the IP3 receptor and store-operated channels. Suramin (100 μM) did not inhibit the effect of UTP (1 μM). The ineffectiveness of suramin at the concentration used was consistent with P2Y4 but not P2Y2. Transcripts for both P2Y2 and P2Y4 were found in the stria vascularis. Sustained exposure to ATP or UTP for 15 min caused a depression of Isc that appeared to have two components but with apparently no chronic desensitization.
Conclusion
The results support the conclusion that regulation of K+ secretion across strial marginal cell epithelium occurs by P2Y4 receptors at the apical membrane. The apparent lack of desensitization of the response is consistent with two processes: a rapid-onset phosphorylation of KCNE1 channel subunit and a slower-onset of regulation by depletion of plasma membrane PIP2.
==== Body
Background
A high concentration of K+ is maintained in the lumen of the cochlea via electrogenic secretion by the strial marginal cell epithelium [1]. One pathway of regulation is the coupling of purinergic receptors on the apical membrane of these cells to the apical potassium channels (IKs) which mediate secretion [2]. These receptors are responsive to both ATP and UTP as agonists and they were found to exert their action via the phospholipase C – protein kinase C intracellular signal pathway [3]. At the time of the original investigations, the purinergic receptor field recognized only 1 receptor responding to uracil nucleotides (P2U, now P2Y2 receptor). The known members of the mammalian metabotropic, G protein-coupled, purinergic receptors has since grown to P2Y1, P2Y2, P2Y4, P2Y6, P2Y11, P2Y12, P2Y13 and P2Y14 [4]. In addition to P2Y2, both the P2Y4 and P2Y6 receptors also respond to uracil nucleotides. On the basis of pharmacologic agonist and antagonist profiles, P2Y6 could be distinguished from P2Y2 and P2Y4 by the greater potency of UDP over UTP [5,6]. The pharmacologic distinction between P2Y2 and P2Y4 was more ambiguous and the accepted criteria changed rapidly [7].
The question of the subtype of P2Y receptor mediating regulation of K+ secretion by strial marginal cells has been addressed by immunohistochemistry, which suggests the presence of P2Y4 in the apical marginal cell membrane [8]. However, a functional demonstration was lacking, the time-course of activation was not investigated and an independent demonstration of gene expression was not available.
The present study was conducted with the goals of 1) obtaining dose-response profiles for potentially definitive agonists of P2Y2, P2Y4 and P2Y6 receptors; 2) refining the profile by the use of antagonists, 3) testing for desensitization of the response to agonist and 4) determining the presence of transcripts for pyridine-sensitive P2Y receptors. The results support the conclusion that regulation of K+ secretion by strial marginal cells occurs by apical P2Y4 receptors.
Results
ATP perfused for 30 s at the apical side of strial marginal cell epithelium caused a monophasic decrease in Isc (Fig. 1A), consistent with previous findings [2]. Removal of ATP led to a recovery of Isc after an initial overshoot. Similar responses were seen from perfusion of other agonists, such as UTP and Ap4A (Fig. 1B &1C).
Figure 1 Summary recordings of Isc (Ussing chamber) of strial marginal cells during apical perfusion of A) ATP (10-3 M; 30 s; n = 6), B) UTP (10-4 M; 30 s; n = 7) and C) Ap4A (3 × 10-4 M; 60 s; n = 4). Vertical bars are SEM (not all shown; spaced for clarity).
The P2Y6 agonist, UDP, was tested after removal of contamination by UTP in the commercial product [6]. Contaminating UTP was digested by hexokinase in the presence of glucose. The enzymatically-purified UDP produced little effect on Isc (Fig. 2).
Figure 2 Concentration-response curves for inhibition of Isc (Ussing chamber) by P2 agonists. Mean ± SEM, n = 4 to 7 for each point. Data fit to Hill equation; fit parameters are given in the text.
Concentration-response curves for the purinergic agonists are summarized in Figure 2. The potency order was UTP ≥ ATP > Ap4A >> UDP with EC50 values for UTP and ATP of about 2.3 and 1.2 × 10-5 M. The relative potencies of the agonists are consistent with an action on P2Y2 and/or P2Y4 receptors.
The purinergic receptor antagonist, suramin, was tested for effectiveness in blocking the response to apical UTP. Suramin at 100 μM had no inhibitory effect on the action of the agonist, UTP (Fig. 3). UTP (1 μM) caused a decrease in Isc by 20.6 ± 5.4% in the absence of suramin and by 21.9 ± 5.3% (P > 0.05, n = 5) in the presence of suramin.
Figure 3 Summary of relative decrease in Isc (vibrating probe) caused by UTP (1 μM) in the absence or presence of suramin (100 μM).
Both P2Y2 and P2Y4 receptor subtypes have been reported to undergo desensitization within 5–15 min [9,10]. Surprisingly, we found that sustained exposure (15 min) of the apical membrane to ATP led to a biphasic but sustained inhibition of Isc (Fig. 4). The recovery of Isc after removal of agonist was much slower, incomplete and without overshoot compared to acute exposure to agonist.
Figure 4 Summary recordings of Isc (Ussing chamber) of strial marginal cells during apical perfusion of ATP (10-3 M) for 15 min. Vertical bars are SEM (not all shown; spaced for clarity; n = 6).
Perfusion of UTP (10 μM) for 30 s led to an increase in intracellular [Ca2+] that had both a peak and plateau phase and was repeatable (Fig. 5, top). Both the peak and plateau were significantly reduced by 2-APB (75 μM) (to 26.5 ± 3.9% and 25.1 ± 4.6%; Fig. 5, middle). 2-APB is an inhibitor of the IP3 receptor at this concentration [11,12] and can also inhibit other transporters including store-operated Ca2+ channels [13]. Store operated channels, however, would not be activated under the present experimental conditions, since the Ca2+ stores were not emptied prior to measurement. In the absence of external Ca2+, both the peak and plateau responses to UTP remained but were significantly reduced (to 68.9 ± 3.0% and 60.5 ± 4.2%; Fig. 5, bottom). The reductions, however, were likely the result of a general decrease in intracellular [Ca2+] seen after removal of bath Ca2+ prior to perfusion of UTP. These findings taken together suggest that intracellular stores are the primary source of the increase in intracellular [Ca2+] induced by UTP.
Figure 5 Changes of intracellular Ca2+ concentration ([Ca2+]) of marginal cell layer in response to three consecutive perfusions of UTP (10 μM; 30 s). Top: Time control, no additional treatments (n = 7); middle: second exposure to UTP in the presence of the IP3 receptor inhibitor 2-APB (n = 9); bottom: second exposure to UTP in the absence of bath Ca2+ (n = 7). Vertical bars are SEM (not all shown; spaced for clarity).
We tested for the presence of transcripts for P2Y2 and P2Y4 receptors in stria vascularis. Primers proven to recognize gerbil P2Y2 and P2Y4 [14] were used to amplify single, gene-specific bands in stria vascularis (Fig. 6); 301 base pairs (bp) for P2Y2 and 447 bp for P2Y4. Controls in which the reactions were run in the absence of reverse transcriptase (-RT) demonstrated the absence of contributions from genomic DNA. PCR products were analyzed for their sequence to confirm the identity of the bands. Sequences were the same as found previously in the gerbil vestibular labyrinth [14] (GenBank accession numbers: P2Y2, AF313448; P2Y4, AF313447).
Figure 6 Gel electrophoresis of reverse transcription-polymerase chain reaction (RT-PCR) products from gerbil stria vascularis. Gene-specific primers were used for detection of transcripts for segments of P2Y2 and P2Y4. +, Reactions performed in the presence of reverse transcriptase; -, reactions performed in the absence of reverse transcriptase. Position of the bands for the expected lengths of the RT-PCR products [447 bp, P2Y4; 301 bp, P2Y2] are indicated and a 100-bp ladder is shown (L).
Discussion
Purinergic agonists control strial marginal cell K+ secretion, a process observed as a transepithelial short circuit current, Isc [1,2]. Strial marginal cells secrete K+ by a constellation of transporters previously described [1]. K+ is taken up across the basolateral membrane by the Na+, K+-ATPase and the Na+, K+, 2Cl--cotransporter. Na+ carried into the cell on the cotransporter is removed by the Na+-pump and Cl- carried into the cell on the cotransporter leaves by passive diffusion across a large Cl- conductance in the basolateral membrane [15-17]. K+ taken up by both the Na+-pump and cotransporter is secreted across the apical membrane by diffusion through IKs channels [18], consisting of KCNQ1 alpha subunits and KCNE1 beta regulatory subunits [19,20]. The epithelium regulates K+ secretion by a variety of signal pathways, including those initiated by apical purinergic receptors, that converge at the IKs channel complex [21].
The previous demonstration of control of Isc in strial marginal cell epithelium by "P2U" receptors included determination of a concentration-response to ATP and comparative activities of other nucleotides (UTP, 2-MeS-ATP and α, β-meth-ATP) at 1 μM [2]. Full concentration-response curves were obtained in the present study to better define the active subtype. The lack of response to UDP eliminated P2Y6 as a candidate subtype [22-24]. The similar potency of ATP and UTP is consistent with hP2Y2 [23], rP2Y2 [22] but not hP2Y4 [6]. In fact, ATP acts as a competitive antagonist at the hP2Y4 receptor [25]. However, these results alone do not identify the strial marginal cell apical purinergic receptor as P2Y2 since it was subsequently found that rP2Y4 has an agonist potency sequence similar to rat and human P2Y2 [7,25]. Since the gerbil is expected to be more closely related to other rodents such as the rat, than to human, our finding is consistent with the contribution of P2Y4 and/or P2Y2.
The potency order of the agonists UTP, ATP, Ap4A and UDP in regulation of Isc from strial marginal cells is the same as for rodent P2Y4 receptors in other systems, including cloned mouse P2Y4 [26] and gerbil vestibular dark cells [2,27] even though the EC50 values are substantially shifted. The absolute values of EC50's are alone not indicative of receptor binding for G protein-coupled receptors, although the relative potency remains constant. Receptor density plays a large role in defining the response of downstream effectors (IKs channels in this case) and differences in density can lead to significant shifts in EC50 curves.
Transcripts for both P2Y2 and P2Y4 were found in stria vascularis. The identity of the cell type(s) within the stria that contain these transcripts is not certain from these findings alone. Recent immunohistochemical findings show staining for the P2Y4 receptor at the apical membrane of strial marginal cells, while the antibody for P2Y2 stained at the basolateral region of the marginal cells and/or the intermediate cells [8], consistent with the functional evidence for apical P2Y4 shown here. The transcript expression confirms the immunohistochemical findings by an independent means and thereby provides an important verification of the specificity of the P2Y antibodies used in the previous study [8].
Ap4A was found to be a potent agonist at the hP2Y2 receptor [23] but much less potent than ATP at the hP2Y4 [28] and rP2Y4 receptors [24], similar to our finding in gerbil strial marginal cells. In spite of this similarity, uncertainty arises in the identification of the apical receptor in strial marginal cells as P2Y4 on this basis alone since there is also a conflicting report from Bogdanov et al. who found Ap4A to be equally potent as ATP at rP2Y4 receptors [7].
Suramin inhibits several P2 receptors, but it was found recently that at a high concentration (100 μM) it can be used to distinguish both rP2Y2 [22] and hP2Y2 [29] from both rP2Y4 [7] and hP2Y4 [29]. This criterion applied to the present results points to the apical P2Y purinergic receptor in gerbil strial marginal cells as the P2Y4 subtype. This conclusion based on function and pharmacology is consistent with the recent report of immunohistochemical localization of P2Y4 at the apical membrane of strial marginal cells [8]. We reported earlier that the response to ATP is increased in the absence of divalent cations [2]. This increased effect suggests that the apical receptor is preferentially activated by an uncomplexed form of ATP, as in aortic endothelial cells [30].
The overshoot of Isc after removal of purinergic agonists from the apical perfusate (Fig. 1) is most likely due to a release of K+ accumulated in the epithelial cells during the inhibition of secretion across the apical membrane. The basolateral K+ uptake mechanisms continue to operate for a time after inhibition of apical IKs channel complexes by activation of purinergic receptors, bringing the cytosolic K+ concentration to a level higher in electrochemical potential above the apical bath than in the absence of agonist. Upon removal of agonist this "extra pool" of K+ is suddenly released, resulting in the observed overshoot of Isc. The same phenomenon was reported previously [31] when K+ secretion was first diminished by raising the apical K+ concentration, thereby reducing the outward gradient across the apical cell membrane. Suddenly returning the apical perfusate K+ concentration to the original low level led to an overshoot of Isc, as observed with the purinergic agonists.
The decrease in Isc observed in response to purinergic agonists could be due a priori to a reduction of electrogenic K+ secretion but could also be accounted for by a stimulation of secretion of Cl- or absorption of Na+. However, it was found in a previous study that the decrease in Isc could be completely accounted for by a decrease in K+ secretion. Apical perfusion of 100 μM ATP lead to a decrease of Isc by 37.1 ± 3.7% and of K+ secretory flux by 22.2 ± 5.5% [3].
The peak and plateau increases in intracellular [Ca2+] during perfusion of agonist in the presence and absence of bath Ca2+, as well as the reduction by 2-APB, are consistent with a release of Ca2+ from intracellular stores. Ikeda et al. found a small monophasic increase in intracellular [Ca2+] of the entire stria vascularis that could not be analyzed further [32]. The same laboratory evaluated the purinergic response of cultured marginal cells and found agonist-induced increases in intracellular [Ca2+] that were not reduced upon removal of Ca2+ from the bath [33]. Those cells were derived from guinea pig, the responses were monophasic and UTP was not tested as agonist.
The apparent absence of desensitization (Fig. 4) was a surprising finding in view of reports of rapid desensitization of both cloned P2Y2 and cloned P2Y4 purinergic receptors [9,10]. Desensitization has typically been observed as a decrease of inositol phosphate production and/or cytosolic Ca2+ increase, either directly [34,35] or via the effect of Ca2+ on transepithelial anion secretion [36]. Since the sustained response to UTP (Figure 4) is biphasic, it therefore may represent the summation of at least 2 processes.
The question arises as to the basis for the observed sustained decrease in IKs. Possible explanations include: 1) The P2Y4 receptor desensitizes several minutes after the onset of agonist, but the initial activation of P2Y4 leads to phosphorylation of KCNE1 via PKC [3] and the channel subunit remains phosphorylated due to inhibition of phosphatase activity. 2) Activation of P2Y4 leads to localized depletion of phosphatidylinositol-4,5-bisphosphate (PIP2) in the plasma membrane and to the consequent deactivation of the PIP2-dependent IKs channel.
Isc is controlled by apical P2Y4 receptors via the protein kinase C pathway by phosphorylation of the beta subunit of the IKs channel complex [3] [vida infra]. The sustained reduction in Isc and the incomplete recovery of Isc after prolonged exposure to agonist can both be explained if there is a decrease of phosphatase activity at the phosphorylation site. The channel subunit would then be expected to remain phosphorylated and the effect on Isc would be sustained even with a desensitization of the receptor. However, dephosphorylation rates have been reported to increase with elevated intracellular [Ca2+] in olfactory sensory neurons [37], arguing against this hypothesis.
The second explanation is more consistent with our findings. It has been found that activation of receptors (such as P2Y4) coupled to Gq/11 deplete PIP2 [38] and that IKs (KCNQ1/KCNE1) activity is controlled by PIP2 [39]. In fact, the time course of desensitization of another K+ channel by phenylephrine induced depletion of PIP2 via α1-adrenergic receptors is similar to the secondary effects seen here [38].
Physiological significance
Purinergic receptors have been identified on the apical membrane of many of the cells forming the border of the cochlear duct [40-42]. Indeed, IKs of strial marginal cells in the cochlea is inhibited by UTP in much the same way as the homologous vestibular dark cells [2,17,27], although the lower potency of agonists on marginal cells suggests a lower density of receptors. The P2Y4 receptors can thereby provide the strial marginal cells with an autocrine as well as paracrine signaling pathway. Autocrine signaling is important for these cells since they have no gap junction communication [43], an unusual occurrence for epithelial cells. Paracrine signaling is also important in this organ since the rate of K+ secretion must be adjusted for variations in K+ efflux during stimulation of the cochlea by sound. In fact, the P2 receptors have been proposed to act as a mechanism to protect the inner ear from noise damage [44].
Conclusion
The response to purinergic agonists at the apical side of strial marginal cells was shown here to be functionally mediated by P2Y4 receptors, previously shown to be present by immunostaining. The decrease in electrogenic K+ secretion evoked by agonists was a) initially accompanied by a biphasic increase in intracellular Ca2+ and b) surprisingly sustained in the continued presence of agonist. The sustained decrease in K+ secretion was best explained by rapid-onset phosphorylation of KCNE1 channel subunit and a slower-onset of regulation by depletion of plasma membrane PIP2. Impaired PIP2 and IKs interaction was recently shown to be the basis of several mutations causing long QT syndrome, a genetic disease characterized by cardiac arrhythmias and deafness [45].
Methods
Animals and tissues
Gerbils (4–5 week old females) were anesthetized by injection of sodium pentobarbital (50 mg/kg, i.p.) and the temporal bones were removed. The method for dissecting strial marginal cell epithelium from the cochlear lateral wall was described previously [1]. Dissected epithelia were either transferred to a micro-Ussing chamber for measurement of the equivalent short circuit current (Isc), to a perfusion chamber for [Ca2+] measurement or were frozen in liquid nitrogen within 10 min of death for reverse transcription-polymerase chain reaction (RT-PCR). All procedures conformed to protocols approved by the Institutional Animal Care and Use Committee.
Short-circuit current measurements
The micro-Ussing chamber for inner ear tissues has been described previously [3]. Briefly, the diameter of the aperture separating the apical and basolateral side half-chambers was 80 μm and each side was continuously perfused independently at 37°C with an exchange of solution accomplished within 1 s. Isc was measured with a 4-wire epithelial current clamp and recorded with a computer data acquisition system with 16 bits resolution. Samples were acquired at 32 Hz and decimated by a factor of 10. Perfusion changes were planned and carefully timed so that experiments from each experimental series could be averaged (Figures 1 &4). Results were analyzed and plotted using Origin software (OriginLab, Northampton, MA).
In one series of experiments (Figure 3), the relative Isc was measured with a vibrating probe as described previously [46]. Briefly, the lateral cochlear wall with stria vascularis was mounted in a perfusion chamber on the stage of an inverted microscope (Nikon TE-300) and continuously perfused with the same solution used for the micro-Ussing chamber at 37°C. The Isc relative to the initial control value was monitored by vibrating a platinum-iridium wire microelectrode coated with Pt-black on the tip. The probe was positioned 20–30 μm from the apical surface of the epithelium. The bath references were 26-gauge Pt-black electrodes. The probe tip was vibrated in the range of 200–400 Hz over an excursion of about 20 μm. Vibration between two positions within the line of current flow yields voltages in the low nanovolt range that correspond to current flow through the resistive physiological saline [47]. The output of the probe amplifier was recorded with ASET software (Science Wares, East Falmouth, MA). Although the single perfusate reached the entire stria vascularis, it was determined that brief changes of solution (≤30 s) effectively reached only the apical surface; the extremely tight junctions of the basal cell layer restricted diffusion of large molecules such as agonists and antagonists to the interior of the stria and to the basolateral membrane of marginal cells [1].
Chemicals and solutions
In all experiments, both sides of the epithelium were perfused with a solution containing (in mM) NaCl 150, KH2PO4 0.4, K2HPO4 1.6, MgCl2 1, CaCl2 0.7, glucose 5, pH 7.4. All experimental agents were dissolved in this solution immediately before use. Suramin was purchased from Calbiochem (San Diego, CA) and adenosine 5'-triphosphate, 2-aminoethoxy diphenyl borate (2-APB), uridine 5'-triphosphate, uridine 5'-diphosphate (UDP), diadenosine tetraphosphate (Ap4A) and hexokinase from Sigma-Aldrich (St. Louis, MO). The 2-APB was predissolved in DMSO and used at a final DMSO concentration of 0.1%.
Intracellular Ca2+ measurements
Calcium was measured as described previously [48]. Briefly, stria vascularis was incubated for 20 min with the indicator dye 5 μM fluo-4-AM at 37°C (Molecular Probes, Eugene, OR) and mounted in the superfusion chamber on the stage of an inverted microscope (Diaphot, Nikon). The stria vascularis was folded into a loop with the marginal cell layer on the outside surface. An optical section of the marginal cells was observed and the recording slit, which defined the field of the Ca2+ measurement, was restricted to this cell layer.
The preparation was alternately illuminated at 600 and 488 nm (Deltascan, Photon Technology International, South Brunswick, N.J.). The fluorescence signal was detected by a photon-counter (Photon Technology International) at a rate of 4 Hz. Changes in the emission intensity were taken as measures of changes in intracellular Ca2+ ([Ca2+]i). The drift of signal due to leakage of dye from the cells was linearly subtracted.
Reverse transcription-polymerase chain reaction (RT-PCR)
Transcripts for P2Y2 and P2Y4 were assayed by RT-PCR using methods previously described for extraction of total RNA, DNase I treatment, PCR amplification, subcloning, and sequencing [14]. First strand cDNA synthesis was also performed as described previously [14] with the exception that 25 pmoles of random hexamers were used to prime the RNA. The sequence of the primers was based on the known sequences in the coding region of the rat, mouse and human receptors. P2Y2 primers: sense, 5'-GCTTCAACGAGGACTTCAAGTA(C/T)GTGC-3'; anti-sense, 5'-AGGTGAGGAAGAGGATGCTGCAGTAG-3'. P2Y4 primers: sense, 5'-CCAGAGGAGTTTGACCACTA-3'; anti-sense, 5'-CACCAAGGCCAGGGAGGA-3'. The primers were expected to yield RT-PCR products of 301 base pairs (bp) for P2Y2 and 447 bp for P2Y4 which were cloned and sequenced (GenBank accession numbers: P2Y2, AF313448; P2Y4, AF313447).
The PCR mixture was incubated as follows: 1 denaturation cycle for 5 min at 98°C; 40 amplification cycles consisting of denaturation for 45 sec at 95°C, annealing for 45 sec at 58°C, and extension for 45 sec at 72°C; and one extension cycle for 7 min at 72°C. PCR products were analyzed by horizontal electrophoresis in 2.0% agarose gels and visualized by ethidium bromide.
Statistics
Data are expressed as the mean ± S.E.M. (n = number of tissues) of the Isc and concentration-response curves from changes in Isc were normalized to the response to 1 mM ATP. The Student's t-test of paired samples was used to determine statistical significance and increases or decreases in Isc were considered significant for P < 0.05.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
DCM conceived, designed and coordinated the study and drafted the manuscript. JL acquired the short-circuit current data with the Ussing chamber. JHL acquired the short-circuit current data with the vibrating probe. EQS acquired and analyzed the intracellular calcium data. MAS designed, performed and carried out the RT-PCR experiments. PW designed, interpreted and supported the intracellular calcium experiments.
Acknowledgements
This work was supported by research grant number R01-DC00212 to DCM and R01-DC01098 to PW from the National Institute on Deafness and Other Communication Disorders, National Institutes of Health.
==== Refs
Wangemann P Liu J Marcus DC Ion transport mechanisms responsible for K+ secretion and the transepithelial voltage across marginal cells of stria vascularis in vitro Hear Res 1995 84 19 29 7642451 10.1016/0378-5955(95)00009-S
Liu J Kozakura K Marcus DC Evidence for purinergic receptors in vestibular dark cell and strial marginal cell epithelia of the gerbil Auditory Neurosci 1995 1 331 340
Marcus DC Sunose H Liu J Bennett T Shen Z Scofield MA Protein kinase C mediates P2U purinergic receptor inhibition of K+ channel in apical membrane of strial marginal cells Hear Res 1998 115 82 92 9472737 10.1016/S0378-5955(97)00180-9
Lee SY O'Grady SM Modulation of ion channel function by P2Y receptors Cell Biochem Biophys 2003 39 75 88 12835530 10.1385/CBB:39:1:75
Chang K Hanaoka K Kumada M Takuwa Y Molecular cloning and functional analysis of a novel P2 nucleotide receptor J Biol Chem 1995 270 26152 26158 7592819 10.1074/jbc.270.44.26152
Nicholas RA Watt WC Lazarowski ER Li Q Harden K Uridine nucleotide selectivity of three phospholipase C- activating P2 receptors: identification of a UDP-selective, a UTP- selective, and an ATP- and UTP-specific receptor Mol Pharmacol 1996 50 224 229 8700127
Bogdanov YD Wildman SS Clements MP King BF Burnstock G Molecular cloning and characterization of rat P2Y4 nucleotide receptor Br J Pharmacol 1998 124 428 430 9647463 10.1038/sj.bjp.0701880
Sage CL Marcus DC Immunolocalization of P2Y4 and P2Y2 purinergic receptors in strial marginal cells and vestibular dark cells J Membr Biol 2002 185 103 115 11891569 10.1007/s00232-001-0116-z
Otero M Garrad RC Velazquez B Hernandez-Perez MG Camden JM Erb L Mechanisms of agonist-dependent and -independent desensitization of a recombinant P2Y2 nucleotide receptor Mol Cell Biochem 2000 205 115 123 10821429 10.1023/A:1007018001735
Robaye B Boeynaems JM Communi D Slow desensitization of the human P2Y6 receptor Eur J Pharmacol 1997 329 231 236 9226417
Broad LM Braun FJ Lievremont JP Bird GS Kurosaki T Putney JW Jr Role of the phospholipase C-inositol 1,4,5-trisphosphate pathway in calcium release-activated calcium current and capacitative calcium entry J Biol Chem 2001 276 15945 15952 11278938 10.1074/jbc.M011571200
Ma HT Patterson RL van Rossum DB Birnbaumer L Mikoshiba K Gill DL Requirement of the inositol trisphosphate receptor for activation of store-operated Ca2+ channels Science 2000 287 1647 1651 10698739 10.1126/science.287.5458.1647
Parekh AB Putney JW Jr Store-operated calcium channels Physiol Rev 2005 85 757 810 15788710 10.1152/physrev.00057.2003
Marcus DC Scofield MA Apical P2Y4 purinergic receptor controls K+ secretion by vestibular dark cell epithelium Am J Physiol Cell Physiol 2001 281 C282 C289 11401851
Sage CL Marcus DC Immunolocalization of ClC-K chloride channel in strial marginal cells and vestibular dark cells Hear Res 2001 160 1 9 11591484 10.1016/S0378-5955(01)00308-2
Takeuchi S Ando M Kozakura K Saito H Irimajiri A Ion channels in basolateral membrane of marginal cells dissociated from gerbil stria vascularis Hear Res 1995 83 89 100 7541786 10.1016/0378-5955(94)00191-R
Wangemann P Comparison of ion transport mechanisms between vestibular dark cells and strial marginal cells Hear Res 1995 90 149 157 8974992 10.1016/0378-5955(95)00157-2
Shen Z Marcus DC Sunose H Chiba T Wangemann P IsK channel in strial marginal cells: Voltage-dependence, ion-selectivity, inhibition by 293B and sensitivity to clofilium Auditory Neurosci 1997 3 215 230
Barhanin J Lesage F Guillemare E Fink M Lazdunski M Romey G KvLQT1 and IsK (minK) proteins associate to form the IKs cardiac potassium current Nature 1996 384 78 80 8900282 10.1038/384078a0
Sanguinetti MC Curran ME Zou A Shen J Spector PS Atkinson DL Coassembly of KvLQT1 and minK (IsK) proteins to form cardiac IKs potassium channel Nature 1996 384 80 83 8900283 10.1038/384080a0
Marcus DC Sperelakis N Acoustic Transduction Cell Physiology Source Book 2001 San Diego: Academic Press 775 794
Chen ZP Krull N Xu S Levy A Lightman SL Molecular cloning and functional characterization of a rat pituitary G protein-coupled adenosine triphosphate (ATP) receptor Endocrinology 1996 137 1833 1840 8612522 10.1210/en.137.5.1833
Nicholas RA Lazarowski ER Watt WC Li Q Boyer J Harden TK Pharmacological and second messenger signalling selectivities of cloned P2Y receptors J Auton Pharmacol 1996 16 319 323 9131407
Webb TE Henderson DJ Roberts JA Barnard EA Molecular cloning and characterization of the rat P2Y4 receptor J Neurochem 1998 71 1348 1357 9751165
Kennedy C Qi AD Herold CL Harden TK Nicholas RA ATP, an agonist at the rat P2Y(4) receptor, is an antagonist at the human P2Y(4) receptor Mol Pharmacol 2000 57 926 931 10779375
Suarez-Huerta N Pouillon V Boeynaems J Robaye B Molecular cloning and characterization of the mouse P2Y4 nucleotide receptor Eur J Pharmacol 2001 416 197 202 11290369 10.1016/S0014-2999(01)00875-5
Marcus DC Sunose H Liu J Shen Z Scofield MA P2U purinergic receptor inhibits apical IsK/KvLQT1 channel via protein kinase C in vestibular dark cells Am J Physiol 1997 273 C2022 C2029 9435509
Communi D Motte S Boeynaems JM Pirotton S Pharmacological characterization of the human P2Y4 receptor Eur J Pharmacol 1996 317 383 389 8997625 10.1016/S0014-2999(96)00740-6
Charlton SJ Brown CA Weisman GA Turner JT Erb L Boarder MR Cloned and transfected P2Y4 receptors: characterization of a suramin and PPADS-insensitive response to UTP Br J Pharmacol 1996 119 1301 1303 8968535
Motte S Pirotton S Boeynaems JM Evidence that a form of ATP uncomplexed with divalent cations is the ligand of P2y and nucleotide/P2u receptors on aortic endothelial cells Br J Pharmacol 1993 109 967 971 8401949
Marcus NY Marcus DC Transepithelial electrical responses to sodium and potassium of nonsensory region of gerbil utricle Hear Res 1990 44 13 23 2157697 10.1016/0378-5955(90)90018-K
Ikeda K Suzuki M Furukawa M Takasaka T Calcium mobilization and entry induced by extracellular ATP in the non-sensory epithelial cell of the cochlear lateral wall Cell Calcium 1995 18 89 99 7585894 10.1016/0143-4160(95)90000-4
Suzuki M Ikeda K Sunose H Hozawa K Kusakari C Katori Y ATP-induced increase in intracellular Ca2+ concentration in the cultured marginal cell of the stria vascularis of guinea-pigs Hear Res 1995 86 68 76 8567423 10.1016/0378-5955(95)00055-9
Brinson AE Harden TK Differential regulation of the uridine nucleotide-activated P2Y4- and P2Y6-receptors: ser-333 and ser-334 in the carboxyl terminus are involved in agonist-dependent phosphorylation desensitization and internalization of the P2Y4 receptor J Biol Chem 2001 276 11939 11948 11114308 10.1074/jbc.M009909200
Garrad RC Otero MA Erb L Theiss PM Clarke LL Gonzalez FA Structural basis of agonist-induced desensitization and sequestration of the P2Y2 nucleotide receptor. Consequences Of truncation of the c terminus J Biol Chem 1998 273 29437 29444 9792648 10.1074/jbc.273.45.29437
Clarke LL Harline MC Otero MA Glover GG Garrad RC Krugh B Desensitization of P2Y2 receptor-activated transepithelial anion secretion Am J Physiol Cell Physiol 1999 276 C777 C787 10199807
Katanosaka K Miwa N Kawamura S Calcium-dependent phosphorylation of a protein in the frog olfactory cilia J Biochem (Tokyo) 2002 132 301 308 12153729
Cho H Hwang JY Kim D Shin HS Kim Y Earm YE Acetylcholine-induced phosphatidylinositol 4,5-bisphosphate depletion does not cause short-term desensitization of G protein-gated inwardly rectifying K+ current in mouse atrial myocytes J Biol Chem 2002 277 27742 27747 12019267 10.1074/jbc.M203660200
Loussouarn G Park KH Bellocq C Baro I Charpentier F Escande D Phosphatidylinositol-4,5-bisphosphate, PIP2, controls KCNQ1/KCNE1 voltage-gated potassium channels: a functional homology between voltage-gated and inward rectifier K+ channels EMBO J 2003 22 5412 5421 14532114 10.1093/emboj/cdg526
Housley GD Physiological effects of extracellular nucleotides in the inner ear Clin Exp Pharmacol Physiol 2000 27 575 580 10901385 10.1046/j.1440-1681.2000.03314.x
Brändle U Zenner HP Ruppersberg JP Gene expression of P2X-receptors in the developing inner ear of the rat Neurosci Lett 1999 273 105 108 10505627 10.1016/S0304-3940(99)00648-5
Housley GD Luo L Ryan AF Localization of mRNA encoding the P2X2 receptor subunit of the adenosine 5'-triphosphate-gated ion channel in the adult and developing rat inner ear by in situ hybridization J Comp Neurol 1998 393 403 414 9550147 10.1002/(SICI)1096-9861(19980420)393:4<403::AID-CNE1>3.0.CO;2-4
Kikuchi T Kimura RS Paul DL Adams JC Gap junctions in the rat cochlea: immunohistochemical and ultrastructural analysis Anat Embryol (Berl) 1995 191 101 118 7726389
Wang JC Raybould NP Luo L Ryan AF Cannell MB Thorne PR Noise induces up-regulation of P2X2 receptor subunit of ATP-gated ion channels in the rat cochlea Neuroreport 2003 14 817 823 12858039 10.1097/00001756-200305060-00008
Park KH Piron J Dahimene S Merot J Baro I Escande D Impaired KCNQ1-KCNE1 and phosphatidylinositol-4,5-bisphosphate interaction underlies the long QT syndrome Circ Res 2005 96 730 739 15746441 10.1161/01.RES.0000161451.04649.a8
Lee JH Marcus DC Endolymphatic sodium homeostasis by Reissner's membrane Neuroscience 2003 119 3 8 12763062 10.1016/S0306-4522(03)00104-0
Marcus DC Chiba T K+ and Na+ absorption by outer sulcus epithelial cells Hear Res 1999 134 48 56 10452375 10.1016/S0378-5955(99)00074-X
Scherer EQ Wonneberger K Wangemann P Differential desensitization of Ca2+ mobilization and vasoconstriction by ETA receptors in the gerbil spiral modiolar artery J Membr Biol 2001 182 183 191 11547341 10.1007/s00232-001-0041-1
|
16266433
|
PMC1298316
|
CC BY
|
2021-01-04 16:39:15
|
no
|
Cell Commun Signal. 2005 Nov 2; 3:13
|
utf-8
|
Cell Commun Signal
| 2,005 |
10.1186/1478-811X-3-13
|
oa_comm
|
==== Front
Clin Pract Epidemiol Ment HealthClinical Practice and Epidemiology in Mental Health : CP & EMH1745-0179BioMed Central 1745-0179-1-221625577410.1186/1745-0179-1-22ResearchPrevalence and co-occurrence of psychiatric symptom clusters in the U.S. adolescent population using DISC predictive scales Chen Kevin W [email protected] Ley A [email protected] William A [email protected] Department of Psychiatry, University of Medicine and Dentistry of New Jersey – Robert Wood Johnson Medical School, 671 Hoes Lane, Piscataway, New Jersey 08854, USA2 Center for Child and Family Policy, Duke University, Box 90545, Durham, NC 27708, USA2005 28 10 2005 1 22 22 17 8 2005 28 10 2005 Copyright ©2005 Chen et al; licensee BioMed Central Ltd.2005Chen 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.Objective
To estimate 12-month prevalence and co-occurrence of symptoms of specific mental problems among US adolescents (12–17 years) by age, sex and racial/ethnic subgroups.
Method
Data from the 2000 National Household Survey of Drug Abuse (NHSDA) adolescent sample are used to estimate prevalence and co-occurrence rates using the DISC predictive scales. Multiple logistic regressions were used to derive significant correlates of each domain of DPS-derived symptom cluster indicators of psychiatric problems and of severe comorbidity, with control of demographics and environmental factors.
Setting
The National Household Survey on Drug Abuse (NHSDA), a national household probability sample, includes a nationally representative sample of 12–17 year-old adolescents (N = 19,430), through in-home surveys.
Results
Three out of five adolescents screened positive for at least one DPS symptom cluster with estimates for specific symptom cluster ranging over 9.7% (substance use disorder), 13.4% (affective), 36.3% (disruptive-behavior), and 40.1% (anxiety). Co-occurrence was high with almost one-third of any DPS symptom cluster reporting multiple positive screens of four or more clusters. Blacks and younger females were most likely to report mental health problems and co-occurrence.
Conclusion
Mental health problems among U.S. youth may be far more common than previously believed, although these symptoms have not yet reached the point of clinical impairment. The data speak to important patterns of age, gender and racial/ethnic differences in mental health problems deserving of further study.
adolescentmental healthsymptom clusterscomorbidityDISC predictive scaledemographic correlates.
==== Body
Introduction
Although mental health has become increasingly important in our healthcare model and definition of health, especially during childhood and adolescence[1,2], little is known about actual prevalence of various psychiatric disorders among adolescents in the general population. Current estimates range from 10% to 60% [3-5], based mostly on local community or treatment samples and dependent upon the case-finding protocol used. Evidence from adult samples suggests that many disorders with psychiatric symptoms begin during childhood or adolescence[6,7], although estimates of the earlier onset of adult disorders may be unreliable due to errors in retrospective recall[8]. It is well documented that childhood and adolescent mental health problems have profound implications for negative adult sequelae [9-13]. By official estimate, at any one time, about 20% of US children and adolescents have at least one diagnosable mental health disorder[1]. However, in their longitudinal study of 9–16 year-old adolescents, Costello et al. reported 36.7% having at least one psychiatric disorder (by DSM-IV criteria) during the study period (although the prevalence of a number of disorders dropped precipitously by age 12)[4]. Turner & Gil reported that 60% of their community youth sample (age 19–21) met lifetime criteria for one or more mental disorders, including substance use disorders (SUD), with full DSM-IV diagnosis[5].
Recent national surveys provided some important epidemiological information about the prevalence of psychiatric disorders among adults [14-18], but our understanding of the scope of psychiatric disorders or symptoms among children and adolescents is limited by a number of methodological constraints, including small samples, or samples from clinics or institutions; overly specific research foci; and screening questions either limited in number or not closely aligned with DSM diagnostic criteria. Thus, the estimates of prevalence in the general U.S. adolescent population cannot be reliably made, presenting a major obstacle for estimating the course and magnitude of adolescents' mental health problems, the degree of both need and unmet need, and for developing effective prevention and intervention programs for this critical age group[4,19] Estimates from adult national probability samples suggest that nearly half of all adult cases report onset by age 14, and three-quarters by age 24[6,18]. However, it is not clear what major types of mental health problems adolescents are confronted with, and to what extent. Moreover, there is a lack of ethnic group-specific estimates of mental health needs that would be required for more targeted services for various subpopulations. Importantly, members of racial/ethnic minority groups report some of the highest rates of unmet need for treatment as adults[20,21].
One way to provide uniform estimates of mental problems would be to include in national surveys of the general population structured diagnostic interviews or selected screening items or scales of symptoms of psychiatric problems that have high predictive value for diagnosis. Although such a survey cannot constitute an actual diagnosis of disorder, it can facilitate identification of groups at high risk, as well as help elucidate differential patterns in important demographic groups, including age, gender and race/ethnicity. An important caveat of using this approach to estimate psychiatric symptoms among adolescents is that this is a group experiencing profound developmental changes across biological, psychological and social domains[2]. Thus, there remains some question of the extent to which psychiatric symptoms are stable indicators of some underlying need or indicative of a developing disorder[22], or whether they are indicative of normal (vs. problematic) behavior[23]. Moreover, cross-sectional surveys can only present a snapshot of prevalence estimates; changes over time and the ways in which such symptoms evolve into diagnosable disorders cannot be measured. Several empirical studies based on the Great Smokey Mountains Study[24] have demonstrated patterns of age effects on disorders and disabilities that suggest psychiatric problems drop substantially by age 12, increase between ages 12 to 15, and drop again at age 16[4,22]. Nonetheless, it remains an important task to derive accurate estimates of prevalence of mental health problems in this age group in the national adolescent population.
Several large-scale national surveys of U.S. adults have included structured diagnostic interviews, for example, the Epidemiological Catchment Area Study (ECA)[25] the National Comorbidity Survey and the National Comorbidity Survey-Replication (NCS, NCS-R, age 15–55)[14,15,18]. Although the parallel National Comorbidity Survey – Adolescents (NCS-A,[18]) is a much-needed addition to the field, it has yet to yield estimates of psychiatric disorder among adolescents in the US. In the NCS, Kessler[14] reported that 12-month prevalence of mental disorders was consistently the highest in their youngest cohort, the 15–24 age-group. Gender differences were also reported in the NCS by Kessler[18,26].
One study that has recently incorporated screening items for mental health problems among adolescents in the United States is the NIMH Methods for the Epidemiology of Childhood and Adolescent Mental Disorders Study (MECA)[27]. This study used probability household samples of children aged 9 to 17 and their adult caretakers from four sites. The final sample included 1285 pairs of child respondent and adult caretaker. The screening instrument used in this study was a recently-developed instrument for diagnostic screening of children and adolescents, the Diagnostic Interview Schedule for Children (DISC-2.3)[28]. The DISC-2.3 is a highly structured diagnostic instrument, which screens for six categories of the most common mental disorders among children and adolescents (DSM-III-R)[29]: anxiety, affective, disruptive behavior, mood, substance use, miscellaneous (e.g., eating disorders), and psychotic disorders. This instrument has demonstrated good criterion validity with independent clinical diagnoses[30] and is a reliable tool for the screening of childhood mental disorders[28]. Almost one third of the youth sample (ages 9–17) met DSM diagnostic criteria for any disorder[28]. However, the MECA samples were small and local, and were not representative of the US population as a whole.
Another study that is in the process of developing estimates of child and adolescent (ages 4 – 17) mental health problems in the U.S., is the National Health Interview Survey (NHIS), an annual household survey with a nationally representative sample. In the 2001 and 2003 Supplements to the NHIS, the parent report version of the Strengths and Difficulties Questionnaire (SDQ) was included, and in the 2002 NHIS, a subset of items from this measure was included. Parent reports for approximately 9000 to 10000 children aged 4 – 17 were collected on five emotional and behavioral disorders[31]. The SDQ used in the NHIS includes five scales of five items each, assessing emotional symptoms, conduct problems, hyperactivity-inattention, peer relationship problems, and prosocial behavior, as well as items that tap impairment and burden[32]. The authors report estimates that approximately 5% of the non-institutionalized U.S. population aged 4 – 17 experienced severe emotional difficulties, with significant gender and age differences in the estimates of problems. Notably, however, these estimates are derived from parent reports, which for older adolescents especially, may underestimate the extent of these problems. It remains an important task, therefore, to derive self-reported estimates of mental health problems across the full range of adolescence.
Lucas and colleagues[33] refined the use of DISC diagnostic scales for mental disorders previously employed in the MECA study[27] by devising brief screening scales that can be used to identify those who are likely to meet diagnostic criteria in full assessment. Based upon secondary analysis of the MECA data, Lucas and colleagues applied logistic regression models with stem items from the DISC (i.e., questions asked of all respondents) as the independent variables and DSM-III-R diagnosis as the dependent variable. For each specific diagnosis, the DISC predictive scale was comprised of all the items that emerged as significant predictors of the diagnosis in the regression[33]. Two cut-off scores were derived, one pre-defined by a positive response to any gate item (i.e., those that suggest the need for further probing in a given area), the other constructed to maximize the positive prediction of the full diagnosis based on the sum of sensitivity and specificity. The final instrument contains 76 items, much reduced from the 206 items in the full DISC scales[33].
For the first time in the 2000 NHSDA[34], the DISC Predictive Scale (DPS) items were included and asked of more than 25000 adolescent respondents. This paper reports the prevalence of symptom clusters of psychiatric problems among adolescents in the U.S. population derived from positive screens using the DPS. Although such positive screens are not equivalent to diagnostic case ascertainment – because the sensitivity and specificity data suggest that positive screens for symptoms may be overestimated for clinically significant disorders – these data are important indicators of mental health problems among age, gender and racial/ethnic subgroups that have not been well-documented elsewhere, and such estimates can extend our understanding of the extent of need and inform prevention efforts[19].
Method
The Data
Prevalence and co-occurrence estimates of symptom clusters of specific psychiatric problems were derived from the 2000 National Household Survey on Drug Abuse (NHSDA)[34]. Specifically, we used the adolescent (12–17 years) sample from public use file (N = 19,430), which is a random sample of the total adolescent sample (N = 25,717)*. In the 2000 survey, a complete module of 22 items that approximates DSM-IV[36] criteria for dependence and abuse on major substances such as alcohol, nicotine, marijuana, cocaine and other drugs was included, making it possible to estimate the prevalence of clinically defined dependence for each substance. Also included was a comprehensive set of 71 mental health questions adapted from the DPS, which provided a unique opportunity to estimate the prevalence and co-occurrence of symptom clusters of major psychiatric problems among this critical age group in a large population sample.
The NHSDA sample is drawn using a multi-stage stratification procedure from the civilian, non-institutionalized U.S. population aged 12 or older. Computer-assisted face-to-face interviews were performed in the selected households. Of the 19,340 adolescent respondents in the public data, approximately half were female (49.3%). The racial/ethnic composition was 67% White, 14% Black, 14% Hispanic, and 6% other race or ethnicity. Table 1 presents the demographic characteristics of the 2000 NHSDA adolescent sample and comparison data for the US population aged 12–17 from the 2000 census[37]. Our sample deviates somewhat from the comparison population only in the distribution of adolescents in small- and non-metropolitan areas (probably due to differing Census and NHSDA sample frames).
Table 1 Characteristics of Adolescents Aged 12 – 17 in NHSDA and U.S. Population (Census 2000)
U.S. Population NHSDA Unweighted NHSDA Weighted
Age 12 – 17 12 – 17 12 – 17
N 24,179,360 19,430 23,367,782
Sex
Male (%) 51.4 50.7 51.2
Female (%) 48.6 49.3 48.8
Race/Ethnicity
White (%) 63.4 66.7 65.5
Black (%) 14.4 13.5 14.3
Hispanic (%) 15.2 13.9 14.2
Other (%) 7.0 5.9 6.0
Region
Northeast (%) 18.0 19.3 17.9
Midwest (%) 23.5 25.6 23.4
South (%) 35.4 32.2 35.6
West (%) 23.1 22.9 23.0
Population Density
MSA (pop≥1 m) 43.5 39.0 43.6
MSA (pop < 1 m) 35.7 34.6 33.4
Not in MSA 20.8 26.4 23.1
The Key Variables
DISC predictive scales (DPS) were used to calculate 14 psychiatric symptom clusters. By symptom cluster we refer to a group of symptoms derived from DSM-III-R (DISC 3.2) disorders without the criteria on impairment or duration of the symptom. To derive the score for each symptom cluster, we summed across the DPS items for each scale that were answered positively by the respondent (see Appendix for a listing of the items used for each scale Additional file: 1). The cutoff point for each symptom cluster is based on previous methodological studies of MECA[33] (see Table 2 for details). Three sub-scales in the NHSDA – elimination, panic and mania – were not included in the previous methodological study. Due to a lack of data on sensitivity and specificity thresholds for these scales, we used the maximum score (positive on all items) to define the symptom cluster for each to minimize over-estimation. Thus, a positive case of symptom cluster was derived only when a respondent met the predetermined cut-off for symptom items in the DISC predictive scales. The DPS symptom clusters measured in this way include seven anxiety problems, two affective problems; three disruptive-behavior problems, and two miscellaneous problems (eating & elimination problems). Four indexes of substance use disorders were derived using the modules for approximate diagnosis included in the NHSDA.
Table 2 Comparison of Item Numbers and Cutoffs for Each Subscale of DISC Predictive Scales in MECA and NHSDA
Total SiPh SoPh Agor OAD OCD SAD Eat MDD Adhd ODD CD Elim Man Panic
Study based on MECAa Not in previous report b
# items in full DISC 206 14 8 4 17 13 18 7 27 44 12 24
# items in DPS 77c 7 5 3 7 7 8 3 9 13 7 9
Optimal DPS cutoff ≥ 2 ≥ 1 ≥ 1 ≥ 4 ≥ 1 ≥ 3 ≥ 2 ≥ 4 ≥ 4 ≥ 4 ≥ 2
Sensitivity 0.77 0.89 0.37 0.85 0.91 0.89 0.90 0.98 1.00 0.96 1.00
Specificity 0.79 0.74 0.96 0.92 0.72 0.85 0.88 0.90 0.85 0.94 0.98
NHSDA 2000
# items in NHSDA 71 7 2 4 4 5 7 4 7 6 7 8 3 5 2
Proposed cutoff ≥ 3 ≥ 2 ≥ 2 ≥ 3 ≥ 3 ≥ 4 ≥ 2 ≥ 5 ≥ 4 ≥ 4 ≥ 2 = 3 = 5 = 2
Note: DISC = Diagnostic Interview Schedule for Children (v2.3); DPS = DISC Predictive Scales; MECA = Methods for the Epidemiology of Child and Adolescent Mental Disorders; SiPh= simple phobia, a.k.a. specific anxiety in DSM-IV; SoPh = social phobia; Agor = agoraphobia; OAD = overall anxious disorder, or general anxiety disorder in DSM-IV; OCD = obsessive-compulsive disorders; SAD = separation anxiety disorder; Eat = eating disorders; MDD = major depressive disorders; ADHD = attention-deficit/hyperactivity disorder; ODD = oppositional defiant disorder; CD = conduct disorders; Elim = elimination disorders; Man = Mania; Panic = Panic disorders.
a. From Lucas et al. 2001. Adjustment was made based on further data analysis and pilot studies of DPS by Lucas.
b. Three sub-scales in the 2000 NHSDA were not reported in previous methodological study. Due to lack of data on sensitivity and specificity for these scales, we used the maximum score (positive on all items) as the cutoff for the predictive diagnosis.
c. Excluding one question on substance use disorder, which was replaced by 19 items in NHSDA.
We further aggregated the DPS symptom clusters into indexes reflecting the presence of any cluster within each category (any anxiety, any affective, any behavior, and any SUD). Lastly, we created two variables, any DPS cluster and ≥4 DPS clusters, to tap the prevalence of having any DPS cluster and the severe co-occurrence of multiple DPS clusters, respectively.
As described earlier, items included in the DPS were derived from the full set of DISC items by logistic regression models to predict DSM diagnoses. The number of gate items for each diagnosis-specific scale ranged from 3 to 7. Further, the cut-off points were established by "any gate item answered positively" and by the criterion that the "sum of sensitivity and specificity was maximized for the positive prediction of each specific diagnosis"[33]. Sensitivity ranged from 0.37 to 1.00, with most above 0.89; specificity ranged from 0.72 to 0.98, with most above 0.90. The subset of gate items on the final DPS scales has been demonstrated to identify with 100% accuracy those respondents who don't have a diagnosis, and further contingent items can be omitted without threat of missing a positive case[33]. Therefore, the DPS provides us with a useful and relatively reliable tool to identify those in the U. S. adolescent population at elevated risk for mental disorders.
However, it is important to keep in mind that sensitivity and specificity information may have different implications for different base rates since they are more or less inherent measurement properties of an instrument, but the predictive value also depends to a great extent on the actual probability or base rate of the measured symptom. In other words, the information may be more relevant or sensitive to the symptom with high prevalence in the population, but less relevant or informative for the symptom with very low prevalence in the population.
Analytic Strategy
First, we estimated the prevalence for each of the 19 DPS clusters (including the 5 SUD clusters) for the total sample and for each subgroup, and then compared them within demographic categories. To avoid type-I errors in the estimates, we only report differences significant at the p < 0.01 level. As the NHSDA used a multiple-level sampling procedure, the standard errors for all prevalence rates were estimated by SUDAAN[38], a software package that uses Taylor Series linearization techniques to adjust for sample design effects. Next, we examined the rates of co-occurrence of clusters among demographic groups in two ways: the proportion of the total sample reporting two or more past-year DPS clusters, and the proportion of those with at least one DPS cluster. Last, we explored and identified significant demographic correlates of each cluster through application of multivariate logistic regression models.
Results
Prevalence Estimates
Twelve-month prevalence of symptom clusters of psychiatric problem was estimated using the DPS. As shown in Table 3, three out of five (58.1%) adolescents aged 12–17 years screened positive for at least one DPS cluster over the 12 months preceding the survey.
Table 3 Twelve-Month Prevalence of Psychiatric Symptom Clusters based on DISC Predictive Scale a Among U.S. Adolescents by Gender, Race/Ethnicity and Age (NHSDA 2000)
Total 12–17 year old By Sex By Race/Ethnicity By Age
Male Female White Black Hispanic 12–14 15–17
% SE b % SE % SE % SE % SE % SE % SE % SE
Anxiety Clusters
Social phobia 16.5 0.32 14.7 0.45 18.4* 0.45 15.1 0.38 22.2 0.95 17.0* 0.94 18.1 0.48 14.8* 0.39
Separation anxiety 7.3 0.22 5.5 0.28 9.1* 0.33 6.1 0.25 11.2 0.79 9.1* 0.59 9.9 0.34 4.6* 0.26
Agoraphobia 9.3 0.24 5.9 0.28 12.9* 0.39 7.5 0.25 13.7 0.75 13.0* 0.84 11.4 0.39 7.2* 0.31
Panic disorder 6.0 0.19 3.8 0.23 8.3* 0.30 5.9 0.22 7.2 0.58 5.9 0.48 6.2 0.28 5.8 0.27
General anxiety 15.0 0.29 11.4 0.38 18.9* 0.43 15.6 0.36 15.6 0.83 13.2* 0.69 14.4 0.39 15.7 0.44
Specific phobia 13.5 0.28 8.4 0.32 18.9* 0.45 11.7 0.31 20.0 0.88 14.1* 0.78 15.5 0.40 11.4* 0.38
OCD 14.4 0.30 11.8 0.40 17.1* 0.47 11.4 0.31 22.8 0.97 17.7* 0.89 15.2 0.42 13.5* 0.43
Any anxiety cluster 40.1 0.41 33.1 0.56 47.4* 0.60 36.8 0.48 50.8 1.13 42.9* 1.17 42.6 0.58 37.5* 0.60
Affective Clusters
Major depression 12.1 0.26 7.9 0.31 16.5* 0.42 12.2 0.31 10.6 0.69 12.0 0.68 10.7 0.37 13.5* 0.40
Mania 3.1 0.15 2.7 0.20 3.6* 0.21 3.2 0.19 2.8 0.35 2.7 0.31 3.0 0.21 3.3 0.21
Any affect cluster 13.4 0.28 9.2 0.34 17.9* 0.44 13.6 0.33 11.7 0.73 13.1* 0.70 12.0 0.39 14.9* 0.41
Substance Use Disorders
Alcohol abuse c 3.3 0.15 3.4 0.21 3.3 0.20 3.8 0.21 1.9 0.31 3.0* 0.35 1.2 0.13 5.4* 0.28
Alcohol dependent 1.9 0.12 1.8 0.17 1.9 0.18 2.1 0.16 0.7 0.18 1.9* 0.30 0.6 0.08 3.1* 0.24
Nicotine dependent 4.1 0.16 3.7 0.22 4.6* 0.24 5.1 0.22 1.5 0.24 3.2* 0.39 1.7 0.15 6.7* 0.30
Drug abuse c 2.0 0.13 2.1 0.18 2.0 0.17 2.1 0.15 1.7 0.31 2.6 0.43 1.0 0.12 3.1* 0.22
Drug dependent 2.4 0.13 2.6 0.20 2.2 0.16 2.5 0.17 1.7 0.28 2.3 0.38 0.9 0.11 3.8* 0.24
Any SUD 9.7 0.25 9.5 0.34 9.9 0.35 10.8 0.33 5.4 0.50 9.1* 0.68 4.0 0.22 15.5* 0.45
Disruptive-behavior Clusters
ADHD 14.7 0.29 13.8 0.41 15.6* 0.44 13.8 0.34 18.3 0.91 15.1* 0.82 16.0 0.44 13.3* 0.37
ODD 27.1 0.38 27.9 0.56 26.2 0.53 27.8 0.47 26.4 1.10 24.9 0.94 27.2 0.53 27.0 0.52
Conduct disorder 11.5 0.28 12.8 0.41 10.1* 0.39 10.9 0.32 12.4 0.79 12.2 0.73 9.3 0.34 13.6* 0.41
Any beh. cluster 36.3 0.42 37.1 0.59 35.5 0.60 36.3 0.51 38.3 1.15 34.7 1.09 36.1 0.59 36.6 0.55
Other Disorder Clusters
Eating disorder 6.1 0.19 3.7 0.23 8.7* 0.32 5.9 0.24 6.5 0.57 6.4 0.50 5.5 0.28 6.8* 0.30
Elimination 0.2 0.05 0.3 0.08 0.2 0.05 0.3 0.06 0.2 0.09 0.2 0.06 0.2 0.06 0.3 0.07
Any DPS Cluster 58.1 0.42 54.8 0.56 61.6* 0.59 56.7 0.51 62.8 1.15 58.8* 1.17 56.6 0.59 59.6* 0.58
* p < 0.01 in Chi-square test of gender, race or age differences. a DISC = Diagnostic Interview Schedule for Children, See appendix Table A; b All standard errors are estimated by SUDAAN, which takes multi-level sampling effects into consideration. c. Abuse only without dependence.
Gender
Although female adolescents usually report lower rates of substance use, the prevalence of SUDs for females in this national sample is as high as that of males. Females were more likely to be nicotine dependent (although males reported more nicotine use than females). Compared to males, females reported higher rates of anxiety, affective, and eating problems, and lower levels of the elimination problems.
Age
Estimates of having any DPS psychiatric symptom cluster were slightly higher for late vs. early adolescents, 59.6% vs. 56.6%, and resulted mainly from age differences in SUD. Late adolescents (ages 15–17) reported higher rates of SUD and affective clusters; younger adolescents (age 12–14) report higher rates of anxiety clusters than older adolescents (age 15–17); and no age differences emerged in rates of behavioral clusters with the exception of ADHD, which was more prevalent among the younger group.
Race/Ethnicity
Blacks reported more DPS psychiatric symptom clusters than Whites and Hispanics. As would be expected from the literature, Blacks reported lower use of licit and illicit substances[39], and exhibited lower estimates of SUD. Blacks also appear to have relatively high risk for anxiety clusters, compared to other ethnic groups. Rates of OCD clusters among Blacks were twice as high as among Whites, and rates for specific anxiety and agoraphobia clusters also approached this degree of difference.
Co-Occurrence
Fifty-eight percent of the adolescent sample met risk-identification criteria for at least one psychiatric symptom cluster in the 12 months preceding the survey. Table 4 shows that 37.7% of the sample met criteria for two or more co-occurring symptom clusters, and 17% reported 4 or more clusters. Of those with at least one identified cluster, almost two-thirds screened positive for an additional one or more clusters: 20.9% reported two, 14.6% reported three, and 29.4% reported four or more (i.e., severe comorbidity). Thus, co-occurrence of mental health problems in this group appears quite high and deserving of more attention. In general, females had higher rates of severe comorbidity than males, with about one third of females and one quarter of males having at least one cluster screening positive for four or more. Black adolescents reported higher rates of comorbid clusters than any other ethnic group; and younger adolescents had a slightly higher rate of comorbidity than their older counterparts.
Table 4 Co-occurrence of 12-Month Psychiatric Symptom Clusters (including SUD) among Adolescents with any DPS Symptom Cluster, by Gender, Age and Race/Ethnicity
No. of 12-mth clusters 0 1 2 3 4+
Among total sample (N = 19,430)
Total (%) 41.9 20.4 12.2 8.5 17.1
Male 45.2 21.5 12.4 7.9 13.0
Female 38.4 19.2 11.9 9.1 21.4
White 43.3 20.8 11.9 8.1 15.9
Black 37.2 18.7 13.5 9.1 21.4
Hispanic 41.2 20.0 12.1 8.5 18.2
Others 39.9 20.3 11.6 10.4 17.7
Age 12–14 43.4 19.3 12.0 8.2 17.1
Age 15–17 40.4 21.5 12.3 8.8 17.1
Among those with any DPS cluster (N = 11,228)
Total (%) -- 35.1 20.9 14.6 29.4
Male -- 39.2 22.6 14.4 23.8
Female -- 31.2 19.4 14.7 34.7
White -- 36.7 21.0 14.4 27.9
Black -- 29.8 21.6 14.5 34.1
Hispanic -- 34.0 20.5 14.5 31.0
Others -- 33.8 19.3 17.3 29.5
Age 12–14 -- 34.1 21.2 14.5 30.2
Age 15–17 -- 36.1 20.6 14.7 28.6
Demographic Correlates of Psychiatric Symptom Clusters
To further understand the relationship of gender, ethnicity, age and other demographic factors on the estimated prevalence of 12-month psychiatric symptom clusters and co-occurrence, we conducted a series of multiple logistic regression analyses, estimated by SUDAAN, to explore the significant demographic correlates for each domain of clusters, for any DPS cluster, and for severe comorbidity (4 or more clusters). Table 5 presents the adjusted odds ratios for each demographic factor in each of these LOGIT models.
Table 5 Odds Ratios of Demographic Correlates of 12-month Psychiatric Symptom Clusters Based on Logit Models
Correlates Any anxiety cluster Any affective cluster Any SUD Any behavior cluster Any DPS symptom cluster 4+ clusters
OR 95% CI OR 95% CI OR 95% CI OR 95% CI OR 95% CI OR 95% CI
Sex
Male 0.55* 0.51–0.59 0.47* 0.42–0.52 0.98 0.88–1.10 1.08* 1.00–1.16 0.76* 0.71–0.81 0.56* 0.51–0.61
Female 1.00 -- 1.00 -- 1.00 -- 1.00 -- 1.00 -- 1.00 --
Age
12–13 1.00 -- 1.00 -- 1.00 -- 1.00 -- 1.00 -- 1.00 --
14–15 0.85* 0.79–0.93 1.20* 1.07–1.35 3.96* 3.23–4.85 1.17* 1.08–1.26 1.16* 1.05–1.23 0.96 0.87–1.06
16–17 0.75* 0.68–0.81 1.34* 1.19–1.51 8.01* 6.54–9.83 1.03 0.95–1.11 1.12* 1.04–1.22 0.95 0.85–1.06
Race/ethnicity
White 1.00 -- 1.00 -- 1.00 -- 1.00 -- 1.00 -- 1.00 --
Black 1.67* 1.50–1.85 0.86 0.73–1.01 0.43* 0.35–0.54 1.08 0.97–1.21 1.24* 1.10–1.38 1.32* 1.13–1.53
Hispanic 1.20* 1.07–1.35 0.97 0.84–1.12 0.79* 0.64–0.98 0.97 0.87–1.09 1.06 0.95–1.18 1.11 0.97–1.28
Others 1.26* 1.06–1.49 1.24 0.98–1.56 0.79 0.57–1.09 1.09 0.91–1.30 1.14 0.96–1.35 1.13 0.89–1.44
Family income ($)
0–19,999 1.32* 1.17–1.49 0.99 0.84–1.17 1.49* 1.22–1.82 1.04 0.93–1.17 1.19* 1.06–1.34 1.52* 1.31–1.78
20,000–39,999 1.33* 1.20–1.48 0.97 0.85–1.12 1.52* 1.28–1.81 1.11* 1.00–1.24 1.24* 1.12–1.38 1.48* 1.28–1.71
40,000–74,999 1.04 0.94–1.15 0.96 0.83–1.10 1.21* 1.03–1.42 1.04 0.94–1.14 1.07 0.97–1.18 1.15* 1.00–1.31
75,000 + 1.00 -- 1.00 -- 1.00 -- 1.00 -- 1.00 -- 1.00 --
Population density
MSA 1 million+ 0.90* 0.82–0.99 0.87* 0.76–0.99 0.84* 0.72–0.99 1.05 0.95–1.15 0.96 0.87–1.06 0.88* 0.78–1.00
MSA <1 million 0.98 0.90–1.08 0.95 0.84–1.07 0.92 0.79–1.06 1.05 0.96–1.15 1.02 0.93–1.11 0.99 0.87–1.12
Non-MSA 1.00 -- 1.00 -- 1.00 -- 1.00 -- 1.00 -- 1.00 --
U.S. Born 0.91 0.78–1.06 0.91 0.75–1.11 1.65* 1.21–2.25 1.41* 1.19–1.67 1.09 0.94–1.26 1.17 0.96–1.43
School dropout 0.86 0.67–1.10 0.95 0.67–1.34 2.20* 1.63–2.97 1.01 0.79–1.30 1.03 0.81–1.31 1.22 0.88–1.69
Models with sex-by-age interaction ¶
Sex *Age interaction, (vs. female, age 12–13)
Male, 14–15 0.73* 0.62–0.86 0.61* 0.48–0.79 0.75 0.49–1.15 0.86 0.73–1.02 0.76* 0.65–0.89 0.76* 0.61–0.95
Male, 16–17 0.67* 0.56–0.80 0.62* 0.49–0.80 0.89 0.60–1.34 0.89 0.75–1.05 0.77* 0.65–0.91 0.75* 0.59–0.94
* p < .05 (two tailed test) – Models estimated by SUDAAN with correction of design effects;
¶ Shows the interactive effects only with presenting other correlates.
Gender
In support of the prevalence estimates reported above, males had a lower likelihood of reporting any DPS cluster. Specifically, they were less likely to report any anxiety or affective cluster, or severe co-occurrence. However, males were more likely to have a significantly elevated chance of attention deficient and behavior problems.
Age
We analyzed the age effect on symptom clusters in three groups, early adolescent (ages 12–13), middle-adolescent (ages 14–15), and late adolescent (ages 16–17). Whereas the middle adolescents were more likely to report either any DPS symptom cluster or disruptive-behavior cluster, there were considerable variations in the odds ratios of reporting clusters across age groups. For example, the younger adolescents were more likely to meet criteria for an anxiety cluster than the older adolescents, whereas the older adolescents were more likely to do so for an affective cluster. Middle-adolescents had the highest risk for attention deficient and behavioral problems, suggesting the transitory and developmental nature of these problems. Confirming the wealth of data in the literature, the odds of SUD increased with age: middle-adolescents were almost four times as likely, and late adolescents eight times as likely as early adolescents to report any SUD.
Gender-Age interactions
A number of significant gender-age interactions appeared in the models predicting any DPS symptom cluster, anxiety clusters, affective clusters, and severe co-occurrence (see models in the lower portion of Table 5). Generally, compared to the youngest females (12–13), middle and late adolescent males (14–17) had lower odds for each of these problems. Taken together with the findings of prevalence reported earlier, it seems that the youngest females are at the highest risk for these clusters and for the more severe levels of comorbidity problems.
Race/Ethnicity
Black adolescents are at greater risk than White not only for any DPS symptom clusters but also for severe co-occurrence, and they are at higher risk for anxiety problems as well. Indeed, all minority groups in this sample were more likely to report anxiety problems than Whites. However, Blacks alone are at increased risk for co-occurrence once they have met criteria for any one symptom cluster.
Family Income
Adolescents from families with lower incomes are more likely to meet criteria for at least one psychiatric symptom cluster and to be at risk for severe comorbidity compared to their more affluent peers. This relationship held for anxiety cluster and SUD, although for SUD there was increased risk for all adolescents save those from the wealthiest families (income ≥ $75 k).
Other Demographics
Interestingly, compared to those from non-metropolitan areas, adolescents from the largest metropolitan areas were at reduced risk for most psychiatric problems, including anxiety and affective domains, SUD and co-occurring problems. US-born adolescents had higher risk for SUD and disruptive behavior problems in comparison with immigrant adolescents. School dropouts were at higher risk for SUD than those currently enrolled in school.
Discussion
Although the data from NHSDA are not sufficient to reach any clinical diagnosis, these unique data offer us important descriptive information on the epidemiology of psychiatric symptoms and their variations in gender and ethnic groups among the U.S. adolescent population. Even a conservative interpretation of these estimates from a nationally representative sample of the U.S. adolescents suggests that adolescent mental health problems and related co-occurrence may be more serious than previously believed. Almost three out of five US adolescents aged 12–17 screened positive for a symptom cluster of specific psychiatric problem using the DPS scales in the 12 months prior to the interview. More than one third reported a disruptive-behavior problem and a slightly larger proportion reported an anxiety problem. Approximately one out of every eight adolescents reported an affective problem, and one in ten had a substance use disorder. Although the high prevalence may be inflated by behavioral symptoms that are likely to be significantly reduced with maturation, these estimates should raise serious concerns about the mental health status of U.S. adolescents. Although these symptom clusters do not fully emulate the diagnostic criteria of clinical disorders, the previous study of specificity and sensitivity of DPS suggested that these symptom clusters have good predictive validity for a disorder ascertained using the DISC[33]. Parenthetically recent epidemiological reports of 12-month DSM-IV rates of psychiatric disorders from the World Mental Health 2000 initiative among adults in European countries have found they are far lower than the U.S. national rates measured with the same DSM-IV diagnostic protocol[40]. In the Oregon Adolescent Depression Project, the concept of sub-threshold psychiatric conditions was introduced to monitor the potential mental health problems among adolescents[41], which is very similar to the symptom cluster of psychiatric problem proposed in this study. They reported that, of the 1704 adolescents in the study, 52.5% had at least one sub-threshold disorder; of those 40% had also experienced a comorbid sub-threshold condition, and 30% had a second comorbid condition[41]. Based on this updated information, our estimates of high prevalence of psychiatric symptom clusters among U.S. adolescents based on the largest national household survey are not farfetched. The key is how we interpret these data and how we use the information. For example, it is possible to view these as estimates of adolescents "at-risk" rather than as ascertained cases since the items used for the estimates were derived from psychiatric symptoms used in the actual DISC diagnostic modules. Even in instances where the number or pattern of symptoms are sufficient for a "screened positive" but not sufficient to constitute a recognized psychological disorder, the evidence suggests that many of these children do indeed have mental health problems [42], and are in need of further, expert assessment. We acknowledge, however, that surveys such as the NHSDA rely upon the anonymity of their respondents to ensure accurate collection of sensitive data; for this reason, such referral is impossible.
Our results show that girls tend to have a higher rate of mental health problems than boys, which is consistent with the literature [15-17] and with estimates for the adult population as reported in the NCS study[14]. However, contradicting findings from community or treatment samples, prevalence of disruptive-behavior problems was similar among boys and girls; additionally, in the present study, there were no gender differences for SUD, in contrast to the findings from the NCS for the US population aged 15–24, where males had a significantly higher rate of SUD than females. Our data also contradict previous findings in this area: although numerous national surveys consistently report lower rates of substance use among females[17,39], we found that females were equally likely to be dependent upon or abuse substances, suggesting the possibility that once they begin use, females are more at risk for developing an SUD.
Another contradictory finding emerged concerning our estimates of disruptive behavior. In treatment samples, males reported higher rates of disruptive-behavior problems; in our study, however, estimates of behavior problems suggest similar rates for male and female adolescents in the general population.
As to the racial/ethnic differences, Blacks reported a higher rate of psychiatric problems than other groups, including anxiety problems. These findings are in marked contrast to those reported from the NCS, where no racial difference was found in the prevalence of anxiety disorders, including simple phobia and agoraphobia[14]. However, the data for adolescents presented here are similar to those reported from the ECA study[43].
It is interesting to notice the reduced risk of mental health problems in large metropolitan areas, since some assume that large metropolitan areas may have more stressors and mental health problems, and are the areas with greater need for mental health services. The finding here raises the prospect that mental health service delivery to the rural areas (e.g. mobile clinics, tele-support network, etc.) should become a priority in health service planning.
Evidence emerging from epidemiological studies suggests high levels of comorbid mental health problems among children and adolescents [44-46]. The proportion of adolescents in the present sample who meet criteria for more than one psychiatric symptom cluster is astonishingly high – about two-thirds of those with one symptom cluster were at risk for other symptom clusters, and half of these met criteria for four or more symptom clusters (see Table 4). Such high estimates of co-occurring mental health problems are surely deserving of attention and raise important issues of providing appropriate treatment and preventive interventions. We identified a number of risk factors for severe psychiatric comorbidity including Black race, lower family income, and being a younger female (12–13 years old). Previous studies of the relationship between psychiatric problems and substance use in adolescents suggests an approximately linear relationship between the intensity of use and the likelihood of having a mental problem, especially conduct disorder [47-49]. Thus this issue remains an important topic for future studies, especially with nationally-representative samples.
Limitations
There are a number of limitations that might affect the findings reported in this paper. First, the NHSDA is a cross-sectional study that relies solely upon retrospective self-reports of symptoms and behaviors; therefore, the reliability of past 12-month recall of adolescent mental health symptoms needs further verification.
Second, we reiterate that the DISC Predictive Scales used to derive prevalence of psychiatric symptom clusters identify positive screens for disorders – they are not clinical diagnoses, and should be considered as indicators of elevated risk for psychiatric disorders. Compared to DSM-IV criteria, these scales do not have the required duration or replication measures (as DPS was developed with DISC criteria). Even with this limitation, however, these data speak to important patterns of age, gender and ethnic differences in the prevalence of mental health problems deserving of further study.
Third, the DPS does not have the impairment or severity measures of each symptom that are frequently used in clinical diagnoses. Estimates based on symptoms alone can vary drastically from those derived using impairment criteria[19]. Using these additional criteria increases the likelihood that the ascertainment in a field interview would more closely resemble a clinically ascertained case in terms of severity. However, collecting this level of information about each respondent was and is not feasible for the type of survey used in our study. It should be noted that using additional impairment criteria does not necessarily affect caseness estimates, and the criteria we have used for case (of at-risk) identification have shown good sensitivity and specificity when compared to the use of a real diagnostic interview (e.g., DISC). Moreover, children who meet criteria for psychiatric disability without meeting full DSM-IV criteria for diagnosis can be considered as having significant psychiatric problems[42].
Finally, despite the use of anonymous and computer-assisted interview (CAI) techniques, household surveys tended to yield lower rates of substance use compared to school-based surveys such as Monitoring the Future[39] and Youth Risk Behavior Surveillance Survey[50]. Youth matched for high-school grade and age in the NHSDA reported lower rates of substance use than those from other school-based surveys[51]. Therefore, although the estimated rates of symptoms of mental health problems in our sample of adolescents are very high, the possibility of under-reporting in NHSDA is still of an unknown degree and needs further studies to verify its extent.
Clinical Implications
Although the United States has a well-established system for monitoring the pattern of drug use in the general adolescent population, through both school-based surveys[39] and the household survey (NHSDA), there is no system existing for estimating rates of substance use disorders and mental disorders in the population on an ongoing basis and among various socio-demographic subgroups. The current estimates of the prevalence and co-occurrence of adolescent psychiatric problems, especially the dramatic gender and ethnic variations in different psychiatric problem domains, may provide clinicians with both a heightened alert and a useful tool with which to identify potential mental health problems in the general adolescent population, which traditionally has been biased by extensive reliance on results from treatment studies.
Black and poor adolescents reported mental health problems more frequently than did those from other groups; unfortunately, both Black and poorer adolescents are less likely to receive diagnosis or treatment for mental health problems[21,52]. Clearly, the present findings strongly suggest a need not only for increased efforts to identify and treat adolescents with psychiatric problems but also to design and implement effective preventative and treatment strategies for the most needy subgroups.
Integrating the DISC predictive scales in the NHSDA (now NHSDUH) would provide an excellent mechanism for including an expanded set of questions designed to systematically monitor and assess substance use disorders and psychiatric problems. Perhaps by using alternating numbers of diagnostic modules per survey iteration, NHSDUH can regularly include the mental health module as it did for SUDs. Such regular inclusion of these modules will allow for an examination of changes over time, an important goal for estimating trends in the prevalence of mental health problems. We recommend that the Substance Abuse and Mental Health Service Administration study and institute the expansion of DPS in future waves so that researchers and policy makers can continuously monitor the trends and patterns in adolescent mental health problems that will provide essential information for the development of more effective prevention and intervention programs.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
• Dr. Kevin Chen initiated the paper plan, performed data analysis and initial literature review, and wrote up the draft of the paper.
• Dr. Ley Killeya-Jones participated in planning of the paper, performed most literature search and review, wrote-up the introduction and discussion, and polished the entire paper.
• Dr. William Vega participated in planning of the paper, defining the concept of symptom cluster of psychiatric problem instead of traditional "disorder", provided key literature and conceptual support, and editing and polishing the entire paper.
Note
* Due to concerns with confidentiality, we were unable to have released to us the entire NHSDA sample by the Substance Abuse and Mental Health Service Administration (SAMHSA).[35]
Supplementary Material
Additional File 1
"appendix_items.doc"
Click here for file
Acknowledgements
We would like to thank Dr. Christopher Lucas of Columbia University College of Physicians and Surgeons for his sharing of additional pilot data on DISC predictive scale to help make more reasonable decisions on cutoff points for each symptom cluster.
==== Refs
Demyttenaere K Bruffaerts R Posada-Villa J Gasquet I Kovess V Lepine JP WHO World Mental Health Survey Consortium. Prevalence, severity, and unmet need for treatment of mental disorders in the World Health Organization World Mental Health Surveys JAMA 291 21 2581 90 2004 Jun 2 15173149
Williams PG Holmbeck GN Grayson N Greenley RN Adolescent health psychology J Clinical and Consulting Psychol 2002 70 3 828 42 10.1037/0022-006X.70.3.828
Roberts RE Attkisson CC Rosenblatt A Prevalence of psychopathology among children and adolescents American Journal of Psychiatry 1998 155 715 725 9619142
Costello EJ Mustillo S Erkanli A Keeler G Angold A Prevalence and development of psychiatric disorders in childhood and adolescence Arch Gen Psychiatry 2003 60 837 844 10.1001/archpsyc.60.8.837 12912767
Turner RJ Gil AG Psychiatric and substance use disorders in South Florida: Racial/ethnic and gender contrasts in a young adult cohort Arch Gen Psychiatry 2002 59 1 43 50 10.1001/archpsyc.59.1.43 11779281
Kessler RC Berglund P Demler O Jin R Merikangas KR Walters EE Lifetime prevalence and age of onset distributions of DSM-IV disorders in the National Comorbidity Replication Arch Gen Psychiatry 2005 62 593 602 10.1001/archpsyc.62.6.593 15939837
Kim-Cohen J Caspi A Moffit TE Harrington H Milne BJ Poulton R Prior juvenile diagnoses in adults with mental disorder: Developmental follow-back of a prospective longitudinal cohort Arch Gen Psychiatry 2003 60 7 709 717 10.1001/archpsyc.60.7.709 12860775
Simon GE Von Korff M Recall of psychiatric history in cross-sectional surveys: implications for epidemiological research Epidemiol Rev 1995 17 221 227 8521941
Giaconia RM Reinherz HZ Paradis AD Hauf AM Stashwick CK Major depression and drug disorders in adolescence: general and specific impairments in early adulthood J Am Acad Child Adolesc Psychiatry 2001 40 12 1426 1433 10.1097/00004583-200112000-00013 11765288
Kessler RC Avenevoli S Merikangas KR Mood disorders in children and adolescents: An epidemiologic perspective Biol Psychiatry 2001 49 12 1002 14 10.1016/S0006-3223(01)01129-5 11430842
Lewinsohn PM Rohde P Seeley JR Klein Gotlib IH Psychosocial functioning of young adults who have experienced and recovered from Major Depressive Disorder during adolescence J Abnormal Psychol 2003 112 3 353 363 10.1037/0021-843X.112.3.353
Lewinsohn PM Seeley JR Hibbard J Rohde P Sack WH Cross-sectional and prospective relationships between physical morbidity and depression in older adolescents J Am Acad Child Adolesc Psychiatry 1996 35 1120 29 10.1097/00004583-199609000-00009 8824055
Woodward LJ Fergusson DM Life course outcomes of young people with anxiety disorders in adolescence J Am Acad Child Adolesc Psychiatry 2001 40 1086 93 10.1097/00004583-200109000-00018 11556633
Kessler RC McGonagle KA Zhao S Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States: results from the National Comorbidity Survey Arch Gen Psychiatry 1994 51 8 19 8279933
Kessler RC Nelson CB McGonagle KA Edlund MJ Frank RG Leaf PJ The epidemiology of co-occurring addictive and mental disorders: implications for prevention and service utilization American Journal of Orthopsychiatry 1996 66 17 31 8720638
Robins LN Regier DA Editors Psychiatric disorders in America The Epidemiological Catchment Area Study 1991 New York: The Free Press
Narrow WE Rae DS Robins LN Regier DA Revised prevalence estimates of mental disorders in the United States: using a clinical significance criterion to reconcile 2 surveys' estimates Arch Gen Psychiatry 2002 59 2 115 23 10.1001/archpsyc.59.2.115 11825131
Kessler RC Chiu WT Demler O Walters EE Prevalence, severity, and comorbidity of 12-month DSM-IV disorders in the National Comorbidity Survey Replication Arch Gen Psychiatry 2005 62 617 627 10.1001/archpsyc.62.6.617 15939839
Costello EJ Egger H Angold A 10-year research update review: The epidemiology of child and adolescent psychiatric disorders: I. Methods and public health burden J Am Acad Child Adolesc Psychiatry 2005 44 10 972 986 10.1097/01.chi.0000172552.41596.6f 16175102
Wang PS Lane M Olfson M Pincus HA Wells KB Kessler RC Twelve-month use of mental health services in the United States Arch Gen Psychiatry 2005 62 629 640 10.1001/archpsyc.62.6.629 15939840
Elster A Jarosik J VanGeest J Fleming M Racial and ethnic disparities in health café fro adolescents: a systematic review of the literature Arch Pediatr Adolesc Med 2003 157 867 874 12963591
Ezpeleta L Erkanli A Costello EJ Angold A Epidemiology of psychiatric disability in childhood and adolescence J Child Psychol Psychiatr 2001 42 7 901 914 10.1111/1469-7610.00786
Cicchetti D Rogosch FA Psychopathology as risk for adolescent substance use disorders: A developmental psychopathology perspective J Clin Child Psychol 1999 28 355 365 10.1207/S15374424jccp280308 10446685
Costello EJ Angold A Burns BJ Stangl DK Tweed DL Erkanli A Worthman CM The Great Smoky Mountains Study of Youth: goals, designs, methods, and the prevalence of DSM-III-R disorders Arch Gen Psychiatry 1996 53 1129 1136 8956679
Robins LN Regier DA Psychiatric Disorders in America 1991 New York: The Free Press
Kessler RC Sex differences in DSM-III-R psychiatric disorders in the United States: Results from the National Comorbidity Survey JAMWA 1998 53 148 58
Lahey B Flagg EW Bird HR The NIMH Methods for the Epidemiology of Child and Adolescent Mental Disorders [MECA] Study: background and methodology J Am Acad Child Adolesc Psychiatry 1996 35 855 864 10.1097/00004583-199607000-00011 8768345
Shaffer D Fisher P Dulcan MK Davies M Piacentini J The NIMH Diagnostic Interview Schedule for Children Version 2 3 [DISC-2 3]: description, acceptability, prevalence rates, and performance in the MECA study J Am Acad Child Adolesc Psychiatry 1996 3 865 77 10.1097/00004583-199607000-00012
American Psychiatry Association Diagnostic and statistical manual of mental disorders [3rd ed. Revised, DSM-III-R] 1987 Washington, DC: American Psychiatric Press
Schwab-Stone ME Shaffer D Dulcan MK Criterion validity of the NIMH Diagnostic Interview Schedule for Children, Version 2 3 [DISC-2 3] J Am Acad Child Adolesc Psychiatry 1996 35 878 888 10.1097/00004583-199607000-00013 8768347
Simpson GA Bloom B Cohen RA Blumberg S U.S. children with emotional and behavioral difficulties: data from the 2001, 2002, and 2003 National Health Interview Surveys Advance Data 306 June 23 2005 10662357
Bourdon KH Goodman R Rae DS Simpson G Koretz DS The Strengths and Difficulties Questionnaire: U. S. normative data and psychometric properties J Am Acad Child Adolesc Psychiatry 2005 44 6 557 564 10.1097/01.chi.0000159157.57075.c8 15908838
Lucas CP Zhang H Fisher PW Shaffer D Regier DA The DISC Predictive Scales [DPS]: Efficiently screening for diagnoses J Am Acad Child Adolesc Psychiatry 2001 40 443 9 10.1097/00004583-200104000-00013 11314570
Substance Abuse and Mental Health Services Administration [SAMHSA] National Household Survey on Drug Abuse: Main Findings 1997 2001 US Department of Health and Human Services, Public Health Service
Substance Abuse and Mental Health Services Administration [SAMHSA] National Household Survey on Drug Abuse: Methodology Reports and Questionnaires: 2000 Methodology Resource Book [On-line] http://oas.samhsa.gov/NHSDA/2kCAI/2000_CAI_Specs.pdfRetrieved September 24, 2005
American Psychiatry Association Diagnostic and statistical manual of mental disorders [4th ed., DSM-IV] 1994 Washington, DC: American Psychiatric Press
United States Bureau of the Census United States Census, 2000, Summary File 4 [SF4] [serial on the Internet] 2003 http://www.census.gov/Press-Release/www/2003/SF4.htmlAccessed 2003, Dec 15
Shah BV Barnwell BG Hunt PN LaVange LM SUDAAN User's Manual, Professional Software for Survey Data Analysis for Multi-stage Sample Designs 1992 North Carolina: Research Triangle Institute
Johnston LD O'Malley PM Bachman JG Monitoring The Future national survey results on drug use, 1975–2001. Secondary school students. [NIH publication No 02-5106] 2002 1 Bethesda[MD]: National Institute on Drug Abues
Demyttenaere K Bruffaerts R Posada-Villa J Gasquet I Kovess V Lepine JP WHO World Mental Health Survey Consortium. Prevalence, severity, and unmet need for treatment of mental disorders in the World Health Organization World Mental Health Surveys JAMA 2004 291 21 2581 90 10.1001/jama.291.21.2581 15173149
Lewinsohn PM Shankman SA Gau JM Klein DN The prevalence and comorbidity of subthreshold psychiatric conditions Psychol Med 2004 34 4 613 622 10.1017/S0033291703001466 15099416
Angold A Costello EJ Farmer EMZ Burns BJ Erkanli A Impaired but undiagnosed J Am Acad Child Adol Psychiatry 1999 38 129 137 10.1097/00004583-199902000-00011
Eaton WW Dryman A Weisman MM Robins LN, Regier DA Panic and phobia Psychiatric disorders in America: The Epidemiological Catchment Area Study 1991 38 New York: The Free Press 155 79
Angold A Costello EJ Erkalani A Comorbidity J Child Psychol Psychiatry 1999 40 57 87 10.1017/S0021963098003448 10102726
Rohde P Lewinsohn PM Seeley JR Comorbidity of unipolar depression: II Comorbidity with other mental disorders in adolescents and adults J Abnorm Psychol 1991 100 214 222 10.1037/0021-843X.100.2.214 2040773
Merikangas KR Mehta RL Molnar BE Walters EE Swendsen JD Comorbidity of substance use disorders with mood and anxiety disorders: results of the international consortium in psychiatric epidemiology Addictive Behaviors 1998 23 893 907 10.1016/S0306-4603(98)00076-8 9801724
Biederman J Wilens T Mick E Faraone SV Weber W Curtis S Is ADHD a risk factor or psychoactive substance use disorders? Findings from a four-year prospective follow up study J Am Acad Child Adolesc Psychiatry 1997 36 1 21 29 10.1097/00004583-199701000-00013 9000777
Brook J Whiteman M Nomura C Gordon AS Personality, family, and ecological influences on adolescents drug use: A developmental analysis Journal of Chemical Dependency Treatment 1998 1 2 123 161 10.1300/J034v01n02_07
Kandel DB Johnson JG Bird HR Psychiatric comorbidity among adolescents with substance use disorders: Findings from the MECA Study J Am Acad Child Adolesc Psychiatry 1999 38 6 693 99 10.1097/00004583-199906000-00015 10361787
Center for Disease Control and Prevention Youth Risk Behavior Surveillance System: Assessing the Health Risk Behaviors of Young People 2001 Atlanta [GA]: Centers for Disease Control and Prevention
Fendrich M Johnson TP Examining prevalence differences in three national surveys of youth: Impact of consent procedures, mode, and editing rules Journal of Drug Issues 2001 31 3 615 42
National Institute on Drug Abuse Drug use among racial/ethnic minorities [revised] [DHHS Pub No NIH 03-3888] 2003 Rockville, MD: National Institute on Drug Abuse
|
16255774
|
PMC1298317
|
CC BY
|
2021-01-04 18:01:02
|
no
|
Clin Pract Epidemiol Ment Health. 2005 Oct 28; 1:22
|
utf-8
|
Clin Pract Epidemiol Ment Health
| 2,005 |
10.1186/1745-0179-1-22
|
oa_comm
|
==== Front
Environ HealthEnvironmental Health1476-069XBioMed Central London 1476-069X-4-251627447510.1186/1476-069X-4-25ResearchIncreasing work-place healthiness with the probiotic Lactobacillus reuteri: A randomised, double-blind placebo-controlled study Tubelius Py [email protected] Vlaicu [email protected] Anders [email protected] Tetra Pak Occupational Health and Safety AB, Ruben Rausings Gata, 221 86 Lund, Sweden2 BioGaia AB, St Lars Väg 42A, 220 09, Lund, Sweden2005 7 11 2005 4 25 25 22 8 2005 7 11 2005 Copyright © 2005 Tubelius et al; licensee BioMed Central Ltd.2005Tubelius 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
Short term illnesses, usually caused by respiratory or gastrointestinal diseases are disruptive to productivity and there is relatively little focus on preventative measures. This study examined the effect of the probiotic Lactobacillus reuteri protectis (ATCC55730) on its ability to improve work-place healthiness by reducing short term sick-leave caused by respiratory or gastrointestinal infections.
Methods
262 employees at TetraPak in Sweden (day-workers and three-shift-workers) that were healthy at study start were randomised in a double-blind fashion to receive either a daily dose of 108 Colony Forming Units of L. reuteri or placebo for 80 days. The study products were administered with a drinking straw. 181 subjects complied with the study protocol, 94 were randomised to receive L. reuteri and 87 received placebo.
Results
In the placebo group 26.4% reported sick-leave for the defined causes during the study as compared with 10.6% in the L. reuteri group (p < 0.01). The frequency of sick-days was 0.9% in the placebo group and 0.4% in the L. reuteri group (p < 0.01). Among the 53 shift-workers, 33% in the placebo group reported sick during the study period as compared with none in the L. reuteri group(p < 0.005).
==== Body
Introduction
The general well-being in work-places is receiving increased attention in Sweden. Not only is the well-being and healthiness important to the individual himself, but also important to fellow co-workers, family members and last but not least to the success of the company. During the last few decades increased focus have been on how to increase well-being by offering company-sponsored health care, memberships in fitness centres and similar programs aiming to prevent and cure disease and to increase physical activity and awareness on health issues.
Comparatively little attention has been on the diet and how a daily healthy diet impacts on general health with the exception of anti-obesity initiatives.
The cost due to short term sick-leave is in Sweden alone estimated to more than 2.2 billion € [1] and a majority (50–60%) of the episodes are caused by diseases in the respiratory tract (common cold) and gastrointestinal infections [2].
Of special interest is the well-being among shift-workers as this group is known to be at significantly higher risk to attract short-term illnesses such as the common cold and gastroenteritis [3].
Probiotics, i.e. naturally occurring bacteria with health benefits are gaining wider acceptance. These bacteria, commonly from the Lactobacillus family have been demonstrated to have numerous potentially important benefits in terms of gut health and immunity [4], but very few studies address how these effects translate into health benefits in normal populations.
Recently, it was demonstrated in a double blind study [5] that gastrointestinal illnesses and febrile episodes could be significantly reduced in babies attending day-care centers. This was achieved by adding the probiotic Lactobacillus reuteri to infant formula during a 12 week long study period.
This study aimed to evaluate if similar effects could be achieved in adults and if the addition of the probiotic L. reuteri as a daily dietary supplement can improve work-place healthiness.
Materials and methods
Healthy volunteers were recruited among employees at Tetra Pak in Sweden. The criteria for participation were that they should be symptom-free at Day 0, between 18–65 years of age, willing to comply with the protocol and that signed informed consent was obtained. The study protocol was approved by the ethical committee at Lund University Hospital.
The overall study period was set to 80 days. The subjects were allowed to miss study treatment occasionally, but not for more than a total of 7 days.
The volunteers were randomised in a double-blind fashion to take a daily probiotic drinking straw together with at least 100 ml liquid. Each straw delivered 108 Colony Forming Units (CFU) of L. reuteri protectis, ATCC55730 or placebo.
During the study period the volunteers were asked to report in a diary format any illness symptoms related to the respiratory tract and/or the gastrointestinal tract resulting in sick-leave, and if so, the duration of the sick-leave. They were also asked to report if they had taken the study product as instructed or, if applicable, how many days they missed taking the study product. The diaries were distributed on Day 0 together with the study product and were collected after finalisation of the study period.
Subjects were randomly assigned to L Reuteri (50%), Placebo (50%) in blocks of 4 by a person handing out a sequentially numbered package, containing either L Reuteri or Placebo, together with a diary marked with the same randomisation number. The packages containing either L Reuteri or Placebo were identical in appearance. The randomisation list was generated bythe data management company and used for sorting and numbering the packages by another person. A list in a sealed envelope was kept by the sponsor. Randomisation envelopes were also generated bythe data management company, and kept for safety by the sponsor. The list of randomisation was kept confidential by the statisticians until all results had been generated and sealed.
Results
After ethical committee approval had been granted, informed consent was obtained from 262 employees who started treatment. 132 were randomised to receive L. reuteri and 130 randomised to placebo. 38 subjects in the L. reuteri group and 43 in the placebo group failed to comply with the full protocol requirements and were withdrawn from further analysis. In all cases the reason for non-compliance was that they failed to take the study treatment in accordance with the protocol.
The demographic data for the remaining 181 subjects are given in table 1. There were no statistically significant demographic differences between the two groups.
Table 1 Demographic data
L. reuteri Group n = 94 Placebo Group n = 87
Mean age (years) 44 44
Male / Female (%) 65/35 71/29
Shift work (%) 28 31
The data for symptoms and sick-leave are given in table 2 and figure 1. In the placebo group 23 of 87 subjects reported sick-leave during the study. The corresponding number in the L. reuteri group was 10 of 94 (p < 0.01). Consequently, the percentage of sick-days of working days fell from 0.9% in the placebo group to 0.4% in the L. reuteri group. This difference was also statistically significant (p < 0.01). However, the median length of sick-leave among the subjects who reported any sick-leave was equal in the two groups, 3 days.
Table 2 Sick-leave.
L. reuteri Group n = 94 Placebo Group n = 87 Significance
No. of subjects reporting sick-days (%) 10 (11) 23 (26) p < 0.01, Pearsons χ2
number of sick-days for individuals reporting sick-leave, median 3 3 n.s.
Frequency of sick days*, whole groups 0.4% 0.9% p < 0.01, Kruskal-Wallis
* Frequency of sick days = (reported number of sick days/total number of study days) × 100
Figure 1 Proportion (%) of subjects reporting sick during the study. ■ = Placebo; = L. reuteri.
Among the 53 shift-workers in the study, 9 of 27 (33%) in the placebo group reported sick-leave as compared with none of the 26 (0%) in the L. reuteri group (p < 0.005, Fisher's exact) (figure 1).
There were no adverse events reported during the study.
Discussion
The proportion of subjects that were withdrawn for reasons of non-compliance was fairly high, 31%. This can most probably be explained by the study design itself in combination with the fairly long study duration. When the study was designed it was decided that the study staff should meet the study subjects as little as possible in order to minimise any placebo effect as it was assumed that frequent contacts could increase the individual subject's awareness of health issues beyond the normal behaviour.
The outcome of the study demonstrates that daily intake of L. reuteri can reduce the proportion of subjects reporting sick from gastrointestinal or respiratory tract diseases by 60%. The effect was highly statistically significant and similar to the findings by Weizman et al [5], where small children in day-care centres had a 70% lower frequency of absence when given L. reuteri as compared with placebo.
As demonstrated elsewhere [6-8], L. reuteri is efficient both in preventing and treating acute diarrhoea and gastroenteritis in young children. In a study on healthy adults it was shown that L. reuteri was able to stimulate the immune system by recruiting CD4+-cells [9]. Such stimulation by L. reuteri has been observed in animal models and is associated with an improved response to pathogen infection [10]. Although the exact mechanism of action cannot be defined from our study it is likely that such an immune-stimulation lies behind the reduced morbidity in the subjects taking L. reuteri. This stimulation may also explain why the beneficial effect of L. reuteri in our study was specifically apparent among shift-workers. This subset consisted of 31% of the total study groups and therefore some caution is warranted when interpreting this result. Nevertheless, shift-workers are known to be at risk for having a weaker immuno-defence as compared to those working day-time shifts only [3]. Consequently it can be argued that shift-workers would benefit relatively more by the immuno-stimulating effect of L. reuteri.
In conclusion, the present study demonstrates that L. reuteri is effective to promote work-place healthiness. In the studied population sick-days caused by respiratory or gastrointestinal diseases could be reduced by 55% by the use of L. reuteri group as compared with the placebo group. Translated to the total Swedish work-force, this translates to a total of 4.3 million working days of improved productivity per year (3.9 million employed, 220 working days per year and 0,5% "saved" days). Our results indicate that the effect on shift-work productivity could probably be even more profound but this issue should be addressed in further studies.
List of abbreviations used
CFU = Colony Forming Units
Competing interests
PT and VS declare that they have no competing interests. AZ is a stockholder and an employee of Biogaia AB that holds several patents on Lactobacillus reuteri.
Authors' contributions
PT and VS participated in the design of the study and were responsible for the execution of the study. AZ conceived the study and drafted the design and the manuscript. All authors read and approved the final manuscript.
Acknowledgements
The authors would like to express their gratitude to Agneta Persson and Ann Lindberg at Tetra Pak and Karin Diderot at BioGaia AB. Their dedication and efforts during the execution of the study contributed greatly to the speedy and smooth completion.
BioGaia AB provided the study material and funding for the statistical analysis of the study.
==== Refs
Confederation of Swedish Enterprise Facts about Health Insurance and Paid Sick Leave 2005
Jeding K Hägg GM Marklund s Nygren Å Theorell T Vingård E Ett friskt arbetsliv Arbete och hälsa vetenskaplig skriftserie 1999 22 1 88
Mohren DCL Jansen NWH Kant I Galama J van den Brandt PA Swaen GMH Prevalence of common infections among employees in different work schedules JOEM 2002 44 1003 1011 12449906
Cummings JH Antoine J-M Azpiroz F Bourdet-Sicard R Brandtzaeg P Calder PC Gibson GR Guarner F Isolauri E Pannemans D Shortt C Tuijtelaars S Watzl B PASSCLAIM – Gut health and Immunity Eur J Nutr 2004 43 118 173
Weizman Z Asli G Alsheikh A Effect of a probiotic infant formula on infections in child care centers: Comparison of two probiotic agents Pediatrics 2005 115 5 9 15629974
Shornikova A-V Casas IA Isolauri E Mykkänen H Vesikari T Lactobacillus reuteri as a therapeutic agent in acute diarrhea in young children J Pediatr Gastroenterol Nutr 1997 24 399 404 9144122 10.1097/00005176-199704000-00008
Shornikova A-V Casas IA Mykkänen H Salo E Vesikari T Bacteriotherapy with Lactobacillus reuteri in rotavirus gastroenteritis Pediatr Infect Dis J 1997 16 1103 1107 9427453 10.1097/00006454-199712000-00002
Ruiz-Palacios G Guerrero ML Hilty M Dohnalek M Newton P Calva JJ Costigan T Tuz F Arteaga F Feeding of a probiotic for the prevention of community-acquired diarrhea in young Mexican children Ped Res 1996 39 184A
Valeur N Engel P Carbajal N Connolly E Ladefoged K Colonization and Immunomodulation by Lactobacillus reuteri ATCC 55730 in the Human Gastrointestinal Tract Appl Environ Microbiol 2004 70 1176 1181 14766603 10.1128/AEM.70.2.1176-1181.2004
Casas IA Dobrogosz WJ Validation of the probiotic concept: Lactobacillus reuteri confers broad-spectrum protection against disease in both humans and animals Microb Ecol Health Dis 2000 12 247 285
|
16274475
|
PMC1298318
|
CC BY
|
2021-01-04 16:36:33
|
no
|
Environ Health. 2005 Nov 7; 4:25
|
utf-8
|
Environ Health
| 2,005 |
10.1186/1476-069X-4-25
|
oa_comm
|
==== Front
Harm Reduct JHarm Reduction Journal1477-7517BioMed Central London 1477-7517-2-231625314510.1186/1477-7517-2-23CommentaryCultural Approach to HIV/AIDS Harm Reduction in Muslim Countries Hasnain Memoona [email protected] Department of Family Medicine, College of Medicine, University of Illinois at Chicago, Chicago, Illinois, USA2005 27 10 2005 2 23 23 30 11 2004 27 10 2005 Copyright © 2005 Hasnain; licensee BioMed Central Ltd.2005Hasnain; 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.
Muslim countries, previously considered protected from HIV/AIDS due to religious and cultural norms, are facing a rapidly rising threat. Despite the evidence of an advancing epidemic, the usual response from the policy makers in Muslim countries, for protection against HIV infection, is a major focus on propagating abstention from illicit drug and sexual practices. Sexuality, considered a private matter, is a taboo topic for discussion. Harm reduction, a pragmatic approach for HIV prevention, is underutilized. The social stigma attached to HIV/AIDS, that exists in all societies is much more pronounced in Muslim cultures. This stigma prevents those at risk from coming forward for appropriate counseling, testing, and treatment, as it involves disclosure of risky practices. The purpose of this paper is to define the extent of the HIV/AIDS problem in Muslim countries, outline the major challenges to HIV/AIDS prevention and treatment, and discuss the concept of harm reduction, with a cultural approach, as a strategy to prevent further spread of the disease. Recommendations include integrating HIV prevention and treatment strategies within existing social, cultural and religious frameworks, working with religious leaders as key collaborators, and provision of appropriate healthcare resources and infrastructure for successful HIV prevention and treatment programs in Muslim countries.
AIDSHIVinjection drug usersrisk factorsharm reductionbehavioral interventions
==== Body
Introduction
AIDS is far more than a medical and biological problem [1]. Around the world, in the year 2003, the AIDS epidemic claimed an estimated three million lives, and almost five million people acquired HIV, 700,000 of them children [2]. The current course of the epidemic is unlikely to change unless the people affected, and those at risk, make a concerted effort to adopt preventive measures. Apart from inadvertent modes of transmission, such as vertical transmission from mother to child and accidental needle stick injuries among health care professionals, certain types of behaviors, such as unprotected sexual intercourse and sharing of hypodermic needles, place individuals at increased risk for HIV and AIDS. The disease is therefore largely avoidable by changes in personal behavior, in other words by voluntary choice. Containment of the AIDS epidemic thus depends on effecting change in behavior and lifestyle to break the chain of transmission. This is all the more challenging because the forces that shape and influence human behavior that is injurious to health are very complex and poorly understood.
In recent years, increasing attention is being paid to the manner in which social and cultural variables influence risk behaviors related to HIV infection transmission. Though the association of contentious ethical and moral issues with HIV risk behaviors exists in all societies, it is much more pronounced in the Muslim world. Thus understanding the role of social and cultural variables affecting HIV transmission in Muslim countries is critical for the development and implementation of successful HIV prevention programs.
Harm reduction is a pragmatic philosophy that aims to reduce risks to the individual and the community associated with some often stigmatized, antisocial or illegal behaviors. For HIV/AIDS prevention in Muslim countries, the concept of harm reduction is just as important as in non-Muslim countries. The perspective of harm reduction, developed primarily from work on AIDS and drug problems in the Netherlands [3,4] and United Kingdom [5,6], is a pragmatic approach to the social and individual problems associated with the misuse of psychoactive drugs. In the context of injection drug users, it translates into making sure that if drug misuse cannot be eliminated, some of the problematic risk behaviors leading to HIV transmission, such as sharing of contaminated injection equipment, be reduced. Harm reduction provides a strong rationale for such services as syringe-exchange programs and methadone maintenance treatment.
The purpose of this paper is to explore the extent of the HIV/AIDS problem in Muslim countries and discuss the modalities of employing a cultural approach as a strategy for harm reduction and, hence, prevention of further spread of the disease.
The Changing Face of HIV/AIDS: An Emerging Problem in Muslim Countries
The reliability of the available HIV/AIDS incidence, prevalence and mortality data for Muslims is low because many Muslim countries either do not report their statistics or are under-reporting. Global epidemiological indicators, including data from the World Health Organization's Global Health Atlas, do indicate evidence of the burgeoning threat of an HIV/AIDS crisis in Muslim countries. Table 1 provides HIV/AIDS prevalence and AIDS-related mortality data in countries with 50 percent or greater Muslim population, for the period 2001 to 2003. A recent report from the National Bureau of Asian Research in the United States also notes that the ever-growing HIV/AIDS crisis in the Muslim world is a problem that poses potentially serious dangers at the national, regional, and international levels [7].
Table 1 HIV/AIDS prevalence and AIDS-related mortality in countries with 50 percent or greater Muslim population, 2001–20031
Country Estimated number of adults and children living with HIV/AIDS Estimated number of deaths due to AIDS
Year Year
2001 2003 2001 2003
1 Afghanistan * * * *
2 Albania * * * *
3 Algeria * <10 000 * <500
4 Azerbaijan <10 000 <10 000 <500 *
5 Bahrain <10 000 <10 000 * <500
6 Bangladesh 10 000 – <100 000 500 – <1 000
7 Brunei Darussalam <10 000 <500
8 Burkina Faso 100 000 – <500 000 100 000 – <500 000 10 000 – <50 000 10 000 – <50 000
9 Chad 100 000 – <500 000 100 000 – <500 000 10 000 – <50 000 10 000 – <50 000
10 Cocos (Keeling Island) * * * *
11 Comoros * * * *
12 Djibouti * <10 000 * 500 – <1 000
13 Egypt <10 000 10 000 – <100 000 * 500 – <1 000
14 Eritrea 10 000 – <100 000 10 000 – <100 000 <500 1 000 – <10 000
15 Ethiopia >=2 M 1 M – <2 M >=100 000 >=100 000
16 Gambia <10 000 <10 000 <500 500 – <1 000
17 Gaza Strip * * * *
18 Guinea * 100 000 – <500 000 * 1 000 – <10 000
19 Guinea-Bissau 10 000 – <100 000 * 1 000 – <10 000
20 Indonesia 100 000 – <500 000 100 000 – <500 000 1 000 – <10 000 1 000 – <10 000
21 Iran (Islamic Republic of) 10 000 – <100 000 10 000 – <100 000 <500 500 – <1 000
22 Iraq <10 000 <10 000 * *
23 Jordan <10 000 <10 000 * <500
24 Kazakhstan <10 000 10 000 – <100 000 <500 <500
25 Kuwait
26 Kyrgyzstan <10 000 <10 000 <500 <500
27 Lebanon <10 000 <500
28 Libyan Arab Jamahiriya <10 000 10 000 – <100 000 * *
29 Malaysia 10 000 – <100 000 10 000 – <100 000 1 000 – <10 000 1 000 – <10 000
30 Maldives <10 000 * * *
31 Mali 100 000 – <500 000 100 000 – <500 000 10 000 – <50 000 10 000 – <50 000
32 Mauritania * <10 000 * <500
33 Mayotte * * * *
34 Morocco 10 000 – <100 000 10 000 – <100 000 * *
35 Niger * 10 000 – <100 000 * 1 000 – <10 000
36 Nigeria >=2 M >=2 M >=100 000 >=100 000
37 Oman <10 000 <10 000 * <500
38 Pakistan 10 000 – <100 000 10 000 – <100 000 1 000 – <10 000 1 000 – <10 000
39 Qatar * * * *
40 Saudi Arabia * * * *
41 Senegal 10 000 – <100 000 10 000 – <100 000 1 000 – <10 000 1 000 – <10 000
42 Sierra Leone 100 000 – <500 000 * 10 000 – <50 000 *
43 Somalia 10 000 – <100 000 * * *
44 Sudan 100 000 – <500 000 100 000 – <500 000 10 000 – <50 000 10 000 – <50 000
45 Syrian Arab Republic * <10 000 * <500
46 Tajikistan <10 000 <10 000 * <500
47 Togo 100 000 – <500 000 100 000 – <500 000 10 000 – <50 000 10 000 – <50 000
48 Tunisia * <10 000 * <500
49 Turkey * * *
50 Turkmenistan <10 000 <10 000 <500 *
51 United Arab Emirates * * * *
52 United Republic of Tanzania 1 M – <2 M 1 M – <2 M >=100 000 >=100 000
53 Uzbekistan <10 000 10 000 – <100 000 <500 <500
54 West Bank * * * *
55 Western Sahara * * * *
56 Yemen * * * *
1 Sources:
a. For HIV/AIDS statistics: World Health Organization Global Health Atlas 2005, available at:
b. For percentage of Muslim population: CIA World Fact Book 2005, available at:
* Data not available
The continent of Africa, particularly the southern region, continues to have the highest HIV/AIDS incidence and prevalence rates globally [8]. The number of HIV-positive adults range from 6–10% in Nigeria, and 10–18% in Ethiopia; both countries have a majority of residents who are Muslims. By the year 2010, 40% of the African population, where the disease burden is highest, will be Muslim [9]. Some Muslim countries, such as Sudan and Nigeria already show evidence of an explosive epidemic (Table 1). In the Eastern Hemisphere, countries like Kazakhstan, Kyrgyzstan, Tajikistan, Turkmenistan, and Uzbekistan, which were part of the former Soviet Union, face a young and rapidly growing epidemic. East and Southeast Asia, which include countries like China and India containing some of the world's largest populations, show indicators of soon surpassing Africa in terms of their absolute number of cases, if HIV/AIDS rates continue to escalate at their current rate [2]. These projections hold particular relevance for HIV/AIDS in Muslim populations; India and China, though not identified as Muslim countries, have a significant number of Muslims (approximately 138 million Muslims in India, and 40 million in China).
The under-reporting of HIV and AIDS cases in Muslim countries has serious bearings on disease surveillance and monitoring. In the Eastern Mediterranean Region, an estimated 700,000 people are currently living with HIV/AIDS but only 14,198 AIDS cases have been officially registered since the start of the epidemic, indicating under detection, under-reporting, and surveillance difficulties [10]. Of the 22 countries of the region, complete data were lacking for nine countries for 2003, and data from two others had to be discarded because of reporting problems. Although the prevalence of HIV infection among adults in the Eastern Mediterranean Region (0.3%) is roughly equivalent to figures for Western Europe, the number of estimated new HIV/AIDS cases for 2003 is about 60% higher (55,000 in the Eastern Mediterranean Region versus 35,000 in Western Europe), demonstrating the alarming increase in the epidemic in the region [10]. Even though the absolute number of HIV/AIDS cases in the majority of Muslim countries, particularly those in the Middle East or South East Asia, such as Pakistan, may still be lower than other countries; complacency toward this issue will be costly, both in terms of lives and health care costs.
Reasons for the spread of HIV in Muslim countries are open to speculations. Islam places a high value on chaste behavior and prohibits sexual intercourse outside of marriage. It specifically prohibits adultery, homosexuality, and the use of intoxicants [11]. Then how can the spread of HIV/AIDS in Muslim countries be explained? A logical explanation is that in spite of Islamic teachings, some Muslims do engage in activities that lead to acquiring HIV; these risky practices include illicit drug use and/or premarital or extra marital sex. Men who engage in risky behaviors have the potential of transmitting the disease to their unsuspecting wives. Women, on the other hand, also are directly susceptible; in many Muslim countries, brothels and other forms of commercial sex trade are prevalent. The sex workers have poor social support and they are not screened for sexually transmitted diseases including HIV, thus contributing to the spread of infection. Injection drug users (IDUs) also are rapidly becoming a population of increasing concern in the transmission of HIV and AIDS, not only in western countries such as the United States [12-15], but also in developing countries, including Muslim countries. Sex- and drug-related behaviors of IDUs can facilitate HIV transmission even when syringes are not directly shared [15-18].
Challenges
With regard to curtailing the spread of disease, it is particularly troublesome that a majority of governments in countries with primarily Muslim populations have been slow to respond to the rapidly spreading disease. Despite the evidence of an advancing epidemic, the typical response from the policy makers in Muslim countries is to propagate Muslim ideals, mainly abstention from illicit drug and sexual practices, for protection against HIV infection. Sexuality, considered a private matter, is taboo for discussion. More importantly, there is a denial by most governments in Muslim countries that they are facing an increasing HIV/AIDS threat.
The issue of HIV/AIDS prevention in Muslim countries is a complex problem and requires a multifaceted approach with particular attention to cultural norms. In order to devise harm reduction strategies for HIV prevention in these countries, it is important to study the social dynamics and practices of the populations at risk. Analysis of the cultural context in which risk behaviors occur provides meaningful insight into those factors that shape and define the external reality within which these behaviors take place. Knowledge of why people behave in certain ways and the resources available to them becomes helpful in assisting them to access and utilize available preventive and therapeutic resources. In the context of high-risk groups, it is important to understand that even within them, some individuals choose to indulge in risk behaviors while others do not.
Philipson and Posner [19] note that human actors make rational choices aimed at maximizing the expected utility of the outcome. The subjective welfare of the actor and presence of uncertainty are two inherent components of expected utility maximization. When acquiring information is costly, an uninformed choice – one that underestimates or overestimates the risk to health of some contemplated action – may still be expected utility maximization. Therefore, when education and counseling services are not readily and cheaply available, or when accessing such services means that the user has to disclose risk behaviors and is afraid to do so, he/she has no course but to make uninformed decisions. Effective counseling and education have been shown to change sexual behavior and reduce the risk of HIV transmission even in high-risk groups.
In the context of HIV/AIDS prevention and treatment in Muslim countries, the principles of harm reduction or harm minimization can certainly be utilized to prevent or minimize the spread of HIV infection. However, a clear distinction needs to be made that this approach does not advocate illicit drug and sex related practices. The harm reduction concept which has been successfully applied in substance abusers can also be applied to other high risk groups, such as commercial sex workers. Because consistent condom use has been linked to reductions in HIV seroincidence [15,20,21], and because reductions in frequency of unprotected sex also predict lower levels of HIV infection incidence, the behavioral effects of the intervention carry considerable public health importance. In addition to counseling, IDUs could be provided needles at reduced prices or even free of charge. Regular screenings for sexually transmitted infections (including HIV), and antiretroviral therapy, should make significant contributions to HIV prevention, early detection, and appropriate treatment where required.
In the Muslim World, religion defines culture and the culture gives meaning to every aspect of an individual's life. The following contentious issues need particular attention when designing HIV prevention programs for Muslim countries:
1. Gender Inequality
In a majority of Muslim societies, there exists an imbalance in power between men and women, which is apparent in heterosexual relations as well as in the economic and social spheres of life – with men having greater power than women. For most women, the private life within the sanctuary of their houses is their whole life. Women remain uneducated and deprived of resources, making them unaware of their civil, legal and sexual rights, economically vulnerable and largely dependent on men. Due to these inequalities, women are more susceptible to contracting HIV/AIDS as they are less likely to be able to negotiate with their partners infected with HIV/AIDS. Women also are easy targets for abusive relationships and are less able to cope with illness once infected.
2. Stigma and Discrimination
The social stigma attached to HIV/AIDS that exists in all societies is much more pronounced in Muslim cultures due to the religious doctrine regarding illicit sex and drug related practices. There are greater negative sanctions for illicit sexual conduct than drug use. Even if there is a suspicion of illicit sexual conduct, the affected person(s) is discriminated against and shunned by the family as well as by the community. The stigma attached to risk behaviors thus prevents those at risk from coming forward for appropriate counseling, testing and treatment, as this would involve disclosure of their risky practices. This results in creating barriers to successful implementation of prevention and treatment strategies where they do exist.
3. Ignorance/Misinformation
In developed countries, a majority of the population is aware of the modes of transmission for HIV infection, whereas in the developing countries, misconceptions about the disease and its causes are rampant. Most persons residing in Muslim countries assume that all HIV infections are transmitted only through immoral sexual behaviors and are unaware that it can also be transmitted inadvertently through mother-to-child, accidental pricking of skin and contact with contaminated blood (as in the case of health care professionals) or the possibility of an innocent spouse getting infected by the husband who may have acquired HIV though sexual or drug related contact with other infected persons. Therefore, due to lack of education, expression of compassion towards HIV/AIDS patients is perceived as tolerance towards the practices that lead to acquiring the infection.
4. Other issues
In addition to the issues outlined above, the main challenges to instituting an HIV prevention approach include poverty and economic instability, lack of education, wars, internal conflicts, refugees, migrant labor forces, intimidating role of religious leaders and activists, and lack of healthcare resources and infrastructure.
In summary, the existing social, cultural and religious frameworks in Muslim countries do not provide an environment for any safe disclosure for persons who are infected. Hence, the development of effective prevention and support services is often impeded. Meanwhile, growing gender imbalances in HIV rates among women, and the tendency for the virus to be found disproportionately among marginalized and disadvantaged populations throughout the Muslim world, mirror deeply entrenched systems of societal inequality that help to fuel further spread of the epidemic. For those who are not educated, cultural expectations are very difficult to disregard. Containment of the HIV/AIDS epidemic in Muslim countries depends on a combination of individual and community level efforts to effect change in behavior and lifestyle to break the chain of transmission.
Recommendations
There is an urgent need for developing and implementing policy and programs that provide AIDS education and awareness, prohibit stigmatization, and advocate compassion. Like most religions, Islam condemns homosexuality, drug use, and sex outside of marriage. Though the most important means of protection is obviously abstinence from sex and to remain faithful to the marriage partner, however, Muslims must recognize that in many instances there is a gap between religious teaching and practice; risky behaviors that may not be allowed by Islam are indeed practiced. The main challenge is how to bridge this gap. Religious scholars seem to be divided on the concept of harm reduction. In countries where HIV/AIDS is a rapidly rising threat, such as Uganda [22] and Indonesia [23], religious scholars are taking a more flexible stance and justify the provision and use of condoms and clean needles through Qur'anic and Hadith passages. They reason that the sanctity of life is greater than the sin of condom use and that this strategy can be used as a short term measure, permissible under a state of emergency. On the other hand, in countries with low incidence and prevalence of HIV/AIDS, religious leaders believe that approving promotion of condoms and clean needles will encourage sexual promiscuity and drug use. To address these controversies, the Organization of Islamic Countries (OIC) should step forward and assume a central role in drafting harm reduction strategies for Muslim countries.
Any effort directed at harm reduction and HIV prevention needs to take into consideration the powerful impact of religious leaders in the community as they play a critical role in Muslim culture. It is important to be cognizant of the reality that religious leaders take issue with harm reduction strategies due to the moral issue involved with the idea of harm reduction. There is a perception that promoting safe injection and sex related practices will promote illicit drug and sex behaviors. Hence, for HIV prevention programs to be successful, collaboration with religious scholars and leaders is a key element. It is critical to win their confidence and educate them. Not all cases of HIV and AIDS are contracted through needle exchange or sexual intercourse, and second, regardless of the route of transmission, once a person is infected, he/she should not be treated as a criminal but should be considered a patient suffering from a disease. Just as patients afflicted with any disease deserve the provision of clinical care and support from their family and the society, patients suffering from HIV/AIDS have all the rights to the same services, support and compassion.
Examples of successful prevention efforts that involved religious leaders in Muslim societies include those of Uganda [24] and Senegal [25,26]. In 1992, the Islamic Medical Association of Uganda designed an AIDS prevention project and after conducting a baseline survey prior to community level activities, instituted prevention activities in local Muslim communities. Twenty-three trainers educated over 3,000 religious leaders and their assistants, who in turn educated their communities about AIDS during home visits and at religious gatherings. After two years, there was a significant increase in accurate knowledge of HIV transmission, methods of preventing HIV infection and the risk associated with ablution of the dead and unsterile circumcision. More importantly, there was a significant reduction in self-reported sexual partners among the young respondents of less than 45 years of age. In addition, there was a significant increase in self-reported condom use among males in urban areas [24]. A recent report notes that there is a tangible decline of HIV/AIDS incidence among members of Uganda's Muslim community from 18 percent in the early 90's to the current rate of 6 percent [22].
Senegal also is one of the best examples regarding HIV/AIDS prevention by engaging religious institutions in a proactive role. In March 1990, 260 religious leaders attended a conference on AIDS and reached a consensus to make AIDS control a national priority. Unlike other African countries, HIV/AIDS prevention is a regularly discussed topic in the Friday prayers in mosques in Senegal. From 1989 to 1996, the levels of HIV infection estimated in four sentinel urban regions remained stable at around 1.2 percent in the population of pregnant women, and at three percent in male STD patients [25]. The current 1.2 percent AIDS prevalence rate in the general population of Senegal is in stark contrast to the rest of the continent which has an average AIDS prevalence rate of 30–35 percent. The level of knowledge of preventive practices relating to HIV/AIDS among the general population exceeded 90 percent in the early 1990s.
The reasons for Senegal's successful HIV control are: 1) a good STD tracking system that has been in place since 1969; and 2) AIDS education, utilizing religious institutions and mass media sources such as the radio [26]. From available data, Senegal can rightfully claim to have contained the spread of HIV by intervening early and comprehensively to increase awareness of and knowledge about HIV/AIDS, and to promote safe sexual behaviors via religion and education.
In the context of the AIDS epidemic, limited attention has been paid to the manner in which political, economic and social variables constrain or enable individual behavior related to AIDS [27]. The association of variables such as social capital [28], human capital (educational attainment) [29], and religiosity [30] with HIV risk compels prevention efforts to look beyond the traditional biomedical model of disease prevention. In order to find workable means of combating this disease, research also needs to be directed towards its critically important cultural dimensions [31-33]. The major focus of preventive efforts should be aimed at behavioral change, minimizing the transmission of HIV through unsafe sexual practices and the sharing of contaminated injection drug equipment. The models developed and successfully implemented in western countries can be tailored according to local culture and norms to address the needs of those at risk of or suffering from HIV/AIDS in Muslim countries. In addition to proper food, housing, education, employment, regardless of country of residence, all persons at risk of or suffering from HIV/AIDS should have the right to safe disclosure and appropriate health care.
There also is an urgent need in Muslim countries for increasing infection surveillance and enhancing HIV preventive and therapeutic services for high-risk groups, such as commercial sex workers, drug abusers, and those with alternative sexual lifestyles, not simply those who identify themselves as being either infected or possibly infected. In addition, legislative and social changes, such as protecting the legal rights of the infected, promoting safer alternative behaviors among high-risk groups, and spreading the message that being a good Muslim can include taking care of those infected by HIV would be helpful in combating the spread of the disease. HIV/AIDS education and control efforts could also become part of each citizen's duty. The international community can also assist by helping poorer countries establish social programs or simply sharing experience in drug treatment and behavioral change efforts [7].
In summary, our recommendations to stem the spread of HIV in Muslim countries include:
1. Addressing the underlying societal problems such as poverty, lack of education and gender imbalance;
2. Developing collaborative prevention and care models (including all possible stakeholders such as, religious scholars, academics, expert health professionals, policy makers, non governmental organization, community based organizations, and HIV positive persons);
3. Development and provision of appropriate healthcare resources and infrastructure including:
• Blood safety and infection control
• Appropriate surveillance and reporting mechanisms
• Drug abuse prevention and rehabilitation services
• Medical care and social support including HIV counseling, testing and treatment facilities
• Adequate number of trained health care workforce
• Appropriate reproductive health care programs
• Broader efforts directed at enhancing information, education and communication.
Conclusion
The challenge of addressing the rising threat of the spread of HIV/AIDS in Muslim countries/societies is significant. The most effective public health method of controlling the spread of AIDS is education and changing the way people behave. Political, financial, and social barriers have often kept the most effective prevention and treatment strategies from reaching persons at the highest risk. There is a need to ensure sustained access to preventive and treatment services for all high-risk groups. The goal of prevention is best achieved through an ongoing process, open to change and flexible to adaptation. Incorporating such change within religious and cultural frameworks is no easy task. This is the challenge we are facing and it is up to us, individually and collectively, as health care professionals and researchers to respond.
To ensure ongoing usefulness of public health policies related to HIV prevention, we must learn to synthesize old knowledge with new, and, at the same time, utilize opportunities to choose new directions. The framework proposed in this paper can serve as an initial model for appropriate HIV prevention and care programs in Muslim countries. Risk needs to be viewed within the context of the social subculture of Muslim countries to design strategies to reduce risk behaviors related to HIV transmission. The social dimension of health mandates that policy and program measures to stop AIDS be a balance of social and biomedical scientific efforts. Our recommendations include education, involvement and mobilization of diverse stakeholders, particularly religious leaders; establishing sustainable financing for AIDS treatment and drug procurement; institutingregulatory mechanisms to ensure blood safety and appropriate delivery of HIV/AIDS counseling, screening and treatment services; improvement in health infrastructure; and training of health care workers. None of the above will be successful without reducing the stigma associated with HIV and AIDS, developing compassion for those afflicted, and designing harm reduction strategies which would be conceptually integrated within the existing social, cultural, and religious frameworks in Muslim countries.
==== Refs
Hasnain M Levy JA Albrecht GA HIV/AIDS Encyclopedia of Disability 2005 2 Thousand Oaks, CA, Sage Publications 875 883
United Nations Programme on HIV/AIDS (UNAIDS) and World Health Organization (WHO) AIDS epidemic update: 2003. Geneva, Switzerland 2003
Buning EC Coutinho RA van Brussel GH van Santen GW van Zadelhoff AW Preventing AIDS in drug addicts in Amsterdam Lancet 1986 1 1435 2872527 10.1016/S0140-6736(86)91570-9
Langendam MW van Brussel GH Coutinho RA van Ameijden EJ Methadone maintenance and cessation of injecting drug use: results from the Amsterdam Cohort Study Addiction 2000 95 591 600 10829334 10.1046/j.1360-0443.2000.95459110.x
Des Jarlais DC Sotheran JL The public health paradigm for AIDS and the drug use: shifting the time frame Br J Addict 1990 85 348 349
Brettle RP HIV and harm reduction for injection drug users AIDS 1991 5 125 136 2031687
Kelley LM Eberstadt N Behind the veil of a Ppublic health crisis: HIV/AIDS in the Muslim world. NBP Special Report: The National Bureau of Asian Research 2005
Hasnain M Antenatal HIV Screening and Treatment in South Africa: Social Norms and Policy Options Afr J Reprod Health 2004 8 77 85 15623122
Sigmon M Africa. Mission Network News. Delivered at the International Coalition for Religious freedom Conference on Religious Freedom and the new millennium 2004
World Health Organization The Work of WHO in the Eastern Mediterranean Region. Annual Report of the Regional Director 1 January 31 December 2003 2004 Cairo, World Health Organization Regional Office for the Eastern Mediterranean
Pickthall MM The Meaning of the Glorious Qur'an. 1930 Hyderabad, India. Chapters and Verses; 17:32, 26:165–166, 5:90
Celentano DD Vlahov D Cohn S Shadle VM Obasanjo O Moore RD Self-reported antiretroviral therapy in injection drug users JAMA 1998 280 544 546 9707145 10.1001/jama.280.6.544
Sherer R Adherence and antiretroviral therapy in injection drug users JAMA 1998 280 567 568 9707152 10.1001/jama.280.6.567
Holmberg SD The estimated prevalence and incidence of HIV in 96 large US metropolitan areas Am J Public Health 1996 86 642 654 8629714
Koester S Booth RE Zhang Y The prevalence of additional injection-related HIV risk behaviors among injection drug users J Acquir Immune Defic Syndr Hum Retrovirol 1996 12 202 207 8680893
Rhodes T Stimson GV Quirk A Sex, drugs, intervention, and research: from the individual to the social Subst Use Misuse 1996 31 375 407 8834267
Needle RH Coyle S Cesari H Trotter R Clatts M Koester S HIV risk behaviors associated with the injection process: multiperson use of drug injection equipment and paraphernalia in injection drug user networks Subst Use Misuse 1998 33 2403 2423 9781822
Thorpe LE Bailey SL Huo D Monterroso ER Ouellet LJ Injection-related risk behaviors in young urban and suburban injection drug users in Chicago (1997–1999) J Acquir Immune Defic Syndr 2001 27 71 78 11404523
Philipson TJ Posner RA Private Choices and Public Health. The AIDS Epidemic in an Economic Perspective 1993 Cambridge, MA, Harvard University Press
de Vincenzi I A longitudinal study of human immunodeficiency virus transmission by heterosexual partners. European Study Group on Heterosexual Transmission of HIV N Engl J Med 1994 331 341 346 8028613 10.1056/NEJM199408113310601
Padian NS O'Brien TR Chang Y Glass S Francis DP Prevention of heterosexual transmission of human immunodeficiency virus through couple counseling J Acquir Immune Defic Syndr 1993 6 1043 1048 8340895
Farrell M Condoms and AIDS Prevention: A comparison of three faith-based organizations in Uganda AIDS and Anthropology Bulletin 2003 15 3
Centers for Disease Control and Prevention International News Indonesia's Insulation from HIV/AIDS Wears Thin Excerpted from: Deutsche Presse-Agentur 19/06/03
Kagimu M Marum E Wabwire-Mangen F Nakyanjo N Walakira Y Hogle J Evaluation of the effectiveness of AIDS health education intervention in the Muslim community in Uganda AIDS Educ Prev 1998 10 215 228 9642420
Meda N Ndoye I M'Boup S Wade A Ndiaye S Niang C Sarr F Diop I Carael M Low and stable HIV infection rates in Senegal: natural course of the epidemic or evidence for success of prevention? AIDS 1990 13 1397 1405 10449294
Lagarde E Pison G Enel C Risk behaviours and AIDS knowledge in a rural community of Senegal: relationship with sources of AIDS information Int J Epidemiol 1998 27 890 896 9839749 10.1093/ije/27.5.890
Fournier AM Carmichael C Socioeconomic influences on the transmission of human immunodeficiency virus infection: the hidden risk Arch Fam Med 1998 7 214 217 9596453 10.1001/archfami.7.3.214
Hasnain M Mensah EK Levy JA Sinacore JM Social Capital as a Predictor of HIV/AIDS Risk Behaviors XIV International AIDS Conference, Barcelona, Spain 2002
Hasnain M Levy JA Sinacore JM Mensah EK Human Capital and HIV Risk in Injection Drug Users XV International AIDS Conference, Bangkok, Thailand 2004
Hasnain M Sinacore JM Levy JA Mensah EK Influence of Religiosity on HIV Risk Behaviors in Active Injection Drug Users AIDS Care 2005 17 892 901 16120505
UNESCO Mexico City Declaration on Cultural Policies World Conference on Cultural Policies, Mexico City 26 July – 6 August 1982
UNESCO A Cultural Approach to HIV/AIDS Prevention and Care. Summary of Country Assessments. An International Overview UNESCO/UNAIDS Research Project Studies and Reports, Special Series 2002 Division of Cultural Policies, UNESCO
Adrien A Godin G Cappon P Singer SM Maticka-Tyndale E Willms D Overview of the Canadian study on the determinants of ethnoculturally specific behaviours related to HIV/AIDS Can J Public Health 1996 87 S4 S10 8705923
|
16253145
|
PMC1298319
|
CC BY
|
2021-01-04 16:36:49
|
no
|
Harm Reduct J. 2005 Oct 27; 2:23
|
utf-8
|
Harm Reduct J
| 2,005 |
10.1186/1477-7517-2-23
|
oa_comm
|
==== Front
Head Face MedHead & Face Medicine1746-160XBioMed Central London 1746-160X-1-101627091210.1186/1746-160X-1-10ReviewPalatal development of preterm and low birthweight infants compared to term infants – What do we know? Part 3: Discussion and Conclusion Hohoff Ariane [email protected] Heike [email protected] Ulrike [email protected] Erik [email protected] Poliklinik für Kieferorthopädie, Universitätsklinikum, Westfälische Wilhelms-Universität, Münster, Germany2 Department of Neonatology, Brighton & Sussex University Hospitals, UK3 Klinik für Kinderheilkunde, Division of Neonatology, Universitätsklinikum, Westfälische Wilhelms-Universität, Münster, Germany2005 2 11 2005 1 10 10 8 9 2005 2 11 2005 Copyright © 2005 Hohoff et al; licensee BioMed Central Ltd.2005Hohoff 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 hypothesized that prematurity and adjunctive neonatal care is 'a priori' a risk for disturbances of palatal and orofacial development which increases the need for later orthodontic or orthognathic treatment. As results on late consequences of prematurity are consistently contradictory, the necessity exists for a fundamental analysis of existing methodologies, confounding factors, and outcomes of studies on palatal development in preterm and low birthweight infants.
Method
A search of the literature was conducted based on Cochrane search strategies including sources in English, German, and French. Original data were recalculated from studies which primarily dealt with both preterm and term infants. The extracted data, especially those from non-English paper sources, were provided unfiltered in tables for comparison (Parts 1 and 2).
Results
Morphology assessment of the infant palate is subject to non-standardized visual and metrical measurements. Most methodologies are inadequate for measuring a three-dimensional shape. Several confounding factors were identified as causes contributing to disturbances of palatal and orofacial development.
Conclusion
Taking into account the abovementioned shortcomings, the following conclusions may be drawn for practitioners and prospective investigators of clinical studies. 1) The lack of uniformity in the anatomical nomenclature of the infant's palate underlines the need for a uniform definition. 2) Metrically, non-intubated preterm infants do not exhibit different palatal width or height compared to matched term infants up to the corrected age of three months. Beyond that age, no data on the subject are currently available. 3) Oral intubation does not invariably alter palatal morphology of preterm and low birthweight infants. 4) The findings on palatal grooving, height, and asymmetry as a consequence of orotracheal intubation up to the age of 11 years are inconsistent. 5) Metrically, the palates of orally intubated infants remain narrower posteriorly, beginning at the second deciduous molar, until the age of 11 years. Beyond that age, no data on the subject are currently available. 6) There is a definite need for further, especially metrical, longitudinal and controlled trials on palatal morphology of preterm and low birthweight infants with reliable measuring techniques. 7) None of the raised confounding factors for developmental disturbances may be excluded until evident results are presented. Thus, early orthodontic and logopedic control of formerly premature infants is recommended up to the late mixed dentition stage.
==== Body
Background
The research on palatal development dates back to the beginning of the last century. Unfortunately, the results of many studies are conflicting in some respects and may be difficult to interpret. Compared to the significant improvements in the survival rate of preterm infants, the knowledge on late consequences of orofacial development in these small patients is still unsatisfactory. A recently published systematic review concluded that further well-designed studies are needed [1]. Therefore, a fundamental analysis of existing methodologies, confounding factors, and outcomes seems motivated.
Traditional reviews are characterized by less stringent inclusion criteria than systematic reviews. Thus, studies from the pre-'evidence based medicine' (EBM) era were included. This has revealed important information about the research on development of the term infant's palate. It could be shown (Part 1) that the newborn's palate can vary considerably and is subject to various influences. Therefore, palatal growth may occur undetected or appear excessive or inadequate.
The general methodologies used for morphometry are comparable between studies from the pre-EBM and EBM eras, whereas the quality of study designs improves over the years. The assessment of palatal alterations is only as accurate as the measurements that are taken from the palate and therefore independent of the general study type. Two-dimensional measurements – however reliable – are of limited value for the description of three-dimensional shape changes, and this measurement technique may contribute conflicting results.
The present part of the review discusses the different results of Parts 1 and 2. Conclusions were drawn based on clinical relevance and the quality of included studies.
Discussion
Clinical relevance of visual inspection of the newborn palate
Palatal height is often employed as a diagnostic criterion of craniofacial syndromes [2], as is an abnormal number of frenula [3]. Thickening of the alveolar ridges may be congenital or acquired, for example following prolonged anticonvulsant therapy [3]. There is the risk that thickened (lateral) palatine and alveolar ridges producing narrowness of the palate may give a false impression of increased palatal height [3,4]. The latter is a less common anomaly, but may be a manifestation of a number of syndromes [2,5]. The distinction is important clinically, since prominent lateral palatine ridges most commonly imply a long-term deficit of neuromuscular function and thus may be an important diagnostic clue to alterations dating back to early prenatal development. Prominent lateral palatine ridges are, for example, a characteristic feature of Smith-Lemli-Opitz syndrome [6]. They are also a feature in Apert's syndrome [7,8] and in disorders in which the tongue is small, displaced or immobilized, including a narrow palate, as well as in mucopolysaccharidosis and other storage diseases with an abnormal accumulation of various metabolic substances within the connective tissue [4].
Clinical relevance of a metric description of the palatal configuration
'Growth is the essence of the developing system. ... Growth of different parts of the body follows a predictable schedule during normal development and maturation. This timetable of development is influenced and controlled by many genetic and environmental factors. Any disturbance in the normal schedule of development and growth may lead to disproportion of physical features. These imbalances may be transient and can sometimes be compensated for by later catch up growth' [3].
In order to establish whether a palate is normal, it is necessary to have reliable information on the extent of variation of the normal gum pad [9]. 'There is a definite need for standards of oral-facial dimensions in children within ... this age range' (6 weeks to 36 months of age) [10]. The information would be extremely valuable for the health care professional treating posttrauma patients or patients with craniofacial anomalies [9,10]. Manufacturers also would benefit from metrical information [11] for example for designing appropriate pacifiers. Data for height and weight are available up to 36 months of age, but there is lack of information on oral dimensions [10].
Palatal height is often employed as a diagnostic criterion of craniofacial syndromes [5]. It was suggested, as a rough guide, that 'when maximum palatal height is greater than twice the height of the teeth, it should be considered abnormally high' [3].
Consequences of intubation on the maxillofacial region
PT and LBW infants often require short or long-term neonatal intubation for resuscitation and to relieve respiratory distress [12].
Complications resulting from intubation, however it is performed, are always to be expected: in cases of nasal intubation, potential problems are nasal deformation [13-16] and subsequent choanal stenosis. Airway obstruction, possible hypoxia [175] or respiratory problems may occur in cases of nasogastric tubes [17]. Neonates and small infants are nasal breathers due to immature coordination of respiratory and oral function [18]; only 6% of PT infants (gestational age 31 – 32 weeks) are able to breathe orally [19]. Because PT neonates are nose breathers orotracheal intubation is often preferred to nasal intubation [20], but also reported to be associated with a higher rate of unplanned extubation [21].
As the palatal bones of fetuses are spongy and connective tissue interspersed at the midline forms a weakened palatal configuration, oral defects can easily result from the trauma of oral intubation [22]. This may result in the inability of the tongue to meet the palate correctly [22] and may give rise to considerable functional impairment [20] like sucking problems and impaired middle ear function [15] or articulation disturbances [15,23], e.g. in the form of a significantly higher incidence of fair or poor speech intelligibility in contrast to non-orally intubated infants [22,24].
The following dental complications are described as potential consequences of oral intubation and can be either caused by lack of oxygen, by the larynoscope blade [25] or by the tube itself [26,27]: enamel hypoplasia in 18 – 70% of preterm neonates [24,25,28,29], severe disruption of the developing enamel organ and deviation of the crown/root angulation [30], dilaceration of primary teeth [28,31], retarded eruption of primary teeth [26,32,33], impaired amelogenesis [24,27,34-36], effects on the position of the central incisors [15].
Palatal complications reported in connection with oral intubation are erosion and indentation of the alveolar ridge [37], notching [30,32,38], a high [12,28,39], and narrow [39,40] palatal shape, asymmetry of the palate [12,22] and cleft palate [41]. It was recommend not to use the term clefting [28], since no oral nasal communication has been demonstrated [41,42].
Alveolar grooving [28], and 'palatal grooving' [12,24,28,30,38,39,39,41-46] have also been described, never occuring in combination [28]. The majority of articles dealing with the phenomenon fail to give a definition of palatal grooving. However, there are three exceptions:
1. Two authors defined a palatal groove as follows: 'Narrow channel of variable depth located near the midline of the palate as identified by visual inspection of the maxillary cast' [39] (Comment: Consider the variability of the term 'narrow').
2. Two other authors, performing intraoral measurement with a micrometer 'from its floor to the surface of the palate at the midpoint of the hard palate', selected a palatal groove of ≥0.5 cm arbitrarily as significant [15] (Comment: Consider how difficult it is to make precise intraoral measurements in a tiny infant).
3. A further group stated: 'By definition, a palatal groove is an architechtural deformity of the palate caused by external pressure from the orotracheal tube' [47].
There are various hypotheses on the cause of grooving:
1. It is an oral manifestation of head flattening commonly seen in very premature infants [48]. The same compressive interplay of forces that contribute to craniofacial narrowing is transferred from the zygomatic structures through the lateral aspects of the hard palate and causes the palatal grooving [49].
2. The deformation results from continuous pressure of the endotracheal tube against the median palatine suture [28,38,39,41,50]. This might be aggravated by the direction of pressure applied to the front of the tube in order to hold it in its desired position [47] and also by sucking [9] and result in a pressure necrosis [37].
3. The groove is caused by constriction of the palate adjacent to the tube [47]. This broadening of the alveolar ridges creates the false impression that the palate has been eroded as a groove; in fact, the palate is intact but partially obscured [50-52]. This finding is confirmed metrically [32]: 'Palatal grooving did not always correspond with relative palatal depth, but did usually occur in intubated infants. We therefore consider that palatal grooving is not caused by the direct pressure of the orotracheal tube. It is more likely that it is due to overgrowth of the lateral palatine ridges'. In their reply to a letter from Ginther [50], Molteni and Bumstead [53] agreed that the term 'palatal groove' might be misleading. 'Groove does not imply a palatal defect or cleft but rather a transient mechanical obstruction to normal ingrowth of the lateral palatine ridges.'
4. Several authors [4,28,50-52] regard an impeded tongue function as the cause of the palatal deformation. Grooving was observed even when the tube did not have a midline location, as there was also an absence of tongue thrust against the palatal shelves, which allowed the shelves to grow together [52].
Unusually prominent lateral palatine ridges have been regarded as a nonspecific feature of a variety of disorders in which there is either a neuromotor dysfunction or a malformation which prevents tongue thrust into the palatal vault, suggesting that a long-standing deficit of tongue thrust is the common pathogenetic mechanism [4,50]. This is frequently associated with reports of a poor sucking reflex in early childhood. In most of these conditions the ridges ultimately appear to attain a normal adult configuration [4,50]. These authors believe, however, that truly narrow, highly arched palates are a very infrequent occurrence and are confused with primarily structural aberrations of the maxilla or the palate, or with prominent lateral palatine ridges.
At age 3–5 a characteristic high palatal vault on the dental casts of formerly intubated children was still observed, and 21% of the intubated infants with high palatal vaults also had palatal grooves; nearly 1/4 of the children suffered from crossbites; neither birthweight nor intubation was related to palatal symmetry [24]. No data was given in the abstract concerning the method nor if preterm or term children were examined.
Quality of studies
Bias, i.e. wrong selection of included and excluded studies could have come over the presented review for several reasons.
Firstly the authors were already strongly involved in the matter and thus not blinded to its subject. Secondly, in only four of the 'metrical control studies' did the authors state explicitly that fullterm infants had been investigated [9,54-56]. In additional two studies, data for term infants could be extracted by the authors of the review because all single figured concerning weight or maturity were given by the authors of these dissertations [57,58]. In most of the metrical studies with term infants, the reliability of the method was not given. We have to comment that the data included in the study are the best evidence we have for the moment concerning a 'control group' of term infants.
Thirdly, non metrical studies and studies without dental casts should be interpreted with caution due to several shortcomings: they suffered from a lack of definition or a non-uniform definition of the term 'palatal grooving', from low case numbers in some studies, from the difficulty of intraoral assessments in very small babies and from subjective assessment of palatal shape. In some studies, the intra-examiner reliabilty was not given [43,44] or statistically significant inter-examiner differences existed [43], whereas in one paper the subjective assessment of relative palatal height turned out to be fairly reliable (approximately 80% inter-and intraexaminer agreement) [59].
Visual descriptions alone cannot always give rise to valid decisions on whether the alterations described in the palatal configuration are in fact palatal grooves respectively deepened palatales or only thickened lateral palatine ridges: putting the visual assessment of the palatal configuration into perspective by means of metric assessments revealed that, although oral intubation may lead to palatal grooving, palatal grooving was not necessarily associated with an increase in palatal depth [32], whereas in another study the subjective assessment of palatal height correlated reasonably well with the palatal index [59].
The above mentioned shortcomings affected the comparability of the non-metrical data and gave rise to vastly varying data on the incidence of grooving (7–90%). Subjective assessments have not the kind of discriminatory power which is nowadays desirable for identifying potential genetic, environmental or developmental associations of deformities. However, a visual inspection of the infant palate may give the pediatrician some important diagnostic clues with respect to syndromes and changed functional patterns. This is why intraoral examinations should be an integral part of postnatal pediatric routine examinations and why non-metric diagnoses were included in this review.
Fourthly, only twelve metrical studies concerning PT infants palates were found, with the methodological quality varying widely [12,15,22,32,48,49,60-65]. One had the exactness of different measuring methods as the primary interest of outcome [64], three examined the effect of protective appliances [48,49,61], four included preschool or school children of a wide age range [12,22,60,65] (one study showed the mean difference in palatal width from 9 – 12 years in girls was 0.9 mm in the molar region [66]), one measured palatal depths intraorally, entailing the risk of being imprecise [15], one study included term and preterm infants [32]. In the majority of studies a problem with the reliabilty of the measuring method was present: Either the reliabilty was not given [15,22,32,62,63], or a significant measuring error for palatal depth was recorded [6], or the coefficent of variation for repeated palatal height measurements ran up to 11.73% [49,64].
Fifthly, the confounding of prematurity, i.e. birtweight and gestational age vs intubation in most cases cannot be resolved, as many preterm infants need the latter.
Sixthly, there is the risk of at least 4 × 2 included papers being 'double publications' reffering to the same group of infants ([67] and [68]; [43] and [44]; [69] and [70]; [62] and [63]). This entails the risk of bias and impact on the conclusions of the review [71].
Seventhly, the following problems are worth to be mentioned:
• The calcium phosphate metabolism has so far been taken into account in only one study [62], the type of milk intake in only one other [49]. As two-thirds of the newborn's stores of calcium and phosphorus are accumulated during the third trimester of pregnancy, and a premature infant born prior to about 28 or 30 weeks gestation would have missed much of his mineral accretion [72], it cannot be exluded that bias came over the metrical studies on PT infants palates due to missing data on nutrition.
• The development of the palate is linked to that of the mandible and can thus not be seen in isolation [73]: the dimensions for the maxillary gum pad do vary considerably beginning with an overjet, i.e. a sagittal distance of upper and lower jaw of >6 mm [74].
• The development of the maxilla is linked to that of the cranial base [73] and cranium. Only two of the metric studies in PT infants [32,49] took this mutual relationship into account.
• The development of the palate is subject to various functions: in comparison with the closed mouth of the term-born infant, the mouth of the PT infant is commonly open [75], which might have a significant implication for orofacial development and was not considered in any of the studies. Attention to the influence of oral feeding was made in only one study [64].
Conclusion
Considering the shortcomings of the currently available articles on palatal development (lack of uniform definitions of palatal morphology, lack of control studies with term infants, lack of studies with determination of the reliability of the measuring method), the following conclusions may be carefully drawn:
The palate of the term newborn
1. The distinctive feature of the infantile palate is the groove system. The lack of uniformity in the nomenclature of the groove system and of the frenula of the infant jaw underlines the need for a uniform definition in the anatomic terminology.
2. The shape of the palate of the term infant can considerably vary, both, visually and metrically.
3. Contradictory information is given with regard to gender differences in palatal shape of term infants.
4. With the exception of one study, in which indian and latino children were included, all studies reported more palatal cysts in term white children compared to black babies. Alveolar notches and alveolar lymphangioma occur more often in black neonates. In term infants, no gender differences were found with respect to alveolar notches, palatal cysts, alveolar cysts or lymphangioma.
5. Contradictory statements were given for term infants with respect to a correlation of birthweight or gender and palatal size at birth.
6. No significant differences between spontaneously and forceps delivered term infants have been described with respect to palatal size.
7. Contradictory statements were given concerning a correlation between nasal deformity and palatal symmetry, thus no conclusions concerning that subject can be drawn in this review.
The palate of the preterm/low birthweight infant
1. Orotracheal intubation has been reported to be harmful for teeth, tooth eruption, palatal shape and speech as early as 12 hours after intubation.
2. Due to a non-uniform definition and a subjective, non-metric evaluation in the majority of the studies there is a marked difference in the percentage data on the incidence of palatal grooving in PT infants (7 – 90%).
3. The following facts have been accused for provoking grooving: head flattening, pressure of an oral tube, pathologic or impeded tongue function and broadening of the alveolar ridges adjacent to the tube. Thickened palatine ridges may give a false impression of palatal height.
4. Metrically, the palates of intubated PT babies remain narrower, what has been examined up to the age of 11 years. Thus, an earlier orthodontic control of formerly orally intubated PT infants compared to non-intubated infants is advisable. From the orthodontic point of view, nasal intubation should be favoured.
5. Contradictory information is given in the literature on PT infants concerning
• the correlation of length of intubation time and amount of grooving.
• the duration of 'grooving' (which was examined up to the age of ten years).
• the incidence of crossbites compared to non-intubated babies.
• a possible difference in palatal asymmetry compared to non-intubated babies.
• palatal depth compared to non-intubated babies.
Thus, no conclusions are possible concerning those subjects on the base of this review.
6. It remained unclear, if gestation or birthweight of preterm infants were related to palatal height, due to confounding with intubation time.
7. Palatal plates have proven to protect palates with inserted tubes from deepening. Pressure dispersing pads for the head, however, did not have a significant impact on palatal height. It remained unclear, if changes in palatal width occurred due to pressure dispersing pads or due to oral feeding. There is a need of prospective studies to assess the infection rate and development of the tooth buds in children with protective plates.
8. PT children seem to have significantly less palatal cysts than term babies (be aware of different examination times for PT and term children and of the difficulty of an oral examination in a tiny infant!).
9. Up to the corrected age of 13.8 weeks, the palatal morphology of non-orally intubated PT infants does not differ from that of (probably) term infants. Beyond that time, no controlled long term studies comparing palatal dimensions of non-intubated PT children with those of non intubated term children are available. Thus, it cannot be excluded, that (e.g. as a consequence of functional impairment) PT infants do have a priori an altered palatal shape, which has been wrongly attributed to oral intubation.
10. No statement can be made on the development of biometric palatal data of term infants in the period from 1930 to present on the basis of the reviewed studies, as age groups were formed over several non-comparable months and data on the body height and weight of the probands were unfortunately lacking in most studies.
11. Further investigations in which the parents are also examined are needed to clarify the implication of genetic factors in the palatal configuration.
12. Parameters such as diet (breast milk versus PT formula), mode of feeding (bottle- versus breast-versus orogastric vs. nasogastric feeding), positioning, habits as well as biometric data and the influence of the mandible must be included more consistently in future studies than they have been to date.
13. Future studies should quote the product of palatal height and width in order to give a numerical expression of relative palatal height. As two palates with apparently different shapes may have an identical palatal index, the palatal length should also be included for a better three-dimensional understanding of palatal shape, too.
List of abbreviations
[PT] preterm infant, [BW] birthweight, [LBW] low birthweight, [NBW] normal birthweight, [VLBW] very low birthweight, [NBW] normal birthweight, [GA] gestational age, [GW] gestational weeks, [NS] not significant
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
AH designed the study, searched the databases, extracted the data, analyzed the results and wrote the manuscript. HR helped with study design, analysis and provided critical input in neonatal associated issues and revised the manuscript. UE and EH formulated the research question, helped with study design, analysis and in revising the manuscript. All authors read and approved the final manuscript.
Acknowledgements
We thank Fiona Lawson for the English language revision.
==== Refs
Paulsson L Bondemark L Soderfeldt B A systematic review of the consequences of premature birth on palatal morphology, dental occlusion, tooth-crown dimensions, and tooth maturity and eruption Angle Orthod 2004 74 269 279 15132456
Gorlin R Pindborg J Syndromes of the head and neck 1964 New York: Blakiston
Hall J Froster-Iskenius U Allanson J Handbook of normal physical measurements 1989 Oxford, New York, Toronto: Oxford University Press
Hanson JW Smith DW Cohen MMJ Prominent lateral palatine ridges: developmental and clinical relevance J Pediatr 1976 89 54 58 932903
Shapiro BL Redman RS Gorlin RJ Measurement of normal and reportedly malformed palatal vaults. 1. Normal adult measurements J Dent Res 1963 42 1039 14046509
Smith DW Lemli L Opitz JM A newly recognized syndrome of multiple congenital anomalies J Pediatr 1964 64 210 217 14119520
Marin-Padilla M Study of the skull in human cranioschisis Acta Anat (Basel) 1965 62 1 20 5864445
Solomon LM Medenica M Pruzansky S Kreiborg S Apert syndrome and palatal mucopolysaccharides Teratology 1973 8 287 291 4272080 10.1002/tera.1420080308
Leighton BC A preliminary study of the morphology of the upper gum pad at the age of 6 months Swed Dent J 1982 115 122
Rubenstein L Disscussion on: Nowak AJ, Casamassimo PS. Oral opening and other selected facial dimensions of children 6 weeks to 36 months of age J Oral Maxillofac Surg 1994 52 848
Nowak AJ Casamassimo PS Oral opening and other selected facial dimensions of children 6 weeks to 36 months of age J Oral Maxillofac Surg 1994 52 845 847 8040739
Macey-Dare LV Moles DR Evans RD Nixon F Long-term effect of neonatal endotracheal intubation on palatal form and symmetry in 8–11-year-old children Eur J Orthod 1999 21 703 710 10665200 10.1093/ejo/21.6.703
Baxter RJ Johnson JD Goetzman BW Hackel A Cosmetic nasal deformities complicating prolonged nasotracheal intubation in critically ill newborn infants Pediatrics 1975 55 884 887 1134889
Pettett G Merenstein GB Letter: Nasal erosion with nasotracheal intubation J Pediatr 1975 87 149 150 1151537
Molteni RA Bumstead DH Development and severity of palatal grooves in orally intubated newborns. Effect of 'soft' endotracheal tubes Am J Dis Child 1986 140 357 359 3953527
Jasmin JR Müller-Giamarchi M Dupont D Velin P Plaque palatine du nouveau-né prématuré Actual Odontostomatol (Paris) 1991 45 63 66 1853745
Stocks J Effect of nasogastric tubes on nasal resistance during infancy Arch Dis Child 1980 55 17 21 6769396
Berry FA Yemen TA Pediatric airway in health and disease Pediatr Clin North Am 1994 41 153 180 8295801
Miller MJ Carlo WA Strohl KP Fanaroff AA Martin RJ Effect of maturation on oral breathing in sleeping premature infants J Pediatr 1986 109 515 519 3746545
Von Gonten AS Meyer JB Kim AK Dental management of neonates requiring prolonged oral intubation J Prosthodont 1995 4 221 225 8601178 10.1111/j.1532-849X.1995.tb00344.x
Metz R Mickell J Unplanned extubation in pediatrics: incidence and related factors [abstract] Ped Res 1993 33 s211
Kopra DE Davis EL Prevalence of oral defects among neonatally intubated 3- to 5- and 7- to 10-year old children Pediatr Dent 1991 13 349 355 1843991
Erenberg A Nowak AJ Appliance for stabilizing orogastric and orotracheal tubes in infants Crit Care Med 1984 12 669 671 6430649
Kopra DE Creighton PR Buckwald S Kopra LF Carter JM The oral effects of neonatal intubation [abstract] J Dent Res 1988 67 s420
Seow WK Effects of preterm birth on oral growth and development Aust Dent J 1997 42 85 91 9153835
Moylan FM Seldin EB Shannon DC Todres ID Defective primary dentition in survivors of neonatal mechanical ventilation J Pediatr 1980 96 106 108 6985657
Seow WK Brown JP Tudehope DI O'Callaghan M Developmental defects in the primary dentition of low birth-weight infants: adverse effects of laryngoscopy and prolonged endotracheal intubation Pediatr Dent 1984 6 28 31 6592545
Angelos GM Smith DR Jorgenson R Sweeney EA Oral complications associated with neonatal oral tracheal intubation: a critical review Pediatr Dent 1989 11 133 140 2668901
Krous H Defective dentition following mechanical ventilation J Pediatr 1980 97 334 7400912
Boice JB Krous HF Foley JM Gingival and dental complications of orotracheal intubation JAMA 1976 236 957 958 988900 10.1001/jama.236.8.957
Seow WK Masel JP Weir C Tudehope DI Mineral deficiency in the pathogenesis of enamel hypoplasia in prematurely born, very low birthweight children Pediatr Dent 1989 11 297 302 2639324
Procter AM Lether D Oliver RG Cartlidge PH Deformation of the palate in preterm infants Arch Dis Child Fetal Neonatal Ed 1998 78 F29 32 9536837
Viscardi RM Romberg E Abrams RG Delayed primary tooth eruption in premature infants: relationship to neonatal factors [abstract] Pediatr Dent 1993 33 s1651
Funakoshi Y Kushida Y Hieda T Dental observations of low birth weight infants Pediatr Dent 1981 3 21 25 6951147
Mellander M Noren J Freden H Kjellmer I Mineralization defects in deciduous teeth of low birthweight infants Acta Paediatr Scand 1982 71 727 733 7180440
Rosenzweig KA Sahar M Enamel hypolpasia and dental caries in the primary dentition of premature Br Dent J 1962 113 279 280
Wetzel RC Defective dentition following mechanical ventilation J Pediatr 1980 97 334 7400913
Saunders BS Easa D Slaughter RJ Acquired palatal groove in neonates. A report of two cases J Pediatr 1976 89 988 989 993929
Erenberg A Nowak AJ Palatal groove formation in neonates and infants with orotracheal tubes Am J Dis Child 1984 138 974 975 6475859
Erenberg A Nowak AJ Palatal groove formation: A complication of orotracheal intubation Factors influencing orofacial development in the ill, preterm low birthweight and term neonate 1984 Iowa City: University of Iowa
Duke PM Coulson JD Santos JI Johnson JD Cleft palate associated with prolonged orotracheal intubation in infancy J Pediatr 1976 89 990 991 993930
Biskinis EK Herz M Acquired palatal groove after prolonged orotracheal intubation J Pediatr 1978 92 512 513 633007
Ginoza G Cortez S Modanlou HD Prevention of palatal groove formation in premature neonates requiring intubation J Pediatr 1989 115 133 135 2738781
Ginoza GW Cortez S Modanlou HD Prevention of palatal groove formation with prolongued orotracheal intubation in preterm infants [abstract] Ped Res 1989 25 s1276
Neal P Bull MJ Jansen RD Palatal groove secondary to oral feeding tubes J Perinatol 1985 5 42 53
Warwick-Brown MM Neonatal palatal deformity following oral intubation Br Dent J 1987 162 258 259 3472556 10.1038/sj.bdj.4806097
Fadavi S Punwani IC Vidyasagar D Adeni S Intraoral prosthetic appliance for the prevention of palatal grooving in premature intubated infants Clin Prev Dent 1990 12 9 12 2376108
Ash SP Moss JP An investigation of the features of the pre-term infant palate and the effect of prolonged orotracheal intubation with and without protective appliances Br J Orthod 1987 14 253 261 3481276
Morris KM Burns YR Reduction of craniofacial and palatal narrowing in very low birthweight infants J Paediatr Child Health 1994 30 518 522 7865266
Ginther WH Palatal grooves in orally intubated newborns Am J Dis Child 1986 140 1097 3766480
Behrstock B Ramos A Kaufman N Does prolonged oral intubation contribute to medical hypertrophy of the lateral palatine ridges and possibly to iatrogenic cleft palate? J Pediatr 1977 91 171 874659
Carrillo PJ Palatal groove formation and oral endotracheal intubation Am J Dis Child 1985 139 859 860 4036908
Molteni RA Bumstead DH Letter to: Ginther WH. Palatal grooves in orally intubated newborns Am J Dis Child 1986 140 1097
Clinch L Variations in the mutual relationships of the maxillary and mandibular gum pads in the newborn child Int J Orthod 1934 20 359 374
Kent SE Rock WP Nahl SS Brain DJ The relationship of nasal septal deformity and palatal symmetry in neonates J Laryngol Otol 1991 105 424 427 2072008
Leighton BC Morphologische Variationen der Alveolarbögen beim Neugeborenen Fortschr Kieferorthop 1976 37 8 14 10.1007/BF02169375
Klemke B Über Kieferform und Bisslage beim Neugeborenen Med Diss Bonn 1939
Neumann M Kieferbezügliche Untersuchungen und Messungen an Neugeborenen Med Diss Kiel 1953
Howell S Assessment of palatal height in children Community Dent Oral Epidemiol 1981 9 44 47 6941877
Fadavi S Adeni S Dziedzic K Punwani I Vidyasagar D The oral effects of orotracheal intubation in prematurely born preschoolers ASDC J Dent Child 1992 59 420 424 1491080
Fadavi S Adeni S Dziedzic K Punwani I Vidyasagar D Use of a palatal stabilizing device in prevention of palatal grooves in premature infants Crit Care Med 1990 18 1279 1281 2225899
Hohoff A Rabe H Ehmer U Harms E Orofacial development in preterm infants – a prospective longitudinal study [abstract] Ped Res 2001 49 s282
Hohoff A Rabe H Ehmer U Harms E Orofacial development in preterm infants – relationship to nutritional intake Ped Res 2001 50 s289
Morris KM Seow WK Burns YR Palatal measurements of prematurely born, very low birth weight infants: comparison of three methods Am J Orthod Dentofacial Orthop 1993 103 368 373 8480703
Kim Seow W Tudehope DI Brown JP O'Callaghan M Effect of neonatal laryngoscopy and endotracheal intubation on palatal symmetry in two- to five-year old children Pediatr Dent 1985 7 30 36 3857557
Knott VB Johnson R Height and shape of the palate in girls: a longitudinal study Arch Oral Biol 1970 15 849 860 5273406 10.1016/0003-9969(70)90157-3
Gray LP Deviated nasal septum. Incidence and etiology Ann Otol Rhinol Laryngol Suppl 1978 87 3 20 99070
Gray LP The development and significance of septal and dental deformity from birth to eight years Int J Pediatr Otorhinolaryngol 1983 6 265 277 6676334
Watterberg KL Munsick-Bruno G Incidence and persistence of acquired palatal groove in preterm neonates following prolongued orotracheal intubation [abstract] Clin Res 1986 34 s113
Watterberg KL Munsick-Bruno G Incidence and persistence of acquired palatal groove in preterm neonates following prolongued orotracheal intubation [abstract] Ped Res 1986 20 s1357
Raju TNK Say it again, Sam: republication of research data and meta-analysis [abstract] Ped Res 1993 33 s542
Tsang RC Donovan EF Steichen JJ Calcium physiology and pathology in the neonate Pediatr Clin North Am 1976 23 611 626 792775
Kjær I Keeling JW Fischer-Hansen B The prenatal human cranium-normal and pathologic development 1999 Copenhagen: Munksgaard
Sillman JH Relationship of maxillary and mandibular gum pads in the newborn infant Am J Orthodont Oral Surg 1938 24 409 424 10.1016/S0096-6347(38)90295-9
Bosma JF Bosma JF Prologue to the symposium Fourth symposium on oral sensation and perception 1973 Washington: Government Printing Office
|
16270912
|
PMC1298320
|
CC BY
|
2021-01-04 16:37:14
|
no
|
Head Face Med. 2005 Nov 2; 1:10
|
utf-8
|
Head Face Med
| 2,005 |
10.1186/1746-160X-1-10
|
oa_comm
|
==== Front
Head Face MedHead & Face Medicine1746-160XBioMed Central London 1746-160X-1-91627090910.1186/1746-160X-1-9ReviewPalatal development of preterm and low birthweight infants compared to term infants – What do we know? Part 2: The palate of the preterm/low birthweight infant Hohoff Ariane [email protected] Heike [email protected] Ulrike [email protected] Erik [email protected] Poliklinik für Kieferorthopädie, Universitätsklinikum, Westfälische Wilhelms-Universität, Münster, Germany2 Department of Neonatology, Brighton & Sussex University Hospitals, UK3 Klinik für Kinderheilkunde, Division of Neonatology, Universitätsklinikum, Westfälische Wilhelms-Universität, Münster, Germany2005 28 10 2005 1 9 9 8 9 2005 28 10 2005 Copyright © 2005 Hohoff et al; licensee BioMed Central Ltd.2005Hohoff 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
Well-designed clinical studies on the palatal development in preterm and low birthweight infants are desirable because the literature is characterized by contradictory results. It could be shown that knowledge about 'normal' palatal development is still weak as well (Part 1). The objective of this review is therefore to contribute a fundamental analysis of methodologies, confounding factors, and outcomes of studies on palatal development in preterm and low birthweight infants.
Methods
An electronic literature search as well as hand searches were performed based on Cochrane search strategies including sources of more than a century in English, German, and French. Original data were recalculated from studies which primarily dealt with both preterm and term infants. The extracted data, especially those from non-English paper sources, were provided unfiltered for comparison.
Results
Seventy-eight out of 155 included articles were analyzed for palatal morphology of preterm infants. Intubation, feeding tubes, feeding mode, tube characteristics, restriction of oral functions, kind of diet, cranial form and birthweight were seen as causes contributing to altered palatal morphology. Changes associated with intubation concern length, depth, width, asymmetry, crossbite, and contour of the palate. The phenomenon 'grooving' has also been described as a complication associated with oral intubation. However, this phenomenon suffers from lack of a clear-cut definition. Head flattening, pressure from the oral tube, pathologic or impaired tongue function, and broadening of the alveolar ridges adjacent to the tube have been raised as causes of 'grooving'. Metrically, the palates of intubated preterm infants remain narrower, which has been examined up to the age of the late mixed dentition.
Conclusion
There is no evidence that would justify the exclusion of any of the raised causes contributing to palatal alteration. Thus, early orthodontic and logopedic control of formerly orally intubated preterm infants is recommended, as opposed to non-intubated infants. From the orthodontic point of view, nasal intubation should be favored. The role that palatal protection plates and pressure-dispersing pads for the head have in palatal development remains unclear.
==== Body
Background
Compared to their term counterparts, prematurely born babies are at risk for postnatal growth and development defects. The general objective of neonatal care of premature infants is to support life and ensure a growth rate sufficient to fulfill the individual's genetic potential. Reaching this goal has undergone a dramatic improvement in the last two decades. Therefore, research into the development of these patients can and must now be extended to other areas, such as their physical and cognitive development. The morbidity potential associated with prematurity needs to be investigated to establish preventive measures. The orofacial region plays an important role in the infant's development. Premature babies must develop the skills needed to begin oral feeding as they reach an age that supports coordination of breathing and swallowing. This normality of oral functions, including nose breathing, induces adequate development of the whole orofacial region. During the time period from initiation of oral functions to full oral feeding in neonatal care, a complex interplay of various internal and external factors exerts an influence that may affect palatal development and lead therefore to a higher risk for malocclusions, facial asymmetries, and other late consequences. The evidence on these possible consequences is weak and the results of one century of research on palatal development are still controversial. This applies also to the knowledge gained on 'normal' palatal development of term babies (Part 1) which is a precondition to recognize abnormalities in the preterm infant's palate.
The objective of Part 2 of this review is therefore to contribute a fundamental analysis of methodologies, confounding factors, and outcomes of studies on palatal development of preterm and low birthweight infants.
Methods
The search strategy, the surveyed medical databases and sources of hand searches, and the assessment of included studies are described in Part 1. Exclusion criteria and excluded articles are listed in detail in Table 1 of Part 1 (see Additional file 1 of Part 1). The general methodologies used for morphological assessment could be divided into visual descriptions and metrical descriptions of the palatal configuration. To elucidate possible mediating and interactive effects on alterations of palatal development in preterm infants, the analysis of studies was ordered as follows.
• Visual descriptions of the palatal configuration of PT/LBW infants
- Incidence of high arched palates
- Incidence of grooving
- Palatal morphology in relation to intubation time
- Duration of intubation associated changes
- Palatal morphology in relation to birthweight
- Palatal morphology in relation to weight at the time of impression taking
- Palatal morphology and characteristics of the tube
- Feeding tubes
- Palatal morphology and tube position
- Palatal morphology and palatal plates
* Palatal morphology and feeding plates
* Palatal morphology and protective plates
- Palatal morphology and oral functions
- Palatal and alveolar cysts
- Influence of positioning on the orofacial development of PT infants
• Metric descriptions of the palatal configuration of PT/LBW infants
- Palates of non-intubated PT infants
- Influence of intubation on the palatal configuration of PT infants
* Length of palate
* Depth of palate
* Asymmetry of palatal depth
* Width of palate
* Asymmetry of palatal width
* Crossbite
* Palatal contour in relation to GA
* Grooving in relation to weight
* Grooving in relation to characteristics of the tube
* Palatal configuration in relation to duration of intubation
* Duration of intubation-associated changes
* Palatal morphology and palatal plates
* Influence of diet on the development of the palatal dimension
* Influence of feeding mode on the development of the palatal dimension
* Influence of cranial form on the palatal dimension
• Comparison of non-intubated PT/LBW and term infants' palatal measurements
Results and Discussion
Seventy-eight articles published between 1940 and 2000 were included in the analysis of descriptions of the palatal morphology and palatal development in preterm and low birthweight infants. The majority of the studies were published between 1980 and 1990.
Visual descriptions of the palatal configuration of PT/LBW infants
Incidence of high arched palates
Very high arched palate, but 'no palatal groove' was seen in 32% of 37 VLBW infants aged 9 to 75 months (72% out of the 37 infants had been intubated for on average 34.5 days) [1].
Incidence of grooving (Table 1, see Additional file 1)
Alveolar grooving [2], and 'palatal grooving' [2-13] have been described as a complication in connection with oral intubation. They never occurred in combination [2]. The various hypotheses on the cause of grooving are discussed in Part 3. The majority of articles dealing with the phenomenon fail to give a definition of palatal grooving. However, there are three exceptions:
1. Two authors defined a palatal groove as follows: 'Narrow channel of variable depth located near the midline of the palate as identified by visual inspection of the maxillary cast' [6] (Comment: Consider the variability of the term 'narrow').
2. Two other authors, performing intraoral measurement with a micrometer 'from its floor to the surface of the palate at the midpoint of the hard palate', selected a palatal groove of ≥0.5 cm arbitrarily as significant [14] (Comment: Consider how difficult it is to make precise intraoral measurements in a tiny infant).
3. A further group stated: 'By definition, a palatal groove is an architechtural deformity of the palate caused by external pressure from the orotracheal tube' [15].
Intubation does not invariably lead to grooving [16,17]. The incidence of palatal grooving in PT infants is quoted at 7 – 90% [6,14,16-18] (n. b. 'grooving' may also be a matter of thickened palatine ridges). Only a few cases of alveolar grooving have been reported [4,19,20]. The deepest of these alveolar grooves divided entirely the alveolar ridge [19]. No evidence concerning the resolution of the defects has been given.
Palatal morphology in relation to intubation time (Table 1, see Additional file 1)
Grooving may occur just 12 hours after intubation; the longer the intubation time, the greater the incidence of groove formation, with a percentage of 87.5% grooves with an intubation time of more than 15 days [6]. In six infants intubated 50 ≥ 89 days, no palatal deformity was detected [5]. Two case reports, however report large or deep palatal grooves after 70 – 95 days of intubation [12].
A significant correlation between severty of groove and length of intubation in a group of infants without protective plates was observed [7,8]. In one of the studies, however, a statistically significant difference among the examaniers was revealed [7], in the other the intra- and inter-examiner reliabilty is not given [8]. In contrast, Raval et al. [1] quoted that palatal arch morphology was not influenced by duration of ventilation (reliability of the method not given).
Duration of intubation associated changes
' ... It is unknown whether the palatal groove is permanent ... ' [6] (Table 1, see Additional file 1). While an almost complete resolution of a palatal groove after an unstated amount of time was reported in a letter [3], other authors did not find a noticable closure of a 'cleft' at four months of age [5] (Table 1, see Additional file 1).
Watterberg and Munsick-Bruno [16,17] observed grooves in 90% of PT infants at the time of extubation. 67% [16], respectively 70% [17] still displayed the grooving six months after extubation, while the other had high arched palates (mean intubation time 67.6 days). The study had a high drop-off rate (Tables 1 and 2, see Additional files 1 and 2).
In the experience of some authors, the prominence of the lateral palatine ridges recedes after extubation, with a normal tongue motion ensuing and the palate having a normal appearance by the age of 2 years [2,21,22].
In an infant that had previously been intubated for 4 months a normal palatal contour in the second year of life was observed, whereas the groove was found to have increased in size in another patient at 21 months [13] (Table 2, see Additional file 2).
By the age of 2 – 5 years, no grooves were observed in 31 previously intubated PT infants in the clinical section of a follow-up study [23] (Table 2, see Additional file 2).
In contrast, in children of the same age group palatal grooving, a high palatal vault and crossbite were found in 28%, 100% and 16%, respectively in another study [24]. The persistence of a palatal groove acquired in the neonatal period for as long as 5 years was observed [25] (Tables 1 and 2, see Additional files 1 and 2).
In 32% of 37 VLBW infants aged 9 – 75 months very high arched palates have been diagnosed (72% of the 37 infants had been intubated for an average of 34.5 days) [1] (Table 1, see Additional file 1).
In 27 four year old VLBW children an equal amount of crossbite (15%) compared to normal controls was found [26].
Palatal morphology in relation to birthweight
Comparing three studies [1,23,24] with inclusion of children with increasing birthweights the neonates in the study with the lowest mean birthweight had the highest incidence of palatal deformation, i.e. 37% very high arched palates [1], while the probands in the study with the highest birthweights had no palatal deformity (Table 3, see Additional file 3).
Palatal morphology in relation to weight at the time of impression taking
No correlation between weight at the time of impression-taking and the incidence of grooving has been reported [6]. No data was given on age, GA and BW of the infants (Table 1, see Additional file 1).
Palatal morphology and characteristics of the tube
In conclusion to two case reports on deep palatal grooves, Saunders et al. [12] stated: 'Although chemical contents of previously used endotracheal tubes have been shown to cause some tissue irritation, the tubes used were made of polyvinyl chloride ... and are unlikely to be the major cause of this injury' (i.e. the grooves) (Table 1, see Additional file 1).
In contrast, other authors argue: 'The rigidity of these PVC tubes is believed to be strongly related to the development of the grooves in the intubated neonates' [27] (Table 1, see Additional file 1).
Warwick-Brown [13] considered that 'the detrimental effect of the orotracheal tube versus orogastric tube may be a reflection of its increased size and rigidity and/or its use in the very early postnatal period' (Table 2, see Additional file 2).
Feeding tubes
Infants of 26 to 32 GW have neither a sucking nor a coordinated swallowing reflex, thus the normal pattern of feeding is impossible because of the risk of inhalation and asphyxia [28]. A nasogastric feeding tube is commonly used to feed PT infants, until the sucking reflex is observed at about 36 weeks gestation age, and some infants require tube feeding even after this point [29].
The neonatal infant is an obligatory nose-breather. In normal circumstances this is a favourable characteristic as it allows the child to suckle while breathing through the nose [28]. Nasogastric feeding tubes, however obstruct the nares; this increases both the danger of hypoxia and the work of breathing [29]. Orogastric tubes used for a period of up to 50 days induced no grooves in 94.3% of cases [6] (Table 1, see Additional file 1). Arens and Reichman [30] described grooving in three VLBW infants at 85, 65 and 65 days after insertion of no. 5 F polyethylene feeding tubes for 108, 75 or 65 days. Simultaneous oral feeding was performed for 15, 26 and 9 days. The grooves were still present at 12 months in 2, at 13 months in 1 infant (Table 1, see Additional file 1).
Palatal morphology and tube position
It has been postulated that the effect of the tube is dependent on its position and the developmental state of the palate-forming bone and that most likely the palatal groove forms due to the continuous pressure of the endotracheal tube against the median palatine suture [6]. Some authors therefore recommended shifting the tubes to one side to prevent grooving [5]. However, one study revealed that grooving occurred even with laterally positioned tubes. The author attributed this to insufficient tongue thrust against the palatal shelves, allowing the shelves to grow together [22]. Even a laterally positioned tube may exert pressure in the rear palatal area, also giving rise to grooving [31].
Palatal morphology and palatal plates
Palatal morphology and feeding plates
According to two case reports an acrylic feeding plate, which is sometimes used to fix the orogastric tube, cannot protect the palate against the earlier assault of an orotracheal tube [13] (Table 2, see Additional file 2).
Palatal morphology and protective plates
Passing tubes through the mouth causes discomfort to child, shown by an increase in movements of the jaw and tongue [28]. In order to stabilize oral ventilation or feeding tubes against displacement from tongue and jaw movements, and thus against accidental extubation, to remedy palatal narrowing or grooving and protect primary teeth from trauma caused by intraoral tubes denture like protective plates are recommended or used by various authors [2,7,8,13,15,24,27-30,32-35]. Such an oral plate has been recommended for any infant requiring an oral tube for more than 24 hours, since 12 hours was the shortest period for palatal groove formation (no information concerning size, depth or severety of that groove was reported) [27].
A 90% reduction of spontaneous extubation and a 100% succes in prevention of palatal groove formation was reported in 30 intubated preterm infants with protective plates; babies who are receive this appliance should be medically stable as determined by the attending neonatologist [15].
In a randomized study a prefabricated palatal stabilizing device was compared with an acrylic, custom-made palatal stabilizing device. In those 34 PT intubated children with the prefabricated device, the appliance turned out to be significantly less retentive, thus requiring a greater monitoring. Accidental extubation occurred significantly more often than in 36 intubated preterm infants with a custom-made appliance. Both groups were medically stable and did not differ statistically significant with respect to birth weight, gestational age and period of intubation [33].
The plates are fabricated in the dental laboratory after an impression is taken from the infants mouth. The act of inserting and seating the tray with the impression material is often associated with a slight increase in the heart rate as displayed on the ECG monitor, however, once the tray has been seated in place, the vital signs remain within normal limits [15,28]. To preserve the child in its favorable environment, impressions are taken while the baby is in the incubator. This complicates the procedure but has the advantage that the infant's well being can be monitored throughout the impression procedure. To make the impression taking process more secure and to avoid the danger of indigestion or aspiration of the impression material, it was recommend to insert the impression taking material with the tray into a condom prior to insertion into the mouth [34]. A covered 'chimney' in the acrylic plate corresponding to the size of the prospective tube may have the advantge of a sticking plaster to fix the tube, which can be the cause for dermal irritations to be unnecessary. On the other hand, this construction holds the plate in place without fixative cream being necessary [34]. Other authors retain the appliance in the mouth by means of adhesive powder [15].
Those plates, which should be secured by dental floss taped to the newborns cheek to prevent aspiration have proved capable of resisting a displacing force of 150 – 200 g, which is more than adequate to support two feeding tubes in position [28,29]. The device should be removed daily for cleaning [28,36] and can be readily manipulated by the neonatal nursing staff [27,28,36]. Care of the infant with such a device is not particularly different from care of babies without such a device [32]. Infants are reported to tolerate the procedure and the appliance well [28,29,35].
Some authors recommend to fabricate a new appliance after 2–3 weeks to allow for growth [29], others make a new plate approximately every 4 weeks [15,27]. Religning of the appliance at an interval between 10–20 days may extent its fit for up to 6–8 weeks [28,36].
In only one study, attention has been made to instances of ulceration or erythema to the palate [28]. Both could be denied. Neither was evidence found for monilial infection, probably because in these infants no food is taken by the mouth, so there is no possibility of food substances contaminating the palate [28]. A retrospective review of the infection rates showed no significant increase in the incidence of nosocomial infections during the period palatal stabilizing devices were used [15].
Preliminary investigations suggested that the appliance may have an important part to play in reducing the breathing effort necessary in 'at risk' premature infants and in lowering the arterial CO2 levels in hypercapnic children [28]. Regrettably, no studies concerning this subject were found during the literature searching process.
Palatal morphology and oral functions
Some orally intubated infants suck energetically at the tubes [5,37], thus shaping the oral tissues to the insertion direction of the tube in addition. This hypothesis is supported by the finding that in term infants a sucking habit narrows the palatal width significantly [38].
In PT infants an aberrant feeding pattern was observed, which might be the cause or else the consequence of a change in the palatal morphology of PT infants [39,40].
Palatal and alveolar cysts
The prevalence of palatal cysts is significantly lower in the PT infant (9%) (examined 12 days after birth) than in the term infant (30%) (examined 1 day after birth), as is the prevalence of maxillary anterior alveolar cysts (PT 27% vs. term 58%); palatal and maxillary alveolar cysts increase with increasing gestational age, post-natal age and birthweight; no significant differences were found in the prevalence of palatal and alveolar cysts for gender, nor while comparing caucasian preterm infants with a group of non-caucasian infants rising from black, latino and indian children [41].
Influence of positioning on the orofacial development of PT infants
It was stated that ' ... Positioning and gravitational forces may interrupt or cause deviation in the development of palatal, cranial and facial bones' [42]. 'The effect of these changes tends to alter the facial appearance of a child' [43,44].
Metric descriptions of the palatal configuration of PT/ LBW infants
Twelve metrical studies dealing exclusively with PT or LBW infants' palates were found [10,14,23,32,37,42,45-50] (Table 4, see Additional file 4). One had the exactness of different measuring methods as the primary interest of outcome [42]; three examined the effect of protective appliances [32,37,49], four included preschool or school children of a wide age range [10,23,45,48] (n. b. the mean difference in palatal width from 9 – 12 years in girls has been reported to be 0.9 mm in the molar region [51]); one measured palatal depths intraorally, entailing the risk of being imprecise [14]; a further study included term and PT infants [50]. In the majority of studies a problem with the reliabilty of the measuring method was present: Either the reliabilty was not given [14,46-48,50] or a significant measuring error for palatal depth was recorded [37], or the coefficent of variation for repeated palatal height measurements ran up to 11.73% [42,49].
Additionally to the above mentioned studies, the authors of the review recalculated data given in two doctoral thesis, and therefore were able to 'extract' figures concerning preterm infants from studies which primaraly included both, preterm and term infants [52,53] (Table 5, see Additional file 5; Part 1: Table 3, see Additional file 3 of Part 1). The measuring method and the reliability of the method was not given in either of the two latter studies. For the above mentioned reasons, criteria for a systematic analysis was not applicable to the retrieved publications.
Palates of non-intubated PT infants
Grooving with humping-up of the lateral palatal margins seen in infants following orotracheal intubation was not observed in any non-intubated infant [37] (Table 4, see Additional file 4).
Recalculation of the data given by Neumann [53], including exclusively spontaneously delivered children (occipito-anterior vertex presentation) revealed no significant difference in palatal width of preterm infants with respect to gender (Table 5, see Additional file 5; Part 1: Table 3, see Additional file 3 of Part 1). A comparison of palatal depth measurements at 36/37.6 and 53/53.8 weeks postmenstrual age reveals a lower palatal depth in non-intubated children [47] compared with intubated children [50] (Table 4, see Additional file 4). This difference between intubated and non-intubated is even more pronounced than expressed by the comparison of the pure figures, taking into account, that in the latter study palatal depth was measured from the lateral alveolar ridges, which are 'lower' than the alveoar crests, from where the measurements of the former study were conducted.
For 6 LBW children which were included in the study of Klemke [52] the authors of this review found a significant correlation between maximum palatal width and body length (Pearson, one sided, p = .009).
Influence of intubation on the palatal configuration of PT infants
Length of palate
Anterior palatal length (measured between the midpoint between the junctions of the lateral grooves with the gingival grooves and the anterior midline point on the alveolus) and maximum palatal length (distance between the midline point on the anterior part of the alveolus and posterior limit of the gingival grooves) were similar for non-intubated and intubated PT infants until term. The presence of prolonged intubation thus had little effect on the increase in length of the preterm palate [37]. The study ended at term.
Depth of palate
One study was excluded for that point, as a significant error of the method has been described by the authors for palatal depths measurements [37] (Tables 4 and 5, see Additional files 4 and 5).
The results of the literature research with respect to palatal depth were heterogeneous (Tables 6 and 7, see Additional files 66 and 7). Procter et al. [50] found only a small and transient effect of oral intubation on palatal depth, which disappeared at term (only 4 infants were intubated > 10 days, term infants were included in that study) (Tables 4 and 5, see Additional files 4 and 5). Visual inspection of the casts in that study revealed that palatal grooving did not always correspond with relative palate depth, but did usually occur in intubated infants. Procter et al. [50] therefore concluded that palatal grooving is not caused by the direct pressure of the orotracheal tube but is more likely to be due to overgrowth of the lateral palatine ridges. Whether the cause is irritation by the tube or the impairment of a normal tongue function could not be clarified within the framework of their study. At extubation time, the incidence and severity of grooves was found to be closely related to BW and total intubation time (mean intubation time > 15 days) [14] (Table 1, see Additional file 1).
Fadavi et al. [32] reported the deepest indentations of the palate in children, who had been intubated for more than 30 days; the prevalence of oral defects increased with increasing intubation time as with decreasing BW and significantly greater palatal depths were recorded in 2 – 5 year old, formerly orally intubated children (mean intubation time 36 days) [45] (Tables 4 and 6, see Additional files 4 and 6). Significantly higher palatal vaults and grooved palates in 3 – 5 and 7 – 10 years old formerly intubated PT children compared to non-intubated term children were also described by other authors (mean intubation time 18.3 and 26.4 days, respectively) [48] (Tables 2, 6, 7, see Additional files 2, 6, 7). Significantly higher palatal depths in the anterior region of formerly intubated children at a mean age of ten were measured (mean intubation time 15.2 days) [10] (Tables 4 and 7, see Additional files 4 and 7). Seow et al. [23] were the only group of authors who did not report any indentations of the palate in association with orotracheal intubation, the intubation time of the sample of that study was the shortest (Table 6, see Additional file 6).
Asymmetry of palatal depth
The only study on that point was excluded for that subject due to contradictory statements in text and tables [48] (Table 2, see Additional file 2).
Width of palate
Palatal width of intubated PT infants was reported to be significantly smaller in comparison to non-intubated PT infants from 32 weeks to term at the lateral grooves (mean intubation time > 30 days, study ended at term) [37] (Tables 2, 4, 5, see Additional files 2, 4, 5), and also in comparison to non-intubated term infants; the latter was true for the deciduous (Table 8, see Additional file 8) as well as for the mixed dentition (Table 9, see Additional file 9): Kopra and Davis [48] reported, that at ages 3 – 5 and 7 – 10 years the palates of intubated PT children were significantly smaller compared to age matched controls (mean orotracheal intubation times 18.3 and 26.4 days, respectively; mean orogastric intubation times 55.6 and 52.4 days, respectively) (Tables 2, 8, 9, see Additional files 2, 8, 9). This is confirmed by another study, in which the palates of formely intubated children at a mean age of ten years were significantly narrower, but only in the region of the second deciduous molars and first permanent molars (mean intubation time 15.2 days) [10] (Tables 4 and 9, see Additional files 4 and 9). Only a small and transient rise of palatal index (depth/width ratio), disappearing at term in children intubated for ≥ 10 days was described in another paper, (only 4 infants were intubated > 10 days, term infants were included in that study) [50] (Tables 4 and 5, see Additional files 4 and 5).
Asymmetry of palatal width
From the deciduous front teeth up to the first primary molar no palatal asymmetry was proven in 2 – 5 year old formerly orally intubated LBW children (mean intubation time not given) [23]. For the mentioned frontal region, this has been proven for 8 – 11 year old infants [10] (Table 4, see Additional file 4). In the region of the second deciduous and first permanent molars two other studies showed significantly more asymmetric palates of formerly orally intubated children compared to non intubated, normal controls with respect to palatal width proportional asymmetry for 3 – 5 year olds [48] and for 8 – 11 year olds [10] (Tables 2 and 4, see Additional files 2 and 4). In contrast to the latter authors, the former could not find significant differences in palatal width proportional asymmetry between 7 – 10 year old intubated and non-intubated children.
Crossbite
Tables 8 and 9 (see Additional files 8 and 9) show the crossbite frequency in the deciduous and mixed dentitions, respectively. Literature research revealed, that the crossbite frequency of intubated PT children did not differ significantly in all studies from that of term, non intubated controls: One study failed to show a significant difference with respect to crossbite in 2 – 5 year old, formerly orally intubated VLBW and LBW children compared to NBW controls (mean intubation time 36 days) [45] (Table 4, see Additional file 4). In contrast, significantly more crossbites in 3 – 5 year and 7 – 10 year old formerly orally intubated children were diagnosed compared to non-intubated controls by another research group (mean intubation time 24.6 days) [48] (Tables 2, 8, 9, see Additional files 2, 8, 9).
Palatal contour in relation to GA
With the same postmenstrual age up to < 40 weeks, relative palate depth tended to be higher in less mature children, but those were in fact the children with the highest percentage and duration of intubation. Depth and width of the palate were related to gestation and postmentrual age, with the most mature babies having the largest palates, but gestation had no effect on palatal index, i.e. the depth/width ratio [50] (Tables 4 and 5, see Additional files 4 and 5). Bias could have come over this study ≥ 40 weeks of gestation, as term infants had been included. Analysis of variance revealed significant relationships between high vaulted palate, palatal grooving, and gestation [48] (Table 2, see Additional file 1).
Grooving in relation to weight
Three studies claim that the development of grooving is closely tied to BW [14,45,48] (Tables 1, 2, 4, 5, see Additional files 1, 2, 4, 5). N. b. the confounding that probably the most immature infants need the longest intubation.
Grooving in relation to characteristics of the tube
The incidence and severity of grooving are not reported to be related to the consistence of the tube; even the use of soft tubes can thus not reduce the incidence and extent of palatal grooving [14]. Although the authors detected no difference in the incidence of commonly recognized complications of endotracheal intubation when hard and soft endotracheal tubes were compared, the flexibility of the soft tubes occasionally entailed a prolonged intubation time (Table 1, see Additional file 1).
Palatal configuration in relation to duration of intubation
Molding of the gum pad can be seen within hours of intubation [37] (Tables 2 and 4, see Additional files 2 and 4). The incidence and severity of grooving have been reported to be closely related to the total intubation time [14]. The authors invariably observed grooving in children below 1000 g bodyweight after 7 days intubation. No neonate developed a palatal groove when mechanical ventilation was continued for seven days or less (Table 1, see Additional file 1). In a letter referring to that paper, other authors [54] express their surprise, that the number of infants in whom palatal grooves developed who were intubated less than seven days was so low. The reply was that the low incidence was most likely tied to the older mean gestational age of the group ventilated for less than seven days [55].
A high palatal vault was recorded in 69% and palatal grooving in 25% of 2 to 5 year-old children, both parameters increasing with longer duration of intubation [45] (Table 4, see Additional file 4). Others found palatal height not to be affected by length of intubation [10,49], nor palatal width and area [49] (Table 4, see Additional file 4).
According to Procter et al. [50] prolonged intubation > 10 days only leads to a small and transient increase in relative palatal height (i.e. palatal index, i.e. depth/ width ratio) until term (Table 4, see Additional file 4). Bias could have come over this study, as only max. n = 4 infants had been intubated ≥ 10 days, and term and NBW infants had been included.
At the age of 2 – 5 years, no influence of duration of intubation on palatal symmetry could be proven from the frontal region up to the region of the first deciduous molars [56] (Tables 2 and 3, see Additional files 2 and 3). Accordingly, no differences between children aged 8 – 11 which had been intubated as neonates for ≤ 15 days and children intubated > 15 days with respect to palatal width asymmetry were recorded [10] (Table 4, see Additional file 4).
Duration of intubation-associated changes
Once the tube was removed, the gum pad remolded in most instances in one study [37] (Tables 2 and 4, see Additional files 2 and 4). At term no more influence of intubation on palatal depth was recorded in another study [50] (Table 4, see Additional file 4). Bias could have come over this study, as only max. n = 4 infants had been intubated ≥ 10 days, and term and NBW infants had been included.
In children aged 2 – 5 years in contrast, significantly greater palatal depths were revealed in orally intubated PT infants compared to age-matched NBW children [45]. Twenty fife percent of the children still had severe to moderate grooves (2/52 < 3 mm, 6/52 = 3 – 5 mm, 5/52 > 5 mm), and 69% very deep to deep palatal vaults. Only 25% had normal, and 6% flat or shallow palates [45] (Table 4, see Additional file 4). In contrast, in the same age group, no palatal grooves and no palatal asymmetry were detected in an intubated group compared with a non-intubated group with repect to reference points at the gingival margins from the central up to the first primary molars [23] (Table 4, see Additional file 4). No data was given as to how many of the previously intubated infants had grooves originally. The authors hypothesized: 'Growth changes and remodeling of the palate and alveolus in the first few years of life probably correct any deformation caused by laryngoscopy and endotracheal intubation. That growth changes can allow for remodeling of the palate is seen in patients' thumb or finger sucking habits. Most cases of uncomplicated palatal deformities resulting from digit sucking are resolved once the habit is discontinued' (Table 4, see Additional file 4).
In 3 – 5 year old, formerly intubated PT children the following significant differences were found compared to age matched, normal controls: smaller palatal widths as well as a higher incidence of high vaulted palates and of palatal width proportional asymmetry in the molar region. No differences were described, however with respect to palatal depth and palatal width asymmetry [48] (Table 2, see Additional file 2). In 4 year old very low birthweight children an equal number (15%) of crossbites compared to NBW controls was diagnosed [26]. In 7 – 10 year old, formerly intubated preterm children, a significantly greater prevalence of a high palatal vault, grooved palate and crossbites was described, as significantly smaller palatal widths. Again, palatal depth did not differ significantly among the groups, neither did palatal width, nor palatal width proportional asymmetry [48] (Table 2, see Additional file 2).
In 8 – 11 year old children no significant differences in mean dental arch widths in the canine and molar region were measured while comparing formely intubated PT children and non intubated controls matched for age and gender [10] (Table 4, see Additional file 4). Like Seow et al. [23] (Tables 2 and 3, see Additional files 2 and 3), the former authors did not detect any significant differences in palatal asymmetry from the centrals to the first primary deciduous molars between intubated and non-intubated children. However, in gingival reference points located more distally (second deciduous molars and first permanent molars) they found significantly smaller palatal widths in formerly intubated vs. non intubated children (n. b. dental arch widths are measured at dental reference points, where buccal tipping of the tooth crown could compensate for a small transvere maxillary dimension, this buccal tipping could also compensate crossbites; palatal widths are measured at gingival reference points.). The left side of the palate in those intubated children was significantly wider than the right side posteriorly at the level of the second primary molars and first permanent molars [10] (Table 4, see Additional file 4). This could happen, because true stabilization of the endotracheal tube is only achieved extra-orally [57] and leaves open the possibilty for intra-oral tube displacement posteriorly towards the side of the prone nursing position, which is often on the right to aid gastric emptying. Prone sleep position has been shown to promote dolichocephyly and cranial moulding on the side on which the infant is nursed [58]. Macey-Dare et al. [10] also found the palatal heights anteriorly to be significantly steeper, but only at the level of the incisor region. Posterior crossbite or crossbite tendency was detected in only 11% of the formerly intubated children, what is comparable to that reported for the general population (7 – 10%) [59].
Palatal morphology and palatal plates
Anterior palatal length (midpoint between the junctions of the lateral grooves with the gingival grooves and the anterior midline point on the alveolus) and maximum palatal length (distance between the midline point on the anterior part of the alveolus and posterior limit of the gingival grooves) were similar for non-intubated and intubated infants as well as for intubated infants with protective plates [37]. A significant difference between a no-plate and a plate group in the percentage changes in lateral growth was recorded only for the anterior part of the palate at the lateral grooves, but not for the posterior region. There was a trend of palates of intubated children to be deeper, which was, however not significant. A significant error in the method of palatal depth determination was recorded in the study, which ended at term [37] (Table 4, see Additional file 4).
Palatal plates are able to protect intubated children from grooving [32] (Table 1, see Additional file 1). The effect of protective plates may consist in stabilization of the palate against deforming forces by the mattress rather than in keeping intraoral tubes in place [49]. However, only a poor correlation between cranial index and palatal index (depth/width ratio) was found [50].
Some authors see in the use of plates only a minimally protective effect: in their study population the mean difference in palate depth at 32 weeks between the intubated and non-intubated infants was 0.35 mm (only 4 infants had been intubated for ≥ 10 days) [50]. In comparison, the mean difference in palate depths at 32 weeks between protected and unprotected palates in the study by Ash and Moss [37] was 0.21 mm acording to Procter et al. [50], a statement which is incorrect: The mean difference in palate depth at 32 weeks between protected and unprotected palates in the study by Ash and Moss [37] at 32 weeks was 0.47 mm; furthermore, mean intubation time was longer in the latter study.
Influence of diet on the development of the palatal dimension
It has been reported that PT infants receiving human milk have lower bone mineralization rates than commercial formula fed infants [60]. This may lead to the hypothesis that commercial formula fed infants have an advantage in craniofacial and palatal bone growth and in resistance to postural deformation. However, any significant differences between breast- and formula fed children with respect to palatal width and depth [46] have been refuted (Table 4, see Additional file 4).
Influence of feeding mode on the development of the palatal dimension
Major benefits for growth in palatal width and palatal area followed the introduction of oral feeding [49]. The authors see the cause in the forming effect of the tongue on the shape of the palate during oral feeding. However, they also observed in PT infants an aberrant feeding pattern which they associated with palatal deformations.
Influence of cranial form on the palatal dimension
PT and LBW infants often have an unusually long, narrow head (dolicocephaly) compared with full term babies [43,61-63]. To prevent cranial deformation and thus any associated narrowing of the palate, the use of foam pressure dispersing pads (PDP) positioned on either side of the infant's head was recommended [49]. In comparison with a control group without pdp, a significantly greater change in the temporomandibular diameter was observed in the PDP group only in the early postnatal period prior to the start of oral feeding, but a significantly larger increase in palate surface and width only after the start of oral feeding. No significant intergroup differences were registered in the change in palatal height.
Although the cranial index (occipitofrontal : biparietal diameter; the greater the cranial index, the flatter the head) showed that the heads of the most immature infants were flattest and thus displayed side to side flattening, no significant correlation was detected between palatal index and cranial index [50]. Thus, the authors conclude that external pressure on the side of the head which causes head flattening cannot contribute to palatal grooving. Lateral head x-rays in children born small for gestational age showed a short anterior cranial base and small maxilla in a retrognathic face [64].
List of abbreviations
[PT] preterm infant, [BW] birthweight, [LBW] low birthweight, [NBW] normal birthweight, [VLBW] very low birthweight, [NBW] normal birthweight, [GA] gestational age, [GW] gestational weeks, [NS] not significant
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
AH designed the study, searched the databases, extracted the data, analyzed the results and wrote the manuscript. HR helped with study design, analysis and provided critical input in neonatal associated issues and revised the manuscript. UE and EH formulated the research question, helped with study design, analysis and in revising the manuscript. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
Table 1 Incidence or severity of palatal grooving, relation to intubation time.
Click here for file
Additional File 2
Table 2 Intubation-associated changes in palatal configuration and length.
Click here for file
Additional File 3
Table 3 Influence of birthweight on palatal morphology of preterm / low birthweight infants.
Click here for file
Additional File 4
Table 4 Measurements on palatal dimension of preterm infants.
Click here for file
Additional File 5
Table 5 Crosstables for palatal measurements: preterm (LBW) vs. term (NBW).
Click here for file
Additional File 6
Table 6 Metrical studies with respect to vertical palatal dimensions of intubated PT infants (deciduous dentition).
Click here for file
Additional File 7
Table 7 Metrical studies with respect to vertical palatal dimensions of intubated PT infants (mixed dentition).
Click here for file
Additional File 8
Table 8 Metrical studies with respect to transverse palatal dimensions of intubated PT infants (deciduous dentition).
Click here for file
Additional File 9
Table 9 Metrical studies with respect to transverse palatal dimensions of intubated PT infants (mixed dentition).
Click here for file
Acknowledgements
We thank Fiona Lawson for the English language revision.
==== Refs
Raval D Loevy H Rastogi A Anyebuno M Pildes RS Teeth eruption patterns and palatal arch morphology in vlbw infants ≤ 1250 g [abstract] Ped Res 1989 25 s801
Angelos GM Smith DR Jorgenson R Sweeney EA Oral complications associated with neonatal oral tracheal intubation: a critical review Pediatr Dent 1989 11 133 140 2668901
Biskinis EK Herz M Acquired palatal groove after prolonged orotracheal intubation J Pediatr 1978 92 512 513 633007
Boice JB Krous HF Foley JM Gingival and dental complications of orotracheal intubation JAMA 1976 236 957 958 988900 10.1001/jama.236.8.957
Duke PM Coulson JD Santos JI Johnson JD Cleft palate associated with prolonged orotracheal intubation in infancy J Pediatr 1976 89 990 991 993930
Erenberg A Nowak AJ Palatal groove formation in neonates and infants with orotracheal tubes Am J Dis Child 1984 138 974 975 6475859
Ginoza G Cortez S Modanlou HD Prevention of palatal groove formation in premature neonates requiring intubation J Pediatr 1989 115 133 135 2738781
Ginoza GW Cortez S Modanlou HD Prevention of palatal groove formation with prolongued orotracheal intubation in preterm infants [abstract] Ped Res 1989 25 s1276
Kopra DE Creighton PR Buckwald S Kopra LF Carter JM The oral effects of neonatal intubation [abstract] J Dent Res 1988 67 s420
Macey-Dare LV Moles DR Evans RD Nixon F Long-term effect of neonatal endotracheal intubation on palatal form and symmetry in 8–11-year-old children Eur J Orthod 1999 21 703 710 10665200 10.1093/ejo/21.6.703
Neal P Bull MJ Jansen RD Palatal groove secondary to oral feeding tubes J Perinatol 1985 5 42 53
Saunders BS Easa D Slaughter RJ Acquired palatal groove in neonates. A report of two cases J Pediatr 1976 89 988 989 993929
Warwick-Brown MM Neonatal palatal deformity following oral intubation Br Dent J 1987 162 258 259 3472556 10.1038/sj.bdj.4806097
Molteni RA Bumstead DH Development and severity of palatal grooves in orally intubated newborns. Effect of 'soft' endotracheal tubes Am J Dis Child 1986 140 357 359 3953527
Fadavi S Punwani IC Vidyasagar D Adeni S Intraoral prosthetic appliance for the prevention of palatal grooving in premature intubated infants Clin Prev Dent 1990 12 9 12 2376108
Watterberg KL Munsick-Bruno G Incidence and persistence of acquired palatal groove in preterm neonates following prolongued orotracheal intubation [abstract] Clin Res 1986 34 s113
Watterberg KL Munsick-Bruno G Incidence and persistence of acquired palatal groove in preterm neonates following prolongued orotracheal intubation [abstract] Ped Res 1986 20 s1357
Spitzer AR Fox WW Postextubation atelectasis-the role of oral versus nasal endotracheal tubes J Pediatr 1982 100 806 810 7069547
Krous H Defective dentition following mechanical ventilation J Pediatr 1980 97 334 7400912
Saunders BS Easa D Slaughter RJ Does prolongued oral intubation contribute to medical hypertrophy of the lateral palatine ridges and possibly to iotrogenic cleft palate? [Letter] J Pediatr 1978 702249
Behrstock B Ramos A Kaufman N Does prolonged oral intubation contribute to medical hypertrophy of the lateral palatine ridges and possibly to iatrogenic cleft palate? J Pediatr 1977 91 171 874659
Carrillo PJ Palatal groove formation and oral endotracheal intubation Am J Dis Child 1985 139 859 860 4036908
Kim Seow W Tudehope DI Brown JP O'Callaghan M Effect of neonatal laryngoscopy and endotracheal intubation on palatal symmetry in two- to five-year old children Pediatr Dent 1985 7 30 36 3857557
Adeni S Fadavi S Dziedzic K Vidayasager D Punwani I Prevalence of oral abnormalties in children following neonatal intubation Pediatr Res 1990 27 s1411
Adeni S Fadavi S Dziedzic K Defects in primary dentition following neonatal intubation [abstract] Clin Res 1989 37 s961
Lai PY A longitudinal controlled study of developmental enamel defects in preterm children MDSc thesis 1994 University of Qeensland, Brisbane
Erenberg A Nowak AJ Appliance for stabilizing orogastric and orotracheal tubes in infants Crit Care Med 1984 12 669 671 6430649
Sullivan PG An appliance to support oral intubation in the premature infant Br Dent J 1982 152 191 195 7044395 10.1038/sj.bdj.4804784
Sullivan PG Haringman H An intra-oral appliance to stabilise orogastric tube in premature infants Lancet 1981 1 416 417 6110044 10.1016/S0140-6736(81)91795-5
Arens R Reichman B Grooved palate associated with prolonged use of orogastric feeding tubes in premature infants J Oral Maxillofac Surg 1992 50 64 65 1727465
Nowak AJ Erenberg A Palatal groove formation in neonates and infants with orotracheal tubes Am J Dis Child 1985 139 860
Fadavi S Adeni S Dziedzic K Punwani I Vidyasagar D Use of a palatal stabilizing device in prevention of palatal grooves in premature infants Crit Care Med 1990 18 1279 1281 2225899
Fadavi S Punwani I Vidyasagar D Use of the Pala-nate device in the prevention of palatal grooves in premature, intubated infants Pediatr Crit Care Med 2000 1 48 50 12813286 10.1097/00130478-200007000-00009
Jasmin JR Müller-Giamarchi M Dupont D Velin P Plaque palatine du nouveau-né prématuré Actual Odontostomatol (Paris) 1991 45 63 66 1853745
Von Gonten AS Meyer JB Kim AK Dental management of neonates requiring prolonged oral intubation J Prosthodont 1995 4 221 225 8601178 10.1111/j.1532-849X.1995.tb00344.x
Collins G Windley HW Arnold RR Offenbacher S Effects of a Porphyromonas gingivalis infection on inflammatory mediator response and pregnancy outcome in hamsters Infect Immun 1994 62 4356 4361 7927695
Ash SP Moss JP An investigation of the features of the pre-term infant palate and the effect of prolonged orotracheal intubation with and without protective appliances Br J Orthod 1987 14 253 261 3481276
Leighton BC A preliminary study of the morphology of the upper gum pad at the age of 6 months Swed Dent J 1982 115 122
Burns Y Rogers Y Neil M Development of oral function in pre-term infants Physiother Pract 1987 3 168 178
Woolridge MW The 'anatomy' of infant sucking Midwifery 1986 2 164 171 3643397 10.1016/S0266-6138(86)80041-9
Donley CL Nelson LP Comparison of palatal and alveolar cysts of the newborn in premature and full-term infants Pediatr Dent 2000 22 321 324 10969441
Morris KM Seow WK Burns YR Palatal measurements of prematurely born, very low birth weight infants: comparison of three methods Am J Orthod Dentofacial Orthop 1993 103 368 373 8480703
Baum JD Searls D Head shape and size of pre-term low-birthweight infants Dev Med Child Neurol 1971 13 576 581 5119914
Cartlidge PH Rutter N Reduction of head flattening in preterm infants Arch Dis Child 1988 63 755 757 3415321
Fadavi S Adeni S Dziedzic K Punwani I Vidyasagar D The oral effects of orotracheal intubation in prematurely born preschoolers ASDC J Dent Child 1992 59 420 424 1491080
Hohoff A Rabe H Ehmer U Harms E Orofacial development in preterm infants – a prospective longitudinal study [abstract] Ped Res 2001 49 s282
Hohoff A Rabe H Ehmer U Harms E Orofacial development in preterm infants – relationship to nutritional intake Ped Res 2001 50 s289
Kopra DE Davis EL Prevalence of oral defects among neonatally intubated 3- to 5- and 7- to 10-year old children Pediatr Dent 1991 13 349 355 1843991
Morris KM Burns YR Reduction of craniofacial and palatal narrowing in very low birthweight infants J Paediatr Child Health 1994 30 518 522 7865266
Procter AM Lether D Oliver RG Cartlidge PH Deformation of the palate in preterm infants Arch Dis Child Fetal Neonatal Ed 1998 78 F29 32 9536837
Knott VB Johnson R Height and shape of the palate in girls: a longitudinal study Arch Oral Biol 1970 15 849 860 5273406 10.1016/0003-9969(70)90157-3
Klemke B Über Kieferform und Bisslage beim Neugeborenen Med Diss Bonn 1939
Neumann M Kieferbezügliche Untersuchungen und Messungen an Neugeborenen Med Diss Kiel 1953
Erenberg A Nowak A Palatal grooves in orally intubated newborns Am J Dis Child 1986 140 973 974 3752028
Molteni RA Bumstead DH Letter to: Erenberg A, Nowak A. Palatal grooves in orally intubated newborns Am J Dis Child 1986 140 974 3752029
Seow WK Effects of preterm birth on oral growth and development Aust Dent J 1997 42 85 91 9153835
Pettett G Merenstein GB Letter: Nasal erosion with nasotracheal intubation J Pediatr 1975 87 149 150 1151537
Updike C Schmidt RE Macke C Cahoon J Miller M Positional support for premature infants Am J Occup Ther 1986 40 712 715 3777108
Björk A Krebs A Solow B A method for epidemilogical registration of malocclusion Acta Odont Scand 1964 22 27 41 14158468
Chan GM Mileur L Hansen JW Effects of increased calcium and phosphorous formulas and human milk on bone mineralization in preterm infants J Pediatr Gastroenterol Nutr 1986 5 444 449 3723265
Elliman AM Bryan EM Elliman AD Starte D Narrow heads of preterm infants – do they matter? Dev Med Child Neurol 1986 28 745 748 3545952
Marsden DJ Reduction of head flattening in pre-term infants Dev Med Child Neurol 1980 22 507 509 7190946
Rutter N Hinchliffe W Cartlidge PH Do preterm infants always have flattened heads? Arch Dis Child 1993 68 606 607 8323368
Van Erum R Mulier M Carels C de Zegher F Short stature of prenatal origin: craniofacial growth and dental maturation Eur J Orthod 1998 20 417 425 9753823 10.1093/ejo/20.4.417
Huddart AG Bodenham RS Evaluation of arch form and occlusion in unilateral cleft palate subjects Cleft Palate J 1972 9 194 209 4505892
Lebret L Der menschliche Gaumen: Sein Wachstum, die auf ihn bezogene Wanderung der Seitenzähne, seine Expansion bei Anwendung zweier verschiedener orthodontischer Behandlungsmethoden Fortschr Kieferorthop 1966 27 121 140 5233533 10.1007/BF02163901
Ashley-Montagu MF The form and dimension of the palate in the newborn Int J Orthod Dent Child 1934 20 694 827 10.1016/S0097-0522(34)90177-5
Bakwin H Bakwin R Form and dimension of the palate during the first year of life Int J Orthod 1936 22 1018 1024
Clinch L Variations in the mutual relationships of the maxillary and mandibular gum pads in the newborn child Int J Orthod 1934 20 359 374
Deffez JP Plante P The maxilla of infants Rev Stomatol Chir Maxillofac 1976 77 403 408 1066765
Dittrich I Untersuchungen über Form and Größe des Neugeborenen – Oberkiefers anhand von 1000 eigens gewonnenen Modellen Med Diss Leipzig 1959
Howell S Assessment of palatal height in children Community Dent Oral Epidemiol 1981 9 44 47 6941877
Kent SE Rock WP Nahl SS Brain DJ The relationship of nasal septal deformity and palatal symmetry in neonates J Laryngol Otol 1991 105 424 427 2072008
Leighton BC Morphologische Variationen der Alveolarbögen beim Neugeborenen Fortschr Kieferorthop 1976 37 8 14 10.1007/BF02169375
Oelschlägl S Gestaltuntersuchungen an Oberkiefermodellen von 1000 Neugeborenen unter Berücksichtigung des Geschlechtsunterschiedes und der Beziehung zum Kopfindex Med Diss Leipzig 1954
Ott H Beitrag zur normalen Entwicklung des Milchgebisses von der Geburt bis zu ca. drei Jahren (anhand von ca. 463 Untersuchungsfällen) Med Diss Leipzig 1961
Robke E Die Gaumenplatte nach Castillo Morales and ihre Wirkung auf die Gaumenentwicklung von Säuglingen mit Trisomie 21 im Vergleich zu gesunden Säuglingen Zahnmed Diss Münster 1998
Sillman JH Relationship of maxillary and mandibular gum pads in the newborn infant Am J Orthodont Oral Surg 1938 24 409 424 10.1016/S0096-6347(38)90295-9
|
16270909
|
PMC1298321
|
CC BY
|
2021-01-04 16:37:14
|
no
|
Head Face Med. 2005 Oct 28; 1:9
|
utf-8
|
Head Face Med
| 2,005 |
10.1186/1746-160X-1-9
|
oa_comm
|
==== Front
Int J Health GeogrInternational Journal of Health Geographics1476-072XBioMed Central London 1476-072X-4-291628197610.1186/1476-072X-4-29ResearchGeographic bias related to geocoding in epidemiologic studies Oliver M Norman [email protected] Kevin A [email protected] Mir [email protected] Fern R [email protected] Linda W [email protected] Department of Family Medicine, University of Virginia, Charlottesville, VA, USA2 Department of Public Health Sciences, University of Virginia, Charlottesville, VA, USA3 Surveillance Research Program, Division of Cancer Control and Population Sciences, National Cancer Institute, Bethesda, MD, USA2005 10 11 2005 4 29 29 2 11 2005 10 11 2005 Copyright © 2005 Oliver et al; licensee BioMed Central Ltd.2005Oliver 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
This article describes geographic bias in GIS analyses with unrepresentative data owing to missing geocodes, using as an example a spatial analysis of prostate cancer incidence among whites and African Americans in Virginia, 1990–1999. Statistical tests for clustering were performed and such clusters mapped. The patterns of missing census tract identifiers for the cases were examined by generalized linear regression models.
Results
The county of residency for all cases was known, and 26,338 (74%) of these cases were geocoded successfully to census tracts. Cluster maps showed patterns that appeared markedly different, depending upon whether one used all cases or those geocoded to the census tract. Multivariate regression analysis showed that, in the most rural counties (where the missing data were concentrated), the percent of a county's population over age 64 and with less than a high school education were both independently associated with a higher percent of missing geocodes.
Conclusion
We found statistically significant pattern differences resulting from spatially non-random differences in geocoding completeness across Virginia. Appropriate interpretation of maps, therefore, requires an understanding of this phenomenon, which we call "cartographic confounding."
bias (epidemiology)confounding factorsepidemiologygeographic information systems
==== Body
Background
Epidemiologists and public health researchers are increasingly using geographic information systems (GIS) to assess the association between population health and the area characteristics of where people reside [1-8]. However, spatial analyses are fraught with challenges. The initial task of assigning geographic locations to study subjects – geocoding – can be difficult [9,10]. As Krieger et al have noted [11], the completeness with which geocoding is performed varies, which can affect the findings of spatial epidemiologic analyses [12].
There are two separate and equally important problems that can arise in the process of matching addresses to locations. First, addresses may be assigned longitudes and latitudes that are unacceptably far from their actual locations. This positional inaccuracy, owing to incorrect addresses or the assignment of incorrect latitudes and longitudes to correctly recorded addresses, can lead to bias in a study's outcomes [13-16]. This paper does not address this issue; rather, it focuses on the second problem, which is differential match rates by geographic region.
Such differential match rates can give biased results because GIS analyses may be based on unrepresentative data and a consequent information [17] bias in which important data are missing in a non-random fashion. Non-random missingness, a term used by statisticians to describe this information bias [18], can result from social, economic, political, and other reasons. One of the central contributions of spatial analyses of disease is that geographic location can be used to help account for unmeasured and unmeasurable risks for disease[19,20]. However, if the risk factors for missing locations are some of the same risks as those for the disease under study, then this confounding of risk factor with place makes an unbiased spatial analysis more difficult to achieve [21]. We illustrate this latter type of bias through an analysis of prostate cancer incidence over a 10-year period in Virginia.
In a study of prostate cancer incidence and race in Virginia [22], we found that the median household income and urban status of an area were associated positively with prostate cancer incidence for both African Americans and whites. The level of poverty and lower education were associated with decreased incidence among whites but not African Americans. Statistically significant associations were found only at the census tract level, disappearing when the analyses were conducted at the county level.
We sought to discern whether the differences we noted at the census-tract level analysis were real or simply an example of the modifiable areal unit problem (MAUP) [20], in which one obtains different results and inferences when the same set of data is grouped in different sized areal units. For example, Krieger et al [4] found in their analysis of cancer incidence and mortality in Massachusetts and Rhode Island that significant findings were lost when the analyses were conducted with areal units larger than the census tract. Gregorio et al [23], however, found in their cluster analysis of incident prostate and breast cancer cases in Connecticut that there were few differences in results across areal units, and there was no compelling need in typical cancer surveillance studies to prepare data at areas finer than the census tract.
In our study, only 74% of the cases were geocoded to the census-tract level, whereas 100% of the cases had county codes. When the statistical analyses were conducted at the county level with the same reduced data set (74% of the cases) as that used at the census-tract level, the results were similar, regardless of the geographic unit of analysis. (The generalized linear mixed modeling used to conduct this analysis and the results are described in detail elsewhere [22].) When the county analyses were conducted using 100% of the cases, the associations between predictor variables and prostate cancer incidence disappeared. These statistical analyses indicated the differences in results were not owing to the MAUP, but possibly to missing data.
In this paper, we report the findings of a study we conducted to evaluate whether unrepresentative data resulted merely from being missing, or whether the "missingness" of the data itself was confounded geographically with our covariates.
Results
County codes were available for all African-American and white cases from the Virginia Cancer Registry (VCR). We successfully geocoded 26,338 (74 percent) of the cases to the census-tract level. The types of unmatched cases did not differ between African Americans and whites, with rural routes and Post Office boxes making up virtually all of the 26 percent of unmatched addresses in both populations. (See Table 1.) As can be seen from Table 1, we geocoded about 94% of the cases that possibly could be geocoded. Gregorio et al [21] note that subject loss between 5 and 16% had been reported in a few studies [24-26]. Our geocoding match rate falls within that same range when measured against the cases with actual addresses.
Table 1 Census Tract geocoding results broken down by address type.
% Of address types (No.)
No. %(No.) Street Addresses Rural Routesa P.O. Boxesa Othera,b
Matchedc
African American 6,060 74.0 93.8 0.0 0.0 0.0
White 20,278 73.4 94.0 0.0 0.0 0.0
Unmatchedc
African American 2,192 26.6 6.2 100.0 100.0 100.0
White 7,136 26.0 6.0 100.0 100.0 100.0
TOTAL NUMBER 35,666 (26,338) (28,039) (4,131) (2,635) (861)
aAccurate geocoding to the Census tract cannot be performed on this address type. bIncludes garbled and incomplete addresses. cTo the Census Tract.
The overall incidence rate for whites was 97/100,000, whereas the rate for African Americans was 157/100,000, using the cases successfully geocoded to the census tract. (Smoothed maps of the annualized, age-adjusted prostate cancer rates for all males in Virginia 1990–99 are shown in Figure 1.) Statistical testing for global clustering was highly significant for the entire time period (Tango's Maximum Excess Events Test, p < 0.008) and for 1990–94 and 1995–99 separately (p < 0.001 for both). We examined local clustering at the county level in both time periods, using either 100 percent of the cases or just those cases geocoded to the census tract level (74 percent of the cases). With 100 percent of the cases, we found 6 such clusters in 1990–94 and 8 such clusters in 1995–99. In both time periods, major clusters appeared in geographically similar locations. For the reduced data set, we found 14 statistically significant clusters in 1990–94 and 10 such clusters in 1995–99. For each time period, patterns appeared markedly different, depending upon whether one used the cases located in the county or those geocoded to the census tract (Figure 2) [22].
Figure 1 Annualized, age-adjusted prostate cancer incidence in Virginia, 1990–99.
Figure 2 Prostate cancer incidence clusters in Virginia, 1990–99.
Figure 3 shows the proportion of missing census tract geocodes for the years 1990–1994 and 1995–1999. During the earlier study period, the proportion of missing geocodes in some areas reached 95%. Clearly, in the second half of the study period, we successfully geocoded more cases. In part, the increased match rate was due to a decreasing rural population, and, also, owing to increased numbers of rural residents receiving addresses rather than rural routes or Post Office box numbers.
Figure 3 Proportion of unmatched prostate cancer cases in Virginia, 1990–99.
However, cases in the most rural portions of the state remained systematically over-represented in the group with missing geocodes. Figure 4 depicts a cluster analysis on proportion of missing geocodes. Significant clusters of missing geocodes are all in rural areas of the study area. In addition to the missing geocodes being concentrated in rural areas, the generalized linear regression analysis showed that, in the most rural counties, the percent of a county's population age 65 or older and adults with less than a high school education were both independently associated with a higher percent of missing geocodes (p = 0.016 and p = 0.003, respectively). One study in California suggested that P.O. Box holders were not necessarily representative of the entire case population [27]. Semivariograms of the residuals from the generalized linear regression models showed no sign of spatial correlation.
Figure 4 Clusters by proportion of unmatched prostate cancer cases in Virginia, 1990–99.
Discussion
One of the most important contributions of using GIS technology in epidemiologic research is to help us discern geographic patterns of disease. We found that geographic patterns of prostate cancer incidence at the census-tract level in Virginia may reflect the distribution of the available data rather than real, underlying disease patterns. In an analysis of cancer incidence, where the census population is the denominator, information bias may result from missing geocoded data. The effect of missing geocoded data could be different, for example, in an analysis of the proportion of late-stage disease, where the denominator would be the successfully geocoded cases. However, even in this situation, if screening practices are different in urban versus rural communities, then the apparent proportion of late-stage disease could be biased.
The percentage of data able to be geocoded at the census-tract level in this analysis increased over the study period, reflecting the progressive implementation of the enhanced 911 rules by the Federal Communications Commission. These rules, among other things, require assigning street addresses to rural locations. Despite these improvements, as well as increasing prostate cancer screening from 1990–94 to 1995–99, the location of high-rate clusters did not differ markedly between these two time periods. However, for each time period the spatial location of clusters among cases geocoded to the census tract versus those located in the county were vastly different.
These findings demonstrate statistically significant pattern differences resulting from spatially non-random differences in geocoding completeness across Virginia. In classic epidemiologic terms, a measure of the effect of one factor on disease risk can be biased because of its association with another factor (confounder) and the disease. Similarly, when the factor of interest is geographic, a factor related to the disease that is not distributed randomly across the study area can confound the appearance of maps of that disease. Appropriate interpretation of maps, therefore, requires an understanding of this phenomenon, which we call "cartographic confounding."
In this study, systematically missing data are a result of location; however, a location's urban or rural status and associated sociodemographic characteristics were found to be associated with the likelihood of missing data from that location, as well as to the likelihood of disease in that area. Spatial patterns of disease incidence, therefore, may confound cartographically the location and sociodemographic risk factors for the disease. In our study of prostate cancer incidence in Virginia, the findings that area-level measures of income and urban status are associated with increased incidence are tempered by the possibility of cartographic confounding. This problem is particularly vexing when evaluating geographic health disparities, as the possible bias of one's statistical analyses depends upon the proportion of rural population in the study [28].
Cartographic confounding is geographically based, i.e., related to location. Methods designed to account for possibly unrepresentative data, therefore, also should account for this geographic component. One approach to dealing with this problem is to minimize missing data through imputation of geocodes [29].
Conclusion
When conducting spatial analyses, sound analyses depend upon assessing for possible bias and cartographic confounding resulting from insufficient geocoding that leads to systematically missing data. In this way, the power of geographic information science can be more effectively brought to bear on important issues of public health and the inferences from the analyses are more likely to be correct.
Methods
Data sources
Incidence data are from the Virginia Cancer Registry (VCR), 1990 – 1999. There were a total of 37,373 malignant neoplasms of the prostate in that period, with 27,414 in whites, 8,252 in African Americans, and 1,707 in others. Incident cases were geocoded to the street level using ArcGIS and its StreetMap USA 2000 database.(Products of the Environmental Science Research Institute, Redlands, CA.) A point-in-polygon methodology [30,31] was used to attribute 1990 census tracts to cases. Case counts were aggregated to the census tract. All cases were assigned county codes by the VCR. County codes were checked against the address-matched geocodes as a quality control measure. County codes were incorrect in only 1% of cases.
For the study period, the North American Association of Central Cancer Registries (NAACCR) reports that the VCR has 90% case ascertainment [32]. The VCR is not given NAACCR's top ranking primarily because the number of cases ascertained by death certificate only is too high or not available, depending on the year.
Area-based measures were derived from the 1990 U.S. Census data (U.S. Bureau of the Census Summary Tape File 3A). These measures were used so that presumed exposures occurred before disease incidence. The poverty variable was a measure of the percentage of persons in a census tract below the poverty level, categorized as <10%, 10 – 19%, and ≥20%. A near-poor variable measured the percent of the tract's population between 100 and 200% of the federal poverty level. The tract's median household income was also used as a variable. A low-education variable measured the percentage of persons in a census tract 25 years or older who had less than a high-school education. A high-education variable did likewise for that percentage with at least 4 years of college. The percent of a tract's population that was rural (≤50%, 51 to <100%, and 100%) was another variable. These cutpoints were chosen based on frequencies for this measure in our data. The percent of female heads of household was another predictor variable.
For both the African-American and white populations during the study period, we used an area allocation method [19,30] to produce population averages over the study period at the tract level. (The 1990 and 2000 census tracts do not match exactly, and this method was utilized to adjust for this fact.) The 1990 and 2000 county boundaries were directly comparable. As a result, we created a direct average of populations without any manipulation. The resulting averages were annualized over the 10-year study period, and these figures were used to calculate age-adjusted incidence rates of prostate cancer by the direct method utilizing the 2000 U.S. standard million [33].
Exploratory spatial data analysis
Annualized, age-adjusted prostate cancer incidence rates for African Americans and whites were calculated at the census tract and county levels. Owing to the low case counts at younger ages, we used three age categories (<50, 50–74, and ≥75). These incidence rates were mapped at the tract and county levels.
Low case counts, sparse populations, or both result in unstable incidence rates. We used a weighted, two-dimensional, median-based smoothing algorithm called "headbanging" to reduce this noise [34], allowing patterns to emerge from the data.
Statistical methods
Hierarchical Poisson regression modeling, using the SAS GLIMMIX macro (SAS 2001), was performed to assess prostate cancer incidence for all census tracts and counties in Virginia by the patient's age at diagnosis and sociodemographic characteristics of the census tracts. Specifically, the number of prostate cancer cases in census tract i (i = 1, ..., 1673), age group j (j = 1, 2, 3), denoted dij, was assumed to be distributed as a Poisson random variable, with a mean nijλij where nij is the corresponding population at risk and λij is the incidence rate in census tract i and age group j. We assumed a log-linear rate structure, with the county or tract intercept of the regression model conceptualized as a random effect with a spatial correlation structure to account for spatial autocorrelation in the data.
The statistical analyses were stratified by racial category. Owing to the sparseness of the VCR data in other racial categories, we only analyzed data for African Americans and whites. All main effects and two-way interactions were initially screened for significance in a logistic regression model, and a final model for the hierarchical Poisson regression was constructed using stepwise, backward variable selection. Only highly significant interaction terms (p < 0.005) were retained in the logistic model to account for the multiple comparisons inherent in the selection process [35]. The full results of this study of prostate cancer incidence in Virginia, 1990–99, is available elsewhere [22].
In the current study, we conducted analyses to assess whether prostate cancer cases clustered within the study area. As noted by Waller and Gotway [20], global clustering indicates clustering exists at some point in one's study area, whereas local clustering refers to the presence of a cluster at a specific site. We evaluated the raw count data for global clustering, using Tango's Maximum Excess Events Test (MEET) [36]. The statistical code to execute the MEET was provided to the authors in R (an open-source statistical package similar to S-Plus [Insightful Corporation]) by Prof. Toshiro Tango. We also assessed the count data with a spatial scan statistic (SaTScan) [37,38] to identify statistically significant local clusters.
The patterns of missing tract identifiers were examined by generalized linear regression models in SAS 9.1 (SAS Institute, Inc. Cary, NC) that included percent of tract population over age 64, percent of tract population aged 25 or older with less than a high school education, percent of tract population aged 25 or older with at least a college education, and the median household income in the area. These factors have been found to be associated with prostate cancer incidence [4-6,22]. Moreover, in our prior Virginia study [22], the geographic distribution of missing data and that of several of these covariates was similar, which, we hypothesized, might be the result of confounding between the two.
Authors' contributions
MNO conceived of the study, designed and coordinated the project, performed the statistical analyses, and drafted the manuscript. KAM carried out the cluster analyses. MS performed the statistical analyses for spatial clustering. FRH helped draft the manuscript. LWP helped conceive of the study, participated in its design and coordination, and helped draft the manuscript. All authors read and approved the final manuscript.
Acknowledgements
This project was supported by grant K07 CA099983 from the National Cancer Institute, HRSA CFDA No. 93.984, Academic Units in Primary Care-Family Medicine, and a grant from the Paul Mellon Prostate Cancer Research Institute. The authors would like to thank David Stinchcomb, Denise Lewis, and Barry Miller for informative discussions on this topic. We would like to especially thank the reviewers of our original manuscript. Their thoughtful comments helped immensely to improve the quality of this paper.
==== Refs
Ricketts TC Geographic Information Systems and Public Health Annu Rev Public Health 2003 24 1 6 12471275 10.1146/annurev.publhealth.24.100901.140924
McLafferty SL GIS and Health Care Annu Rev Public Health 2003 24 25 42 12668754 10.1146/annurev.publhealth.24.012902.141012
Rushton G Public Health, GIS, and Spatial Analytic Tools Annu Rev Public Health 2003 24 43 56 12471269 10.1146/annurev.publhealth.24.012902.140843
Krieger N Chen JT Waterman PD Soobader MJ Subramanian SV Carson R Geocoding and monitoring of U.S. socioeconomic inequalities in mortality and cancer incidence: Does the choice of area-based measure and geographic level matter? The Public Health Disparities Geocoding Project. Am J Epidemiol 2002 156 471 482 12196317 10.1093/aje/kwf068
GK S BA M B H Edwards BK Area socioeconomic variations in U.S. cancer incidence, mortality, stage, treatment, and survival, 1975-1999 NCI Cancer Surveillance Monograph Series, No. 4 2003 NIH Pub No 03-5417 Bethesda, MD, National Cancer Institute
Krieger N Chen JT Waterman PD Rehkopf DH Subramanian SV Painting a truer picture of U.S. socioeconomic and racial/ethnic health inequalities: The Public Health Disparities Geocoding Project Am J Public Health 2005 95 312 323 15671470 10.2105/AJPH.2003.032482
Pickle LW Feuer EJ Edwards BK Prediction of incident cancer cases in non-SEER counties. ASA Proceedings of the Biometrics Section 2000
Pickle LW Feuer EJ Edwards BK US predicted cancer incidence, 1999: Complete maps by county and state from spatial projection models NCI Cancere Surveillance Monograph Series No 5 2003 Bethesda, MD, National Cancer Institute, NIH Publication No. 03-5435
McElroy JA Remington PL Trentham-Dietz A Robert SA Newcomb PA Geocoding addresses from a large population-based study: lessons learned Epidemiology 2003 14 399 407 12843762
Boscoe FP Kielb MS Schymura MJ Bolani TM Assessing and improving census tract completeness. J Registry Management 2002 29 117 123
Krieger N Waterman PD Lemieux K Zierler S Hogan JW On the wrong side of the tracts? Evaluating the accuracy of geocoding in public health research Am J Public Health 2001 91 1114 1116 11441740
Krieger N Waterman PD Chen JT Soobader MJ Subramanian SV Carson R Zip code caveat: Bias due to spatitemporal mismatches between zip codes and U.S. census-defined geographic areas -- The Public Health Disparities Geocoding Project. Am J Public Health 2002 92 1100 1102 12084688
Cayo MR Talbot TO Positional error in automated geocoding of residential addresses Int J Health Geogr 2003 2 10 14687425 10.1186/1476-072X-2-10
Bonner MR Han D Nie J Rogerson P Vena JE Freudenheim JL Positional accuracy of geocoded addresses in epidemiologic research Epidemiology 2003 14 408 412 12843763
Rushton G Selecting appropriate geocoding methods for cancer control and prevention program activities. 2005
Whitsel EA Rose KM Wood JL Henley AC Liao D Heiss G Accuracy and repeatability of commercial geocoding Am J Epidemiol 2004 160 1023 1029 15522859 10.1093/aje/kwh310
Rothman KJ Greenland S Modern epidemiology 1998 Second Philadelphia, PA, Lippincott Williams & Wilkins
Vach W Some issues in estimating the effect of prognostic factors from incomplete covariate data Stat Med 1997 16 57 72 9004383 10.1002/(SICI)1097-0258(19970115)16:1<57::AID-SIM471>3.3.CO;2-J
EK C McLafferty SL GIS and public health 2002 New York, The Guilford Press
Waller LA Gotway CA Applied spatial statistics for public health data 2004 Hoboken, NJ, John Wiley & Sons, Inc.
Gregorio DI Cromley E Mrozinski R Walsh SJ Subject loss in spatial analysis of breast cancer Health Place 1999 5 173 177 10670998 10.1016/S1353-8292(99)00004-0
Oliver MN Smith E Siadaty M Hauck FR Pickle LW A spatial analysis of prostate cancer incidence and race in Virginia, 1990-1999 Am J Prev Med 2005
Gregorio DI Dechello LM Samociuk H Kulldorff M Lumping or splitting: seeking the preferred areal unit for health geography studies Int J Health Geogr 2005 4 6 15788100 10.1186/1476-072X-4-6
Beyers N Gie RP Zietsman HL Kunneke M Hauman J Tatley M Donald PR The use of a geographical information system (GIS) to evaluate the distribution of tuberculosis in a high-incidence community S Afr Med J 1996 86 40 1, 44 8685780
Miles-Doan R Kelly S Geographic concentration of violence between intimate partners Public Health Rep 1997 112 135 141 9071275
Rushton G Lolonis P Exploratory spatial analysis of birth defect rates in an urban population Stat Med 1996 15 717 726 9132899 10.1002/(SICI)1097-0258(19960415)15:7/9<717::AID-SIM243>3.0.CO;2-0
Hurley SE Saunders TM Nivas R Hertz A Reynolds P Post office box addresses: a challenge for geographic information system-based studies Epidemiology 2003 14 386 391 12843760
Kravets N Hadden WC The accuracy of address coding and the effects of coding errors Health Place 2005 16162420
Klassen AC Curriero FC Hong JH Williams C Kulldorff M Meissner HI Alberg A Ensminger M The role of area-level influences on prostate cancer grade and stage at diagnosis Prev Med 2004 39 441 448 15313082 10.1016/j.ypmed.2004.04.031
PA L MF G DJ M DW R Geographic information systems and science 2001 New York, John Wiley & Sons
Centers for Disease Control and Prevention (CDC) Behavioral Risk Factor Surveillance System Survey Data 2000 Atlanta, GA, U.S. Department of Health and Human Services, Centers for Disease Control and Prevention
North American Association of Central Cancer Registries Cancer in North America, 1988-2002. Appendix C: Indicators of data quality for all participating registries. 2005
Fleiss JL Statistical methods for rates and proportions 1981 New York, NY, John Wiley and Sons
Mungiole M Pickle LW Simonson KH Application of a weighted headbanging algorithm to mortality data maps Statistics in Medicine 1999 18 3201 3209 10602145 10.1002/(SICI)1097-0258(19991215)18:23<3201::AID-SIM310>3.3.CO;2-L
LR C Simultaneous statistical inference in the normal multiple linear regression model. JASA 1973 68 457 461
Tango T A test for spatial disease clustering adjusted for multiple testing Statistics in Medicine 2000 19 191 204 10641024 10.1002/(SICI)1097-0258(20000130)19:2<191::AID-SIM281>3.0.CO;2-Q
Kuldorff M Nagarwalla N Spatial disease clusters: detection and inference Statistics in Medicine 1995 14 799 810 7644860
Kuldorff M A spatial scan statistic Communications in Statistics Theory and Methods 1997 26 1481 1496
|
16281976
|
PMC1298322
|
CC BY
|
2021-01-04 16:39:06
|
no
|
Int J Health Geogr. 2005 Nov 10; 4:29
|
utf-8
|
Int J Health Geogr
| 2,005 |
10.1186/1476-072X-4-29
|
oa_comm
|
==== Front
Immun AgeingImmunity & ageing : I & A1742-4933BioMed Central London 1742-4933-2-151627114710.1186/1742-4933-2-15ResearchConstitutive degradation of IκBα in human T lymphocytes is mediated by calpain Ponnappan Subramaniam [email protected] Sarah J [email protected] Usha [email protected] Department of Geriatrics, University of Arkansas for Medical Sciences, Little Rock, AR, USA2 Department of Microbiology and Immunology, University of Arkansas for Medical Sciences, Little Rock, AR, USA3 VA Medical Research, Central Arkansas Veterans Health care system, Little Rock, AR, USA2005 4 11 2005 2 15 15 11 8 2005 4 11 2005 Copyright © 2005 Ponnappan et al; licensee BioMed Central Ltd.2005Ponnappan 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
Activation-induced induction of transcription factor NFκB in T lymphocytes is regulated by its inhibitor IκBα. NFκB activation has been demonstrated to occur either by phosphorylation on serine residues 32 and 36 of the inhibitor, IκBα, followed by ubiquitination and degradation of the inhibitor by the 26S proteasome, or by a proteasome-independent mechanism involving tyrosine phosphorylation, but not degradation. However, the mechanism underlying constitutive regulation of the levels of the inhibitor, IκB, in primary human T lymphocytes, remains to be fully delineated.
Results
We demonstrate here, the involvement of a proteasome-independent pathway for constitutive regulation of IκBα levels in primary human T lymphocytes. Pretreatment with a cell permeable calpain inhibitor, E64D, but not with a proteasome specific inhibitor, lactacystin, blocks stimulus-independent IκBα degradation in primary human T cells. However, E64D pre-treatment fails to impact on IκBα levels following stimulation with either TNFα or pervanadate. Other isoforms of the inhibitor, IκBβ, and IκBγ, appear not to be subject to a similar ligand-independent regulation. Unlike the previously reported decline in ligand-induced degradation of IκBα in T cells from the elderly, constitutive degradation does not exhibit an age-associated decline, demonstrating proteasome-independent regulation of the activity.
Conclusion
Our studies support a role for an E64D sensitive protease in regulating constitutive levels of IκBα in T cells, independent of the involvement of the 26S proteasome, and suggests a biological role for constitutive degradation of IκBα in T cells.
==== Body
Background
Transcription factor NFκB exists as homo-or-hetero-dimeric complexes, consisting of the Rel family of proteins [1]. These dimers operate as transcriptional regulators essential for a variety of cellular processes ranging from cell cycle progression to immune response gene induction [2]. In human T lymphocytes, like most other cells, NFκB exists in the cytoplasm coupled to its inhibitor IκBα or IκBβ, predominant members of IκB family of proteins [3]. A high affinity for RelA and c-Rel molecules enables these inhibitory proteins to associate with and thus restrict nuclear localization of the NFκB molecules. Most stimuli responsible for NFκB induction have been demonstrated to either invoke serine phosphorylation of the inhibitory proteins followed by ubiquitination and degradation via the 26S proteasome pathway, or involve the activation of tyrosine phosphorylation, as in the case of oxidative stress mediated stimuli, which is independent of proteasomal degradation mechanism [4,5]. While stimulation-induced modification of IκB has been studied extensively, little is known about the constitutive regulation of IκB protein in T cells under resting conditions.
Recent studies in B cell lines have demonstrated that IκBα, but not IκBβ, is constitutively degraded and is important for the induction of constitutive NFκB activity [6,7]. These studies indicate that constitutive degradation of IκB is mediated by a proteasome independent pathway. Studies also suggest that a calcium-dependent protease, calpain, may be important in regulating levels of IκBα [8-10]. These studies prompted us to investigate whether a similar regulation of "constitutive" i.e., stimulus-independent levels of IκBα occurs in primary T lymphocytes. Unlike B cells, NFκB induction and IκB regulation reported in T cells is clearly mediated by exogenous activating stimuli, with little or no constitutive nuclear NFκB present under basal condition.
Further, as NFκB regulation is significantly altered during aging in human T cells, we examined whether abnormal constitutive regulation may underlie lowered activation-mediated induction. Employing primary T cells obtained from human donors, we evaluated whether aging affects the regulation of constitutive levels of IκBα. We hypothesized that, as constitutive levels of IκBα were relatively unaffected by age in primary T cells, this may reflect minimal effect of age on calpain activity in T cells. We now report that E64D sensitive protease, calpain, is indeed responsible for regulating constitutive levels of IκBα, but not IκBβ or IκBγ, in human T cells. Further, calcium-ionophore mediated increase in calpain activity induced in T cells from young donors showed consistently higher activity at early time points after activation, when compared to the elderly. However, total calpain activity measured at the end of 60 minutes demonstrated no significant modulation based on the age of the donor. Thus, while the kinetics of calpain activation appears to be altered in T cells from the elderly, cumulative activity over a period of time remains unaffected. Additionally, we demonstrate that aging does not significantly affect stimulus-independent degradation of IκBα mediated by calpain, demonstrating proteasome-independent regulation. Thus, the calpain system is involved in the constitutive regulation of IκBα, and hence the NFκB signaling pathway, under resting conditions, in primary human T lymphocytes.
Results
Treatment with a cell permeable cysteine protease inhibitor, E64D, inhibits constitutive degradation of IκBα but not IκBβ or IκBγ in T lymphocytes from young and elderly donors
We examined the effect of a cell permeable cysteine protease inhibitor, E64D, on the basal levels of IκBα, IκBβ and IκBγ, inhibitors of the ubiquitous transcription factor NFκB. Employing primary T cells and western blotting using antibody specific to the inhibitor isoforms, we now demonstrate that constitutive levels of IκBα, but not IκBβ or IκBγ are modulated by pretreatment with E64D. Results presented in fig. 1, demonstrate that pretreatment with E64D significantly inhibits constitutive degradation of IκBα with little or no effect on either IκBβ (fig. 2) or IκBγ (fig. 3). Levels of IκBα before treatment with E64D are significantly lower, than that observed following treatment, indicating inhibition of degradation.
Figure 1 Effect of treatment with E64D on constitutive levels of IκBα in T cells from young and elderly donors. T cells obtained from young and elderly donors were either treated with E64D (+), 50 μM, or left untreated (-) for 45 minutes. At the end of incubation, cells were washed and cell lysates prepared. Lysates, equalized for protein, were resolved using SDS-PAGE, followed by electroblotting. Resolved proteins were detected using specific antibody to IκBα and ECL. Representative data from one donor pair are presented (A). Values obtained from a minimum of 4 donor pairs following densitometric scanning, are presented as mean integrated density units ± S.D.(B); ** indicates statistical difference from untreated controls.
Figure 2 Effect of age on constitutive degradation of IκBβ. T cells obtained from young and elderly donors were either treated with E64D (+), 50 μM, or left untreated (-) for 45 minutes. At the end of incubation, cells were washed and cell lysates prepared. Lysates, equalized for protein, were resolved using SDS-PAGE, followed by electroblotting. Resolved proteins were detected using specific antibody to IκBβ and ECL. Representative data from one donor pair are presented (A). Values obtained from a minimum of 4 donor pairs following densitometric scanning, are presented as mean integrated density units ± S.D.(B).
Figure 3 Effect of age on constitutive degradation of IκBγ. T cells obtained from young and elderly donors were either treated with E64D (+), 50 μM, or left untreated (-) for 45 minutes. At the end of incubation, cells were washed and cell lysates prepared. Lysates, equalized for protein, were resolved using SDS-PAGE, followed by electroblotting. Resolved proteins were detected using specific antibody to IκBγ and ECL. Representative data from one donor pair are presented (A). Values obtained from a minimum of 4 donor pairs following densitometric scanning, are presented as mean integrated density units ± S.D.(B).
As IκBα degradation induced by exogenous signals has been reported to be differentially regulated during aging [8,9], we next assessed the effect of treatment with E64D on IκB-α in T cells obtained from young and elderly donors. Results presented in fig. 1, demonstrate that, irrespective of the age of the T cell donor, E64D pretreatment significantly protected IκBα levels from constitutive degradation, and had little impact on isoforms, IκBβ (fig. 2) and IκBγ (fig. 3). Thus, aging does not influence cysteine protease-sensitive constitutive degradation of IκBα in T cells.
Inhibition of cysteine protease activity by E64D does not affect TNFα-induced degradation of IκBα in T cells
Treatment of T cells by activating stimuli such as, anti-CD3 or TNFα, have been demonstrated to induce transcription factor NFκB activation by a signal-induced, proteasome-mediated degradation of the inhibitor IκBα. To determine whether such signal-induced degradation of IκBα was subjected to regulation by cysteine proteases, we next examined the role of E64D on activation-induced levels of IκBα. Results depicted in fig. 4 and 5A, clearly demonstrate that activation of T cells with TNFα induces degradation of IκB-α, irrespective of treatment with E64D. This indicates that E64D sensitive protease does not modulate or impact on activation-induced, proteasome-dependent degradation of IκBα. It should be noted that a slower mobility IκBα band appears at later time-points (20 and 30 min.) in TNF activated cells from the elderly. We believe that this represents modified IκBα. Future experiments will determine the precise nature of this modification. As proteasome dependent degradation of IκBα is clearly differentially regulated in T cells from young and elderly donors, we next examined, whether E64D mediated inhibition of constitutive levels impact differentially on the induced degradation of IκBα in T cells from the elderly. Results presented in fig. 5A, demonstrate that pre-treatment with E64D failed to influence activation-induced degradation of IκBα. Thus, irrespective of the age of the donor, treatment with E64D failed to modulate activation-induced degradation of IκBα.
Figure 4 Effect of E64D on stimulus-dependent modification of IκBα (Western Blots). T cells obtained from young and elderly donors were either left untreated or treated with E64D (50 μM) for 45 min. At the end of incubation cells were stimulated with 100 μM pervanadate or TNFα (10 ng/ml) for 10, 20 or 30 min. Cell lysates were prepared and resolved by SDS-PAGE, using equal amounts of protein (30 μg / lane). Resolved proteins were transferred to nitrocellulose membrane and Western blotted with antibody to IκBα and detected using ECL. Representative results obtained from one donor pair are presented. Arrows represent IκBα specific bands detected and used in the analyses.
Figure 5 Effect of E64D on stimulus-dependent modification of IκBα (Integrated Densities). T cells obtained from young and elderly donors were either left untreated or treated with E64D (50 μM) for 45 min. At the end of incubation cells were stimulated with 100 μM pervanadate or TNFα (10 ng/ml) for 10, 20 or 30 min. Cell lysates were prepared and resolved by SDS-PAGE, using equal amounts of protein (30 μg / lane). Resolved proteins were transferred to nitrocellulose membrane and Western blotted with antibody to IκBα and detected using ECL. Mean integrated densities of IκBα specific bands obtained from a minimum of 4 donor pairs were used to determine fold change when compared to their respective controls and are presented as Mean (fold-change) ± S.D. [TNF-α (A); Pervanadate (B)].
Pretreatment with E64D does not interfere with activation-induced modification mediated by pervanadate treatment in T cells
As a next step in the analyses, we examined the effect of pretreatment with E64D on modes of activation that does not involve signal-induced degradation of IκBα, such as those mediated by pervanadate. Similar to the observation with TNFα, pretreatment with E64D failed to impact on activation-induced IκBα modification in cells pre-treated with pervanadate, irrespective of the age of the donor (fig. 4 and 5B). It is important to note that IκBα in cells treated with pervanadate clearly demonstrate a slower mobility band representing tyrosine-phosphorylated IκBα. Thus, irrespective of the activation stimuli (TNFa or Pervanadate), E64D appeared not to impact on the activation-induced modulation of IκBα. NFκB dependent luciferase activity was also assayed following pervanadate treatment, in the presence or absence of E64D. In keeping with the data obtained with IκBα, E64D pretreatment, failed to impact on NFκB dependent luciferase activity, (data not shown).
Treatment of T cells with proteasome inhibitor, Lactacystin, does not influence constitutive IκB-α levels, unlike that mediated by E64D
As IκB-α is constitutively degraded by E64D sensitive cysteine protease, we next assessed whether treatment with a proteasome inhibitor also interfered with this basal degradation. Results presented in fig. 6A, clearly demonstrate that pretreatment with lactacystin, a proteasome specific inhibitor, failed to influence basal levels of IκBα. Thus suggesting that basal or constitutive regulation of IκB-α is not dependent on the proteasome. To ensure that lactacystin did inhibit proteasome at the dose employed, i.e. positive control, T cells from young and elderly donors were either pretreated with lactacystin or left untreated. These cells were then subjected to treatment with TNFa. As seen in fig. 6B, TNFa treatment in the young induced degradation of IκBα, when compared to untreated controls. Pretreatment with lactacystin and then TNFα (L+T), inhibited TNFa mediated degradation. As reported previously, lactacystin is only minimally effective in T cells from the elderly.
Figure 6 Treatment with proteasome inhibitor, Lactacystin, fails to affect constitutive levels of IκBα in human T cells. T cells obtained from young and elderly donors were either left untreated (-) or treated with lactacystin (5 μM) for 2 h to inhibit the proteasome. At the end of incubation cells were washed and lysates were prepared as described. Equal amounts of protein from cell lysates were resolved using SDS-PAGE and electroblotted. Resolved IκBα was detected using antibody to IκBα and ECL. To ensure equal protein loading, blots were stripped and reprobed with β-actin. Representative data from one donor pair are presented (A). As a positive control for lactacystin, T cells from young and elderly donors were either pretreated with Lactacystin (5 μM, 2 h) and then activated with TNFα for 10 min (L+T) or activated with TNFα without lactacystin pretreatment (T). US represent untreated T cells. Cell lysates were resolved using SDS-PAGE and IκBα detected as indicated above (B).
Treatment of T cell lysates with purified calpain, mimics constitutive degradation of IκBα
As E64D pretreatment specifically inhibited basal degradation of IκBα, implicating a role for cysteine proteases such as calpain in regulating the constitutive levels of IκBα, we tested for a direct role for calpain by treating cytosolic lysates obtained from T cells with a purified preparation of calpain in an in vitro assay. T cell lysates equalized for protein from young donors were pooled and subjected to lysis in the presence of purified calpain, as indicated. Degradation of IκBα was determined by measuring the detectable levels of IκBα by SDS-PAGE and western blotting using antibody to IκBα. Results presented in fig. 7, show that upon exposure to 0.01 U of calpain, lysates from T cells demonstrate lowered levels of IκBα, indicating calpain-mediated degradation, which is inhibited by pretreatment with E64D.
Figure 7 Pretreatment with E64D inhibits calpain-mediated degradation of IκBα in T cell lysates. T cell lysates obtained from young donors were pooled and 50 μg protein was incubated with 0.01 U purified calpain. The mixture was incubated at 37°C for 15 min. At the end of incubation samples were denatured and resolved using SDS-PAGE. IκBα was detected usingspecific antibody and ECL (A). Values obtained from a minimum of 4 donor pairs following densitometric scanning, are presented as mean integrated density units ± S.D. (B). * indicates significant degradation (p < 0.05).
Kinetics of induction of calpain activity following calcium ionophore treatment, is modulated by age, but has no impact on overall effective calpain activity in T cells
To assess the effect of age on calpain activity, we obtained T cells from young and elderly healthy volunteers and examined them for endogenous calpain activity. As demonstrated in fig. 8, calpain specific protease hydrolyzing activity appeared to be slightly higher at time 0 in T cells from young donors; however, calcium ionophore induced increase in calpain activity was not significantly different between T cells from young and elderly donors. In fact, total hydrolyzing activity measured at the end of 60 minutes was not statistically different between cells obtained from the two age groups. Therefore, effective calpain activity remained unaffected by age of the T cell donor.
Figure 8 Basal Calpain activity in T cells obtained from young and elderly donors. T cells obtained from young and elderly donors were adjusted to 2.5 × 105 cells /ml. Calpain activity was measured using a fluorogenic substrate Boc-Met-AMC at 380 nm exc and 460 nm emm using a fluorometer. Specific activity was determined in the presence of calpain inhibitor E64D. Values represent calpain specific activity and represent Mean fluorescence intensity ± S.D of data obtained from 4 independent experiments.
Discussion
Calpain system has been demonstrated to be the main protease involved in constitutive degradation of IκBα [7-9]. Delineation of the exact region of IκBα necessary for degradation by calpain resulted in the identification of the C-terminal 39 amino acid sequence containing the PEST sequence to be critical for degradation in vitro by Shumway and Miyamoto [6]. However, their studies, demonstrated that calpain was not responsible for the degradation of IκBα in primary B cells [6]. Unlike B cells, that express constitutive nuclear NFκB, significant nuclear expression of NFκB in T cells occurs predominantly following stimulus-induced activation. Few studies, to date, have delved into constitutive regulation of NFκB in resting T cells. In our current studies, employing E64D, we demonstrate specificity of constitutive degradation of IκBα mediated by calpain in human primary T lymphocytes. The inhibition of this degradation in the presence of E64D, a cell permeable, cysteine protease inhibitor, supports the potential involvement of calpain activity in this process. Given the central role of NFκB in cell survival and signaling [11], constitutive degradation of the inhibitor IκBα is vital in understanding steady state kinetics of T cell regulation in the context of immune activation.
Our studies also demonstrate that degradation of IκBα under resting condition is refractory to proteasome inhibitor, Lactacystin, but not to calpain inhibitor, E64D. Therefore, unlike that reported for activation-induced degradation [12], constitutive levels of IκBα appear not to be subject to proteosomal regulation. This is particularly important given that our previous findings clearly showed that inducible degradation of IκBα is subject to an "age-effect" due to the inhibitory action of aging on proteasome- associated proteolytic activity [13,14].
Calpain-dependent degradation of IκBα has been demonstrated to occur in other cell types, which are refractory to proteasome activity [11]. Thus, it appears, that the degradation of IκBα can occur through two mutually exclusive pathways, dependent on the state of the cells, i.e., resting versus activated. Calpain system plays a role in constitutive, but not induced IκBα degradation, while proteasome degradation dictates induced levels in T cells [6,12]. Calpain activity has been demonstrated to be involved in the degradation of IκBα under certain conditions of viral infection [8]. It is therefore likely that this ability of constitutive degradation may be exploited by certain pathogens.
Unlike the most predominant inhibitor IκBα; IκBβ and IκBγ isoforms, appear not to be susceptible to this calpain-mediated degradation. Recent elegant experiments by Miyamoto et al implicate similar degradation kinetics for IκBα isoform in B cell lines [6]. Drawing upon the significance of such degradation events in the constitutive induction of NFκB in B cells, the role for constitutive regulation of NFκB by the calpain pathway in primary T cells was examined here. Results from these experiments clearly provide a biological basis for stimulus-independent degradation and its importance in the maintenance of NFκB in cell survival, which is not evident, unless challenged by stimuli capable of inducing apoptosis (data not shown).
Our studies on the effect of advancing age on constitutive degradation of IκBα, clearly implicate absence of any effect of donor age on the maintenance of E64D protease sensitive/calpain activity responsible for this degradation. Experiments conducted to determine the impact of aging on calpain activity clearly indicate that the effective activity is unaltered during aging. This is also reflected in the levels of IκBα in resting T cells from young and elderly donors, which are unaffected by age. Thus, despite loss in proteasome activity accompanying aging, calpain-mediated degradation of IκBα remains unaltered, demonstrating little or no role for the proteasomal regulation in calpain-mediated pathway that regulates IκBα levels. This observation is in keeping with earlier reports from our laboratory that demonstrated minimal effect of age on overall cellular proteolytic activity, especially, T cell chymotryptic activity [15]. It is also interesting to note that reports on calpain activity as a function of advancing age have been conflicting, with some demonstrating increased activity, [16,17] and others, decreased activity [18,19], however, these studies either used other cell types, employed exogenous substrates or cell lysates for the evaluation of the activity. Using a fluorogenic model substrate that is cell permeable, we now demonstrate that, ionomycin-inducible specific calpain activity, inhibitable by E64D, is unaffected; however, proteolytic activity observed in T cell lysates appeared to follow different kinetics in cells from the young than those observed in the elderly. Importantly, while 90% of the activity in cells from the young was clearly inhibitable by treatment with E64D, only 50% of the activity was inhibitable in cells from the elderly (data not shown).
While constitutive degradation of IκBα is clearly regulated by E64D sensitive calpain in T cells, activation-induced degradation, appears unaffected by pretreatment with E64D. Similarly, while activation-induced degradation of IκBα is sensitive to proteasome inhibition, constitutive degradation is unaffected by pretreatment with lactacystin, a proteasome inhibitor. Clearly, susceptibility of IκBα to degradation is not only dependent on the state of activation but also on the specificity of the protease. The physiologic significance of the degradation of the inhibitor clearly dictates induction of NFκB levels, and thus anti-apoptotic or survival ability, under uninduced conditions. Thus, regulation of IκBα levels in basal state of a cell is crucial and sets the stage for activation-induced survival signals. These results also indicate that the calpain pathway works independently of phosphorylation, since neither TNF nor pervanadate that induce serine and tyrosine phosphorylation, respectively, were affected by the inhibitor. Further, while proteasome dependent activation-induced degradation pathway, as well as proteasome pathway has been demonstrated to be compromised in T cells during aging, calpain activity clearly appears to be still functional, and is minimally affected by advancing age.
Conclusion
In summary, we have demonstrated that basal levels of IκBα, but not IκBβ or IκBγ are subject to regulation by E64D sensitive protease, and can be mimicked by pretreatment with calpain. The regulation of IκBα levels by cysteine protease appears to have no effect on activation-induced IκBα or on other isoforms of IκB, irrespective of the stimuli employed. Additionally, it appears that interference with this decrease in basal degradation of IκBα does not impact on cell survival under resting conditions.
Methods
Materials
Fluorochrome labeled anti-CD3, and FITC- and PE-labeled isotype controls were obtained from Sigma Chemical Co. (St. Louis, MO). Anti IgG coupled to horseradish peroxidase was obtained from BD-Transduction Laboratories (Lexington, KY). All other antibodies were from Santa Cruz Biotech (Carlsbad,CA). Enhanced Chemi-luminescence reagents were from Amersham (Arlington Heights, IL). All fine chemicals unless otherwise mentioned were obtained from Sigma Chemical Company, (St. Louis, MO.), Electrophoresis supplies and Molecular weight standards were from BioRad (Richmond, CA.). E64D and lactacystin were from Calbiochem (CA). Substrate for Calpain was from Molecular probes, (Eugene, OR).
Human subjects
Blood was obtained from healthy individuals living in the community. Young donors were between 21 and 30 years and old donors were between 65 and 85 years of age. A minimum of at least four donor pairs were used in each experiment. Both young and elderly donors were in good physical and mental health, had no apparent illness as suggested by an elaborate screening history and were not on any medication directly impacting the immune system during the course of this study.
T Lymphocyte Isolation
Peripheral blood was obtained and T cells were purified and maintained in RPMI 1640 culture medium as previously described (20). Magnetic sorting by negative selection was used to isolate CD3+ T cells. Purity of the isolated T cells was determined by flow cytometry using anti-CD3 conjugated to FITC. Populations were 90–95% pure. Treatment of T cells (20 × 106cells/ml) with pervanadate (100 μM, freshly prepared before use) was carried out for indicated times at 37°C, before cell lysates were prepared. For experiments involving the use of inhibitor, cells were treated with E64D at 50 μM for 45 minutes.
Western blotting
Cytosolic extracts for Western blotting were prepared by homogenization of cells in lysis buffer (1 mM Hepes, 10 mM KCl, 1.5 mM MgCl2, and 1 mM sodium orthovanadate, and 0.5% NP-40) (20). The following reagents were added to all buffers prior to their use: 0.5 mM dithiothreitol (DTT), 0.5 mM phenylmethylsulfonyl fluoride (PMSF), and 10 μg/ml each of aprotinin, leupeptin, and soybean trypsin inhibitor. Protein content of cytosolic extracts was determined using BioRad protein assay. Cell lysates equalized for protein (40 μg) were resolved by SDS-polyacrylamide gel electrophoresis (PAGE), transferred to nitrocellulose, immuno-blotted with specific antibody/s, and detected using anti-IgG coupled to horseradish peroxidase followed by Enhanced Chemi-luminescence (ECL). Where possible samples from young and elderly donors were resolved on the same gel, but in experiments where different treatments were analyzed, samples from young and the elderly were resolved on different gels, but were run simultaneously, to avoid inter and intra experimental variability. Resolution of samples on different gels did not influence the outcome of the results.
Calpain activity assays
Using a cell permeable fluorescent substrate Boc-Met-AMC, calpain activity within live cells was measured using a spectrofluorometer. Fluorescence was measured in cell suspension (2.5 × 105cells/ml) following the addition of fluorophore, using the LS-50 model, Perkin-Elmer Spectrofluorometer. The fluorometer was equipped with a magnetic stirrer and warmed with recirculating water at 37°C using a pump. Fluorescence was measured using excitation and emission wavelengths of 380 and 460 nm, respectively. Values were obtained using a time drive mode, for up to 60 minutes.
Statistical analyses
Data were analyzed using student's t-test. Differences were considered significant, if p < 0.05.
Acknowledgements
This work was supported by Grants provided by NIH RO1 AG13081, MO1RR1288 and in part by the use of facilities at the VA Medical Center, Little Rock, Arkansas. We gratefully acknowledge the technical assistance of Mrs. Virginia Fitzhugh and assistance provided by the General Clinical Research Center, in sample collection.
==== Refs
Hayden MS Ghosh S Signaling to NF-kappaB Gene Dev 2004 18 2195 2224 15371334 10.1101/gad.1228704
Hoffmann A Levchenko A Scott ML Baltimore D The IkappaB-NF-kappaB signaling module: temporal control and selective gene activation Science 2002 298 1241 1245 12424381 10.1126/science.1071914
Israel A The IKK complex: an integrator of all signals that activate NF-kappaB? Trends Cell Biol 2000 10 129 133 10740266 10.1016/S0962-8924(00)01729-3
Imbert V Rupec RA Livolsi A Pahl HL Traenckner EB Mueller-Dieckmann C Farahifar D Rossi B Auberger P Baeuerle PA Peyron JF Tyrosine phosphorylation of I kappa B-alpha activates NF-kappa B without proteolytic degradation of I kappa B-alpha Cell 1996 86 787 798 8797825 10.1016/S0092-8674(00)80153-1
Glickman MH Ciechanover A The ubiquitin-proteasome proteolytic pathway: destruction for the sake of construction Physiol Rev 2002 82 373 428 11917093
Shumway SD Miyamoto S A mechanistic insight into a proteasome-independent constitutive inhibitor kappaBalpha (IkappaBalpha) degradation and nuclear factor kappaB (NF-kappaB) activation pathway in WEHI-231 B-cells Biochem J 2004 380 173 180 14763901 10.1042/BJ20031796
Huang TT Kudo N Yoshida M Miyamoto S A nuclear export signal in the N-terminal regulatory domain of IkappaBalpha controls cytoplasmic localization of inactive NF-kappaB/IkappaBalpha complexes Proc Natl Acad Sci USA 2000 97 1014 1019 10655476 10.1073/pnas.97.3.1014
Waris G Livolsi A Imbert V Peyron JF Siddiqui A Hepatitis C virus NS5A and subgenomic replicon activate NF-kappaB via tyrosine phosphorylation of IkappaBalpha and its degradation by calpain protease J Biol Chem 2003 278 40778 40787 12909638 10.1074/jbc.M303248200
Schaecher K Goust JM Banik NL The effects of calpain inhibition on IkB alpha degradation after activation of PBMCs: identification of the calpain cleavage sites Neurochem Res 2004 29 1443 1451 15202778 10.1023/B:NERE.0000026410.56000.dd
Han Y Weinman S Boldogh I Walker RK Brasier AR Tumor necrosis factor-alpha-inducible IkappaBalpha proteolysis mediated by cytosolic m-calpain. A mechanism parallel to the ubiquitin-proteasome pathway for nuclear factor-kappab activation J Biol Chem 1999 274 787 794 9873017 10.1074/jbc.274.2.787
Poulaki V Mitsiades CS Joussen AM Lappas A Kirchhof B Mitsiades N Constitutive nuclear factor-kappaB activity is crucial for human retinoblastoma cell viability Am J Pathol 2002 161 2229 2240 12466137
Alkalay I Yaron A Hatzubai A Orian A Ciechanover A Ben-Neriah Y Stimulation-dependent I kappa B alpha phosphorylation marks the NF-kappa B inhibitor for degradation via the ubiquitin-proteasome pathway Proc Natl Acad Sci USA 1995 92 10599 10603 7479848
Trebilcock GU Ponnappan U Evidence for lowered induction of nuclear factor kappa B in activated human T lymphocytes during aging Gerontology 1996 42 137 146 8796372
Trebilcock GU Ponnappan U Nuclear factor-kappaB induction in CD45RO+ and CD45RA+ T cell subsets during aging Mech Ageing Dev 1998 102 149 163 9720648 10.1016/S0047-6374(97)00160-7
Ponnappan U Ubiquitin-proteasome pathway is compromised in CD45RO+ and CD45RA+ T lymphocyte subsets during aging Exp Gerontol 2002 37 359 367 11772523 10.1016/S0531-5565(01)00203-0
Hinman JD Duce JA Siman RA Hollander W Abraham CR Activation of calpain-1 in myelin and microglia in the white matter of the aged rhesus monkey J Neurochem 2004 89 430 441 15056286
Averna M De Tullio R Salamino F Minafra R Pontremoli S Melloni E Age-dependent degradation of calpastatin in kidney of hypertensive rats J Biol Chem 2001 276 38426 38432 11485997 10.1074/jbc.M101936200
Guttmann RP Johnson GV Oxidative stress inhibits calpain activity in situ J Biol Chem 1998 273 13331 13338 9582380 10.1074/jbc.273.21.13331
Witkowski JM Bryl E Paradoxical age-related cell cycle quickening of human CD4 (+) lymphocytes: a role for cyclin D1 and calpain Exp Gerontol 2004 39 577 585 15050293 10.1016/j.exger.2003.10.028
Ponnappan U Trebilcock GU Zheng MZ Studies into the effect of tyrosine phosphatase inhibitor phenylarsine oxide on NFkappaB activation in T lymphocytes during aging: evidence for altered IkappaB-alpha phosphorylation and degradation Exp Gerontol 1999 34 95 107 10197731 10.1016/S0531-5565(98)00059-X
|
16271147
|
PMC1298323
|
CC BY
|
2021-01-04 16:36:31
|
no
|
Immun Ageing. 2005 Nov 4; 2:15
|
utf-8
|
Immun Ageing
| 2,005 |
10.1186/1742-4933-2-15
|
oa_comm
|
==== Front
J Autoimmune DisJournal of Autoimmune Diseases1740-2557BioMed Central London 1740-2557-2-101628009210.1186/1740-2557-2-10ResearchAutoantibody profiles in the sera of patients with Q fever: characterization of antigens by immunofluorescence, immunoblot and sequence analysis Camacho MT [email protected] I [email protected] A [email protected]í J [email protected] Departamento de Orientación Diagnóstica. Centro Nacional de Microbiologia. Instituto de Salud Carlos III. Ctra. Majadahonda -Pozuelo Km 12,5. 28080-Madrid. Spain2 Departamento de Respuesta Inmune. Centro Nacional de Microbiologia. Instituto de Salud Carlos III. Ctra. Majadahonda -Pozuelo Km 12,5. 28080-Madrid. Spain3 Servicio de Inmunología. Hospital Carlos III. Imsalud. c/ Sinesio Delgado n° 10. 28029-Madrid. Spain2005 10 11 2005 2 10 10 4 4 2005 10 11 2005 Copyright © 2005 Camacho et al; licensee BioMed Central Ltd.2005Camacho 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.
Recent reports have shown that some of the immunological aspects of Q fever, a rickettsiosis caused by Coxiella burnetii, could be related to self-antigen responses. The aim of this study was to determine the specificity of the autoantibody response of patients with acute and chronic Coxiella infections. Smooth muscle and cardiac muscle-specific autoantibodies were observed in significant percentages in acutely or chronically affected Q fever patients when compared to healthy volunteers. Moreover, the incidence of cardiac muscle-specific autoantibody was significantly higher among chronically ill patients compared to acutely ill patients. Moreover, a band of 50 kD of a HeLa extract was detected in most of the sera of individuals with chronic infections and previous sequence analysis suggests that this antigen presents a high degree of homology with the human actin elongation factor 1 alpha. Further research would be necessary to confirm if antibodies to human cytoskeletal proteins could be of clinical importance in chronically infected Q fever patients.
AutoantibodiesC. burnetiiQ fever.
==== Body
Background
Q fever is a worldwide distributed human rickettsiosis that was described by Derrick in 1937. Burnett in Australia and Cox in the United States first isolated its etiological agent almost simultaneously so it was referred to as Coxiella burnetii [1,2]. Q fever infection is usually asymptomatic or acute self-limited but Coxiella is an intracellular bacterium that may persist within host macrophages leading to chronic complications such as endocarditis, hepatitis, meningitis, bone infections or chronic fatigue syndrome [3-6]. Diagnosis of Q fever is usually based on serological procedures because isolation of the bacterium is difficult and hazardous and requires level 3 biosafety laboratories [7]. Unique to Coxiella is its antigenic phase variation that appears to involve changes on lipopolyssacharide [2,4,8]. Virulent phase I bacteria are isolated from natural infection while avirulent phase II occurs after serial passages. Although the factors that determine the clinical presentation of Q fever are still not well understood, variation in C. burnetii strains, route of infection, host immunity and size of the inoculum have been implicated in disease evolution [2,4,5].
Autoimmune diseases are defined as the pathologic sequelae of autoimmune responses. The precise mechanisms by which they are induced are still under discussion, but genetic, hormonal and environmental factors have all been implicated [9,10]. The idea that infectious agents may represent one of the major environmental factors initiating autoimmune responses is now generally accepted and the mechanisms of induction are currently being re-examined [11-13]. Intracellular microorganisms, particularly those associated with persistent or latent infection, have developed strategies to modulate immune responses and hence survive within host cells [14]. Under these circumstances, these microorganisms would not only initiate but also sustain an anomalous reaction that would lead to the appearance of autoimmune features in predisposed patients. Molecular homology with host sequences, polyclonal stimulation of B and T clones, cytokine stimulation, over expression of major histocompatibility complex molecules (MHC II), antigen modification and host cell damage with the release of self-antigens, are the most commonly described mechanisms that could lead to autoimmune responses due to infectious agents [11,13,15-17]. Links between autoimmunity and infection have been described in several situations, as in the case of myocarditis following coxsackievirus and trypanosomal infections [18-20], rheumatic fever in relation to streptococci [21], ankylosing spondylitis and klebsiella [22], reactive arthropathies and Epstein Barr viruses [23,24] and recently heart disease and chlamydia [25]. Direct evidence that confirms the diagnosis of autoimmune diseases requires test that are still beyond the capacity of most clinical laboratories so autoimmune disease diagnosis is frequently based on circumstantial evidence [26], such as the demonstration and quantitation of the various autoantibodies.
In Q fever patients, the immune responses elicited by C. burnetii are associated with inflammatory responses, rheumatoid factor, cryoglobulins or immune-complexes [8,27]. Antibodies to cardiolipin, nuclear antigens, leukocyte, platelet, mitochondrial and smooth muscle antigens are commonly found in chronically infected patients [1,8,27]. The study of IgG subclass distributions has revealed a significant increase in IgG1 and IgG3 levels in sera of patients similarly to those described in some autoimmune diseases [28]. Based on all these findings it could be proposed that some pathological aspects of the long-term chronic Q fever disease, such as heart disease, could be induced or maintained by autoimmune mechanisms. The aim of this study was to characterize the specificity of the autoantigens recognised by circulating antibodies in sera of patients with acute and chronic Coxiella infections and identify differences between chronic diseases with and without cardiac involvement. This may bring new insights to the mechanisms of induction of possible auto-reactive cardiac complications.
Methods
Study Subjects
The work was done in accordance with the regulations of internal review board of the Institute of Health, Carlos III. Spain. The patients and volunteers gave a written informed consent on the use of their sera to participate in the study, which was approved by the Human Research Committee.
Serum samples of patients with Q fever. Convalescent sera from 58 serologically confirmed acute (n = 24) and chronic (n = 34) Q fever cases were studied. All patients were adults, equally divided between males and females, aged 26 – 60 years (mean 45 years). Diagnosis was done by immunofluorescence using phase I and phase II Coxiella antigens. A serum sample with phase II IgG titre ≥ 200 and phase II IgM ≥ 50 was considered acute and a serum sample with a phase I IgG titre ≥ 800 was classified as chronic [1,2]. All chronic cases had phase I IgA antibodies titre ≥ 50 [29]. Patients with acute infection had fever and/or atypical pneumonia but those demonstrating serological evidence of chronic Q fever disease could be divided in two groups: those with cardiac involvement (n = 26) or those with other pathologies (hepatic diseases, fever or bone disease) (n = 8).
Normal control group
Twenty serum samples from healthy volunteers were studied, (males 10, females 10) aged 18 – 36 years (mean 29 years).
Detection of autoantibodies by indirect immunofluorescence (IIF)
Cryostat sections of monkey cardiac muscle (The Binding Site, UK) were used as a substrate to determine cardiac muscle antibodies (CMA). Sections of rat liver, kidney or stomach (BioSystems, Sp) were used as a substrate to detect specific antibodies (anti-parietal cell, PCA, and anti-smooth muscle, ASMA). Human carcinoma cell line (Hep-2) (MarDx Carlsbad, USA) was used as a substrate to study anti-nuclear antibodies (ANA). Anti-neutrophil antibodies (ANCA) were screened using neutrophil slides (The Binding Site, UK) and anti-double-stranded DNA antibodies (dsDNA) were assayed using Crithidia luciliae slides (MarDx Carlsbad, USA).
IIF was done as previously described [30]. Briefly, sera of patients were diluted in phosphate buffered saline (PBS) pH 7.4 (1:40 in ANCA and Hep-2 slides, 1:20 in CMA, dsDNA and rat liver, kidney and stomach slides) and incubated for 30 min at room temperature with each of the sections. Bound antibodies were detected with a fluorescent rabbit anti-human IgG immunoglobulin (Dako, DK). After each incubation, the slides were washed three times for five minutes with PBS. After the final wash, the slides were mounted in 70% glycerol in PBS and examined in a Leitz Diaplan fluorescence microscope. Slides were coded and scored blind by two independent examiners.
Detection of HeLa antigens specific autoantibodies by Western blot
Assay was carried out using methods previously described [31,32]. Briefly, HeLa cells were cultured at 37°C in 5% CO2 to logarithmic phase in RPMI 1640 supplemented with 10 % heat inactivated fetal calf serum, 1 % glutamin, 10 U/ml penicillin and 60 μg/ml streptomycin. Then, the cells were incubated for 30 min at 4°C in lysis buffer (1 % Nonidet P-40, 150 mM NaCl, 20 mM Tris-HCl pH 8, 1 mM phenylmethylsulfonyl fluoride, 1 μ/ml aprotinin, 1 μ/ml pepstatin, 1 μ/ml leupeptin and 2 mM EDTA). After centrifugation at 20,000 × g for 30 min, the supernatants were stored at -80°C. Protein concentration was determined by BCA assay (Pierce, Pennsylvania, USA). About 400 μg of HeLa whole cell extracts were separated on 12 % SDS-PAGE. Proteins were electrotransferred to polyvinyl difluoride (PVDF) or nitrocellulose membranes in a semi-dry system for 1 h at 5.5 mA/cm2 in carbonate buffer, however ethanol 20 % (v/v) was added instead of methanol. Membranes were soaked for 1 hour in PBS-5 % bovine serum albumin (BSA) and incubated overnight with sera diluted 1:100 in PBS-BSA. After washing, the membranes were incubated for 1 hour with an alkaline phosphatase-conjugated anti-human IgG antibody (The Binding Site. UK) at 1:8000 dilution. After three washes, antigen-bound antibody was visualised with nitroblue tetrazolium and 5-bromo-4chloro-3-indolyl phosphate (Bio-Rad, Richmond, CA) following manufacturer's instructions.
Protein isolation and N-terminal sequence analysis
HeLa proteins were separated by 8% SDS-PAGE and electrotransferred to PVDF membranes as described above. The membranes were then stained with Coomassie Brilliant Blue R-250 and destained in 50 % methanol. The proteins of interest were cut out from the air-dried membrane and incubated for 30 min with 0.5 % polivinilpirrolidone-40 in 100 mM acetic acid, digested at a enzyme-substrate ratio 1:20 in weight with trypsin in 100 mM sodium bicarbonate at pH 8.2, and then incubated in acetonitrile 95:5 (v/v) for 16 hours at 37°C. After centrifugation at 12.000 × g for 10 min the digestion mixture was acidified with 2 % trifluoroacetic acid and the supernatant injected in a C18 Vydac 2.1 × 250 mm microbore column (Beckman HPLC "System Gold") at a flow rate of 0.15 ml/min in a chromatographic gradient of 7% acetonitrile and 0.09 % of trifluoroacetic acid in water. For peptide micro-sequencing a pulse liquid phase automatic sequencer was used (Applied Biosystems model 473). Sequence homologies obtained with the other proteins were analysed with FASTA and TFASTA programs using the update database from Swissprotein and GenEMBL.
Data analysis
The results of IIF were analysed using Fisher's exact test whenever appropriate. The overall difference was considered significant when p < 0.05.
Results
Detection of autoantibodies by IIF
The substrate specific responses detected in sera of patients with acute and chronic Q fever infections are shown in Table 1.
Table 1 Distribution of patients with acute and chronic Q fever that showed specific autoantibodies (IgG isotype) by indirect immunofluorescence.
Specific autoantibodies Acute Q fever (n = 24)(%) Chronic Q fever (n = 34)(%)
ANA 3(12.5) 5(14.7)
dsDNA 1(4.1) 3(8.8)
ANCA 3(12.5) 4(11.8)
ASMA 7(29.2)* 9(26.5)*
PCA 0 3(8.8)
CMA 3(12.1)* 13(38.3)#
Total positive patients 13(54.2) 23(67.6)
ANA: Anti-nuclear antibodies;
dsDNA: Anti-double strand DNA antibodies;
ANCA: Anti-neutrophil antibodies;
ASMA: Anti-smooth muscle antibodies;
PCA: Anti-parietal cells antibodies;
CMA: Anti-cardiac muscle antibodies.
*: p < 0.05 as compared with healthy controls
#: p < 0.05 as compared with acute Q fever patients and healthy controls
A total of 54.2% of patients with acute infection had demonstrable responses to at least one of the antigens tested. One patient had antibodies to dsDNA and three had ANCA or ANA but only ASMA and CMA were detectable in a significant number of patients (29.2 % and 12.1% respectively).
On the other hand, 67.2% of the patients with chronic Q fever infections showed positive results by IIF, most of them to various antigens tested and no differences were found between patients with or without cardiac involvement. Three patients (8.8%) elicited antibodies to dsDNA or parietal cells, four had ANCA (11.8%) and five ANA (14.7%), and, as in acute patients, the presence of ASMA was detected in a significantly higher percentage of sera (26.5%). CMA were found in 13 patients (38.3%) and both fibrillar and sarcolemma fluorescence stains were found (figure 1). The high frequency of CMA-specific antibody response detected on chronically infected patients was statistically significant (p < 0.05) as compared to patients with acute infection. Sera samples from healthy controls showed no reactivity at the dilutions tested.
Figure 1 Fibrillar indirect immunofluorescence stain with obtained with sera of Q fever patients using monkey cardiac muscle sections. Magnification × 400.
Detection of antibodies by Western blot and sequence analysis
Fourteen different HeLa bands, with molecular masses ranging from 20 to 100 kD, were detected by Western blot. Normal sera showed minimal if any reactivity with HeLa proteins. Twelve patients with acute Coxiella infections (57%) showed antibodies that reacted with one to three different bands per sample. By contrast, twenty-five sera of patients with chronic Coxiella infections (75%) showed antibody reactivities to a broad spectrum of up to seven different proteins. A predominant and consistent 50 kD band was observed in 14 of the patients with chronic infections (42%) and in 3 of the acute cases (14%) (Figure 2). No reactivity to this band was found in healthy controls. Sequence analysis of a fragment of this 50 kD protein showed a 98% homology with human elongation factor 1 alpha (Figure 3).
Figure 2 IgG antibodies detected in sera of Q fever patients against partially purified HeLa cell antigens by Western blot analysis. Incubated with sera from acute (number. 1,5,6,10,15,20,24) and chronic (number: 2,7,10,11,15,22,27,28,31,32) Q fever patients. Results obtained with representative sera from each group are depicted. Location of the 50 kD bands are indicated.
Figure 3 Sequence alignement of a fraction of the chromatogram of the 50 kD HeLa protein trypsin digestion and human elongation factor 1alpha.
Discussion
It must be noted that the presence of natural autoantibodies is a normal feature in sera of healthy individuals and their presence may only be circumstantial or transitory. Such natural autoantibodies are usually present at low titre, have poor affinity for their corresponding antigen and usually belong to the IgM class. In autoimmune diseases, self-antigens are recognised by pathological autoantibodies with specificities that differ from those that may occur naturally in healthy humans [26].
As in other infectious diseases, some immunological aspects of the host response to Coxiella in patients with Q fever could be related to self-antigen responses. In our hands, the presence of PCA, ANA, dsDNA and ANCA does not seem to clearly correlate with any of our group of patients and could be only circumstantial. Only ASMA and CMA antibodies were found in a significant percentage of acute and chronically infected Q fever patients as compared to the normal population. In the literature, CMA have rarely been found in normal individuals and their presence has been associated with autoimmune myocarditis, idiopathic dilated cardiomyopathy or rheumatic carditis [33-37]. Many different specificities have been described but mitochondrial and contractile cytoplasmic proteins such as myosin or actin seem to be widely implicated [33,38]. CMA detection following microbial infections has been related to the release of antigens after tissue damage, but in some cases, as in chlamydia infections, Chagas disease, post-streptoccocal rheumatic fever or coxsackie myopericarditis, molecular homologies between host antigens and the etiological agent have been described [11,18,21,25]. CMA presence in the sera of Q fever patients without cardiac involvement was unexpected and suggested that they were not elicited by tissue damage. The possible existence of molecular homologies between coxiella and host antigens was studied by immunoblotting with human HeLa extracts. Results revealed that sera from most of the chronically infected patients and some of acute patients reacted strongly with a 50 kD antigen. Preliminary results of the sequence analysis of an internal fragment of this protein showed molecular homology with human elongation factor 1 alpha. This ubiquitously expressed protein is phylogenetically conserved and an abundant member of the actin binding proteins family, responsible for binding the aminoacyl-tRNA to the ribosome during polypeptide elongation [39-41]. It co-localises intracellularly with the F-actin and is related to changes in the actin cytoskeleton. High levels of its expression are correlated with cell proliferation, oncogenic transformation, metastasis and are been considered a molecular market for injured muscle [42,43] and it has been described as a common IgG auto-antibody target in atopic dermatitis and Felty's syndrome [44,45].
This preliminary study shows that, as in other infectious diseases, patients with Coxiella infections have an autoantibody profile with specificities that resemble those found in autoimmune disease. Further studies would be needed to evaluate whether these findings could only be considered a circumstantial evidence of natural response or have pathological implications.
Acknowledgements
We would like to thank A.I. Marina, J. Gonzalez, J.M. Vazquez of the Department of Quimica de Proteinas of Centro de Biología Molecular of the University Autónoma de Madrid for performing amino acid sequencing and M.J. Ramos for technical assistance.
This work was supported by a grant from the Fondo de Investigaciones Sanitarias (Expt n° 97/0191).
==== Refs
Madariaga MG Rezai K Trenholme GM Weinstein RA Q fever: a biological weapon in your backyard Lancet Infect Dis 2003 3 709 721 14592601 10.1016/S1473-3099(03)00804-1
Maurin M Raoult D Q Fever Clin Microbiol Rev 1999 12 518 553 10515901
Harris RJ Storm PA Lloyd A Arens M Marmion BP Long-term persistence of Coxiella burnetii in the host after primary Q fever Epidemiol Infect 2000 124 543 549 10982079 10.1017/S0950268899003763
Raoult D Marrie TJ Mege JL Natural history and pathophysiology of Q fever Lancet Infect Dis 2005 5 219 226 15792739 10.1016/S1473-3099(05)70052-9
Arricau-Bouvery N Rodolakis A Is Q fever and emerging or re-emerging zoonosis? Vet Res 2005 36 327 349 15845229 10.1051/vetres:2005010
Raoult D Tissot-Dupont H Foucault C Gouvernet J Fournier PE Bernit E Stein A Nesri M Harle J Weiller PJ Q fever 1985–1998 Medicine 2000 79 109 123 10771709 10.1097/00005792-200003000-00005
Bossi P Tegnell A Baka A van Loock F Hendriks J Werner A Maidhof H Gouvras G Bichat Guidelines for the Clinical Management of Q Fever and Bioterrorism-Related Q Fever Eurosurveillance 2004 9 E19 E20 15677840
Fournier PE Marrie TJ Raoult D Diagnosis of Q fever J Clin Microbiol 1998 36 1823 1933 9650920
Dighiero G Rose NR Critical self-epitopes are key to the understanding of self-tolerance and autoimmunity Immunol Today 1999 20 423 428 10462743 10.1016/S0167-5699(99)01509-1
Klareskog L Lorentzen J Padyukov L Alfredsson L Genes and environment in arthritis: can RA be prevented? Arthritis Res 2002 4 31 36 10.1186/ar566
Kamradt T Goggel R Erg KJ Induction, exacerbation and inhibition of allergic and autoimmune diseases by infection Trends Immunol 2005 26 260 267 15866239 10.1016/j.it.2005.03.009
Slev PR Potts WK Disease consequences of pathogen adaptation Curr Opin Immunol 2002 14 609 614 12183161 10.1016/S0952-7915(02)00381-3
Behar SM Porcelli SA Mechanisms of autoimmune disease induction. The role of the immune response to microbial pathogens Arthritis Rheum 1995 38 458 476 7718001
Small PLC Ramakrishnan L Falkow S Remodelling schemes of intracellular pathogens Science 1994 263 637 639 8303269
Aichele P Bachmann MF Hengartner H Zinkernagel RM Immunopathology or organ-specific autoimmunity as a consequence of virus infection Immunol Rev 1996 152 21 45 8930666
Naucler CS Larsson S Moller E A novel mechanism for virus-induced autoimmunity in humans Immunol Rev 1996 152 175 192 8930673
Von Herrath MG Oldstone MBA Virus-induced autoimmune disease Curr Opin Immunol 1996 8 878 885 8994870 10.1016/S0952-7915(96)80019-7
Kaplan D Ferrari I Lopez Bergami P Mahler E Levitus G Chiale P Hoebekes J Van Regenmortel MHV Levin MJ Antibodies to ribosomal P proteins of Trypanosoma cruzi in Chagas disease possess functional autoreactivity with heart tissue and differ from anti-P autoantibodies in lupus Proc Natl Acad USA 1997 94 10301 10306 10.1073/pnas.94.19.10301
Rose NR Herskowitz A Neumann DA Neu N Autoimmune myocarditis: a paradigm of post-infection autoimmune disease Immunol Today 1988 9 117 120 3076766 10.1016/0167-5699(88)91282-0
Wolfgram LJ Beisel KW Rose NR Heart-specific autoantibodies following murine Coxsackievirus B3 myocarditis J Exp Med 1965 161 1112 1121 2985729 10.1084/jem.161.5.1112
Kaplan MH Meyeserian M An immunological cross-reaction berween Group-A Streptococcal cells and human heart tissue Lancet 1962 1 706 710 14453769 10.1016/S0140-6736(62)91653-7
Isenberg DA Maddison P Swana G Skinner RP Swana M Jones M Addison I Dudeney C Shall S Roiey AEL Shoenfeld Y Profile of autoantibodies in the serum of patients with tuberculosis, Klebsiella and other Gram-negative infections Clin Exp Immunol 1987 67 516 523 3301097
Vaughan JH Valbracht JR Nguyen MD Handley HH Smith RS Patrick K Rhodes GH Epstein-Barr virus-induced autoimmne responses. I Immunoglobulins M autoantibodies to proteins mimicking and not mimicking Epstein-Barr Virus Nuclear Antigen 1 J Clin Invest 1995 95 1306 1315 7533788
Vaughan JH Nguyen MD Valbracht JR Smith RS Patrick K Rhodes GH Epstein-Barr virus-induced autoimmune responses. II Immunoglobulin G autoantibodies to mimicking and notmimicking epitopes. Presence in autoimmune disease J Clin Invest 1995 95 1316 1327 7533789
Bachmaier K Neu N de la Maza LM Pal S Hassel A Penninger JM Chlamydia infections and heart disease linked through antigenic mimicry Science 1999 283 1335 1339 10037605 10.1126/science.283.5406.1335
Rose NR Bona C Defining criteria for autoimmune diseases (Witebsky's postulates revisited) Immunol Today 1993 14 426 430 8216719 10.1016/0167-5699(93)90244-F
Levy P Raoult D Razongles JJ Q fever and autoimmunity Eur J Epidemiol 1989 5 447 453 2691275 10.1007/BF00140139
Camacho MT Outschoorn I Téllez A Role of IgG subclasses in Q fever Eur J Clin Microbiol Infect Dis 1995 14 1070 1075 8681982 10.1007/BF01590941
Camacho MT Outschoorn I Echevarría C Kovachova E Yebra M Maté I Auffray P Téllez A Distribution of IgA subclass response to C.burnetii in patients with acute and chronic Q fever Clin Immunol Immunopathol 1998 88 80 83 9683553 10.1006/clin.1998.4547
Tan EM Relationship of nuclear staining patterns with precipitating antibodies in systemic lupus erythematosus J Lab Clin Med 1967 70 800 812 4964289
Laemmli UK Cleavage of structural proteins during assembly of thew head of bacteriophage T4 Nature 1970 227 680 685 5432063 10.1038/227680a0
Garcia-Lerma JG Mendoza AZ Ramos MJ Sequi J Evaluation of Recombinant Ro/SSA, La/SSB, Sm and U1 RNP Autoantigens in Clinical Diagnosis J Clin Lab Anal 1995 9 52 58 7722773
Scheerder I Buyzere M Algoed L De Lange M Delanghe J Bogaert AM Clement DL Characteristic anti-heart antibody patterns in post-cardiac injury syndrome, endocarditis and acute myocarditis Eur Heart J 1987 8 237 238
Caforio AL Bonifacio E Stewart JT Neglia D Parodi O Bottazzo GF McKenna WJ Novel organ-specific circulating cardiac autoantibodies in dilated cardiomyopathy J Am Coll Cardiol 1990 15 1527 1534 2188986
Caforio AL Mahon NJ McKenna WJ Cardiac autoantibodies to myosin and other heart-specific autoantigens in myocarditis and dilated cardiomyopathy Autoimmunity 2001 34 199 204 11908778
Latif N Baker C Dunn MJ Rose ML Brady P Yacoub MH Frequency and specificity of antiheart antibodies in patients with dilated cardiomyopathy detected using SDS-PAGE and Western-blotting J Am Coll Cardiol 1993 22 1378 1384 7901255
Tontsch D Pankuweit S Maish B Autoantibodies in the sera of patients with rheumatic heart disease: characterization of myocardial antigens by two-dimensional immunoblotting and N-terminal sequence analysis Clin Exp Immunol 2000 121 270 274 10931141 10.1046/j.1365-2249.2000.01283.x
Neu N Rose NR Beisel KW Herskowitz A Gurri-Glass G Craig SW Cardiac Myosin induces myocarditis in genetically predisposed mice J Immunol 1987 139 3630 3636 3680946
Moldave K Discovery, resolution, purification and function of elongation factors Methods Enzymol 1990 182 809 818 2179673
Negrutskii BS El'skaya AV Eukaryotic translation elongation factor 1 alpha: structure, expression, functions and possible role in aminoaciyl-tRNA channelling Prog Nucleic Acid Res Mol Biol 1998 60 47 78 9594571
Bektas M Guncer B Guven C Nurten R Bermek E Actin-an inhibitor of eukariotic alongation factor activities Biochem Biophys Res Commun 2004 317 1061 1066 15094376 10.1016/j.bbrc.2004.03.153
Edmonds BT Wyckoff J Yeung Y Wang Y Stanley R Jones J Segall J Condeelis J Elongation factor-1α is an overexpressed actin binding proteins in metastatic rat mammary adenocarcinoma J Cell Sci 1996 109 2705 2714 8937988
Khalyfa A Carlson BM Carlson JA Wang E Toxin injury-dependent switched expression between EF-1 alpha and its sister S1 in rat skeletal muscle Dev Dyn 1999 216 267 273 10590478 10.1002/(SICI)1097-0177(199911)216:3<267::AID-DVDY5>3.0.CO;2-V
Ditzel HJ Masaki Y Nielsen H Farnaes L Burton RD Clonning and expression of a novel human antibody-antigen pair associated with Felty's syndrome PNAS 2000 97 9234 9239 10922075 10.1073/pnas.97.16.9234
Ohkouchi K Mizutani H Tanaka M Takahashi M Nakashima K Shimizu M Anti-elongation factor-1α autoantibody in adult athopic dermatitis patients Intern Immunol 1999 11 1635 1640 10.1093/intimm/11.10.1635
|
16280092
|
PMC1298324
|
CC BY
|
2021-01-04 16:38:03
|
no
|
J Autoimmune Dis. 2005 Nov 10; 2:10
|
utf-8
|
J Autoimmune Dis
| 2,005 |
10.1186/1740-2557-2-10
|
oa_comm
|
==== Front
J NeuroinflammationJournal of Neuroinflammation1742-2094BioMed Central London 1742-2094-2-241625962810.1186/1742-2094-2-24ReviewThe microglial "activation" continuum: from innate to adaptive responses Town Terrence [email protected] Veljko [email protected] Jun [email protected] Section of Immunobiology, Yale University School of Medicine, 300 Cedar St., New Haven, CT 06520-8011, USA2 Neuroimmunology Laboratory, Silver Child Development Center, Department of Psychiatry and Behavioral Medicine, University of South Florida, 3515 E. Fletcher Ave., Tampa, FL 33613, USA2005 31 10 2005 2 24 24 24 10 2005 31 10 2005 Copyright © 2005 Town et al; licensee BioMed Central Ltd.2005Town 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.
Microglia are innate immune cells of myeloid origin that take up residence in the central nervous system (CNS) during embryogenesis. While classically regarded as macrophage-like cells, it is becoming increasingly clear that reactive microglia play more diverse roles in the CNS. Microglial "activation" is often used to refer to a single phenotype; however, in this review we consider that a continuum of microglial activation exists, with phagocytic response (innate activation) at one end and antigen presenting cell function (adaptive activation) at the other. Where activated microglia fall in this spectrum seems to be highly dependent on the type of stimulation provided. We begin by addressing the classical roles of peripheral innate immune cells including macrophages and dendritic cells, which seem to define the edges of this continuum. We then discuss various types of microglial stimulation, including Toll-like receptor engagement by pathogen-associated molecular patterns, microglial challenge with myelin epitopes or Alzheimer's β-amyloid in the presence or absence of CD40L co-stimulation, and Alzheimer disease "immunotherapy". Based on the wide spectrum of stimulus-specific microglial responses, we interpret these cells as immune cells that demonstrate remarkable plasticity following activation. This interpretation has relevance for neurodegenerative/neuroinflammatory diseases where reactive microglia play an etiological role; in particular viral/bacterial encephalitis, multiple sclerosis and Alzheimer disease.
brainmicrogliainnate immunityadaptive immunityToll-like receptorinflammationencephalitismyelinamyloidvaccineimmunotherapy
==== Body
Introduction
Microglia are somewhat enigmatic central nervous system (CNS) cells that have been traditionally regarded as CNS macrophages (MΦs). They represent about 10% on average of the adult CNS cell population [1]. In mice, microglial progenitors can be detected in neural folds at the early stages of embryogenesis. Murine microglia are thought to originate from the yolk sac at a time in embryogenesis when monocyte/Mφ progenitors (of hematopoeitic origin) are also found [1,2]. Based on this observation, it is now generally accepted that adult mouse microglia originate from monocyte/MΦ precursor cells migrating from the yolk sac into the developing CNS. Once CNS residents, these newly migratory cells actively proliferate during development, thereby giving rise to the resident CNS microglial cell pool. More recently however, it has been shown that bone marrow-derived cells can enter the CNS and become cells that phenotypically resemble microglia in the adult mouse [3-5]. Interestingly, under conditions of CNS damage such as stroke, cholinergic fiber degeneration, or motor neuron injury, Priller and colleagues found that green fluorescent protein-labeled bone marrow cells could enter the CNS and take up a microglial phenotype [6].
Microglia normally exist in a quiescent (resting) state in the healthy CNS, and are morphologically characterized by a small soma and ramified processes. However, upon "activation" in response to invading viruses or bacteria or CNS injury, microglia undergo morphological changes including shortening of cellular processes and enlargement of their soma (sometimes referred to as an "amoeboid" phenotype). Activated microglia also up-regulate a myriad of cell surface activation antigens and produce innate cytokines and chemokines (discussed in detail below). As the microglial lineage originates from peripheral myeloid precursor cells, it is helpful to consider the activation states of such peripheral innate immune cells to better understand the nature of microglial activation.
Classical roles of peripheral innate immune cells
It is now widely accepted that both innate and adaptive arms of the immune system play important roles in maintaining immune homeostasis. However, little attention was paid to the evolutionarily much older innate immune system until the late Charlie Janeway proposed the involvement of innate mechanisms in vertebrate immunity. Specifically, Janeway pioneered the idea that lymphocyte activation could be critically regulated by the evolutionarily ancient system of antigen clearance by phagocytic cells of myeloid origin. Together with Ruslan Mezhitov, he originated the concept that these phagocytic innate immune cells recognize pathogen-associated molecular patterns (PAMPs) through pattern recognition receptors, the most notable examples being a set of phylogenetically conserved, germ-line encoded Toll-like receptors (TLRs, currently 11–12 members, [7-10]), resulting in expression of cell-surface activation molecules [for example, major histocompatibility complex (MHC) class I and II, B7.1, B7.2, and CD40] and secretion of innate cytokines [i.e., tumor necrosis factor α (TNF-α), interleukin (IL)-1, IL-6, IL-12, and IL-18] [11,12]. Once activated, the innate arm of the immune response calls adaptive immune cells into action, and both branches act in concert to promote neutralization and clearance of invading pathogens. Thus, innate immune cells are able to discriminate "non-infectious self" from "infectious non-self" and thereby form the first line of defense against invading bacteria and viruses (for reviews see [13-15]).
The macrophage: prototypical phagocyte
MΦs are quintessential phagocytes whose primary role is to engulf pathogens such as invading bacteria and to remove debris and detritus, i.e., remnants of apoptotic cells. Tissue MΦs develop when blood monocytes enter into the various organs and tissues and differentiate into specialized, site-specific MΦs depending on their anatomical location, such as alveolar MΦs (lung), histiocytes (connective tissue), kupffer cells (liver), mesangial cells (kidney), osteoclasts (bone), or microglia (brain) [16]. Resting MΦs are both weak phagocytes and weak lymphocyte activators [17]. Upon activation however, for example in response to TLR stimulation by PAMPs, their phagocytic potential greatly enhances [18] and they up-regulate cell-surface co-stimulatory molecules and produce pro-inflammatory innate cytokines as mentioned above. Typically, engulfment of the pathogen by phagocytosis triggers a "respiratory burst" involving production of reactive oxygen species such as superoxide and peroxinitrite that kill the pathogen [17,19]. In addition, activated MΦs up-regulate cell-surface Fc receptors that aid in phagocytosis of pathogens opsonized by antibodies produced by plasma cells [20,21]. On the other hand, in response to debris from apoptotic cells, the MΦ mounts a phagocytic response essentially in the absence of pro-inflammatory cytokines [22]. The most likely reason for this anti-inflammatory phagocytic response is that pro-inflammatory cytokines such as TNF-α promote bystander injury which may further damage tissues in which the apoptotic cells reside. Thus, MΦs are highly capable of "innate activation" characterized by a strong phagocytic response sometimes accompanied by pro-inflammatory cytokine production (for a review see [23]).
The dendritic cell: professional antigen presenting cell
Whereas MΦs have limited ability to process and present antigen to T cells, dendritic cells (DCs) are considered professional antigen presenting cells (APCs). DCs can be found under the epithelia and in most organs where they capture and process non-self antigens, migrate to lymphoid organs, and present antigen in the context of MHC to CD4+ and CD8+ T lymphocytes. With their many finger-like cellular processes, DCs are morphologically optimized to simultaneously display antigen to many T cells. Like MΦs, DCs respond to invading pathogens by recognizing PAMPs through TLRs, and subsequently phagocytose and process antigen. DCs then up-regulate cell-surface co-stimulatory molecules and secrete innate cytokines and chemokines (typically at levels an order of magnitude higher than those secreted by MΦs) to promote recruitment and activation of CD4+ and/or CD8+ T lymphocytes. There are three generally accepted classifications of DCs in mice: plasmacytoid (p) DCs (CD11clo, CD11blo, B220+, CD8-), lymphoid (l) DCs (CD11c+, CD11b-, CD8+), and myeloid (m) DCs (CD11c+, CD11b+, B220-, CD8-, there are several subtypes, [24]). In humans, there are clearly two distinct subsets of DCs: pDCs (CD11c-, CD11b-, CD14-, CD45RA+) and monocyte DCs (CD11c+, CD11b+, CD14+, CD45RA-) (for a review see [25]). DC classes differ from each other predominately in tissue distribution, production of specific cytokines, TLR expression, and ability to promote innate versus adaptive immune responses (for a review see [15]). For example, freshly isolated human pDCs express TLR7 and 9, whereas mDCs express TLR1, 2, 3, 5, 6, and 8 [26-28]. Stimulation of human pDCs or monocytic DCs with synthetic TLR7 ligands induces the secretion of interferon (IFN)-α (important for anti-viral innate immunity) or IL-12 [a key inducer of the adaptive T helper (Th) type I response], respectively [29]. Similarly, stimulation of TLR9 via DNA containing unmethylated CpG motifs results in IFN-α secretion by pDCs and IL-12 production by murine mDCs [30]. Despite these relative differences between DC classes, the major role of DCs on the whole remains; they act as potent APCs capable of strongly activating T lymphocytes. Their APC capacity is much stronger than that of MΦs, as DCs are able to directly activate naïve T cells whereas MΦs are not [15]. Thus, by virtue of their ability to promote T cell activation responses, DCs are highly capable of "adaptive activation". Activation markers of phagocytosis and APC responses in various innate immune cells are presented in Table 1.
Table 1 Phagocytic and antigen presenting cell responses of immune cells. Note: ND, assay not performed; +/-, weak response; +, modest response; ++, strong response
Surface marker Phagocytosis APC function References
CD36 ++ + [91] [92] [93]
CD40 ND ++ [58] [94] [95]
CD80 ND ++ [94] [95] [96]
CD86 ND ++ [94] [95] [96]
MHC II ND ++ [94]
CD11c ND ++ [58]
CD40L +/- ++ [58] [80]
Cytokine
IL-1β ND ++ [97]
IL-6 ND ++ [98]
IL-12 ND ++ [98] [97]
TNF-α + ++ [98] [80] [97]
IFN-γ -/+ ++ [80]
IL-4 ++ -/+ [80]
IL-10 ++ -/+ [80]
TGF-β1 ++ ND [99]
Microglial activation after toll-like receptor stimulation: a mixed response
Recent evidence indicates that microglia, like their peripheral innate immune cell counterparts, express a wide array of TLRs, including mRNA for TLRs 1–9 in mice [31] and in humans [32]. Furthermore, many of these TLRs have been shown to be functional, allowing microglial recognition of a variety of PAMPs. For example, Kielian and coworkers found that heat-killed Staphylococcus aureus and its cell wall product peptidoglycan (PGN) are able to stimulate innate activation of microglia characterized by pro-inflammatory cytokine and chemokine production [33]. Those authors found that the effect of PGN was critically dependent on TLR2, as TLR2-deficient mice demonstrated reduced cytokine and chemokine production after PGN challenge [34]. Furthermore, murine microglia respond to the TLR9 agonist, unmethylated CpG-DNA, by secreting numerous pro-inflammatory innate cytokines (probably responsible for neurotoxicity in oganotypic brain slice cultures treated with CpG-DNA [35]), by up-regulating co-stimulatory cell surface molecules, and by promoting adaptive activation by secreting IL-12 to affect T cell activation [36]. In two recent studies, murine microglial pro-inflammatory responses to bacterial lipopolysaccharide (LPS), a known TLR4 ligand, resulted in dramatic injury to cultured oligodendrocytes [37] and neurons [38], further demonstrating microglial bystander injury after TLR stimulation (probably mediated by over-production of innate pro-inflammatory cytokines). It has recently been shown that microglia respond to poly I:C [a synthetic double-stranded (ds) RNA analog thought to be recognized by TLR3, [39]] by producing pro-inflammatory cytokines and chemokines [40], and microglial pro-inflammatory responses to dsRNA seem to be dependent on TLR3, as TLR3-deficient microglia have blunted innate cytokine responses in vitro and markedly reduced cell surface activation markers in brain after poly I:C stimulation (Town et al., submitted). Finally, infection with West Nile virus, a retrovirus that produces dsRNA during its life cycle, results in profound microglial activation as assessed by pro-inflammatory cytokine production in vitro and cell surface activation markers in vivo, effects that are dramatically reduced in TLR3-deficient animals [41].
In addition to production of pro-inflammatory cytokines and up-regulation of cell surface activation antigens, phagocytosis is a hallmark indicator of innate immune cell activation. We have recently investigated microglial phagocytosis in response to PAMP stimulation using both the N9 microglial cell line and primary cultured microglia derived from neonatal C57BL/6 mice (for methods see [42]). We pre-treated either N9 cells or primary cultured microglia with poly (I:C) (50 μg/mL), LPS (50 ng/mL), PGN (50 μg/mL), or CpG-DNA (1 μM) for 6 hours, rinsed the cells multiple times in complete RMPI 1640 media, and then cultured cells in the presence of a 1:1000 dilution of yellow-green fluorescent latex beads (Sigma) at 37°C or at 4°C (a control for non-specific incorporation of beads) for 1 hour. After this final culture period, cells were rinsed multiple times in complete media and subjected to fluorescence-activated cell sorter analysis, and mean fluorescence intensity values obtained from cells at 4°C were subtracted from figures obtained from cells cultured at 37°C. These corrected figures were then normalized for each treatment condition to values obtained from untreated control cells, yielding a percentage of phagocytosis increase over baseline. As shown in Fig. 1, both N9 and primary cultured microglia increased phagocytosis after stimulation with each PAMP studied by as much as ~190% over baseline (see PGN treatment of N9 cells, Fig. 1A).
Figure 1 PAMP stimulation results in enhanced microglial phagocytosis. N9 cells (A) or primary cultured microglia from C57BL/6 mice (B) were pre-stimulated with the PAMPs indicated (Poly I:C, 50 μg/mL; LPS, 50 ng/mL; PGN, 50 μg/mL; CpG-DNA, 1 μM) for 6 hours. Cells were rinsed in complete RPMI 1640 media (containing 5% fetal calf serum and 1 mM penicillin/streptomycin) and then cultured for an additional 1 hour at 37°C or at 4°C (control) with yellow-green fluorescent latex beads (1:1000, Sigma). After extensive rinsing, microglia were subjected to fluorescence-activated cell sorter analysis, and mean fluorescence intensity of cells cultured at 4°C was subtracted from values from cells cultured at 37°C. These figures were then normalized to untreated control microglia to obtain percentage of phagocytosis increase over baseline. Unpaired t-test was used to assess statistical significance for each treatment condition compared to control, with n = 3 wells for each condition presented; ** p < 0.001, * p < 0.05. Abbreviations used: Poly (I:C), polyinosinic : polycytidylic acid; LPS, lipopolysaccharide; PGN, peptidoglycan; CpG-DNA, unmetylated DNA containing CpG motifs.
When taking these results together with the above-mentioned reports, it seems that PAMP stimulation of TLRs produces a "mixed" microglial activation phenotype. In terms of innate responses, PAMP-stimulated microglia clearly secrete pro-inflammatory innate cytokines (i.e., TNF-α and IL-6), up-regulate cell-surface activation markers (i.e., MHC I and II, B7.1, B7.2, CD40), and increase phagocytosis. Regarding adaptive responses, particularly in the case of CpG-DNA stimulation of TLR9, reactive microglia activate T lymphocytes and may bias CD4+ T cells towards a pro-inflammatory T helper type I response by secreting IL-12 [36].
In peripheral innate immune cells, TLR responses to PAMPs seem to be dependent on at least four different TLR intracellular adapter molecules: MyD88 (involved in TLR1, 2, 4, 6, 7, 8, and 9 signaling), TRIF/TICAM-1 (mediates TLR3 and 4 signaling), TIRAP/Mal (involved in TLR1, 2, 4, and 6 responses) and TIRP/TRAM/TICAM-2 (mediates TLR4 signaling). These adapter molecules bind to the intracellular leucine-rich repeat region of the TLR and promote recruitment of additional factors such as IRAKs and TRAF6 that allow for activation of transcription factors including IRF-3 and NF-κB, which are responsible for activation of numerous innate cytokines and cell-surface activation antigen genes (for review see [43,44]). It is still unclear how different TLR responses in innate immune cells (i.e., promotion of innate versus adaptive responses) can be achieved when many TLRs share intracellular signaling molecules. While little work has been done on intracellular signaling following TLR stimulation in microglia, it is likely that microglia utilize the same signaling cascades described for MΦs and DCs.
Adaptive response of activated microglia in demyelinating disease via CD40-CD40 ligand interaction
Brain inflammation in demyelinating disease
Experimental autoimmune encephalomyelitis (EAE) is a mouse model of the human disease multiple sclerosis (MS), an autoimmune disease characterized by inflammatory CNS demyelinating lesions accompanied by motor disturbances. EAE can be induced in different strains of mice by subcutaneous or intraperioteneal inoculation with adjuvant plus epitopes found in myelin such proteolipid protein, myelin basic protein, or myelin oligodendrocyte glycoprotein. The disease is critically dependent on activation of pro-inflammatory CD4+ T helper type I (Th1) cells by APCs, and these auto-aggressive Th1 cells can be adoptively transferred to non-diseased recipient mice that subsequently develop disease. EAE is characterized by paralysis, typically beginning in the tail and hind limbs and progressing to the fore limbs. In the SJL mouse strain, animals develop a relapsing-remitting form of the disease while C57BL/6 mice manifest paralysis that progressively worsens until death. Upon histopathological analysis, brains from EAE mice generally show infiltration of Th1 cells (and other lymphocytes including MΦs and DCs) and activation of microglia, typically in white matter regions where demyelinating lesions are found (for a review see [45-47]).
CD40-CD40 ligand interaction in experimental autoimmune encephalomyelitis
Immune/inflammatory cells receiving a primary stimulus (i.e., MHC-T cell receptor interaction between APCs and T lymphocytes, respectively) typically require co-stimulatory signals via other pairs of molecules in order to become activated [for instance, the B7-CD28 and/or CD40-CD40 ligand (L) dyads in APC/T-cell activation; [48]. CD40L is a key immunoregulatory molecule that plays a co-stimulatory role in the activation of immune cells from both the innate and adaptive arms of the immune system, and is typically expressed by activated CD4+ and some CD8+ T cell subsets [49]. CD40 receptor, a member of TNF and nerve growth factor super-family, is expressed on many professional and non-professional APCs, including DC's, B cells, monocytes/MΦs and microglial cells [42,50-53]. Nearly 10 years ago, activated Th cells that expressed CD40 ligand (CD40L) were found in brains of MS patients, and these cells were found in close apposition to CD40-bearing cells in active demyelinating lesions [54]. The authors determined that the CD40-expressing cells were either MΦ or microglia based on staining for acid phosphatase or CD11b.
To evaluate whether the CD40-CD40L interaction was pathogenic in EAE, Gerritse and co-workers administered a CD40L neutralizing antibody to SJL mice that were given proteolipid protein with adjuvant to induce EAE. Strikingly, EAE was prevented in a prophylactic treatment regimen of anti-CD40L, and, when EAE was induced in another cohort of animals, CD40L antibody treatment significantly reduced disease severity in an active treatment paradigm [54]. It was later shown that genetic deficiency in CD40L [55] or antibody-mediated blockade of CD40L [56] resulted in attenuation of Th1 differentiation and effector function, including marked inhibition of the Th1 cytokine IFN-γ and reduced numbers of encephalitogenic effector T cells. In an effort to further understand the nature of the CD40-CD40L interaction responsible for promotion of EAE, Becher and colleagues used a bone marrow reconstitution system to determine which CD40-expressing cells were responsible for promoting EAE [57]. In that report, the authors showed that CD40 expression by parenchymal microglia was responsible for recruitment/retention of encephalitogenic T cells in EAE. Strikingly, treatment of microglia with a combination of granulocyte macrophage-colony stimulating factor and CD40L has been shown to promote differentiation of these cells into cells that (1) express the pan-DC marker CD11c, (2) morphologically resemble DCs, and (3) secrete the Th1-promoting cytokine IL-12 p70 [58]. Such CD11c+ CD11b+ "DC-like" microglia were found in EAE brain lesions in inflammatory foci containing T cells, and exhibited potent stimulation of allogeneic T cell proliferation versus CD11c- CD11b+ microglia [58]. Although their origin was not determined, it was recently shown that "CNS DCs" (possibly "DC-like" microglia) are responsible for activation of naïve T cells in response to endogenous myelin epitopes (termed "epitope spreading"), and this process was initiated in the CNS as opposed to the peripheral lymphoid organs [59]. Thus, in the context of EAE, CD40-CD40L interaction on microglia seems to promote adaptive function of these cells, resulting in a "DC-like" activated microglia phenotype that promotes encephalitogenic Th1 cell differentiation and effector function.
Activation of microglia after CD40 ligation in Alzheimer disease: a shift from innate to adaptive response
Alzheimer disease and microglial responses to β-amyloid
Alzheimer disease (AD) is the most common dementia and is characterized by insidious onset in late life with progressive decline in memory and other cognitive functions. Definitive diagnosis of AD is made at autopsy, based on the neuropathological hallmarks of extracellular amyloid plaques [largely comprised of β-amyloid (Aβ) peptides, derived from the proteolysis of amyloid precursor protein (APP)] and intracellular neurofibrillary tangles. In addition, brain inflammation, characterized by reactive astrocytes and microglia (but very low levels of infiltrating T cells), is found in close vicinity of amyloid plaques in AD and in transgenic mouse models of the disease (for a review see [60]). It has been suggested that activated microglia play a key role in AD pathogenesis as they secrete pro-inflammatory innate cytokines such as TNF-α and IL-1β, which have been shown to promote neuronal injury at high levels [61,63]. Furthermore, there is a large body of evidence that non-steroidal anti-inflammatory drug (NSAID) use is associated with reduced risk for AD in humans [64-66], (for a review see [67]), and NSAID treatment of AD mice results in reduced amyloid plaque burden concomitant with ameliorated microglial activation [68-70]. Work done in Maxfield's laboratory showed that challenge of microglia with labeled Aβ peptides promotes phagocytosis but poor degradation of soluble or fibrillar Aβ via scavenger receptors [71-73]. Using knockout mice, his laboratory showed that the class A scavenger receptor (type I and II) is the predominant scavenger receptor responsible for Aβ uptake by microglia, with other scavenger receptors playing a more minor role (including the class B scavenger receptor CD36) [74].
Microglial responses to β-amyloid in the context of CD40 ligation
We previously showed that, while murine microglial challenge with soluble Aβ peptides alone does not elicit TNF-α secretion, co-stimulation provided in the form of CD40 ligation (either via CD40L or an agonistic CD40 antibody) results in TNF-α production being synergistically affected [41]. Further, microglia cultured from AD mice deficient in CD40L demonstrate reduced TNF-α secretion versus CD40L-sufficient AD mouse microglia [42]. This form of microglial activation in CD40L-sufficient AD mice is pathogenic, as CD40L-deficient AD mice demonstrate reduced activated (CD11b+) microglia, an effect that is associated with mitigated abnormal hyper-phosphorylation of tau protein (a key indicator of neuronal stress) [42]. Furthermore, genetic ablation of CD40L or administration of a CD40L-neutralizing antibody markedly reduces amyloid plaques in mouse models of AD, effects that are associated with mitigated astrocytosis and microgliosis ([75], for review see [76,77]). Recently, overproduction of microglia-associated CD40 and of astrocyte-derived CD40L was found in and around β-amyloid plaques in AD patient brain [78,79], raising the possibility that the CD40-CD40L interaction may contribute to AD pathogenesis by promoting brain inflammation.
In order to better understand the form of microglial activation affected by Aβ plus CD40L stimulation, we examined innate and adaptive activation of murine microglia challenged with Aβ in the presence or absence of CD40L co-stimulation [80]. When microglia were challenged with fluorescent-tagged synthetic human Aβ alone, they mounted a time-dependent phagocytic response (from 15 min to 60 min) which could be enhanced by Fc receptor stimulation using an anti-human Aβ antibody (clone BAM-10). This phagocytic response to Aβ alone was not associated with production of the pro-inflammatory innate cytokines TNF-α, IL-6, or IL-1β, a result similar to that seen when microglia are challenged with apoptotic cells and mount an anti-inflammatory, pro-phagocytic innate response [81]. Importantly, CD40L treatment opposed this phagocytic response, as determined by measuring both cell-associated Aβ and free extracellular Aβ. As mentioned above, Maxfield's laboratory demonstrated that microglia slowly degrade phagocytosed Aβ peptides [71-73]. We examined the ability of microglia to degrade Aβ peptides by first pulsing them with Aβ and then chasing these cells after 1 hour of culture in the presence or absence of CD40L stimulation. Using this experimental approach, we found that CD40L also retarded microglial clearance of the peptide. We further assessed putative modulation of microglial Aβ phagocytosis by cytokines known to promote effector T cell function, and found that the pro-inflammatory Th1-type cytokines IFN-γ and TNF-α inhibited Aβ phagocytosis whereas the anti-inflammatory Th2-type cytokines IL-4 and IL-10 boosted this response.
Having established that CD40 ligation attenuates innate (phagocytic) activation of microglia challenged with Aβ, we then examined the role of CD40 ligation in APC function of Aβ-treated microglia by first determining if Aβ peptides could be co-localized with MHC II. Interestingly, CD40 ligation promoted "loading" of Aβ peptides onto the MHC II molecule as determined by double immunofluorescence microscopy or immunoprecipitation assays. Finally, we determined whether this Aβ-MHC II co-localization was functional by first pre-treating microglia with Aβ in the presence or absence of CD40L, co-culturing these microglia with CD4+ T cells, and then measuring cytokine levels in co-cultured media. Interestingly, Aβ plus CD40L pre-treatment of microglia resulted in markedly enhanced levels of the Th1-promoting cytokines IL-6, TNF-α, IL-2, and IFN-γ. These effects on enhanced cytokine production could be blocked by the addition of an antagonistic CD40 antibody, confirming the requirement of the CD40-CD40L interaction per se in this phenomenon. It is interesting that another group found that IFN-γ treatment of microglia promotes APC function of these cells when they are challenged with Aβ [82]. Thus, it seems that when microglia encounter Aβ in the context of co-simulation (e.g., CD40L), their activation phenotype is biased away from innate, phagocytic activation and towards adaptive, APC function.
Microglial activation in Alzheimer disease immunotherapy: differences between mice and men
In a seminal report, Schenk and colleagues showed that peripheral immunization of the PDAPP mouse model of AD with Aβ1–42 peptide resulted in high antibody titers, a small fraction of which (0.1%, [83]) crossed the blood-brain-barrier and entered the brain parenchyma [84]. Most importantly, these authors found that Aβ1–42 vaccination markedly diminished β-amyloid plaque burden [84]. These authors also found evidence of cells in the brains of the Aβ1–42 immunized animals that contained Aβ. Many of these cells stained for the activated microglia marker MHC II and phenotypically resembled activated microglia, suggesting that these cells were able to phagocytose Aβ deposits. In a follow-up report, Bard and colleagues supported this hypothesis by showing ex vivo that certain antibodies against Aβ peptides could trigger microglial phagocytosis and subsequent clearance of Aβ through the Fc receptor [83-85]. Clearance of brain amyloid-β deposits was beneficial, as Aβ1–42-vaccinated mice had markedly reduced cognitive impairment as assayed by behavioral testing in AD mice [86,87]. Thus, in mouse models of AD, innate (phagocytic) microglial activation mediated by the Fc receptor in the presence of antibody-opsonized Aβ appears beneficial rather than deleterious.
Based on the above-mentioned data, a human clinical trial was begun to peripherally administer a synthetic Aβ1–42 peptide (AN-1792) with an adjuvant to AD patients. Unfortunately, the trial was halted when a small percentage of patients developed aseptic T cell meningoencephalitis. This response most likely occurred because of an immune reaction to Aβ mediated by infiltrating T cells [88]. In the post-mortem brain of one patient who died as a consequence of this side-effect of treatment, there was significant clearance of Aβ plaques in parts of the neocortex and, in other areas where plaques remained, Aβ-immunoreactivity was associated with microglia [89]. It is not yet clear whether this fulminate infiltration of T cells in AD patients who developed aseptic T cell meningoencephalitis was due to adaptive activation of microglia, but this is a distinct possibility given that microglia did seem to recognize antibody-opsonized Aβ [89,90]. These results indicate the potentially damaging and overwhelming effects of a full-blown T cell autoimmune response, which does not normally occur in AD, and which may have been mediated by adaptively activated microglia.
Conclusion
Accumulating evidence has revealed that microglial "activation" is not simply one phenotypic manifestation. Here, we suggest a model wherein microglial cells exist in at least two functionally discernable states once "activated", namely a phagocytic phenotype (innate activation) or an antigen presenting phenotype (adaptive activation), as governed by their stimulatory environment. When challenged with certain PAMPs (particularly CpG-DNA), murine microglia seem to activate a "mixed" response characterized by enhanced phagocytosis and pro-inflammatory cytokine production as well as adaptive activation of T cells. In the EAE model, murine microglia seem to largely support an adaptive activation of encephalitogenic T cells in the presence of the CD40-CD40 ligand interaction. In the context of Aβ challenge, CD40 ligation is able to shift activated microglia from innate to adaptive activation. Further, it seems that the cytokine milieu that microglia are exposed to biases these cells to innate activation (i.e., anti-inflammatory Th2-associated cytokines such as IL-4, IL-10, and perhaps TGF-β1) or an adaptive form of activation (i.e., pro-inflammatory Th1-associated cytokines such as IFN-γ, IL-6, and TNF-α; summarized in Fig. 2). Not all forms of microglial activation are deleterious, as activated microglia may serve a protective role as was shown in Aβ1–42-immunized mouse models of AD. It seems that enhanced microglial phagocytosis of β-amyloid plaques is at least partly responsible for the therapeutic benefit in these animals, so perhaps stimulation of innate microglial activation contributes to these reported benefits. In conclusion, if we can learn how to better harness microglia in order to produce specific forms of microglial activation, this could be key in turning a pathogenic cell into a therapeutic modality.
Figure 2 Model for innate versus adaptive microglial activation responses. In the context of β-amyloid challenge, microglia activate a phagocytic response. If co-stimulated with CD40 ligand, a shift from innate activation to adaptive antigen-presenting cell response ensues. Additionally, certain anti-inflammatory Th2-type cytokines shift this balance back towards innate phagocytic response, while some pro-inflammatory Th1-associated cytokines tip the balance further towards adaptive activation of microglia. See the text and Table 1 for references. Abbreviations used: APC, antigen presenting cell; CD40L, CD40 ligand; Th1, CD4+ T helper cell type I response; Th2, Th type II response; TGF, transforming growth factor; IL, interleukin; IFN, interferon, TNF, tumor necrosis factor.
Competing interests
The author(s) declare that they have no completing interests.
Authors' contributions
T.T. provided an initial outline of the areas to be covered. V.N. and J.T. wrote the first draft. T.T. performed the experiments described in Fig. 1. V.N. and T.T. edited the references. T.T. and J.T. revised and edited the final manuscript.
Acknowledgements
This work is supported by grants from the NIH/NINDS (to J. Tan). T. Town is supported by a Ruth L. Kirschstein NIH/NRSA/NIA post-doctoral fellowship and an Alzheimer Association Investigator-Initiated Research Grant. We thank K. Townsend (Department of Pharmacology, Center for Experimental Therapeutics, University of Pennsylvania) for helpful discussion.
==== Refs
Pessac B Godin I Alliot F [Microglia: origin and development] Bull Acad Natl Med 2001 185 337 346 discussion 346-337 11474589
Alliot F Godin I Pessac B Microglia derive from progenitors, originating from the yolk sac, and which proliferate in the brain Brain Res Dev Brain Res 1999 117 145 152 10567732 10.1016/S0165-3806(99)00113-3
Eglitis MA Mezey E Hematopoietic cells differentiate into both microglia and macroglia in the brains of adult mice Proc Natl Acad Sci USA 1997 94 4080 4085 9108108 10.1073/pnas.94.8.4080
Brazelton TR Rossi FM Keshet GI Blau HM From marrow to brain: expression of neuronal phenotypes in adult mice Science 2000 290 1775 1779 11099418 10.1126/science.290.5497.1775
Mezey E Chandross KJ Harta G Maki RA McKercher SR Turning blood into brain: cells bearing neuronal antigens generated in vivo from bone marrow Science 2000 290 1779 1782 11099419 10.1126/science.290.5497.1779
Priller J Flugel A Wehner T Boentert M Haas CA Prinz M Fernandez-Klett F Prass K Bechmann I de Boer BA Targeting gene-modified hematopoietic cells to the central nervous system: use of green fluorescent protein uncovers microglial engraftment Nat Med 2001 7 1356 1361 11726978 10.1038/nm1201-1356
Qureshi ST Medzhitov R Toll-like receptors and their role in experimental models of microbial infection Genes Immun 2003 4 87 94 12618855 10.1038/sj.gene.6363937
Yamamoto M Takeda K Akira S TIR domain-containing adaptors define the specificity of TLR signaling Mol Immunol 2004 40 861 868 14698224 10.1016/j.molimm.2003.10.006
Zhang D Zhang G Hayden MS Greenblatt MB Bussey C Flavell RA Ghosh S A toll-like receptor that prevents infection by uropathogenic bacteria Science 2004 303 1522 1526 15001781 10.1126/science.1094351
Tabeta K Georgel P Janssen E Du X Hoebe K Crozat K Mudd S Shamel L Sovath S Goode J Toll-like receptors 9 and 3 as essential components of innate immune defense against mouse cytomegalovirus infection Proc Natl Acad Sci USA 2004 101 3516 3521 14993594 10.1073/pnas.0400525101
Janeway CA JrMedzhitov R Innate immune recognition Annu Rev Immunol 2002 20 197 216 11861602 10.1146/annurev.immunol.20.083001.084359
Medzhitov R Janeway C Jr The Toll receptor family and microbial recognition Trends Microbiol 2000 8 452 456 11044679 10.1016/S0966-842X(00)01845-X
Medzhitov R Janeway CA Jr How does the immune system distinguish self from nonself? Semin Immunol 2000 12 185 188 discussion 257-344 10910738 10.1006/smim.2000.0230
Medzhitov R Janeway CA Jr Decoding the patterns of self and nonself by the innate immune system Science 2002 296 298 300 11951031 10.1126/science.1068883
Iwasaki A Medzhitov R Toll-like receptor control of the adaptive immune responses Nat Immunol 2004 5 987 995 15454922 10.1038/ni1112
Goldsby R Kindt T Osborne B Kuby J Goldsby R Mononuclear Phagocytes Immunology 2002 5 New York: Freeman and Co 38 19 11972630
Adler H Peterhans E Jungi TW Generation and functional characterization of bovine bone marrow-derived macrophages Vet Immunol Immunopathol 1994 41 211 227 7941305 10.1016/0165-2427(94)90098-1
Blander JM Medzhitov R Regulation of phagosome maturation by signals from toll-like receptors Science 2004 304 1014 1018 15143282 10.1126/science.1096158
Forman HJ Torres M Redox signaling in macrophages Mol Aspects Med 2001 22 189 216 11679166 10.1016/S0098-2997(01)00010-3
Tsukada N Miyagi K Matsuda M Yanagisawa N Expression of Fc epsilon R2/CD23 and p55 IL-2R/CD25 on peripheral blood macrophages/monocytes in multiple sclerosis J Neuroimmunol 1994 55 127 133 7530259 10.1016/0165-5728(94)90002-7
Blom AB Radstake TR Holthuysen AE Sloetjes AW Pesman GJ Sweep FG van de Loo FA Joosten LA Barrera P van Lent PL van den Berg WB Increased expression of Fcgamma receptors II and III on macrophages of rheumatoid arthritis patients results in higher production of tumor necrosis factor alpha and matrix metalloproteinase Arthritis Rheum 2003 48 1002 1014 12687542 10.1002/art.10871
Gregory CD Devitt A The macrophage and the apoptotic cell: an innate immune interaction viewed simplistically? Immunology 2004 113 1 14 15312130 10.1111/j.1365-2567.2004.01959.x
Fujiwara N Kobayashi K Macrophages in inflammation Curr Drug Targets Inflamm Allergy 2005 4 281 286 16101534 10.2174/1568010054022024
Banchereau J Briere F Caux C Davoust J Lebecque S Liu YJ Pulendran B Palucka K Immunobiology of dendritic cells Annu Rev Immunol 2000 18 767 811 10837075 10.1146/annurev.immunol.18.1.767
Shortman K Liu YJ Mouse and human dendritic cell subtypes Nat Rev Immunol 2002 2 151 161 11913066 10.1038/nri746
Hornung V Rothenfusser S Britsch S Krug A Jahrsdorfer B Giese T Endres S Hartmann G Quantitative expression of toll-like receptor 1–10 mRNA in cellular subsets of human peripheral blood mononuclear cells and sensitivity to CpG oligodeoxynucleotides J Immunol 2002 168 4531 4537 11970999
Jarrossay D Napolitani G Colonna M Sallusto F Lanzavecchia A Specialization and complementarity in microbial molecule recognition by human myeloid and plasmacytoid dendritic cells Eur J Immunol 2001 31 3388 3393 11745357 10.1002/1521-4141(200111)31:11<3388::AID-IMMU3388>3.0.CO;2-Q
Kadowaki N Ho S Antonenko S Malefyt RW Kastelein RA Bazan F Liu YJ Subsets of human dendritic cell precursors express different toll-like receptors and respond to different microbial antigens J Exp Med 2001 194 863 869 11561001 10.1084/jem.194.6.863
Ito T Amakawa R Kaisho T Hemmi H Tajima K Uehira K Ozaki Y Tomizawa H Akira S Fukuhara S Interferon-alpha and interleukin-12 are induced differentially by Toll-like receptor 7 ligands in human blood dendritic cell subsets J Exp Med 2002 195 1507 1512 12045249 10.1084/jem.20020207
Hemmi H Kaisho T Takeda K Akira S The roles of Toll-like receptor 9, MyD88, and DNA-dependent protein kinase catalytic subunit in the effects of two distinct CpG DNAs on dendritic cell subsets J Immunol 2003 170 3059 3064 12626561
Olson JK Miller SD Microglia initiate central nervous system innate and adaptive immune responses through multiple TLRs J Immunol 2004 173 3916 3924 15356140
Bsibsi M Ravid R Gveric D van Noort JM Broad expression of Toll-like receptors in the human central nervous system J Neuropathol Exp Neurol 2002 61 1013 1021 12430718
Kielian T Mayes P Kielian M Characterization of microglial responses to Staphylococcus aureus: effects on cytokine, costimulatory molecule, and Toll-like receptor expression J Neuroimmunol 2002 130 86 99 12225891 10.1016/S0165-5728(02)00216-3
Kielian T Esen N Bearden ED Toll-like receptor 2 (TLR2) is pivotal for recognition of S. aureus peptidoglycan but not intact bacteria by microglia Glia 2005 49 567 576 15593098 10.1002/glia.20144
Iliev AI Stringaris AK Nau R Neumann H Neuronal injury mediated via stimulation of microglial toll-like receptor-9 (TLR9) Faseb J 2004 18 412 414 14688201
Dalpke AH Schafer MK Frey M Zimmermann S Tebbe J Weihe E Heeg K Immunostimulatory CpG-DNA activates murine microglia J Immunol 2002 168 4854 4863 11994434
Lehnardt S Lachance C Patrizi S Lefebvre S Follett PL Jensen FE Rosenberg PA Volpe JJ Vartanian T The toll-like receptor TLR4 is necessary for lipopolysaccharide-induced oligodendrocyte injury in the CNS J Neurosci 2002 22 2478 2486 11923412
Lehnardt S Massillon L Follett P Jensen FE Ratan R Rosenberg PA Volpe JJ Vartanian T Activation of innate immunity in the CNS triggers neurodegeneration through a Toll-like receptor 4-dependent pathway Proc Natl Acad Sci USA 2003 100 8514 8519 12824464 10.1073/pnas.1432609100
Alexopoulou L Holt AC Medzhitov R Flavell RA Recognition of double-stranded RNA and activation of NF-kappaB by Toll-like receptor 3 Nature 2001 413 732 738 11607032 10.1038/35099560
Jack CS Arbour N Manusow J Montgrain V Blain M McCrea E Shapiro A Antel JP TLR Signaling Tailors Innate Immune Responses in Human Microglia and Astrocytes J Immunol 2005 175 4320 4330 16177072
Wang T Town T Alexopoulou L Anderson JF Fikrig E Flavell RA Toll-like receptor 3 mediates West Nile virus entry into the brain causing lethal encephalitis Nat Med 2004 10 1366 1373 15558055 10.1038/nm1140
Tan J Town T Paris D Mori T Suo ZM Crawford F Mattson MP Flavell RA Mullan M Microglial activation resulting from CD40-CD40L interaction after beta-amyloid stimulation Science 1999 286 2352 2355 10600748 10.1126/science.286.5448.2352
Vogel SN Fitzgerald KA Fenton MJ TLRs: differential adapter utilization by toll-like receptors mediates TLR-specific patterns of gene expression Mol Interv 2003 3 466 477 14993454 10.1124/mi.3.8.466
Hemmi H Akira S TLR signalling and the function of dendritic cells Chem Immunol Allergy 2005 86 120 135 15976491
Olsson T Cytokine-producing cells in experimental autoimmune encephalomyelitis and multiple sclerosis Neurology 1995 45 S11 15 7540264
Swanborg RH Experimental autoimmune encephalomyelitis in rodents as a model for human demyelinating disease Clin Immunol Immunopathol 1995 77 4 13 7554482 10.1016/0090-1229(95)90130-2
Cornet A Vizler C Liblau R [Experimental autoimmune encephalomyelitis] Rev Neurol (Paris) 1998 154 586 591 9809373
van Kooten C Banchereau J CD40-CD40 ligand J Leukoc Biol 2000 67 2 17 10647992
Grewal IS Flavell RA CD40 and CD154 in cell-mediated immunity Annu Rev Immunol 1998 16 111 135 9597126 10.1146/annurev.immunol.16.1.111
Nguyen VT Walker WS Benveniste EN Post-transcriptional inhibition of CD40 gene expression in microglia by transforming growth factor-beta Eur J Immunol 1998 28 2537 2548 9710231
Carson MJ Reilly CR Sutcliffe JG Lo D Mature microglia resemble immature antigen-presenting cells Glia 1998 22 72 85 9436789 10.1002/(SICI)1098-1136(199801)22:1<72::AID-GLIA7>3.0.CO;2-A
Havenith CE Askew D Walker WS Mouse resident microglia: isolation and characterization of immunoregulatory properties with naive CD4+ and CD8+ T-cells Glia 1998 22 348 359 9517567 10.1002/(SICI)1098-1136(199804)22:4<348::AID-GLIA4>3.0.CO;2-#
Tan J Town T Paris D Placzek A Parker T Crawford F Yu H Humphrey J Mullan M Activation of microglial cells by the CD40 pathway: relevance to multiple sclerosis Journal of Neuroimmunology 1999 97 77 85 10408982 10.1016/S0165-5728(99)00053-3
Gerritse K Laman JD Noelle RJ Aruffo A Ledbetter JA Boersma WJ Claassen E CD40-CD40 ligand interactions in experimental allergic encephalomyelitis and multiple sclerosis Proc Natl Acad Sci USA 1996 93 2499 2504 8637903 10.1073/pnas.93.6.2499
Grewal IS Foellmer HG Grewal KD Xu J Hardardottir F Baron JL Janeway CA JrFlavell RA Requirement for CD40 ligand in costimulation induction, T cell activation, and experimental allergic encephalomyelitis Science 1996 273 1864 1867 8791592
Howard LM Miga AJ Vanderlugt CL Dal Canto MC Laman JD Noelle RJ Miller SD Mechanisms of immunotherapeutic intervention by anti-CD40L (CD154) antibody in an animal model of multiple sclerosis J Clin Invest 1999 103 281 290 9916140
Becher B Durell BG Miga AV Hickey WF Noelle RJ The clinical course of experimental autoimmune encephalomyelitis and inflammation is controlled by the expression of CD40 within the central nervous system J Exp Med 2001 193 967 974 11304557 10.1084/jem.193.8.967
Fischer HG Reichmann G Brain dendritic cells and macrophages/microglia in central nervous system inflammation J Immunol 2001 166 2717 2726 11160337
McMahon EJ Bailey SL Castenada CV Waldner H Miller SD Epitope spreading initiates in the CNS in two mouse models of multiple sclerosis Nat Med 2005 11 335 339 15735651 10.1038/nm1202
Akiyama H Barger S Barnum S Bradt B Bauer J Cole GM Cooper NR Eikelenboom P Emmerling M Fiebich BL Inflammation and Alzheimer's disease Neurobiol Aging 2000 21 383 421 10858586 10.1016/S0197-4580(00)00124-X
Meda L Cassatella MA Szendrei GI Otvos L JrBaron P Villalba M Ferrari D Rossi F Activation of microglial cells by beta-amyloid protein and interferon-gamma Nature 1995 374 647 650 7715705 10.1038/374647a0
Barger SW Harmon AD Microglial activation by Alzheimer amyloid precursor protein and modulation by apolipoprotein E Nature 1997 388 878 881 9278049 10.1038/42257
McGeer EG McGeer PL The importance of inflammatory mechanisms in Alzheimer disease Exp Gerontol 1998 33 371 378 9762518 10.1016/S0531-5565(98)00013-8
Stewart WF Kawas C Corrada M Metter EJ Risk of Alzheimer's disease and duration of NSAID use Neurology 1997 48 626 632 9065537
in t' Veld BA Ruitenberg A Hofman A Launer LJ van Duijn CM Stijnen T Breteler MM Stricker BH Nonsteroidal antiinflammatory drugs and the risk of Alzheimer's disease N Engl J Med 2001 345 1515 1521 11794217 10.1056/NEJMoa010178
Zandi PP Anthony JC Hayden KM Mehta K Mayer L Breitner JC Reduced incidence of AD with NSAID but not H2 receptor antagonists: the Cache County Study Neurology 2002 59 880 886 12297571
Szekely CA Thorne JE Zandi PP Ek M Messias E Breitner JC Goodman SN Nonsteroidal anti-inflammatory drugs for the prevention of Alzheimer's disease: a systematic review Neuroepidemiology 2004 23 159 169 15279021 10.1159/000078501
Lim GP Yang F Chu T Chen P Beech W Teter B Tran T Ubeda O Ashe KH Frautschy SA Cole GM Ibuprofen suppresses plaque pathology and inflammation in a mouse model for Alzheimer's disease Journal of Neuroscience 2000 20 5709 5714 10908610
Lim GP Yang F Chu T Gahtan E Ubeda O Beech W Overmier JB Hsiao-Ashe K Frautschy SA Cole GM Ibuprofen effects on Alzheimer pathology and open field activity in APPsw transgenic mice Neurobiology of Aging 2001 22 983 991 11755007 10.1016/S0197-4580(01)00299-8
Lim GP Chu T Yang FS Beech W Frautschy SA Cole GM The curry spice curcumin reduces oxidative damage and amyloid pathology in an Alzheimer transgenic mouse Journal of Neuroscience 2001 21 8370 8377 11606625
Paresce DM Ghosh RN Maxfield FR Microglial cells internalize aggregates of the Alzheimer's disease amyloid beta-protein via a scavenger receptor Neuron 1996 17 553 565 8816718 10.1016/S0896-6273(00)80187-7
Paresce DM Chung H Maxfield FR Slow degradation of aggregates of the Alzheimer's disease amyloid beta-protein by microglial cells J Biol Chem 1997 272 29390 29397 9361021 10.1074/jbc.272.46.29390
Brazil MI Chung H Maxfield FR Effects of incorporation of immunoglobulin G and complement component C1q on uptake and degradation of Alzheimer's disease amyloid fibrils by microglia J Biol Chem 2000 275 16941 16947 10747968 10.1074/jbc.M000937200
Chung H Brazil MI Irizarry MC Hyman BT Maxfield FR Uptake of fibrillar beta-amyloid by microglia isolated from MSR-A (type I and type II) knockout mice Neuroreport 2001 12 1151 1154 11338182 10.1097/00001756-200105080-00020
Tan J Town T Crawford F Mori T DelleDonne A Crescentini R Obregon D Flavell RA Mullan MJ Role of CD40 ligand in amyloidosis in transgenic Alzheimer's mice Nat Neurosci 2002 5 1288 1293 12402041 10.1038/nn968
Town T Tan J Mullan M CD40 signaling and Alzheimer's disease pathogenesis Neurochem Int 2001 39 371 380 11578772 10.1016/S0197-0186(01)00044-4
Tan J Town T Mullan M CD40-CD40L interaction in Alzheimer's disease Curr Opin Pharmacol 2002 2 445 451 12127879 10.1016/S1471-4892(02)00180-7
Togo T Akiyama H Kondo H Ikeda K Kato M Iseki E Kosaka K Expression of CD40 in the brain of Alzheimer's disease and other neurological diseases Brain Res 2000 885 117 121 11121537 10.1016/S0006-8993(00)02984-X
Calingasan NY Erdely HA Altar CA Identification of CD40 ligand in Alzheimer's disease and in animal models of Alzheimer's disease and brain injury Neurobiol Aging 2002 23 31 39 11755016 10.1016/S0197-4580(01)00246-9
Townsend KP Town T Mori T Lue LF Shytle D Sanberg PR Morgan D Fernandez F Flavell RA Tan J CD40 signaling regulates innate and adaptive activation of microglia in response to amyloid beta-peptide Eur J Immunol 2005 35 901 910 15688347 10.1002/eji.200425585
Minghetti L Ajmone-Cat MA De Berardinis MA De Simone R Microglial activation in chronic neurodegenerative diseases: roles of apoptotic neurons and chronic stimulation Brain Res Brain Res Rev 2005 48 251 256 15850664 10.1016/j.brainresrev.2004.12.015
Monsonego A Imitola J Zota V Oida T Weiner HL Microglia-mediated nitric oxide cytotoxicity of T cells following amyloid beta-peptide presentation to Th1 cells J Immunol 2003 171 2216 2224 12928365
Bard F Cannon C Barbour R Burke RL Games D Grajeda H Guido T Hu K Huang J Johnson-Wood K Peripherally administered antibodies against amyloid beta-peptide enter the central nervous system and reduce pathology in a mouse model of Alzheimer disease Nat Med 2000 6 916 919 10932230 10.1038/78682
Schenk D Barbour R Dunn W Gordon G Grajeda H Guido T Hu K Huang J Johnson-Wood K Khan K Immunization with amyloid-beta attenuates Alzheimer-disease-like pathology in the PDAPP mouse Nature 1999 400 173 177 10408445 10.1038/22124
Bard F Barbour R Cannon C Carretto R Fox M Games D Guido T Hoenow K Hu K Johnson-Wood K Epitope and isotype specificities of antibodies to beta-amyloid peptide for protection against Alzheimer's disease-like neuropathology Proc Natl Acad Sci USA 2003 100 2023 2028 12566568 10.1073/pnas.0436286100
Janus C Pearson J McLaurin J Mathews PM Jiang Y Schmidt SD Chishti MA Horne P Heslin D French J A beta peptide immunization reduces behavioural impairment and plaques in a model of Alzheimer's disease Nature 2000 408 979 982 11140685 10.1038/35050110
Morgan D Diamond DM Gottschall PE Ugen KE Dickey C Hardy J Duff K Jantzen P DiCarlo G Wilcock D A beta peptide vaccination prevents memory loss in an animal model of Alzheimer's disease Nature 2000 408 982 985 11140686 10.1038/35050116
Pfeifer M Boncristiano S Bondolfi L Stalder A Deller T Staufenbiel M Mathews PM Jucker M Cerebral hemorrhage after passive anti-Abeta immunotherapy Science 2002 298 1379 12434053 10.1126/science.1078259
Nicoll JA Wilkinson D Holmes C Steart P Markham H Weller RO Neuropathology of human Alzheimer disease after immunization with amyloid-beta peptide: a case report Nat Med 2003 9 448 452 12640446 10.1038/nm840
Monsonego A Weiner HL Immunotherapeutic approaches to Alzheimer's disease Science 2003 302 834 838 14593170 10.1126/science.1088469
Albert ML Pearce SF Francisco LM Sauter B Roy P Silverstein RL Bhardwaj N Immature dendritic cells phagocytose apoptotic cells via alphavbeta5 and CD36, and cross-present antigens to cytotoxic T lymphocytes J Exp Med 1998 188 1359 1368 9763615 10.1084/jem.188.7.1359
Tait JF Smith C Phosphatidylserine receptors: role of CD36 in binding of anionic phospholipid vesicles to monocytic cells J Biol Chem 1999 274 3048 3054 9915844 10.1074/jbc.274.5.3048
Coraci IS Husemann J Berman JW Hulette C Dufour JH Campanella GK Luster AD Silverstein SC El-Khoury JB CD36, a class B scavenger receptor, is expressed on microglia in Alzheimer's disease brains and can mediate production of reactive oxygen species in response to beta-amyloid fibrils Am J Pathol 2002 160 101 112 11786404
Brawand P Fitzpatrick DR Greenfield BW Brasel K Maliszewski CR De Smedt T Murine plasmacytoid pre-dendritic cells generated from Flt3 ligand-supplemented bone marrow cultures are immature APCs J Immunol 2002 169 6711 6719 12471102
Kim WK Ganea D Jonakait GM Inhibition of microglial CD40 expression by pituitary adenylate cyclase-activating polypeptide is mediated by interleukin-10 J Neuroimmunol 2002 126 16 24 12020953 10.1016/S0165-5728(02)00059-0
Prilliman KR Lemmens EE Palioungas G Wolfe TG Allison JP Sharpe AH Schoenberger SP Cutting edge: a crucial role for B7-CD28 in transmitting T help from APC to CTL J Immunol 2002 169 4094 4097 12370335
Quaranta MG Tritarelli E Giordani L Viora M HIV-1 Nef induces dendritic cell differentiation: a possible mechanism of uninfected CD4(+) T cell activation Exp Cell Res 2002 275 243 254 11969293 10.1006/excr.2002.5497
Spisek R Bretaudeau L Barbieux I Meflah K Gregoire M Standardized generation of fully mature p70 IL-12 secreting monocyte-derived dendritic cells for clinical use Cancer Immunol Immunother 2001 50 417 427 11726136 10.1007/s002620100215
Wyss-Coray T Lin C Yan F Yu GQ Rohde M McConlogue L Masliah E Mucke L TGF-beta1 promotes microglial amyloid-beta clearance and reduces plaque burden in transgenic mice Nat Med 2001 7 612 618 11329064 10.1038/87945
|
16259628
|
PMC1298325
|
CC BY
|
2021-01-04 16:38:21
|
no
|
J Neuroinflammation. 2005 Oct 31; 2:24
|
utf-8
|
J Neuroinflammation
| 2,005 |
10.1186/1742-2094-2-24
|
oa_comm
|
==== Front
J NeuroinflammationJournal of Neuroinflammation1742-2094BioMed Central London 1742-2094-2-251628588810.1186/1742-2094-2-25ResearchA refined in vitro model to study inflammatory responses in organotypic membrane culture of postnatal rat hippocampal slices Huuskonen Jari [email protected] Tiina [email protected] Riitta [email protected] Groen Thomas [email protected] Antero [email protected] Department of Neuroscience and Neurology, University of Kuopio, PO Box 1627, FIN-70211 Kuopio, Finland2 Department of Cell Biology, University of Alabama at Birmingham, Birmingham, AL 35294-0006, USA3 Department of Neurology, University Hospital of Kuopio, PO Box 1627, FIN-70211 Kuopio, Finland2005 15 11 2005 2 25 25 4 11 2005 15 11 2005 Copyright © 2005 Huuskonen et al; licensee BioMed Central Ltd.2005Huuskonen 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
Propagated tissue degeneration, especially during aging, has been shown to be enhanced through potentiation of innate immune responses. Neurodegenerative diseases and a wide variety of inflammatory conditions are linked together and several anti-inflammatory compounds considered as having therapeutic potential for example in Alzheimer's disease (AD). In vitro brain slice techniques have been widely used to unravel the complexity of neuroinflammation, but rarely, has the power of the model itself been reported. Our aim was to gain a more detailed insight and understanding of the behaviour of hippocampus tissue slices in serum-free, interface culture per se and after exposure to different pro- and anti-inflammatory compounds.
Methods
The responses of the slices to pro- and anti-inflammatory stimuli were monitored at various time points by measuring the leakage of lactate dehydrogenase (LDH) and the release of cytokines interleukin 6 (IL-6) and tumour necrosis factor alpha (TNF-α) and nitric oxide (NO) from the culture media. Histological methods were applied to reveal the morphological status after exposure to stimuli and during the time course of the culture period. Statistical power analysis were made with nQuery Advisor®, version 5.0, (Statistical Solutions, Saugus, MA) computer program for Wilcoxon (Mann-Whitney) rank-sum test.
Results
By using the interface membrane culture technique, the hippocampal slices largely recover from the trauma caused by cutting after 4–5 days in vitro. Furthermore, the cultures remain stable and retain their responsiveness to inflammatory stimuli for at least 3 weeks. During this time period, cultures are susceptible to modification by inflammatory stimuli as assessed by quantitative biochemical assays and morphological characterizations.
Conclusion
The present report outlines the techniques for studying immune responses using a serum-free slice culture model. Statistically powerful data under controlled culture conditions and with ethically justified use of animals can be obtained as soon as after 4–5 DIV. The model is most probably suitable also for studies of chronic inflammation.
==== Body
Background
The discovery of upstream sensors, the Toll-like receptors (TLRs) [1,2], greatly multiplied our understanding of innate and adaptive immune interactions and responses. Downstream, a well known family of transcription factors, the nuclear factor kappa B (NF-κB), is one of the key players in the regulation of inflammatory responses [3,4]. Recent studies have revealed that this unique interplay also exists in the brain macrophages, i.e., the microglial cells [5]. These cells, which can present antigen and are responsible for the production and release of a variety of cytokines and chemokines, interact with immune cells and are intimately involved in immunoregulation within the CNS [6]. Whereas the role of microglia in the brain has been studied extensively [7-11], most progress on the understanding of the role of microglia in inflammation has come from cell culture and slice culture studies. The behaviour of microglia in different culture models has been shown to be affected by the culture time and the composition of culture media [12-14]. It has been emphasized that the presence of serum in the culture media potentiates the LPS-induced microglial response [15,16]. On the other hand, even though they exhibit amoeboid, "active" morphology under serum-free culture conditions, microglia are suggested to be functionally in an "inactive" or "resting" state [12].
In the slice culture systems, whether supplemented with serum or not, microglial cells revert to a "resting", ramified phenotype after a prolonged culture time [17,18]. This morphological transformation starts at around 4 DIV and from approximately 10 DIV on, the overall population of microglia appear for the most part as a ramified type. It has been assumed that this "resting", ramified phenotype of microglia would have reduced functional status but a recent in vivo study by Nimmerjahn and co-workers [19] convincingly demonstrates how microglia cells constantly monitor their immediate environment by extending and retracting their projections in a minute-to-minute time scale. Furthermore, time-lapse imaging of live hippocampal slices [20,21] have also revealed the capacity of microglia to undergo highly dynamic behaviour.
As these observations demonstrate, the microglia are capable of complex behaviour and therefore it is of crucial importance to pay attention to the factors that contribute to the consistency of in vitro models used to mimic in vivo situations.
In the present study, we used hippocampus tissue slices in serum-free culture conditions to examine the behaviour of microglia per se and to investigate how these slices respond to pro- and anti-inflammatory stimuli. This in vitro culture of postnatal brain provide a model where the cytoarchitecture and connectivity of different anatomical regions, as well as the functional relationships and interactions with neighbouring cell types (i.e., neurons and astrocytes) are preserved [22,23]. Organotypic cultures offer also the advantage of controlled manipulations in living tissue and thus they might represent an analogously feasible intermediate between simpler cell lines and in vivo models.
Moreover, by carefully planning the experimental set-up, it should be possible to carry out slice culture studies where a minimum number of animals need to be sacrificed and yet to gain sufficient and reliable data to eliminate the risk of performing unsuccessful, more expensive, preclinical in vivo experiments. Therefore, we also wanted to study the possibility of refining the model itself by taking into the account the experimental design involving appropriate sample size and statistical power analysis.
Methods
Preparation of hippocampal slice cultures
Organotypic slice cultures from hippocampus were prepared using the modified interface culture method described by Stoppini et al. [24]. Postnatal day 7–8 (P7–P8) Wistar rat pups were decapitated, and the brains were rapidly dissected and placed in a petri dish in ice-cold 1 × Dulbecco's Phosphate Buffered Saline without calcium and magnesium (Biowhittaker™, Belgium). The hippocampi of both sides were isolated and sectioned into 400-μm transverse slices with a McIlwain tissue chopper (Mickle Laboratory Engineering Co. Ltd, Goose Green, UK). The slices were then carefully separated and transferred on to porous membrane inserts (one slice per insert) of 12-well culture plates (Transwell TR 3462; Costar, Corning, NY, USA). To reach the level of insert membrane, some 600 μL culture medium, consisting of Neurobasal medium with 1 × B27-supplement (both from Gibco, Rockville, MD, USA), 1 mM L-glutamine, 100 U/mL penicillin and 100 μg/mL streptomycin, was added to the lower compartment of each well and the culture plates were then placed in a 37°C humified incubator enriched with 5% CO2. On the first day of culture, inactivated fetal bovine serum (FBSi, Gibco) was added to the culture medium at a concentration of 10%. On the next day, the culture medium was replaced with fresh medium without serum and from then on serum-free media was changed twice a week.
Animals were obtained from National Laboratory Animal Center (NLAC), University of Kuopio, Finland. The experiments were conducted according to the Council of Europe (Directive 86/609) and Finnish guidelines, and approved by the State Provincial Office of Eastern Finland.
Exposure of slices to different stimuli
A series of experiments was carried out to extend the culture time up to 1 month. First, to induce inflammation, we exposed the slices to 5 μg/ml of lipopolysaccharide (LPS from E.coli 055:B5, L6529, Sigma, St Louis, USA) for 24 h at 4 DIV, 7 DIV, 14 DIV and 21 DIV. Then we performed a concentration series study in order to determine whether the amount of LPS had any effect on cytokine production and LDH leakage at the 4 DIV time point. Next, the slices were exposed either to proinflammatory or anti-inflammatory compounds together with or without LPS induction at 4 DIV. As a proinflammatory stimulus we used trichostatin A (TSA, 20 nM), a well characterised histone deacetylase inhibitor [25] and as an anti-inflammatory stimulus we used a known NF-κB inhibitor helenalin [26] at a concentration of 0.5 μM. TSA was purchased from Sigma and helenalin from BIOMOL Research Laboratories (Plymouth Meeting, PA, USA). We also pre-exposed slices to 5 μg/ml of LPS for 3 h, 6 h, 12 h, 24 h and 48 h to determine the possible time dependent effects either at 4 DIV or 7 DIV. In all experiments, the results were collected from parallel cultures with six individual slices per treatment group. Each study was replicated at least three times.
Histological methods
For revealing microglia, hippocampal slices were stained with Alexa Fluor 488 conjugated fluorescent Griffonia simplicifolia isolectin IB4 (Molecular Probes, Eugene, OR). Prior to staining slices were fixed with 4% paraformaldehyde for 1–2 h and rinsed three times with 0.05 M TBS-T, pH 7.6 (Tris-buffered saline + 0.1% Triton X-100). IB4 was applied at a concentration of 0.5 μg/ml in TBS-T and slices were incubated overnight at 4°C on a shaker. Before visualization, samples were rinsed with 0.1 M phosphate buffer and mounted on slides. We also performed immunocytochemical staining with a microglia marker OX-42 (against CD11b surface Ag, MCA 275R, Serotec, Oxford, UK), an antibody that recognizes type 3 complement receptors CR3 on mononuclear phagocytes. Primary antibody was diluted 1:20 000 with 0.05 M TBS-T, pH 7.6, slices were incubated for 1 week at 4°C on a shaker, rinsed thoroughly, followed by overnight incubation (4°C) with biotinylated secondary Ab (sheep anti-mouse, 1:1000, Serotec) and 2 h incubation (room temperature) with avidin peroxidase (1:1000 ExtrAvidin E-2886, Sigma). The immunoreactive product was visualised with 3,3'-diaminobenzidine tetrahydrochloride dihydrate (DAB, 0.5 mg/ml, Sigma) in a nickel solution containing hydrogen peroxidase (25 μg/ml).
For epifluorescence immunodetection, primary antibodies to glial fibrillary acidic protein for astrocytes (GFAP: dilution 1:10 000; DAKO, Denmark) and doublecortin for neurons (DCX C-18: dilution 1:2500; Santa Cruz Laboratories, Santa Cruz, CA, USA) were used. After an overnight incubation with primary antibody at 4°C and thorough rinsing, 1:1000 diluted goat anti-rabbit Alexa Fluor 488 secondary antibody (Molecular Probes) to GFAP and 1:500 diluted biotinylated rabbit anti-goat secondary (Vector BA-5000; Burlingame, CA, USA) followed by tertiary antibody (goat anti-rabbit Alexa Fluor 488; 1:1000 dilution) to DCX were applied, respectively, also overnight at 4°C. All the antibodies were diluted with 0.05 M TBS-T, pH 7.6.
Monitoring morphological and biochemical recovery and responses
First, some hippocampi were rapidly sliced at 400 μm and immediately fixed and stained for microglia, astrocytes and neurons with IB4, GFAP and DCX, respectively, to reveal the morphological status at the beginning of the culture. Then, the culture medium was collected and tissues fixed at 24 h intervals during the 7 days in vitro (7 DIV) culture period. Cytokines IL-6 and TNF-α, NO and LDH were analysed from the media and tissue samples were stained as described before. Subsequently, samples of slices and culture medium were collected at appropriate time points to further monitor the morphological and biochemical status of cultures in the 1 month time period. To visualize the morphological integrity and both dead or dying cells and living cells, we used standard Nissl staining and the Live/dead-cytotoxicity kit L-3224 (Molecular Probes), respectively, according to the manufacturer's protocol.
LDH leakage to the culture medium was measured with a CytoTox 96 nonradioactive cytotoxicity assay kit obtained from Promega (Madison, WI, USA). The nitrite concentration in the medium was measured by the Griess reaction. Briefly, to a 100-μL of sample an equal amount of the Griess reagent (1:1 0.1% naphthylethylene diamide in H2O and 1% sulfanilamide in 5% concentrated H2PO4) was added and the optical density (OD) was measured at 550 nm using an ELISA microplate reader after incubation for 10 min. The concentrations of cytokines IL-6 and tumour necrosis factor (TNF)-a released into the medium were measured by an enzyme linked immunosorbent assay (ELISA) using OptEIA™ kits or sets obtained from Pharmingen (BD Biosciences, San Diego, CA, USA).
Pictures from stained slices were collected with an Olympus DP50 microscope digital camera system connected to an Olympus BX40 microscope (Olympus Optical Co, Ltd, Japan) with appropriate filters. Except for adjustment to the contrast and brightness levels, no other manipulations were done in any of the images.
Statistical power analysis and estimation of optimal sample size
To determine the minimum number of animals, yet appropriate sample sizes for our experiments, we undertook power analysis calculations. First, based on our previous results, we estimated the effect size among different experimental units. Thereafter, the sample size was set to six per group and the significance level to 5% to see the effect on statistical power. In order to reduce the within-group variation we used lognormal distribution and carried out the statistical power analysis calculations using nQuery Advisor®, version 5.0, (Statistical Solutions, Saugus, MA) computer program for Wilcoxon (Mann-Whitney) rank-sum test.
Results
Slices recover from explantation by 4 DIV
First, we wanted to know how well hippocampal slices would recover from the trauma caused by the isolation procedure. Therefore, we collected the medium after every 24 h during 7 DIV and used standardised protocols to measure the secretion of IL-6, TNF-α and the leakage of LDH. The secreted level of IL-6 was highest at 48 h after explantation and returned to the basal level by 4 DIV (Fig. 1A). The peak levels of LDH and TNF-α were recorded at 1 DIV and reverted to control levels by 3 DIV (Fig. 1B &1C). From that time, the levels of cytokines and LDH exhibited some variability but remained at rather low levels. The Live/dead-assay showed that there were numerous dead/dying cells during the first days of culture (not shown) but gradually the number of dead/dying cells decreased and at 4 weeks in vitro mostly live cells were visible (Fig. 2A–C)
Figure 1 Biochemical recovery of P7 hippocampus slices after explantation. Culture medium was collected at 24 h intervals during the 7 DIV culture period. Values are means ± SD (n = 6 in each group).
Figure 2 Live/dead-assay showing the viability of the P7 hippocampus slice cultured for 4 weeks in serum-free media. Green fluorescence (A) is an indicator of live cells and red (B) indicates the dead-cell population. An overlay (C) shows that mainly live cells are present. Scale bar 200 μm.
Effect of LPS is not concentration dependent
It has been shown that microglia express Toll-like receptor 4 (TLR4) [27,28], which mediates the LPS induced intracellular NF-κB signaling pathway and evokes the release of cytokines. To investigate whether the concentration of LPS had any effect on the slices during the 24 h exposure, we pre-exposed the cultures to 0.1, 0.5, 1, 5 and 10 μg/ml of LPS at day 4. The LDH leakage did not differ significantly at any of these LPS-concentration levels and the control slices showed only minimally lower level of secretion than LPS treated slices (Fig. 3A). The NO and IL-6 levels were clearly higher compared to control slices but LPS did not induce any prominent concentration-dependent effect at any treatment groups (Fig. 3B &3C).
Figure 3 Effect of the dose of LPS on the inflammatory response in P7 hippocampus slices. LPS was added to the medium after 4 DIV and the exposure time was 24 h. LDH secretion remained close to control levels in all LPS groups (A). NO and IL-6 secretion increased prominently already at the dose of 0.1 μg/ml but there was no significant dose-dependent difference in any treatment group. Values are means ± SD (n = 6 in each group).
Effect of LPS is exposure but not culture time dependent
To address the question of whether the slice cultures respond to LPS differently during the culture time, we added LPS (5 μg/ml) to the medium for 24 h at days 7, 14 and 21 in vitro. LPS evoked extensive secretion of IL-6 at all the time points (Fig. 4), thus reflecting the capability of slices to respond to inflammatory stimuli with similar manner as at 4 DIV.
Figure 4 Effect of culture time prior to LPS exposure on LPS-induced IL-6 response. LPS was added to the medium after 7, 14 and 21 DIV; exposure time was 24 h. LPS evoked extensive secretion of IL-6 at all of the time points. Values are means ± SD (n = 6 in each group).
We also determined the temporal profile of LPS response by measuring the levels of IL-6, TNF-α, NO and LDH after 3 h, 6 h, 12 h, 24 h and 48 h exposures to 5 μg/ml of LPS at DIV 4 and 7. The secreted level of IL-6 was clearly increased after 6 hours and the highest levels were found after 48 hours both at 4 DIV and 7 DIV (Fig. 5A). TNF-α secretion was already prominently higher at 3 hours, continued rising at 6 hours and was highest after 12 hours of exposure at 4 DIV. The pattern of TNF-α secretion at 7 DIV was similar to that seen at 4 DIV, but the overall levels were significantly lower and the value after 48 hours was higher than the value of 24 hour exposure (Fig. 5B). The NO levels at 7 DIV increased by degrees at 3 h, 12 h and 48 h time points but at 4 DIV ascended more evenly (Fig. 5D). At 7 DIV a similar elevation, as seen in NO levels, was seen after 12 h when LDH was measured from the same samples. The 4 DIV levels of LDH compared to those at 7 DIV were approximately three- to four-fold higher at 3 and 6 hour time points and two-fold higher after 12 hour exposure to LPS (Fig. 5C).
Figure 5 Effect of LPS exposure time at 4 DIV and 7 DIV. Medium was collected after 3, 6, 12, 24 and 48 h of exposure to LPS. Values are means ± SD (n = 6 in each group).
Slices respond well to pro and anti-inflammatory stimuli during the culture
Next, we wanted to see if either proinflammatory histone deacetylase inhibitor TSA or anti-inflammatory NF-κB inhibitor helenalin could influence the LPS-induced response. Interestingly, TSA significantly potentiated the IL-6 response of LPS and helenalin seemed to downregulate the response to both LPS and LPS/TSA exposure after 24 hour exposure at DIV 4. Helenalin also downregulated the nitric oxide levels when combined with LPS alone and together with LPS and TSA (Fig. 6A &6B).
Figure 6 Potentiation and downregulation of LPS-induced inflammatory response by TSA and helenalin. Exposure time was 24 h at 4 DIV. LPS concentration was 5 μg/ml, TSA concentration was 20 nM and that of helenalin 0.5 μM. Values are means ± SD (n = 6 in each group). * p < 0.05 (Mann-Whitney U test).
Morphological changes of microglia, neurons and astrocytes during culture period
At the beginning of culture, immediately after sectioning, microglia displayed an idiotypical "resting", ramified morphology with small cell bodies and numerous branching processes (Fig. 7A). After 1 DIV, the microglia started to revert gradually into an intermediate, "reactive" form with larger cell bodies and several thicker branches (Fig. 7B) and by 4–5 DIV a number of the IB4-positive cells appeared as the characteristic rounded, "amoeboid" phenotype, though also pleomorphic cells exhibiting projections were visible (Fig. 7C &7D). The vascular endothelium and the amoeboid cells with protruding filopodia were also labeled by IB4. As the culture time extended, the morphological polymorphism of microglial-cell population continued, e.g., all shapes of microglia from "amoeboid" like to "resting", ramified were found in all of the IB4 labeled slices. The same phenomenon was seen when the slices were stained with OX-42. Morphologically active "phagocytic" cells were found in clusters mainly in the cornu ammonis of hippocampus, and ramified cells were scattered throughout the slice with no specific localisation. Despite the differences in cytokine secretion, microglia exhibited a heterogenous morphology also in all the experimental groups (Fig. 8A–B.)
Figure 7 Labeling of microglia with Alexa Fluor 488 conjugated lectin IB4. (A) Hippocampus slice of P7 rat stained immediately after sectioning. Microglia demonstrate typical "resting", ramified morphology with small cell bodies and branching processes (arrows). (B) After 1 DIV, the microglia started to revert to a more intermediate "reactive" form with larger cell bodies and thicker branches (arrow) and by 4 DIV a number of cells appeared with a rounded "amoeboid" phenotype (C, arrows), but also more pleomorphic cells with projections were visible (C&D, dashed arrows). Note also labeling of vascular endothelium (arrowheads in A and C). Scale bar equals 50 μm.
Figure 8 Hippocampus slice stained with OX-42 at 8 DIV after 24 h exposure to 5 μg/ml of LPS. The heterogenous morphology of microglia continued despite the extensive culture time and LPS exposure (A-C, arrows). A, "resting", ramified microglia, B, "reactive", activated microglia, and C, "phagocytic", macrophage microglia. Scale bar equals 50 μm.
GFAP staining revealed the characteristic star-shaped morphology of astrocytes immediately after the explantation. In addition, the expansions of astrocytic processes, the "end-feet", enveloping the microvessels were clearly visible (Fig. 9A), thus mimicking the situation in vivo. After 2 DIV, GFAP-positive cells started to transform into fibrous cells with long processes (Fig. 9B), eventually extending throughout the whole slice by 7 DIV (Fig. 9C). As the culture time lengthened, astrocytes remained throughout the slice displaying the fibrous and star-shaped morphology (Fig. 10A–C). The dentate area remained prominently covered with cells expressing doublecortin, showing the typical morphology of neurons (Fig. 11A and 11B; inserts). DCX-positive cells were found, though in lesser numbers, also in the CA-areas. Regardless of LPS or LPS/TSA exposure either at 7 or 14 DIV, both astrocytes and neurons showed a similar staining pattern as control slices, as judged from the morphological point of view (Fig. 11A–D). Also the overall morphological integrity of the slices remained mostly well preserved (Fig. 12A–C), i.e. the neuronal layers could be recognized despite the different treatments.
Figure 9 Astrocytes in P7 hippocampus slice stained with a GFAP antibody. (A) Immediately after sectioning, astrocytes displayed a star-shaped phenotype (arrows), also the "end-feet" enveloping microvessels were clearly visible (dashed arrows). (B) After 2 DIV the GFAP-positive cells started to transform into fibrous cells with long processes and (C) by 5 DIV, astrocytes with long processes covered the whole slice, with the semblance of an astrocytic scar. Scale bars in A and B equal 50 μm, C = 100 μm.
Figure 10 P7 hippocampus slice cultured for 14 DIV and stained with a GFAP antibody for astrocytes. (A) GFAP-positive cells continued to extend throughout the whole tissue (B) displaying the fibrous long processes and (C) star-shaped morphology through the slice. Scale bars: A = 200 μm, B = 100 μm and C = 50 μm.
Figure 11 Neurons stained with DCX and astrocytes with GFAP. (A) DCX-positive cells in the dentate gyrus of P7 hippocampus slice treated for 24 h with LPS (5 μg/ml) and (B) LPS and TSA (20 nM) at 7 DIV. The treatments had no visible effect on morphology (inserts in A & B, scale bars equal 50 μm), staining intensities or number of labeled neurons. (C) Astrocytes in the CA-area of control slice and (D) LPS treated slice after 14 DIV. Similar long processes were visible in both groups as at 5 DIV (Fig. 9C). Scale bars: A and B = 100 μm, C and D = 50 μm.
Figure 12 Nissl stained sections demonstrating the overall morphological integrity of the P7 hippocampus slices cultured in serum-free media. (A) Control slice cultured for 5 DIV. (B) LPS (5 μg/ml/24 h) treated slice at 5 DIV. (C) LPS (5 μg/ml/24 h) and TSA (20 nM/24 h) treated slice at 14 DIV. The neuronal layers could be recognized despite the different treatments. DG = dentate gyrus, CA3 = cornu ammonis 3 and CA1 = cornu ammonis 1. Scale bars equal 200 μm.
Statistical power analysis and optimal sample size
Our results from parallel cultures indicate that with randomly chosen slices (one per well) and a 5% test significance level it is possible to obtain over 80% statistical power with a sample size of six, when control and LPS groups are compared (Table 1). Likewise, the analysis yielded a biologically satisfactory 77–80% power with the same sample size for the TSA groups. The effect of helenalin without TSA (53% power for IL-6 and 57% for nitric oxide) was not statistically prominent and TSA clearly had no effect on nitric oxide secretion when combined with LPS (4% power).
Table 1 Statistical power analysis showing the effect of sample size (n = 6) and test significance level (5%) on statistical power. The effect size was calculated as the difference between two population means (e.g. control and treatment groups) divided by the standard deviation of either population. The analysis was done with nQuery Advisor®, version 5.1, computer program for Wilcoxon (Mann-Whitney) rank-sum test.
Test significance level, α Power (%) n per group Effect size, δ = μ1 - μ2/σ
IL-6 NO IL-6 NO IL-6 NO IL-6 NO
C vs. LPS 0.05 0.05 85 85 6 6 -11.2 -9.2
LPS vs. LPS+TSA 0.05 0.05 78 4 6 6 -2.4 -0.2
LPS vs. LPS+Hele 0.05 0.05 53 57 6 6 1.4 1.5
LPS+TSA vs. LPS+TSA+Hele 0.05 0.05 77 80 6 6 2.3 2.7
Discussion
Cell culture media have been routinely supplemented with animal serum as a source of nutrients. Over the past years the trend towards serum-free culture conditions has received much attention since there are both ethical and technical concerns related to the use of serum [29,30]. Obtaining the serum from fetuses removed from pregnant cows at slaughter has raised questions about animal suffering. Moreover, the possibility of batch-to-batch variability in the undefined serum composition may add interference to the reproducibility of results between studies. At the same time, there is an evolving concern and commitment among the regulatory authorities, in the European Union and the USA, to find ways to reduce the number of animals used in laboratories for experimental and other scientific purposes to the minimum level [31].
We have previously observed that, in the presence of serum, the LPS response is higher than without serum (Fig. 13). Also, our earlier studies have shown that the efficient concentrations of LPS are dependent on the type of LPS product. In this study, especially to ensure the penetration of LPS throughout the whole slice, we used a saturation level of Sigma L6529 product since we have tested the concentration dependence of that LPS product in our different models (see e.g. [32]). As our results demonstrate, the concentration used did not show any toxic response (Fig. 3A) and did not prevent the pro- and anti-inflammatory responses (Fig. 6A &6B). The vehicle effect (i.e. water) was also carefully excluded.
Figure 13 Effect of serum-free (Neurobasal + B27) vs. serum supplemented (Neurobasal + B27 + 10% FBSi) culture media on LPS-induced IL-6 response. LPS (5 μg/ml) was added to the culture media at 4 DIV for 24 h. LPS response was enhanced with serum supplementation. Values are means ± SD (n = 6 in each group).
Our present findings demonstrate that slices cultured in serum-free medium still respond significantly and in a similar manner to an inflammatory stimulus between the culture period from 4 DIV to 21 DIV. We demonstrate that in our model the phenotype of the microglial-cell population remains heterogenous throughout the culture time after a recovery period of approximately 4–5 DIV. Despite the nonuniform microglial appearance, the non-treated control slices seem to retain stability as what comes to lysis of cells (LDH leakage) and to their ability to release cytokine IL-6 and NO.
In addition, the slice cultures seem to be able to switch from the immunologically resting state to activated modes and vice versa when exposed to pro- and anti-inflammatory stimuli at different time points. As microglia possess the Toll-like receptor 4, and thus recognized as the major LPS-responsive cell type [32], we conclude that these responses are mediated mainly by microglia. A number of studies support the view that indeed the microglia and their cell surface receptors, the TLRs, form a defence system that encounters the microbial attack and make up the immune system of the CNS [27,28,33,34].
Moreover, based on our finding that TSA together with LPS was able to enhance the inflammatory response, we hypothesize that the microglia population, even though subjected to severe stress by 5 μg/ml of LPS alone, still possesses some reserve to counter an even more profound stimulus. As our DCX immunostaining revealed, the TSA/LPS induction was not toxic to neurons, thus suggesting that the slices, at least to some degree, have the capacity to cope with the severe inflammation.
As new methods appear, we have the possibility to refine the design of animal experiments. With the aid of sophisticated computer programs it is nowadays relatively easy to explore how manipulation of different parameters can affect the sample size or statistical power [35]. Keeping this and the inevitable inherent biological variability in mind, we estimated the power of the statistical tests related to defined sample sizes in order to provide biologically meaningful results (see Table 1.) If one wishes to minimize the number of animals to be sacrificed and, yet, to get enough statistical power for analysis, then we propose a "one slice per well/six slices per group" model. Hence the adequate sample size can be reached with an ethically and statistically justified number of animals. Our observations indicate that the distribution of data is often more skewed in the "treatment groups", this being probably due to the biological variation within these groups. Therefore we suggest that before applying an appropriate test (in our case the nonparametric Mann-Whitney U-test) the data should be carefully analysed.
We also argue that by applying this model it is possible to avoid the potential bias emerging from interactions between slices when a number of slices are placed in proximity on the same culture membranes. The possibility of error related to the use of littermates is a matter to be considered when deciding how many study replicates are required. Our replicated studies using different litters for parallel cultures indicate that the variation between different litters is minimal and has no relevance when isogenic rat pups are used.
We suggest that this model is well suited for revealing different biochemical immune responses and these studies can be carried out as soon as after 4–5 days in culture since the slices seem to have largely recovered from the trauma caused by dissection procedure and they seem to respond to anti- and proinflammatory stimuli in a similar manner as they do later on. Also, based on the data from the temporal profile of LPS response, we conclude that NF-κB-mediated inflammatory markers can be measured any time between 6 to 48 hour exposure time. On the other hand, since the slices remain stable for at least 4 weeks, also long term, more chronic exposures to stimuli can be studied with this model.
Conclusion
In this report, we used inflammation as a model to illustrate the efficacy of in vitro hippocampal slice culture as an intermediate model between single cell lines and in vivo models. Furthermore, we wanted to examine the possibility of refining the model by taking into the account the experimental design involving the culture conditions and appropriate sample size. Our study highlights the potential of the currently widely used organotypic hippocampal slice culture (OHSC) as a reliable, defined model to be used in preclinical drug studies involving immune reactions and inflammation.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
JH conceived the design of the study, carried out all experiments, performed data analysis and drafted the manuscript. TS aided in slice culture experiments, participated in study design and gave critical analysis of the manuscript. RM aided in manuscript preparation, experimental design and supervised the histological studies. TvG helped especially in histology and provided the facilities, participated in study design and preparation of manuscript. AS aided in study design, especially with regards to inflammatory experiments, supervised all experiments and helped to draft the manuscript. All authors read and approved the final manuscript.
Acknowledgements
We thank Dr Ewen MacDonald for checking the language of the manuscript and Mr Pasi Miettinen and Mrs Airi Boman for technical assistance. This study was financially supported by the Academy of Finland and University of Kuopio.
==== Refs
Medzhitov R Janeway C Jr The Toll receptor family and microbial recognition Trends Microbiol 2000 8 452 456 11044679 10.1016/S0966-842X(00)01845-X
Janeway CA JrMedzhitov R Innate immune recognition Annu Rev Immunol 2002 20 197 216 11861602 10.1146/annurev.immunol.20.083001.084359
Sanz O Acarin L Gonzalez B Castellano B NF-kappaB and IkappaBalpha expression following traumatic brain injury to the immature rat brain J Neurosci Res 2002 67 772 780 11891791 10.1002/jnr.10140
Hawiger J Innate immunity and inflammation: a transcriptional paradigm Immunol Res 2001 23 99 109 11444396 10.1385/IR:23:2-3:099
Rivest S Molecular insights on the cerebral innate immune system Brain Behav Immun 2003 17 13 19 12615045 10.1016/S0889-1591(02)00055-7
Hanisch UK Microglia as a source and target of cytokines Glia 2002 40 140 155 12379902 10.1002/glia.10161
Streit WJ Microglia as neuroprotective, immunocompetent cells of the CNS Glia 2002 40 133 139 12379901 10.1002/glia.10154
Streit WJ Mrak RE Griffin WS Microglia and neuroinflammation: a pathological perspective J Neuroinflammation 2004 1 14 15285801 10.1186/1742-2094-1-14
Vilhardt F Microglia: phagocyte and glia cell Int J Biochem Cell Biol 2005 37 17 21 15381143 10.1016/j.biocel.2004.06.010
Graeber MB Scheithauer BW Kreutzberg GW Microglia in brain tumors Glia 2002 40 252 259 12379912 10.1002/glia.10147
McGeer EG McGeer PL Inflammatory processes in Alzheimer's disease Prog Neuropsychopharmacol Biol Psychiatry 2003 27 741 749 12921904 10.1016/S0278-5846(03)00124-6
Tanaka J Toku K Matsuda S Sudo S Fujita H Sakanaka M Maeda N Induction of resting microglia in culture medium devoid of glycine and serine Glia 1998 24 198 215 9728766 10.1002/(SICI)1098-1136(199810)24:2<198::AID-GLIA5>3.0.CO;2-0
Nakamura Y Si QS Kataoka K Lipopolysaccharide-induced microglial activation in culture: temporal profiles of morphological change and release of cytokines and nitric oxide Neurosci Res 1999 35 95 100 10616913 10.1016/S0168-0102(99)00071-1
Lombardi VR Etcheverria I Fernandez-Novoa L Cacabelos R In vitro regulation of rat derived microglia Neurotox Res 2003 5 201 212 12835124
Cooper CL Jeohn GH Tobias P Hong JS Serum-dependence of LPS-induced neurotoxicity in rat cortical neurons Ann N Y Acad Sci 2002 962 306 317 12076983
Si Q Nakamura Y Kataoka K A serum factor enhances production of nitric oxide and tumor necrosis factor-alpha from cultured microglia Exp Neurol 2000 162 89 97 10716891 10.1006/exnr.2000.7334
Czapiga M Colton CA Function of microglia in organotypic slice cultures J Neurosci Res 1999 56 644 651 10374819 10.1002/(SICI)1097-4547(19990615)56:6<644::AID-JNR10>3.0.CO;2-9
Coltman BW Ide CF Temporal characterization of microglia, IL- 1 beta-like immunoreactivity and astrocytes in the dentate gyrus of hippocampal organotypic slice cultures Int J Dev Neurosci 1996 14 707 719 8960978 10.1016/S0736-5748(96)00071-8
Nimmerjahn A Kirchhoff F Helmchen F Resting Microglial Cells Are Highly Dynamic Surveillants of Brain Parenchyma in Vivo Science 2005 308 1314 1318 15831717 10.1126/science.1110647
Grossmann R Stence N Carr J Fuller L Waite M Dailey ME Juxtavascular microglia migrate along brain microvessels following activation during early postnatal development Glia 2002 37 229 240 11857681 10.1002/glia.10031
Petersen MA Dailey ME Diverse microglial motility behaviors during clearance of dead cells in hippocampal slices Glia 2004 46 195 206 15042586 10.1002/glia.10362
Skibo GG Nikonenko IR Savchenko VL McKanna JA Microglia in organotypic hippocampal slice culture and effects of hypoxia: ultrastructure and lipocortin-1 immunoreactivity Neuroscience 2000 96 427 438 10683583 10.1016/S0306-4522(99)00562-X
Gahwiler BH Capogna M Debanne D McKinney RA Thompson SM Organotypic slice cultures: a technique has come of age Trends Neurosci 1997 20 471 477 9347615 10.1016/S0166-2236(97)01122-3
Stoppini L Buchs PA Muller D A simple method for organotypic cultures of nervous tissue J Neurosci Methods 1991 37 173 182 1715499 10.1016/0165-0270(91)90128-M
Suuronen T Huuskonen J Pihlaja R Kyrylenko S Salminen A Regulation of microglial inflammatory response by histone deacetylase inhibitors J Neurochem 2003 87 407 416 14511118 10.1046/j.1471-4159.2003.02004.x
Lyss G Knorre A Schmidt TJ Pahl HL Merfort I The anti-inflammatory sesquiterpene lactone helenalin inhibits the transcription factor NF-kappaB by directly targeting p65 J Biol Chem 1998 273 33508 33516 9837931 10.1074/jbc.273.50.33508
Lehnardt S Massillon L Follett P Jensen FE Ratan R Rosenberg PA Volpe JJ Vartanian T Activation of innate immunity in the CNS triggers neurodegeneration through a Toll-like receptor 4-dependent pathway Proc Natl Acad Sci USA 2003 100 8514 8519 12824464 10.1073/pnas.1432609100
Olson JK Miller SD Microglia initiate central nervous system innate and adaptive immune responses through multiple TLRs J Immunol 2004 173 3916 3924 15356140
van der Valk J Mellor D Brands R Fischer R Gruber F Gstraunthaler G Hellebrekers L Hyllner J Jonker FH Prieto P Thalen M Baumans V The humane collection of fetal bovine serum and possibilities for serum-free cell and tissue culture Toxicol in Vitro 2004 18 1 12 14630056 10.1016/j.tiv.2003.08.009
Sandra C Balls M Bowe G Davis J Gstraunthaler G Hartung T Hay R Merten OW Price A Schechtman L Stacey G Stokes W Patlewicz G Second ECVAM Task Force on Good Cell Culture Practice: Guidance on good cell culture practice. a report of the second ECVAM task force on good cell culture practice Altern Lab Anim 2005 33 261 287 16180980
Schechtman LM Stokes WS ECVAM-ICCVAM: prospects for future collaboration Altern Lab Anim 2002 30 227 236 12513682
Huuskonen J Suuronen T Nuutinen T Kyrylenko S Salminen A Regulation of microglial inflammatory response by sodium butyrate and short chain fatty acids Br J Pharmacol 2004 141 874 880 14744800 10.1038/sj.bjp.0705682
Lehnardt S Lachance C Patrizi S Lefebvre S Follett PL Jensen FE Rosenberg PA Volpe JJ Vartanian T The toll-like receptor TLR4 is necessary for lipopolysaccharide-induced oligodendrocyte injury in the CNS J Neurosci 2002 22 2478 2486 11923412
Lee SJ Lee S Toll-like receptors and inflammation in the CNS Curr Drug Targets Inflamm Allergy 2002 1 181 191 14561199 10.2174/1568010023344698
van Wilgenburg H van Schaick Zillesen PG Krulichova I Sample Power and ExpDesign: tools for improving design of animal experiments Lab Anim (NY) 2003 32 39 43 12601388
|
16285888
|
PMC1298326
|
CC BY
|
2021-01-04 16:38:21
|
no
|
J Neuroinflammation. 2005 Nov 15; 2:25
|
utf-8
|
J Neuroinflammation
| 2,005 |
10.1186/1742-2094-2-25
|
oa_comm
|
==== Front
Mol CancerMolecular Cancer1476-4598BioMed Central London 1476-4598-4-401627766710.1186/1476-4598-4-40ResearchMolecular and cytological features of the mouse B-cell lymphoma line iMycEμ-1 Su Han Seong [email protected] Arthur L [email protected] Liangping [email protected] Seung Tae [email protected] Jae Hwan [email protected] Sungho [email protected] Kim Joong [email protected] Nicole [email protected] Thomas [email protected] Louis M [email protected] Siegfried [email protected] Laboratory of Genetics, Center for Cancer Research (CCR), National Cancer Institute (NCI), NIH, Bethesda, MD, USA2 Metabolism Branch, CCR, NCI, NIH, Bethesda, MD, USA3 Laboratory of Metabolism, CCR, NCI, NIH, Bethesda, MD, USA4 Laboratory of Cellular Carcinogenesis and Tumor Promotion, CCR, NCI, NIH, Bethesda, MD, USA5 Genetics Branch, CCR, NCI, NIH, Bethesda, MD, USA6 Korea Research Institutes of Bioscience and Biotechnology, Daejeon (J. S. K.) and Department of Biological Sciences, Andong National University, Andong, South Korea (J. H. L.)2005 9 11 2005 4 40 40 31 8 2005 9 11 2005 Copyright © 2005 Su Han et al; licensee BioMed Central Ltd.2005Su Han 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
Myc-induced lymphoblastic B-cell lymphoma (LBL) in iMycEμ mice may provide a model system for the study of the mechanism by which human MYC facilitates the initiation and progression of B cell and plasma cell neoplasms in human beings. We have recently shown that gene-targeted iMycEμ mice that carry a His6-tagged mouse Myc cDNA, MycHis, just 5' of the immunoglobulin heavy-chain enhancer, Eμ, are prone to B cell and plasma cell tumors. The predominant tumor (~50%) that arose in the iMycEμ mice on the mixed genetic background of segregating C57BL/6 and 129/SvJ alleles was LBL. The purpose of this study was to establish and characterize a cell line, designated iMycEμ-1, for the in-depth evaluation of LBL in vitro.
Methods
The morphological features and the surface marker expression profile of the iMycEμ-1 cells were evaluated using cytological methods and FACS, respectively. The cytogenetic make-up of the iMycEμ-1 cells was assessed by spectral karyotyping (SKY). The expression of the inserted MycHis gene was determined using RT-PCR and qPCR. Clonotypic immunoglobulin gene arrangements were detected by Southern blotting. The global gene expression program of the iMycEμ-1 cells and the expression of 768 "pathway" genes were determined with the help of the Mouse Lymphochip© and Superarray© cDNA micro- and macroarrays, respectively. Array results were verified, in part, by RT-PCR and qPCR.
Results
Consistent with their derivation from LBL, the iMycEμ-1 cells were found to be neoplastic IgMhighIgDlow lymphoblasts that expressed typical B-cell surface markers including CD40, CD54 (ICAM-1), CD80 (B7-1) and CD86 (B7-2). The iMycEμ-1 cells harbored a reciprocal T(9;11) and three non-reciprocal chromosomal translocations, over-expressed MycHis at the expense of normal Myc, and exhibited gene expression changes on Mouse Lymphochip© microarrays that were consistent with MycHis-driven B-cell neoplasia. Upon comparison to normal B cells using eight different Superarray© cDNA macroarrays, the iMycEμ-1 cells showed the highest number of changes on the NFκB array.
Conclusion
The iMycEμ-1 cells may provide a uniquely useful model system to study the growth and survival requirements of Myc-driven mouse LBL in vitro.
==== Body
Background
Gene-targeted iMycEμ mice contain a single-copy mouse MycHis (c-myc) cDNA that has been inserted in opposite transcriptional orientation in the mouse immunoglobulin heavy-chain gene cluster, Igh. The specific insertion site of the MycHis transgene is in the intervening region of the Igh joining gene locus, JH, and the intronic heavy-chain enhancer, Eμ. The inserted transgene encodes a C-terminal His6 tag that is useful to distinguish message and protein encoded by MycHis and normal Myc [1]. The iMycEμ mice provide a model system for the study of the molecular and oncogenic consequences of the human MYC- and mouse Myc-deregulating chromosomal t(8;14)(q24;q32) and T(12;15) translocations that are widely accepted as the crucial initiating oncogenic events in the great majority of human Burkitt lymphomas (BL) and mouse plasmacytomas, respectively [2]. Specifically, the iMycEμ mice mimic the type of t(8;14)(q24;q32) and T(12;15) translocation that is found in the endemic form of BL [3] and a subset (~20%) of IL-6 transgenic mouse plasmacytomas [4], respectively. We have recently shown that heterozygous transgenic iMycEμ mice on the mixed genetic background of segregating C57BL/6 and 129/SvJ alleles are genetically prone to mature B cell and plasma cell neoplasms, ~50% of which are IgM+ lymphoblastic B-cell lymphomas (LBL) [1]. We now report on a newly established LBL-derived cell line, iMycEμ-1, which was developed to study the growth and survival requirements of LBL in vitro.
Results and Discussion
Features of iMycEμ-1 cells
The iMycEμ-1 cell line, which demonstrated the typical cytological features of mouse LBL (Fig. 1A top), was derived from a primary IgM+LBL (Fig. 1A bottom) that exhibited moderate plasmacytic differentiation potential in situ (not shown). FACS analysis using a panel of antibodies to B cell surface markers (Fig. 1B) showed that iMycEμ-1 cells were positive for CD40, CD48, CD54, CD80/86 (B7-1/2) and CD138 (syndecan 1). Expression of class I and II MHC antigens and CD45 (B220) was detectable at low or very low levels, respectively, but CD95 (Fas) was absent. Treatment of iMycEμ-1 cells with antibody to CD40 led to the induction of Fas and upregulation of CD45 and CD54, activation markers CD80/86, and CD138, indicating that CD40 signaling was functional (Additional File 1). Southern blotting of genomic DNA from the LBL from which the cell line was derived demonstrated V(D)J rearrangement at the Ig heavy-chain and κ light chain loci (Fig. 1C top). The expression of surface IgMhighIgDlow by the derivative cell line was consistent with this and indicated that the rearrangement was productive (Fig. 1C bottom). SKY analysis of metaphase chromosomes from iMycEμ-1 cells (Fig. 2) uncovered four chromosomal translocations that took the form of a reciprocal T(9;11) exchange and three non-reciprocal exchanges: T(13;16), T(14;13) and T(17;6). Although it is unclear whether these translocations occurred during tumor development or establishment of the cell line [5], repeat karyotyping showed that the present iMycEμ-1 line is cytogenetically stable.
Figure 1 Features of iMycEμ-1 cells. A, cytofuge specimen of cultured cells stained according to May-Grünwald-Giemsa (top). Tissue section of the LBL from which the cell line was derived after immunostaining for μ H-chain (bottom). B, B-cell surface marker expression evaluated by FACS in cells treated with specific antibodies (purple histograms) or isotype controls (green lines). C, H/L rearrangements and surface Ig expression. Southern blots of Igh (top left) and Igk (top right) rearrangements of the LBL from which the cell line was derived. Included as control is liver DNA from homozygous (Tg/Tg) or heterozygous (Tg/+) transgenic iMycEμ mice (left panel) or inbred C57BL/6 and 129SvJ mice (right panel). Recombination at the Igh locus was detected by the reduction of the normal, H chain-encoding upper fragment (upper arrowhead) in the face of comparable amounts of the mutated, MycHis-harboring lower fragment (lower arrowhead). Thus, the 6.2 kb long upper fragment was diminished in the LBL compared to the Tg/+ sample (and absent, as expected, in the Tg/Tg sample), whereas the MycHis-harboring lower fragment was comparable. The MycHis-bearing Igh locus cannot encode H chain because of the gene insertion. Recombination at the Igκ locus resulted in an enlarged fragment (~7.8 kb) compared to the germ line fragment that is indicated by the arrowhead pointing left. Detection of surface IgMhiIgDlow using FACS analysis (bottom).
Figure 2 Spectral karyotype of iMycEμ-1 cells. A, representative metaphase chromosome spread in SKY display (left) and classification colors (center) and as an inverted DAPI image (right). B, complete, near-diploid tumor karyotype depicting each chromosome in SKY display (left) and classification (right) colors and after staining with DAPI (center): 38–40, XX, Del(4)(C4)[2], Del(6)(D2)[6], Der(9)T(9A4;11E2)[4], Der(11)T(11A4;9E4)[4], Del(12)[3], Der(13)T(13D11;16B5)[5], Der(14)T(14;13)[3], Der(17) T(17D;6D3)[5], +19[6]. The chromosomes are arranged in numerical order from left to right and top to bottom. C, chromosomal translocations that took the form of a reciprocal T(9;11) exchange (left) and three different non-reciprocal exchanges (right).
Gene expression profile of iMycEμ-1 cells on cDNA microarray
The Mouse Lymphochip, a microarray of hematopoietic mouse cDNA clones, provides a powerful tool to evaluate the similarity of primary mouse B cells, B-cell tumors, and tumor-derived cell lines at the level of global gene expression [6]. To compare the gene expression profile of LBL and iMycEμ-1 cells, RNA was obtained from primary B cells, fresh-frozen tumors and iMycEμ-1 cells. The RNA was labeled with Cy5-dUTP, and hybridized to the cDNA microarray spotted on a glass slide. An RNA control pool labeled with Cy3-dUTP was co-hybridized to the same array and used as a common denominator by which all samples were compared to one another. Further information on microarray make-up, analysis and data interpretation is available at: .
Three independent RNA samples of iMycEμ-1 cells and ten primary LBL were analyzed together with a collection of normal, resting mouse B cells and T cells, mouse embryonic fibroblasts (MEF), and peritoneal plasmacytomas that arose in pristane-treated BALB/c mice. A total of 414 well-characterized array elements that clustered across these samples based on gene expression patterns (Fig. 3) demonstrated a clear distinction of lymphoid and non-lymphoid cell types (B and T cells versus MEF), lymphocyte lineages (B versus T cells), and transformation- and development-associated differences within the B-cell lineage (normal B cells versus LBL and PCT). The gene expression profiles of LBL and iMycEμ-1 cells, which clustered in one tight group (Fig. 3 top, blue rectangle), exhibited a remarkable homogeneity. Compared to the plasmacytomas, LBL and iMycEμ-1 cells maintained many hallmark genes in the B cell signature (e.g., those encoding CD19, CD79 and μ heavy-chain) but under-expressed numerous genes in the plasma cell signature (e.g., Sec61, Ssr4 and DNAjc3) and matrix signature (e.g., those encoding vinculin, gelsolin and integrin B1) [7,8]. A more detailed analysis of the plasma cell signature revealed that in contrast to Xbp1 and its target genes, the iMycEμ-1 cells expressed Prdm1 (Blimp1). This suggested that the cells underwent neoplastic transformation at the early stage of plasmacytic differentiation (Additional File 2).
Figure 3 Similar gene expression profile of iMycEμ-1 cells and LBL using comparative cDNA microarray measurements. Relative gene expression levels are depicted according to the color scale shown below the cluster.
Myc expression in LBL and iMycEμ-1 cells
The expression levels of Myc, as measured by the arrays, was clearly elevated in LBL and iMycEμ-1, intermediate in "premalignant" B cells from tumor-free iMycEμ mice, and absent, as expected, in resting lymphocytes from normal mice (Fig. 4A). Because the inserted Myc cDNA in iMycEμ mice also encodes a C-terminal His6 tag, it is possible to distinguish message and protein encoded by MycHis and normal Myc. Allele-specific RT-PCR analysis of MycHis and Myc mRNA demonstrated that, in common with LBL, iMycEμ-1 cells expressed predominantly the transgene (Fig. 4B top, lanes 2–3). This pattern of suppression of the normal Myc gene [9] is also a feature of human B-cell lymphomas containing constitutively deregulated MYC [10]. Western blotting with an anti-Myc antibody detecting both MycHis and normal Myc proteins (Fig. 4B bottom) showed that LBL and iMycEμ-1 cells over-expressed Myc at comparable levels (lanes 2–3) relative to B splenocytes from non-transgenic littermates (lane 1). To compare the levels of Myc in LBL and iMycEμ-1 cells more precisely, we performed qPCR using Aktb mRNA levels as internal standard. The iMycEμ-1 cells expressed nearly twice as much Myc as the LBL (Fig. 4C). The levels of Myc also correlated with the expression of genes from the proliferation cluster when the gene expression from the proliferation signature, as defined in Figure 3, was averaged for each cell type. Proliferation gene expression was low in unstimulated cells (MEF and B/T cell samples), intermediate in pre-malignant B cells from iMycEμ mice, and upregulated in LBL and iMycEμ-1 (Fig. 4D).
Figure 4 Myc expression and Myc-driven gene expression changes in iMycEμ-1 cells. A, expression of Myc in iMycEμ-1 cells, LBL, and normal controls. The average gene expression was calculated and plotted according to the microarray measurements shown in Figure 3. B, RT-PCR of Myc, MycHis and Aktb mRNA levels (top) and Western blotting of Myc protein (bottom) in normal B cells (lane 1), LBL from an iMycEμ mouse (lane 2), and the iMycEμ-1 cell line (lane 3; SM, size marker). C, real-time qPCR analysis of Myc mRNA levels in iMycEμ-1 and LBL cells. Mean values and standard deviations based on three independent iMycEμ-1 and five LBL samples are shown. D, expression of proliferation signature genes in iMycEμ samples and normal controls. The expression of genes that fall in the proliferation signature defined in Figure 3 was averaged for each cell population and plotted. E, differentially expressed Myc targets in iMycEμ-1 cells (left column) and LBL (right column) compared to normal resting B cells. Relative gene expression levels are depicted according to the color scale at the bottom. Gene designations and names are listed to the right.
Myc target genes in LBL and iMycEμ-1 cells
To further examine the contribution of the MycHis transgene to the gene expression profile of LBL and iMycEμ-1, we performed a statistical analysis (Student's T test) of genes differentially expressed between normal B cells versus LBL and iMycEμ-1 cells. A total of 122 array elements from Figure 3 were significantly differential in their expression when B cells were compared to iMycEμ-1 cells and LBL (p < 0.015, 1.5-fold minimal difference in average expression). The vast majority (97%) of these elements, many of them previously identified as proliferation-associated Myc targets , behaved similarly in both the cell line and LBL (Fig. 4E). Twenty-four known Myc targets were up-regulated in the LBL and iMycEμ-1 cells (Ahcy, Apex1, Cbfb, Cdk4, Ctps, Eef2, Gapd, Hdgf, Hnrpa1, Hnrpd, Idh1, Myc, Ncl, Nme1, Npm1, Pa2g4, Pcna, Pim1, Pkm2, Ppia, Sfrs2Slc7a5, Tfdp1), two were down-regulated (Btg1, Igk), and five were undetermined as to the effect of Myc on their expression (Aldh2, Hint1, Mcl1, Rheb, Slc1a4). These findings were in accordance with the nature of LBL and iMycEμ-1 as Myc-driven B-cell tumors and firmly established the similarity of LBL and iMycEμ-1at the level of a single gene (Fig. 4A), a gene expression signature (Fig. 4D), and globally (Fig. 3).
Validation of gene expression changes in iMycEμ-1 cells
To further compare the gene expression profiles of LBL and iMycEμ-1, and validate the cDNA microarray results with an independent method, we used cDNA macroarrays on nylon membranes to assess the expression of selected "pathway" genes in LBL, iMycEμ-1 and normal B cells. Included in the analysis were RNA samples of iMycEμ-1 and LBL previously analyzed on the microarray. Freshly prepared RNA from normal, MACS purified, B220+ splenocytes were used as control. RNA samples were labeled with 32P-dUTP and individually hybridized to the macroarrays. Individual expression profiles were determined and compared with each other. Reproducible two-fold or higher changes in hybridization signal intensity were used as threshold for gene expression changes. Eight different macroarrays, each containing 96 genes involved in cell cycle regulation, apoptosis, cancer, signal transduction, stress and toxicity responses, and the NFκB and MAPK pathways, were used. The primary data set is depicted in Additional File 3.
The changes on the macroarrays were remarkably consistent in LBL and iMycEμ-1 cells compared to B cells. This is illustrated in Figure 5A, using the apoptosis array as the example. Among a total of 768 genes present on eight different macroarrays, 121 (16%) genes were concordantly up- or down-regulated in LBL and iMycEμ-1 cells relative to B cells. The NFκB array showed the highest number of changes (n = 22) among the eight different macroarrays, followed by the MAPK (n = 19), cell cycle (n = 18) and apoptosis arrays (n = 17). The stress and toxicity array contained the lowest number of changes (n = 9). Altogether, down-regulated genes (83/121, 69%) outnumbered up-regulated genes (38/121, 31%) by a factor of 2.2. The presence of some genes on two or more arrays afforded an opportunity for additional quality control. Genes of that sort exhibited the same trend on different arrays, either up or down relative to normal B cells, thus adding confidence in the results. This is illustrated in Additional File 4 using one representative array each of iMycEμ-1 and normal B cells.
Figure 5 Concordant gene expression changes in iMycEμ-1 cells and LBL compared to normal B cells. A, gene expression changes were assessed by comparative filter cDNA macroarray measurements using a representative apoptosis array of normal B cells (left), LBL (center) and iMycEμ-1 cells (right) as the example. Compared to B cells, LBL and iMycEμ-1 cells under-expressed Bcl2a1d (array position D2), Birc2 (G3), Cflar (G4), Ripk1 (G8) and Traf5 (E12; indicated by green squares). Additional File 3 shows four additional LBL arrays that exhibit the same changes. B, expression changes of 5 up-regulated and 11 down-regulated genes in iMycEμ-1 cells compared to normal B cells (see Additional File 6 for names, functions and groupings of these genes). Similar changes were seen when LBL and B cells were compared using cDNA macroarrays (not shown) or when iMycEμ-1 cells and/or LBL were compared to B cells using cDNA microarrays (Fig. 3; results not shown). C, verification of gene array results using RT-PCR. Shown are ethidium bromide-stained PCR fragments of the differentially regulated genes plotted in panel B except Myc, which was verified in the experiments presented in Figure 4A-C. The iMycEμ-1 and B-cell samples are shown in the right and left lane, respectively. Up and down regulated genes are depicted on the pink and green background, respectively.
Among the differentially regulated genes that exhibited concordant changes in LBL and iMycEμ-1 cells relative to normal B cells on both the gene micro- and macroarrays were 16 genes that were selected for further confirmation using semi-quantitative RT-PCR (Additional File 5). These genes were of particular interest to us because of possible follow-up studies on signaling pathways in iMycEμ-1 cells. Eleven of the 16 genes were down regulated (Bcl2a1, Birc2, Cflar, Cdkn1b, Grb2, Irf1, Jun, Map2k1, Rb1, Ripk1, Traf5) and five genes were up regulated (Ccna2, Ccnb, Myc, Nfkb1, Odc). Figure 5B presents average quantitative changes on the macroarrays when iMycEμ-1 cells were compared with normal B cells. These changes were readily confirmed by RT-PCR in all cases (Fig. 5C) except Myc, which was not included because it was confirmed in previous work (Fig. 4).
Conclusion
This study reports the molecular, cytogenetic and morphological features of a stable cell line, designated iMycEμ-1. The iMycEμ-1 cells are surface IgMhighIgDlow and cytogenetically stable using SKY. The cells express high levels of the inserted MycHis transgene and exhibit a global gene expression profile consistent with that of MycHis-driven B-cell neoplasia. The iMycEμ-1 cells may be useful for in-depth studies on the growth and survival requirements of MycHis-driven mouse B-cell tumors in vitro. Specifically, the cells may facilitate the elucidation of the signal transduction pathways that appear to maintain high Myc protein levels in mouse LBL [1]. These studies may results in new approaches to treat and prevent MYC-induced B cell and plasma cell neoplasms in human beings.
Methods
Mouse lymphomas and derivation of iMycEμ-1 cells
Transgenic iMycEμ mice develop a high incidence of B cell and plasma cell tumors with LBL being the predominant phenotype (8). Tumor samples obtained at autopsy were fixed in formalin for later histopathology or frozen for later preparation of protein, DNA and RNA. Histological criteria used for diagnosing mouse LBL are detailed elsewhere [11]. Highly enriched splenic B cells were prepared from C57BL/6 mice using CD45R (B220) microbeads and MACS separation columns (Miltenyi Biotec, Auburn, CA). All mice were maintained under Animal Study Protocol LG-028. The iMycEμ-1 cell line was derived from a LBL and maintained at 37°C and 5% carbon dioxide in RPMI 1640 medium supplemented with 10% fetal calf serum, 200 mM L-glutamine, 50 μM 2-mercaptoethanol and penicillin/streptomycin (Gibco-BRL, Rockville, MD).
Characterization of iMycEμ-1 cells
For cytological analysis, cytofuge specimens were stained according to May-Grünwald-Giemsa and inspected by microscopy. For detection of chromosomal aberrations, cells were analyzed by spectral karyotyping (SKY) as previously described [12]. For flow cytometry, single-cell suspensions were stained and analyzed on a FACSort® using the CELLQuest™ software (BD Pharmingen, San Diego, CA). Rat anti-mouse CD16/CD32 was used to block FcγII and FcγIII receptors. Antibodies to mouse CD45 (catalog number 553076), CD80 (553766), Fas (CD90, 554255), CD86 (553689), CD40 (553787), I-Ab (MHC class II, 553551), H-2Kb (MHC class I, 553569), CD48 (557483), CD54 (553250), CD138 (553712), IgD (553438) and IgM (53519) were purchased from BD Biosciences. For the evaluation of surface marker changes upon ligation of CD40, cells were incubated with rat anti-mouse CD40 (553787) using 3.5 μg antiboy per 5 × 105 cells. For Southern blot hybridization of clonotypic V(D)J rearrangements, genomic DNA (20 μg) was digested with BamHI and EcoRI, fractionated on a 0.7 % agarose gel, transferred to a nylon membrane, and crosslinked under UV light. Following pre-hybridization (Hybrisol I, Intergen) at 42°C, the membrane was hybridized to a 1.5-kb HindIII/EcoRI fragment of Igh spanning JH2 and Eμ or to a 1.1-kb Cκ probe, which was generated by PCR using a primer pair obtained from Dr. Michael Kuehl (NCI): 5'-GAT GCT GCA CCA ACT GTA TCC A-3' and 5'-GGG GTG ATC AGC TCT CAG CTT-3'. Probes were labeled with [32P]-CTP using a random priming kit.
Allele-specific RT-PCR of Myc and MycHis mRNA
For semi-quantitative determination of Myc and MycHis mRNA, total RNA was isolated using TRIzol (Sigma, St. Louis, MO, USA). Double stranded cDNA was synthesized from 1 μg of total RNA, using the AMV Reverse Transcriptase kit (Roche, Indianapolis, IN). A common 5' primer for both MycHis and Myc (5'-TCT CCA CTC ACC AGC ACA AC-3') was combined with a specific 3' primer for MycHis (5'-CCT CGA GTT AGG TCA GTT TA-3') and Myc (5'-ATG GTG ATG GTG ATG ATG AC-3') to distinguish the two messages. Thermal cycling conditions were as follows: 95°C for 5 min followed by 20 cycles of amplification at 57°C, 72°C and 95°C, each for 1 min. PCR amplification of Aktb cDNA was performed as control using the following primer pair: 5'-GCA TTG TTA CCA ACT GGG AC-3' and 5'-AGG CAG CTC ATA GCT CTT CT-3'. PCR products were analyzed by electrophoresis in 1% agarose gel and visualized by staining with ethidium bromide.
Real-time qPCR of Myc mRNA
For quantitative Taqman RT-PCR of Myc (Myc plus MycHis), total RNA was isolated from cells using TRIzol Reagent (Invitrogen). Serial dilutions of input RNA (100 ng - 1.56 ng) were analyzed in triplicates using the ABI PRISM 7900HT sequence detector system, primers, probes, and the Taqman One-Step RT-PCR Master Mix Reagents kit, all purchased from Applied Biosystems. The reaction mixture was held at 48°C for 30 min for reverse transcription of RNA into cDNA. This was followed by incubation at 95°C for 10 min to activate the Taq polymerase. PCR amplification of cDNA was performed for 40 cycles using the following cycling conditions: denaturing for 15 s at 95°C and annealing and extending for 1 min at 60°C. All samples were tested in triplicates, and average values were used for quantification. Analysis was performed using SDS v2.1 software (Applied Biosystems) according to the manufacturer's instruction. Aktb was used as internal reference gene. The comparative CT method (ΔΔCT) was used for quantification of gene expression.
Gene microarray hybridization and analysis
cDNA made from total RNA (50 μg) from each tumor, primary cell sample, or iMycEμ-1 cells was labeled with cyanine 5-conjugated dUTP (Cy5). cDNA made from pooled mouse cell line RNA (50 μg) was labeled with cyanine 3-conjugated dUTP (Cy3) and used as reference. Microarray hybridizations were performed on Mouse Lymphochip microarrays [7]. After washing, the slides were scanned using an Axon GenePix 4.0 scanner (Axon Instruments Inc., Union City, CA). After normalization, those elements that failed to meet confidence criteria based on signal intensity and spot quality were excluded from analysis. In addition, data were discarded for any gene for which measurements were missing on >30% of the arrays or were not sequence-verified. The Cy5:Cy3 intensity ratios of the remaining spots were log2 transformed. To compare normal samples, hierarchical cluster analysis was performed using the Gene Cluster and Treeview programs [8].
Gene macroarray hybridization and analysis
The relative mRNA expression of genes involved in regulation of apoptosis, cell cycle progression, NFkB signaling, and cellular stress and toxicity responses was analyzed with GEArray (SuperArray Inc., Bethesda, MD) according to the manufacturer's protocol. Cells were treated for 24 hrs with 0.4 mM and 1 mM CDDO-Im, respectively, followed by preparation of total RNA using TriReagent (Sigma). Five μg from each sample were reverse transcribed into 32P-labeled cDNA using MMLV reverse transcriptase (Promega, Madison, WI) and 32P-dCTP (NEN, Boston, MA). The resulting cDNA probes were hybridized to gene-specific cDNA fragments spotted in quadruplicates on the GEArray membranes. After stringent washing of the arrays, the signal of the hybridized spots was measured with a STORM PhosphorImager (Molecular Dynamics, Sunnyvale, CA) and normalized to the signal of the housekeeping gene Gapd. Array results on six CDDO-Im inducible genes were validated using semi-quantitative RT-PCR.
Gene array validation using RT-PCR
For semi-quantitative determination of mRNA levels, total RNA was isolated and double stranded cDNA was synthesized as described above for Myc. Information on PCR primers and thermal cycling conditions is available in Additional File 6. PCR products were analyzed by electrophoresis in 1% agarose gel and visualized by staining with ethidium bromide.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
Seong-Su Han determined gene expression using Superarray© cDNA macroarrays. Arthur L. Shaffer and Louis M. Staudt evaluated global gene expression profiles using Mouse Lymphochip© microarrays. Liangping Peng performed FACS studies. Seung-Tae Chung harvested and transplanted tumors, cultured cells, and prepared histo- and cytological specimens. Sungho Maeng and Jae-Hwan Lim validated gene array results using RT-PCR and qPCR. Joong-Su Kim performed Southern analysis and Nicole McNeil and Thomas Ried performed SKY analysis. Siegfried Janz designed the study and wrote and approved the article.
Supplementary Material
Additional File 1
FACS histograms.
Click here for file
Additional File 2
Heat map (panel A) and bar graph (panel B).
Click here for file
Additional File 3
Images of cDNA gene arrays.
Click here for file
Additional File 4
Images of cDNA gene arrays (panel A) and gene table (panel B).
Click here for file
Additional File 5
Gene list.
Click here for file
Additional File 6
PCR primers and conditions.
Click here for file
Acknowledgements
We thank Wendy duBois, Nicole Wrice and Vaishali Jarral, NCI, for assistance with the in vivo studies; Michael Kuehl, NCI, for the Cκ probe; R. Eric Davis, NCI, for helpful scientific discussions; and Beverly A. Mock, NCI, for support. This research was supported by the Intramural Research Program of the NIH, NCI, CCR.
==== Refs
Park SS Kim JS Tessarollo L Owens JD Peng L Han SS Tae Chung S Torrey TA Cheung WC Polakiewicz RD McNeil N Ried T Mushinski JF Morse HC Janz S Insertion of c-Myc into Igh induces B-cell and plasma-cell neoplasms in mice Cancer Res 2005 65 1306 1315 15735016 10.1158/0008-5472.CAN-04-0268
Boxer LM Dang CV Translocations involving c-myc and c-myc function Oncogene 2001 20 5595 5610 11607812 10.1038/sj.onc.1204595
Joos S Falk MH Lichter P Haluska FG Henglein B Lenoir GM Bornkamm GW Variable breakpoints in Burkitt lymphoma cells with chromosomal t(8;14) translocation separate c-myc and the IgH locus up to several hundred kb HumMolGenet 1992 1 625 632
Kovalchuk AL Kim JS Park SS Coleman AE Ward JM Morse HC Kishimoto T Potter M Janz S IL-6 transgenic mouse model for extraosseous plasmacytoma Proc Natl Acad Sci USA 2002 99 1509 1514 11805288 10.1073/pnas.022643999
Coleman AE Schröck E Weaver Z du Manoir S Yang F Ferguson-Smith MA Ried T Janz S Previously hidden chromosome aberrations in T(12;15)-positive BALB/c plasmacytomas uncovered by multicolor spectral karyotyping Cancer Res 1997 57 4585 4592 9377573
Alizadeh A Eisen M Davis RE Ma C Sabet H Tran T Powell JI Yang L Marti GE Moore DT Hudson JRJ Chan WC Greiner T Weisenburger D Armitage JO Lossos I Levy R Botstein D Brown PO Staudt LM The lymphochip: a specialized cDNA microarray for the genomic-scale analysis of gene expression in normal and malignant lymphocytes Cold Spring Harb Symp Quant Biol 1999 64 71 78 11232339 10.1101/sqb.1999.64.71
Shaffer AL Yu X He Y Boldrick J Chan EP Staudt LM BCL-6 represses genes that function in lymphocyte differentiation, inflammation, and cell cycle control Immunity 2000 13 199 212 10981963 10.1016/S1074-7613(00)00020-0
Shaffer AL Shapiro-Shelef M Iwakoshi NN Lee AH Qian SB Zhao H Yu X Yang L Tan BK Rosenwald A Hurt EM Petroulakis E Sonenberg N Yewdell JW Calame K Glimcher LH Staudt LM XBP1, downstream of Blimp-1, expands the secretory apparatus and other organelles, and increases protein synthesis in plasma cell differentiation Immunity 2004 21 81 93 15345222 10.1016/j.immuni.2004.06.010
Penn LJ Brooks MW Laufer EM Land H Negative autoregulation of c-myc transcription EMBO J 1990 9 1113 1121 2182320
Eick D Polack A Kofler E Lenoir GM Rickinson AB Bornkamm GW Expression of P0- and P3-RNA from the normal and translocated c-myc allele in Burkitt's lymphoma cells Oncogene 1990 5 1397 1402 2216463
Morse HC Anver MR Fredrickson TN Haines DC Harris AW Harris NL Jaffe ES Kogan SC MacLennan IC Pattengale PK Ward JM Bethesda proposals for classification of lymphoid neoplasms in mice Blood 2002 100 246 258 12070034 10.1182/blood.V100.1.246
Liyanage M Coleman A du Manoir S Veldman T McCormack S Dickson RB Barlow C Wynshaw-Boris A Janz S Wienberg J Ferguson-Smith MA Schröck E Ried T Multicolour spectral karyotyping of mouse chromosomes NatGenet 1996 14 312 315
|
16277667
|
PMC1298327
|
CC BY
|
2021-01-04 16:36:35
|
no
|
Mol Cancer. 2005 Nov 9; 4:40
|
utf-8
|
Mol Cancer
| 2,005 |
10.1186/1476-4598-4-40
|
oa_comm
|
==== Front
Nutr JNutrition Journal1475-2891BioMed Central London 1475-2891-4-301625578110.1186/1475-2891-4-30ResearchAssessment of soy phytoestrogens' effects on bone turnover indicators in menopausal women with osteopenia in Iran: a before and after clinical trial Haghighian Roudsari Arezoo [email protected] Farideh [email protected] Arash [email protected] Bahram [email protected] Bagher [email protected] Seyed Masoud [email protected] School of Nutrition and Food Technology, Shaheed Beheshti University of Medical Sciences, Tehran, Iran2 Endocrinology & Metabolism Research center, Tehran university of Medical science, Tehran, Iran3 Department of Nutritional Sciences, Oklahoma State University, Stillwater, OK, USA2005 29 10 2005 4 30 30 13 7 2005 29 10 2005 Copyright © 2005 Roudsari et al; licensee BioMed Central Ltd.2005Roudsari 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
Osteoporosis is the gradual declining in bone mass with age, leading to increased bone fragility and fractures. Fractures in hip and spine are known to be the most important complication of the disease which leads in the annual mortality rate of 20% and serious morbidity rate of 50%. Menopause is one of the most common risk factors of osteoporosis. After menopause, sex hormone deficiency is associated with increased remodeling rate and negative bone balance, leading to accelerated bone loss and micro-architectural defects, resulting into increased bone fragility.
Compounds with estrogen-like biological activity similar to "Isoflavones" present in plants especially soy, may reduce bone loss in postmenopausal women as they are similar in structure to estrogens.
This research, therefore, was carried out to study the effects of Iranian soy protein on biochemical indicators of bone metabolism in osteopenic menopausal women.
Materials and methods
This clinical trial of before-after type was carried out on 15 women 45–64 years of age. Subjects were given 35 g soy protein per day for 12 weeks. Blood and urine sampling, anthropometric measurement and 48-h-dietary recalls were carried out at zero, 6 and 12 weeks. Food consumption data were analyzed using Food Proccessor Software. For the study of bone metabolism indicators and changes in anthropometric data as well as dietary intake, and repeated analyses were employed.
Results
Comparison of weight, BMI, physical activity, energy intake and other intervening nutrients did not reveal any significant changes during different stages of the study. Soy protein consumption resulted in a significant reduction in the urinary deoxypyridinoline and increasing of total alkaline phosphatase (p < 0.05), although the alterations in osteocalcin, c-telopeptide, IGFBP3 and type I collagen telopeptide were not significant.
Conclusion
In view of beneficial effect of soy protein on bone metabolism indicators, inclusion of this relatively inexpensive food in the daily diet of menopausal women, will probably delay bone resorption, thereby preventing osteoporosis.
soy proteinIsoflavonespostmenopausal womenbone metabolism markersosteopenia
==== Body
Backgrounds
Menopause is a period normally occupying one-third of women's life [1]. Reduced bone density is one of the most prominent symptoms during menopause [2].
Osteoporosis is a serious problem for postmenopausal women which increases the risk of bone fracture and worsens with age, increasing from 4% in 50–59 year age bracket to 50% in 80 years old women. Bone fractures are also prevalent in these women [3]. Today estrogen therapy (ERT) and drugs like bisphosphonates, calcitonin and raloxifene is employed to prevent and treat osteoporosis [4]. However, side effects such as breast cancer, endometrial adenocarcinoma [5] have limited the acceptance of these medications among women [6] and only 35% to 40% of women ever start ERT, and many do not continue it [7].
Epidemiologic studies have shown osteoporotic fractures, cardiovascular disease, postmenopausal symptoms and some cancers to be less prevalent in Asians compared to their western counterparts. Hip fracture, for example, is 50–60% less frequent among Asian compared to western women [8]. This advantage is gradually anihilated as Asian adapt western lifestyle [9]. These observations, prompted researchers to scrutinize Asian dietary habits. Soy is a part of Asian traditional diet [10], showing some relationship with the above-mentioned diseases [9].
Estrogen-like compounds such as isoflavones existing in plant foods specially soy [11,12] can curb reduced bone density in menopausal women, due to their structural similarity [13]. Some studies have not, however, supported clearly the role of soy isoflavones in preventing osteoporosis [14].
Isoflavones are phyto-estrogens similar to women's estrogens and are bound to cellular estrogen receptors in various organs, thus phytoestrogens affinity is weak compared to human's estrogens. Recent studies have shown that cells have two types estrogen receptors α and β. Human estrogens have more affinity to α-receptors, whereas, isoflavones have high affinity to β-receptors. β-receptors exist in brain, bone, bladder and vascular epithelium, being important in the function of non-steroid estrogens [15].
As soy cultivated in Iran is of different variety, namely "Gorgan" compared to other studies and because the Iranian food habits and food pattern is different [16] which might affect the metabolism of nutrients and isoflavones, this study was conducted to assess its effects compared to other countries. Furthermore there are still contradiction, in the literature regarding the role of soy isoflavones which justify implementation of this study.
Methods
This clinical trial of before and after type was carried out in 2003. Women referring to the osteoporosis clinic of Endocrinology & Metabolism Research Centre of Tehran University of Medical Sciences for bone density measurement were screened to find osteopenic subjects and 15 postmenopausal 45–64 year old women were selected. Those women between 1 to 10 years postmenopause who were non-smokers and free from diseases entered this study.
Information on weight, height, body mass index, two 24-hr food consumption recall and physical activity were collected at the start, 6 and 12 weeks of the study. Soy protein at 35 g level containing 98.3 mg isoflavones were given to subjects daily. Subjects were provided with a special cup for measuring soy. Cooking instructions were also given to the subjects.
Blood and urine samplings were done in 3 stages, in the beginning and at the end of 6th and 12th week. Blood and urine samples were kept frozen until the end of the twelfth week at -80°C. Serum biochemical indicators were measured on the same day for all samples. Total alkaline phosphatase was assayed calorimetrically with Hitachi 902 autoanalyzer, osteocalcin by IRMA method using Biosource kit and Wizard gamacounter, IGFBP3 and c-telopeptide by ELISA using Biosources and Bioscience diagnostics respectively. Type I collagen telopeptide was determined by RIA method using Orion-Diagnostica on Wizard gama counter and urinary creatinine by colorimetrically method [17]. Food Processor software was used for food consumption survey and SPSS (version 11.5) was employed for statistical analysis of the data. All quantitative variables were then examined by Kolmogrof-Smirnof (KS) to ensure normality of distribution. To analyze any possible changes in food intake, intervening and biochemical variables in 3 stages, repeated measurement analysis was utilized. The purpose of this analysis was to ensure lack of significant changes of the variables. Significance level was set at below 5 percent (P < 0.05).
Results
Subjects' anthropometric data are shown in Table 1. Mean age was 52.9 ± 4.3 years, years post menopause 5.47 ± 3.4 years and mean height 157.4 ± 7.2 centimeters. Mean body mass index and physical activity level remained unchanged. Mean food consumption figures were not different at 6 and 12 weeks compared to the start of the study (Table 2). Mean bone metabolic indicators for the 3 stages are given in Table 3. After 12 weeks of soy consumption total serum alkaline phosphatase (TALP) significantly increased while urinary deoxypyridinoline (DPD) decreased (P < 0.05). Other indicators namely osteocalcin, insulin growth factor binding protein (IGFBP3), c-telopeptide and type-I collagen telopeptides did not change significantly.
Table 1 Subjects anthropometric data in 3 stages of the study (n = 15).
Indicators At the start 6 weeks 12 weeks
Weight (kg) 68 ± 7.5* 68 ± 7.6 68 ± 7.8
BMI (kg/m2) 27.4 ± 3** 27.4 ± 3 27.4 ± 3
* Mean ± SD
** Mean height was 157.2 ± 7.2.
Table 2 Subjects mean food intake in 3 stages of the study (n = 15).
Variables at the start 6 weeks 12 weeks P value**
Energy (kcal) 1933.4 ± 302.5* 2033.2 ± 420.3 1902.8 ± 308.6 NS***
Protein (g) 74.6 ± 12.3 71.9 ± 17.4 74.8 ± 10.2 NS
Calcium (mg) 999.3 ± 460.2 966.8 ± 443.7 1014 ± 436 NS
Phosphore (mg) 873.2 ± 228.9 853.1 ± 273.9 866.4 ± 200.6 NS
* Mean ± SD
** significant level was set at below 5 percent (P < 0.05).
*** To analyze changes in 3 stages, repeated measurement analysis was utilized.
Table 3 Bone metabolism biomarkers in 3 stages of the study (n = 15).
Variables* At the start 6 weeks 12 weeks P value†
TALP (IU/l) 237.5 ± 85.4** 300.4 ± 294.1 281.3 ± 80.5 <0.05
OC (ng/ml) 11.4 ± 4.8 12.7 ± 5.7 11.5 ± 4.8 NS††
IGFBP3 (ng/ml) 3304.6 ± 728.6 3221.6 ± 534.4 3103.9 ± 639.8 NS
DPD (nmol/mmol) 7 ± 1.2 5.9 ± 1.1 5.1 ± 2.1 <0.05
C-TX (ng/ml) 0.79 ± 0.49 0.8 ± 0.4 0.8 ± 0.3 NS
ITCP (μg/l) 4.6 ± 0.9 4.3 ± 0.9 5.9 ± 1.1 NS
* TALP: Total alkaline phosphatase (IU/l); OC: Osteocalcin (ng/ml); IGFBP3: Insulin like growth factor binding protein 3 (ng/ml); DPD: Deoxypyridinoline (nmol/mmol); C-TX: C-Telopeptides (ng/ml); ITCP: Carboxy terminal telopeptides of type I collagen (μg/l)
** Mean ± SD
† To analyze changes in 3 stages, repeated measurement analysis was utilized.
†† significant level was set at below 5 percent (P < 0.05).
Discussion
The results demonstrated soy protein consumption to have caused increase in TALP and reduction in DPD in menopausal women with osteopenia, while other parameters were not significantly difference. Comprehensive human studies on the effect of soy on TALP have not been carried out. In Arjmandi et al studies on rats a slight but insignificant increase in TALP was seen [18-20]. The crucial role of gut microflora in the metabolism of isoflavones in human beings has also been previously explored [21]. In vivo studies proved that bacteria in the gastrointestinal tract play an important role in determining the magnitude and pattern of isoflavone bioavailability [22]. However, only 30 to 40% of the population can produce equol from daidzein and interindividual differences in the bacteria responsible for equol production [23,24]. Register et al observed a significant fall in TALP in monkeys after 12 weeks [25]. Animal models such as monkey may convert daidzein into equol more readily than 30 to 50% of humans. It has been shown that equol possesses more estrogen-like properties than daidzein. This is why isoflavone efficacy has been less pronounced in monkeys and the results have been reported as reduced TALP levels [26,27].
TALP is an insensitive marker for bone formation compared to bone-specific alkaline phosphatase (BAP) or osteocalcin (OC). In this study, serum osteocalcin as a sensitive marker for bone formation has not changed during interventional period, indicating that soy protein may not enhance bone formation.
With regard to bone resorption, our results showed reduced urinary DPD levels following soy consumption which agrees with the finding of other investigators [28-30]. The effect of isoflavones on this indicator is so strong that Uesagi et al [31] observed consuming 61.8 mg of isoflavone for 4 weeks results in a significant reduction in urinary DPD. It can be said that DPD acts as a bridge between collagen fibrils which enter urine with collagen breakdown. As this is a very specific marker for bone resorption, its significant reduction in our study suggests soy consumption may prevent degradation of collagen the major protein in bone matrix [17].
Other serum indicators of bone metabolism were not affected in our study. In most studies on the effects of isoflavones in rats these phytochemicals, have been reported to have caused rises [32], no change [19,20] and even reduction [14] in bone formation as well as reduction or no change [33,34] on bone resorption. The changes observed in this study, therefore, are not contradictory to other studies and slight differences observed may be attributed to sample size, isoflavone dosing, period of intervention and dissimilarity of the studied groups.
Conclusion
In conclusion it seems soy protein can be effective in protecting bone mass through curbing bone resorption specially in high risk groups as was demonstrated in our osteopenic subjects but not enhance bone formation. Different studies have reported intake of 70–90 milligrams of isoflavones per day to be effective. Soy protein in our study provided 98 mg of isoflavones which is in accordance with other studies [35,36]. Some have reported lesser amounts can be effective in longer periods of time [37]. Soy protein consumption, thus, is a valuable plant estrogen which can be recommended for osteoporosis prevention.
List of abbreviations
BAP: Bone-specific Alkaline Phosphatase
BMI: Body Mass Index
CTX: Collagen type I cross-Linked C-telopeptide
Dpd: Deoxypyridinoline
ELISA: Enzyme Linked Immunosorbent assay
IU/l: International Unit/liter
IGFBP3: Insulin Like Growth Factor-Binding Protein 3
IRMA: Immuno Radiometric Assay
ITCP: Serum Carboxyterminal telopeptide of type I collagen
Kcal: Kilocalorie
Kg/m2: kilogram/meter2
Ks: Kolmogrof-smirnof
mg: miligram
n mol/m mol: nano mol/mili mol
ng/ml: nanogram/mililiter
Oc: Osteocalcin
RIA: Radioimmunoassay
TALP: Total Alkaline Phosphatase
μg/L: microgram/Liter
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
This research is the title of AHR's thesis. She has designed the frame of research and performed the subjects' selection, intervention, follow up of subjects, data gathering and data analysis and finally writing and editing of this manuscript was carried out by her.
FT: Advisor of thesis.
AHN assisted with subjects collection and data analysis. He also contributed in performing the laboratory tests.
BA Measured the soy phytoestrogens.
BL and MK: conceived of the study and participated in its design.
Acknowledgements
Grant of this project was supplied by Endocrinology & Metabolism Research Center of Tehran Medical University.
==== Refs
Barrett CE Epidemiology and the menopause:A global overview Int J Fertil 1993 38 6 14
Dempster DW Lindsay R Pathogenesis of osteoporosis Lancet 1996 341 797 801 8096008 10.1016/0140-6736(93)90570-7
Position statement of the north american menopause society Management of postmenopausal osteoporosis Menopause 2002 9 84 101 11875327
Scharbo DM Hormone replacement therapy Nurse Pract 1996 21 1 13 9238356
Groeneveld FP Bareman FP Barensen r Determinants of first prescription of hormone replacement therapy. A follow up study among 1689 women aged 45–60 years Maturitas 1994 20 81 89 7715478 10.1016/0378-5122(94)90003-5
Lignieres DB Hormone replacement therapy: clinical benefits and side-effects Maturitas 1996 31 36 10.1016/0378-5122(96)01009-2
Glazier MG Bowman MA A review of the evidence for the use of phytoestrogens as a replacement for traditional estrogen replacement therapy Arch Intern Med 2001 161 1161 1172 11343439 10.1001/archinte.161.9.1161
Ross PD Norimatsu H Davis JW Yano K Wasnich RD A comparison of hip frcture incidence amonge native Japanese, Japanese Americans, and American Caucasians Am J Epidemiol 1991 133 801 809 2021147
Adlercreutz H Mazur W Phytoestrogens and Western diseases Ann Med 1997 29 95 120 9187225
Simoon F Food in china: a cultural and historical inquiry 1991 Boca Raton, FL: CRC press Inc
Kurzer MS Xu X Dietary phytoestrogens Ann Rev Nutr 1997 17 353 381 9240932 10.1146/annurev.nutr.17.1.353
Coward L Barnes NC Setchell EDR Barnes S Genistein, daidzein and their β-glucoside conjugates: antitumor isoflavones in soybean foods from American diets and Asian diets J Agri Food Chem 1993 41 1961 1967 10.1021/jf00035a027
Messina MJ Legumes and soybeans: Overview of their nutritional profiles and health effects Am J Clin Nutr 1999 439 450 10500011
Wangen KE Duncan AM Merz-Demlow BE Effects of soy isoflavones on markers of bone turnover in premenopausal and postmenopausal women J Clin Endocrinol Metab 2000 85 3043 3048 10999783 10.1210/jc.85.9.3043
Kuiper GG Carisson B Grandien K Comparison of the ligand binding specifity and transcript tissue distribution of estrogen receptors alpha and beta Endocrinology 1997 138 863 870 9048584 10.1210/en.138.3.863
National Nutrition Institute National food consumption survey report. Tehran, Iran 1995
Watts NB Clinical utility of biochemical markers of bone remodeling Clin Chem 1999 45 1359 1368 10430819
Arjmandi BH Alekel L Hollis BW Amin D Dietary soybean protein prevents bone loss in an ovariectomized rat model of osteoporosis J Nut 1996 126 161 167
Arjmandi BH Getlinger MJ Goyal NV Alekel l Hasler CM Role of soy protein with normal or reduced isoflaonecontent in reversing bone loss induced by ovarian hormone deficiency in rats Am J Clin Nutr 1998 1358 1363
Arjmandi BH Birnbaum R Goyal NV Getlinger MJ Bone sparing effect of soy protein in ovarian hormone-deficient rats is related to its isoflavone content Am J Clin Nutr 1998 1364 1368
Chang YC Nair MG Metabolism of daidzein and genistein by intestinal bacteria J Nutr Proc 1995 58 1892 1896
Xu X Harris KS Wang HJ Murphy PA Hendrich S Bioavailability of soybean isoflavones depends upon gut microflora in women J Nutr 1995 125 2307 2315 7666247
Sathyamoorthy N Wang TTY Differential effects of dietary phyto-oestrogens daidzein and equol on human breast cancer MCF-7 cells Eur J Cancer 1997 33 2384 2389 9616286 10.1016/S0959-8049(97)00303-1
Setchell KDR Adlercreutz H Rowland IR Mammalian lignans and phytoestrogens: recent studies on their formation, metabolism, and biological role in health and disease Role of the gut flora in toxicity and cancer 1988 Academic Press. London, United Kingdom 315 345
Register TC Jayo MJ Anthony MS Soy phytoestrogens do not prevent bone loss in postmenopausal monkeys J Clin Endocrinol Metab 2003 88 4362 4370 12970311 10.1210/jc.2003-030493
Clarkson TB Anthony MS Morgan TM Inhibition of postmenopausal atherosclerosis progression: a comparison of the effects of conjugated equine estrogens and soy phytoestrogens J Clin Endocrinol Metab 2001 86 41 47 11231976 10.1210/jc.86.1.41
Setchell KD Brown NM Lydeking-Olsen E The clinical importance of the metabolite equol-a clue to the effectiveness of soy and its isoflavones J Nutr 2002 132 3577 3584 12468591
Arjmandi Bh Khalil DA Smith BJ soy protein has a greater effect on bone in postmenopausalwomen not on hormon replacement therapy as evidenced by reducing bone resorption and urinary calcium excretion J Clin Endocrinol Metab 2003 88 1048 1054 12629084 10.1210/jc.2002-020849
Yamori Y Moriguchi EH Teramoto T Miura A Soybean isoflavones reduce postmenopausal bone resorption in female Japanese immigrants in Brazil:A ten week study J Am Coll Nutr 2002 21 560 563 12480802
Picherit C Pelissero CB Chanteranne B Lebecque P Soybean isofavones dose-dependently reduce bone turnover but do not reverse established osteopenia in adult ovariectomized rats J Nutr 2001 131 723 728 11238750
Uesugi T Fukui Y Yamori Y Beneficial effects of soybean isoflavone supplementation on bone metabolism and serum lipids in postmenopausal Japanese women: a four week study J Am Coll Nutr 2002 21 97 102 11999549
Fanti P Monier-Faugere MC Geng Z The phytoestrogen genistein reduces bone loss in short term ovariectomized rats Osteoporosis Int 1998 8 274 281 10.1007/s001980050065
Agnusdei D Crepaldi G Isaia G A double blind, placibo controlled trial of ipriflavone for prevention of postmenopausal spinal bone loss Calcif Tissue Int 1997 61 142 147 9236262 10.1007/s002239900312
Gennari C Agnusdei D Crepaldi G Effect of ipriflavone-a synthetic derivative of natural isoflavones-on bonemass loss in the early years after menopause Menopause 1998 5 9 15 9689189
Potter SM Baum JA Teng H Stillman RJ Soy protein and isoflavone: their effects on blood lipids and bone density in postmenopausal women Am J Clin Nutr 1998 68 1375 1379
Alekel DL Germain AS Peterson CT Hanson KB Isoflavone-rich soy protein isolate attenuates bone loss in the lumbar spine of perimenopusal women Am J Clin Nutr 2000 72 844 852 10966908
Barnes S Phytoestrogens and osteoporosis – What is a safe dose? Br J Nutr 2003 39 101 108
|
16255781
|
PMC1298328
|
CC BY
|
2021-01-04 16:39:30
|
no
|
Nutr J. 2005 Oct 29; 4:30
|
utf-8
|
Nutr J
| 2,005 |
10.1186/1475-2891-4-30
|
oa_comm
|
==== Front
Nutr JNutrition Journal1475-2891BioMed Central London 1475-2891-4-311625578510.1186/1475-2891-4-31ResearchA comparative study of food habits and body shape perception of university students in Japan and Korea Sakamaki Ruka [email protected] Rie [email protected] Yoshie [email protected] Naotaka [email protected] Kenji [email protected] Seinan Jo Gakuin University, Faculty of Health and Welfare, Department of Nutritional Sciences, Kitakyusyu, 803-0835, Japan2 Seinan Gakuin University, School of Human Sciences, Faculty of Social Welfare, Fukoka, 814-8511, Japan2005 31 10 2005 4 31 31 21 7 2005 31 10 2005 Copyright © 2005 Sakamaki et al; licensee BioMed Central Ltd.2005Sakamaki 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
Abnormal body weight, dietary concerns, and unhealthy weight loss behaviors are increasingly being observed in young females in Japan. Our previous research has shown that the irregular lifestyles of female Japanese and Chinese students are significantly related to their desire to be thinner. In the present study, we compare the food habits and body shape preferences of female university students in South Korea and Japan to explore body shape perceptions in those populations.
Methods
A total of 265 female university students aged 19 – 25 years participated in this study. University students in Korea (n = 141) and university students in Japan (n = 124) completed a self-reported questionnaire. Data were analyzed using SPSS statistical software. Descriptive statistics were used to identify the demographic characteristics of the students and parametric variables were analyzed using the Student's t-test. Chi-square analyses were conducted for non-parametric variables.
Results
Comparison of body mass index (BMI) distributions in Japan and Korea showed the highest value in the normal category (74%) together with a very low obesity rate (1.2%). Significant differences were observed between the two countries in terms of eating patterns, with more Japanese eating breakfast daily and with Japanese students eating meals more regularly than Korean students. A difference was also observed in frequency of meals, where Korean students reported eating meals two times per day (59%) and the majority of Japanese students reported eating meals three times per day (81%). Although most subjects belonged to the normal BMI category, their ideal BMI classification was the underweight category (BMI: 18.4 ± 3.4).
Conclusion
Few studies have compared the health related practices of Japanese and Korean university students. The present results suggest the necessity of nutrition and health promotion programs for university students, especially programs emphasizing weight management.
==== Body
Background
South Korea has experienced rapid and varied socioeconomic change during the past three decades. Similar to the experience of Japan, the South Korean nutritional transition has also been very rapid. A large increase in the consumption of animal food products and a reduction in total cereal intake have been reported [1]. Also, incidence of metabolic syndrome is now more than 15% in South Korea despite a low prevalence of obesity [2]. Previously, we studied the health related attitudes and body shape perceptions of female Japanese and Chinese university students and compared them with those of other Asian populations [3,4]. Our results showed that despite a very low prevalence of overweight students, the majority of female subjects in both countries have a desire to be thinner.
Nutritional knowledge, food habits, and body-shape preferences vary across cultures. While information about these health-related factors are important for health educators when implementing health-related education programs, little is known about these factors in Korean University students. Therefore, the purpose of this study was to identify and to compare nutritional knowledge, food habits, and body-shape preferences among female university students in Japan and Korea.
Materials and methods
In 2004, a cross-sectional study was carried out on 150 female Japanese students in Kitakyushu and 130 female Korean students in Seoul. A self-reported questionnaire was administered to 280 students from 19 to 25 years of age. The questionnaire was comprised of three major sections regarding eating, drinking and smoking habits (19 questions), with an additional 4 questions related to body weight. Self-reported height and weight were used to calculate BMI (kg/m2). In the present study, we used the BMI classification of the Japan Society for the Study of Obesity (2000) [5] since use of the BMI classification according to the World Health Organization is based on Caucasian populations and is therefore the subject of debate [6,7]. For reference, a comparison of BMI classification according to the Japan Society for the Study of Obesity and the WHO are shown in Table 1. The questionnaire was designed by the authors and is based on a national dietary survey conducted by the Health and Labor Ministry of Japan. Some of the authors also traveled to Korea to investigate the dietary life of Korean people in order to facilitate questionnaire design. The questionnaire was first written in Japanese and then translated to Korean by a native Korean who teaches the Japanese language in Korea. The translated Korean version was then back-translated to insure accuracy. Informed consent was obtained from all participants of this study, according to the Declaration of Helsinki. The statistical software package SPSS 10.0 was used for all data analysis [8]. Parametric variables were analyzed using the Student's t-test while chi-squared analyses were conducted for non-parametric variables. All analyses were two-tailed, and 'p' values less than 0.05 were considered statistically significant.
Table 1 The comparison of BMI classifications between Japan and WHO. BMI classification of Japan and WHO. Comparison of BMI classification of obesity in Japan and obesity classification according to the WHO
BMI Japan (2000) WHO (1998)
<18.5 underweight underweight
18.5≦-<25 normal weight normal weight
25≦-<30 obese 1 preobese
30≦-<35 obese 2 obese class I
35≦-<40 obese 3 obese class II
40≦ obese 4 obese class III
Results
Sample characteristics and BMI distributions
Survey responses were collected from 95% (265 / 280) of the students. Characteristics of the subjects are shown in Table 2. Of a total of 265 female students, with a mean age of 20 ± 1.9 years, who completed the survey, 124 were Japanese students and 141 were Korean students, with mean ages of 19 ± 1.8 and 20 ± 1.3 years, respectively. The average height was 159.6 ± 5.4 cm, while the average weight was 50.6 ± 6.1 kg. Comparison of BMI distributions between both countries indicated the highest value in the normal category, and a very low rate of obesity (Table 3). According to BMI classifications of the Japan Society for the Study of Obesity (2000), 74.1% of students were classified into the normal BMI range, 24.7% (61/247) students were underweight (BMI < 18.5) and 1.2% (3/247) of students were obese (BMI ≥ 25). No significant difference in BMI was observed between the two countries. The average BMI for Japanese and Korean students was 20.0 ± 2.1 and 19.7 ± 2.0, respectively. Though more Korean students (28.6%) than Japanese students (20.2%) belonged to the underweight BMI category, this observation was not statistically significant.
Table 2 Characteristics of participants. Demographic data of study sample. BMI is based on self-reported height and weight. BMI = weight [kg] / height [m]2
Characteristics of Participants
Total Japan Korea
Variable n = 248 n = 113 n = 135
Age (y) 20.0 ± 1.7 19.0 ± 1.8 20.0 ± 1.3
weight (kg) 50.6 ± 6.1 49.6 ± 6.2 51.3 ± 5.9
height (cm) 159.6 ± 5.4 157.1 ± 5.2 161.7 ± 4.7
BMI (kg/m2) 19.9 ± 2.0 20.0 ± 2.1 20.0 ± 2.1
Table 3 BMI distribution of Japanese and Korean university students. The BMI of Japanese and Korean students was categorized into 3 groups (underweight, normal and obese class1), according to the BMI classification of the Japan Society for the Study of Obesity.
Classification BMI Total Japan (%) Korea (%) p values
Underweight <18.5 61(24.7) 23 (20.2) 38 (28.6) n.s
Normal 18.5≦-<25 183 (74.1) 89 (78.1) 94 (70.7) n.s
Obese class 1 25≦-<30 3 (1.2) 2 (1.8) 1 (0.8) n.s
BMI classification as defined by Japan Society for the Study of Obesity (2000)
Eating habits
Life style practices, in particular food habits, were compared (Table 4). Meal patterns were found to be significantly different between the two countries. Compared to Korean subjects, Japanese reported eating meals more frequently and were also more likely to eat breakfast daily (Japan, 79.0%; Korea, 36.2%; p < 0.01). More than half of the Korean students reported eating meals 2 times per day (58.9%), while the majority of Japanese students (81.0%) eat proper meals three times per day (p < 0.01). Korean students were found to eat fruits and drink alcohol more frequently than Japanese students. Both Japanese (85.4%) and Korean (77.0%) students tend to eat with friends or family members daily or three to four times per week.
Table 4 The life style practices of students in Japan and Korea. Table 4 shows the results of questions related to dietary practices with special reference to eating habits. Meal patterns, consumption of fruits and vegetables, consumption of fried foods, consumption of alcohol were assessed for Japan and Korean students. Behavioral differences between two countries were compared using chi-square analyses. Statistical significance was established at p < 0.05.
Questions Levels Total (%) Japan (%) Korea (%) p values
Do you take your meals regularly always regular 120 (45.3) 74 (59.7) 46 (32.6) **
irregular 145 (54.7) 50 (40.3) 95 (67.4)
Do you always take breakfast daily 149 (56.2) 98 (79.0) 51 (36.2) **
three or four times per week 22 (8.3) 3 (2.4) 19 (13.5)
once or twice per week 37 (14.0) 12 (9.7) 25 (17.7)
rarely 57 (21.5) 11 (8.9) 46 (32.6)
How many times do you eat meals except snacks one time 19 (7.3) 5 (4.1) 14 (9.9) **
two times 100 (38.2) 17 (14.0) 83 (58.9)
three times 141 (53.8) 98 (81.0) 43 (30.5)
four times 2 (0.8) 1 (0.8) 1 (0.7)
How often do you take snacks apart from regular meals daily 92 (34.7) 41 (33.1) 51 (36.2) n.s
three or four times per week 70 (26.4) 28 (22.6) 42 (29.8)
once or twice per week 64 (24.2) 34 (27.4) 30 (21.3)
rarely 39 (14.7) 21 (16.9) 18 (12.8)
How often do you eat green, red or yellow colored vegetables daily 126 (48.1) 70 (56.5) 56 (40.3) *
three or four times per week 46 (17.6) 18 (14.5) 28 (20.1)
once or twice per week 76 (29.0) 30 (24.2) 46 (33.1)
rarely 14 (5.3) 5 (4.0) 9 (6.5)
How often do you eat fruits daily 62 (23.6) 18 (14.5) 44 (31.7) **
three or four times per week 85 (32.3) 44 (35.5) 41 (29.5)
once or twice per week 62 (23.6) 21 (16.9) 41 (29.5)
rarely 54 (20.5) 41 (33.1) 13 (9.4)
How often do you eat fried food daily 12 (4.6) 9 (7.3) 3 (2.2) n.s
three or four times per week 137 (52.1) 67 (54.0) 70 (50.4)
once or twice per week 47 (17.9) 29 (23.4) 18 (12.9)
rarely 67 (25.5) 19 (15.3) 48 (34.5)
How often do you take alcohol daily 2 (0.8) 1 (0.8) 1 (0.7) **
two or three times per week 63 (24.2) 18 (14.5) 45 (33.1)
rarely 195 (75.0) 105 (84.7) 90 (66.2)
How often do you eat with friends and family daily 183 (69.8) 91 (74.0) 92 (66.2) n.s
three or four times per week 29 (11.1) 14 (11.4) 15 (10.8)
once or twice per week 31 (11.8) 5 (4.1) 26 (18.7)
always alone 19 (7.3) 13 (10.6) 6 (4.3)
Please state your smoking history current smoker 37 (14.2) 24 19.8 13 9.4 **
never smoke 223 (85.8) 97 80.2 126 90.6
What type of food do you think you should eat to have a balanced nutrition mainly meat 2 (0.8) 1 (1.4) 1 (0.7) n.s
mainly vegetable 46 (18.0) 21 (9.1) 25 (18.0)
meat, vegetable and other variety of food 181 (70.7) 78 (87.8) 103 (74.1)
others 27 (10.5) 17 (1.7) 10 (7.2)
Differences between current and ideal body weight
Subjects were asked to report a healthy body weight and ideal body weight for their current height (Table 5). Prior to completing the survey, it was explained to respondents that healthy weight maintains the health and wellbeing of individuals while ideal body weight refers to a desired body image figure. Most respondents were found to have a BMI in the normal BMI category. Students indicated a healthy BMI for their current height as 19.2 ± 1.3 for Japanese students and 18.8 ± 4.2 for Korean students. Healthy BMI of Japanese subjects (19.2 ± 1.3) was almost 1 point higher than the healthy BMI of Korean subjects (18.4 ± 4.4). Students' ideal BMI for their current height was also obtained (Japan: 18.4 ± 1.1, Korea: 18.4 ± 4.4); however, no significant differences were observed between the two countries. On average, the ideal weight was 4.0 kg (Japan) and 3.1 kg (Korea) lower than current weight. It should be noted that respondents' ideal BMI values can be classified into the underweight BMI category.
Table 5 Differences between ideal and current body weight
Variable Total Japan Korea p values
n = 248 n = 113 n = 135
Current body weight(kg) 50.6 ± 6.1 49.6 ± 6.2 51.5 ± 5.9 n.s
Healthy body weight (kg) 48.5 ± 8.7 47.6 ± 4.4 49.3 ± 11.0 n.s
Healthy BMI calculated from current height(kg/m2) 19.0 ± 3.3 19.2 ± 1.3 18.8 ± 4.2 n.s
Ideal body weight (kg) 47.0 ± 8.9 45.6 ± 4.0 48.2 ± 11.3 n.s
Ideal BMI calculated from current height(kg/m2) 18.4 ± 3.4 18.4 ± 1.1 18.4 ± 4.4 n.s
Discussion
This study aimed to determine and compare the dietary behavior and body shape perception of university students in Japan and Korea. Accordingly, we recorded the distribution of BMI among Japanese and Korean students and found a significantly low prevalence of obesity, a finding that is consistent with a study of Chinese and Japanese students (BMI ≥ 25 overweight 5.8%; BMI > 30 obese 0%) [4]. Previous reports [9,10] have also indicated a low prevalence of obesity in South Korean adults. As South Korea's economic growth accelerated during the past 3 decades, life style changes have included a unique nutrition transition [9]. Although fast food has become very popular among young Koreans, the traditional dietary patterns and intake of staple foods have been maintained at a higher rate than other Asian countries. A report from the Korea National Health and Nutrition Examination Survey (1998) indicated that the rate of overweight (BMI ≦ 25.0 to < 30.0) and obese (BMI ≦ 30) individuals were low among Korean adults; 23.4% and 1.7% in men and 24.9% and 3.2% in women, respectively. However, high rates of diabetes, hypertension, and dyslipidemia were noted in middle-aged and elderly Koreans, even among individuals with relatively low BMI [10]. According to The National Nutrition Survey in Japan (J-NNS), Japan also experienced dietary change from 1950 to 1970 as a result of rapid economic growth. During the past 50 years, the diet of Japanese people has changed remarkably, with the proportion of fat intake in total energy rising to more than 25% [11]. Our results show a low prevalence of overweight and obese conditions among young female subjects in Japan and Korea; however, health issues related to these conditions certainly exist in the middle-aged and elderly generations. Thus, the importance of health promotion at the disease prevention stage can not be overstated, and similar health education programs should be implemented for university students.
The present research shows that meal patterns for the two countries were significantly different. Japanese students reported eating meals regularly and eating breakfast daily. In contrast, Korean students were significantly less likely to eat breakfast daily and ate meals less frequently. South Korea has shown a unique nutrition transition. A range of government and nutrition specialists have made efforts to retain the traditional diet in Korea. This has resulted in a high consumption of vegetables and low level of fat intake [1]. However, few reports have been published to date regarding the food habits and nutrition knowledge of young adults. Publicity and education programs at schools should also emphasize that a healthy eating pattern parallels the beneficial effects of traditional foods.
Our results revealed that Japanese and Korean students desire body weights that are lower than their actual body weight, with Japanese students desiring thinner figures than Korean students. Similarly, previous research on young Japanese women reported that an ideal weight for their current height was an average of 5.2 kg less than current weight [12]. Body shape perception and ideal body shape are strongly influenced by socioeconomic factors. In western society, many young females are extremely concerned with their body weight and shape. Mass media and pictures in fashion magazines have a strong impact on girls' perceptions of their weight and shape [13]. In addition, weight concern is a predictor of the development of eating disorders of at least subsyndromal severity in young females [14]. Therefore, it is vital that educators guide their students to understand that an ideal weight should take into account optimal physiological function. Instilling young women with this knowledge is of particular importance because excessive weight reduction adversely affects their health and reproductive systems.
Conclusion
The findings of the present study show that BMI distributions of female students in both Japan and Korea have the highest values in the normal BMI category, together with very low obesity rates. In terms of eating patterns, significant differences were observed between the two countries, with more Japanese students reporting eating meals regularly and eating breakfast daily than Korean students. A difference was also observed in meal frequency, where Korean students reported eating meals two times per day and the majority of Japanese students reported eating meals three times per day. Although most subjects belonged to the normal BMI category, their ideal BMI values were classified into the underweight category. Little research has been carried out comparing the dietary habits and body figure perceptions of Japan and Korean students. The present results and previous data from female Chinese university students show a desire for a thinner figure similar to that observed in western society. These findings suggest the need for improved nutrition education for female students, especially education regarding body weight management.
Authors' contributions
R.S carried out questionnaire design, manuscript drafting and total coordination of the study. R.A contributed to the data entry and its analysis. Y. M was actively involved in the study's implementation and in data collection. S.N and K.T contributed to final approval of the manuscript.
Acknowledgements
The authors express their appreciation for the invaluable partnership and support of Dr. Cho Chan-baek of Bae Wha College and all the study participants. We also thank Dr. Shigeki Minakami for valuable comments on the manuscript. This work was supported by a grant from Seinan Jo Gakuin University.
==== Refs
Lee MJ Popkin BM Kim S The unique aspects of the nutrition transition in South Korea: the retention of healthful elements in their traditional diet Public Health Nutr 2002 5 197 203 12027285 10.1079/PHN2001294
Park HS Oh SW Cho SI Choi WH Kim YS The metabolic syndrome and associated lifestyle factors among South Korean adults Int J Epidemiol 2004 33 328 36 15082635 10.1093/ije/dyh032
Rie Amamoto Mina Dozono Kenji Toyama The relationship between dietary life and indefinite complaint in female Nutrition department students Seinan Jo Gakuin Bulletin 2004 75 85
Sakamaki R Toyama K Amamoto R Liu CJ Shinfuku N Nutritional knowledge, food habits and health attitude of Chinese university students -a cross sectional study- Nutr J 2005 4 4 15703071 10.1186/1475-2891-4-4
Japanese Society of Obesity A new judgment and diagnostic criteria for obesity 2000 6 18 28 (in Japanese)
Wang J Thornton JC Russell M Burastero S Heymsfield S Pierson RN Jr Asians have lower body mass index (BMI) but higher percent body fat than do whites: comparisons of anthropometric measurements Am J Clin Nutr 1994 60 23 8 8017333
WHO Expert Consultation Appropriate body-mass index for Asian populations and its implications for policy and intervention strategies Lancet 2004 363 157 63 14726171 10.1016/S0140-6736(03)15268-3
SPSS Inc SPSS Base 7.5 Application guide Chicago 1997 IL: SPSS 53 56
Kim S Moon S Popkin BM The nutrition transition in South Korea Am J Clin Nutr 2000 71 44 53 10617945
Kim Y Suh YK Choi H BMI and metabolic disorders in South Korean adults: 1998 Korea National Health and Nutrition Survey: Obes Res 2004 12 445 53 15044661
Yoshiike N Matsumura Y Iwaya M Sugiyama M Yamaguchi M National Nutrition Survey in Japan J Epidemiol 1996 6 S189 200 8800293
Takasaki Y Fukuda T Watanabe Y Kurosawa T Shigekawa K Ideal body shape in young Japanese women and assessment of excessive leanness based on allometry J Physiol Anthropol Appl Human Sci 2003 22 105 10 12672974 10.2114/jpa.22.105
Field AE Cheung L Wolf AM Herzog DB Gortmaker SL Colditz GA Exposure to the mass media and weight concerns among girls Pediatrics 1999 103 E36 10049992 10.1542/peds.103.3.e36
Taylor CB Sharpe T Shisslak C Bryson S Estes LS Gray N McKnight KM Crago M Kraemer HC Killen JD Factors associated with weight concerns in adolescent girls Int J Eat Disord 1998 24 31 42 9589309 10.1002/(SICI)1098-108X(199807)24:1<31::AID-EAT3>3.0.CO;2-1
|
16255785
|
PMC1298329
|
CC BY
|
2021-01-04 16:39:31
|
no
|
Nutr J. 2005 Oct 31; 4:31
|
utf-8
|
Nutr J
| 2,005 |
10.1186/1475-2891-4-31
|
oa_comm
|
==== Front
Popul Health MetrPopulation Health Metrics1478-7954BioMed Central London 1478-7954-3-111628864810.1186/1478-7954-3-11ResearchDescribing the longitudinal course of major depression using Markov models: Data integration across three national surveys Patten Scott B [email protected] Robert C [email protected] Department of Community Health Sciences, University of Calgary, 3330 Hospital Drive NW, Calgary, Canada2 Health Technology Implementation Unit, Calgary Health Region. Foothills Medical Centre, South Tower, Room 602. 1403 29th Street NW, Calgary, Canada3 Department of Psychiatry, University of Calgary, 1403 – 29 Street NW Calgary, Canada4 Department of Community Health Sciences. University of Calgary, 3330 Hospital Drive NW, Calgary, Canada2005 15 11 2005 3 11 11 7 3 2005 15 11 2005 Copyright © 2005 Patten and Lee; licensee BioMed Central Ltd.2005Patten and Lee; 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
Most epidemiological studies of major depression report period prevalence estimates. These are of limited utility in characterizing the longitudinal epidemiology of this condition. Markov models provide a methodological framework for increasing the utility of epidemiological data. Markov models relating incidence and recovery to major depression prevalence have been described in a series of prior papers. In this paper, the models are extended to describe the longitudinal course of the disorder.
Methods
Data from three national surveys conducted by the Canadian national statistical agency (Statistics Canada) were used in this analysis. These data were integrated using a Markov model. Incidence, recurrence and recovery were represented as weekly transition probabilities. Model parameters were calibrated to the survey estimates.
Results
The population was divided into three categories: low, moderate and high recurrence groups. The size of each category was approximated using lifetime data from a study using the WHO Mental Health Composite International Diagnostic Interview (WMH-CIDI). Consistent with previous work, transition probabilities reflecting recovery were high in the initial weeks of the episodes, and declined by a fixed proportion with each passing week.
Conclusion
Markov models provide a framework for integrating psychiatric epidemiological data. Previous studies have illustrated the utility of Markov models for decomposing prevalence into its various determinants: incidence, recovery and mortality. This study extends the Markov approach by distinguishing several recurrence categories.
Depressive DisorderEpidemiologic MethodsMarkov Chain
==== Body
Introduction
In a series of previous reports, we have described the use of Markov models in major depression epidemiology. In two initial papers, we described a general approach to modeling, in which prevalence was depicted as a steady state outcome of inflow and outflow from a prevalence pool [1,2]. Subsequently, the approach was extended to include strata for age and sex categories [3], other demographic variables [4] and chronic conditions [5]. Broadly speaking, Markov models are useful for medical decision modeling and economic analyses. In the case of major depression, many of the most important health policy decisions relate to the longitudinal course of the condition. Clinical practice guidelines, for example, frequently distinguish between high recurrence and low recurrence groups, subjects with a high risk of recurrence being candidates for long-term treatment. In this report, we describe an application of Markov modeling to description of the longitudinal course of major depression.
Markov models are important in this context because the current literature presents few other options for modeling. To our knowledge, the only other example of an epidemiological general population model is that reported by Kruijshaar et al. [6] using microsimulation. This study integrated data from European (NEMESIS) [7] and Australian studies [8]. The use of more than one data source in this study illustrates the potential usefulness of modeling as a means of integrating the best available epidemiological data into a coherent epidemiological description. For purposes such as surveillance, policy development and cost-effectiveness analysis, data integration using epidemiological modeling is a promising approach.
Markov models, also known as health state transition models, divide a target population into a series of mutually exclusive health states. Transitions between these health states are assigned probabilities and the model's predictions are evaluated over a series of stages [9].
Data Sources
In a series of previous reports, we used data from a Canadian study called the National Population Health Survey (for additional information about the NPHS; see, ) to model the relationship between prevalence, incidence and mortality. The longitudinal component of the NPHS uses a nationally representative probability sample initially consisting of 17,262 subjects who have subsequently been followed through four biannual data collection cycles. The NPHS provides a valuable source of longitudinal health data. A limitation, however, is that the NPHS utilized only a brief predictive instrument for major depression, the Composite International Diagnostic Interview Short Form [10] for Major Depression (CIDI-SFMD). This instrument identifies subjects with a high probability of having met DSM-IV [11] criteria for major depression in the preceding year, but does not provide fully specified data about recurrence. The CIDI-SFMD instrument covers only one half (12 months) of the interval between NPHS interviews (conducted 24 months apart). The CIDI Short Form only covers the final 12 months of this 24-month interval. The proportion of subjects without major depression at one interview who have an episode at the next interview can be directly estimated from the NPHS data, but this estimate provides only an approximation of the annual incidence proportion.
The CIDI-SFMD also includes an item that assesses episode duration. This item elicits the number of weeks during the previous year that a subject with CIDI-SFMD major depression was depressed. Notably, the number of weeks depressed during the preceding year only precisely assesses episode duration in subjects whose episode began and ended within that year. A final limitation of the NPHS is its sample size. Although the longitudinal cohort was relatively large, the number of subjects with major depressive episodes during any particular interview cycle was small enough to result in appreciable imprecision in estimating the weeks depressed in the past year variable. Fortunately, Statistics Canada conducted a related survey, called the Canadian Community Health Survey (CCHS 1.1) during the same time frame and using the same sampling frame and which also included the CIDI-SFMD. The CCHS 1.1 had a sample size of 130,880 subjects, and therefore offered much greater precision for estimating episode duration.
In order to estimate the proportion of the population falling into various recurrence categories, neither of the two data sources listed above were adequate. The NPHS, as noted, provides incomplete longitudinal follow-up and the CCHS 1.1 was strictly cross-sectional. In Canada, the best source of lifetime data is the Canadian Study of Mental Health and Wellbeing, also known as the CCHS 2.1 . This study used the same sampling frame as the other two, but evaluated major depression using the World Health Organization's Mental Health (WMH2000) project WMH-CIDI [12]. This instrument provided a basis for dividing the population into three broad categories of recurrence risk based on the pattern of recurrence reported by the CIDI. Subjects with no prior episodes of major depression (low risk of developing an episode), those having had one episode (moderate risk) and with multiple prior episodes (high risk).
Approach to Markov Modeling
The Markov models developed here adopted the general format of an incidence-prevalence model, modeling the "prevalence pool" [13] for major depression as a function of the inflow to the pool (incidence and recurrence) and the outflow through recovery. For simplicity, mortality was not included in the models presented here since our previous work indicated that this variable did not have an important influence on the epidemiological dynamics [1]. Changes in health state (depressed and not-depressed) were evaluated over a series of one week stages. To deal with declining probabilities of recovery with mounting episode length, a Markov tunnel [9] was used to depict the process of recovery. The Markov tunnel is a way of adding flexibility to a Markov model. The Markov tunnel used in this model consisted of a series of depressed health states, so that at the onset of an episode (by definition at 2-weeks after the onset of symptoms) the subject occupies a health state that represents the first week in an episode. At the next stage, there can be a transition either to the non-depressed state or, alternatively, the subject can progress into a health state representing the next week of the episode. By selecting transitions back to the non-depressed health state that decline with each stage in the tunnel, the pattern of recovery can be flexibly depicted.
As noted above, it was not possible to directly estimate incidence and episode duration from the available data sources. Using tracking variables in the Markov model, it was, however, possible to define variables depicting parameters that are directly estimable: the proportion without an episode at one interview with an episode in the year prior to a subsequent interview and weeks depressed in the past year. The Markov models could then be evaluated by Monte Carlo simulation across a series of possible values for incidence and recovery to find the values most consistent with the observed estimates. The model is presented in Figure 1. The population is depicted in three strata, with size of the strata fixed according to the CCHS 1.2 results.
Figure 1 Framework for the Markov Model Employed in the Project, Depicting Three Recurrence Strata.
Estimation from the Epidemiological Data Sources
Data from the three source studies were employed as described above: the NPHS to calibrate the transition probabilities for episode incidence, the CCHS 1.1 for weeks depressed in past year data and the CCHS 1.2 for the lifetime recurrence pattern. The sample sizes for the three studies were: 17,262 (at baseline), 130,880 and 36,525, respectively. Each survey used the same complex sampling frame. All reported estimates incorporated sampling weights and appropriate statistical procedures to deal with the resulting design effects. The estimates were made using SAS Version 8.1.
Monte Carlo Simulation
The simulation period was set at 312 weeks (i.e. 6 years) in order to depict the duration of data available from the NPHS (data from the first four cycles, 1994 to 2000, had been released at the time when the analysis was conducted). Each simulation used 50,000 Monte Carlo trials to reduce random variation in the simulation output. Tracker variables were used to link the model output to directly estimable parameters, as described above.
A single Markov tunnel was used to describe the recovery pattern in each of the three incidence/recurrence risk categories. Essentially, this represents an assumption that episode prognosis is similar in the various recurrence categories. Successive Monte Carlo simulations were used to identify values for weekly incidence within each category that could explain both the observed overall incidence and duration data, as well as the pattern of recurrence observed over the six year NPHS follow-up period (the proportions of subjects having one, two or three detected episodes).
Results
In the NPHS, the estimated proportion of the population without major depression at baseline who had one or more episodes during the subsequent six years (keeping in mind that the three interviews conducted during the six year follow-up period only covered three of these six years) was 9.3%. Of this population, 7.8% had one episode, 1.3% had two episodes and 0.3% had three episodes (these add to 9.4% due to approximation in rounding). The latter group represented those subjects who were positive on the CIDI-SFMD at each of the three follow-up visits. Some of these CIDI-SFMD positive instances may have represented persistence rather than recurrence, a distinction that could not be made using the NPHS data set.
According to data from the CCHS 1.2, the lifetime prevalence of major depression in the Canadian general population is 12.2% (95% C.I. 11.7% – 12.7%), consistent with European estimates [14] and somewhat less than American [15] estimates using the WMH2000 CIDI instrument. In the NPHS, 12.5% of the subjects had an episode at one or more of the follow-up interviews.
The pattern for "number of weeks depressed" for subjects reporting an episode of major depression is presented as a cumulative proportion in Figure 2. Consistent with existing literature, many subjects had brief episodes, but as expected, the pattern suggested a declining probability of recovery with increasing episode duration. For example, the difference between the proportion of the population reporting n weeks depressed in the past year and n+1 weeks depressed in the past year became smaller as n became larger.
Figure 2 Cumulative Distribution of Reported Number of Weeks Depressed in Past Year, Subjects with CIDI-SFMD Major Depressive Episode.
As expected, the use of higher values for the incidence and recurrence resulted in a larger proportion of subjects with recurrent episodes. Because the proportions falling into the moderate and high recurrence groups were constrained by the CCHS 1.2 estimates, it was possible to identify, using a series of simulations, values for incidence within each recurrence category that resulted in the pattern observed in the NPHS. Figure 3 presents observed and simulated recurrence data for the parameters selected; Figure 4 is a depiction of the model, including values for the transition probabilities. Weekly transition probabilities were 0.00028 per week in the low risk, 0.0010 per week in the moderate risk and 0.00575 in the high risk categories.
Figure 3 Observed and Simulated Proportions in NPHS Recurrence Categories.
Figure 4 Specifications for the Markov Model.
The Markov tunnel describing the weekly recovery probabilities (pr) followed the pattern of pr = 0.18*e-0.09*week so that the probability of recovery was initially very high, but declined by approximately 9% with each week spent in the depressed state. The tunnel was truncated after 26 weeks, such that the probability of recovery once one half of a year was spent in the depressed state remained constant at approximately one half of one percent per week. Figure 5 juxtaposes the observed and simulated number of weeks depressed in the past year.
Figure 5 Observed and Simulated Weeks Depressed in the Past Year.
For ease of interpretation, it is possible to express the weekly transition probabilities as an expected annual cumulative incidence for depression using the formula: annual cumulative incidence = 1 - (1-weekly transition probability)52. Using this formula, the annual incidence in the low, moderate and high recurrence categories are: 1.4%, 5.1% and 25.9%. The weighted overall annual incidence is 3.3%, consistent with existing literature [16]. The mean number of weeks spent in the depressed state was estimated by Monte Carlo simulation in those with a single episode over the 312 week simulation interval, and was found to be 16.8 weeks (median = 9.0 weeks). These results are broadly consistent with the literature, where a range for mean duration of 12 to 30 weeks has been reported [17]. The considerable difference between median and mean duration has also been reported [18], and reflects the very long duration of a minority of episodes. Additional File 1 is the Markov model in the form of a Treeage® Data Pro file, the Markov tunnel is contained in Additional File 2.
Discussion
The model presented here projects a 9.3% overall new-episode frequency over a period of 6 years. Intuition suggests that since the duration of follow-up in the NPHS was essentially 6 years, lifetime prevalence should considerably exceed the observed occurrence of one or more episodes in the NPHS. This was not found to be the case, since lifetime prevalence from the CCHS 1.2, which used the WMH-CIDI was only 12.2%. This may be due to a lack of specificity of the CIDI-SFMD [19], which could bias the NPHS-derived incidence and prevalence estimates upwards. It is also possible that recall bias affected the WMH-CIDI lifetime prevalence estimate, which could bias the lifetime prevalence estimate downward. This interpretation is consistent with Kruijshaar et al.'s microsimulation results, which suggested that lifetime prevalence may actually be 20–30% [6] and is also consistent with the observation that approximately 50% of episodes are forgotten after a 25 year period [20]. Many previous authors have speculated that recall bias may result in underestimation of lifetime prevalence [21-23].
Since lifetime data from the WMH-CIDI were used to calibrate the Markov model presented here, and since the Kruijshaar et al. [6] microsimulation model suggested that recall bias may impact upon such estimates, the Markov model presented here may have been calibrated against an imperfect standard. Since recall bias would probably have the largest impact on remote episodes, the impact on our Markov model would probably be an underestimation of the proportion of the population in the low recurrence categories. In the future, studies with longer-term follow-up will be helpful for clarifying these dynamics more decisively.
Another limitation of these data was the assumption that the three categories identified by the CCHS 1.2 represented the proportion of the population falling into three broad recurrence categories. Subjects with no lifetime major depression were used to approximate the size of a low-risk category in the population, single episodes a moderate risk category and multiple episodes a high risk category. This division was somewhat arbitrary, but needed to be adopted because the NPHS data was also used to estimate incidence (it would have been tautological to estimate incidence from the same proportions that defined recurrence risk).
Figure 4 presents a more advanced Markov-based framework describing major depression epidemiology than previously described Markov models, and the model appears to provide a good description of available Canadian major depression data, despite the specified limitations. To our knowledge, such a method for estimating incidence and recurrence of major depression has not been previously reported. Markov models are commonly used in cost-effectiveness modeling (e.g. see, Sorensen et al. [24]). More recently, integration of data from psychiatric epidemiological surveys and clinical trials for cost-utility analyses has been described [25,26]. While available modeling approaches have limitations, they do provide a methodological framework which should support increasingly meaningful descriptions of major depression epidemiology. Of most importance, it should be possible to refine these models since they provide a platform for integrating the best available information as this becomes available.
Disclaimer
The analyses reported here are based on data collected by Statistics Canada , but do not represent the opinions or interpretations of Statistics Canada.
Supplementary Material
Additional File 1
This is the Markov model, as a Treeage® Data Pro file.
Click here for file
Additional File 2
The Markov model contains a Markov tunnel describing the pattern of recovery. This is the tunnel required – in the form of a table file for Treeage® Data Pro.
Click here for file
Acknowledgements
Dr. Patten is a Health Scholar with the Alberta Heritage Foundation for Medical Research, and a Fellow with the Institute of Health Economics. This study was funded by a grant from the Canadian Institutes of Health Research.
==== Refs
Patten SB Lee RC Refining Estimates of Major Depression Incidence and Episode Duration in Canada using a Monte Carlo Markov Model Med Decis Making 2004 24 351 358 15271273 10.1177/0272989X04267008
Patten SB Lee RC Towards a dynamic model of major depression epidemiology Epidemiologia e Psychiatria Sociale 2004 13 21 28
Patten SB Lee RC Epidemiological theory, decision theory and health services research Soc Psychiatry Psychiatr Epidemiol 2004 39 893 898 15549241 10.1007/s00127-004-0872-z
Patten SB Markov models of major depression for linking psychiatric epidemiology to clinical practice Clinical Practice & Epidemiology in Mental Health 2005 1 2 15967052 10.1186/1745-0179-1-2
Patten SB An analysis of data from two general health surveys found that increased incidence and duration contributed to elevated prevalence of major depression in person with chronic medical conditions J Clin Epidemiol 2005 58 184 189 15680753 10.1016/j.jclinepi.2004.06.006
Kruijshaar ME Barendregt J Vos T de Graaf R Spijker J Andrews G Lifetime prevalence estimates of major depression: An indirect estimation method and a quantification of recall bias Eur J Epidemiol 2005 20 103 111 15756910 10.1007/s10654-004-1009-0
Bijl RV van Zessen G Ravelli A de Rijk C Langendoen Y The Netherlands Mental Health Survey and Incidence Study (NEMESIS): objectives and design Soc Psychiatry Psychiatr Epidemiol 1998 33 581 586 9857790 10.1007/s001270050097
Andrews G Henderson S Hall W Prevalence, comorbidity, disability and service utilisation. Overview of the Australian National Mental Health Survey Br J Psychiatry 2001 178 145 153 11157427 10.1192/bjp.178.2.145
Sonnenberg FA Beck JR Markov models in medical decision making: a practical guide Med Decis Making 1993 13 322 338 8246705
Kessler RC Andrews G Colpe LJ E. H Mroczek DK Normand SL Short screening scales to monitor population prevalences and trends in non-specific psychological distress Psychol Med 2002 32 959 976 12214795 10.1017/S0033291702006074
Association AP Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR) 2000 Washington, American Psychiatric Association
Kessler RC Ustun TB The World Mental Health (WMH) Survey Initiative Version of the World Health Organization (WHO) Composite International Diagnostic Interview (CIDI) Int J Methods Psychiatr Res 2004 13 83 121
Rothman KJ Greenland S Rothman KJ and Greenland S Measures of Disease Frequency Modern Epidemiology 1998 3 2 Philadelphia, Lippincott-Raven 29 64
Investigators ESEMDMHEDEA Prevalence of mental disorders in Europe: results from the European Study of the Epidemiology of Mental Disorders (ESEMeD) project Acta Psychiatr Scand 2004 109(Suppl. 420) 21 27
Kessler RC Berglund P Demler O Jin R Koretz D Merikangas KR Rush JA Waters EE Wang PS The epidemiology of major depressive disorder: results from the National Comorbidity Survey Replication (NCS-R) JAMA 2003 289 3095 3105 12813115 10.1001/jama.289.23.3095
Waraich PS Goldner EM Somers JM Hsu L Prevalence and incidence studies of mood disorders: a systematic review of the literature Can J Psychiatry 2004 49 124 138 15065747
Üstün TB Kessler RC Global burden of depressive disorders: the issue of duration Br J Psychiatry 2002 181 181 183 12204918 10.1192/bjp.181.3.181
Eaton WW Anthony JC Gallo J Cai G Tien A Romanoski A Lyketsos C Natural history of diagnostic interview schedule/DSM-IV major depression. The Baltimore Epidemiological Catchment Area follow-up. Arch Gen Psychiatry 1997 54 993 999 9366655
Patten SB Brandon-Christie J Devji J Sedmak B Performance of the Composite International Diagnostic Interview Short Form for Major Depression in a Community Sample Chron Dis Can 2000 21 68 72
Andrews G Anstey K Brodaty H Issakidis C Luscombe G Recall of depressive episode 25 years previously Psychol Med 1999 29 787 791 10473305 10.1017/S0033291799008648
Eaton WW Anthony JC Gallo J Cai G Tien A Romanoski A Lyketsos C Natural history of diagnostic interview schedule/DSM-IV major depression. The Baltimore epidemiological catchment area follow-up. Arch Gen Psychiatry 1997 54 993 999 9366655
Newman SC Bland RC Incidence of mental disorders in Edmonton: estimates of rates and methodological issues J Psychiatr Res 1998 32 273 282 9789205 10.1016/S0022-3956(98)00011-9
Regier DA Kaelber CT Rae DS Farmer ME Knauper B Kessler RC Norquist GS Limitations of diagnostic criteria and assessment instruments for mental disorders. Arch Gen Psychiatry 1998 55 109 115 9477922 10.1001/archpsyc.55.2.109
Sorensen J Kind P Modelling cost-effectiveness issues in the treatment of clinical depression IMA J Math Appl Med Biol 1995 12 369 385 8919571
Vos T Haby MM Berendregt JJ Kruijshaar M Corry J Andrews G The burden of major depression avoidable by longer-term treatment strategies Arch Gen Psychiatry 2004 61 1097 1103 15520357 10.1001/archpsyc.61.11.1097
Andrews G Issakidis C Sanderson K Corry J Lapsley H Utilising survey data to inform public policy: comparison of the cost-effectiveness of treatment of ten mental disorders Br J Psychiatry 2005 184 526 533 15172947 10.1192/bjp.184.6.526
|
16288648
|
PMC1298330
|
CC BY
|
2021-01-04 16:37:42
|
no
|
Popul Health Metr. 2005 Nov 15; 3:11
|
utf-8
|
Popul Health Metr
| 2,005 |
10.1186/1478-7954-3-11
|
oa_comm
|
==== Front
Reprod Biol EndocrinolReproductive biology and endocrinology : RB&E1477-7827BioMed Central London 1477-7827-3-611628007210.1186/1477-7827-3-61ResearchMixed protocols: Multiple ratios of FSH and LH bioactivity using highly purified, human-derived FSH (BRAVELLE) and highly purified hMG (MENOPUR) are unaltered by mixing together in the same syringe Scobey M Joseph [email protected] Elizabeth [email protected] Dennis C [email protected] Ferring Pharmaceuticals Inc., Suffern, New York, USA2 Qualtech Laboratories Inc., Ocean, New Jersey, USA2005 9 11 2005 3 61 61 6 6 2005 9 11 2005 Copyright © 2005 Scobey et al; licensee BioMed Central Ltd.2005Scobey 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 mixed or blended protocols, that utilize both FSH and hMG, for controlled ovarian hyperstimulation is increasing in use. To reduce the number of injections a patient must administer, many physicians instruct their patients to mix their FSH and hMG together to be given as a single injection. Therefore, the goal of this study was to definitively determine if the FSH and LH bioactivities of highly purified, human-derived FSH (Bravelle(R)) and highly purified hMG (Menopur(R)) were altered by reconstituting in 0.9% saline and mixing in the same syringe.
Methods
Bravelle(R) and Menopur(R) were reconstituted in 0.9% saline and mixed in a Becton Dickinson plastic syringe. The FSH and LH bioactivities of the products were determined after injecting female and male rats, respectively, with Bravelle(R), Menopur(R), or a mixture of Bravelle(R) and Menopur(R). Ratios of FSH:LH activity tested were 150:75 IU (1 vial Bravelle(R): 1 vial Menopur(R)), 300:75 IU (3 vials Bravelle(R): 1 vial Menopur(R)) or 300:225 IU (1 vial Bravelle(R): 3 vials of Menopur(R)).
Results
There were no statistically significant changes in either FSH or LH bioactivity that occurred after mixing Bravelle(R) with Menopur(R) in the same syringe. The theoretical vs. actual FSH bioactivity for Bravelle(R) and Menopur(R) were 75 vs. 76.58 IU/mL and 75 vs. 76.0 IU/mL, respectively. For the 3 ratios of FSH:LH activity tested, 150:75 IU (1 vial Bravelle(R): 1 vial Menopur(R)), 300:75 IU (3 vials Bravelle(R): 1 vial Menopur(R)) or 300:225 IU (1 vial Bravelle(R): 3 vials of Menopur(R)) tested, the theoretical vs. actual FSH bioactivities were 150 vs. 156.86 IU/mL, 300 vs. 308.69 IU/mL and 300 vs. 306.58 IU/mL, respectively. The theoretical vs. actual LH bioactivity for Menopur(R) in the above mentioned ratios tested were 75 vs. 77.50 IU/mL. For the 3 ratios of FSH:LH activity tested, 150:75 IU (1 vial Bravelle(R): 1 vial Menopur(R)), 300:75 IU (3 vials Bravelle(R): 1 vial Menopur(R)) or 300:225 IU (1 vial Bravelle(R): 3 vials of Menopur(R)), the theoretical vs. actual LH bioactivities were 75 vs. 78.38 IU/mL, 75 vs. 78.63 IU/mL and 225 vs. 233.48 IU/mL, respectively.
Conclusion
Mixing human-derived FSH (Bravelle(R)) with highly purified hMG (Menopur(R)) in the same diluent, 0.9% NaCL, does not alter the FSH or LH bioactivity of either gonadotropin preparation.
==== Body
Introduction
Greep and co-workers [1] were the first to demonstrate that follicle stimulating hormone (FSH) and luteinizing hormone (LH) were distinct chemical moieties and that both were necessary for follicular growth, ovulation and formation of the corpus luteum. Sixteen years later, pituitary-derived gonadotropin extracts were used successfully to induce ovulation in anovulatory women [2]. Since then, gonadotropins have been widely used to successfully treat infertile women with oligomenorrhea and amenorrhea that is not secondary to ovarian failure [3].
The first commercially available gonadotropin, human menopausal gonadotropin (hMG; Pergonal®), was purified from the urine of postmenopausal women and contained approximately equal amounts of FSH and LH activity. By adsorbing LH using anti-hCG antibodies and filtration through gel chromatography columns, an FSH only preparation was subsequently produced from postmenopausal urine [4]. In 1986, urinary FSH with <1% LH activity became available for clinical use. A preparation devoid of LH was speculated to have a theoretical advantage over hMG in controlled ovarian hyperstimulation (COH) protocols in women with polycystic ovary syndrome (PCOS) since these women have elevated levels of endogenous LH [5]. However, to date there is no convincing evidence to support that in PCOS patients, FSH alone is more effective than hMG for COH.
Today, COH protocols that utilize both hMG and FSH, known as mixed or blended protocols, are commonly used in an attempt to obtain higher pregnancy rates. To reduce the number of daily injections, physicians often recommend that patients reconstitute the different hormone preparations in the accompanying diluent and then mix them in the same syringe. This practice raises the question of gonadotropin compatibility, especially if products from different manufacturers and/or different diluents are used. Although amino acid sequences of human and recombinant derived FSH are identical, they differ with respect to their glycosylation profile [6]. This can affect the chemical nature of these glycoproteins and impact their interaction with each other in a heterogeneous mixture.
In a recent study [7], it was demonstrated that highly purified, human-derived FSH (Bravelle®; Ferring Pharmaceuticals Inc., Suffern, New York) could be reconstituted in 0.9% saline and mixed with hMG (Repronex®; Ferring Pharmaceuticals Inc., Suffern, New York) in the same syringe, without any alteration of the FSH and LH bioactivities of either product. The objective of the present investigation was to determine if the FSH and LH bioactivities of highly purified, human-derived FSH (Bravelle®) and a new, highly purified hMG (Menopur®; Ferring Pharmaceuticals Inc., Suffern, New York) were altered when reconstituted in 0.9% saline and mixed in the same syringe.
Materials and methods
Study Design
Validated methods [8], specified by the United States Pharmacopeia (USP), were used to determine the bioactivities of each product individually and subsequent mixtures of the two. The bioassays were performed at Qualtech Laboratories Inc., Ocean, New Jersey. Ovarian weights (FSH bioassay) or seminal vesicle weights (LH bioassay) were obtained after treating prepubertal female or male rats with Bravelle® (75 IU FSH; ≤2% LH activity), Menopur® (75 IU FSH; 75 IU LH activity), or one of three mixtures of Bravelle® and Menopur®: 150:75 IU (1 vial Bravelle®:1 vial Menopur®); 300:75 IU (3 vials Bravelle®: 1 vial Menopur®); and 300:225 IU (1 vial of Bravelle®:3 vials of Menopur®). These ratios represent a low combination of FSH: LH activity (150:75 IU) as well as combinations with a high FSH and low LH activity (300:75 IU) and high FSH and high LH activity (300:225 IU). The FSH and LH bioactivities were compared to a menotropin Reference Standard traceable to the National Institute for Biological Standards and Control (NIBSC).
Test Animals
In-bred Wistar female rats (n = 296) 21 days of age were used for the FSH bioassay and Sprague Dawley male rats (n = 248) 21 days of age were used for the LH bioassay. The rats, certified to be free from murine viruses, were purchased from Hilltop Lab Animals Inc. (Scottdale, Pennsylvania). Upon arrival, the animals were weighed, their health was assessed and they were randomized into groups of eight. The animals were housed in polypropylene solid bottom cages with stainless steel wire lids containing shredded Aspen shavings. The animal room was maintained at 18–19°C with a relative humidity of approximately 50% and 12:12 hour, light: dark cycle. Rats were given Purina rodent diet and tap water ad libitum.
For both the FSH and LH assays, one group of rats (N = 8) was assigned to each of the low, middle, and high concentration groups for each gonadotropin, gonadotropin mixture and Reference Standard. Each assay was done in duplicate, Therefore, for the FSH assay, 48 rats each were assigned to six groups: Bravelle®, Menopur®, the three ratios of Bravelle® plus Menopur®, and the Reference Standard. For the LH assay, 48 rats each were assigned to five groups: Menopur®, the three ratios of Bravelle® plus Menopur®, and the Reference Standard. In addition, there was a control group in each assay that had eight rats treated with diluent alone.
Reference Standards for the FSH and LH assays
The Reference Standard diluent used for both gonadotropins was prepared by dissolving 10.75 g disodium hydrogen phosphate, 7.6 g sodium chloride and 1.0 g bovine serum albumin in 1.0 liter of distilled water. The pH of the solution was adjusted to 7.2 ± 0.2 with 1 N sodium hydroxide.
As per the USP, for the FSH assay 70,000 units of hCG were added to the Reference Standard diluent. A menotropin Reference Standard (Ferring, lot-DPH12250397, 122.2 FSH IU/mg, traceable to NIBSC) was reconstituted and diluted in the Reference Standard diluent to the final concentrations of 1.9, 3.8 and 7.6 IU/0.6 mL.
For the LH assay, a menotropin Reference Standard (Ferring, lot-DPH12250397, 99.3 LH IU/mg, traceable to NIBSC) was reconstituted and diluted in the Reference Standard diluent to the final concentrations of 7, 14, and 28 IU/0.8 mL.
The Reference Standard concentrations were established in a geometric progression for the low, middle and high doses respectively, based on previous dose response studies. The lowest concentration in this three-dose range produces a definite response in some of the rats as compared to the control group and the highest concentration produces a submaximal to maximal response.
Gonadotropin Assays
Bravelle® and Menopur® single solutions
For the FSH assay, four vials of Bravelle® (lot-FMA001) or Menopur® (lot-FHA004ULB) were each individually reconstituted with 1.0 mL of 0.9% saline and pooled. Each solution was adjusted with Reference Standard diluent to the same three concentrations as the reference standard (1.9, 3.8 and 7.6 IU/0.6 mL). For the LH assay, fourteen vials of Menopur® were each individually reconstituted with 1.0 mL of 0.9% saline and pooled. The solution was adjusted with Reference Standard diluent to the same three concentrations as the reference standard (7, 14 and 28 IU/0.8 mL).
Analyte (Mixture of Bravelle® and Menopur®)
150:75 IU; FSH: LH
For both the FSH and LH bioassay 20 vials of Bravelle® (lot-FMA001) were each individually reconstituted with 1.0 mL of 0.9% saline and pooled. For the FSH assay, two vials of Menopur® (lot-FHA004ULB) were each individually reconstituted with 1.0 mL of the Bravelle mixture and pooled. For the LH assay analyte, 14 vials of Menopur® (lot-464-891) were each individually reconstituted with 1.0 mL of the Bravelle® mixture and pooled.
300:75 IU; FSH: LH
For both the FSH and LH bioassay 56 vials of Bravelle® (lot-FMA001) were each individually reconstituted with 0.5 mL of 0.9% saline and pooled (diluent A). Another 20 vials of Bravelle® (lot-FMA001) were each individually reconstituted with 1.0 mL of diluent A and pooled (diluent B). For the FSH assay, two vials of Menopur® (lot-FHA004ULB) were each individually reconstituted with 1.0 mL of diluent B and pooled. For the LH assay analyte, 14 vials of Menopur® (lot-FHA004ULB) were each reconstituted with 1.0 mL of diluent B and pooled.
300:225 IU; FSH: LH
For both the FSH and LH bioassay, 29 vials of Menopur® (lot-FHA004ULB) were each individually reconstituted with 0.5 mL of 0.9% saline and pooled (diluent A). Another 10 vials of Menopur® (lot-FHA004ULB) were individually reconstituted with 1.0 mL of diluent A and pooled (diluent B). For the FSH assay, two vials of Bravelle® (lot-FMA001) were each reconstituted with 1.0 mL of diluent B and pooled. For the LH assay, five vials of Bravelle® (lot-FMA001) were each individually reconstituted with 1.0 mL of diluent B and pooled.
For all mixtures, reconstitution and pooling were done with commercially available BD plastic syringes. The solutions were further diluted to the same three concentrations as the Reference Standard with Reference Standard diluent: 1.9; 3.8; and 7.6 IU/0.6 mL for the FSH assay and 7; 14; and 28 IU/0.8 mL for the LH assay.
Control Solution
The Reference Standard diluent was used as the control solution for both assays.
FSH assay
Each female rat was injected subcutaneously with 0.2 mL of the assigned dose at approximately the same time of day for three consecutive days. Twenty-four hours after the last injection, rats were sacrificed in a carbon dioxide chamber. Left and right ovaries were carefully dissected from each rat, freed of fat or fibrous tissue, dried by gentle blotting on absorbent paper and weighed on an analytical balance. For each rat, the left and right ovarian weights were recorded.
LH assay
Each male rat was injected with 0.2 mL of the assigned dose at approximately the same time of day for four consecutive days. Twenty-four hours after the last injection, rats were sacrificed in a carbon dioxide chamber. Seminal vesicles were carefully dissected from each rat, freed of fat or fibrous tissue, dried by gentle blotting on absorbent paper and weighed on an analytical balance.
Determination of FSH and LH bioactivity: potency calculations
The FSH and LH assays were 3 × 3 parallel line assays performed in duplicate. For each assay, the response to three concentrations of the test hormones and the analyte were compared to the response of the same three concentrations of the Reference Standard. The result from each replicate was combined for the final result. For each replicate, ovarian weights or seminal vesicle weights were used to calculate potency values for FSH and LH, respectively.
For potency values to be accepted, the combined L-value (the confidence limit) from the duplicate assays had to be less than 0.08 (L-value set by Ferring Pharmaceuticals Inc.) which means that the true bioactivity of the preparation was within 93–107% of the obtained result. By Ferring imposed standards, in order to have accurate, consistent assesments of bioactivity, if an L-value was not met, the bioassay was to be repeated until the combined L-value was less than 0.08, providing a more exacting bioassay. This is a more stringent criterion than specified in the USP monograph on menotropins, which only requires an L-value of 0.18 (a value ±21% of the obtained result) [8].
Potency calculations were the same for the FSH and LH bioassays. The following equations were used: M' = ciTa/Tb; where M' = log-potency of an unknown relative to its assumed potency; c = 4/3; i = interval in logarithms between successive log-doses (same for standard and test concentration); Ta = difference in responses between the standards and the test concentrations, Tb = combined slope of the dose-response curves for the standards and test concentrations. After calculating M', the log-potency was determined by the equation: M = M'+log R, (M = log potency; for the present experiment: R = 1 making log R = 0). Therefore, potency = antilog M, and % claim = antilog M × 100 [8]. FSH and LH bioactivities obtained by these assays were compared with the FSH and LH bioactivities set forth in the product label (labeled claim).
Statistical analysis
An analysis of variance (ANOVA) was performed to determine if there were differences among ovarian or seminal vesicle weights between replicates. This analysis included: replicate; treatment; dose; replicate × treatment; replicate × dose; and replicate × treatment × dose, as sources of variation. Since replicate × treatment × dose analyses were not significant (P > 0.05), ovarian weights from both replicates were combined as were seminal vesicle weights for both replicates.
An ANOVA for a randomized block design was next performed to determine if there were differences in ovarian or seminal vesicle weights among treatments. This analysis included: replicate (block); treatment; dose and treatment × dose, as sources of variation.
Results
FSH Bioassay
The ovarian weights (mean ± SEM) were not significantly different among treatments (Table 1). As expected, ovarian weights increased in a dose-dependent manner with increasing doses of FSH. The magnitude of increase in ovarian weights was similar across treatment groups (treatment × dose, P > 0.05; Table 1), indicating that FSH bioactivity was unaffected by either reconstitution of hormones with 0.9% saline or by mixing the reconstituted hormones in the same syringe. The theoretical and actual FSH bioactivities for Bravelle®, Menopur® and a mixture of Bravelle® and Menopur® are shown in Table 2. The bioactivities of the five groups expressed as a percent of the labeled claims are shown in Figure 1a.
Table 1 Ovarian weights in rats injected with a low, middle and a high dose of the Reference Standard, Bravelle®, Menopur®, or a mixture of Bravelle® and Menopur® reconstituted in 0.9% saline and mixed in the same syringe.
Hormone (FSH:LH)a Total no. rats Ovarian weights (mean ± SEM)b*
Lowc dose Middled dose Highe dose
Reference Standard 48 93.55 ± 4.17 151.37 ± 4.00 193.01 ± 5.02
Bravelle® (75:0) 48 92.31 ± 3.02 149.60 ± 3.29 201.37 ± 4.56
Menopur® (75:75) 48 92.19 ± 3.45 151.51 ± 2.62 194.92 ± 4.03
Bravelle® + Menopur®
(150:75) 48 97.29 ± 2.80 148.82 ± 3.65 199.59 ± 4.30
(300:75) 48 92.38 ± 3.37 157.08 ± 3.82 192.62 ± 5.50
(300:225) 48 95.75 ± 2.94 149.50 ± 6.11 195.39 ± 6.13
* The magnitude of this increase was similar across treatment groups (treatment × dose P = 0.71)
a Ratio of FSH:LH bioactivity expressed in IUs
b Combined ovarian weights from the 2 replicates (replicate × treatment × dose)
c Low dose = 1.9 IU FSH/0.6 mL
d Middle dose = 3.8 IU FSH/0.6 mL
e High dose = 7.6 IU FSH/0.6 mL
Table 2 Theoretical and actual FSH bioactivities for Bravelle®, Menopur®, and a mixture of Bravelle® and Menopur® when reconstituted in 0.9% saline and mixed in the same syringe.
Hormone (FSH:LH)a Theoretical bioactivity IU/mL (%) Actual bioactivity IU/mL (% claim) L-valueb
Bravelle® (75:0) 75 (100%) 76.58 (102.1%) 0.072
Menopur® (75:75) 75 (100%) 76.0 (101.3%) 0.072
Bravelle® + Menopur®
(150:75) 150 (100%) 156.86 (104.6%) 0.072
(300:75) 300 (100%) 308.69 (102.9%) 0.072
(300:225) 300 (100%) 306.58 (102.2%) 0.072
a Ratio of FSH:LH bioactivity expressed in IUs
b L-value refers to the confidence limit obtained after combining two replicates. The USP mandated L-value = 0.18; the more stringent L-value of 0.08 is set by Ferring Pharmaceuticals Inc. to more precisely and consistently assess bioactivity.
Figure 1 Theoretical vs. actual FSH and LH activities. A) FSH bioactivities of Bravelle, Menopur and the three ratios of Bravelle and Menopur expressed as a percent of labeled claim. B) LH bioactivities of Menopur and the three ratios of Bravelle and Menopur expressed as a percent of labeled claim.
LH Bioassay
The seminal vesicle weights (mean ± SEM) were not significantly different among treatments (Table 3). As expected, seminal vesicle weights increased with increasing doses of LH (Table 3). The magnitude of this increase was similar across treatment groups (treatment × dose, P > 0.05; Table 3), indicating that LH bioactivity was unaffected by either reconstitution of hormones with 0.9% saline or by mixing the hormones in the same syringe. The theoretical and actual bioactivities for Menopur® and a mixture of Bravelle® and Menopur® are shown in Table 4. The bioactivities of the four groups expressed as a percent of the labeled claims are shown in Figure 1b.
Table 3 Seminal vesicle weights in rats injected with a low, middle and a high dose of the Reference Standard, Menopur®, or a mixture of Bravelle® and Menopur® reconstituted in 0.9% saline and mixed in the same syringe.
Hormone (FSH:LH)a Total no. rats Seminal Vesicle weights (mean ± SEM)b*
Lowc dose Middled dose Highe dose
Reference Standard 48 35.82 ± 1.64 68.88 ± 2.40 96.39 ± 2.95
Menopur® (75:75) 48 36.49 ± 1.54 72.55 ± 1.61 95.69 ± 2.87
Bravelle®+ Menopur®
(150:75) 48 36.47 ± 1.23 73.94 ± 2.42 96.62 ± 1.70
(300:75) 48 37.80 ± 1.25 74.57 ± 1.57 94.49 ± 2.09
(300:225) 48 36.28 ± 1.50 74.78 ± 1.78 95.73 ± 2.01
* The magnitude of this increase was similar across treatment groups (treatment × dose P = 0.74)
a Ratio of FSH:LH bioactivity expressed in IUs
b Combined seminal vesicle weights from the 2 replicates (replicate × treatment × dose)
c Low dose = 7 IU LH/0.8 mL
d Middle dose = 14 IU LH/0.8 mL
e High dose = 28 IU LH/0.8 mL
Table 4 Theoretical and actual LH bioactivities for Menopur®, and a mixture of Bravelle® and Menopur® when reconstituted in 0.9% saline and mixed in the same syringe.
Hormone (FSH:LH)a Theoretical bioactivity IU/mL (%) Actual bioactivity IU/mL (% claim) L-valueb
Menopur® (75:75) 75 (100%) 77.50 (103.3%) 0.061
Bravelle® + Menopur®
(150:75) 75 (100%) 78.38 (104.5%) 0.061
(300:75) 75 (100%) 78.63 (104.8%) 0.061
(300:225) 225 (100%) 233.48 (103.8%) 0.061
a Ratio of FSH:LH bioactivity expressed in IUs
b L-value refers to the confidence limit obtained after combining two replicates. The USP mandated L-value = 0.18; the more stringent L-value of 0.08 is set by Ferring Pharmaceuticals Inc. to more precisely and consistently assess bioactivity.
While Bravelle® contains up to 2% LH, due to the small amount, for this assay it is assumed that Bravelle® does not contain LH activity.
For both the FSH and LH bioassays, the L-value obtained was less than 0.08 which means that the true bioactivity is within 93–107% of the obtained result. Since all of the required USP specifications for the bioassays were met, it is concluded that the resulting bioactivity for LH and FSH in the mixed preparations was unaffected by reconstituting Bravelle® and Menopur® in 0.9% saline and mixing in a BD plastic syringe.
Discussion
In many cases, physicians are administering FSH and hMG as separate injections because of concerns that different gonadotropins may not be compatible, which could result in unexpected results during controlled ovarian hyperstimulaiton (COH) cycles. This concern for compatibility has merit since different gonadotropins are either from animal or human origin and use different diluents for reconstitution and/or require different delivery systems. It is well known that even structurally similar proteins, expecially those of high molecular weight can be incompatible. FSH, LH and hCG gonadotropins have high molecular weight (27–31 kD) and require surfactants to prevent sticking. Nonetheless, in an effort to minimize the number of injections that a woman must take while undergoing COH, some physicians are combining FSH and hMG in the same syringe. The charge distribution (described by the isoelectric point) and the carbohydrate complexity (simple, intermediate and complex carbohydrates) of recombinant FSH was found to be quite different when compared with serum FSH throughout the menstrual cycle [11]. Therefore, when hormones with different chemical configurations are reconstituted in different diluents and mixed, their compatibility and possibly their bioactivity may be affected.
A previous study demonstrated that the activity of peptides may be significantly changed by the diluents in which they are dissolved [12] or the characteristic of the vessel used for mixing. Studies have shown that peptides and proteins adhere to certain surfaces [13]. Surface adsorption of calcitonin on soda lime silica glass is pH dependent [13]. Fibrinogen adheres to both dimethyldichlorosilane-treated glass and low-density polyethylene [14].
We must emphasize that the results from this study and a previously published study [7] were obtained with FSH and LH activity containing preparations from a single manufacturer (Ferring Pharmaceuticals Inc.) and that the hormone preparations were reconstituted and mixed according to the manufacturer's directions. These data cannot be extended to other preparations of gonadotropins, particularly if recombinant hormones that require bacteriostatic water for reconstitution or other delivery vehicles are mixed with human-derived hormone preparations.
In summary, the results of this study confirm those of a previous study [7] combining Bravelle® with Repronex®. The present study was conducted because Menopur is a more highly-purified hMG than Repronex®, thereby resulting in significantly fewer injection site reactions [15]. Since Menopur is now commercially available for use in ART, it is important for doctors to have confidence that Bravelle® and Menopur® can be reconstituted with 0.9% saline (diluent provided by the manufacturer) and mixed together in a commercially available syringe without altering the FSH or LH bioactivity. The health care provider can be assured that the expected doses of FSH and LH will be delivered, with no alteration in bioactivity of either Bravelle® or Menopur®.
==== Refs
Greep RO Van Dyke HB Chan BF Gonadotropins of the swine pituitary: various biological effects of purified thylakentrin (FSH) and pure metakentrin (ICSH) Endocrinolology 1942 30 635 649
Gemzell CA Diczfalusy E Tillinger KG Clinical effects of human pituitary follicle stimulating hormone (FSH) J Clin Endocrinol Metab 1958 18 1333 1348 13611018
March CM Diamond MP, De Cherney AH Human menopausal gonadotropins Infertility and reproductive medicine Clinics of North America 1990 1 Philadelphia: WB Saunders 59 77
Donini P Puzzuoli D D'Alessio I Lunenfeld B Eshkol A Parlow AF Purification and separation of follicle stimulating hormone (FSH) and luteinizing hormone (LH) from human postmenopausal gonadotropin (HMG). II. Preparation of biological apparently pure FSH by selective binding of the LH with an anti-hCG serum and subsequent chromatography Acta Endocrinol (Copenh) 1966 52 186 198 5952830
Homburg R Armar NA Eshel A Adams J Jacobs HS Influence of serum luteinizing hormone concentrations on ovulation, conception, and early pregnancy loss in polycystic ovary syndrome BMJ 1988 297 1024 1026 3142595
Lambert A Rodgers M Mitchell R Wood AM Wardle C Hilton B Robertson WR In-vitro biopotency and glycoform distribution of recombinant human follicle stimulating hormone (Org 32489), Metrodin and Metrodin-HP Hum Reprod 1995 10 1928 1935 8583012
Marshall DC Raike E De Silva M Nardi RV Mixed protocol same syringe combination of gonadotropins: compatibility of Ferring's new highly-purified, human-derived FSH (Bravelle®) and hMG (Repronex) Today's Therapeutic Trends 2001 19 213 224
The United States Pharmacopeia, 26th rev, and The National Formulary 2003 21 Rockville: The United States Pharmacopeial Convention 2033 2045
Dickey RP Nichols JE Steinkampf MP Gocial B Thornton M Webster BW Bello SM Crain J Marshall DC Highly purified human-derived follicle-stimulating hormone (Bravelle®) has equivalent efficacy to follitropin-beta (Follistim®) in infertile women undergoing in vitro fertilization Reprod Biol Endocrinol 2003 1 63 14609434 10.1186/1477-7827-1-63
Gocial B Keye WR Fein SH Nardi RV Subcutaneously administered Repronex in female patients undergoing in vitro fertilization is as effective and well tolerated as intramuscular menotropin treatment Fert Ster 2000 74 73 79 10.1016/S0015-0282(00)00605-1
Horsman G Talbot JA McLoughlin JD Lambert A Robertson WR A biological, immunological and physico-chemical comparison of the current clinical batches of the recombinant FSH preparations Gonal-F and Puregon Hum Reprod 2000 15 1898 1902 10966982 10.1093/humrep/15.9.1898
Corbo DC Suddith RL Sharma B Naso RB Stability, potency, and preservative effectiveness of epoetin alfa after addition of a bacteriostatic diluent Am J Hosp Pharm 1992 49 1455 1458 1529989
Law SL Shih CL Adsorption of calcitonin to glass Drug Dev Ind Pharm 1999 25 253 256 10065361 10.1081/DDC-100102168
Amiji M Park K Prevention of protein adsorption and platelet adhesion on surfaces by PEO/PPO/PEO triblock copolymers Biomaterials 1992 13 682 692 1420713 10.1016/0142-9612(92)90128-B
Keye WR Webster B Dickey R Somkuti S Crain J Scobey MJ Subcutaneously Administered Menopur®, A New Highly Purified Human Menopausal Gonadotropin Causes Significantly Fewer Injection Site Reactions Than Repronex® in Subjects Undergoing In Vitro Fertilization Reprod Biol Endocrin
|
16280072
|
PMC1298331
|
CC BY
|
2021-01-04 16:37:13
|
no
|
Reprod Biol Endocrinol. 2005 Nov 9; 3:61
|
utf-8
|
Reprod Biol Endocrinol
| 2,005 |
10.1186/1477-7827-3-61
|
oa_comm
|
==== Front
Reprod Biol EndocrinolReproductive biology and endocrinology : RB&E1477-7827BioMed Central London 1477-7827-3-631628197310.1186/1477-7827-3-63ReviewZebrafish sex determination and differentiation: Involvement of FTZ-F1 genes von Hofsten Jonas [email protected] Per-Erik [email protected] Department of Molecular Biology, Umeå University, SE-901 87 Umeå, Sweden2 Örebro Life Science Center, Department of Natural Science, Örebro University, SE-701 82 Örebro, Sweden2005 10 11 2005 3 63 63 21 10 2005 10 11 2005 Copyright © 2005 von Hofsten and Olsson; licensee BioMed Central Ltd.2005von Hofsten and Olsson; 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.
Sex determination is the process deciding the sex of a developing embryo. This is usually determined genetically; however it is a delicate process, which in many cases can be influenced by environmental factors. The mechanisms controlling zebrafish sex determination and differentiation are not known. To date no sex linked genes have been identified in zebrafish and no sex chromosomes have been identified. However, a number of genes, as presented here, have been linked to the process of sex determination or differentiation in zebrafish. The zebrafish FTZ-F1 genes are of central interest as they are involved in regulating interrenal development and thereby steroid biosynthesis, as well as that they show expression patterns congruent with reproductive tissue differentiation and function. Zebrafish can be sex reversed by exposure to estrogens, suggesting that the estrogen levels are crucial during sex differentiation. The Cyp19 gene product aromatase converts testosterone into 17 beta-estradiol, and when inhibited leads to male to female sex reversal. FTZ-F1 genes are strongly linked to steroid biosynthesis and the regulatory region of Cyp19 contains binding sites for FTZ-F1 genes, further linking FTZ-F1 to this process. The role of FTZ-F1 and other candidates for zebrafish sex determination and differentiation is in focus of this review.
==== Body
Sex determination
Among mammals sex is usually defined by the presence or absence of the sex specific chromosome Y. In many, but not all, fish species there is also a chromosomal background to sex determination. Several fishes, including most salmonids, have heterogametic males and homogametic females, similar to the mammalian XY/XX-system [1-3]. Other species, such as Poecilia, have homogametic males and heterogametic females (ZZ/ZW), which also is the case for birds [4]. Some species of the Poecilid platyfish Xiphophorus, utilize a system with three sex chromosomes [5]. In yet other species sex determination is influenced by environmental factors such as the temperature surrounding the developing embryo [6-8]. Hermaphroditism is also a common feature of several fish species. Several studies have shown that species with genetic sex determination can be directed to produce genetically sex-reversed offspring. This is accomplished either by treating the fish with hormones, which can induce sex reversal in synchronous hermaphroditic fish [9,10] and masculinization/feminization in gonochoristic species, or by incubating embryos in certain temperatures or pH [11]. The proportion of males usually increases with temperature whereas lower temperatures favour females. In the case of pH, species differences have been observed.
There are few studies of sex determination in fish and the genetic mechanisms behind sex determination in fish remain largely unknown. Environmental factors, including endocrine disrupters such as diethylstilbestrol, PCBs or dioxins, can affect both teleost and mammalian reproductive systems, but do not seem to alter sex ratios or cause sex reversals in mammals. This indicates that mammalian sex determination is more strictly genetic, and shows less gonad plasticity than teleosts. However, it has been observed that a number of genes, both sex-linked and autosomal, display dosage effects in mammals (Table 1), suggesting that allelic variants could account for differences in gene function.
Table 1 Chromosomal location and dose effects. Several genes involved in mammalian sex determination have dose effects leading to sex reversal.
Chromosomal location Gene Number of copies Phenotype Reference
Y Chromosome SRY 0 Female
1 Male [12]
X Chromosome Dax-1 1 Normal
2 Female XY [13]
Autosomal Sox9 1 Female XY [14,15]
Masculinizing 2 Normal
3 Male XX [16]
Autosomal SF-1 1 Female XY [17]
Masculinizing 2 Normal
Autosomal WT1 1 Female XY [18]
Masculinizing 2 Normal
Autosomal Dmrt1 1 Female XY [19]
Masculinizing 2 Normal
While the developmental mechanisms by which the mammalian gonads are formed have been thoroughly studied and several genes involved have been identified, only a few of these genes have been identified in fish. The functions of these genes have not been fully elucidated in fish and both conserved and divergent functions between mammals and fish have been suggested. As zebrafish is an important vertebrate model for developmental biology it is vital that the basic developmental mechanisms of sex determination are further studied in this species. In the present review we discuss the roles of genes involved in sex determination with a focus on the potential role of FTZ-F1 genes in zebrafish sex determination and differentiation. From the present knowledge of these genes in zebrafish we attempt to present a model for zebrafish sex determination and differentiation.
Formation and differentiation of gonads
There is a close anatomical relationship between the development of the genital ridge and the excretory system during early ontogeny of all vertebrates, including fish. A mesodermal layer ventral to the somites differentiates into structures involved in excretion and reproduction. There are species differences in how closely connected these structures are in regard to sharing ducts for secretion [20]. The teleost gonads are similar to those in mammals. The testis contains Sertoli and Leydig cells in addition to germ cells, and the ovary consists of thecal cells and granulosa cells surrounding the ovum. In both teleosts and mammals the interstitial cells (Leydig and theca), Sertoli cells and granulosa cells are of the same origin. An important difference is that the mammalian gonads are terminally developed into either testis or ovary, while fish gonads often retain the ability to change, making them sequential hermaphrodites [21]. Immature teleost gonads can be directed to develop into testes or ovaries, regardless of chromosomal background, by hormone treatment [22,11]. Far too few fish species have been studied with regard to gonad development to be able to develop a general model of how this occurs.
The zebrafish has become a useful vertebrate model system and is probably the most studied fish in developmental biology. The zebrafish diploid genome consists of 50 chromosomes and no specific sex chromosomes have been identified. The use of synaptonemal complex studies rather indicates that no sex chromosomes exist in zebrafish [23,24]. Teleosts have a partially duplicated genome, which in zebrafish has been determined by studying HOX-clusters [25], and this further complicates the elucidation of potential sex linked genes. In theory, there may be smaller genomic differences that account for or direct the development toward two separate sexes. By studying patterns of inheritance, zebrafish have been suggested to have XY-like chromosomal background [26]. However, the opposite system, with heterogametic females in zebrafish has also been suggested [27]. The inconsistency of results regarding sex chromosomes in zebrafish suggests that the sex determining system is labile and that no clear sex determining chromosome exist. The zebrafish is sexually mature after approximately three months, but separate sexes can be detected after 21–23 days post fertilization (dpf) [26]. Prior to sex differentiation all zebrafish develop ovary-like gonads, regardless of chromosomal background. Ovarian development is the default pathway, which is initiated after 10 dpf and progress until 20 dpf. At 21 dpf until approximately 30 dpf testis development is initiated in males simultaneously with ovarian apoptosis.
The gonad development in zebrafish begins during embryogenesis. Using Vasa as a marker gene, germ cells can be detected in the area ventral to the third to fifth somite at the six-somite stage [28]. While germ cells can be detected earlier, they are not properly positioned until around the 6 somite stage. So far no studies have been made regarding markers for gonadal steroidogenic precursor cells, rendering it difficult to know exactly where these cells are located in the embryo. But, it is likely that these cells derive from an area close to where Vasa is detected at the six-somite stage. The zebrafish Wilms Tumour-1 (WT1) is also initially detected in an area corresponding to that of Vasa but is later expressed in the pronephric ducts [29], making this region probable for the development of the rest of the zebrafish gonadal cells.
Candidate genes in zebrafish sex determination
In zebrafish, little information exists regarding sex determination and the potential presence of sex chromosomes. To date, no sex-linked genes have been identified. However, a number of genes, as presented here, have been linked to the process of sex determination or differentiation in zebrafish. Since none of the genes have been shown to be sex-linked, it is not likely that any of the below listed genes is the single factor responsible for specifying sex in zebrafish. Still, the expression patterns and regulatory mechanisms of these genes leads to the conclusion that they are part of a signalling network responsible for the development of sex specific gonads. In line with observations on mammals (Table 1) gene dosage effects may be a factor involved in zebrafish sex determination. Since no sex-linked genes have been found in zebrafish, allelic variants and dosage effects of autosomal genes, such as the Fushi Tarazu factor-1 (FTZ-F1) genes, SRY HMG box related gene 9 (Sox9), WT-1, Anti-Mullerian Hormone (AMH), doublesex-mab 3 related gene (Dmrt1) and GATA4 (a zinc finger transcription factor) may be involved in determining sex and directing gonad development. The dosage dependent region on X (Dax-1) is highly involved in female sex determination in mammals, but no dax-1 gene homologue has so far been identified in zebrafish. The Dax-1 gene has however been identified in the Nile Tilapia [30], suggesting that other fish species may also have Dax-1 homologues.
SOX9
Even when sex determination in teleost fish has a genetic background, they lack an equivalent to the testis-determining factor SRY found in mammals. However, several HMG-box containing genes, Sox-genes, have been identified in fish [31-33]. In zebrafish, two Sox9 genes, termed Sox9a and Sox9b, have been identified. Both contain the HMG-box and are able to bind the AACAAAG recognition site in a similar manner as murine Sox9 [34]. The expression patterns of Sox9a and b are dissimilar in adult zebrafish. Sox9a displays a broad expression pattern and has been found in brain, kidney, muscle, testis and pectoral fin, whereas Sox9b is only found in ovary. During embryogenesis Sox9a and b are both expressed in cells involved in craniofacial development and in the brain [34]. In addition Sox9a has been shown to be essential for chondrogenic development [35] and Sox9b has been implicated to be involved in neural crest development [36]. Whether Sox9a and/or b are involved in sex determination or differentiation has so far not been studied. However, an HMG-Box cis element has been identified in gene promoter of fushi tarazu factor 1a (ff1a) [37]. Sox9a is also able to specifically bind this site in vitro (von Hofsten et al., unpublished) indicating that a regulatory connection between Sox9a and ff1a is present in zebrafish. Zebrafish embryos homozygous for jellyfish (jef) (mutations in sox9a) show craniofacial defects and lack of cartilage similar to humans with campomelic dysplasia [35]. The jef strain is still able to reproduce, which leads to the conclusion that Sox9a alone does not direct sex determination and differentiation in zebrafish.
AMH
AMH may not be excluded as a factor involved in the sex determining process in zebrafish. Although fish lack Müllerian ducts, other AMH functions may be important for gonad development. In mammals AMH is, in addition to Müllerian degeneration, involved in regulation of gonadal steroidogenesis. AMH inhibits the expression of aromatase in developing gonads [38]. It also negatively modulates the differentiation and function of Leydig cells [39] by down regulating several enzymes involved in the steroidogenic pathway. Ovarian cell growth is inhibited by AMH in vitro [40]. An AMH related gene, eel spermatogenesis related substances 21 (eSRS21) identified in the Japanese eel is mainly expressed in Sertoli cells and down regulates 11 KT induced spermatogenesis. This indicates that eSRS21 and genes related to AMH have important reproductive functions and are involved in sex determination and differentiation in fish [41]. In zebrafish, we recently cloned an AMH cDNA and observed that it was expressed exclusively in gonads [42]. AMH expression was, by in situ hybridization, found predominantly in Sertoli cells in testis and in the follicular layer in ovaries. Interestingly, AMH is co-expressed both with the Steroidogenic Factor-1 (SF-1) homologue ff1d and Sox9a within these cells [42,43]. AMH displays complex regulation in mammals, involving several factors, including the FTZ-F1 related gene SF-1, GATA4, Sox9 and WT1 [44,45]. The transcriptional regulation of zebrafish AMH has so far not been elucidated. However, the AMH gene promoter sequence contains putative binding sites for the same transcription factors that regulate mammalian AMH, indicating a conserved regulatory mechanism in vertebrates.
WT1
As in mammals, the anlagen for the excretion and reproductive systems both derive from intermediate mesoderm [46-48]. WT1 was originally found to be a suppressor of Wilms tumour, as individuals with inactivated WT1 developed the Wilms tumour condition [49]. WT1 is also a crucial factor in the differentiation of renal tissue. In zebrafish, WT1 has been shown to be expressed in the intermediate mesoderm prior to and during renal tissue differentiation [29]. It is also essential for the steroidogenic interrenal development together with ff1b [50]. WT1 is thereby an important factor in the early events during development of gonadal primordium.
FTZ-F1 (NR5A)
The Drosophila homeobox gene fushi tarazu (ftz) was initially identified as a central factor for segmentation, as inhibition of ftz resulted in the development of fewer segments [51,52]. The fushi tarazu factor-1 (FTZ-F1) was later identified as the key regulator of ftz expression [53,54]. Genes homologous to the Drosophila FTZ-F1 have subsequently been identified in several species in different phyla [55-63]. Several different names have been given to these homologues, including steroidogenic factor-1 (SF-1), adrenal-4-binding protein (Ad4BP), embryonal long terminal repeat-binding protein (ELP), α-fetoprotein transcription factor (FTF) and liver receptor homologue-1 (LRH-1). However, lately a nomenclature system presented by the nuclear receptor committee groups the FTZ-F1 homologues under the name NR5A [64].
The mammalian genome contains two FTZ-F1 homologues (NR5A1 and NR5A2). NR5A1 contains the SF-1 related genes, which are closely connected to steroidogenesis. In mammals, the NR5A1 genes are expressed in steroidogenic tissues, are key regulators of steroidogenesis and are involved in the testis determining pathway during sex determination [65-67]. The NR5A2 genes are linked to regulation of the estrogen-binding α-fetoprotein [60]. The mammalian NR5A2 genes are expressed in steroidogenic tissues as well as liver, pancreas and intestine, but appear to be more involved in cholesterol metabolism than steroid synthesis or sex determination.
Zebrafish FTZ-F1
FTZ-F1 homologues have been identified in a number of teleost species [55-57,68-70]. Several teleosts have multiple variants of FTZ-F1 genes. The roles and functions of these genes are not completely elucidated, but all studies conducted so far indicate an involvement in the reproductive axis or in steroidogenesis. The ff1 proteins share the general structure of other nuclear receptors. They contain a DNA binding domain (DBD) with two Zn-fingers and the FTZ-F1 box for DNA interaction and recognition, a hinge domain that connects to the ligand-binding domain (LBD), which contains the I-box for protein-protein interaction and the activator function-2 (AF-2) domain for transcriptional activation (Fig. 1). Zebrafish is the most extensively studied teleost and four FTZ-F1 genes have been identified (ff1a, b, c and d). The arrangement of FTZ-F1 genes into the nuclear receptor 5A subgroups is a suitable system for genes in higher vertebrates, as no indistinguishable genes have been described so far in these animals. However, teleosts and particularly zebrafish are different in more than one way compared to higher vertebrates. Zebrafish have four different FTZ-F1 genes, whereas mammals and higher vertebrates only possess two. The zebrafish genes are not easily arranged within the NR5A subgroups. Zebrafish ff1a and Arctic char ff1 aligns well within the NR5A2 subgroup (Fig. 2), but their expression patterns and suggested functions do not fit the description of the mammalian NR5A2 genes. The zebrafish ff1c does not align well with any of the described subgroups, which further raises the question of how appropriate the subdivisions really are for teleost FTZ-F1. Ff1d and ff1b are similar and aligns together with medaka FTZ-F1 in a subgroup within the NR5A1 clade. This subgroup has previously been named NR5A4, but recent data indicates that the genes in the NR5A4 subgroup are NR5A1 homologous. Recently it was suggested that ff1b and ff1d are of the same origin and arose from ancestral gene duplication [71]. This was supported by the overlapping expression patterns found in embryonic interrenal and pituitary cells [42]. The tissue distribution of ff1b and ff1d is identical, while it differs from ff1a and ff1c (Fig. 3). The combined information of ff1b and ff1d expression patterns, function and sequence similarities to other genes in the NR5A1 group suggest that these genes should be considered as homologues.
Figure 1 General structure of zebrafish FTZ-F1 proteins. The zebrafish FTZ-F1 proteins consist of four main regions, the modulator domain, DNA-binding domain (DBD), hinge region and the ligand-binding domain. The DBD contains a Zink-finger region, an A- and P-box for recognition of the FTZ-F1 response element, and a T-box for stabilising the DNA-binding. The proximal repressive- and interactive domains (PRD and PID) are used for interactions with co-repressors and co-activators. The ligand binding-domain containing the I-box and AF-2 region, which both are involved in ligand binding and transactivation, and a distal repressive domain (DRD) for co-repressor binding.
Figure 2 NR5A sequence similarity analysis displayed in a radial tree. Clades containing subgroups NR5A1, NR5A2, NR5A3 and NR5A4 are indicated. Arctic char FF1a (acFF1a); Mouse LRH-1 (mLRH-1); Rat SF-1 (rSF-1); Mouse ELP (mELP); Rana rugosa FTZ-F1 (rrFTZ-F1); Zebrafish ff1b (zff1b); Zebrafish ff1a (zff1a); Zebrafish ff1c (zff1c); Zebrafish ff1d (zff1d); Rat FTF (rFTF); Medaka FTZ-F1 (mFTZ-F1); Rainbow trout FTZ-F1 (rtFTZ-F1); Chick SF-1 (cSF-1); Chick FTF (cFTF) and Drosophila melanogaster ftz-f1 (dmFTZ-F1). Modified from [42].
Figure 3 Tissue distribution of ff1a, ff1b, ff1c and ff1d in adult zebrafish, detected by RT-PCR. The four ff1 genes show differential expression with the ff1a gene being expressed in most tissues with high expression in liver. The ff1b and ff1d genes are both expressed in gonads and brain with the ff1d showing higher expression in testis than in ovary. The ff1c gene is primarily expressed in the liver. m: male tissue, f: female tissue. Modified from [42].
Ff1a
The first FTZ-F1 gene described in zebrafish was ff1a [55]. The gene was named zff1, as no other zebrafish ff1 genes were known at that time. With the identification of additional ff1 genes it was later renamed ff1a. The gene possesses two splicing variants, now designated ff1a-A and ff1a-B. Ff1a-A was, in synergy with ER, shown to transcriptionally activate a gonadotropin promoter, whereas ff1a-B acted as an inhibitor of ff1a-A due to its lack of the AF-2 trans-activation domain. The expression of zebrafish ff1a was later shown to be driven by two separate gene promoters, giving rise to a total of four separate gene transcripts, ff1a-IA, ff1a-IB, ff1a-IIA and ff1aIIB [72].
The functional difference between ff1a proteins consisting of exon one, transcribed via promoter I, and proteins transcribed from promoter II has so far not been studied. However, they differ slightly in their tissue distribution, where promoter I derived transcripts are lacking in brain and heart [72]. The two ff1a promoters contain different putative response elements [37,72]. This suggests that the two promoters regulate ff1a in tissue specific manners and during different developmental stages rather than rendering them separate functions. Also mouse and rat have two separately regulated NR5A promoters [65,73].
Two of the putative response elements in promoter I indicate an involvement in somitogenesis. MyoD and Snail are both transcription factors shown to be involved in somite development [74,75], indicating that promoter I may drive the ff1a expression during somitogenesis. The ff1a IIA gene product is involved in muscle differentiation during somitogenesis. Microinjection of ff1a AII mRNA into the ubo-mutant strain, which lack slow-twitch muscle cells, results in partially restored myofibers [76]. Promoter II contains an HMG-Box response element 24–31 bp up stream of the transcription start. An identical response element has been shown to bind Sox9a in vitro [34] and Sox9 is hence a putative regulator of gonadal expression of ff1a. Expression of ff1a can be detected in the uro-genital and pronephric duct region during early somitogenesis [77] and by linking the ff1a dual promoter to GFP, early gonadal expression can be detected 5 dpf after microinjection [37]. This indicates a role in early gonad development and differentiation. The Arctic char ff1 homologue has been linked to steroidogenesis by showing a cyclic expression pattern during the reproductive maturation process and 17β-estradiol mediated down-regulation of testicular expression [56]. The phylogenetic relationship indicates that the Arctic char ff1 belongs to the ff1a related genes and should be named ff1a (Fig. 2). Furthermore, the developmental expression pattern of Arctic char ff1a is similar to zebrafish ff1a [78], indicating that teleost ff1a homologues are involved both in steroidogenesis and gonad development.
Ff1b
The ff1b gene was initially assigned to functions related to pancreatic development as it was co-expressed with pancreas duodenum homeobox-1 (pdx-1) and proinsulin [56]. However, more recent publications suggest that ff1b is an important factor for steroidogenic cell development and that ff1b is required for the differentiation of the interrenal organ [79,80]. The expression of ff1b precedes that of cyp11A and 3βHSD in the embryonic interrenal cells and ff1b morpholino knock down experiments abolishes the expression of these two genes [79].
The down stream transcriptional activation function of ff1b is modulated by protein-protein interactions with homeodomain protein Prox1 [80]. Two domains are needed for the interaction, the I-box and the AF-2 domain, both situated in the LBD. Binding to Prox1 leads to a repression of down stream trans-activation. There is also a co-localization of ff1b and Prox1 expression in the developing interrenal. Due to the conserved I-boxes and AF-2 domains, both ff1a and ff1c are probably able to interact with Prox1, although less efficiently than ff1b.
Ff1c
There is little information available regarding ff1c functions, regulation or expression patterns. Except for a weak interaction between Prox1 and ff1c presented in Liu et al. [80], the sequence published on GenBank, (AF327373) is the only published data available so far. Expression of ff1c can be detected in numerous tissues in adult zebrafish and its highest expression is found in liver and intestine, indicating a role in cholesterol metabolism, similar to ff1a (see Fig. 3). No specific ff1c expression domains have been identified during embryogenesis. Both ff1c and ff1d are similar to ff1a and b in their DNA-binding domains where the FTZ-F1 box is situated and in the ligand binding domains, but are less conserved in their hinge regions. All zebrafish ff1 have highly conserved AF-2 domains and I-Boxes in their LBD.
Ff1d
Ff1b and ff1d display an overlapping expression pattern during embryogenesis. They share protein domains important for co-factor interactions and have been suggested to be the result of an ancestral gene duplication [42,71]. Even though the two zebrafish NR5A1 genes are similar in several aspects, the shared sequence identities are 62%, which suggests that functional differences are likely to exist. Expression of ff1d in adult zebrafish is restricted to brain, gonads and liver [42]. There also seem to be sexual differences, as ff1d expression is higher in testis than in ovary. In the testis ff1d is highly expressed in interstitial Leydig cells and Sertoli cells, but cannot be detected in germ cells. In ovary ff1d is located to the follicular layer and inside the oocyte [42]. Although the function and regulatory mechanisms of ff1d in these cells needs to be further studied, a possible target of ff1d is AMH, which is co-expressed with ff1d in Sertoli cells and in the follicular layer. During mammalian sex determination and differentiation, the ff1d homolog SF-1 regulates the expression of AMH, leading to the development of male sex characteristics. Therefore it is intriguing that this may be a conserved vertebrate developmental mechanism.
Dmrt1
The lack of testis-determining factor similar to SRY in fish is not a unique phenomenon. This is also the case for many other lower vertebrates. Genes containing a DM-domain (Dmrt1) have however been identified in fish. DM-domain containing genes are involved in sex determination in both vertebrates and invertebrates [81], which is a unique conservation of function between phyla, not seen in any other gene involved in sex determination. Different Dmrt1 homologues have been shown to be involved in gonad development [82,83] and somitogenesis [84].
The teleost Japanese medaka has specific sex chromosomes (XX/XY), where the DM-domain gene DMY has been mapped to the Y-chromosome and has been shown to be essential for testis differentiation [85,86]. This was determined by isolating the sex-determining region on the male specific Y-chromosome. Except for the DMY-containing region, the X and Y-chromosomes are very similar. This indicates that the medaka Y-chromosome and DMY are, in an evolutionary prospective, relatively new. This theory was later confirmed and DMY was discarded as a universal teleost sex-determining gene, as it was shown to be a species-specific sex-determining strategy [87]. No target genes for DMY have been identified and DMY function in testis development remains unresolved. Treating birds with the aromatase inhibitor fadrozole lead to elevated Dmrt1 levels indicating that Dmrt1 may be down regulated by aromatase [88]. This indicates that Dmrt1 may have an important role in testis determination in teleosts, since alteration of aromatase levels during sex differentiation can cause sex reversals. The regulation of Dmrt1 related genes in teleosts remains unknown, but testicular expression of Dmrt1 is in mammals regulated by GATA4 [89].
GATA
GATA factors are divided into two families based on expression patterns, structure, and function [90]. GATA-1/2/3 is most commonly associated with haematopoietic cell and neuronal development [91,92]. GATA-4/5/6 are usually linked to organ development, including the urogenital system [93,94]. GATA factors recognize and bind the DNA consensus motif, WGATAR, and closely related sequences. GATA-4 plays an important role as transcriptional regulator of SRY and AMH during mammalian sex determination and differentiation [95]. Studies of zebrafish GATA factors have so far been associated with the development of organ systems other than the urogenital, but binding sites for GATA4 have been found in the cyp19 gene promoter [96,97] suggesting a role in regulating aromatase expression.
Aromatase
Steroidogenesis, sex determination and differentiation are closely related to each other. SF-1 is one of the crucial factors essential for steroid biosynthesis as well as sex determination and differentiation in mammals. The terminal sex-hormone products in the steroid biosynthesis pathway are androgens and estrogens, and the balance between them leads to the development of proper sex characteristics. Aromatase (Cyp 19) is the product of the cyp19 gene, and is an important regulator of this balance. Aromatase is produced in the gonads and directs the conversion of testosterone into 17β-estradiol.
Like many other fish species, the zebrafish genome contain two aromatase genes designated cyp19a and cyp19b [98]. Cyp19a is highly expressed in the steroidogenic Theca and granulosa cell layer surrounding the oocytes in the ovary, whereas cyp19b is mainly expressed in brain. Thus, while one aromatase gene appears to be involved in gonadal development the other gene may be involved in neuronal development. However, both genes contribute to the regulation of estrogenic responses and may thus influence sex differentiation. The regulation of teleost cyp19 transcription is not completely elucidated, but the zebrafish cyp19a promoter region contains binding sites for Ftz-F1, which suggests a role for ff1 genes in the regulation of cyp19a expression in gonadal tissue [97,99]. Ftz-F1 dependent cyp19 transcription has also been documented in species from turtles to humans [100-102], indicating that this mechanism is conserved in all vertebrates. In many vertebrates, reptiles in particular, the level or activity of aromatase is the deciding factor for sex during development [103,104]. The temperature surrounding the developing embryos influences the activity of aromatase leading to variations in sex ratio [105]. A similar scenario has been documented for several fish species, including zebrafish, suggesting that aromatase is an important factor in sex determination and differentiation in fish. By using an aromatase inhibitor, or by increasing water temperature to 35°C–37°C, oocyte apoptosis can be induced in zebrafish [106]. However, during normal breeding conditions the temperature is not an important factor deciding sex ratios in zebrafish. The role of aromatase remains important in zebrafish sex determination as exposure to the aromatase inhibitor fadrozole results in sex-reversion of female zebrafish [106].
Sex determination and differentiation pathway
Studies of mammals have shown that genes involved in sex determination have multiple functions but that the presence of the SRY gene in animals with XY/XX chromosomal systems leads to male development. As discussed earlier in this review many of the genes involved in mammalian sex determination show dosage dependent effects on sex determination and differentiation. A generalised genetic regulatory pathway can be extracted from the studies conducted on different vertebrate species (Fig. 4). Most identified genes have a primary function in male development. The regulatory pathway includes, and is based on, the presence of SRY and Dax-1, which in an antagonistic manner direct the development of male and female phenotypes in mammals. Lack of the SRY gene, as is the case in teleosts, suggest that when chromosomal sex determination exist it must be regulated by alternative genes. In medaka the Dmrt1 gene was found to be the switch. However, this role for Dmrt1 is specific for medaka and it remains that any of the other known genes in the sex determination cascade could potentially develop into the genetic switch in other teleost species. Or, there may be yet unidentified genes and pathways that participate in teleost sex determination.
Figure 4 Involvement of a hierarchy of genes in mammalian sex determination and differentiation. In XY/XX systems where SRY is the key regulator of sex determination its absence leads to activation of Dax1 and female development. The presence of SRY results in a hierarchy of activation of genes leading to the development of testis. In this hierarchy SF-1 (FTZ-F1) is a key regulator of steroidogenesis and AMH, demonstrating its central role in sex determination and differentiation. ⊥: Inhibition, ↓: stimulation.
Proposed model
From the information obtained to date it is not possible to define a hierarchy of regulation during sex determination in teleosts. Therefore, the proposed model for zebrafish sex determination is based on the same group of male determining genes, excluding SRY and Dax-1, found in mammals while the interplay between the genes remains undefined (Fig. 5). Theoretically, any one of the described genes may become the sex determining gene, as several of them have been shown to cause sex reversals in mammalian model systems in a dosage dependent way (see table 1). The male determining switch may also be dependent on combinatory effects of allelic variants among the genes involved. Furthermore, the regulation of aromatase appears to be crucial for zebrafish sex determination.
Figure 5 A generalized model of the involvement of different genes in zebrafish sex determination and differentiation. While little is known of the hierarchy of genes involved in zebrafish sex determination and differentiation several genes have been identified. While aromatase has been shown to play a central role in zebrafish sex differentiation the environmental and/or genetic mechanisms have not been fully elucidated. ⊥: Inhibition, ↓: stimulation.
The model suggests that ff1a and WT1 are important for the differentiation of the uro-genital tissue, which subsequently develops into renal and gonadal tissue. WT1 is essential for the differentiation of pronephros, and a battery of genes, including the FTZ-F1 genes, Sox9a, GATA4, Dmrt1 and AMH, are involved in the differentiation of gonads. During the critical time period around 25 dpf this battery of genes may direct the development towards male gonads in individuals with the allelic combinations predestined to become male. This would lead to a decrease or absence of aromatase and subsequent reduced estrogen levels and activity, resulting in the onset of ovarian apoptosis, the differentiation of testicular Sertoli cells and increased testosterone levels. The model is based on the observation that adult female zebrafish can be sex-reversed by inhibiting aromatase [106]. This suggests that zebrafish has a high plasticity in their mechanism of gender development and that steroidogenesis plays an essential part in the sex determining process. The complex mechanism of sex determination and differentiation is still far from elucidated and the biochemistry behind it must be further studied to establish protein interactions controlling it. The FTZ-F1 genes are important, as they are involved in the early development of uro-genital tissue and as regulators of steroidogenic cells and their gene expression.
Acknowledgements
The present study was financed by generous grants from the Swedish Research Council, the Magnus Bergwall foundation and the Kempe memorial foundation.
==== Refs
Thorgaard GH Heteromorphic sex chromosomes in male rainbow trout Science 1977 196 900 902 860122
Thorgaard GH Sex chromosomes in the sockeye salmon: a Y-autosome fusion Can J Genet Cytol 1978 20 349 354 570442
Phillips RB Ihssen PE Identification of sex chromosomes in lake trout (Salvelinus namaycush) Cytogenet Cell Genet 1985 39 14 18 3979115
Volff JN Schartl M Variability of genetic sex determination in poeciliid fishes Genetica 2001 111 101 110 11841158 10.1023/A:1013795415808
Kallman KD Evidence for the existence of transformer genes for sex in the teleost Xiphophorus maculatus Genetics 1968 60 811 828 5732420
Bull JJ Vogt RC Temperature-dependent sex determination in turtles Science 1979 206 1186 1188 505003
Conover DO Heins SW Adaptive variation in environmental and genetic sex determination in a fish Nature 1987 326 496 498 3561487 10.1038/326496a0
Pavlidis M Koumoundouros G Sterioti A Somarakis S Divanach P Kentouri M Evidence of temperature-dependent sex determination in the European sea bass (Dicentrarchus labrax L.) J Exp Zool 2000 287 225 232 10900442 10.1002/1097-010X(20000801)287:3<225::AID-JEZ4>3.0.CO;2-D
Tang F Chan ST Lofts B Effect of mammalian luteinizing hormone on the natural sex reversal of the rice-field eel, Monopterus albus (Zuiew) Gen Comp Endocrinol 1974 24 242 248 4473397 10.1016/0016-6480(74)90178-6
Yeung WS Chen H Chan ST In vivo effects of oLH and LHRH-analog on sex reversal and plasma sex steroid profiles in the female Monopterus albus Gen Comp Endocrinol 1993 90 23 30 8504919 10.1006/gcen.1993.1056
Baroiller JF Guiguen Y Fostier A Endocrine and environmental aspects of sex differentiation in fish Cell Mol Life Sci 1999 55 910 931 10.1007/s000180050344
Berta P Hawkins JR Sinclair AH Taylor A Griffiths BL Goodfellow PN Fellous M Genetic evidence equating SRY and the testis-determining factor Nature 1990 348 448 450 2247149 10.1038/348448A0
Swain A Narvaez V Burgoyne P Camerino G Lovell-Badge R Dax1 antagonizes Sry action in mammalian sex determination Nature 1998 391 761 767 9486644 10.1038/35799
Wagner T Wirth J Meyer J Zabel B Held M Zimmer J Pasantes J Bricarelli FD Keutel J Hustert E Wolf U Tommerup N Schempp W Scherer G Autosomal sex reversal and campomelic dysplasia are caused by mutations in and around the SRY-related gene SOX9 Cell 1994 79 1111 1120 8001137 10.1016/0092-8674(94)90041-8
Foster JW Dominguez-Steglich MA Guioli S Kowk G Weller PA Stevanovic M Weissenbach J Mansour S Young ID Goodfellow PN Brook JD Schafer AJ Campomelic dysplasia and autosomal sex reversal caused by mutations in an SRY-related gene Nature 1994 372 525 530 7990924 10.1038/372525a0
Huang B Wang S Ning Y Lamb AN Bartley J Autosomal XX sex reversal caused by duplication of SOX9 Am J Med Genet 1999 87 349 353 10588843 10.1002/(SICI)1096-8628(19991203)87:4<349::AID-AJMG13>3.0.CO;2-N
Achermann JC Ito M Hindmarsh PC Jameson JL A mutation in the gene encoding steroidogenic factor-1 causes XY sex-reversal and adrenal failure in humans Nat Genet 1999 22 125 126 10369247 10.1038/9629
Hammes A Guo JK Lutsch G Leheste JR Landrock D Ziegler U Gubler MC Schedl A Two splice variants of the Wilms' tumor 1 gene have distinct functions during sex determination and nephron formation Cell 2001 106 319 329 11509181 10.1016/S0092-8674(01)00453-6
Raymond CS Parker ED Kettlewell JR Brown LG Page DC Kusz K Jaruzelska J Reinberg Y Flejter WL Bardwell VJ Hirsch B Zarkower D A region of human chromosome 9p required for testis development contains two genes related to known sexual regulators Hum Mol Genet 1999 8 989 996 10332030 10.1093/hmg/8.6.989
Blum V Comparative anatomy of the urogenital sysytem Vertebrate Reproduction 1986 Springer Verlag, Berlin 53 63
Francis RC Sexual lability in teleosts: developmental factors Quart Rev Biol 1992 67 1 18 10.1086/417445
Kobayashi T Kajiura-Kobayashi H Nagahama Y Induction of XY sex reversal by estrogen involves altered gene expression in a teleost, tilapia Cytogenet Genome Res 2003 101 289 294 14684997 10.1159/000074351
Traut W Winking H Meiotic chromosomes and stages of sex chromosome evolution in fish: zebrafish, platyfish and guppy Chromosome Res 2001 9 659 672 11778689 10.1023/A:1012956324417
Wallace BM Wallace H Synaptonemal complex karyotype of zebrafish Heredity 2003 90 136 140 12634819 10.1038/sj.hdy.6800184
Amores A Force A Yan YL Joly L Amemiya C Fritz A Ho RK Langeland J Prince V Wang YL Westerfield M Ekker M Postlethwait JH Zebrafish hox clusters and vertebrate genome evolution Science 1998 282 1711 1714 9831563 10.1126/science.282.5394.1711
Uchida D Yamashita M Kitano T Iguchi T Oocyte apoptosis during the transition from ovary-like tissue to testes during sex differentiation of juvenile zebrafish J Exp Biol 2002 205 711 718 11914381
Devlin RH Nagahama Y Sex determination and sex differentiation in fish: an overview of genetic, physiological, and environmental influences Aquaculture 2002 208 191 364 10.1016/S0044-8486(02)00057-1
Yoon C Kawakami K Hopkins N Zebrafish vasa homologue RNA is localized to the cleavage planes of 2- and 4-cell-stage embryos and is expressed in the primordial germ cells Development 1997 124 3157 3166 9272956
Drummond IA Majumdar A Hentschel H Elger M Solnica-Krezel L Schier AF Neuhauss SC Stemple DL Zwartkruis F Rangini Z Driever W Fishman MC Early development of the zebrafish pronephros and analysis of mutations affecting pronephric function Development 1998 125 4655 4667 9806915
Wang DS Kobayashi T Senthilkumaran B Sakai F Sudhakumari CC Suzuki T Yoshikuni M Matsuda M Morohashi K Nagahama Y Molecular cloning of DAX1 and SHP cDNAs and their expression patterns in the Nile tilapia, Oreochromis niloticus Biochem Biophys Res Commun 2002 297 632 640 12270141 10.1016/S0006-291X(02)02252-0
Takamatsu N Kanda H Ito M Yamashita A Yamashita S Shiba T Rainbow trout SOX9: cDNA cloning, gene structure and expression Gene 1997 202 167 170 9427561 10.1016/S0378-1119(97)00483-6
Kanda H Kojima M Miyamoto N Ito M Takamatsu N Yamashita S Shiba T Rainbow trout Sox24, a novel member of the Sox family, is a transcriptional regulator during oogenesis Gene 1998 211 251 257 9602142 10.1016/S0378-1119(98)00100-0
Wang R Cheng H Xia L Guo Y Huang X Zhou R Molecular cloning and expression of Sox17 in gonads during sex reversal in the rice field eel, a teleost fish with a characteristic of natural sex transformation Biochem Biophys Res Commun 2003 303 452 457 12659838 10.1016/S0006-291X(03)00361-9
Chiang EF Pai CI Wyatt M Yan YL Postlethwait J Chung B Two sox9 genes on duplicated zebrafish chromosomes: expression of similar transcription activators in distinct sites Dev Biol 2001 231 149 163 11180959 10.1006/dbio.2000.0129
Yan YL Miller CT Nissen RM Singer A Liu D Kirn A Draper B Willoughby J Morcos PA Amsterdam A Chung BC Westerfield M Haffter P Hopkins N Kimmel C Postlethwait JH Nissen R A zebrafish sox9 gene required for cartilage morphogenesis Development 2002 129 5065 5079 Erratum in: Development 129:5551 12397114
Li M Zhao C Wang Y Zhao Z Meng A Zebrafish sox9b is an early neural crest marker Dev Genes Evol 2002 212 203 206 12012235 10.1007/s00427-002-0235-2
von Hofsten J Modig C Larsson A Karlsson J Olsson PE Determination of the expression pattern of the dual promoter of zebrafish fushi tarazu factor-1a following microinjections into zebrafish one cell stage embryos Gen Comp Endocrinol 2005 142 222 226 15862566 10.1016/j.ygcen.2004.12.020
di Clemente N Ghaffari S Pepinsky RB Pieau C Josso N Cate RL Vigier B A quantitative and interspecific test for biological activity of anti-mullerian hormone: the fetal ovary aromatase assay Development 1992 114 721 727 1319894
Racine C Rey R Forest MG Louis F Ferre A Huhtaniemi I Josso N di Clemente N Receptors for anti-mullerian hormone on Leydig cells are responsible for its effects on steroidogenesis and cell differentiation Proc Natl Acad Sci U S A 1998 95 594 599 9435237 10.1073/pnas.95.2.594
Ha TU Segev DL Barbie D Masiakos PT Tran TT Dombkowski D Glander M Clarke TR Lorenzo HK Donahoe PK Maheswaran S Mullerian inhibiting substance inhibits ovarian cell growth through an Rb-independent mechanism J Biol Chem 2000 275 37101 109 10958795 10.1074/jbc.M005701200
Miura T Miura C Konda Y Yamauchi K Spermatogenesis-preventing substance in Japanese eel Development 2002 129 2689 2697 12015296
von Hofsten J Larsson A Olsson PE Novel steroidogenic factor-1 homolog (ff1d) is coexpressed with anti-Mullerian hormone (AMH) in zebrafish Dev Dyn 2005 233 595 604 15768398 10.1002/dvdy.20335
Rodriguez-Mari A Yan YL Bremiller RA Wilson C Canestro C Postlethwait JH Characterization and expression pattern of zebrafish Anti-Mullerian hormone (Amh) relative to sox9a, sox9b, and cyp19a1a, during gonad development Gene Expr Patterns 2005 5 655 667 15939378 10.1016/j.modgep.2005.02.008
de Santa Barbara P Bonneaud N Boizet B Desclozeaux M Moniot B Sudbeck P Scherer G Poulat F Berta P Direct interaction of SRY-related protein SOX9 and steroidogenic factor 1 regulates transcription of the human anti-Mullerian hormone gene Mol Cell Biol 1998 18 6653 6665 9774680
Rey R Lukas-Croisier C Lasala C Bedecarras P AMH/MIS: what we know already about the gene, the protein and its regulation Mol Cell Endocrinol 2003 211 21 31 14656472 10.1016/j.mce.2003.09.007
Kimmel CB Ballard WW Kimmel SR Ullmann B Schilling TF Stages of embryonic development of the zebrafish Dev Dyn 1995 203 253 310 8589427
Serluca FC Fishman MC Pre-pattern in the pronephric kidney field of zebrafish Development 2001 128 2233 2241 11493543
Molyneaux K Wylie C Primordial germ cell migration Int J Dev Biol 2004 48 537 544 15349828 10.1387/ijdb.041833km
Haber DA Buckler AJ Glaser T Call KM Pelletier J Sohn RL Douglass EC Housman DE An internal deletion within an 11p13 zinc finger gene contributes to the development of Wilms' tumor Cell 1990 61 1257 1269 2163761 10.1016/0092-8674(90)90690-G
Hsu HJ Lin G Chung B Parallel early development of zebrafish interrenal glands and pronephros:differential control by wt1and ff1b Development 2003 130 2107 2116 12668625 10.1242/dev.00427
Kuroiwa A Hafen E Gehring WJ Cloning and transcriptional analysis of the segmentation gene fushi tarazu of Drosophila Cell 1984 37 825 31 6430567 10.1016/0092-8674(84)90417-3
Wakimoto BT Turner FR Kaufman TC Defects in embryogenesis in mutants associated with the antennapedia gene complex of Drosophila melanogaster Dev Biol 1984 102 147 72 6421639 10.1016/0012-1606(84)90182-9
Ueda H Sonoda S Brown JL Scott MP Wu C A sequence-specific DNA-binding protein that activates fushi tarazu segmentation gene expression Genes Dev 1990 4 624 635 2113881
Lavorgna G Ueda H Clos J Wu C FTZ-F1, a steroid hormone receptor-like protein implicated in the activation of fushi tarazu Science 1991 252 848 851 1709303
Liu D Le Drean Y Ekker M Xiong F Hew CL Teleost FTZ-F1 homolog and its splicing variant determine the expression of the salmon gonadotropin IIβ subunit gene Mol Endocrinol 1997 11 877 890 9178748 10.1210/me.11.7.877
Chai C Chan WK Developmental expression of a novel Ftz-F1 homologue, ff1b (NR5A4), in the zebrafish Danio rerio Mech Dev 2000 91 421 426 10704877 10.1016/S0925-4773(99)00312-3
von Hofsten J Karlsson J Jones I Olsson PE Expression and regulation of fushi tarazu factor-1 and steroidogenic genes during reproduction in Arctic char (Salvelinus alpinus) Biol Reprod 2002 67 1297 1304 12297548 10.1095/biolreprod67.4.1297
Lala DS Rice DA Parker KL Steroidogenic factor I, a key regulator of steroidogenic enzyme expression, is the mouse homolog of fushi tarazu-factor I Mol Endocrinol 1992 6 1249 1258 1406703 10.1210/me.6.8.1249
Sun GC Hirose S Ueda H Intermittent expression of BmFTZ-F1, a member of the nuclear hormone receptor superfamily during development of the silkworm Bombyx mori Dev Biol 1994 162 426 437 8150206 10.1006/dbio.1994.1099
Galarneau L Paré JF Allard D Hamel D Levesque L Tugwood JD Green S Belanger L The alpha1-fetoprotein is activated by a nuclear receptor of the Drosophila FTZ-F1 family Mol Cell Biol 1996 16 3853 3865 8668203
Wong M Ramayya MS Chrousos GP Driggers PH Parker KL Cloning and sequence analysis of the human gene encoding steroidogenic factor 1 J Mol Endocrinol 1996 17 139 147 8938589 10.1677/jme.0.0170139
Kudo T Sutou S Molecular cloning of chicken FTZ-F1-related orphan receptors Gene 1997 197 261 268 9332374 10.1016/S0378-1119(97)00270-9
Kawano K Miura I Morohashi K Takase M Nakamura M Molecular cloning and expression of the SF-1/Ad4BP gene in the frog, Rana rugosa Gene 1998 2 169 76 9831646 10.1016/S0378-1119(98)00498-3
Nuclear receptors committee A unified nomenclature system for the nuclear receptor superfamily Cell 1999 97 161 163 10219237 10.1016/S0092-8674(00)80726-6
Ikeda Y Lala DS Luo X Kim E Moisan MP Parker KL Characterization of the mouse FTZ-F1 gene, which encodes a key regulator of steroid hydroxylase gene expression Mol Endocrinol 1993 7 852 860 8413309 10.1210/me.7.7.852
Ikeda Y Shen WH Ingraham HA Parker KL Developmental expression of mouse steroidogenic factor-1, an essential regulator of the steroid hydroxylases Mol Endocrinol 1994 8 654 662 8058073 10.1210/me.8.5.654
Sadovsky Y Crawford PA Woodson KG Polish JA Clements MA Tourtellotte LM Simburger K Milbrandt J Mice deficient in the orphan receptor steroidgenic factor 1 lack adrenal glands and gonads but express P450 side-chain-cleavage enzyme in the placenta and have normal embryonic serum levels of corticosteroids Proc Natl Acad Sci 1995 92 10939 10943 7479914
Ito M Masuda A Yumoto K Otomo A Takahashi Y Takamatsu N Kanda H Yamashita S Shiba T cDNA cloning of a new member of the FTZ-F1 subfamily from a rainbow trout Biochim Biophys Acta 1998 1395 271 274 9512659
Watanabe M Tanaka M Kobayashi D Yoshiura Y Oba Y Nagahama Y Medaka (Oryzias latipes) FTZ-F1 potentially regulates the transcription of P-450 aromatase in ovarian follicles: cDNA cloning and functional characterization Mol Cell Endocrinol 1999 149 221 228 10375033 10.1016/S0303-7207(99)00006-4
Higa M Kanda H Kitahashi T Ito M Shiba T Ando H Quantitative analysis of fushi tarazu factor 1 homolog messenger ribonucleic acids in the pituitary of salmon at different prespawning stages Biol Reprod 2000 63 1756 1763 11090446 10.1095/biolreprod63.6.1756
Kuo MW Postlethwait J Lee WC Lou SW Chan WK Chung BC Gene duplication, gene loss and evolution of expression domains in the vertebrate nuclear receptor NR5A (Ftz-F1) family Biochem J 2005 389 19 26 15725073 10.1042/BJ20050005
Lin WW Wang HH Sum C Liu D Hew C Chung B Zebrafish ftz-f1 gene has two promoters, is alternatively spliced, and is expressed in digestive organs Biochem J 2000 348 439 446 10816440 10.1042/0264-6021:3480439
Nomura M Bartsch S Nawata H Omura T Morohashi K An E box element is required for the expression of the ad4bp gene, a mammalian homologue of ftz-f1 gene, which is essential for adrenal and gonadal development J Biol Chem 1995 270 7453 7461 7706291 10.1074/jbc.270.13.7453
Weinberg ES Allende ML Kelly CS Abdelhamid A Murakami T Andermann P Doerre OG Grunwald DJ Riggleman B Developmental regulation of zebrafish MyoD in wild-type, no tail and spadetail embryos Development 1996 122 271 280 8565839
Thisse C Thisse B Schilling TF Postlethwait JH Structure of the zebrafish snail1 gene and its expression in wild-type, spadetail and no tail mutant embryos Development 1993 119 1203 1215 8306883
Sheela SG Lee WC Lin WW Chung BC Zebrafish ftz-f1a (nuclear receptor 5a2) functions in skeletal muscle organization Dev Biol 2005 Sep 12 16162335
von Hofsten J Jones I Karsson J Olsson PE Developmental expression patterns of FTZ-F1 homologues in zebrafish (Danio rerio) Gen Comp Endocrinol 2001 121 146 155 11178880 10.1006/gcen.2000.7582
von Hofsten J Karlsson J Olsson PE Fushi tarazu factor-1 mRNA and protein is expressed in steroidogenic and cholesterol metabolising tissues during different life stages in Arctic char (Salvelinus alpinus) Gen Comp Endocrinol 2003 132 96 102 12765648 10.1016/S0016-6480(03)00034-0
Chai C Liu YW Chan WK Ff1b is required for the development of steroidogenic component of the zebrafish interrenal organ Dev Biol 2003 260 226 244 12885566 10.1016/S0012-1606(03)00219-7
Liu Y-W Gao W Teh H-L Tan J-H Chan WK Prox1 Is a Novel Coregulator of Ff1b and Is Involved in the Embryonic Development of the Zebra Fish Interrenal Primordium Mol Cell Biol 2003 23 7243 7255 14517294 10.1128/MCB.23.20.7243-7255.2003
Raymond CS Shamu CE Shen MM Seifert KJ Hirsch B Hodgkin J Zarkower D Evidence for evolutionary conservation of sex-determining genes Nature 1998 391 691 695 9490411 10.1038/35618
Guan G Kobayashi T Nagahama Y Sexually dimorphic expression of two types of DM (Doublesex/Mab-3)-domain genes in a teleost fish, the Tilapia (Oreochromis niloticus) Biochem Biophys Res Commun 2000 272 662 666 10860811 10.1006/bbrc.2000.2840
Marchand O Govoroun M D'Cotta H McMeel O Lareyre J Bernot A Laudet V Guiguen Y DMRT1 expression during gonadal differentiation and spermatogenesis in the rainbow trout, Oncorhynchus mykiss Biochim Biophys Acta 2000 1493 180 187 10978520
Meng A Moore B Tang H Yuan B Lin S A drosophila doublesex-related gene, terra, is involved in somitogenesis in vertebrates Development 1999 126 1259 1268 10021344
Matsuda M Nagahama Y Shinomiya A Sato T Matsuda C Kobayashi T Morrey CE Shibata N Asakawa S Shimizu N Hori H Hamaguchi S Sakaizumi M DMY is a Y-specific DM-domain gene required for male development in the medaka fish Nature 2002 417 559 563 12037570 10.1038/nature751
Nanda I Kondo M Hornung U Asakawa S Winkler C Shimizu A Shan Z Haaf T Shimizu N Shima A Schmid M Schartl M A duplicated copy of DMRT1 in the sex-determining region of the Y chromosome of the medaka, Oryzias latipes Proc Natl Acad Sci U S A 2002 99 11778 11783 12193652 10.1073/pnas.182314699
Volff JN Kondo M Schartl M Medaka dmY/dmrt1Y is not the universal primary sex-determining gene in fish Trends Genet 2003 19 196 199 12683972 10.1016/S0168-9525(03)00051-9
Smith CA Katz M Sinclair AH DMRT1 is upregulated in the gonads during female-to-male sex reversal in ZW chicken embryos Biol Reprod 2003 68 560 570 12533420 10.1095/biolreprod.102.007294
Lei N Heckert LL Gata4 regulates testis expression of Dmrt1 Mol Cell Biol 2004 24 377 388 14673170 10.1128/MCB.24.1.377-388.2004
Lavoie H Debeane F Trinh QD Turcotte JF Corbeil-Girard LP Dicaire MJ Saint-Denis A Page M Rouleau GA Brais B Polymorphism, shared functions and convergent evolution of genes with sequences coding for polyalanine domains Hum Mol Genet 2003 12 2967 2979 14519685 10.1093/hmg/ddg329
Ohneda K Shimizu R Nishimura S Muraosa Y Takahashi S Engel JD Yamamoto M A minigene containing four discrete cis elements recapitulates GATA-1 gene expression in vivo Genes Cells 2002 7 1243 1254 12485164 10.1046/j.1365-2443.2002.00595.x
Patient RK McGhee JD The GATA family (vertebrates and invertebrates) Curr Opin Genet Dev 2002 12 416 422 12100886 10.1016/S0959-437X(02)00319-2
Morrisey EE Ip HS Lu MM Parmacek MS GATA-6: a zinc finger transcription factor that is expressed in multiple cell lineages derived from lateral mesoderm Dev Biol 1996 177 309 322 8660897 10.1006/dbio.1996.0165
Molkentin JD Tymitz KM Richardson JA Olson EN Abnormalities of the genitourinary tract in female mice lacking GATA5 Mol Cell Biol 2000 20 5256 5260 10866681 10.1128/MCB.20.14.5256-5260.2000
Tevosian SG Albrecht KH Crispino JD Fujiwara Y Eicher EM Orkin SH Gonadal differentiation, sex determination and normal Sry expression in mice require direct interaction between transcription partners GATA4 and FOG2 Development 2002 129 4627 4634 12223418
Tchoudakova A Kishida M Wood E Callard GV Promoter characteristics of two cyp19 genes differentially expressed in the brain and ovary of teleost fish J Steroid Biochem Mol Biol 2001 78 427 439 11738553 10.1016/S0960-0760(01)00120-0
Tong SK Chung BC Analysis of zebrafish cyp19 promoters J Steroid Biochem Mol Biol 2003 86 381 386 14623534 10.1016/S0960-0760(03)00347-9
Chiang EF Yan YL Guiguen Y Postlethwait J Chung BC Two Cyp19 (P450 aromatase) genes on duplicated zebrafish chromosomes are expressed in ovary or brain Mol Biol Evol 2001 18 542 550 11264405
Watanabe M Tanaka M Kobayashi D Yoshiura Y Oba Y Nagahama Y Medaka (Oryzias lapites) FTZ-F1 potentially regulates the transcription of P-450 aromatase in ovarian follicles: cDNA cloning and functional characterization Mol Cell Endocrinol 1999 149 221 228 10375033 10.1016/S0303-7207(99)00006-4
Crews D Fleming A Willingham E Baldwin R Skipper JK Role of steroidogenic factor 1 and aromatase in temperature-dependent sex determination in the red-eared slider turtle J Exp Zool 2001 290 597 606 11748608 10.1002/jez.1110
Lynch JP Lala DS Peluso JJ Luo W Parker KL White BA Steroidogenic factor 1, an orphan nuclear receptor, regulates the expression of the rat aromatase gene in gonadal tissues Mol Endocrinol 1993 7 776 786 8395654 10.1210/me.7.6.776
Gurates B Amsterdam A Tamura M Yang S Zhou J Fang Z Amin S Sebastian S Bulun SE WT1 and DAX-1 regulate SF-1-mediated human P450arom gene expression in gonadal cells Mol Cell Endocrinol 2003 208 61 75 14580722 10.1016/S0303-7207(03)00198-9
Jeyasuria P Place AR Embryonic brain-gonadal axis in temperature-dependent sex determination of reptiles: a role for P450 aromatase (CYP19) J Exp Zool 1998 281 428 449 9662830 10.1002/(SICI)1097-010X(19980801)281:5<428::AID-JEZ8>3.0.CO;2-Q
Wibbels T Cowan J LeBoeuf R Temperature-dependent sex determination in the red-eared slider turtle, Trachemys scripta J Exp Zool 1998 281 409 416 9662828 10.1002/(SICI)1097-010X(19980801)281:5<409::AID-JEZ6>3.0.CO;2-S
Bogart MH Sex determination: a hypothesis based on steroid ratios J Theor Biol 1987 128 349 357 3444342
Uchida D Yamashita M Kitano T Iguchi T An aromatase inhibitor or high water temperature induce oocyte apoptosis and depletion of P450 aromatase activity in the gonads of genetic female zebrafish during sex-reversal Comp Biochem Physiol A Mol Integr Physiol 2004 137 11 20 14720586 10.1016/S1095-6433(03)00178-8
|
16281973
|
PMC1298332
|
CC BY
|
2021-01-04 16:37:13
|
no
|
Reprod Biol Endocrinol. 2005 Nov 10; 3:63
|
utf-8
|
Reprod Biol Endocrinol
| 2,005 |
10.1186/1477-7827-3-63
|
oa_comm
|
==== Front
Reprod Biol EndocrinolReproductive biology and endocrinology : RB&E1477-7827BioMed Central London 1477-7827-3-641628197410.1186/1477-7827-3-64ResearchBiochemical characterization of the Arctic char (Salvelinus alpinus) ovarian progestin membrane receptor Berg A Håkan [email protected] Peter [email protected] Per-Erik [email protected] The University of Texas at Austin Marine Science Institute, Port Aransas, Texas, USA2 Örebro Life Science Center, Department of Natural Sciences, Örebro University, Sweden2005 10 11 2005 3 64 64 5 10 2005 10 11 2005 Copyright © 2005 Berg et al; licensee BioMed Central Ltd.2005Berg 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.
Membrane progestin receptors are involved in oocyte maturation in teleosts. However, the maturation-inducing steroid (MIS) does not appear to be conserved among species and several progestins may fulfill this function. So far, complete biochemical characterization has only been performed on a few species. In the present study we have characterized the membrane progestin receptor in Arctic char (Salvelinus alpinus) and show that the 17,20beta-dihydroxy-4-pregnen-3-one (17,20beta-P) receptor also binds several xenobiotics, thus rendering oocyte maturation sensitive to environmental pollutants. We identified a single class of high affinity (Kd, 13.8 ± 1.1 nM), low capacity (Bmax, 1.6 ± 0.6 pmol/g ovary) binding sites by saturation and Scatchard analyses. Receptor binding displayed rapid association and dissociation kinetics typical of steroid membrane receptors, with t1/2 s of less than 1 minute. The 17,20beta-P binding also displayed tissue specificity with high, saturable, and specific 17,20beta-P binding detected in ovaries, heart and gills while no specific binding was observed in muscle, brain or liver. Changes in 17,20beta-P binding during oocyte maturation were consistent with its identity as the oocyte MIS membrane receptor. Incubation of fully-grown ovarian follicles with gonadotropin induced oocyte maturation, which was accompanied by a five-fold increase in 17,20beta-P receptor binding. In addition, competition studies with a variety of steroids revealed that receptor binding is highly specific for 17,20beta-P, the likely maturation-inducing steroid (MIS) in Arctic char. The relative-binding affinities of all the other progestogens and steroids tested were less than 5% of that of 17,20beta-P for the receptor. Several ortho, para derivatives of DDT also showed weak binding affinity for the 17,20beta-P receptor supporting the hypothesis that xenobiotics may bind steroid receptors on the oocyte's surface and might thereby interfere with oocyte growth and maturation.
==== Body
Introduction
Meiosis is arrested at prophase 1 in vertebrate oocytes during their growth phase and a surge in gonadotropin secretion is required to induce the resumption of meiosis and oocyte maturation (OM). It has been demonstrated that gonadotropin (luteinizing hormone, LH) initiates oocyte maturation and ovulation in teleost fish and amphibians by stimulating the production of a maturation inducing substance (MIS) by the ovarian follicles [1]. The MISs have been identified as progesterone in a variety of amphibians and as hydroxylated progestins in teleost fishes [2,3]. Two C21 steroids, 17, 20β-dihydroxy-4-pregnen-3-one (17,20β-P) [4] and 17, 20β, 21-trihydroxy-4-pregnen-3-one (20β-S) [5,6], have been positively identified as the MISs in amago salmon and Atlantic croaker, respectively. While 17,20β-P is the major MIS for salmonids and cyprinids [2], 20β-S has been shown to be the predominant MIS in sciaenids and some other perciform fishes [3].
MIS does not induce OM in amphibians and teleosts by the classical mechanism of steroid action, instead it acts at the cell surface by binding to receptors located on the oocyte plasma membrane [3,7]. Activation of the MIS receptor results in induction of OM via a non-genomic mechanism [8] mediated by G-proteins and second messengers such as cAMP [9,10]. Progestin membrane receptors (mPRs) have been identified and characterized in several amphibian and teleost species [11-15]. Moreover, membrane progestin receptor (mPR) upregulation by gonadotropins during OM has been demonstrated in several teleost species [14,16] and has been associated with development of the ability of oocytes to become responsive to the MIS (oocyte maturational competence, OMC) and complete OM [17]. A two-stage model of the gonadotropic control of OM in teleosts has been proposed based on the results with several teleost species showing that priming of fully grown follicle-enclosed oocytes by gonadotropin is required to induce OMC [18]. Early studies in rainbow trout (Oncorhynchus mykiss) showing increased OM in response to the MIS after in vivo treatment with pituitary extracts provided an initial indication that priming with pituitary factors is necessary for the development of OMC in salmonids [19]. However, most studies have been conducted with perciform fishes, so that the mechanisms regulating the development of OMC and its timing relative to other processes during OM remain poorly understood in other teleosts.
The present study describes a comprehensive characterization of the ovarian mPR in a salmonid, Arctic char (Salvelinus alpinus). The Arctic char is of great commercial value in countries in the northern hemisphere, mainly due to its ability to grow at low temperatures, but the species also displays high sensitivity to environmental change. Functional reproduction is of outmost importance for species survival and successful breading in fish farms. Thus it is of great importance to obtain information on the basic mechanisms of reproduction in this species. We demonstrate that the receptor is upregulated during OM. The mPR abundance increase during OM, suggests its involvement in the development of OMC in salmonid fishes. Furthermore, the receptor display binding to xenobiotics, indicating that it is a target for endocrine disruptors that through binding to the mPR may interfere with oocyte maturation and thereby disrupt reproduction.
Materials and methods
Chemicals
[1,2,6,7-3H]17α-hydroxyprogesterone (specific activity 97 Ci/mmol) was obtained from New England Nuclear (Boston, MA) and enzymatically converted to 17,20β-P as described by [20]. The unlabeled steroids were purchased from either Steraloids, Inc. (Wilton, NH) or from Sigma Chemical Company (St. Louis, MO). The xenobiotics o,p'-DDT, o,p'-DDE, o,p'-DDD, Kepone and methoxychlor were obtained from Chem Services (Westchester, PA). 4-Nonylphenol was obtained from the Huntsman Corporation (Port Neches, TX). Flutamide and cimetidine were purchased from Sigma (St Louis, MO). The antiprogestin ORG31710 was a generous gift of Organon (Oss, The Netherlands) while ZK98,299 and ZK112,993 were generous gifts of Schering AG (Berlin, Germany). All steroids, antihormones and xenobiotics were dissolved in 95% ethanol to appropriate concentrations and stored at -20°C. Chemicals and salts used for making the buffers were purchased from Sigma Chemical Company (St Louis, MO) and from Fisher Scientific (Pittsburgh, PA). The scintillation cocktail was prepared without methanol according to [21].
Animals
Adult female Arctic char were obtained from Fiskeriverkets Forskningsstation, Kälarne, Sweden and held in indoor tanks with a continuous flow-through water system and fed a commercial Arctic char pelleted diet (Skretting, Norway) daily. A total of 30 fish were used in the present study. Temperature and photoperiod were adjusted to mimic natural conditions during the period of ovarian recrudescence. Fish were maintained under these conditions until their oocytes were fully grown and had diameters of approximately 5 mm.
Tissue sampling and preparation
Fish were anesthetized with MS222 (Sigma Chemical Company, St Louis, MO), sacrificed, and the ovaries were removed and immediately frozen in liquid nitrogen and stored at -80°C for up to 6 months prior to analysis. Thawed ovarian tissue (1–2 g) was placed between two 700 nm nylon mesh screens and clamped between two boards to rupture the oocytes and expel the yolk. The yolk was subsequently rinsed from the remaining tissue twice with 20 ml of ice-cold HAEW buffer (25 mM Hepes, 1 mM NaCl, 1 mM EDTA, pH 7.4). The tissue fragments were weighed and placed in 15 ml (~1:15 w/v) of ice-cold HAEP buffer (25 mM Hepes, 10 mM dithiothreitol (DTT), 1 mM NaCl, 1 mM EDTA, 1 mM PMSF, pH 7.4) and homogenized using a Polytron (Tekmar Tissuemizer). Homogenization was performed in two steps: first at low speed (setting 4) for 10 seconds and then at moderate speed (setting 7) for 30 seconds. The homogenate was centrifuged at 3,000 g for 5 minutes to remove the nuclear and heavy mitochondrial fractions [22]. The lipid layer was removed from the top and the supernatant was transferred to a new tube. The supernatant was centrifuged at 20,000 g for 5 minutes to pellet the plasma membrane fraction. The final pellet was resuspended in 10 ml of HAEP buffer and stored at -80°C until analysis.
Receptor Binding Assay
The filtration assay method originally developed by Patiňo and Thomas [12] to measure the mPR in spotted seatrout ovarian membranes was used with minor modifications. Radiolabeled 17,20β-P (final concentration 1 – 20 nM) was dissolved in HAEP buffer and a 250 μl aliquot was added to each assay tube with or without a 100-fold excess of cold steroid. The cold steroids were dissolved in ethanol, added to the assay tubes and the ethanol was dried down under N2. To each tube, an aliquot (250 μl) of the membrane preparation was added, the tubes were vortexed and incubated for 30 minutes at +4°C. The binding reaction was stopped by taking a 250 μl aliquot from each tube and passing it through a presoaked glass microfiber filter (Whatman GF/B, 2.1 cm, Whatman, Denmark) on a microfilter holder (Hoefer Scientific Instruments; FH124) attached to a vacuum pump. Each filter was rinsed with 12.5 ml of HAEW buffer and then placed in a 7 ml scintillation vial. Scintillation cocktail (5 ml) was added to each tube, the tubes were shaken for 5 minutes and the radioactivity was measured in a Beckman LS 6000SC scintillation counter (Beckman Instruments, Fullerton, CA). Each sample was assayed in triplicate. Specific binding in each sample was calculated by deducting the non-specific binding from the total binding.
Saturation analysis
Radiolabeled 17,20β-P (0.625–30 nM) was added to each reaction tube with or without 3 μM cold steroid. Membrane samples were incubated with steroids for 30 minutes at +4°C. The reactions were terminated by filtration and the radioactivity in the filter was determined as described above. Non-linear curve fitting procedures (GraphPad Prism, version 3.03, GraphPad Software Inc) were used to estimate the concentration of 17,20β-P binding sites (Bmax) and to calculate the dissociation constant (Kd). To determine the time necessary to reach binding equilibrium, membrane preparations were incubated in 7 nM radiolabeled 17,20β-P with or without 700 nM unlabeled 17,20β-P. The reaction was terminated at different time-points ranging between 15 seconds to 4 hours. The specific binding for each time-point was determined as described above.
Dissociation kinetics for [3H]-17,20β-P binding to the receptor was determined using standard procedures. Ovarian membrane preparations were first incubated with 7 nM radiolabeled 17,20β-P in the absence (total binding) or presence (nonspecific binding) of 700 nM unlabeled 17,20β-P for 30 minutes to ensure maximum association of the radioligand to the receptor. Aliquots (500 μl) of the membrane receptor suspensions were subsequently added to tubes containing 700 nM unlabeled 17,20β-P (500 μl) and incubated at +4°C. At various time points ranging between 15 seconds and 4 hours the reactions were terminated and the specific binding for each time point was determined as described above.
Tissue specificity
Plasma membrane suspensions were made from ovaries, heart, gill and muscle tissue as described above. All samples were assayed in triplicate and the resuspended membrane preparations were incubated with 7 nM of radiolabeled 17,20β-P in the presence or absence of 700 nM unlabeled steroid. After 30 minutes incubation at +4°C the reaction was terminated by filtration and the specific binding for each membrane preparation was determined as described above.
Steroid specificity
The steroid specificity of receptor binding was examined using a competitive binding assay. Ovarian tissue membrane preparations were incubated for 30 minutes in tubes containing 7 nM [3H]-17,20β-P with or without 11 different steroids at concentrations ranging between 0.1 nM and 10 μM. The following steroids were tested: 17,20β-P, 20β-S, progesterone (P4), cortisol (F), 17β-estradiol (E2), 11-ketotestosterone (11-KT), testosterone (T), 11-deoxycorticosterone (11-DOC), 17α-hydroxyprogesterone (17α-OH-P), 11-deoxycortisol (DC), and pregnenolone (P5). All cold steroids were dissolved in ethanol and added to the assay tubes. The ethanol was subsequently dried down under N2. After 30 minutes incubation at +4°C the reaction was terminated by filtration and the specific binding was determined as described above.
Effects of in vitro gonadotropin treatment on 17,20β-P binding to ovarian mPR
The possible hormonal regulation of ovarian 17,20β-P receptor by gonadotropin was investigated in vitro. Freshly prepared ovarian fragments (3 g) were incubated in Dulbecco's Modified Eagle's medium with Ham's nutrient mixture F-12 (DME/F12) (30 ml) containing 1, 7 and 14 IU human chorionic gonadotropin (hCG)/ml. The fragments were incubated at +15°C for 20 hours under an atmosphere of oxygen. At the end of the incubation the medium was discarded and the tissue fragments stored at -80°C until assayed for 17,20β-P binding.
To investigate the time course of 17,20β-P receptor regulation by hCG, ovarian fragments (3 g) were incubated for time-points ranging between 0 and 20 hours in DME/F12 containing 14 IU hCG/ml. At the end of each incubation the ovarian fragments were removed from the medium, some of the oocytes were examined for germinal vesicle migration (GVM) and germinal vesicle breakdown (GVBD), and the remaining ovarian tissue was immediately stored at -80°C until analysis. Estimations of 17,20β-P binding were obtained as described above.
Interactions of xenobiotics and hormone antagonists with the mPR
The binding affinities of a variety of xenobiotic chemicals for the 17,20β-P membrane receptor were investigated to determine whether the 17,20β-P action mediated by this receptor was potentially susceptible to chemical disruption. The xenoestrogens nonylphenol, Kepone, and o,p'-DDT with its metabolites o,p'-DDE and o,p'-DDD, the androgen antagonist flutamide, the progestin antagonists ZK 98299, ZK112992 and ORG 31710 and the H2 histamine receptor antagonist cimetidine were tested for receptor binding. All xenobiotics were added to the assay tubes dissolved in ethanol, which was dried down under N2 prior to addition of tissue preparation. The ovarian tissue suspensions were prepared as described above and incubated with 7 nM [3H]-17,20β-P in absence or presence of the xenobiotics over a broad range of concentrations (0,1 nM – 1 mM). After a 30 minutes incubation at +4°C free steroid was removed by filtration. The specific binding for each tissue suspension was determined as described above. The binding of xenobiotics to the 17,20β-P receptor was expressed as a percentage of the maximum specific binding of 17,20β-P binding to its receptor.
Data analysis
The significance (P) was calculated using a one way ANOVA followed by Bonferroni's multiple comparison test with a P < 0.05. All statistical analysis was performed using GraphPad Prism version 3.03 for Windows (GraphPad Software, San Diego California USA).
Results
Receptor characterization
High affinity (Kd, 13.8 ± 1.1 nM), specific 17,20β-P binding sites were detected on Arctic char ovarian plasma membranes by saturation analysis. Receptor binding was saturated at a concentration of 7 nM 17,20β-P and Scatchard analysis showed a single class of limited capacity (Bmax, 1.6 ± 0.6 pmol/mg ovary) binding sites for 17,20β-P (Fig. 1). Specific binding was detected over the pH range of 7.0 to 7.8 with maximum receptor binding at pH 7.4 (data not shown).
Figure 1 Representative saturation curve (A) and Scatchard analysis (B). Specific binding of [3H]-17,20β-P to the plasma membrane fraction of Arctic char ovaries using membrane suspensions incubated with [3H]-17,20β-P at concentrations ranging between 1 nM – 20 nM with or without a 100 fold excess non radiolabeled 17,20β-P. Specific binding (▼) was determined by subtracting non-specific binding (■) from total binding (▲). TB: total binding, NSB: nonspecific binding, SB: specific binding. All points are means of triplicate determinations.
The association kinetics of 17,20β-P binding to the receptor was rapid and saturation of the binding sites was achieved after 3 minutes at +4°C, with a t1/2 of 42 ± 3.4 seconds (Fig. 2). The non-specific binding did not change throughout the incubation time. The binding of [3H]-17,20β-P to the receptor was readily displaced with unlabeled 17,20β-P. The kinetics of dissociation were rapid and [3H]-17,20β-P was completely dissociated from the receptor after 5 minutes of incubation at +4°C, with a t1/2 of 30 ± 4.2 seconds (Fig. 2).
Figure 2 Association (●) and dissociation (■) kinetics. [3H]-17,20β-P binding to Arctic char ovarian plasma membrane preparations. The reactions were terminated at time-points ranging between 15 seconds and 4 hours. The data are expressed as percentages of maximum specific binding. Each point represents the average of three replicate assays.
Specific 17,20β-P binding was detected in membrane fractions of ovary, heart and gill Arctic char tissues, whereas no binding was detected in muscle tissue (Fig. 3). Highest 17,20β-P receptor binding was measured in ovarian tissue. High amounts of 17,20β-P binding were also observed in brain and liver tissues but subsequent saturation and Scatchard analyses showed that these binding moieties did not have the binding characteristics of steroid receptors (data not shown). In vitro treatment of ovarian fragments with gonadotropin (hCG) for 20 hours caused a concentration-dependent increase in 17,20β-P binding to the mPR. Maximum induction of the mPR, a five-fold increase over initial levels, was observed following treatment with 14 IU hCG/ml (Fig. 4). The time course study revealed that 17,20β-P binding to the mPR increased after 6–8 h of hCG treatment (Fig. 5), but that the increase was not significant until 12 hours of hCG treatment. Maximum receptor binding was not observed until after 20 hours of hCG exposure. These changes in mPR binding were accompanied by the induction of maturation of the follicle-enclosed oocytes. GVM, an early stage of OM, was first observed in about 10% of the oocytes after 6 hours of hCG incubation (Fig. 5). The proportion of oocytes undergoing GVM reached a maximum of 75% after 12 hours and subsequently declined at 20 hours as the oocytes proceeded to the later GVBD stage of OM.
Figure 3 Relative receptor abundance in different tissues. Specific [3H]-17,20β-P binding to membrane preparations of Arctic char ovary, heart, gill and muscle tissues. Data are presented as means ± SEM, obtained from three females.
Figure 4 Concentration-dependent effects of in vitro gonadotropin (hCG) treatment on ovarian 17,20β-P binding to mPR. Ovarian fragments were incubated with 0, 1, 7, or 14 IU hCG for 20 hours. Each bar represents the mean ± SEM of triplicate determinations. Significant difference at the p < 0.01 level compared to control (a) and the 7 IU group (b) are indicated in the figure.
Figure 5 Time course of increased ovarian 17,20β-P binding to mPR and oocyte maturation. Arctic char ovarian fragments were incubated with 14 IU hCG/ml. At the end of each incubation period, oocytes were examined for 17,20β-P binding and the occurrence of GVM and GVBD. Each bar represents the mean ± SEM from 5 different fishes, each one assayed in triplicate. * P < 0.01 compared to other time points.
Competition studies showed that steroid binding to the Arctic char ovarian membrane fraction was specific for C21 Δ4 (4-ene) steroids lacking a functional group at the 11 position, cortisol and pregnenolone displaying negligible affinity for the receptor (Fig. 6a,c). The common fish androgens (C19 steroids) T and 11-KT also showed little or no affinity for the receptor (Fig. 6b). The C18 steroid E2 displayed the highest relative binding affinity (5%) to the receptor of the other major steroids present in fish plasma (Table 1). 17,20β-P, the MIS in salmonids, showed 20× higher binding affinity for the receptor than any of the other steroids or xenobiotics tested. Removal of the OH group at the 20 position of the side chain of 17,20β-P, resulting in the formation of 17αOH-P, resulted in a forty-fold decrease in the relative binding activity (RBA) which was not further altered by the additional removal of the OH group on the 17 position (P4). Addition of an OH group at the 21 position of 17,20β-P, to produce 20β-S, the other major fish MIS, caused an 80 fold decline in binding affinity. Other 11-deoxycorticosteroids (11-DOC, DC) had similar RBAs as 20β-S. The presence of a functional (OH) group at the 11 position of a C21 steroid, cortisol (F), resulted in a further loss of binding affinity. Finally alterations of the ketone at the 3 position of progesterone to a hydroxyl and repositioning of the double bond to the 5 position (P5, Δ5, 5-ene steroid) resulted in a complete loss of binding affinity.
Figure 6 Competition by various steroids for binding of [3H]17,20β-P to ovarian plasma membranes. Ovarian membrane preparations were incubated for 30 minutes with [3H]-17,20β-P in the presence of 10 pM – 10 μM competitor. For explanation of abbreviations see Materials and Methods. Each point represents the mean of triplicate determinations. A) progestogens. B) androgens and estrogen. C) corticosteroids. D) xenobiotics.
Table 1 EC 50 and relative binding affinity (RBA) of various steroids and xenobiotics for Arctic char ovarian membrane fractions.
Competitor Concentration (nM) causing 50% displacement RBA (%)
17,20β-P 1.25 100.0
E2 37.9 4.68
P4 52.6 3.37
17α-OH-P 69.3 2.56
11-DOC 132.3 1.34
20β-S 145.2 1.22
DC 207.1 0.86
Flutamide 256.1 0.49
o,p'-DDT 377.5 0.37
o,p'-DDD 549.2 0.30
ZK 112992 1179 0.11
ZK 98299 12879.8 0.01
F 14421.2 0.01
T 18347.21 0.01
11-KT ND 0.0
P5 ND 0.0
Kepone ND 0.0
Nonylphenol ND 0.0
o,p'-DDE ND 0.0
Cimetidine ND 0.0
ORG 31710 ND 0.0
ND indicates substances that did not display 50% displacement of 17,20β-P from its receptor in the range of concentrations examined (steroids: 0.1 nM – 10μM, RBA < 0.01%; xenobiotics, hormone antagonists: 0.1 nM – 1 mM, RBA < 0.0001%).
All synthetic hormone antagonists and xenobiotics tested showed low or no binding affinity for the 17,20β-P receptor (Table 1; Fig. 6d). The highest binding affinity was displayed by the androgen antagonist flutamide, which had a RBA of 0.49%, a 50× higher affinity than T, the strongest binding androgen. The progestin antagonist ZK112992 also showed weak binding affinity (0.11%) for the mPR. Among the xenobiotics tested o,p'-DDT and its metabolite o,p'-DDD had a RBA less than 1% that of the natural ligand, 17,20β-P (0.37% and 0.30% respectively), while the other DDT metabolite o,p'-DDE, Kepone and nonylphenol did not bind the 17,20β-P receptor.
Discussion
In the present study, a 17,20β-P binding moiety was identified in Arctic char ovarian plasma membranes that fulfil all the criteria for its designation as a steroid membrane receptor. A single class of high affinity (Kd, 13.8 ± 1.13 nM, N = 6), saturable (Bmax, 1.6 ± 0.6 pmol/g ovary) and displaceable, specific 17,20β-P binding sites are detectable on ovarian membranes. The kinetics of 17,20β-P association and dissociation to the membrane binding sites are very rapid, with t1/2 s less then a minute, which is typical of steroid membrane receptors. The binding is highly specific for 17,20β-P and other progestins and is negligible for other steroids. The binding also shows tissue specificity, with highest levels of specific binding in the ovaries, a known target of 17,20β-P action. It was also found that 17,20β-P binding to mPR in Arctic char ovaries increase dramatically during OM which supports the suggestion that gonadotropin-induced upregulation of the MIS receptor is a common occurrence amongst teleosts and is of physiological importance during OM [17].
Specific binding of the salmonid MIS, 17,20β-P, and the synthetic progestin R5020 has been demonstrated with ovarian and oocyte plasma membranes from brook trout (Salvelinus fontinalis) [23] and rainbow trout [13]. Overall, the binding characteristics of the Arctic char ovarian mPR are similar to those reported previously. The affinity of 17,20β-P binding (Kd, 13.8 nM, to the mPR on Arctic char ovaries is similar to that reported for MIS binding to the mPR in rainbow trout (Oncorhynchus mykiss, Kd-18 nM) and yellowtail (Seriola quinqueradiata Kd, 22.9 nM), but somewhat lower than that reported in spotted seatrout (Cynoscion nebulosus, ovaries: Kd, 1.5–6.0 nM) and striped bass (Morone saxatilis Kd, 1.4 nM) gonadal membrane fractions [12-15]. The slightly higher Kds in rainbow trout tissues could be related to the higher amounts of MIS present in the plasma of this species [24] compared with Arctic char [25], while the lower Kd found in spotted seatrout and striped bass correlates well with lower MIS plasma levels during OM in these species [17,26].
The Arctic char ovarian mPR has a limited binding capacity for 17,20β-P (1.6 ± 0.6 pmol/g ovary), similar to the 17,20β-P receptor binding capacity reported previously in yellowtail (2.1 pmol/g ovary; [15]). The membrane receptor demonstrated rapid association and dissociation kinetics to the ligand as well as high ligand and tissue specificity, all specific characteristics of a steroid membrane receptor. The association and dissociation rates (T1/2 < 1 minute, equilibrium reached within 3 minutes) were similar to previously reported ovarian membrane receptor binding for MIS in some sciaenids [12], but differs from the reported t1/2 from carangids [15] where the t1/2 s were found to be slower (15 minutes).
An interesting finding was that gonadotropin induction of OM in vitro was associated with a dramatic five-fold increase in mPR concentrations in Arctic char ovarian tissues. Upregulation of mPR concentrations by gonadotropin during induction of OM in vitro has been observed previously in spotted seatrout ovaries [17,27]. It has been suggested that exposure to gonadotropin is necessary for the development of oocyte maturational competence and the onset of OM [28]. The Arctic char 17,20β-P membrane receptor displayed dose-dependent induction when preincubated in hCG for 20 hours. A time-course experiment showed that 17,20β-P binding activity was significantly increased after 12 hours treatment with hCG. The timing of the mPR upregulation was associated with GVM and the onset of GVBD after gonadotropin treatment. GVM occurred in a high percentage of the oocytes after 12 hours, and GVBD was observed between 12 and 20 hours of hCG treatment of whole Arctic char ovary fragments. These results are consistent with a physiological role for increased 17,20β-P binding to the mPR on the oocyte surface during oocyte maturation.
Determination of steroid specificity showed that 17,20β-P had the highest binding affinity, followed by E2 (4.68%) while 11-DOC and 20 β-S only had RBAs of 1.34% and 1.22%, respectively. These RBAs are comparable to results obtained with the yellowtail mPR, where 17,20β-P is the ligand with highest affinity [15]. In yellowtail the receptor binding-affinity for 11-DOC was slightly higher (3.2%) and 20β-S was slightly lower (0.8%) than observed for Arctic char.
It has been found that xenobiotic chemicals can interfere with nongenomic, cell surface-mediated actions of steroids [29,30]. Low, environmentally realistic, concentrations of a variety of xenobiotics has been shown to impair OM in vitro in response to the MIS in spotted seatrout and Atlantic croaker ovarian fragments [29,31], and also antagonize MIS upregulation of sperm motility [32]. The observation that these compounds could bind to the mPR on seatrout oocytes and croaker sperm provided the first indication that chemicals could interfere with nongenomic steroid actions by a receptor-mediated mechanism. Xenobiotic chemical binding to mPR was confirmed in a salmonid species for several of these compounds in the present study. In Arctic char both o,p'DDT and o,p'DDD showed significant binding to the mPR, with RBAs of 0.37 ± 0.04% and 0.30 ± 0.04% respectively. However, Kepone and the other xenobiotics tested did not show measurable binding to the membrane 17,20β-P receptor at concentrations up to 1 mM. In contrast, Kepone displayed the highest binding affinity of the chemicals tested for the spotted seatrout mPR (RBA ~ 0.1% relative to 20β-S), followed by o,p'DDD, methoxychlor and o,p'-DDT which showed weak displacement of 20β-S binding [29]. These differences in the binding affinities of xenobiotics for the spotted seatrout and Arctic char mPRs are not surprising, since these receptors also display marked differences in their RBAs for steroids. The spotted seatrout mPR shows the highest binding affinity for 20β-S and 17,20β-P has a RBA of 0.64% [32], while Arctic char mPR shows highest binding affinity for 17,20β-P and 20β-S has a RBA of 1.2%. These results indicate that the mPR located on Arctic char ovarian plasma membranes is a possible target for endocrine disruption. However, further research is needed to determine the extent of xenobiotic interactions with mPRs and their effects on reproduction.
In conclusion, this study shows the presence of a highly specific 17,20β-P receptor located in the ovarian plasma membrane in Arctic char. The receptor was identified as a high affinity and a low capacity mPR. This study also suggests that abundance of the Arctic char mPR has biological relevance since both the receptor density on the oocyte surface and the onset of OM clearly are regulated by gonadotropin. Furthermore, the interaction of xenobiotics with mPR suggests that it may be a target for endocrine disruption resulting in reduced OM thereby reducing the reproductive success.
Acknowledgements
This research was partially supported by Environmental Protection Agency STAR Grant R-82902401 (to P.T.), the Swedish Foundation for International Cooperation in Research and Higher Education (STINT) (to A.H.B.), Knut and Alice Wallenberg Foundation (to A.H.B.), the Swedish Environmental Protection Agency (to P.-E.O.), the Swedish Animal Welfare Agency (to P.-E.O.) and the Swedish Knowledge Foundation (to P.-E.O.).
==== Refs
Nagahama Y Norris DO, Jones RE Endocrine control of oocyte maturation Hormone and Reproduction in Fishes Amphibians and Reptiles 1987 New York:PlenumPress 171 202
Scott AP Canario AVM Status of oocyte maturation-inducing steroids in teleosts Proc Third Int Symp Reprod Physiol Fish 1987 223 234
Thomas P Davey KG, Peter RE, Tobe SS Hormonal control of oocyte maturation in sciaenid fishes Perspectives in Comparative Endocrinology 1994 Nat Res Council of Canada, Ottawa 619 625
Nagahama Y Adachi S Identification of maturation-inducing steroid in a teleost, the amago salmon (Oncorhynchus rhodurus) Dev Biol 1985 109 428 435 3996758
Trant JM Thomas P Shackleton CHL Identification of 17-alpha-20-beta 21 trihydroxy-4-pregnen-3-one as the major ovarian steroid produced by the teleost Micropogonias undulatus during final oocyte maturation Steroids 1986 47 89 100 3564088
Trant J Thomas P Isolation of a novel maturation-inducing steroid produced in vitro by ovaries of Atlantic croaker Gen Comp Endocrinol 1989 75 397 404 2792725
Liu Z Patiño R High-affinity binding of progesterone to the plasma membrane of Xenopus oocytes: Characteristics of binding and hormonal and developmental control Biol Reprod 1993 49 980 988 8286594
Nagahama Y Yoshikuni M Yamashita M Yokumoto T Katsu Y Regulation of oocyte growth and maturation in fish Curr Top Dev Biol 1995 30 103 145 7555044
Yoshikuni M Nagahama Y Involvement of an inhibitory G-protein in the signal transduction pathway of maturation-inducing hormone (17α, 20β-dihydroxy-4-pregnen-3-one) action in rainbow trout (Oncorhynchus mykiss) oocytes Dev Biol 1994 166 615 622 7813780
Thomas P Zhu Y Pace M Progestin membrane receptors involved in the meiotic maturation of teleost oocytes: a review with some new findings Steroids 2002 67 511 517 11960629
Sadler SE Maller JL Identification of a steroid receptor on the surface of Xenopus oocytes by photoafinity labeling J Biol Chem 1982 257 355 361 7198121
Patiño R Thomas P Characterisation of a membrane receptor activity for 17α,20β,21-trihydroxy-4-pregenen-3-one in ovaries of spotted seatrout (Cynoscion nebulosus) Gen Comp Endocrinol 1990 78 204 217 2354763
Yoshikuni M Shibata N Nagahama Y Specific binding of [3H]17α,20β-dihydroxy-4-pregnen-3-one to oocyte cortices of rainbow trout Fish Physiol Biochem 1993 11 15 24
King W Ghosh S Thomas P Sullivan CV A receptor for the oocyte maturation-inducing hormone 17α,20β,21-trihydroxy-4-pregnen-3-one on ovarian membranes of striped bass Biol Reprod 1997 56 266 271 9002659
Rahman M Ohta K Yoshikuni M Nagahama Y Chuda H Matsuyama M Characterization of ovarian membrane receptor for 17,20beta-dihydroxy-4-pregnen-3-one, a maturation-inducing hormone in yellowtail, Seriola quinqueradiata Gen Comp Endocrinol 2002 127 71 79 12161204
Thomas P Patiño R Scott AP, Sumpter J, Kime D, Rolge MS Changes in 17α,20β,21-trihydroxy-4-pregnen-3-one membrane receptor concentrations in ovaries of spotted seatrout during final oocyte maturation Proc Fourth Int Symp Reprod Physiol Fish 1991 122 124
Thomas P Pinter J Das S Upregulation of the maturation-inducing steroid membrane receptor in spotted seatrout ovaries by gonadotropin during oocyte maturation and its physiological significance Biol Reprod 2001 64 21 29 11133654
Patiño R Yoshozaki G Thomas P Kagawa H Gonadotropic control of ovarian follicle maturation: the two-stage concept and its mechanism Comp Biochem Physiol B 2001 129 427 439 11399477
Jalabert B Breton B Fostier A Precocious induction of oocyte maturation and ovulation in rainbow trout (Salmo gairdneri): problems when using 17α,20β-dihydroprogesterone Annu Biol Anim Biochem Biophys 1978 18 977 984
Scott AP Sheldrick EL Flint AP Measurement of 17 alpha, 20 beta-dihydroxy-4-pregnen-3-one in plasma of trout (Salmo gairdneri Richardson): seasonal changes and response to salmon pituitary extract Gen Comp Endocrinol 1982 46 444 451 7095406
Pinter J Thomas P Characterisation of a progestogen receptor in the ovaries of the spotted seatrout, Cynoscion nebulosus Biol Reprod 1995 52 667 675 7756460
Graham JM Graham J, Higgins J Isolation of mitochondria, mitochondrial membranes, lysosomes, peroxisomes, and Golgi membranes from rat liver Biomembrane Protocols 1 isolation and analysis 1993 Totowa, NJ: Human Press 29 40
Maneckjee A Idler DR Weisbart M Demonstration of putative membrane and cytosol steroid receptors for 17α,20β-dihydroxy-4-pregnene-3-one in brook trout Salvelinus fontinalis oocytes by photoaffinity labeling using synthetic progestins 17,21-dimethyl-19-nor-pregn-4,9-diene-3.20-dione (R5020) Fish Physiol Biochem 1991 9 123 135
Scott AP Sumpter JP A comparison of the female reproductive cycles of autumn-spawning and winter-spawning strains of rainbow trout (Salmo gairdneri Richardson) Gen Comp Endocrinol 1983 52 79 85 6628981
Mayer I Schmitz M Borg B Schultz R Seasonal endocrine changes in male and female Arctic charr (Salvelinus alpinus). I. Plasma levels of three androgens, 17α-hydroxy-20β-dihydroprogesterone, and 17β-estradiol Canadian Journal of Zoology 1992 70 37 42
Mylonas CC Woods LC 3rdThomas P Zohar Y Endocrine profiles of female striped bass (Morone saxatilis) in captivity, during postvitellogenesis and induction of final oocyte maturation via controlled-release GnRHa-delivery systems Gen Comp Endocrinol 1998 110 276 289 9593648
Zhu Y Rice CD Pang Y Pace M Thomas P Cloning, expression, and characterization of a membrane progestin receptor and evidence it is an intermediary in meiotic maturation of fish oocytes Proc Natl Acad Sci USA 2003 100 2231 2236 12574519
Patiño R Thomas P Effects of gonadotropin on ovarian intrafollicular processes during the development of oocyte maturational competence in a teleost, the Atlantic croaker: Evidence for two distinct stages of gonadotropic control of final oocyte maturation Biol Reprod 1990 43 818 877 2291916
Das S Thomas P Pesticides interfere with the nongenomic action of a progestogen on meiotic maturation by binding to its plasma membrane receptor on fish oocytes Endocrinology 1999 140 1953 1956 10098537
Thomas P Chemical interference with genomic and nongenomic action of steroids in fishes: role of receptor binding Mar Environ Res 2000 50 127 134 11460678
Ghosh S Thomas P Antagonistic effects of xenobiotics on steroid-induced final maturation of Atlantic croaker oocytes in vitro Mar Environ Res 1995 39 159 163
Thomas P Das S Correlations between binding affinities of C21 steroids for the maturation-inducing steroid membrane receptor in spotted seatrout ovaries and their agonist and antagonist activities in an oocyte maturation bioassay Biol Reprod 1997 57 999 1007 9369163
|
16281974
|
PMC1298333
|
CC BY
|
2021-01-04 16:37:13
|
no
|
Reprod Biol Endocrinol. 2005 Nov 10; 3:64
|
utf-8
|
Reprod Biol Endocrinol
| 2,005 |
10.1186/1477-7827-3-64
|
oa_comm
|
==== Front
Reprod Biol EndocrinolReproductive biology and endocrinology : RB&E1477-7827BioMed Central London 1477-7827-3-651628588210.1186/1477-7827-3-65ResearchIncrease of mitochondrial DNA content and transcripts in early bovine embryogenesis associated with upregulation of mtTFA and NRF1 transcription factors May-Panloup Pascale [email protected] Xavier [email protected]étien Marie-Françoise [email protected] Yvan [email protected] Manoel [email protected]èry Yves [email protected] Pascal [email protected] Biologie de la Reproduction, Labo FIV, Centre Hospitalier Universitaire d'Angers, 4 rue Larrey, F-49033 Angers, France2 Inserm, U694, F-49033 Angers, France3 INRA, Biologie du Développement et Reproduction, UMR 1198 INRA/ENVA, F-78352 Jouy en Josas cedex, France4 University of Illinois, Dept of Veterinary Clinical Medicine, 1008 West Hazelwood Dr. Urbana, IL 61802, USA5 Centre Hospitalier Universitaire d'Angers, Laboratoire de Biochimie et Biologie Moléculaire, 4 rue Larrey, F-49033 Angers, France2005 14 11 2005 3 65 65 29 8 2005 14 11 2005 Copyright © 2005 May-Panloup et al; licensee BioMed Central Ltd.2005May-Panloup 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 work has shown that mitochondrial biogenesis and mitochondrial functions are critical determinants of embryonic development. However, the expression of the factors controlling mitochondrial biogenesis in early embryogenesis has received little attention so far.
Methods
We used real-time quantitative PCR to quantify mitochondrial DNA (mtDNA) in bovine oocytes and in various stages of in vitro produced embryos. To investigate the molecular mechanisms responsible for the replication and the transcriptional activation of mtDNA, we quantified the mRNA corresponding to the mtDNA-encoded cytochrome oxidase 1 (COX1), and two nuclear-encoded factors, i.e. the Nuclear Respiratory Factor 1 (NRF1), and the nuclear-encoded Mitochondrial Transcription Factor A (mtTFA).
Results
Unlike findings reported in mouse embryos, the mtDNA content was not constant during early bovine embryogenesis. We found a sharp, 60% decrease in mtDNA content between the 2-cell and the 4/8-cell stages. COX1 mRNA was constant until the morula stage after which it increased dramatically. mtTFA mRNA was undetectable in oocytes and remained so until the 8/16-cell stage; it began to appear only at the morula stage, suggesting de novo synthesis. In contrast, NRF1 mRNA was detectable in oocytes and the quantity remained constant until the morula stage.
Conclusion
Our results revealed a reduction of mtDNA content in early bovine embryos suggesting an active process of mitochondrial DNA degradation. In addition, de novo mtTFA expression associated with mitochondrial biogenesis activation and high levels of NRF1 mRNA from the oocyte stage onwards argue for the essential function of these factors during the first steps of bovine embryogenesis.
==== Body
Background
Mitochondria, which are maternally inherited organelles, perform several cellular functions, e.g. energetic metabolism, calcium and iron homeostasis, signal transduction, and apoptosis, and play a role in metabolic pathways such as those involved in the biosynthesis of heme, lipids, amino acids and nucleotides [1]. These mitochondrial functions are therefore likely to be critical determinants of early embryonic development at various levels including spindle organization, chromosomal segregation, cell-cycle regulation, and morpho-dynamic processes such as compaction, cavitation and blastocyst hatching [2].
Pre-existing oocyte components are critical during the interval between fertilization and the so-called maternal-embryonic transition (MET) when the transcriptional activity of the embryonic genome becomes fully functional. During this period the development of the embryo is supported by maternal RNAs, proteins and organelles stored in the ooplasm. The transcription of the embryonic genome start at the 2-cell stage in the cow, defining a step called minor activation of the embryonic genome [3]. During the first cell divisions there is a balance between maternal and embryonic transcripts. Indeed, embryonic transcription and the degradation of maternal mRNA are gradual processes [4]. When the embryo reaches the 8/16-cell stage, the MET occurs, marking the major activation of the embryonic transcription and explaining the sharp increase in the RNA level at the blastocyst stage [3]. Throughout the preimplantation period the gene expression pattern is not constant but varies according to the gene considered.
The active transcription of the mitochondrial genome starts at different developmental stages depending on the species. In mice, the mtDNA transcription occurs in the late 2-cell stage, whereas it occurs in the 4/8-cell stage in humans and in the 8/16-cell stage in cattle [5,6]. The molecular mechanisms responsible for this transcriptional activation of mtDNA during early embryogenesis are not well understood. Ubiquitous transcription factors, such as the nuclear respiratory factor 1 (NRF1) and the mitochondrial transcription factor A (mtTFA), are well known to regulate mtDNA transcription in various tissues. NRF1 transactivates the promoters of a number of mitochondrial-related genes including genes coding for respiratory chain subunits and mtTFA [7]. Mitochondrial TFA is a nuclear-encoded high-mobility group (HMG) box protein, which binds upstream of the light- and heavy-strand mtDNA promoters [8]. This transcription factor also regulates mtDNA replication, since the initiation of replication of the leading strand of mtDNA depends on an RNA primer produced by transcription from the light-strand promoter. Moreover, there is new evidence that mtTFA plays a role in the direct regulation of the mtDNA copy number [9]. A recent study on early mouse embryogenesis shows that sharp changes in the abundance of NRF1 and mtTFA mRNAs occur in the 8-cell stage, which is one cell cycle before changes appear in mitochondrial oxidative phosphorylation transcripts, although mtDNA replication does not occur until later in the development [10]. In contrast to mouse embryos, in vitro fertilized bovine embryos showed a significantly higher mtDNA copy number at the blastocyst stage [11]. Since the bovine blastocyst has a high mtDNA copy number, it offers a good mammalian model to study the regulation of the transcription of factors controlling mtDNA replication in the preceding stage.
The purpose of this study was to explore the variation of mtDNA and mitochondrial RNA (mtRNA) content through the different stages of early bovine embryogenesis and to investigate the possible role of NRF1 and mtTFA in the activation of mtDNA replication and transcription. To achieve this, we used the real-time polymerase chain reaction (PCR) to quantify mtDNA, and real-time reverse-transcription PCR (RT-PCR) to quantify mtDNA transcripts as well as NRF1 and mtTFA transcripts in metaphase II bovine oocytes and in bovine embryos at early stages of development.
Methods
In vitro production of bovine oocytes and embryos
Cumulus oocyte complexes (COCs) were obtained from bovine ovaries collected immediately after slaughter and transported to the laboratory in container maintained at 30°C. The content of antral follicles 2–8 mm in diameter was aspirated and recovered in a conical 50-ml tube containing 10 ml of HEPES-buffered M199 medium at 39°C. Oocytes were selected on the basis of their morphology and rinsed before in vitro maturation. COCs containing degenerated oocytes, oocytes with irregular ooplasm, and COCs with abnormal or expanded cumulus investments were discarded.
The maturation of the COCs was carried out as described in the literature [12]. Briefly, the COCs were matured in vitro for 22–24 h at 39°C under a humidified atmosphere of 5% CO2 and air in M199 medium supplemented with 10% fetal calf serum (FCS) (Life Technologies, Cergy, France), 10 μg/ml FSH (Stimufol, Mérial, Lyon France), 1 μg/ml LH and 1 μg/ml estradiol 17β (Sigma). At the end of the maturation period, cumulus-expanded oocytes where either inseminated in vitro with frozen-thawed semen, or dechoronized and retained if they present their first polar body (group of metaphase II oocytes before insemination). One ejaculate from a single bull was used throughout all the experiments. Eighteen hours after fertilization, presumptive zygotes had their cumulus cells removed by vortexing and were transferred into 50 μl microdrops of B2 medium (CCD-laboratories, Paris, France) supplemented with 2.5% of FCS and containing a layer of Vero cells for coculture according to a technique described elsewhere [12]. A total number of 610 in vitro matured oocytes were inseminated through 5 replicate experiments. Inseminated oocytes were distributed by groups of about 30 presumptive zygotes in microdrops. Oocytes that failed to divide at 28 hpi were collected at that time to constitute the group of uncleaved oocytes. Each microdrop was then allocated to the collection of one of the following specific developmental stage : 2-cell at 28 hpi, 4/8-cell at 48 hpi, 8/16-cell at 72 hpi, morula at 120 hpi and blastocyst at 168 hpi. Only the more advanced embryos in development for each stage, and morulas and blastocysts of highest visual quality were included in the study. Each sample was constituted by a single oocyte or embryo rinsed in 50 μl of PBS – immediately frozen in liquid nitrogen and individually stored at -80°C until assay.
DNA extraction and mtDNA quantification was performed on 105 oocytes and embryos: 15 single metaphase II oocytes collected just before insemination, 15 metaphase II oocytes, which had failed to cleave, and 15 isolated embryos at each of the developmental stages. RNA extraction and RNA transcript quantification were carried out on a similar series of 105 oocytes and embryos at the same developmental stage.
DNA extraction
DNA was extracted from each single oocyte or embryo by means of the High Pure PCR Template Preparation Kit (Roche Diagnostics, Mannheim, Germany) according to the manufacturer's recommendations. The DNA was bound specifically to glass fibers following the combined action of a chaotropic agent (guanidine), a detergent (Triton X-100) and the enzyme proteinase K. After washing, the silica-bound DNA was eluted with 200 μl of pre-warmed (72°C) elution buffer and maintained at 4°C. The extraction efficiency, assessed as described elsewhere [13], was greater than 90%.
RNA extraction and Reverse transcription (RT)
Poly(A) RNA was prepared from isolated single oocytes and embryos using the High Pure Viral RNA Kit (Roche Diagnostics, Manheim, Germany) following the manufacturer's instructions. Briefly, lysis was accomplished by incubation of the sample in a special Binding Buffer (4.5 M guanidine-HCl, 50 mM Tris-HCl, 30% Triton® X-100) supplemented with poly(A) carrier RNA. The nucleic acids then bound specifically to the surface of glass fibers in the presence of a chaotropic salt. After washing, the silica-bound RNA was eluted with 50 μl of elution buffer and stored at -80°C until use. To confirm the absence of contaminating DNA, each RNA extract was subjected to the amplification protocol with the COX1 primer (see below) before reverse transcription.
Ten microlitres of each resultant poly(A) mRNA sample were used in duplicate. The RT-PCR reaction was carried out with the Advantage RT for PCR Kit (Becton Dickinson, Franklin Lakes, NJ, USA) following the manufacturer's instructions using a random hexamer mix to prime the RT reaction and to produce cDNA. Tubes were heated to 70°C for 2 mn to denature the secondary RNA structure and the RT mix was completed with 200 U of the MMLV RT enzyme. They were then incubated at 42°C for 1 hour to promote the reverse transcription of RNA, followed by incubation at 94°C for 1 mn to denature the RT enzyme. Each sample was completed to 50 μl with RNAse-free sterile water and stored at -80°C until use.
Primer design
For mtDNA quantification, we used a couple of primers located in the COX1 gene (Table 1). The PCR product was a 190-bp DNA fragment. mtRNA quantification was performed using the same couple of primers. Because bovine sequences for mtTFA and NRF1 are currently unknown, we first used primers located in sequences highly conserved between species in order to amplify bovine sequences. The conserved sequences of NRF1 and mtTFA mRNAs were evaluated by nucleotide multiple-sequence alignments of several orthologues using Clustal W 1.83 [14]. Alignments of complete NRF1 coding sequences were performed using sequences available in RefSeq from Homo sapiens (NM_005011), Mus musculus (NM_010938) and Danio rerio (NM_131680). Alignments of complete mtTFA coding sequences were performed using sequences available in RefSeq from Homo sapiens (NM_003201), Mus musculus (NM_010938), and Rattus norvegicus (NM_131680).
Table 1 Primer couples and PCR conditions
Gene Primer sequence D Primer sequence R PCR1 AT2 CN3
COX1 5'-AAA-TAA-TAT-AAG-CTT-CTG-ACT-CC-3' 5'-TCC-TAA-AAT-TGA-GGA-AAC-TCC-3' 190 56 4.8
mtTFA 5'-CAA-ATG-ATG-GAA-GTT-GGA-CG-3' 5'-AGC-TTC-CGG-TAT-TGA-GAC-C-3' 148 58 6.1
NRF1 5'-CCC-AAA-CTG-AGC-ACA-TGG-C-3' 5'-GTT-AAG-TAT-GTC-TGA-ATC-GTC-3' 162 58 5.6
1 PCR: PCR product length/pb
2 AT: Annealing temperature/degree C°
3 CN: Copy Number in 1 ng of PCR product × 109
After purification (High Pure Purification Kit, Roche Diagnostics, Mannheim, Germany) and sequencing of the PCR products, we designed bovine-specific primer couples. The PCR products were a 148-bp DNA fragment for mtTFA and a 162-bp DNA fragment for NRF1 (Table 1).
We also tried to quantify the housekeeping genes β-actin and histone H2A using primers described in the literature [15,16].
Preparation of external standards
For each gene studied, PCR reactions were carried out under standard conditions with 100 ng of total bovine DNA, extracted from a piece of bovine muscle, in a 50 μl volume: 1.5 mM MgCl2, 75 mM Tris-HCl (pH 9 at 25°C), 20 mM (NH4)2SO4, 0.01% Tween 20, 50 pmol of each primer, 200 μM of each dNTP and 2 units of GoldStar DNA polymerase (Eurogentec, Belgium). Each of the 30 cycles consisted of a denaturation step of 30 seconds at 94°C, a hybridization step of 30 seconds at 58°C, and an extension step of 1 min at 72°C. The PCR products were purified using the High Pure Purification Kit (Roche Diagnostics, Mannheim, Germany) and quantified by spectrophotometry. The quality of purification was checked by means of the 260/280 ratios, values between 1.8 and 2.0 being considered acceptable. It was assumed that 1 ng of a 100 bp product contained 9.1 × 109 molecules of double-stranded DNA. Table 1 shows the number of molecules of double-stranded DNA per nanogram of each of the PCR products obtained. Several serial dilutions were then made in order to assess the concentrations of a known number of templates. These were used as external standards for real-time PCR. The serial dilutions were all stored at -20°C in single-use aliquots.
Quantification of mtDNA and cDNA
We used a Roche LightCycler to determine the mtDNA and the cDNA copy number using the LightCycler FastStart DNA master SYBR Green 1 kit (Roche, Mannheim, Germany) as described elsewhere [13]. Briefly, 20-μl PCR reaction mixtures were prepared as follows: 1× buffer containing 4 mM MgCl2, 0.2 mM dNTPs, 0.5 μM of both primers for each gene, SYBR Green I dye, 0.25 U HotStart Taq DNA polymerase and 10 μl of the extracted mtDNA or 10 μl of the cDNA obtained or 10 μl of standard with a known copy number. The reactions were performed as follows: initial denaturing at 95°C for 7 min and 40 cycles at 95°C for 1 s, 56–59°C for 5 s, and 72°C for 13 s. The SYBR Green fluorescence was read at the end of each extension step (72°C). A melting curve (loss of fluorescence at a given temperature between 66°C and 94°C) was analyzed in order to check the specificity of the PCR product. For each run, a standard curve (log of the initial template copy number on the abscissas, and the cycle number at the crossing point on the ordinates) was plotted using five 10-fold serial-dilutions (100–1,000,000 copies) of the external standard. This standard curve, which depends on the efficiency of the PCR reaction, allowed the determination of the starting copy number of mtDNA or of the cDNAs in each sample. All samples were tested twice. The raw data was then multiplied by 20 to calculate the total mtDNA content in each oocyte or embryo. For the transcripts studied, we multiplied the raw data by 25 to express the cDNA level for each oocyte or embryo treated. The precision of the real-time PCR quantification was assessed as described elsewhere [13]. The CV of the intra-assay and inter-assay values ranged from 3.9% to 9.1% and from 9.3% to 12.7% respectively.
Statistical analysis
Since the distribution of the variables analyzed was non-Gaussian, all comparisons were made using the non-parametric Mann-Whitney and Kruskal-Wallis U-tests. Results are given as mean values ± SE. Statistical analysis was performed with SPSS software, version 10.1 (SPSS, Chicago, IL, USA) and differences were considered significant at p < 0.05.
Results
In this study, the in vitro cleavage rate was 88% (as assessed by the number of embryos with 2 cells or more at 48 hpi) and 54,4% of cleaved oocytes developed to the blastocyst stage at day 7.
mtDNA copy number
The mean mtDNA copy number at each embryonic stage is shown in Figure 1. There was no statistical difference between the mean mtDNA copy number in metaphase II oocytes (373,000 ± 63,000) and 2-cell embryos (371,000 ± 52,000). In contrast, the mtDNA content was significantly higher in 2-cell embryos compared to 4/8-cell embryos (p = 0.0008). There was no significant variation of the mean mtDNA copy number between the 4/8-cell stage (135,000 ± 28,000), the 8/16-cell stage (163,000 ± 36,000) and the morula stage (180,000 ± 26,000). However, there was a considerable increase in the mtDNA copy number at the blastocyst stage (688,000 ± 50,000) (p < 0.0001).
Figure 1 Comparison of mtDNA content in bovine oocytes and embryos at various stages of development. The mtDNA content decreases between the 2-cell and the 4/8-cell stages (p = 0.0008); in contrast, it increases sharply between the morula and the blastocyst stages (p < 0.0001). Bars with different superscript differ significantly.
mRNAs quantification
It seems illusory to seek a detectable and constant housekeeping gene during early embryogenesis. Indeed, this is a much debated subject [10,15-17]. In our study β-actin levels remained low or undetectable during the first stages (oocyte, 2-cell, 4/8-cell, 8/16-cell stages), and then increased dramatically during the morula stage. Moreover, histone H2A was not detectable until the morula stage (data not shown). The reproducibility of the results (two RT-PCRs for each sample tested twice for each gene) and the homogeneity at a given embryonic stage led us to express our results in arbitrary units per oocyte or embryo.
The levels of mitochondrial COX1 mRNA, which remained roughly constant from the oocyte to the 8/16-cell stage, increased sharply after the morula stage (p = 0.002) (Figure 2). The mtTFA mRNA was undetectable until it appeared at the morula stage. The quantity of this transcript increased dramatically at the blastocyst stage (p < 0.0001) (Figure 2). The quantity of NRF1 transcripts remained practically constant from the oocyte to the morula stage, after which it increased significantly up to the blastocyst stage (p < 0.0001) (Figure 2).
Figure 2 Comparison of COX1, mtTFA and NRF1 transcript levels in oocytes and at various stages of bovine embryonic development. Levels of COX1 mRNA remain constant from the oocyte to the 8/16-cell stage, and then increase sharply from the morula stage onwards (p = 0.002). mtTFA mRNA was not detected before the morula stage. The abundance of this transcript increased dramatically at the blastocyst stage (p < 0.0001). The abundance of NRF1 transcripts remained practically constant from the oocyte to the morula stage, after which it increased significantly up to the blastocyst stage (p < 0.0001).
Uncleaved oocyte
There was no significant difference between mean mtDNA copy number of post insemination uncleaved oocytes (415,000 ± 24,000) and 2-cell embryos (371,000 ± 52,000) both collected 28 hpi. However, the number of COX1 and NRF1 RNA transcripts was significantly lower in the uncleaved oocytes as compared to the embryos that cleaved or to the oocyte before insemination (p < 0.0001) (Figure 3). Moreover, the mean mtDNA copy number and transcripts levels are similar in oocytes collected before insemination and 2-cell embryos (Figure 3).
Figure 3 Comparison between uncleaved oocytes and 2-cell embryos (both collected at 28 hpi), and oocytes collected before insemination. Although, the mean mtDNA copy numbers did not differ between these three groups, COX1 and NRF1 transcripts were significantly fewer in uncleaved oocytes (p < 0.0001).
Discussion
The mean mtDNA copy number per bovine metaphase II oocyte reported here (373,000) is comparable to the 260,000 copies/oocyte first determined in 1982 [18] using the hybridization technique, and very close to the 377,000 copies/oocyte recently found by our group using real-time quantitative PCR [19]. Experiments on bovine oocytes have shown that the mtDNA content, in at least some species, is related to the competence of development to the blastocyst stage [19]. The progression of the mtDNA content during early embryogenesis in vitro has been performed only in mouse eggs [20]. This study, performed on pooled oocytes and embryos with the Southern-blot technique, showed that the mtDNA content remained constant from the oocyte to the implantation stage. These results have led to the general belief that mtDNA replication does not occur until after implantation [21]. This result has just been confirmed by quantitative PCR analysis on mouse embryos [10]. However, the fact that the mtDNA copy number remains stable during early mouse embryogenesis could be due to a balance between the degradation and synthesis of mtDNA. Indeed, a recent report has indicated that mtDNA replication occurs in pronuclear and 2-cell stage mouse embryos [22]. In contrast to the mouse model, a marked increase of mtDNA replication from the blastocyst stage onwards was found in bovine embryos. Thus, DNA replication is disconnected from the implantation event (21 days in the bovine), and occurs at an earlier embryonic stage in this species. We observed a significant reduction of about 60% in the mtDNA content between the 2-cell and the 4/8-cell stages. This finding is reinforced by the fact that the metaphase II oocytes and the 2-cell embryos had similar high mtDNA levels, whereas the 4/8-cell, the 8/16-cell embryos and the morulas had similar low mtDNA levels (Figure 2). This drastic reduction of mtDNA content argues in favor of active destruction rather than a reduced turnover of mtDNA molecules. It has been demonstrated that in the course of mammalian embryogenesis, the paternal mtDNA is destroyed at the same stage by a mechanism involving the proteasome [23]. The active destruction of mtDNA would be compatible with the bottleneck hypothesis proposed to explain the homogeneity of the transmitted mitochondrial genomes. This phenomenon of restriction-amplification in the mtDNA copy number seems to occur in multiple steps during oogenesis and embryogenesis [24].
COX1 is a respiratory chain protein encoded by mtDNA. We found that COX1 mRNA increases sharply from the morula stage onwards. The same pattern of expression has been described for cytochrome b mRNA, which is another respiratory chain transcript encoded by the mitochondrial genome [25]. According to several reports [5,6], the onset of mitochondrial transcriptional activity appears to occur at the same time as the MET. Before this, during the early stages of bovine embryogenesis, the level of mitochondrial transcripts remains roughly constant. This observation is supported by other studies in which the inhibition of mitochondrial transcription permitted embryonic development in the mouse up to the blastocyst stage [8,26]
In mammals, mtTFA has been isolated only in humans [27], in mice [28], and in rats [29]. The mouse mtTFA gene, estimated to span about 10 kb, consists of 7 exons and 6 introns [30]. The NRF1 gene sequence in mice, sheep, fish and humans is partially or totally known. The human NRF1 gene spans 65 Kb and comprises 11 exons and 10 introns [31]. Since the sequences of bovine mtTFA and NRF1 genes remain unknown, we selected nucleotide sequences of primer couples among the exon sequences that are highly conserved between species to obtain bovine PCR products. We retained only the primer couples that yielded single, pure PCR products. Upon sequencing, these PCR products showed a homology of 76% with human mtTFA and a homology of 89% with human NRF1.
NRF1 and mtTFA are ubiquitous factors well known to regulate mtDNA transcription and replication in various tissues. The critical role of mtTFA in embryogenesis has been demonstrated in transgenic experiments. Indeed, in knockout mice the implication of mtTFA in the regulation of the mtDNA copy number has been demonstrated together with its essential involvement in mitochondrial biogenesis and embryonic development [8]. In our study, we found no mtTFA expression before the morula stage. The appearance of mtTFA transcripts is concomitant with the increase of mitochondrial mRNA and just precedes the increase of the mtDNA copy number. This result strongly suggests that the activation of mitochondrial biogenesis in the bovine embryo occurs between the 8/16-cell and the morula stages under the impulse of mtTFA.
Moreover, it has been shown that the homozygous disruption of the mouse NRF1 gene leads to embryonic death around the time of implantation. The depletion of mtDNA occurring between fertilization and the blastocyst stage suggests that NRF1 is required for mitochondrial maintenance in vivo. In this mouse model, the transcription of NRF1 occurred not only during oogenesis but also in early embryogenesis [32]. We found that NRF1 expression was constant up to the blastocyst stage. Thus, it is likely that NRF1 mRNA pre-exists in the oocyte and that a balance is established between the degradation of maternal transcripts and the synthesis of embryonic mRNA. The expression of NRF1 all through early embryogenesis may be necessary to maintain mitochondrial activity and other vital embryonic functions without the intervention of mtTFA [33]. This hypothesis is supported by the finding that homozygous NRF1 knockout embryos died significantly earlier than homozygous mtTFA knockout mouse embryos (at an average age of 6.5 days versus 10.5 days). Conversely, the onset of mitochondrial biogenesis at the MET stage under the dependence of mtTFA may be initiated either by NRF1, progressively unmasked to become functional, or by other transcription factors known, at least in humans, to act on the mtTFA promoter [34].
We found that bovine oocytes that failed to cleave at 28 hpi contained significantly fewer transcripts implicated in mitochondrial biogenesis (COX1 and NRF1 mRNAs) than 2-cell stage embryos (collected at the same time) as well as potentially fertilizable oocytes collected before insemination. This finding substantiates the hypothesis that mitochondrial quality is closely related to the fertilizability of the oocyte and to the developmental capacity of the embryo [35]. Indeed, in the case of human oocytes, the developmental potential of the embryo has been shown to be related to the ATP content of the cells [36]. Furthermore, the injection of a small number of mitochondria into mouse oocytes prevents these cells from undergoing apoptosis [37]. However, further investigation will be needed to establish whether the impairment of the factors of mitochondrial biogenesis is the central cause of fertilization failure or merely incidental to a vaster death process.
To our knowledge, this is the first study on a bovine model and using isolated oocytes and embryos. We have determined the kinetics of mtDNA replication and transcription during early bovine embryogenesis in vitro and studied the expression of mtTFA and NRF1, the two main regulators of mitochondrial biogenesis. Our results support the hypothesis that these factors play a critical role in mitochondrial biogenesis during early embryogenesis.
Authors' contributions
MPP carried out the DNA and RNA extraction, the PCR (and the RT-PCR) reactions and drafted the manuscript. VX and HY participated in collecting the oocytes and embryos and collaborated in the design and the coordination of this study. CMF and TM have made substantial contributions to the analysis and interpretation of the data. They have been involved in revising the manuscript critically for its content. MY has been involved in revising the manuscript critically for his content. RP conceived the study and participated in its design and coordination and helped to draft the manuscript. All the authors have read and approve of the final manuscript.
Acknowledgements
The authors wish to thank Ms Y. Lavergne for the technical preparation of bovine oocytes and embryos, and Dr K. Malkani for his critical reading of the manuscript.
==== Refs
Delbart c Rôle des mitochondries Les mitochondries: biologie et incidences physiopathologiques 2000 Paris: Tec & Doc 61 93
Van Blerkom J Mitochondria in human oogenesis and preimplantation embryogenesis: engines of metabolism, ionic regulation and developmental competence Reproduction 2004 128 269 280 15333778 10.1530/rep.1.00240
Memili E Dominko T First NL Onset of transcription in bovine oocytes and preimplantation embryos Mol Reprod Dev 1998 51 36 41 9712315 10.1002/(SICI)1098-2795(199809)51:1<36::AID-MRD4>3.0.CO;2-X
Gandolfi TA Gandolfi F The maternal legacy to the embryo: cytoplasmic components and their effects on early development Theriogenology 2001 55 1255 1276 11327683 10.1016/S0093-691X(01)00481-2
Telford NA Watson AJ Schultz GA Transition from maternal to embryonic control in early mammalian development: a comparison of several species Mol Reprod Dev 1990 26 90 100 2189447 10.1002/mrd.1080260113
Thompson JG McNaughton C Gasparrini B McGowan LT Tervit HR Effect of inhibitors and uncouplers of oxidative phosphorylation during compaction and blastulation of bovine embryos cultured in vitro J Reprod Fertil 2000 118 47 55 10793625 10.1530/reprod/118.1.47
Scarpulla RC Nuclear control of respiratory chain expression in mammalian cells J Bioenerg Biomembr 1997 29 109 119 9239537 10.1023/A:1022681828846
Larsson NG Wang J Wilhelmsson H Oldfors A Rustin P Lewandoski M Barsh GS Clayton DA Mitochondrial transcription factor A is necessary for mtDNA maintenance and embryogenesis in mice Nat Genet 1998 18 231 236 9500544 10.1038/ng0398-231
Ekstrand MI Falkenberg M Rantanen A Park CB Gaspari M Hultenby K Rustin P Gustafsson CM Larsson NG Mitochondrial transcription factor A regulates mtDNA copy number in mammals Hum Mol Genet 2004 13 935 944 15016765 10.1093/hmg/ddh109
Thundathil J Filion F Smith LC Molecular control of mitochondrial function in preimplantation mouse embryos Mol Reprod Dev 2005 71 405 413 15895466 10.1002/mrd.20260
Smith LC Thundathil J Filion F Role of the mitochondrial genome in preimplantation development and assisted reproductive technologies Reprod Fertil Dev 2005 17 15 22 15745628 10.1071/RD04084
Menck MC Guyader-Joly C Peynot N Le Bourhis D Lobo RB Renard JP Heyman Y Beneficial effects of Vero cells for developing IVF bovine eggs in two different coculture systems Reprod Nutr Dev 1997 37 141 150 9178355
Reynier P May-Panloup P Chretien MF Morgan CJ Jean M Savagner F Barriere P Malthiery Y Mitochondrial DNA content affects the fertilizability of human oocytes Mol Hum Reprod 2001 7 425 429 11331664 10.1093/molehr/7.5.425
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
Lonergan P Rizos D Gutierrez-Adan A Moreira PM Pintado B De La Fuente J Boland MP Temporal Divergence in the Pattern of Messenger RNA Expression in Bovine Embryos Cultured from the Zygote to Blastocyst Stage In Vitro or In Vivo Biol Reprod 2003 69 1424 1431 12826577 10.1095/biolreprod.103.018168
De Sousa PA Watson AJ Schultz GA Bilodeau-Goeseels S Oogenetic and zygotic gene expression directing early bovine embryogenesis: a review Mol Reprod Dev 1998 51 112 121 9712325 10.1002/(SICI)1098-2795(199809)51:1<112::AID-MRD14>3.0.CO;2-9
Robert C McGraw S Massicotte L Pravetoni M Gandolfi F Sirard MA Quantification of housekeeping transcript levels during the development of bovine preimplantation embryos Biol Reprod 2002 67 1465 1472 12390877 10.1095/biolreprod.102.006320
Michaels GS Hauswirth WW Laipis PJ Mitochondrial DNA copy number in bovine oocytes and somatic cells Dev Biol 1982 94 246 251 6295849 10.1016/0012-1606(82)90088-4
Tamassia M Nuttinck F May-Panloup P Reynier P Heyman Y Charpigny G Stojkovic M Hiendleder S Renard JP Chastant-Maillard In vitro embryo production efficiency in cattle and its association with oocyte adenosine triphosphate content, quantity of mitochondrial DNA, and mitochondrial DNA haplogroup Biol Reprod 2004 71 697 704 15084486 10.1095/biolreprod.103.026104
Piko L Taylor KD Amounts of mitochondrial DNA and abundance of some mitochondrial gene transcripts in early mouse embryos Dev Biol 1987 123 364 374 2443405 10.1016/0012-1606(87)90395-2
Cummins JM The role of mitochondria in the establishment of oocyte functional competence Eur J Obstet Gynecol Reprod Biol 2004 115 S23 29 15196712 10.1016/j.ejogrb.2004.01.011
McConnell JM Petrie L Mitochondrial DNA turnover occurs during preimplantation development and can be modulated by environmental factors Reprod Biomed Online 2004 9 418 424 15511342
Sutovsky P Moreno RD Ramalho-Santos J Dominko T Simerly C Schatten G Ubiquitinated sperm mitochondria, selective proteolysis, and the regulation of mitochondrial inheritance in mammalian embryos Biol Reprod 2000 63 582 590 10906068 10.1095/biolreprod63.2.582
Smith LC Bordignon V Couto MM Garcia SM Yamazaki W Meirelles FV Mitochondrial genotype segregation and the bottleneck Reprod Biomed Online 2002 4 248 255 12709275
Bilodeau-Goeseels S Schultz GA Changes in the relative abundance of various housekeeping gene transcripts in in vitro-produced early bovine embryos Mol Reprod Dev 1997 47 413 420 9211425 10.1002/(SICI)1098-2795(199708)47:4<413::AID-MRD7>3.0.CO;2-N
Piko L Chase DG Role of the mitochondrial genome during early development in mice. Effects of ethidium bromide and chloramphenicol J Cell Biol 1973 58 357 378 4738106 10.1083/jcb.58.2.357
Fisher RP Clayton DA Purification and characterization of human mitochondrial transcription factor 1 Mol Cell Biol 1988 8 3496 3509 3211148
Larsson NG Garman JD Oldfors A Barsh GS Clayton DA A single mouse gene encodes the mitochondrial transcription factor A and a testis-specific nuclear HMG-box protein Nat Genet 1996 13 296 302 8673128 10.1038/ng0796-296
Gadaleta G D'Elia D Capaccio L Saccone C Pepe G Isolation of a 25-kDa protein binding to a curved DNA upstream the origin of the L strand replication in the rat mitochondrial genome J Biol Chem 1996 271 13537 13541 8662779 10.1074/jbc.271.23.13537
Larsson NG Barsh GS Clayton DA Structure and chromosomal localization of the mouse mitochondrial transcription factor A gene (Tfam) Mamm Genome 1997 8 139 140 9060414 10.1007/s003359900373
Gopalakrishnan L Scarpulla RC Structure, expression, and chromosomal assignment of the human gene encoding nuclear respiratory factor 1 J Biol Chem 1995 270 18019 18025 7629110 10.1074/jbc.270.26.15675
Huo L Scarpulla RC Mitochondrial DNA instability and peri-implantation lethality associated with targeted disruption of nuclear respiratory factor 1 in mice Mol Cell Biol 2001 21 644 654 11134350 10.1128/MCB.21.2.644-654.2001
Scarpulla RC Transcriptional activators and coactivators in the nuclear control of mitochondrial function in mammalian cells Gene 2002 286 81 89 11943463 10.1016/S0378-1119(01)00809-5
Virbasius JV Scarpulla RC Activation of the human mitochondrial transcription factor A gene by nuclear respiratory factors: a potential regulatory link between nuclear and mitochondrial gene expression in organelle biogenesis Proc Natl Acad Sci U S A 1994 91 1309 1313 8108407
Hsieh RH Au HK Yeh TS Chang SJ Cheng YF Tzeng CR Decreased expression of mitochondrial genes in human unfertilized oocytes and arrested embryos Fertil Steril 2004 81 912 918 15019829 10.1016/j.fertnstert.2003.11.013
Van Blerkom J Davis PW Lee J ATP content of human oocytes and developmental potential and outcome after in-vitro fertilization and embryo transfer Hum Reprod 1995 10 415 424 7769073
Perez GI Trbovich AM Gosden RG Tilly JL Mitochondria and the death of oocytes Nature 2000 403 500 501 10676949 10.1038/35000651
|
16285882
|
PMC1298334
|
CC BY
|
2021-01-04 16:37:13
|
no
|
Reprod Biol Endocrinol. 2005 Nov 14; 3:65
|
utf-8
|
Reprod Biol Endocrinol
| 2,005 |
10.1186/1477-7827-3-65
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-1301626909010.1186/1465-9921-6-130ResearchIsolation of human β-defensin-4 in lung tissue and its increase in lower respiratory tract infection Yanagi Shigehisa [email protected] Jun-ichi [email protected] Hiroshi [email protected] Yukari [email protected] Hiroshi [email protected] Naoyoshi [email protected] Masamitsu [email protected] Third Department of Internal Medicine, Miyazaki University School of Medicine, Miyazaki 889-1692, Japan2 Second Department of Internal Medicine, Nagasaki University School of Medicine, Nagasaki 852-8501, Japan3 Peptide Institute, Inc. Osaka 562-8686, Japan2005 4 11 2005 6 1 130 130 21 7 2005 4 11 2005 Copyright © 2005 Yanagi et al; licensee BioMed Central Ltd.2005Yanagi 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
Human β-defensin-4 (hBD-4), a new member of the β-defensin family, was discovered by an analysis of the genomic sequence. The objective of this study was to clarify hBD-4 expression in human lung tissue, along with the inducible expression in response to infectious stimuli, localization, and antimicrobial activities of hBD-4 peptides. We also investigated the participation of hBD-4 in chronic lower respiratory tract infections (LRTI) by measuring the concentrations of hBD-4 peptides in human bronchial epithelial lining fluid (ELF).
Methods
The antimicrobial activity of synthetic hBD-4 peptides against E. coli and P. aeruginosa was measured by radial diffusion and colony count assays. We identified hBD-4 in homogenated human lung tissue by reverse-phase high-performance liquid chromatography coupled with a radioimmunoassay (RIA). Localization of hBD-4 was studied through immunohistochemical analysis (IHC). We investigated the effects of lipopolysaccharide (LPS) on hBD-4 expression and its release from small airway epithelial cells (SAEC). We collected ELF from patients with chronic LRTI using bronchoscopic microsampling to measure hBD-4 concentrations by RIA.
Results
hBD-4 exhibited salt-sensitive antimicrobial activity against P. aeruginosa. We detected the presence of hBD-4 peptides in human lung tissue. IHC demonstrated the localization of hBD-4-producing cells in bronchial and bronchiolar epithelium. The levels of hBD-4 peptides released from LPS-treated SAECs were higher than those of untreated control cells. ELF hBD-4 was detectable in 4 of 6 patients with chronic LRTI, while the amounts in controls were all below the detectable level.
Conclusion
This study suggested that hBD-4 plays a significant role in the innate immunity of the lower respiratory tract.
==== Body
Background
Bronchial epithelial lining fluid (ELF) contains various antimicrobial substances to protect against pathogenic insult. The antimicrobial components of the ELF are lysozyme, lactoferrin, secretory phospholipase-A2, and antimicrobial peptides, including defensins [1]. Defensins, which are single-chain, strongly cationic antimicrobial peptides with a molecular weight of 3,000–4,500, have broad-spectrum antimicrobial activities against various Gram-positive and Gram-negative bacteria, mycobacteria, fungi, and certain enveloped viruses [1]. Defensins are classified as α-and β-defensins based on the connectivity of their six cystein residues [1]. Human β-defensins (hBDs) are expressed mainly in epithelial cells. hBD-1 is expressed constitutively in the epithelia of the urogenital tract, trachea, and respiratory tract [2-4]. hBD-2 and hBD-3, isolated from psoriatic scale extracts [5,6], are expressed mainly in the respiratory tract, and their expression increases in response to infections and inflammatory mediators [6-11]. In addition, these two hBDs show strong antimicrobial activity against pathogens of respiratory infections, including P. aeruginosa, and thus they seem to function in airway mucosal defense [6-11].
hBD-4, a new member of the β-defensin family, was identified by analysis of genomic sequence mapping at chromosome 8p23, where all known α- and β-defensins are clustered [12]. hBD-4 mRNA is expressed in human testis, stomach, neutrophils, lung, and other organs [12], but neither hBD-4 peptide expression in human lung tissue nor its pathophysiological significance in respiratory tract infections has been clarified. We here studied the role of hBD-4 in lower respiratory tract infections (LRTI). We showed the existence, localization, and inducible expression of hBD-4 in response to infectious stimuli. In addition, we determined the concentrations of hBD-4 in human ELF collected by the bronchoscopic microsampling (BMS) method to investigate the significance of hBD-4 in respiratory tract infections.
Methods
Peptide synthesis
The reduced peptide of hBD-4, designed by García et al. and composed of 37 amino acid residues, was obtained by the chemical ligation method [12]. An oxidative folding reaction of the reduced peptide was carried out in 0.1 M ammonium acetate buffer (pH 7.8) in the presence of reduced and oxidized glutathione (GSH/GSSG) in a molar ratio of 1/100/10 (reduced hBD-4/GSH/GSSG) at 4°C overnight. Reversed-phase high-performance liquid chromatography (RP-HPLC) analysis revealed a single distinct main product, which was purified by preparative RP-HPLC on a YMC C18 column and ion-exchange chromatography on CM-Sepharose. The peptide thus obtained was passed through columns of Muromac and then Sephadex LH-20 to obtain hBD-4 in the acetate form (the yield of the oxidized peptide was 56% based on the reduced peptide). The purity of synthetic hBD-4 was confirmed to be sufficiently high by RP-HPLC, IEX-HPLC, capillary zone electrophoresis, amino acid analysis, sequence analysis, elemental analysis, and matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry (observed m/z was 4367.3, theoretical [M+H]+ = 4367.0). The synthetic products of hBD-2 and hBD-3 were purchased from Peptide Institute Inc. (Osaka, Japan).
Bactericidal assay
Radial diffusion and colony count assays were used to examine antimicrobial activity [13,14]. We studied the antimicrobial ability of synthetic hBD-4 as well as hBD-2, hBD-3, and penicillin G (Sigma, St. Louis, MO, USA) by radial diffusion assay with E. coli strain HB101 and P. aeruginosa strain PAO1 (supplied by T. Hayashi, Department of Microbiology, Miyazaki University). Briefly, bacteria were cultured at 37°C overnight in trypticase soy broth (TSB; Nissui Pharmaceutical Co., Ltd., Tokyo, Japan). An aliquot of this culture was transferred to fresh TSB and incubated for 4 h at 37°C to obtain cells in logarithmic-phase growth. Following the precipitation of bacteria by centrifugation at 800 × g for 10 min, the samples were washed with phosphate-buffered saline (PBS) and quantified spectrophotometrically at 620 nm. A culture volume containing 1 × 106 bacterial colony-forming units (CFU) was then added to 10 ml warm (40°C) autoclaved PBS containing 3.0 g of TSB medium and 1% low electroendosmosis-type agarose. After a rapid dispersion of bacteria, the bacteria-containing agar was poured into a plate to form a uniform layer. Wells measuring 3 mm in diameter were then created in the agar using a gel punch. After 5 μl of each control samples and each diluted peptides to each well, the samples were incubated for 18 h at 37°C. The antimicrobial activity was taken as the difference between the size of the clear zone surrounding the wells containing defensins, penicillin G, and those containing control sample.
The antimicrobial activities of hBD-2, hBD-3, and hBD-4 were also examined by colony count assay using E. coli HB101 and P. aeruginosa PAO1. Then, 5000 CFU of bacteria was incubated for 2 h at 37°C with defensin in concentrations ranging in tenfold steps from 0.1 to 1000 μg/ml. The final volume of the incubation medium was 50 μl. To measure antibacterial activity more precisely, some series were performed by repeating the analysis with defensin concentrations that ranged in twofold steps from 0.625 to 40 μg/ml. Since the differences in salt sensitivity in the antimicrobial activity of hBDs were previously reported [3,6,7,15], we evaluated the salt sensitivity of the antimicrobial activity of the defensins using two incubation media conditions: 1) a high salt condition (Na+ 137 mEq/L, Cl- 130 mEq/L, K+ 4.2 mEq/L, osmolarity 270 mOsm/kg, pH 7.4) and 2) a low salt condition (Na+ 95 mEq/L, Cl- 90 mEq/L, K+ 25 mEq/L, osmolarity 210 mOsm/kg, pH 7.1). The incubation mixtures were serially diluted, spread on nutrient agar plates, and incubated for 18 h at 37°C. The antimicrobial activity was expressed as the colony reduction ratio, defined as the number of killed bacteria to that of control bacteria.
Preparation of antiserum
hBD-4 (2.5 mg) was conjugated to bovine thyroglobulin (15 mg) using 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide HCL (400 mg) as described previously [16], then dialyzed five times against two liters of 0.9% sodium chloride to remove unconjugated material. An antigenic conjugate solution (0.9–3.0 ml) was used to immunize three New Zealand white rabbits by multiple intra- and sub-cutaneous injections. The animals were given booster shots every 2 weeks, then were bled 7 days after each injection. All experimental protocols were approved by the Ethics Review Committee for Animal Experimentation of Miyazaki University.
Study population
For immunohistochemistry, we obtained human normal lung tissues from 2 patients at surgery: a 38-year-old female with pulmonary mucormycosis and a 70-year-old male with bullae. The patient with mucormycosis also exhibited insulin-dependent diabetes mellitus, while the other patient had no complications that induced an immunosuppressive condition. The patient with mucormycosis was a smoker, and the other patient was not. To evaluate the localization of hBD-4 in chronic LRTI, we also obtained human lung tissue from a 63-year-old female with middle lobe syndrome.
For radioimmunoassay experiments, 6 controls (2 males and 4 females, ranging from 30 to 78 years old, 1 smoker and 5 nonsmokers) and 6 patients (2 males and 4 females, ranging from 64 to 83 years old, all 6 nonsmokers) with chronic LRTI who had persistent productive cough with purulent sputum for more than 6 months were enrolled in this study. The following exclusion criteria were adopted for the patient group: (i) steroids, immunosuppressive drugs, or any antibiotics prescribed within 3 months; (ii) cancer or diabetes mellitus. The pathogens of patients with chronic LRTI consisted of the mucoid phenotype of P. aeruginosa in 3 cases and the nonmucoid phenotype of P. aeruginosa in 3 cases. The controls underwent bronchoscopy to identify the causes of small solitary peripheral nodules. The final diagnoses of the controls consisted of the healing stage of pulmonary suppuration in 1 case and lung nodule of unidentified etiology in 5 cases. According to the results of the histological study, laboratory data, clinical course, and radiological findings including positron emission tomography, we confirmed strongly that the pulmonary diseases in the 6 controls were all benign. In the controls, no bacterial compounds were detected in samples obtained from the respiratory tract. All controls and patients gave written informed consent to participate in the study, which was approved by the Research Ethics Committee of Miyazaki University.
Immunohistochemical study
Normal lung tissues from the 2 patients mentioned above, as well as lung tissues with chronic LRTI from a 63-year-old female with middle lobe syndrome, were obtained at surgery for immunohistochemical study. The tissues were fixed in 3.7% formaldehyde in 10 mM PBS (pH 7.2), dehydrated in a graded ethanol series, and embedded in paraffin. Cut sections (3 μm thick) were deparaffinized in xylene, rehydrated in a graded ethanol series, and then washed in Tris-buffered saline containing Tween 20 (TBST; DakoCytomation Co., Ltd., Kyoto, Japan). For antigen retrieval, the sections were incubated in 1 μg/ml proteinase K (DakoCytomation) for 30 min at 37°C and treated with 6% hydrogen peroxidase for 60 min to inactivate endogenous peroxidases. Nonspecific binding was inhibited by an incubation in Protein Block (DakoCytomation) for 3 h at 37°C. Preparations were incubated overnight at 4°C with anti-hBD-4 antiserum at a final concentration of 1/10000. Staining was visualized using the Dako CSA system (DakoCytomation) according to the manufacturer's protocol. Control studies utilized normal rabbit serum or anti-hBD-4 antiserum that had been pre-absorbed with 1 μg hBD-4.
Radioimmunoassay (RIA) procedure
hBD-4 was radioiodinated by the lactoperoxidase method [17]. The 125I-labeled peptide was purified by RP-HPLC using a TSK ODS 120A column (Tosoh Co., Ltd., Tokyo, Japan). RIA reaction mixtures were incubated in 50 mM sodium phosphate (pH 7.4) containing 0.25% N-ethylmaleimide-treated BSA, 80 mM NaCl, 25 mM EDTA·2Na, 0.05% NaN3, 0.1% Triton X-100, and 3.1% Dextran T-40. Diluted samples or standard peptide solutions (100 μl) were incubated for 24 h in 100 μl of antiserum no. 1–4 (final concentration: 1/2,100,000). A solution of the tracer, 16,000–18,000 cpm of 125I-labeled peptide in 100 μl reaction buffer, was then added. After 24 h incubation, normal rabbit serum and anti-rabbit IgG goat serum were added for an additional 12 h incubation. Bound and free ligands were separated by centrifugation. All procedures were performed at 4°C. Samples were assayed in duplicate. In the RIA for hBD-4, antiserum no. 1–4 recognized hBD-4 with high affinity at final dilutions of 1/2,100,000 (35% binding). Half-maximum inhibition occurred at 7 pg/tube. The peptide remained detectable at the low level of 0.7 pg/tube. At 50% binding, the respective intra- and inter-assay coefficients of variation were 3.9% and 4.2%. This antiserum did not exhibit any cross-reactivity for human neutrophil peptide-1, hBD-1, hBD-2, or hBD-3.
Chromatographic characterization of immunoreactive hBD-4 in lung
Normal human lung tissue, isolated as described above for immunohistochemical studies, was heated at 95–100°C for 10 min in a 10-fold volume of water to inactivate intrinsic proteinases. After cooling to 4°C, CH3COOH and HCL were added at final concentrations of 1 M and 20 mM, respectively. Following homogenization in a Polytron for 15 min, the homogenate was centrifuged at 18,500 × g for 30 min at 4°C. The resulting supernatant was applied to a Sep-Pak C-18 cartridge (Waters, Milford, MA, USA) pre-equilibrated in 0.5 M CH3COOH. Peptides were eluted in 35% acetonitrile (CH3CN) containing 0.1% trifluoroacetic acid (TFA). The eluate was examined by RP-HPLC on a TSK ODS SIL 120A (Tosoh Co. Ltd., Tokyo, Japan) column using a linear gradient of 10–35% CH3CN containing 0.1% TFA at a rate of 1.0 ml/min for 40 min. All fractions were assayed for hBD-4 by RIA.
Cell culture and induction of hBD-4 expression
Small airway epithelial cells (SAECs) were purchased from Clonetics and grown to monolayers in tissue culture flasks at 37°C in a 5% CO2-humidified atmosphere. SAECs were maintained in SAGM (Cambrex Bioscience Walkersville, Inc., Walkersville, MD, USA). Hydrocortisone and bovine serum albumin were removed from this medium before treatment with stimulants and during the time of the study. All experiments were performed between the third and fifth passages.
For the analysis of hBD-4 peptide expression and release, SAECs were grown in a 175 cm2 flask (Falcon). When 70–80% confluence was reached, SAECs were incubated for 24 h with culture medium alone (control) or medium containing 100 μg/ml P. aeruginosa-derived lipopolysaccharide (LPS). After stimulation, 70 ml of each medium (derived from approximately 5 × 107 SAECs) was collected and centrifuged (3500 rpm, 30 min), then the supernatants were transferred to a new tube and stored at -20°C until use. The cells were washed twice with cold PBS. Then 10 ml of PBS was added to the flask, and the cells were scraped and collected into a centrifuge tube. After centrifugation (3500 rpm, 30 min), the PBS was aspirated off. The cell pellet was frozen in liquid nitrogen, weighed, and heated at 95–100°C for 10 min in a tenfold volume of water to inactivate intrinsic proteases. After cooling to 4°C, CH3COOH and HCL were added to the respective final concentrations of 1 M and 20 mM, after which the cell pellet was homogenized in a Polytron for 10 min. The homogenate was centrifuged at 18,500 × g for 30 min at 4°C. Both supernatants and extracts from the cells were applied to a Sep-Pak C-18 cartridge pre-equilibrated in 0.5 M CH3COOH. The peptides were eluted in 35% acetonitrile (CH3CN) containing 0.1% trifluoroacetic acid (TFA). The eluate was lyophilized, and the residue was dissolved in 0.1 M sodium phosphate buffer (pH 7.4) containing 0.05% Triton X-100. The peptides were then measured by RIA for hBD-4.
Bronchoscopic microsampling of ELF
Using the BMS method, we obtained ELF from patients with chronic LRTI and controls to measure the concentrations of hBD-4. The BMS probe (Olympus Co., Tokyo, Japan) and sampling procedure were described previously [18]. In brief, after routine premedication, a flexible BF-XT40 fiberoptic bronchoscope (Olympus) was inserted into the lungs. After flushing with air to minimize contamination of the samples, the BMS probe was inserted through the channel into the right lower lobe bronchus. Then the inner probe was advanced slowly into the distal airway, and ELF was sampled by placing the probe gently at a site on the target bronchial wall for 10 seconds. The inner probe was withdrawn into the outer tube, and both devices were withdrawn simultaneously. The wet inner probe was sectioned 2 cm from its tip. Three sectioned probes at one time point from each subject were placed in a preweighed tube and weighed. A dilute solution was prepared by adding 3 ml of saline to the tube and vortexing it for 1 min. The solution was transferred to a new tube and stored at -20°C until use. The probe was then dried and weighed again to measure the ELF volume. The saline-diluted sample (3 ml) was applied to a Sep-Pak C-18 cartridge pre-equilibrated in 0.5 M CH3COOH. Adsorbed peptides were eluted in 35% CH3CN containing 0.1% TFA. The eluate was lyophilized and assayed by hBD-4-specific RIA. The concentrations of hBD-4 in ELF (hBD-4ELF) were determined as follows:
hBD-4ELF = hBD-4BMS × (3 + ELF volume) / ELF volume,
where hBD-4BMS is the measured concentration of hBD-4 in the saline-diluted sample. We also assayed the serum concentrations of hBD-4 in both groups. A serum sample (1 ml) of each groups was collected just before the ELF was obtained. Both ELF and the serum were applied to a Sep-Pak C-18 cartridge pre-equilibrated in 0.5 M CH3COOH. Adsorbed peptides were eluted in 35% CH3CN containing 0.1% TFA. The eluate was lyophilized and assayed by hBD-4-specific RIA.
Statisitical analysis
Data were expressed as means ± standard deviations (SD). Differences between groups were examined using the analysis of variance (ANOVA) and Scheffe's test. A p value of < 0.05 was considered statistically significant.
Results
Antimicrobial activity of hBD-4
We performed a radial diffusion assay with synthesized defensins and penicillin G. hBD-4 exhibited dose-dependent antimicrobial activity, and this activity was stronger against P. aeruginosa than against E. coli (Fig. 1). The antimicrobial activity of hBD-4 against P. aeruginosa was stronger than that of hBD-2. We next studied the antimicrobial activity of hBD-4 by a colony count assay under two different electrolyte concentrations (Table 1). Under the low salt condition (Na+ 95 mEq/L, Cl- 90 mEq/L, K+ 25 mEq/L, osmolarity 210 mOsm/kg, pH 7.1), the concentration of hBD-4 at which the population of E. coli colony was reduced by 50% was 9.1 ± 3.5 μg/ml, which was higher than that for hBD-2 (1.1 ± 0.7 μg/ml). In contrast, hBD-4 had an antimicrobial effect as strong as those of hBD-2 and hBD-3 (1.0 ± 0.5 μg/ml and 0.6 ± 0.2 μg/ml, respectively) against P. aeruginosa under the low salt condition (1.3 ± 0.6 μg/ml). The antimicrobial activity of hBD-4, like that of hBD-2, decreased under the high salt condition (Na+ 137 mEq/L, Cl- 130 mEq/L, K+ 4.2 mEq/L, osmolarity 270 mOsm/kg, pH 7.4), although the activity of hBD-3 did not change substantially under these two conditions.
Figure 1 Antimicrobial activities of hBD-2 (open circles), hBD-3 (closed circles), hBD-4 (open squares), and penicillin G (closed squares). (A) E. coli HB101, (B) P. aeruginosa PAO1. An increase in zone size represents the zone size measured at each antimicrobial compound concentration minus the zone size of the central control well (3 mm). Data represent the means ± SD of three independent experiments.
Table 1 Concentration of human β defensins effective in reducing 50% colony of bacteria.
MIC (μg/ml)
hBD-2 hBD-3 hBD-4
Organism H-salt L-salt H-salt L-salt H-salt L-salt
E. coli 26.6 ± 7.6 1.1 ± 0.7 5.7 ± 2.6 4.1 ± 0.8 147 ± 31 9.1 ± 3.5
P. aeruginosa 11.6 ± 1.6 1.0 ± 0.5 0.6 ± 0.2 0.6 ± 0.2 >500 1.3 ± 0.6
The bacteria were incubated with defensin in concentrations ranging tenfold in steps from 0.1 to 1000 μg/ml. To measure antibacterial activity more precisely, some values were determined by repeating the analysis with defensin concentrations that ranged in twofold steps from 0.625 to 40 μg/ml. Two incubation media conditions were tested: H-salt was a high salt condition (Na+ 137 mEq/L, Cl- 130 mEq/L, K+ 4.2 mEq/L, osmolarity 270 mOsm/kg, pH 7.4), and L-salt was a low salt condition (Na+ 95 mEq/L, Cl- 90 mEq/L, K+ 25 mEq/L, osmolarity 210 mOsm/Kg, pH 7.1). Values represent the means ± SD of three experiments. (MIC: minimum inhibitory concentration)
Identification of hBD-4 peptide in the lung
In the two normal lung samples examined, hBD-4-immunoreactive cells were diffusely observed in the bronchial and bronchiolar epithelium (Fig. 2A and 2B, respectively). Airway epithelial cells showed strong and granular cytoplasmic immunostaining. hBD-4 immunoreactivity was not detected in alveolar epithelial cells (Fig. 2C). Tissue immunoreactivity was abrogated by preabsorption of the antiserum with 1 μg/ml hBD-4 peptide (Fig. 2D). Immunoreactive hBD-4 was also identified in the human lung by RP-HPLC combined with RIA (Fig. 3). hBD-4-immunoreactive peaks in the samples were eluted at the same position as the synthetic hBD-4 peptide. We also performed immunohistochemical analysis obtained from one patient with chronic LRTI. Bronchial epithelial cells showed strong and granular cytoplasmic immunostaining (Fig. 4A). Additionally, hBD-4 immunoreactivity was detected in neutrophils and suppurative exudates within the bronchial lumen (Fig. 4B).
Figure 2 Immunohistochemical study of hBD-4 expression in the human lung. For each pair of images, the upper panels (A1, B1, and C1) are the results of the immunohistochemical study of the lung tissue obtained from a 38-year-old female with pulmonary mucormycosis, and the lower panels (A2, B2, and C2) are those obtained from a 70-year-old male with bullae. Immunoreactive cells are present around the bronchial surface (A1, A2) and bronchiolar surface (B1, B2). hBD-4 immunoreactivity is not detected in alveolar epithelial cells (C1 and C2). No immunoreactivity is detected in tissues following preadsorption of antiserum with 1 μg/ml hBD-4 peptide (D). The bar represents a length of 50 μm in all panels.
Figure 3 Representative RP-HPLC profiles of hBD-4 immunoreactivity. Samples were obtained from 300 mg human lung tissue. Fraction volumes of 0.5 ml were obtained by RP-HPLC using a TSK ODS SIL 120A (4.6 Å × 150 mm) column and a linear gradient of 10–60% CH3CN containing 0.1% TFA at a rate of 1.0 ml/min for 40 min. Arrows indicate the elution position of synthetic hBD-4. "ir-hBD-4" on the Y-axis means immunoreactive hBD-4.
Figure 4 Immunohistochemical study of hBD-4 expression in patients with chronic lower respiratory tract infection. hBD-4 immunoreactivity presented in bronchial epithelial cells (A), neutrophils and suppurative exudates within bronchial lumen (B). The bar represent a length of 50 μm in (A, B).
Induction of hBD-4 peptides from lung epithelial cells by LPS in vitro
We next assessed whether or not infectious stimuli up-regulate the release of hBD-4 peptide in bronchial epithelial cells in vitro. Figure 5 shows the hBD-4 peptide concentrations in the supernatant of SAECs incubated for 24 h with medium alone or with 100 μg/ml of P. aeruginosa-derived LPS. The concentrations of hBD-4 peptide released from LPS-treated SAECs were higher than those of untreated control cells (P <0.05). Moreover, there was little content of hBD-4 peptide in either the untreated or LPS-treated SAECs (data not shown).
Figure 5 Expression profiles of hBD-4 SAECs. hBD-4 peptide concentrations in supernatants of SAECs after 24 h incubation with medium alone (control; open bars), and 100 μg/ml of LPS (solid bar). Values represent the means ± SD of three experiments. (SAECs: small airway epithelial cells, LPS: lipopolysaccharide)
hBD-4 levels in ELF in patients with chronic LRTI
Since β-defensins are expressed constitutively or inducibly in response to infection, we measured the ELF and serum concentrations of hBD-4 in patients with chronic LRTI and controls. ELF hBD-4 was detectable in 4 of 6 patients with chronic LRTI, while the amounts in the controls were all below the detectable level (Fig. 6). The mean ELF concentration of hBD-4 in patients with chronic LRTI was 181.6 pg/ml (range, 0 to 380 pg/ml). All 3 patients infected with the mucoid phenotype of P. aeruginosa demonstrated high ELF concentrations of hBD-4, while hBD-4 was not detectable in the ELF of 2 of the 3 patients infected with the nonmucoid phenotype. The serum hBD-4 concentrations of both groups were below the detectable level (data not shown).
Figure 6 Epithelial lining fluid levels of hBD-4 in controls (n = 6) and patients with chronic lower respiratory tract infection (n = 6). In the CLRTI group, open circles indicate patients infected with the mucoid phenotype of P. aeruginosa, and closed circles indicate patients infected with the nonmucoid phenotype of P. aeruginosa. The horizontal bar represents the mean value. (CLRTI: chronic lower respiratory tract infection, ELF: epithelial lining fluid)
Discussion
The present study indicates that hBD-4 plays a significant role in the innate immunity of the lower respiratory tract. The strong antimicrobial activity of hBD-4 against P. aeruginosa rather than against E. coli, along with its stronger antimicrobial activity relative to that of hBD-2, underscores the deep involvement of hBD-4 in the innate immunity of the lower respiratory tract. The localization of hBD-4 and other hBD peptides in the bronchial and bronchiolar epithelium also supports that they contribute to the mucosal defenses of the lung [6,7,9,11,15]. We here demonstrated that hBD-4 is induced in the ELF of patients with chronic LRTI, making this study the first investigation of antimicrobial peptide expression in the human respiratory tract in vivo. In the present study, the ELF hBD-4 concentrations were not high enough to suppress bacterial proliferation in any patients with respiratory tract infections. However, hBD-4 acts synergistically with lysozyme [12], which is released from neutrophils in P. aeruginosa infections [19]. Moreover, hBD-4 is found to have a strong additive effect with hBD-3 [12]. Together, these findings indicate that hBD-4 collaborates with other antimicrobial substances to defend the airway mucosa against P. aeruginosa infections.
hBD-4 exhibited salt-sensitive antimicrobial activity. All defensins are strongly cationic, which facilitates their interaction with bacteria and allows the formation of multimeric pores within the negatively charged cell membrane [7]. Previous reports show that the antimicrobial activities of desalted ELF obtained from both cystic fibrosis (CF) and normal xenografts were higher than that of crude ELF obtained from their xenografts. This suggests that high NaCl concentrations inactivate defensin antimicrobial activity by weakening the electrostatic interactions between defensins and the cytoplasmic membrane [20]. The salt sensitivity of hBD-4 strengthens the concept that inactivation of these peptides is one of the major factor in recurrent airway infections in patients with CF.
Compared with the localization of hBD-4 within the cytoplasm of airway epithelial cells in normal lung tissues, hBD-4 immunostaining in lung tissue of chronic LRTI was observed in the bronchial lumen as well as in the cytoplasm of epithelial cells. Also, IHC and ELF findings suggested there was no spontaneous release of hBD-4 into the airway from epithelial cells in the absence of any infectious stimuli. Hence, there is a possibility that hBD-4 in the ELF of the controls was present in amounts too small to be detected. The controls selected here for RIA were not completely healthy. However, pulmonary diseases that are known to induce the expression of defensins, such as malignant diseases, were excluded from the controls. hBD-4 was thought to be released in response to specific stimulation such as infection.
The high hBD-4 levels in supernatant, combined with little content of hBD-4 in LPS-treated SAECs after 24 h, means that SAECs biosynthesized hBD-4 only after being stimulated and released promptly into the extracellular space. It remains unknown whether a direct or indirect action of P. aeruginosa is responsible for the biosynthesis and release of hBD-4. Previous reports show that P. aeruginosa up-regulates hBD-4 mRNA expression in SAECs [12], but there is a possibility that these phenomena occur indirectly via cytokines produced from airway epithelial cells. However, inflammatory cytokines such as IL-1α, IL-6, interferon-γ, and TNF-α did not induce up-regulation of hBD-4 mRNA expression in SAECs [12]. Therefore, further investigation is needed to clarify the mechanism underlying these phenomena.
Interestingly, the ELF in all patients infected with the mucoid phenotype of P. aeruginosa demonstrated high hBD-4 concentrations, while hBD-4 was not detectable in the ELF of 2 of the 3 patients infected with the nonmucoid phenotype. The high hBD-4 levels in ELF may have originated from airway epithelial cells and neutrophils in chronic LRTI, since hBD-4 immunoreactivity was also detected in neutrophils. However, the high hBD-4 levels in ELF could not be explained solely by neutrophilsmediated inflammation because of a significant difference between the mucoid and nonmucoid phenotypes of P. aeruginosa. A difference in hBD expression in response to P. aeruginosa between the mucoid and nonmucoid phenotypes has also been shown in hBD-2 in vitro [10]. The mucoid phenotype of P. aeruginosa may contain unique signaling molecules that stimulate respiratory epithelial cells for the production of hBDs. hBD-2 exhibits cytotoxic effects at >50 μg/ml concentrations against airway epithelial cells in vitro [21]. And colonization of the mucoid phenotype of P. aeruginosa in the respiratory tracts has been related to the progression of bronchial airway pathology [19]. Although it remains uncertain whether or not hBD-4 is cytotoxic to airway epithelial cells, the mucoid phenotype of P. aeruginosa can damage the respiratory tracts both directly and via the release of hBDs from bronchial epithelial cells.
The expression of hBD-4 and the release of hBD-4 from bronchial epithelial cells are both up-regulated in response to infectious stimuli [12], while hBD-1 is constitutively expressed in the absence of infectious stimulation [9]. Interestingly, hBD-4 immunoreactivity is not detected in alveolar epithelial cells where hBD-2 is expressed [22]. Furthermore, hBD-4 has specific signal pathways; hBD-4 induction is mediated by protein kinase C, but not by NF-κB or STAT, which are associated with up-regulation of hBD-2 and hBD-3, respectively [11,12,23]. In the present study, hBD-4 as well as hBD-2 exhibited salt-sensitive antimicrobial activity, whereas hBD-3 did not. Finally, although the members of the hBD peptide family have similar amino acid structures, hBD-4 is suggested to play a different role than the other hBDs in the defense against respiratory tract infections.
The hBD-4 peptide exhibited strong antimicrobial activities against P. aeruginosa, which is the most virulent pulmonary pathogen because of its intrinsic resistance to multiple classes of antibiotics [24,25]. Antimicrobial peptides have many of the desirable features of a novel antibiotic class. They have a broad spectrum of activity, kill bacteria quickly, are unaffected by classical antibiotic resistance mutations, and have selective toxicity. Although further investigation is required, including in vivo study, hBD-4 may be an attractive candidate for a new therapeutic agent against P. aeruginosa infection.
Conclusion
hBD-4 plays a significant role in the innate immunity of the lower respiratory tract. Further molecular analyses of hBD-4 activity will provide a better understanding of the physiological role and pathophysiological significance of this molecule in respiratory infectious disease.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
SY evaluated the antimicrobial activity of peptides, performed immunohistochemical study, cultured the SAECs, did the BMS, drafted the manuscript, and participated in the design of the study. HI prepared antiserum, established RIA, and performed RP-HPLC. CN synthesized hBD-4 peptide. JA, YD, HM, and NM conceived the study and helped to draft the manuscript. All authors read and approved the manuscript.
Acknowledgements
The authors wish to thank Dr. K. Toshinai, Dr. T. Simbara, Dr. M.S. Mondal, and Dr. T. Kawagoe of the Miyazaki University School of Medicine, Japan, for their invaluable advice in the experiment on antimicrobial activities, RIA, and cell culture. We also would like to thank S. Tajiri for her excellent technical assistance. This study was supported in part by the 21st Century COE Program.
==== Refs
Bals R Epithelial antimicrobial peptides in host defense against infection Respir Res 2000 1 141 150 11667978 10.1186/rr25
Bensch KW Raida M Mägert HJ Schulz-Knappe P Forssmann WG hBD-1: a novel β-defensin from human plasma FEBS Lett 1995 368 331 335 7628632 10.1016/0014-5793(95)00687-5
Valore EV Park CH Quayle AJ Wiles KR McCray PB JrGanz T Human β-defensin-1: An antimicrobial peptide of urogenital tissues J Clin Invest 1998 101 1633 1642 9541493
Zhao C Wang I Lehrer RI Widespread expression of beta-defensin hBD-1 in human secretory glands and epithelial cells FEBS Lett 1996 396 319 322 8915011 10.1016/0014-5793(96)01123-4
Harder J Bartels J Christophers E Schröder JM A peptide antibiotic from human skin Nature 1997 387 861 9202117 10.1038/43088
Harder J Bartels J Christophers E Schröder JM Isolation and characterization of human β-defensin-3, a novel human inducible peptide antibiotic J Biol Chem 2001 276 5707 5713 11085990 10.1074/jbc.M008557200
Bals R Wang X Wu Z Freeman T Bafna V Zasloff M Wilson JM Human β-defensin 2 is a salt-sensitive peptide antibiotic expressed in human lung J Clin Invest 1998 102 874 880 9727055
Hiratsuka T Nakazato M Date Y Ashitani J Minematsu T Chino N Matsukura S Identification of human β-defensin-2 in respiratory tract and plasma and its increase in bacterial pneumonia Biochem Biophys Res Commun 1998 249 943 947 9731241 10.1006/bbrc.1998.9239
Singh PK Jia HP Wiles K Hesselberth J Liu L Conway BA Greenberg EP Valore EV Welsh MJ Ganz T Tack BF McCray PB Jr Production of β-defensins by human airway epithelia Proc Natl Acad Sci USA 1998 95 14961 14966 9843998 10.1073/pnas.95.25.14961
Harder J Meyer-Hoffert U Teran LM Schwichtenberg L Bartels J Maune S Schröder JM Mucoid Pseudomonas aeruginosa, TNF-α, and IL-1β, but not IL-6, induce human β-detensin-2 in respiratory epithelia Am J Respir Cell Mol Biol 2000 22 714 721 10837369
García JR Jaumann F Schulz S Krause A Rodríguez-Jiménez J Forssmann U Adermann K Klüver E Vogelmeier C Becker D Hedrich R Forssmann WG Bals R Identification of a novel, multifunctional β-defensin (human β-defensin 3) with specific antimicrobial activity. Its interaction with plasma membranes of Xenopus oocytes and the induction of macrophage chemoattraction Cell Tissue Res 2001 306 257 264 11702237 10.1007/s004410100433
García JR Krause A Schulz S Rodríguez-Jiménez FJ Klüver E Adermann K Forssmann U Frimpong-Boateng A Bals R Forssmann WG Human β-defensin 4: a novel inducible peptide with a specific salt-sensitive spectrum of antimicrobial activity FASEB J 2001 15 1819 1821 11481241
Takemura H Kaku M Kohno S Hirakata Y Tanaka H Yoshida R Tomono K Koga H Wada A Hirayama T Kamihira S Evaluation of susceptibility of gram-positive and -negative bacteria to human defensins by using radial diffusion assay Antimicrob Agents Chemother 1996 40 2280 2284 8891130
Harwig SS Ganz T Lehrer RI Neutrophil defensins: purification, characterization, and antimicrobial testing Methods Enzymol 1994 236 160 172 7968608
Goldman MJ Anderson GM Stolzenberg ED Kari UP Zasloff M Wilson JM Human β-defensin-1 is a salt-sensitive antibiotic in lung that is inactivated in cystic fibrosis Cell 1997 88 553 560 9038346 10.1016/S0092-8674(00)81895-4
Miyata A Kangawa K Toshimori T Hatoh T Matsuo H Molecular forms of atrial natriuretic polypeptides in mammalian tissues and plasma Biochem Biophys Res Commun 1985 129 248 255 4039933 10.1016/0006-291X(85)91429-9
Miyata A Mizuno K Minamino N Matsuo H Regional distribution of adrenorphin in rat brain: comparative study with PH-8P Biochem Biophys Res Commun 1984 120 1030 1036 6732783 10.1016/S0006-291X(84)80210-7
Ishizaka A Watanabe M Yamashita T Ogawa Y Koh H Hasegawa N Nakamura H Asano K Yamaguchi K Kotani M Kotani T Morisaki H Takeda J Kobayashi K Ogawa S New bronchoscopic microsample probe to measure the biochemical constituents in epithelial lining fluid of patients with acute respiratory distress syndrome Crit Care Med 2001 29 896 898 11373491 10.1097/00003246-200104000-00043
Fick RB JrSonoda F Hornick DB Emergence and persistence of Pseudomonas aeruginosa in the cystic fibrosis airway Semin Respir Infect 1992 7 168 178 1475541
Bals R Weiner DJ Meegalla RL Accurso F Wilson JM Salt-independent abnormality of antimicrobial activity in cystic fibrosis airway surface fluid Am J Respir Cell Mol Biol 2001 25 21 25 11472971
Sakamoto N Mukae H Fujii T Ishii H Yoshioka S Kakugawa T Sugiyama K Mizuta Y Kadota J Nakazato M Kohno S Differential effects of α- and β-defensin on cytokine production by cultured human bronchial epithelial cells Am J Physiol Lung Cell Mol Physiol 2005 288 L508 513 15557089 10.1152/ajplung.00076.2004
Hiratsuka T Mukae H Iiboshi H Ashitani J Nabeshima K Minematsu T Chino N Ihi T Kohno S Nakazato M Increased concentrations of human β-defensins in plasma and bronchoalveolar lavage fluid of patients with diffuse panbronchiolitis Thorax 2003 58 425 430 12728165 10.1136/thorax.58.5.425
Diamond G Kaiser V Rhodes J Russell JP Bevins CL Transcriptional regulation of β-defensin gene expression in tracheal epithelial cells Infect Immun 2000 68 113 119 10603376
Kollef MH Gram-negative bacterial resistance: evolving patterns and treatment paradigms Clin Infect Dis 2005 Suppl 2 85 88
Cunha BA Nosocomial pneumonia. Diagnostic and therapeutic considerations Med Clin North Am 2001 85 79 114 11190353 10.1016/S0025-7125(05)70305-9
|
16269090
|
PMC1298335
|
CC BY
|
2021-01-04 16:36:25
|
no
|
Respir Res. 2005 Nov 4; 6(1):130
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-130
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-1311627114410.1186/1465-9921-6-131ResearchReference values for methacholine reactivity (SAPALDIA study) Jayet Pierre-Yves [email protected] Christian [email protected]ünzli Nino [email protected] Jean-Pierre [email protected]ändli Otto [email protected] André Paul [email protected] Roland [email protected] Joel [email protected] Ursula [email protected] Philippe [email protected] team 1 Service of Pulmonology, University Hospital Lausanne, Switzerland2 Institute of Social and Preventive Medicine, University of Basle, Switzerland3 Division of Environmental Health, University of Southern California, USA4 Zürcher Höhenklinik Wald, Switzerland5 Department of Internal Medicine, University Hospital of Basle, Switzerland6 Klinik Barmelweid, Aarau, Switzerland7 Department of Environmental Health, Harvard School of Public Health, USA2005 4 11 2005 6 1 131 131 3 6 2005 4 11 2005 Copyright © 2005 Jayet et al; licensee BioMed Central Ltd.2005Jayet 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 distribution of airway responsiveness in a general population of non-smokers without respiratory symptoms has not been established, limiting its use in clinical and epidemiological practice. We derived reference equations depending on individual characteristics (i.e., sex, age, baseline lung function) for relevant percentiles of the methacholine two-point dose-response slope.
Methods
In a reference sample of 1567 adults of the SAPALDIA cross-sectional survey (1991), defined by excluding subjects with respiratory conditions, responsiveness during methacholine challenge was quantified by calculating the two-point dose-response slope (O'Connor). Weighted L1-regression was used to estimate reference equations for the 95th , 90th , 75th and 50th percentiles of the two-point slope.
Results
Reference equations for the 95th , 90th , 75th and 50th percentiles of the two-point slope were estimated using a model of the form a + b* Age + c* FEV1 + d* (FEV1)2 , where FEV1 corresponds to the pre-test (or baseline) level of FEV1. For the central half of the FEV1 distribution, we used a quadratic model to describe the dependence of methacholine slope on baseline FEV1. For the first and last quartiles of FEV1, a linear relation with FEV1 was assumed (i.e., d was set to 0). Sex was not a predictor term in this model. A negative linear association with slope was found for age. We provide an Excel file allowing calculation of the percentile of methacholine slope of a subject after introducing age – pre-test FEV1 – and results of methacholine challenge of the subject.
Conclusion
The present study provides equations for four relevant percentiles of methacholine two-point slope depending on age and baseline FEV1 as basic predictors in an adult reference population of non-obstructive and non-atopic persons. These equations may help clinicians and epidemiologists to better characterize individual or population airway responsiveness.
==== Body
Background
Description of normal airway responsiveness in a general population is a recent concept [1]. However its use in clinic or in epidemiological studies is limited by the lack of established norms (as percentiles) of the distribution of airway reactivity [2] according to the age, sex and airway caliber of the subjects [3].
The conventional method to measure bronchial responsiveness is to perform a bronchochallenge test where FEV1 is measured at increasing levels of methacholine [4] up to a maximal dose of 2 mg and to evaluate the resulting dose-response curve. Results of the test are usually expressed by an index of responsiveness, the provocating dose (PD20) or concentration (PC20) producing a 20% fall of FEV1. A subject is defined to be hyperreactive if, at any of the methacholine levels tested, his/her FEV1 falls below 80% of the baseline value. In epidemiological studies, however, the concept of hyperreactivity has substantial limitations since the majority of subjects do not reach the critical threshold level so that their degree of responsiveness cannot be defined in terms of a critical dose [5].
In order to obtain a simple index of non-specific airway reactivity for every subject (hyperreactive or normal), O'Connor et al [6] defined the slope of the dose-response curve as the ratio between percent decline of FEV1 (from the post-saline value to the value measured after the final methacholine dose administered) and the final cumulative dose of methacholine. For both asthmatic and normal people this simple dose-response slope provides a good summary of each subject's dose-response curve [7].
The distribution of hyperreactivity or of airway responsiveness in a general population sample has been described in several studies [6,8,9]. For tests performed with methacholine or with histamine, non-specific airway responsiveness shows a unimodal skewed distribution. Although asthmatic subjects tend to lie in the "reactive" tail of the distribution, there is a considerable overlap between the distributions of asthmatic and non-asthmatic subjects. Some authors suggest that this unimodal distribution reflects several overlapping clinical states between normal subjects and symptomatic asthmatics [10]. However, apart from clinical state many individual predictive factors influence the degree of bronchial responsiveness. Whereas age has been investigated in many studies [2,5,11-16], the exact influence of aging on reactivity is still not clear. Its estimated effect appears to depend on whether other possible confounding variables such as baseline lung function or smoking status are simultaneously taken into account. Sex appears to be another important predictive factor: women seem to be more reactive than men [5,12,14-16], but adjusting for possible confounding factors may explain some of this difference. Pre-test FEV1 is considered as a major parameter influencing bronchial responsiveness [11,12,14-16]. However many other potential variables appear to play a role, such as smoking status [11,13-15,17], geographic characteristics [2,11], atopic status [14-16], occupational exposure to inhalation irritants [18], presence of chronic respiratory conditions or prior asthma [19], or recent upper airway infection [19]. These findings indicate that bronchial responsiveness, as described by PD20, PC20 or dose-response slope, may be influenced by a wide range of factors that in turn, may substantially affect its interpretation.
Data from the asymptomatic never smoking participants of the SAPALDIA cross-sectional study (1991) have already been used by Brändli [20,21] to derive reference equations for mean values and lower limits of normal of spirometric lung function. In this paper we use data of the methacholine challenge test from a selected sample of "normal" participants of the SAPALDIA sample to establish reference equations for some important percentiles of methacholine slope depending on important individual characteristics (i.e., sex, age and baseline lung function).
Methods
SAPALDIA (Swiss Study on Air Pollution and Lung Diseases in Adults) is a multicenter study designed to investigate the relationship between exposure to air pollutants and respiratory symptoms or diseases. The eight study areas participating in the project were chosen to represent the variety of environmental conditions found in Switzerland concerning geography, climate, degree of urbanisation and air pollution. The study was approved by the institutional review board for human investigations of the different areas. In the cross-sectional part performed in 1991, a random sample of adults 18 to 60 years old were invited to take part in the study. 9651 subjects were included in the study, representing 59% of all eligible subjects. Health assessment included a detailed questionnaire, computer-based spirometric tests, methacholine bronchial challenge and skin allergy tests to 8 inhalative allergens. Details on the methodology of these assessments are given elsewhere [22].
Spirometry measurements were done using a Sensor-Medics 2200 pulmonary function system SP (Bilthoven, The Netherlands). This is an open sensor device which meets the quality criteria of the American Thoracic Society. The Sensor-Medics spirometer displays an error code after each forced expiration to inform the technician about the acceptability of the maneuver and the reproducibility between the trials using the standard quality criteria defined by the American Thoracic Society [23]. The trials were recorded electronically on a personal computer as they were done. Calibration was done at least once daily, using a 3-liter syringe. All the spirometry technicians were trained together according to a standardized protocol and were tested on volunteers [24]. Each of the following criteria was sufficient for excluding a subject from the methacholine test: a) a baseline FEV1 / FVC ratio of less than 80% of the ECCS-norm [25], b) a baseline FEV1 of less than 70% of the ECCS-norm, c) pregnancy or breast feeding, d) a myocardial infarction within the three months preceding the SAPALDIA examination, e) severe heart failure under treatment, f) treatment with β-blockers including eye-drops, g) refusal to participate. These exclusions and the requirement of having complete and valid data on lung function and bronchial responsiveness reduced the sample size to 6942. Non-specific bronchial reactivity was tested using methacholine chloride (Provocholine® , Roche, Nutley, New Jersey, USA) prepared in 0.39, 1.56, 6.25, and 25.0 mg/ml solutions in a phosphate buffer without phenol. Increasing concentrations of methacholine were administered through an aerosol dosimeter (Mefar MB3, Bovezzo, Italy) up to a cumulative dose of 2 mg (8.37 μmol). With each inhalation, approximately 0.01 ml was delivered to the subject. The first dose inhaled by the subject was a saline control. The schedule was then 4 inhalations of methacholine of 0.39 mg/ml (total dose 0.016 mg), 3 inhalations of 1.56 mg/ml (cumulative dose 0.062 mg), 3 inhalations of 6.25 mg/ml (cumulative dose 0.25 mg), 3 inhalations of 25 mg/ml (cumulative dose 1 mg), and 4 inhalations of 25.0 mg/ml (total cumulative dose 2 mg). If a decrease in FEV1 of more than 10% from the baseline level occurred at any intermediate point of the test, smaller increments (i.e., halving the doses and doubling the number of inhalations) were introduced. Testing continued until the final dose of 2 mg was administered or until FEV1 had fallen by 20% or more. Under this protocol the cumulative doses of methacholine converted in micromoles at each level were 0, 0.065, 0.26, 1.05, 4.18, and 8.37. At each level, the subjects were asked to inhale slowly from their functional residual capacity up to their vital capacity. The subjects were instructed to keep a full inspiration for 4 seconds before a slow normal exhalation. After each dose level of methacholine, 2 forced expiratory maneuvers were performed at 1 and 2 minutes after the end of the methacholine inhalation and the best of the two FEV1 values was considered [26].
Methacholine responsiveness was quantified by calculating the two-point dose-response slope as defined by O'Connor [6]. Slope is defined as the percentage of decline of FEV1 from the post-saline value to the value measured after the final methacholine dose administered divided by the final cumulative methacholine dose administered. Figure 1 provides a schematic diagram illustrating the relationship between the two-point dose response slope (expressed in % decline of FEV1 divided by the final cumulative methacholine dose administered) and PD20 (provocating dose in mg producing a 20% fall of FEV1). The figure demonstrates that higher reactivity is indicated by a higher value of slope. The horizontal line drawn at a slope of 2.39% decrease/μmol represents the threshold commonly used to define bronchial hyperreactivity (20% decrease of FEV1 after a cumulative methacholine dose of ≤ 2 mg).
Figure 1 Graphic representation of the relationship between the two-point dose response slope and PD20. This figure shows the relationship between the two-point dose response slope and PD20. The horizontal line drawn at a slope of 2.39% decrease/μmol represents the "cut-off" threshold commonly used to define bronchial hyperreactivity (20% decrease of FEV1 after a cumulative methacholine dose of ≤ 2 mg).
Of the participants who performed the methacholine test, only 1567 were included in the reference sample after applying the following exclusion criteria: a) current or former smoking: (i.e., having smoked 20 or more packs of cigarettes or more than 360 g of tobacco); b) a prior diagnosis of asthma or report of symptoms related to asthma or bronchitis (i.e., wheezing in the last 12 months and/or shortness of breath at rest in the last 12 months and/or nocturnal attacks of shortness of breath in the last 12 months and/or attacks of asthma in the last 12 months and/or current asthma medication and/or cough or phlegm on most days of at least three months of the year); c) atopy: defined by the presence of at least one positive reaction to the eight inhalant allergens tested in a skin prick test (subjects with missing results in this test were also excluded); d) recent respiratory infection (i.e., anamnesis of a respiratory infection within three weeks prior to the methacholine test).
Weighted L1-regression was used to estimate percentile functions. This method consists of finding the model parameters which minimize a given weighted sum of absolute residual values. For instance, estimating the model for the 75th percentile is achieved by assigning the absolute values of positive residuals three times the weight of the absolute values of negative residuals. In general, if the m-th percentile is to be estimated, absolute values of positive residuals are given a weight proportional to 1/(100-m) and absolute values of negative residuals a weight proportional to 1/m. Details of this method are described elsewhere [27-29]. To test whether a given model could be improved by adding an additional predictor term, we defined a dichotomous variable U taking the value 1 for observations with methacholine slopes exceeding the respective percentile estimates and the value 0 for all other observations. A logistic regression model incorporating the covariate part of the underlying percentile model along with the additional predictor term was then computed. If the additional predictor term was significant then it was added to the percentile model. These methods have already been applied in a similar context to estimate percentile equations for lung function [21].
We tested the performance of this approach in identifying asthmatics using the 90th percentile of slope as threshold in subjects who answered positively to the double question: "Have you ever had asthma? Was this confirmed by a doctor?" and performed methacholine test (i.e. fulfilled initial inclusion criteria mentioned above). For both men and women of this subsample, the percentage of subjects whose slopes exceeded this threshold was compared to the percentage of subjects usually defined as hyperreactive (i.e., with a positive response to the methacholine test based on a fall of 20% of FEV1 during the test).
Results
The different stages leading to the selection of the reference sample are described in Table 1. Only 1567 persons, representing 20.9% of all participants of the methacholine bronchial challenge fulfilled all criteria. The major part of subjects excluded were current or former smokers.
Table 1 Definition of the study sample, SAPALDIA cross-sectional study, 1991
Men Women Total
Whole SAPALDIA sample 4743 (100%) 4908 (100%) 9651 (100%)
- subjects with incomplete data on lung function and bronchial responsiveness* 3446 (72.7%) 3496 (71.2%) 6942 (71.9%)
- current or former smokers 1278 (26.9%) 1770 (36.1%) 3048 (31.6%)
- subjects with a prior diagnosis of asthma or symptoms related to asthma or bronchitis 1052 (22.2%) 1428 (29.1%) 2480 (25.7%)
- subjects with a positive or missing skin test 733 (15.5%) 1107 (22.6%) 1840 (19.1%)
- subjects with recent respiratory infection 612 (12.9%) 955 (19.5%) 1567 (16.2%)
Total of the study sample 612 (12.9%) 955 (19.5%) 1567 (16.2%)
* exclusion criteria from methacholine testing were FEV1/FVC ratio less than 80% of the ECCS-norm, FEV1 of less than 70% of the ECCS-norm, incomplete data on lung function, pregnancy or breast feeding, a myocardial infarction within the 3 months preceding the examination, being treated for severe heart failure, being treated with β-blockers including eye-drops, or refusal to participate; subjects with incomplete data on methacholine test were also excluded.
Characteristics of the study population are provided in Table 2. It included a higher proportion of women (60.9%) than in the whole methacholine test sample (49.4%), explained by their lower prevalence of current or former smoking. A scatter plot of methacholine slope vs. baseline FEV1 (all subjects) is given in Figure 2.
Table 2 Distribution of basic predictor variables in the reference sample, SAPALDIA cross-sectional study, 1991
Men (n = 612) Women (n = 955) Entire reference sample (n = 1567)
<30 yrs 31.7% 20.8% 25.1%
30–40 yrs 24.8% 20.5% 22.2%
40–50 yrs 25.5% 27.9% 26.9%
≥50 yrs 18.0% 30.8% 25.8%
Height, mean (SD) 176.1 (6.7) 163.5 (6.5) 168.4 (9.0)
Weight, mean (SD) 75.2 (10.2) 61.6 (10.6) 66.9 (12.4)
FEV1, mean (SD) 4.33 (0.67) 3.10 (0.54) 3.58 (0.84)
PD20 prevalence* 4.4% 14.6% 10.6%
* PD20 prevalence denotes prevalence of subjects with a fall of 20% or more in FEV1 during the methacholine test
Figure 2 Scatter plot of methacholine slope vs. pretest level of FEV1 for our study sample (n = 1567) (excluding 5 observations with slopes >30%/μmol)
Prediction equations of 95th , 90th , 75th and 50th percentiles of the two-point slope are given in Table 3. The corresponding curves for 40 years old subjects are represented in Figure 3. Prediction equations were derived involving age and pre-test (or baseline) FEV1. Between the lower and upper quartile of FEV1, these models are of the form: a + b* Age + c* FEV1 + d* FEV12 , whereas no quadratic term in FEV1 is used below the 1st and above the 3rd quartile. We thus used natural quadratic splines with knots at the lower and upper quartiles of FEV1 to describe the dependency of percentiles of methacholine slope on baseline FEV1. Therefore, up to the first quartile of FEV1, each percentile curve of slope for a given age is described by a straight line. Another straight line describes the percentile curve for FEV1-values above the upper quartile. These two straight line segments are connected by a parabola segment in such a way that the transition between the different pieces is smooth. Although the coefficients a and c have to vary between the three intervals, the smoothness requirement imposes linear restrictions on them. On the other hand, the coefficient b has the same value everywhere, since the association between slope and age appeared to be approximately linear for all percentiles considered. Consequently, the curves for figure 3 would have to be shifted downward and upward for ages higher and lower than 40 years, respectively. The model shows that, with lower pre-test values of FEV1, level and spread of the percentiles increases. A horizontal line drawn at y= 2.39% decrease/μmol represents the threshold commonly used to define bronchial hyperreactivity (20% decrease of FEV1 after a cumulative methacholine dose of ≤ 8.37 μmol). A higher proportion of subjects belong to this "hyperreactive" category at lower values of FEV1 or lower values of age. Consequently a higher proportion of women are defined as "hyperreactive" (Table 3). We provide an additional Excel file allowing calculation of the percentile of methacholine slope of a subject after introducing his/her age, pre-test FEV1, and results of methacholine challenge (i.e. methacholine total cumulative dose and percentage of FEV1 decline at this total cumulative dose) (Additional file 1).
Table 3 Percentiles of methacholine slope* among men and women of the reference sample, SAPALDIA cross-sectional study, 1991
minimum P5 P10 P25 P50 P75 P90 P95 maximum
men (n = 612) -2.81 -0.55 -0.25 0.13 0.48 0.98 1.60 2.25 40.5
women (n = 955) -3.69 -0.13 0.06 0.41 0.90 1.67 3.25 5.72 78.5
entire reference sample (n = 1567) -3.69 -0.30 -0.07 0.26 0.72 1.41 2.40 4.85 78.5
* final %decrease in FEV1 from baseline divided by highest dose of methacholine administered
Figure 3 Percentiles of methacholine slope as a function of pretest level of FEV1 (among persons of reference sample aged 40 years). This figure shows the percentiles of methacholine slope as a function of pretest level of FEV1 (among persons of reference sample aged 40 years). The horizontal line defines the threshold between "hyperreactive" and "normal" subjects as defined by a 20% fall of FEV1 from the baseline value before or at the maximal methacholine dose. The scale of pretest level of FEV1 extends from below the 1 st percentile to above the 99 th percentile of pretest level of FEV1 in our reference sample.
Among subjects with physician-diagnosed asthma (n = 411), the percentage of subjects with a fall of 20% or more during the methacholine test was significantly higher in women than in men (58.8% vs. 43.8%, p < 0.01). In the same population, percentages of subjects above the 90th percentile of methacholine slope from the model including FEV1 did not differ between both sexes (51.0% vs. 51.2%, p = 0.98).
Discussion
Previous studies have demonstrated that non-specific bronchial responsiveness to methacholine may be influenced by a number of factors [2,5,11-19]. On the basis of a review of the literature we excluded subjects presenting characteristics that may influence bronchial reactivity in a "non-physiological" way from our study population: smokers and former smokers, anamnestic asthmatic or bronchitic subjects, atopics, and persons who reported a recent respiratory infection. Moreover, the methacholine challenge was not performed in subjects with spirometric evidence of airway obstruction. Our preliminary analysis showed that among the potential predictor variables considered (i.e., sex, age, height, weight, FEV1, FVC, FEV1 / FVC, FEF25–75%, FEF25–75% / FVC), sex, age and either FEV1, FEF25–75%, or FEF25–75% / FVC had the strongest explanatory power (results not shown). Using pre-test FEV1 in addition to basic variables (sex, age, and height) improves prediction equations for methacholine reactivity, probably due to multiple factors. In subjects with restrictive syndrome, whatever the etiology, airway calibre is better described by absolute values of FEV1 than by the height or weight of subjects. Moreover, the underlying mechanisms of bronchial responsiveness to a pharmacological agent are complex and multifactorial. Several studies suggested that, apart from lung size, other important determinants of non specific bronchial hyperresponsiveness are airway geometry and properties of smooth muscles. Wassmer [15] showed in an adult German population that BHR (defined by a fall in 10% or 20% of FEV1 in methacholine challenge) or bronchial responsiveness (described by dose-response slope) is most strongly predicted by lung function parameters. In a study analyzing hyperreactivity in a large random adult population, Britton [16] showed that FEV1, FEV1 %predicted and FEV1 / FVC were strongly and independently related to BHR, identifying with varying degrees of overlap separate groups of individuals at increased risk of hyperreactivity. In our analysis, however, FEV1 / FVC was not significantly associated with methacholine slope. This may be explained by the exclusion of obstructive and atopic subjects.
An independent significant effect of age on bronchial methacholine dose-response slope is seen in our population study even after correction for FEV1, showing a negative cross-sectional association between slope and age after adjustment for differences in FEV1. This is an interesting result per se, given that an independent effect of age on BHR has not been consistently documented in the literature [2,5,11-13,16].
Our percentile equations may be used in epidemiological studies to define more valid individual measures of responsiveness (i.e. severity) because they incorporate inherent confounding factors such as age and pre-test airway calibre. Moreover, the equations may enable clinicians to assess the degree of bronchial responsiveness in their patients with greater validity. We provide a simple Excel file enabling the computation of the percentile of a subject's bronchial responsiveness provided that this value lies between the 50th and the 95th percentile of the distribution in our adult reference population.
In clinical practice, methacholine challenge is currently used primarily to exclude asthma in atypical situations, being recognized as a useful but imprecise test. Using the 90th percentile as a "cut-off" level for identifying asthmatics in our sample of subjects with self-reported physician diagnosed asthma provided a sensitivity of 51.1% which did not differ between sexes; this percentage was very similar to the percentage of subjects with a fall of 20% or more during the methacholine test in the same population (50.9%), where a significant difference was, however, present between sexes (58.8% in women vs. 43.8% in men). We therefore hypothesize that our equations and index provide a more valid individual marker of the clinical severity, enabling better characterization and quantification of bronchial responsiveness. While receiver operator characteristic (ROC) studies would be needed to evaluate the best "cut-off" percentile for asthma diagnosis, using the 90 th percentile yielded the same sensitivity in our subsample of asthmatics as the PD20 criterion in a similar study population of subjects with self-reported physician diagnosed asthma [30].
Conclusion
The present study provides equations for four relevant percentiles of methacholine slope (defined according to O'Connor) depending on the age and baseline FEV1 in an adult reference population of non-obstructive and non-atopic persons. In addition to the fact that such models may help to better understand the underlying mechanisms of BHR, they may be of use in future epidemiological studies to better identify subjects whose bronchial hyperreactivity is caused by extrinsic factors or by obstructive or atopic conditions. It may be of interest to both clinicians and epidemiologists that the sensitivity of our method in identifying subjects with a doctor's diagnosis of asthma is the same in men and women whereas the traditional method based on PD20 has a lower sensitivity in men. More generally, our equations may help physicians to better characterize and follow bronchial responsiveness of individual patients, based on simple predictive factors.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
PYJ, CS and PL conducted the analyses and drafted the article. CS, NK, JPZ, OB, APP, RK, JS, UAL and PL contributed to the design of the study, the acquisition of data and the interpretation of data. All authors contributed to the conception of the research question, made important intellectual contributions during the drafting process and have given approval for the final version.
Table 4 Estimated equations of the 95th , 90th , 75th and 50th percentiles of methacholine slope given age and pretest level of FEV1 (litres), SAPALDIA cross-sectional study, 1991
Slope95 = 34.70 - 0.0167 age - 9.001 FEV1 + 0 FEV12 (FEV1≤2.93)
= 65.69 - 0.0167 age - 30.152 FEV1 + 3.6095 FEV12 (2.93<FEV1≤4.14)
= 3.82 - 0.0167 age - 0.266 FEV1 + 0 FEV12 (FEV1>4.14)
Slope90 = 14.81 - 0.0160 age - 3.523 FEV1 + 0 FEV12 (FEV1≤2.93)
= 26.48 - 0.0160 age - 11.483 FEV1 + 1.3584 FEV12 (2.93<FEV1≤4.14)
= 3.19 - 0.0160 age - 0.236 FEV1 + 0 FEV12 (FEV1>4.14)
Slope75 = 4.90 - 0.0056 age - 0.997 FEV1 + 0 FEV12 (FEV1≤2.93)
= 7.53 - 0.0056 age - 2.796 FEV1 + 0.3071 FEV12 (2.93<FEV1≤4.14)
= 2.27 - 0.0056 age - 0.253 FEV1 + 0 FEV12 (FEV1>4.14)
Slope50 = 3.03 - 0.0039 age - 0.642 FEV1 + 0 FEV12 (FEV1≤2.93)
= 4.77 - 0.0039 age - 1.828 FEV1 + 0.2025 FEV12 (2.93<FEV1≤4.14)
= 1.30 - 0.0039 age - 0.152 FEV1 + 0 FEV12 (FEV1>4.14)
Supplementary Material
Additional File 1
Calculation of percentiles of methacholine slope as a function of pre-test FEV1 and age. This additional Excel file allows calculation of the percentile of methacholine slope of a subject after introducing his/her age, pre-test FEV1, and results of methacholine challenge (i.e. methacholine total cumulative dose and percentage of FEV1 decline at this total cumulative dose).
Click here for file
Acknowledgements
Supported by grants from the National Research Program 26A (Grant No 4026–28099) of the Swiss National Science Foundation and from the Swiss Federal Office of Education and Science. SAPALDIA Basle is part of the European Respiratory Health Survey.
The authors wish to thank the SAPALDIA team fieldworkers at Aarau, Basle, Davos, Geneva, Lugano, Montana, Payerne and Wald. They are grateful to the collaborators of the central team at Basle (Institute of Social and Preventive Medicine), Lausanne (direction of the project) and Zurich (Allergology Unit, Department of Dermatology). They would like to thank Dr Sara Downs for commenting on the manuscript. The authors thank the authorities of the participating cantons of Aarau, Basle, Geneva, Vaud, Valais, Zurich, Ticino and Grisons for their logistic and financial support.
The SAPALDIA team includes: Ph. Leuenberger (p) (Study director), U. Ackermann-Liebrich (e) (Programme director), P. Alean (am), K. Blaser (a), G. Bolognini (p), J.P. Bongard (p), O. Brändli (p), P. Braun (p), C. Bron (l), M. Brutsche (l), C. Defila (m), G. Domenighetti (p), S. Elsasser (l), L. Grize (s), P. Guldimann (l), P. Hufschmid (l), W. Karrer (p), H. Keller-Wossidlo (o), R. Keller (p), N. Künzli (e), J.C. Lüthi (l), B.W. Martin (e), T. Medici (p), Ch. Monn (am), A.G. Peeters (pa), A.P. Perruchoud (p), A. Radaelli (l), Ch. Schindler (s), J. Schwartz (s), G. Solari (p), M.H. Schöni (p), J.M. Tschopp (p), B. Villiger (p), B. Wüthrich (a), J.P. Zellweger (p), E. Zemp (e). (a) : allergology; (am) : air pollution monitoring; (e) : epidemiology; (l) : local assistant; (m) : meteorology; (o) : occupational medicine; (p) : pulmonology; (pa) : palynology; (s) : statistics.
==== Refs
Rogers DF O'Connor BJ Airway hyperresponsiveness: relation to asthma and inflammation? Thorax 1993 48 11 1095 1096 8296249
Anto JM Methods to assess and quantify BHR (bronchial hyperresponsiveness) in epidemiological studies Clin Exp Allergy 1998 28 Suppl 1 13 4; discussion 32-6 9641584 10.1046/j.1365-2222.1998.0280s1013.x
Savoy J Junod AF [Is non-allergenic bronchial hyperreactivity a good diagnostic test for asthma?] Rev Mal Respir 1994 11 2 201 207 8202610
Cockroft DW Airway responsiveness Asthma 1997 Philadelphia , Lippincott-Raven Publishers 1253 1266
Neukirch F Cartier A [Measure of bronchial hyperreactivity in epidemiology] Rev Mal Respir 1994 11 2 101 109 8202601
O'Connor G Sparrow D Taylor D Segal M Weiss S Analysis of dose-response curves to methacholine. An approach suitable for population studies Am Rev Respir Dis 1987 136 6 1412 1417 3318599
Banik AN Holgate ST Problems and progress in measuring methacholine bronchial reactivity Clin Exp Allergy 1998 28 Suppl 1 15 9; discussion 32-6 9641585 10.1046/j.1365-2222.1998.0280s1015.x
Cockcroft DW Berscheid BA Murdock KY Unimodal distribution of bronchial responsiveness to inhaled histamine in a random human population Chest 1983 83 5 751 754 6839815
Cerveri I Bruschi C Zoia MC Zanon P Maccarini L Grassi M Rampulla C Distribution of bronchial nonspecific reactivity in the general population Chest 1988 93 1 26 30 3275530
Lockhart A Malo JL [Bronchial hyperreactivity. Definitions and history] Rev Mal Respir 1994 11 2 85 92 8202613
Bakke PS Baste V Gulsvik A Bronchial responsiveness in a Norwegian community Am Rev Respir Dis 1991 143 2 317 322 1990946
Trigg CJ Bennett JB Tooley M Sibbald B D'Souza MF Davies RJ A general practice based survey of bronchial hyperresponsiveness and its relation to symptoms, sex, age, atopy, and smoking Thorax 1990 45 11 866 872 2256016
Burney PG Britton JR Chinn S Tattersfield AE Papacosta AO Kelson MC Anderson F Corfield DR Descriptive epidemiology of bronchial reactivity in an adult population: results from a community study Thorax 1987 42 1 38 44 3497466
Schwartz J Schindler C Zemp E Perruchoud AP Zellweger JP Wuthrich B Leuenberger P Ackermann-Liebrich U Predictors of methacholine responsiveness in a general population Chest 2002 122 3 812 820 12226018 10.1378/chest.122.3.812
Wassmer G Jorres RA Heinrich J Wjst M Reitmeir P Wichmann HE The association between baseline lung function and bronchial responsiveness to methacholine Eur J Med Res 1997 2 2 47 54 9085014
Britton J Pavord I Richards K Knox A Wisniewski A Wahedna I Kinnear W Tattersfield A Weiss S Factors influencing the occurrence of airway hyperreactivity in the general population: the importance of atopy and airway calibre Eur Respir J 1994 7 5 881 887 8050544
O'Connor GT Sparrow D Segal MR Weiss ST Smoking, atopy, and methacholine airway responsiveness among middle-aged and elderly men. The Normative Aging Study Am Rev Respir Dis 1989 140 6 1520 1526 2690701
Leuenberger P Schindler C Schwartz J Ackermann-Liebrich U Tara D Perruchoud AP Wuthrich B Zellweger JP Blaser K Bolognini G Bongard JP Brandli O Domenighetti G Elsasser S Grize L Karrer W Keller R Kunzli N Medici T Schoni MH Solari G Tschopp JM Villiger B Zemp E Occupational exposure to inhalative irritants and methacholine responsiveness Scand J Work Environ Health 2000 26 2 146 152 10817380
Zellweger JP Fitting JW [Ventilatory function and bronchial responsiveness in army recruits] Schweiz Med Wochenschr 1990 120 40 1466 1472 2218468
Brandli O Schindler C Kunzli N Keller R Perruchoud AP Lung function in healthy never smoking adults: reference values and lower limits of normal of a Swiss population Thorax 1996 51 3 277 283 8779131
Brandli O Schindler C Leuenberger PH Baur X Degens P Kunzli N Keller R Perruchoud AP Re-estimated equations for 5th percentiles of lung function variables Thorax 2000 55 2 173 174 10702094 10.1136/thorax.55.2.172a
Martin BW Ackermann-Liebrich U Leuenberger P Kunzli N Stutz EZ Keller R Zellweger JP Wuthrich B Monn C Blaser K Bolognini G Bongard JP Brandli O Braun P Defila C Domenighetti G Grize L Karrer W Keller-Wossidlo H Medici TC Peeters A Perruchoud AP Schindler C Schoeni MH Villiger B SAPALDIA: methods and participation in the cross-sectional part of the Swiss Study on Air Pollution and Lung Diseases in Adults Soz Praventivmed 1997 42 2 67 84 9151378 10.1007/BF01318136
Standardization of spirometry--1987 update. Statement of the American Thoracic Society Am Rev Respir Dis 1987 136 5 1285 1298 3674589
Kunzli N Ackermann-Liebrich U Keller R Perruchoud AP Schindler C Variability of FVC and FEV1 due to technician, team, device and subject in an eight centre study: three quality control studies in SAPALDIA. Swiss Study on Air Pollution and Lung Disease in Adults Eur Respir J 1995 8 3 371 376 7789479 10.1183/09031936.95.08030371
Quadrelli S Roncoroni A Montiel G Assessment of respiratory function: influence of spirometry reference values and normality criteria selection Respir Med 1999 93 8 523 535 10542984 10.1016/S0954-6111(99)90150-6
Burney PG Luczynska C Chinn S Jarvis D The European Community Respiratory Health Survey Eur Respir J 1994 7 5 954 960 8050554
Koenker R Basset G Regression quantiles Econometrica 1978 46 33 50
Hastie TJ Tibshirani RJ Generalized additive models Monographs on statistics and applied probability 43 1990 Chapman & Hall/CRC
Yu K Lu.Z. Stander J Quantile regression:applications and current research areas The Statistician 2003 52 331 350
Lewis SA Weiss ST Britton JR Airway responsiveness and peak flow variability in the diagnosis of asthma for epidemiological studies Eur Respir J 2001 18 6 921 927 11829097 10.1183/09031936.01.00211801
|
16271144
|
PMC1298336
|
CC BY
|
2021-01-04 16:36:26
|
no
|
Respir Res. 2005 Nov 4; 6(1):131
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-131
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-1321627115510.1186/1465-9921-6-132ResearchDownregulation of peroxisome proliferator-activated receptors (PPARs) in nasal polyposis Cardell Lars-Olaf [email protected]ägge Magnus [email protected] Rolf [email protected] Mikael [email protected] Laboratory of Clinical and Experimental Allergy Research, Department of Otorhinolaryngology, Lund University, Malmö University Hospital, Malmö, Sweden2005 7 11 2005 6 1 132 132 3 4 2005 7 11 2005 Copyright © 2005 Cardell et al; licensee BioMed Central Ltd.2005Cardell 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
Peroxisome proliferator-activated receptor (PPAR) α, βδ and γ are nuclear receptors activated by fatty acid metabolites. An anti-inflammatory role for these receptors in airway inflammation has been suggested.
Methods
Nasal biopsies were obtained from 10 healthy volunteers and 10 patients with symptomatic allergic rhinitis. Nasal polyps were obtained from 22 patients, before and after 4 weeks of local steroid treatment (fluticasone). Real-time RT-PCR was used for mRNA quantification and immunohistochemistry for protein localization and quantification.
Results
mRNA expression of PPARα, PPARβδ, PPARγ was found in all specimens. No differences in the expression of PPARs were obtained in nasal biopsies from patients with allergic rhinitis and healthy volunteers. Nasal polyps exhibited lower levels of PPARα and PPARγ than normal nasal mucosa and these levels were, for PPARγ, further reduced following steroid treatment. PPARγ immunoreactivity was detected in the epithelium, but also found in smooth muscle of blood vessels, glandular acini and inflammatory cells. Quantitative evaluation of the epithelial immunostaining revealed no differences between nasal biopsies from patients with allergic rhinitis and healthy volunteers. In polyps, the PPARγ immunoreactivity was lower than in nasal mucosa and further decreased after steroid treatment.
Conclusion
The down-regulation of PPARγ, in nasal polyposis but not in turbinates during symptomatic seasonal rhinitis, suggests that PPARγ might be of importance in long standing inflammations.
==== Body
Background
Seasonal allergic rhinitis is the result of an immunologically mediated hypersensitivity reaction of the nasal mucosa, initiated by exposure to specific allergens. The reaction is characterized by the infiltration of various inflammatory cells, like eosinophils, neutrophils, basophils, monocytes, and lymphocytes. When the season is over, symptoms and signs of inflammation disappear and the nasal mucosa gradually returns to a situation close to that in healthy non-allergic subjects [1-4]. Nasal polyposis is another inflammatory disorder of the upper airways and is, like allergic rhinitis, related to an infiltration of inflammatory cells [5]. The allergic inflammation is driven by a network of pro-inflammatory cytokines [6] and several of these mediators also appear to be involved in nasal polyposis [7]. In addition, there is an emerging concept that anti-inflammatory pathways affect the outcome of inflammatory diseases, including those in the airways [8]. Peroxisome proliferator-activated receptors (PPARs) have been suggested to play such an anti-inflammatory role [9-11].
Peroxisome proliferator activated receptors (PPARs) are DNA-binding nuclear hormone receptors that are up-regulated in response to high fat diets [12]. PPARs are structurally related to the type II nuclear receptors, including the thyroid hormone receptors and occur intracellularly both in the cytosol and in the nucleus. Although today there are no proven high-affinity pathways for endogenous ligands in-vivo, all PPARs are activated by fatty acids. To date, three mammalian PPAR subtypes have been identified, referred to as α, β or δ (here named βδ), and γ, which are encoded by separate genes [13,14]. They share a 60–80% homology in their binding domains and have subtle differences in ligand specificity in that e.g. eocosanoid products of the lipooxygenase pathway such as leukotriene B4 and 8-S-hydroxytetraenoic acid (8-S-HETE) activates PPARα [15], prostaglandin (PG) I2 activates PPARβδ and, 15-HETE and the PGD2 derivative, 15-deoxy-Δ12,14-PGJ2, activates PPARγ [16,17]. PPARs are generally expressed at a high level in adipose tissue, and play a prominent role in several physiological processes including the control of lipid and lipoprotein metabolism, and glucose homoeostasis [18]. PPARs might also be involved in cardiovascular disease [19] and cancer [20]. A role for PPARγ in allergic conditions as well as asthma has been suggested, but remains controversial [21]. In murine models of human asthma, it has been demonstrated that local administration of PPARγ agonists decreases serum levels of IgE, and have beneficial effects on airway hyperresponsiveness and lung eosinophilia [22,23]. One study of PPARγ in samples from inflamed human airways has demonstrated that the immunoreactivity for PPARγ is augmented in the bronchial submucosa, the airway epithelium and the smooth muscle cells of asthmatics compared to healthy subjects [24]. The increasing number of reports describing an anti-inflammatory role for PPARs (especially PPARγ) in various disease models [9-11], have prompted us to examine the expression and localization of PPARs in allergic rhinitis and nasal polyposis using nasal polyps as model to investigate the effects of local steroid treatment.
Methods
Subjects
The study included 10 patients (five women) with symptomatic birch or grass pollen-induced allergic rhinitis and 10 healthy volunteers (four women), serving as controls. The median age of the patients and controls was 36 (28–50) years and 32 (22–46) years, respectively. In addition, 22 patients with bilateral nasal polyposis (four women) were included before or after treatment with steroids (median age 52 [22–79] years). In 7 of these patients, two sets of polyps were obtained, one before and one after steroid treatment (fluticasone, see below).
The diagnosis of birch and grass pollen induced allergic rhinitis was based on a positive history of intermittent allergic rhinitis and positive skin prick tests to birch and/or timothy. Exclusion criterias included a history of perennial symptoms, upper airway infection during the time of visit, positive skin prick tests to house dust mite (Dermatophagoides Pteronyssimus and D. Farinae) and molds (Cladosporium and Alternaria), 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 skin prick tests to a panel of allergens including birch, timothy, mugwort, house dust mite horse, dog, cat and molds.
Nasal polyposis was identified on the basis of clinical symptoms (nasal obstruction and anosmia) and the visualization of polyps by anterior rhinoscopy. A complete ear, nose-, and throat examination was performed before inclusion. Patients with cystic fibrosis and ciliary dyskinesia were excluded from the study along with subjects with a history of concurrent purulent nasal infection in the six weeks before the study or any kind of nasal surgery during the last year. None of the patients suffered from asthma that required continuous medication.
All patients were recruited through physician referrals and the healthy volunteers were recruited via advertisements in the local press. The study was approved by the Ethics Committee of the Medical Faculty, Lund University.
Study design for obtaining nasal biopsies
The patients were seen either during the birch (5 patients) or during the grass pollen season (5 patients). They were included when they had experienced substantial symptoms of rhinoconjunctivitis (itchy nose and eyes, sneezing, nasal secretion and nasal blockage) during 3–5 consecutive days. All patients were seen within 5–10 days after the first appearance of symptoms. A local pollen count confirmed the presence of relevant pollen during this period. During the visit patients were asked to evaluate their nasal symptoms, itching/sneezing, secretion and blockage, individually using an arbitrary scale from 0 to 3 (0 = no, 1 = mild, 2 = moderate, 3 = severe symptoms) A total nasal symptom score was then calculated by addition of the three scores. Anterior rhinoscopy was performed and oedema and secretion in each nostril were scored from 0–2 (0 = no, 1 = mild, 2 = severe). Total oedema/secretion scores were then computed by adding the scores for each sign from each nostril. All participants had at least 6 in total symptom score and at least in 5 total oedema/secretion score. The healthy volunteers were seen during the same period.
Biopsies were taken from the inferior turbinate after topical application of local anesthesia containing lidocainhydrochloride-nafazoline for 20 minutes. The specimens for mRNA extraction were immediately placed in RNA-later (QIAGEN) and frozen. For immunohistochemistry, specimens were immersed in an ice-cold fixative solution composed of 2% formaldehyde and 0.2% picric acid, buffered to pH 7.2 with 0.1 M phosphate buffer.
Study design for obtaining nasal polyps
Patients with nasal polyposis participated in the study during the autumn/winter (outside the pollen season). In patients supplying polyps before treatment all steroids (systemic, inhaled and intranasal) were withheld during a minimum of six weeks before the study (three patients were steroid naïve). All patients supplying polyps after treatment were seen by the surgeon and medication with fluticasone, 200 μg twice a day initiated. After four weeks on this course one set of polyps was removed. Polyps were removed using topical application of local anesthesia containing lidocainhydrochloride-nafazoline for about 20 minutes. In patients providing two sets of polyps a washing out period of two weeks were used after the first set was removed and the Fluticasone medication started.
RNA extraction and RT-PCR
RNA was extracted from homogenized biopsies using the RNeasy Mini Kit (QIAGEN GmbH), according to the supplier's protocol including an optional DNaseI (Qiagen) treatment. 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 h.
Quantitative real time-PCR
Quantitative real-time PCR assays were performed using the Smartcyckler II detection system (Cephied, USA). Intron over-spanning oligonucleotide primers for detection of PPARα, PPARβδ, PPARγ and β-actin were designed using Primer Express® 2.0 software (Applied Biosystem, USA) and synthesized by DNA Technology A/S (Aarhus, Denmark, table 1). PCR was performed using QuantiTect SYBR® Green RT-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 a 2°C decrease of the first 6 cycles). Standard curves for the PCR reactions were prepared using half 10log dilutions of PCR product generated from target cDNA. Specific PCR products were analysed by running melting curve and visualized by agarose electrophoresis.
Table 1 Intron over-spanning oligonucleotide primers
Target Accession nr Primer Sequence (5'-3')
PPARα NM005036 forward ACTCAACAGTTTGTGGCAAGACA
reverse GGAAGCACGTCCTCACATGA
PPARβδ NT007592 forward GCACATCTACAATGCCTACCTGAA
reverse CTCGATGTCGTGGATCACAAA
PPARγ NM005037 forward AAGTTCAATGCACTGGAATTAGATGA
reverse TGTAGCAGGTTGTCTTGAATGTCTTC
β-actin NM001101 forward GCCAACCGCGAGAAGATG
reverse ACGGCCAGAGGCGTACAG
Gene expression changes were assessed using the comparative cycle threshold (Ct) method . The relative amounts of mRNA for PPARα, PPARβδ and PPARγ were determined by subtracting Ct values for these genes from the β-actin Ct value (housekeeping gene) and expressed as the amount of mRNA in relation to 100,000 mRNA molecules of β-actin (100,000·2ΔCt).
Immunohistochemistry
The sections from nasal biopsies and nasal polyps were processed for the immunocytochemial demonstration of PPARγ. The PPARγ antibody (Cayman Chemical Company, Ann Arbor, Mi, USA) was raised in rabbit against a peptide corresponding to amino acids 82–101 of human PPARγ1. It cross-reacts with PPARγ2. The antibody was used in dilution 1:800. For the demonstration of the antigen-antibody reaction indirect immunofluorescence was used. Briefly, the cryostat sections were first washed with PBS and then rinsed in PBS for 15 min followed by incubation for 45 min with secondary antibodies raised against rabbit IgG and conjugated to FITC (1:80; swine anti-rabbit FITC, DAKO, Copenhagen, Denmark). Slides were cover-slipped in glycerol/PBS 2:1 (v/v) containing DAP 1 (1 mg/μL) and observed under microscope with chromefluorescence filters.
Negative controls for non-specific binding included normal rabbit serum without primary antibody and secondary antibody alone. Since cross-reactions with other proteins containing amino acid sequences recognized by the antisera could not be excluded, it is appropriate to refer to the immunoreactive material as "PPARγ-like". For brevity, the immunoreactive material is referred to as PPARγ in the text. For quantification, sections were analyzed with Visiopharm Integrator System® v2.1.2 (Visiopharm, Hørsholm, Denmark). The whole batch was immunostained and processed at the same time and the slides were analysed at the same time. Since the background level can differ between the specimens, each slide was individually analysed in that the background level was used as a reference to the induced immunostaining. An intensity reaching a certain threshold was regarded as positive and the area of this staining was measured in relation to the length of the epithelium. The computer program does not analyse the intensity of the staining but only staining that reached a certain level of intensity.
Statistical analysis
All data sets were analysed by Kolmogorov Smirnov test and since the data for PCR expression predominantly not was Gaussian distributed, Kruskal-Wallis test or Wilcoxon signed rank test was performed and expressed as median value (minimum-maximum), whereas the immunohistochemistry data that were found Gaussian distributed were analyzed by t-test and expressed as mean value ± s.e.m. The null hypothesis was rejected at P < 0.05.
Results
The standard curves for PPARα, PPARβδ, PPARγ and β-actin had correlation coefficients ranging between 0.93 and 0.99 and generated slope values not significantly different from each other. The efficiency of the PCR reaction was calculated and ranged between 1.96 and 2.01. The RT-PCR analysis of total RNA extracted from nasal biopsies and nasal polyps demonstrated the presence of PPARα, PPARβδ, PPARγ and β-actin in all samples. Melting curve analysis revealed a single peak in each sample and agarose electrophoresis generated expected PCR products with a single band close to the 100 bp marker.
Of the three PPARs, the expression of PPARα and γ were generally higher than the levels of PPARβδ. No differences were obtained when expression levels for the different PPARs in nasal biopsies from healthy volunteers were compared with biopsies derived from patients with symptomatic allergic rhinitis (Figure 1): PPARα (mRNA in relation to 100,000 mRNA molecules of β-actin) 170 (53–4512) and 82 (43–491), PPARβδ 52 (14–481) and 43 (18–464) and PPARγ 305 (144–2628) and 321 (171–699) in controls and patients with rhinitis, respectively.
Figure 1 Expression levels of PPARα, PPARβδ and PPARγ in biopsies of the nasal mucosa from 10 healthy volunteers (control) and 10 patients with symptomatic allergic rhinitis (allergic) together with biopsies of nasal polyps from 11 patients with bilateral nasal polyposis (polyp). Levels of PPAR mRNA are calculated in relation to 100,000 mRNA molecules of β-actin. Bold lines represent the median values. The expression of PPARα and PPARγ was lower in polyps than in normal nasal mucosa (**p < 0.01).
All three PPARs were also detected in nasal polyps obtained from patients not subjected to steroid treatment (Figure 1). The mRNA levels for PPARα and PPARγ were significantly lower in polyps than in normal nasal mucosa (mRNA in relation to 100,000 mRNA molecules of β-actin; 31 (12–232) and 132 (52–243) for PPARα and PPARγ, respectively). No corresponding differences were seen for and PPARβδ
In order to evaluate if local steroid treatment could affect the expression of the different PPARs we managed to obtain one set before and another set after treatment with steroids from seven of the polyposis patients (Figure 2). Four weeks of treatment resulted in a reduction in the expression of PPARγ (mRNA in relation to 100,000 mRNA molecules of β-actin; 154 (87–244) before and 72 (50–111) after treatment). The expression of PPARα and PPARβδ was not affected by steroid treatment.
Figure 2 Expression levels of PPARα, PPARβδ and PPARγ in polyps from 7 patients with bilateral nasal polyposis before (control) and after steroid treatment. Levels of PPAR mRNA are calculated in relation to 100,000 mRNA molecules of β-actin. Bold lines represent the median values. The PPARγ levels were reduced following steroid treatment (*p < 0.05).
Using immunohistochemistry the protein expression of PPARγ was localized and evaluated (Figure 3A–D). In the nose, PPARγ immunofluorescence was prominent in the surface epithelium, but was also detected in smooth muscle around blood vessels and in acini of small seromucous glands. In addition, PPARγ immunofluorescence was seen in infiltrating inflammatory cells. No differences in PPAR staining could be calculated between nasal biopsies obtained from healthy controls and in biopsies derived from patients with symptomatic allergic rhinitis. In polyps, the PPARγ staining was most prominent in the basal epithelial cells. Quantitative computerized analysis revealed a higher immunoreactivity in the epithelium from biopsies than polyps (area/length units: 94.3 ± 16.4 and 52.6 ± 6.7, respectively; 5 patients from each group; Figure 4). In sections from 5 patients without and 6 patients with steroids, the treatment revealed a reduction after the treatment (area/length units: 52.6 ± 6.7 and 27.5 ± 8.1, respectively).
Figure 3 Immunohistochemical localization of PPARγ in biopsies of the nasal mucosa from control subjects (A) and patients with allergic rhinitis (B), and in polyps before (C) and after treatment with steroids (D). Magnification: ×200.
Figure 4 Quantitative computerized analysis of PPARγ immunostained epithelium in sections from biopsies of nasal mucosa and nasal polyps from 5 patients, and 6 patients with fluticasone treatment. The area of those parts of the epithelium that were immunostained by PPARγ antibodies was measured in relation to the length of the epithelium. Bold lines represent the median values. * P < 0.05.
Discussion
In the present study mRNA expression of PPARα, PPARβδ, PPARγ was found in all specimens. The expression of PPARs between patients with allergic rhinitis and healthy volunteers was more or less identical. Nasal polyps exhibited lower mRNA expression levels of PPARα and PPARγ than normal nasal mucosa and these levels were, for PPARγ, further reduced following four weeks of treatment with local steroids. PPARγ immunofluorescence was prominent in the epithelium of both normal nasal mucosa and polyps. The epithelial PPARγ immunoreactivity was similar in nasal biopsies from patients with allergic rhinitis and healthy volunteers, but was lower in polyps and further decreased after treatment with fluticasone.
In concordance with the present findings PPARγ has been found to be expressed in cultured human epithelial cells and its activation has been shown to antagonize pro-inflammatory events in this system [25]. In monocytes and macrophages PPARγ-agonists inhibit the expression of proinflammatory cytokines, such as TNF-α, IL-1β, and IL-6 [26,27]. PPARγ is also expressed by eosinophils and agonists inhibit eosinophil chemotaxis and antibody-dependent cellular cytotoxicity reactions in vitro [22].
Similar in vivo findings have been seen in a murine model of asthma, where treatment with a PPARγ agonist inhibited the development of allergic inflammation, including pulmonary eosinophilia and airway hyperreactivity [22,28]. Furthermore, it was recently demonstrated that cultured human airway smooth muscle cells express PPARα and PPARγ [29], which might relate to the present finding of PPARγ positive cells in conjunction with the vascular smooth muscle in biopsies from both the inferior turbinate and polyps.
Based on animal studies and in vitro data, an anti-inflammatory role for PPARγ has been suggested. However, human data to support this idea are still limited. Benayoun and colleagues have shown that PPARγ is augmented in the bronchial submucosa, the airway epithelium, and the smooth muscle of asthmatic patients, as compared with control subjects [24]. The enhanced PPARγ expression is accompanied by increased proliferation and apoptosis of airway epithelial and submucosal cells. In addition, several studies have demonstrated that PPARγ plays an important role in the control of the inflammatory response [21,30], acting on T cells, macrophages, dendritic cells, and mast cells [31-34]. Therefore, an increased expression of PPARγ could have been expected in conjunction with the increased amount of inflammatory cells seen during symptomatic allergic rhinitis. This alteration could not be found in the present study. Since PPARγ appears to be expressed in response to the ongoing inflammation, it might be that it takes more than 3–4 days of pollen exposure to fully activate this putative "defense system".
Nasal polyposis represents a chronic type of inflammation and the lower levels of PPARγ in comparison with normal nasal mucosa might reflect a reduced ability of the diseased mucosa to respond to airway inflammation, thereby facilitating the polyp formation. Steroids (locally administrated, with or without an oral supplement) have, in analogy with our experiments, been reported to down regulate PPARγ expression in bronchial epithelium, mucosa and smooth muscle [24]. Thus, the beneficial effect of glucocorticoid treatment on nasal polyposis may adversely affect the down-regulation of PPARγ. On the other hand, if the inflammatory response is reduced, there will be less need for anti-inflammatory mediators. Notwithstanding whether this is beneficial or not, these studies indicate that PPARγ might be regulated by steroid therapy and that increased knowledge of the physiological effect of PPARγ within the airways might be of importance for our understanding of airway regulation
Neither PPARα nor βδ exhibited any difference in their expression when specimens from healthy volunteers were compared with samples obtained from patients with symptomatic allergic rhinitis. Nor did local steroid treatment affect the expression of these PPARs in nasal polyposis. Inflammation induced by LTB4, a PPARα ligand, has been shown to be prolonged in PPARα-deficient mice [15], suggesting an anti-inflammatory role for this receptor. In contrast, in mice injected with lipopolysaccharide (LPS), activation of PPARα induced a significant increase in plasma tumour necrosis factor-α (TNFα) levels [35]. Pro-differentiation and anti-proliferative effects in conjunction with PPARα-stimulation have been demonstrated in various skin models, as well as an ability for this type of stimulation to reduce cutaneus inflammation in vivo [36,37]. Albeit the lower expression of PPARα in polyps, the present study did not give any further evidence for a role for PPARα in airway inflammation. For PPARβδ, which is ubiquitously expressed in the human body, the eventual function in inflammation remains uncertain. The relatively low expression level and the unaltered expression seen in the present study add no further information.
The present polyp data could be interpreted as a support for the widespread idea of an anti-inflammatory role for PPARγ within the human airways. However, data contradicting an anti-inflammatory role for PPARγ has been published [38,39]. This discrepancy has been attributed to the use of nonselective ligands [38] or the use of very high concentrations of more selective ligands [39]. In this context, it is essential to recognize that inflammation is normally a self-resolving process with the existence of both positive and negative regulators that ultimately allow complete resolution and homeostasis. In the absence of resolution and clearance or in the event of a dampened healing response, persistent inflammation can arise in the form of tissue damage as associated with chronic disease.
The down-regulation of PPARγ, in nasal polyposis but not in turbinates during symptomatic seasonal rhinitis, suggests that PPARγ might be of importance in long standing inflammations, causing polyps, whereas an eventual role in allergic rhinitis remains to be established. It is tempting to speculate in a therapeutic future for PPARγ activating agonists in the treatment of long standing airway inflammation.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
LOC performed the sample preparation and together with MA analyzed the data and drafted the manuscript. RU and MA performed the immunohistochemistry. MH participated in the design of the study and revising the manuscript.
Acknowledgements
The present work was supported by the Swedish Medical Research Council, the Swedish Heart Lung Foundation, the Swedish Association for Allergology, the Swedish Foundation for Health Care Science and Allergic Research, Magn. Bergvall Foundation, Tore Nilsson Foundation, Crafoord Foundation and the Royal Physiographic Society.
==== Refs
Wang D Clement P Smitz J De Waele M Derde MP Correlations between complaints, inflammatory cells and mediator concentrations in nasal secretions after nasal allergen challenge and during natural allergen exposure Int Arch Allergy Immunol 1995 106 278 285 7888790
Bachert C Hauser U Prem B Rudack C Ganzer U Proinflammatory cytokines in allergic rhinitis Eur Arch Otorhinolaryngol 1995 252 Suppl 1 S44 9 7537566
Baraniuk JN Pathogenesis of allergic rhinitis J Allergy Clin Immunol 1997 99 S763 72 9042069 10.1016/S0091-6749(97)70125-8
Kinhult J Egesten A Benson M Uddman R Cardell LO Increased expression of surface activation markers on neutrophils following migration into the nasal lumen Clin Exp Allergy 2003 33 1141 1146 12911790 10.1046/j.1365-2222.2003.01682.x
Pawankar R Nasal polyposis: an update: editorial review Curr Opin Allergy Clin Immunol 2003 3 1 6 12582307 10.1097/00130832-200302000-00001
Benson M Adner M Cardell LO Cytokines and cytokine receptors in allergic rhinitis: how do they relate to the Th2 hypothesis in allergy? Clin Exp Allergy 2001 31 361 367 11260146 10.1046/j.1365-2222.2001.01045.x
Bernstein JM Ballow M Rich G Allen C Swanson M Dmochowski J Lymphocyte subpopulations and cytokines in nasal polyps: is there a local immune system in the nasal polyp? Otolaryngol Head Neck Surg 2004 130 526 535 15138416 10.1016/j.otohns.2003.12.022
Dinarello CA Role of pro- and anti-inflammatory cytokines during inflammation: experimental and clinical findings J Biol Regul Homeost Agents 1997 11 91 103 9498158
Law RE Goetze S Xi XP Jackson S Kawano Y Demer L Fishbein MC Meehan WP Hsueh WA Expression and function of PPARgamma in rat and human vascular smooth muscle cells Circulation 2000 101 1311 1318 10725292
Ellis CN Varani J Fisher GJ Zeigler ME Pershadsingh HA Benson SC Chi Y Kurtz TW Troglitazone improves psoriasis and normalizes models of proliferative skin disease: ligands for peroxisome proliferator-activated receptor-gamma inhibit keratinocyte proliferation Arch Dermatol 2000 136 609 616 10815854 10.1001/archderm.136.5.609
Saubermann LJ Nakajima A Wada K Zhao S Terauchi Y Kadowaki T Aburatani H Matsuhashi N Nagai R Blumberg RS Peroxisome proliferator-activated receptor gamma agonist ligands stimulate a Th2 cytokine response and prevent acute colitis Inflamm Bowel Dis 2002 8 330 339 12479648 10.1097/00054725-200209000-00004
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
Kliewer SA Forman BM Blumberg B Ong ES Borgmeyer U Mangelsdorf DJ Umesono K Evans RM Differential expression and activation of a family of murine peroxisome proliferator-activated receptors Proc Natl Acad Sci U S A 1994 91 7355 7359 8041794
Braissant O Foufelle F Scotto C Dauca M Wahli W Differential expression of peroxisome proliferator-activated receptors (PPARs): tissue distribution of PPAR-alpha, -beta, and -gamma in the adult rat Endocrinology 1996 137 354 366 8536636 10.1210/en.137.1.354
Devchand PR Keller H Peters JM Vazquez M Gonzalez FJ Wahli W The PPARalpha-leukotriene B4 pathway to inflammation control Nature 1996 384 39 43 8900274 10.1038/384039a0
Hatae T Wada M Yokoyama C Shimonishi M Tanabe T Prostacyclin-dependent apoptosis mediated by PPAR delta J Biol Chem 2001 276 46260 46267 11551955 10.1074/jbc.M107180200
Forman BM Tontonoz P Chen J Brun RP Spiegelman BM Evans RM 15-Deoxy-delta 12, 14-prostaglandin J2 is a ligand for the adipocyte determination factor PPAR gamma Cell 1995 83 803 812 8521497 10.1016/0092-8674(95)90193-0
Duval C Fruchart JC Staels B PPAR alpha, fibrates, lipid metabolism and inflammation Arch Mal Coeur Vaiss 2004 97 665 672 15283041
Hsueh WA Bruemmer D Peroxisome proliferator-activated receptor gamma: implications for cardiovascular disease Hypertension 2004 43 297 305 14732733 10.1161/01.HYP.0000113626.76571.5b
Michalik L Desvergne B Wahli W Peroxisome-proliferator-activated receptors and cancers: complex stories Nat Rev Cancer 2004 4 61 70 14708026 10.1038/nrc1254
Daynes RA Jones DC Emerging roles of PPARs in inflammation and immunity Nat Rev Immunol 2002 2 748 759 12360213 10.1038/nri912
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
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 139 163 171 12746235 10.1038/sj.bjp.0705232
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
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
Ricote M Li AC Willson TM Kelly CJ Glass CK The peroxisome proliferator-activated receptor-gamma is a negative regulator of macrophage activation Nature 1998 391 79 82 9422508 10.1038/34178
Jiang C Ting AT Seed B PPAR-gamma agonists inhibit production of monocyte inflammatory cytokines Nature 1998 391 82 86 9422509 10.1038/35154
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
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
Delerive P Fruchart JC Staels B Peroxisome proliferator-activated receptors in inflammation control J Endocrinol 2001 169 453 459 11375115 10.1677/joe.0.1690453
Harris SG Phipps RP The nuclear receptor PPAR gamma is expressed by mouse T lymphocytes and PPAR gamma agonists induce apoptosis Eur J Immunol 2001 31 1098 1105 11298334 10.1002/1521-4141(200104)31:4<1098::AID-IMMU1098>3.0.CO;2-I
Faveeuw C Fougeray S Angeli V Fontaine J Chinetti G Gosset P Delerive P Maliszewski C Capron M Staels B Moser M Trottein F Peroxisome proliferator-activated receptor gamma activators inhibit interleukin-12 production in murine dendritic cells FEBS Lett 2000 486 261 266 11119715 10.1016/S0014-5793(00)02319-X
Sugiyama H Nonaka T Kishimoto T Komoriya K Tsuji K Nakahata T Peroxisome proliferator-activated receptors are expressed in human cultured mast cells: a possible role of these receptors in negative regulation of mast cell activation Eur J Immunol 2000 30 3363 3370 11093153 10.1002/1521-4141(2000012)30:12<3363::AID-IMMU3363>3.0.CO;2-B
Sugiyama H Nonaka T Kishimoto T Komoriya K Tsuji K Nakahata T Peroxisome proliferator-activated receptors are expressed in mouse bone marrow-derived mast cells FEBS Lett 2000 467 259 262 10675550 10.1016/S0014-5793(00)01169-8
Hill MR Clarke S Rodgers K Thornhill B Peters JM Gonzalez FJ Gimble JM Effect of peroxisome proliferator-activated receptor alpha activators on tumor necrosis factor expression in mice during endotoxemia Infect Immun 1999 67 3488 3493 10377130
Thieringer R Fenyk-Melody JE Le Grand CB Shelton BA Detmers PA Somers EP Carbin L Moller DE Wright SD Berger J Activation of peroxisome proliferator-activated receptor gamma does not inhibit IL-6 or TNF-alpha responses of macrophages to lipopolysaccharide in vitro or in vivo J Immunol 2000 164 1046 1054 10623855
Kuenzli S Saurat JH Peroxisome proliferator-activated receptors in cutaneous biology Br J Dermatol 2003 149 229 236 12932225 10.1046/j.1365-2133.2003.05532.x
Wiesenberg I Chiesi M Missbach M Spanka C Pignat W Carlberg C Specific activation of the nuclear receptors PPARgamma and RORA by the antidiabetic thiazolidinedione BRL 49653 and the antiarthritic thiazolidinedione derivative CGP 52608 Mol Pharmacol 1998 53 1131 1138 9614218
Spiegelman BM PPARgamma in monocytes: less pain, any gain? Cell 1998 93 153 155 9568708 10.1016/S0092-8674(00)81567-6
Chawla A Barak Y Nagy L Liao D Tontonoz P Evans RM PPAR-gamma dependent and independent effects on macrophage-gene expression in lipid metabolism and inflammation Nat Med 2001 7 48 52 11135615 10.1038/83336
|
16271155
|
PMC1298337
|
CC BY
|
2021-01-04 16:36:25
|
no
|
Respir Res. 2005 Nov 7; 6(1):132
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-132
|
oa_comm
|
==== Front
Theor Biol Med ModelTheoretical Biology & Medical Modelling1742-4682BioMed Central London 1742-4682-2-471629723510.1186/1742-4682-2-47CommentaryComment on and reply to "Analysis of variation of amplitudes in cell cycle gene expression" by Liu, Gaido and Wolfinger: On the analysis of gene expression during the normal, eukaryotic, cell cycle Cooper Stephen [email protected] Department of Microbiology and Immunology, University of Michigan Medical School, Ann Arbor, Michigan 48109-0620, USA2005 18 11 2005 2 47 47 13 11 2005 18 11 2005 Copyright © 2005 Cooper; licensee BioMed Central Ltd.2005Cooper; 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 paper of Liu, Gaido and Wolfinger on gene expression during the division cycle of HeLa cells using the data of Whitfield et al. are discussed in order to see whether their analysis is related to gene expression during the division cycle.
Results
The results of Liu, Gaido and Wolfinger demonstrate that different inhibition methods proposed to "synchronize" cells lead to different levels of gene expression. This result, in and of itself, should be taken as evidence that the original work of Whitfield et al. is flawed and should not be used to support the notion that the cells studied were synchronized or that the microarray analyses identify cell-cycle-regulated genes. Furthermore, the DNA content evidence presented by Whitfield et al. supports the proposal that the cells described as 'synchronized' are not synchronized. A comparison of the gene expression amplitudes from two different experiments indicates that the results are not reproducible.
Conclusion
It is concluded that the analysis of Liu, Gaido, and Wolfinger is problematic because their work assumes that the cells they analyze are or were synchronized. The very fact that different inhibition methods lead to different degrees of gene expression should be taken as additional evidence that the experiments should be viewed skeptically rather than accepted as an approach to understanding gene expression during the cell cycle.
==== Body
Introduction
The recent paper by Liu, Gaido, and Wolfinger entitled "Analysis of Variation of Amplitudes for Cell Cycle Gene Research" [1] requires comment. Because the subject of gene expression variation during the cell cycle is such an important topic of current interest, it is necessary that any work supporting cycle-specific gene expression be beyond reproach and criticism. If the paper by Liu, Gaido and Wolfinger [1] remains unchallenged, it will merely be used as another reference supporting the data of Whitfield et al. [2] regarding gene expression variation during the division cycle. Because I believe that such a conclusion is unwarranted, I now summarize my objections to this analysis so that readers may be able to compare two alternative views of the cell cycle and gene expression during the cell cycle.
As the reader will gather, the view I present here is not widely held, and is even a minority viewpoint. But I hope that readers will at least agree that science does not proceed by majority vote. The most important point I make is a critique of the fundamental paper analyzed by Liu, Gaido and Wolfinger [1], that of Whitfield et al. [2]. I point out that the two experimental approaches analyzed, experiments 2 and 4 of Whitfield et al., are not analyses of synchronized cells. I propose that the cells studied in the Whitfield et al. paper are not synchronized – at all. They may be "aligned" for certain cell properties, and even this may be problematic. But as will be pointed out in detail below, such alignment is not equivalent to synchronization.
Theoretical critique of whole-culture synchronization
The two Whitfield et al. experiments [2] analyzed by Liu, Gaido and Wolfinger [1] are a double thymidine block experiment (thy-thy), and a thymidine-nocodazole block (thy-noc). I propose is that a truly synchronized culture is one where the starting cells are all of the same genome content and cell size and thus reflect the properties of cells of a particular cell-cycle age while growing in unlimited medium. If one does not narrow the size distribution, or does not have the starting cells mimic a particular cell (with respect to DNA content, cell size and cell composition) of a particular age, then the cells are not synchronized. A complete analysis of this idea has been presented [3-16] and I merely refer the readers to these papers. (Some of these papers can be read at .)
Criteria for cell synchronization and analysis of gene expression
One should apply clearly stated and stringent criteria for synchronization and microarray experiments. Such criteria are listed here so the reader can at least see why it is proposed that the Whitfield et al. experiments are problematic.
Criteria for successful synchronization
1) If newborn cells are produced by the synchronization method, there should be a minimal increase in cell number for a period of time covering a significant fraction of the interdivision time.
2) The rise in cell numbers during division should occur over a relatively small fraction of the total interdivision time. It may be as small as 10% for 90% of the final rise in cell number, or it may be as large as 20–25%. Knowing this value is important in judging a synchronization procedure.
3) At the time of synchronous division, the cell number should double. If cell number does not double, that means some cells are dead or altered; this minority of cells could be giving results that obfuscate the results emanating from the majority of dividing cells.
4) There should be at least two successive cycles available for analysis. If only one cycle is analyzed, the results may merely reflect artifacts or perturbations resulting from synchronization. Presumably, but not necessarily, these artifacts would be eliminated in the second cycle.
5) Successive generations (i.e. the time between rises in cell number) should be of equal length and equal to the doubling time of cells in exponential growth.
6) Data points should show synchrony without any need to connect points or draw a suggestive line indicating synchrony. The data should speak for themselves.
7) The DNA distribution of cells should be narrow in the synchronized cells and these distributions should then reflect the movement of cells through the division cycle. Thus, newborn cells should be essentially pure cells with a G1-phase amount of DNA, the DNA content should then move through S-phase contents, there should be a period of time when cells have only G2-phase DNA contents, and then there should be a return to essentially pure G1-phase DNA contents.
8) The size distribution of newly synchronized cells should be narrower than the size distribution of the original population, cell size should increase as the cells move through the cell cycle, and during the period of cell division there should be a bimodal distribution of cell sizes.
9) Cell numbers should be determined by a method that eliminates investigator bias. For example, electronic cell counting is to be preferred to microscope counting chambers.
10) Only selection methods can give synchrony. Whole-culture methods, using inhibition or starvation, cannot synchronize cells. This is not so much a criterion as a theoretical rule regarding synchronization in general.
11) Alignment of cells so that cells all have a particular property in common (e.g. all cells have a G1-phase DNA content) does not mean that the cells are synchronized. Synchronized divisions are the sine qua non of synchrony.
Criteria for successful analysis of gene expression during the division cycle
12) Gene expression results should be replicated (with allowance for normal synchrony decay) in successive cycles. If data do not repeat over two or more cycles, the cells are very likely perturbed by the synchronization method.
13) Peaks in gene expression should decay when expression is studied over more than one cycle. This is because synchrony, if normal and unperturbed, should decay.
14) If a selection method is used, a mock selection should be performed where the selection procedure is performed but the cells are all recombined together and analyzed. These combined cells should not give a variable pattern of gene expression. This controls for perturbation of the culture by the selection method.
15) Results using different synchronization methods should give the same results. Different experiments should be reproducible in cyclicity and in phasing, and thus independent of synchronization methods. That is, the results should not depend on the particular synchronization method used.
Criteria for successful use of microarrays to analyze cycle-related gene expression
16) Analyses should be performed more than once, and the results should be "reproducible". The qualification on reproducible is related to the acceptance of some degree of statistical variation.
17) The data should be published in accordance with the MIAME (Minimum Information About a Microarray Experiment) or MAGE (microarray gene expression object model) standards, so that the public data can be analyzed.
18) Microarray results should be compared to randomized data to show that the observed cyclicities are not the result of random noise or experimental variation. Satisfaction of this criterion, however, does not mean the results are necessarily related to the cell cycle, as perturbations of cells by a synchronization procedure may still be present.
19) Criteria for successful identification of cyclicity should be determined before microarray analysis.
20) Both false positives and false negatives should be considered in the analysis. Just because a particular gene result fits pre-existing data collected by classical means, one must not consider this to support the results unless the previous synchronization method is different. Otherwise the microarray experiment just repeats the same experiment, with a repetition of the artifacts of synchronization in two independent experiments.
21) If some genes are expressed differently in two successive cycles this should invalidate the entire experiment – even for those genes that are expressed similarly in two successive cycles – because the non-repeating patterns indicate that there are artifacts produced by the synchronization.
Are the HeLa cells analyzed truly synchronized?
If one looks at the experiments of Whitfield et al. [2] it is clear that the cells are not synchronized. The DNA patterns presented by Whitfield et al. testify to the fact that the cells are not only not synchronized but are also perturbed. The DNA patterns in the thy-thy and thy-noc experiments are quite inconsistent with the proposal that the cells are synchronized. The flow cytometric data on DNA contents during growth of the arrested cells shows the cells are not synchronized. The DNA patterns start with an 8N (it appears) value for DNA content, which goes down to 2N and back to 4N and never repeats the 8N. In between the start and the end there is no systematic variation of cells through G1, S and G2.
In summary, the work of Whitfield et al. is not a synchrony experiment, and the results in the paper [2] show the cells are not synchronized. What we have here is a perturbation experiment. If one wishes to look at the response of genes to perturbation, then it is a good experiment. The problem is whether the methods of Whitfield et al. lead to an understanding of what happens in the normal, unperturbed cell. Theory says they are not synchronized and experimental results from the Whitfield et al. paper itself support this proposal.
Variation in gene expression: meaning and relevance to the cell cycle
It is of interest to look at the basic data in Table 1 of Liu, Gaido and Wolfinger [1]. In Table 1 they present results on the gene expression amplitudes in two different experiments. In Table 1 they are trying to explain differences between different experiments in terms of variation in the phase angles of peaking of gene expression in different individual cells. Thus, given a particular variation in gene expression in different cells, one can account for different amplitudes based on the variation in phase angle of the peak of expression. If one had a particular cell growing with a 20 hour doubling time, and in some cells the peak was at 9 hours, others 10 hours, and other 11 hours, the amplitude would be reduced based on the distribution and frequency of the peak of expression in different cells.
At this point I have philosophical viewpoint to express, which should, at a minimum, be made explicit. I propose that for a given cell, if there is a particular pattern of gene expression during the cell cycle (something, I may add, that I am skeptical of at the start – that is, I am skeptical of whether there are a large number of genes with cell-cycle dependent expression), then it is the object of an experiment to know what that pattern is. Thus, given a single cell growing in an unlimited supply of fresh medium, if there was a sinusoidal expression in this cell that had a certain amplitude and a certain time of peaking, a good experiment would get this result. Two different true synchronization experiments should lead to the same results.
In the Liu, Gaido and Wolfinger paper the authors talk about comparing "different conditions". I apologize for being so fastidious and demanding, but I want to distinguish between what I consider different "conditions" and different "experimental approaches" to determining something about cells. For example, "different conditions" means growing cells in one case at 25°C and in another case at 37°C and seeing what happens to them or what is different in these different conditions. Or one may have Medium A and Medium B and compare the cells for some property. These are "different conditions." But when you take the same cells, growing in a given single condition, and then try to analyze them using two different methods, what you should get, in an ideal world, are results that agree with each other. To summarize, the Whitfield et al. experiments are using different methods of analysis, not "different conditions."
I don't know how people measure the speed of light, but I do remember that there are many different approaches. But when all is said and done, all of these methods give rather reproducible results. That is, different analytical "methods" give the same result. I propose that we should analogize the thy-thy and the thy-noc to different methods examining same condition.
Returning now to the comparison of the two experiments, I present a graph (Figure 1) of the K values in a scatter diagram. The R2 value is 0.37, which some may say is "rather strong" and others may say is "rather weak". In measuring the size of an atom, this value would be rather weak. In sociological measurements the data would be considered rather strong. There is no absolute measure of how one should accept the scatter as being strong. For a person who is inclined to believe, the data, if correct, should show a clear 45 degree line from lower left to upper right. For a person who says that the results of the two Whitfield et al. experiments are not reproducible, the data allow that. I look at the data in this scatter diagram and see non-reproducible results.
Figure 1 Replotting of the K values of the two experiments analyzed by Liu, Gaido and Wolfinger [1].
Figure 1, in a sense, is a redoing of the Figure 3 of Liu, Gaido, and Wolfinger [1] in a more intuitive manner. What Liu et al. are saying is that if one has broad enough error bars on the data one can say that many of the points do have similarity or even "identity". Perhaps. But that may be, and in my view is, a judgment call.
One important point that I wish to bring up is that if the two experiments have "different strengths of synchronization", then one would expect that one would have a systematic difference in amplitudes. Thus, if synchrony were sharper in one than the other, the amplitudes would be higher in that experiment than in the broader synchronization. This does not appear to be the case, which is again support for the notion that one should be wary of accepting the results of Whitfield et al. as a synchrony experiment.
Summary
The work of Liu, Gaido and Wolfinger [1] is a statistical analysis of a well-regarded paper [2] on the pattern of gene expression during the cell cycle of eukaryotic cells. There are no fundamental problems with the statistical analysis, but there are problems with the underlying experiments that form the data base for analysis.
==== Refs
Liu D Gaido KW Wolfinger R Analysis of variation of amplitudes in cell cycle gene expression Theor Biol Med Mod 2005 2 46 10.1186/1742-4682-2-46
Whitfield ML Sherlock G Saldanha AJ Murray JI Ball CA Alexander KE Matese JC Perou CM Hurt MM Brown PO Botstein D Identification of genes periodically expressed in the human cell cycle and their expression in tumors Mol Biol Cell 2002 13 1977 2000 12058064 10.1091/mbc.02-02-0030.
Cooper S G1 and S phase gene expression cannot be analyzed in mammalian cells synchronized by inhibition Microb Comp Genomics 1997 2 269 273 9689225
Cooper S Mammalian cells are not synchronized in G1-phase by starvation or inhibition: considerations of the fundamental concept of G1-phase synchronization Cell Prolif 1998 31 9 16 9666815 10.1046/j.1365-2184.1998.00110.x
Cooper S The continuum model and G1-control of the mammalian cell cycle Prog Cell Cycle Res 2000 4 27 39 10740812
Cooper S Shayman JA Revisiting retinoblastoma protein phosphorylation during the mammalian cell cycle Cell Mol Life Sci 2001 58 580 595 11361093
Cooper S The Schaechter-Bentzon-Maaløe experiment and the analysis of cell cycle events in eukaryotic cells Trends Microbiol 2002 10 169 173 11912022 10.1016/S0966-842X(02)02322-3
Cooper S Reappraisal of G1-phase arrest and synchronization by lovastatin Cell Biol Int 2002 26 715 727 12175675 10.1006/cbir.2002.0925
Cooper S Cell cycle analysis and microarrays Trends Genet 2002 18 289 290 12044356 10.1016/S0168-9525(02)02694-X
Cooper S Shedden K Microarray analysis of gene expression during the cell cycle Cell Chromosome 2003 2 1 12 14577836 10.1186/1475-9268-2-1
Cooper S Rethinking synchronization of mammalian cells for cell-cycle analysis Cell Mol Life Sci 2003 6 1099 1106 12861378
Cooper S Reappraisal of serum starvation, the restriction point, G0, and G1-phase arrest points FASEB J 2003 17 333 340 12631573 10.1096/fj.02-0352rev
Cooper S Control and maintenance of mammalian cell size BMC Cell Biol 2004 5 35 15456512 10.1186/1471-2121-5-35
Cooper S Whole-culture synchronization can not, and does not, synchronize cells Trends Biotechnol 2004 22 274 276 15158054 10.1016/j.tibtech.2004.04.011
Cooper S Is whole-culture synchronization biology's "Perpetual Motion Machine"? Trends Biotechnol 2004 26 266 269 15158052 10.1016/j.tibtech.2004.04.009
Cooper S Witkowski J DNA replication: The 30th anniversary of the bacterial model and the "Baby Machine" The Inside Story, DNA to RNA to Protein; Readings from Trends in Biochemical Sciences 2005 Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press 109 118
|
16297235
|
PMC1298338
|
CC BY
|
2021-01-04 16:39:25
|
no
|
Theor Biol Med Model. 2005 Nov 18; 2:47
|
utf-8
|
Theor Biol Med Model
| 2,005 |
10.1186/1742-4682-2-47
|
oa_comm
|
==== Front
J Transl MedJournal of Translational Medicine1479-5876BioMed Central London 1479-5876-3-381623616410.1186/1479-5876-3-38ResearchThe natural history of EGFR and EGFRvIII in glioblastoma patients Heimberger Amy B [email protected] Dima [email protected] David [email protected] Weiming [email protected] Kenneth [email protected] Department of Neurosurgery, The Brain Tumor Center, The University of Texas M. D. Anderson Cancer Center, Houston, Texas, USA2 Pathology, The Brain Tumor Center, The University of Texas M. D. Anderson Cancer Center, Houston, Texas, USA2005 19 10 2005 3 38 38 1 8 2005 19 10 2005 Copyright © 2005 Heimberger et al; licensee BioMed Central Ltd.2005Heimberger 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 epidermal growth factor receptor (EGFR) is over expressed in approximately 50–60% of glioblastoma (GBM) tumors, and the most common EGFR mutant, EGFRvIII, is expressed in 24–67% of cases. This study was designed to address whether over expressed EGFR or EGFRvIII is an actual independent prognostic indicator of overall survival in a uniform body of patients in whom gross total surgical resection (GTR; ≥ 95% resection) was not attempted or achieved.
Methods
Biopsed or partially/subtotally resected GBM patients (N = 54) underwent adjuvant conformal radiation and chemotherapy. Their EGFR and EGFRvIII status was determined by immunohistochemistry and Kaplan-Meier estimates of overall survival were obtained.
Results
In our study of GBM patients with less than GTR, 42.6% (n = 23) failed to express EGFR, 25.9% (n = 14) had over expression of the wild-type EGFR only and 31.5 % (n = 17) expressed the EGFRvIII. Patients within groups expressing the EGFR, EGFRvIII, or lacking EGFR expression did not differ in age, Karnofsky Performance Scale (KPS) score, extent of tumor resection. They all had received postoperative radiation and chemotherapy. The median overall survival times for patients with tumors having no EGFR expression, over expressed EGFR only, or EGFRvIII were 12.3 (95% CI, 8.04–16.56), 11.03 (95% CI, 10.18–11.89) and 14.07 (95% CI, 7.39–20.74) months, respectively, log rank test p > 0.05). Patients with tumors that over expressed the EGFR and EGFRvIII were more likely to present with ependymal spread, 21.4% and 35.3% respectively, compared to those patients whose GBM failed to express either marker, 13.0%, although the difference was not statistically significant. There was no significant difference in multifocal disease or gliomatosis cerebri among EGFR expression groups.
Conclusion
The over expressed wild-type EGFR and EGFRvIII are not independent predictors of median overall survival in the cohort of patients who did not undergo extensive tumor resection.
Gliomasurvivalsub-total resection
==== Body
Background
Glioblastoma multiforme (GBM) is the most common primary malignant neoplasm of the central nervous system in adults. Despite multimodal therapies, the median survival time of patients with GBM is approximately 1 year; however, there is considerable variability among these patients. Prognostic indicators have included age [1,2], Karnofsky Performance Scale (KPS) score [3], and extent of surgical resection [4,5]. The most frequent genetic alteration associated with GBM is amplification of the epidermal growth factor receptor (EGFR) gene, which results in over expression of the EGFR, a transmembrane tyrosine kinase receptor [6]. The majority of GBMs with EGFR amplification also contain the mutant EGFR gene, EGFRvIII [7], which is characterized by the deletion of exons 2–7, resulting in an in frame deletion variant that has a truncated extracellular domain with ligand-independent constitutive activity [8]. Previous work has shown that EGFR amplification is evident in all GBMs expressing EGFRvIII and GBMs lacking the amplified EGFR are not positive for EGFRvIII protein. In addition, we have previously shown that the positive staining with the 528 antibody, which recognizes an unspecified extracellular epitope of the EGFR, is highly correlated with EGFR amplification [9].
The role of the over expressed EGFR (wild type) and the variant (vIII) receptor in malignant progression of glial tumors and their respective impacts on overall survival have been debated in the literature. Over expression of wild-type EGFR was not found to be an independent prognostic indicator of survival in several studies [10-12], and one study was inconclusive [13]. Four studies identified EGFR as a negative prognostic indicator of survival [14-16], one of which showed the effect only in patients younger than 45 [17]. In some of these studies, analysis was limited by small sample size, uncharacterized extent of surgical resection, and variable postoperative treatment. The prognostic impact of EGFRvIII has not been as extensively studied, but in the study that addressed this variable, the presence of EGFRvIII was found to be an independent and significant unfavorable prognosticator of survival [18]. In contrast, our group has shown that the presence of the over expressed EGFR and EGFRvIII were not independent negative prognostic indicators in patients who were able to undergo gross total resection (GTR)(>95% MRI-based volumetric resection) [19] when confounding variables were accounted for. However, our previous study had an inherent bias in that more invasive, infiltrative tumors were less likely to be selected for surgery and less likely to have a GTR. Thus, the primary purpose of this study was to determine if EGFR and EGFRvIII are negative prognostic indicators in patients who either receive a biopsy, partial resection (<85%) or subtotal resection (85–95%).
EGFR amplification and EGFRvIII have been shown to increase glioma proliferation and invasion in vitro [8,9]; therefore logically EGFR and/or EGFRvIII expression could exhibit a proclivity towards the development of multifocal disease, gliomatosis cerebri or ependymal seeding. Therefore, the secondary purpose of this study was to address the natural history of EGFR and EGFRvIII expression in GBM patients undergoing less than GTR.
Methods
Study Population
The study was conducted according to an IRB-approved protocol (LAB03-0228). 54 GBM (WHO grade IV) patients received conformal irradiation and adjuvant chemotherapy and were retrospectively reviewed to determine whether tumor expression of EGFR or EGFRvIII conferred a poor prognosis. Clinical and survival information was obtained from the Department of Neurosurgery Clinical and Imaging Database and The University of Texas M. D. Anderson Cancer Center Tumor Registry. All tissue specimens were acquired at initial diagnosis and resection and were classified morphologically and graded according to WHO criteria.
Immunohistochemical Detection of Over Expressed EGFR and EGFRvIII
Immunostaining was performed as previously described [17]. Briefly, 5-μm tumor-tissue sections were mounted on positively charged slides, deparaffinized, and rehydrated in phosphate-buffered saline (PBS). Endogenous peroxidase activity was blocked with 3% hydrogen peroxide in PBS/0.05% Tween 20 for 20 min. Sections were washed in PBS and blocked for 20 min in the appropriate serum (from the same species as the secondary antibody) diluted to 10% in PBS. The primary antibody for EGFR detection was the monoclonal mouse anti-human pan-EGFR clone 528 (Oncogene Research; 1:50 dilution) [20] and for EGFRvIII detection was a rabbit anti-human polyclonal antibody (Zymed, San Francisco, CA; 1:1200 dilution). For EGFRvIII staining, microwave antigen retrieval was performed by placing the slides in 50 mM citrate buffer (pH 6.0) and microwaving for 12 min at full power and 10 min at 20% power, followed by cooling for 15 min and two to three 5-min washes in PBS. For EGFR staining, pretreatment consisted of placing 0.025% trypsin on the tissue and incubating for 30 min at room temperature. Primary antibodies, diluted in PBS/10% serum, were applied to the sections in a humid chamber overnight at 4°C. Sections were washed two to three times in PBS, and secondary antibodies were applied using the Dako (Carpinteria, CA) Envision kit, according to the manufacturer's instructions. Detection of bound secondary antibody was performed with diaminobenzadine for 5 min. Sections were then counterstained with hematoxylin and mounted. Each batch of stained slides was accompanied by a positive control (as determined by both positive staining for EGFR and EGFRvIII), as well as demonstration of gene amplification using samples previously described as positive by both genetic analyses (Southern analysis and RT-PCR) as well as immunohistochemistry [9]. Non-neoplastic brain tissue was used as a negative control, and positive staining was never seen in this tissue. As additional controls, 20 randomly selected cases were stained using equal concentrations of mouse (anti-cytokeratin 14, Biogenex, San Ramon, CA) and rabbit antibodies (anti-cytokeratin 19, Neomarkers of Lab Vision Corporation, Fremont, CA) at concentrations matching the EGFR and EGFRvIII staining conditions, respectively. These irrelevant controls were uniformly negative in each of the 20 cases tested. Scoring was accomplished using a simple positive-negative scoring system. Any detectable cytoplasmic-membrane staining in tumor cells was scored as positive/overexpressed.
Statistical Analysis
The frequencies and descriptive statistics of demographic and clinical variables were performed for the patients in this study. The chi-square or exact test (StatXact 3 for Windows) was used for categorical variables as appropriate. The analysis of variance was used for continuous variables. Cumulative survival times from the time of surgery at our institution were computed using the Kaplan-Meier method [21]. Overall survival curves for the various subgroups were compared using the log rank test. The Cox proportional hazards model was used to obtain crude rate ratios, adjusted rate ratios and their 95% confidence intervals for the various EGFR categories [22]. Adjustments were done for age, sex, KPS score, radiographic enhancement, radiographic necrosis, extent of edema, midline shift, multifocal disease or gliomatosis cerebri and ependymal involvement.
Results
Demographic Characteristics
In our study of 54 GBM patients, 43% (n = 23) failed to express the EGFR, 57% (n = 31) were positive for the pan-EGFR stain and of those that expressed EGFR, 31% (n = 17) also expressed the EGFRvIII variant while 26% (n = 14) failed to express EGFRvIII. Staining for EGFR was typically diffuse, while the staining for EGFR vIII was generally more focal (not shown) as has been previously reported [23]. This distribution of expression was similar to those GBM patients who underwent GTR [19]. There was no significant difference in age, KPS score, or extent of surgical resection (biopsy, subtotal or partial resection) among the patients whose tumors failed to express EGFR, over expressed the wild-type EGFR, or expressed EGFRvIII (Table 1). Interestingly, men were more likely to over express EGFR only (79%) or EGFRvIII (88%). This was not a trend observed in GBM patients who had undergone GTRs [19].
Table 1 Demographic characteristics of patients with glioblastoma multiforme categorized according to epidermal growth factor receptor expression of the tumor
Parameter EGFRa negative EGFRwt positive only EGFRvIII positive
Total 54, N (%) 23 (43) 14 (26) 17 (31)
*Sex, N (%) M 15 (65) 11 (79) 15 (88)
F 8 (35) 3 (21) 2 (22)
*Age, years, Median (range) 57 (15–79) 65 (25–71) 59 (54–73)
*KPS score, Median (range) 90 (50–100) 80 (50–90) 90 (60–100)
aEGFR, epidermal growth factor receptor; wt, wild type; vIII, vIII mutant; KPS, Karnofsky Performance Scale.
While there was a trend towards increased post-surgical incidence of pulmonary embolus in patients who expressed the EGFRvIII (17.6%) compared to those who over expressed the EGFR only (7.1%) or failed to express EGFR (8.7%), it was not statistically significant (p = 0.36). No such trend was observed with DVT.
Radiographic Characteristics
There was no significant difference in the location, extent of necrosis, amount of MR image contrast-enhancement, extent of edema, or amount of brain midline shift among the three EGFR expression categories of GBMs (Table 2). There was no significant difference in the ratio of MRI T2-bright volume to T1-enhancing volume between EGFR expression groups.
Table 2 Radiographic characteristics and post surgical events in glioblastomas according to epidermal growth factor receptor expression category
Parameter EGFRa negative EGFRwt positive EGFRvIII positive
Total of 54, N (%) 23 (43) 14 (26) 17 (31)
Cortical matter involvement, N (%)
Yes 17 (74) 10 (71) 11 (65)
No 6 (26) 4 (29) 6 (35)
Ependymal, N (%) yes 3 (13) 3 (21) 6 (35)
Multifocality, N (%) Yes 2 (9) 2 (14) 2 (12)
Gliomatosis cerebri, N (%) Yes 5 (22) 4 (29) 3 (18)
*Severe necrosis (grade 3), N (%) Yes 5 (24) 6 (46) 7 (44)
aEGFR, epidermal growth factor receptor; wt, wild type; vIII, vIII mutant; *radiographic imaging not available for all patients.
Natural Disease History Based on EGFR and EGFRvIII Expression in GBMs
Patients that over expressed the EGFR or expressed the EGFRvIII presented with a greater incidence of ependymal spread, 21.4% and 35.3% respectively, when compared to tumors that failed to express either marker 13.0% but this was not statistically significant. Ependymal spread negatively impacted median survival in EGFR expressing (5.3 versus 11. months) and EGFRvIII expressing (8.3 versus 15 months) GBM patients. There were no significant differences in the percentages of patients who had multifocal disease or gliomatosis cerebri, irrespective of EGFR or EGFRvIII expression status. However, there was an 18–22% incidence of gliomatosis cerebri in all EGFR expression categories in the less than GTR patient group compared to only 3–5% incidence in those patient who underwent GTR.
Impact of EGFR or EGFRvIII on Survival
Over expressed wild-type EGFR or EGFRvIII were not independent predictors of overall survival and did not confer a worse prognosis (Figure 1). The median overall survival times for patients with tumors having no EGFR expression, over expressed EGFR only, or EGFRvIII were 12.3 (95% CI, 8.04–16.56), 11.03 (95% CI, 10.18–11.89) and 14.07 (95% CI, 7.39–20.74) months, respectively, indicating that neither EGFR or EGFRvIII were negative prognostic indicators in GBM patients unable to undergo GTR. The median overall survival times for patients who underwent GTR with tumors having no EGFR expression, over expressed EGFR only, or mutant EGFRvIII was 11.68, 11.9 and 13.0 months, respectively indicating that the median survival were not significantly different from those patients who underwent GTR. Our group has previously demonstrated that the extent of surgical resection impacts survival in a series of 416 patients with glioblastoma multiforme [4]. The statistically significant impact on survival started in those patients who received 89% volumetric resections or greater. The vast majority of patients in our current study were subtotal resections defined as extents of resection of 85% to <95%, as opposed to <85%. This could account for the comparable survival to the GTR cohort (≥ 95% resection). There was no significant difference in age or KPS between patients who underwent GTR and those who did not.
Figure 1 Graph showing Kaplan-Meier estimates of overall survival in glioblastoma multiforme patients who underwent sub-total resection followed by standard-of-care radiation therapy and chemotherapy. Patients with tumors not expressing the epidermal growth factor receptor (EGFR; n = 23; solid black line), expressing amplified EGFR (n = 14; dashed grey line), and expressing EGFRvIII (n = 17; dotted black line) had median overall survival times of 12.3 (95% CI, 8.04–16.56), 11.03 (95% CI, 10.18–11.89), and 14.07 (95% CI, 7.39–20.74) months, respectively, which were not statistically significantly different.
Established Prognostic Factors
Our results were consistent with prior findings demonstrating that age ≥ 65 years has a negative prognostic impact on survival although this difference did not reach statistical significance, p = 0.33). This was likely due to our small sample size and a paucity of young and very old patients (n = 4 under 40 years; n = 2 over 75 years). KPS score was found to be an independent prognostic indicator in our study, consistent with previous studies that have validated KPS as a prognostic indicator in GBM patients. Radiographically visualized necrosis was more common in EGFR-expressing tumors but the sample size was insufficient to ascertain if necrosis was a negative prognosticator.
Discussion
Despite previous reports indicating that EGFR and EGFRvIII were negative prognostic indicators, we found that neither the over expressed wild-type EGFR nor EGFRvIII were independent predictors of median overall survival in GBM patients who underwent GTR (≥ 95% volumetric resection). The median overall survival times for patients who had tumors devoid of EGFR expression, with over expression of EGFR, or with mutant (EGFRvIII) expression were 0.96, 0.98, and 1.07 years, respectively. In the study by Shinojima et al. [18], the authors concluded that EGFR amplification in GBMs was associated with shorter patient survival in a heterogeneous group of patients who underwent a wide variety of treatments including gross-total resection, partial resection, and biopsy. We therefore hypothesized that the EGFR and EGFRvIII may have a negative prognostic impact in patients that are not able to undergo GTR. This would be one potential variable to account for the discrepancies between the results. However, we did not see a difference in median survival across EGFR expression categories in the subcategory of patients unable to undergo GTR.
As others have found there was a range in the positive cases with respect to the proportion of tumor cells which were positive. This was especially true for EGFRvIII [23]. The number of samples in this study was relatively small given the restriction to subtotally resected cases, and therefore stratification by proportion of positively stained cells was not attempted. However, future studies examining this issue in EGFRvIII-positive tumors are warranted.
We investigated the possibility that the absence of an EGFR prognostic effect could be explained by the age distribution of our sample. Simmons et al. [17] had demonstrated that EGFR over expression was an unfavorable prognostic factor in patients less than 55 years of age. In the study by Shinojima et al. [18], 97% of the patients were <70 years old. If EGFR over expression is truly an unfavorable prognostic factor in younger age patients, such an age distribution may have been sufficient to influence the authors' conclusion that EGFR over expression impacted survival rates independently of age. In our study, however, the respective median overall survival times for our patients under 55 years of age whose tumors expressed neither type of EGFR, overexpressed wild type EGFR or expressed EGFRvIII, were not statistically different. In our earlier study of patients who underwent GTR we saw a trend toward a negative effect of EGFR and EGFRvIII expression on survival in patients under 40 years of age. However, no conclusion could be drawn for this subgroup in this study as there were only 4 patients aged <40. Though age bias is an unlikely explanation for the negative findings in our study, the differential effect of EGFR within different age groups requires further investigation.
EGFR and EGFRvIII expression have been shown to increase the infiltrative and invasive properties of glioma cells [24,25], therefore, one could hypothesize that patients expressing these markers may be more likely to present with multifocal disease, ependymal dissemination, or gliomatosis cerebri and perhaps this is the confounding variable between the studies accounting for the discrepancy in prognostic impact. GTR is not possible when there is multifocal disease, extensive ependymal spread or gliomatosis cerebri. Patients deemed "unresectable" due to extensive tumor invasion and multifocality of disease may have been more likely to have the over expression of the EGFR or the EGFRvIII. Additionally, these characteristics on radiographic presentation typically influence the treatment options toward biopsy and palliative treatment modalities. Although there was a trend of increased ependymal involvement within tumors expressing EGFR (21%) and EGFRvIII (35%) compared with tumors not expressing EGFR (13%) in patients not undergoing GTR, this was not statistically significant. The study sample was too small to determine if there was a statistically significant difference in the incidence of gliomatosis cerebri or multifocal disease. We did observe an increased incidence of gliomatosis cerebri in GBM patients that did not undergo GTR compared to those that were able to undergo GTR. These factors were not addressed in the other studies and could potentially account for the differences in EGFR and EGFRvIII prognostic impact between these studies, especially if there was a significant incidence of these in a small series. Alternatively, there may be unidentified confounding variable that our studies and the other published ones did not account for.
Conclusion
Within the subcategory of patients with GBM who underwent less than GTR, the presence of EGFR or EGFRvIII as a single mutation does not account for poor prognosis; however, simultaneous molecular genetic changes with prognostic significance have not been addressed in this study.
Declaration of competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
ABH conceived the project, designed and coordinated the study and drafted the manuscript. DS designed the database and carried out the data analysis. DY participated in the database design and assisted with the immunohistochemistry. WS performed the detailed volumetric and radiographic analysis. KA performed and interpreted the immunohistochemistry. All authors read and approved the final manuscript.
Acknowledgements
We thank Betty L. Notzon for their editorial assistance in preparation of this manuscript and Dr. Jeff Weinberg for constructive criticisms. The data contained within this manuscript was presented at the 13th World Congress of Neurological Surgeons, Marrakesh, Morocco, June 2005. This work was supported by The University of Texas M. D. Anderson Cancer Center Department of Neurosurgery recruitment start-up funds to ABH.
==== Refs
Kallio M Sankila R Jaaskelainen J Karjalainen S Hakulinen T A population-based study on the incidence and survival rates of 3857 glioma patients diagnosed from 1953 to 1984 Cancer 1991 68 1394 1400 1873791
Davis FG Freels S Grutsch J Barlas S Brem S Survival rates in patients with primary malignant brain tumors stratified by patient age and tumor histological type: an analysis based on Surveillance, Epidemiology, and End Results (SEER) data, 1973-1991 J Neurosurg 1998 88 1 10 9420066
Curran WJJ Scott CB Horton J Nelson JS Weinstein AS Fischbach AJ Chang CH Rotman M Asbell SO Krisch RE Recursive partitioning analysis of prognostic factors in three Radiation Therapy Oncology Group malignant glioma trials J Natl Cancer Inst 1993 85 704 710 8478956
Lacroix M Abi-Said D Fourney DR Gokaslan ZL Shi W DeMonte F Lang FF McCutcheon IE Hassenbusch SJ Holland E Hess K Michael C Miller D Sawaya R A multivariate analysis of 416 patients with glioblastoma multiforme: prognosis, extent of resection, and survival J Neurosurg 2001 95 190 198 11780887
Winger MJ Macdonald DR Cairncross JG Supratentorial anaplastic gliomas in adults. The prognostic importance of extent of resection and prior low-grade glioma J Neurosurg 1989 71 487 493 2552044
Ekstrand AJ James CD Cavenee WK Seliger B Pettersson RF Collins VP Genes for epidermal growth factor receptor, transforming growth factor alpha, and epidermal growth factor and their expression in human gliomas in vivo Cancer Res 1991 51 2164 2172 2009534
Wikstrand CJ McLendon RE Friedman AH Bigner DD Cell surface localization and density of the tumor-associated variant of the epidermal growth factor receptor, EGFRvIII Cancer Res 1997 57 4130 4140 9307304
Ekstrand AJ Longo N Hamid ML Olson JJ Liu L Collins VP James CD Functional characterization of an EGF receptor with a truncated extracellular domain expressed in glioblastomas with EGFR gene amplification Oncogene 1994 9 2313 2320 8036013
Aldape KD Ballman K Furth A Buchner JC Giannini C Burger PC Scheithauer BW Jenkins RB James CD Immunohistochemical detection of EGFRvIII in high malignancy grade astrocytomas and evaluation of prognostic significance J Neuropathol Exp Neurol 2004 63 700 707 15290895
Galanis E Buckner J Kimmel D Jenkins R Alderete B O'Fallon J Wang CH Scheithauer BW James CD Gene amplification as a prognostic factor in primary and secondary high-grade malignant gliomas Int J Oncol 1998 13 717 724 9735401
Newcomb EW Cohen H Lee SR Bhalla SK Bloom J Hayes RL Miller DC Survival of patients with glioblastoma multiforme is not influenced by altered expression of p16, p53, EGFR, MDM2 or Bcl-2 genes Brain Pathol 1998 8 655 667 9804374
Waha A Baumann A Wolf HK Fimmers R Neumann J Kindermann D Astrahantseff K Blumcke I von Deimling A Schlegel U Lack of prognostic relevance of alterations in the epidermal growth factor receptor-transforming growth factor-alpha pathway in human astrocytic gliomas J Neurosurg 1996 85 634 641 8814167
Bouvier-Labit C Chinot O Ochi C Gambarelli D Dufour H Figarella-Branger D Prognostic significance of Ki67, p53 and epidermal growth factor receptor immunostaining in human glioblastomas Neuropathol Appl Neurobiol 1998 24 381 388 9821169 10.1046/j.1365-2990.1998.00137.x
Zhu A Shaeffer J Leslie S Kolm P El-Mahdi AM Epidermal growth factor receptor: an independent predictor of survival in astrocytic tumors given definitive irradiation Int J Radiat Oncol Biol Phys 1996 34 809 815 8598357 10.1016/0360-3016(95)02184-1
Etienne MC Formento JL Lebrun-Frenay C Gioanni J Chatel M Paquis P Bernard C Courdi A Bensadoun RJ Pignol JP Francoual M Grellier P Frenay M Milano G Epidermal growth factor receptor and labeling index are independent prognostic factors in glial tumor outcome Clin Cancer Res 1998 4 2383 2390 9796969
Smith JS Tachibana I Passe SM Huntley BK Borell TJ Iturria N O'Fallon JR Schaefer PL Scheithauer BW James CD Buckner JC Jenkins RB PTEN mutation, EGFR amplification, and outcome in patients with anaplastic astrocytoma and glioblastoma multiforme J Natl Cancer Inst 2001 93 1246 1256 11504770 10.1093/jnci/93.16.1246
Simmons ML Lamborn KR Takahashi M Chen P Israel MA Berger MS Godfrey T Nigro J Prados M Chang S Barker FG Aldape K Analysis of complex relationships between age, p53, epidermal growth factor receptor, and survival in glioblastoma patients Cancer Research 2001 61 1122 1128 11221842
Shinojima N Tada K Shiraishi S Kamiryo T Kochi M Nakamura H Makino K Saya H Hirano H Kuratsu J Oka K Ishimaru Y Ushio Y Prognostic value of epidermal growth factor receptor in patients with glioblastoma multiforme Cancer Res 2003 63 6962 6970 14583498
Heimberger AB Hlatky R Suki D Yang D Weinberg J Gilbert M Sawaya R Aldape K Prognostic effect of epidermal growth factor receptor and EGFRvIII in glioblastoma multiforme patients Clin Cancer Res 2005 11 1462 1466 15746047 10.1158/1078-0432.CCR-04-1737
Reist CJ Archer GE Kurpad SN Wikstrand CJ Vaidyanathan G Willingham MC Moscatello DK Wong AJ Bigner DD Zalutsky MR Tumor-specific anti-epidermal growth factor receptor variant III monoclonal antibodies: use of the tyramine-cellobiose radioiodination method enhances cellular retention and uptake in tumor xenografts Cancer Res 1995 55 4375 4382 7671250
Kaplan EL Meier P Nonparametric estimation from incomplete observations J Am Stat Assoc 1958 53 457 481
Cox DR Regression models and life-tables J R Stat Soc 1972 34 187 220
Biernat W Huang H Yokoo H Kleihues P Ohgaki H Predominant expression of mutant EGFR (EGFRvIII) is rare in primary glioblastomas Brain Pathol 2004 14 131 136 15193025
Lund-Johansen M Bjerkvig R Humphrey PA Bigner SH Bigner DD Laerum OD Effect of epidermal growth factor on glioma cell growth, migration, and invasion in vitro Cancer Res 1990 50 6039 6044 2393868
Tysnes BB Haugland HK Bjerkvig R Epidermal growth factor and laminin receptors contribute to migratory and invasive properties of gliomas Invasion Metastasis 1997 17 270 280 9876221
|
16236164
|
PMC1298339
|
CC BY
|
2021-01-04 16:39:27
|
no
|
J Transl Med. 2005 Oct 19; 3:38
|
utf-8
|
J Transl Med
| 2,005 |
10.1186/1479-5876-3-38
|
oa_comm
|
==== Front
J Transl MedJournal of Translational Medicine1479-5876BioMed Central London 1479-5876-3-401626289410.1186/1479-5876-3-40ResearchViral load responses to HAART is an independent predictor of a new AIDS event in late stage HIV infected patients: prospective cohort study Kazanjian Powel [email protected] Wei [email protected] Morton [email protected] Tejal [email protected] Kamal [email protected] Department of Internal Medicine, University of Michigan Health System, Ann Arbor, Michigan, USA2 Department of Epidemiology, University of Michigan School of Public Health, Ann Arbor, Michigan, USA2005 31 10 2005 3 40 40 12 7 2005 31 10 2005 Copyright © 2005 Kazanjian et al; licensee BioMed Central Ltd.2005Kazanjian 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 sizeable number of HIV-infected patients receiving HAART do not maintain prolonged virologic suppression. We evaluated long-term HIV viral load (VL) responses to HAART as a risk factor for AIDS events (AE) that is independent of CD4 responses.
Methods
A cohort of patients with pre-therapy CD4 < 200/mm3 who had CD4 and VL measurements for > one year after receiving HAART at a university clinic were prospectively enrolled. Cox proportional multivariate regression model was used to determine whether CD4 and VL responses were independently associated with new AE.
Results
The patient (N = 214) mean baseline CD4 = 92/mm3, VL = 219,000 c/mL and follow-up duration 42.3 months (range 13–72 months). A new AE occurred in 56 patients; CD4 cell count response to HAART that remained < 200/mm3 throughout the study period was a significant risk factor for new AE (RR = 9.7–12.5; p < 0.001). Similarly, VL responses that remained > 5,000 c/mL during this period was also a significant risk factor (RR = 6.7–12.8; p = 0.001) that was independent of CD4 response adjusted for <> 200/mm3.
Conclusion
Maintaining adequate long-term virologic responses to HAART provides a clinical benefit independent of CD4 responses.
Viral LoadCD4HAARTAIDS
==== Body
Background
Despite an overall decline in AIDS-associated illnesses since the introduction of HAART [1,2], some patients receiving combination antiretrovirals remain at risk for developing new AIDS events (AE) [3,4]. Those who have a pre-treatment CD4 cell count < 200/mm3 [5-9] or have insufficient CD4 cell responses to HAART [6] are at risk for progressing to AIDS or dying while receiving therapy. Persistent viremia [10], independent of insufficient CD4 cell count responses [13-16], has also been shown to be a predictor of disease progression. These studies, however, have differed in regards to the VL that they identify as being predictive of disease progression-- > 1,000 c/mL [10], > 7,000 c/mL [13], or > 20,000 c/mL [11]. Furthermore, although these studies have linked short virologic markers in response to HAART with a successful clinical outcome [11-14], the importance of sustaining immunologic and virologic responses to HAART over a more prolonged time period remains to be addressed.
It is important to evaluate whether maintaining adequate responses to long-term HAART treatment has clinical significance for several reasons. First, evaluating whether long-term virologic response is an independent predictor of new AE has relevance for the sizeable number of patients receiving HAART who do not maintain prolonged virologic suppression [1-4,15]. Also, the meaning of persistent viremia is especially pertinent to those who have high-level resistance to antivirals, but nonetheless sustain an immunologic benefit while continuing their failing HIV regimen [16]. Furthermore, evaluating whether immunologic responses to HIV treatment is an independent risk factor for disease progression is applicable to those patients who are unable to mount significant CD4 cell elevations, regardless of whether they have undetectable VL [17]. Thus, our aim was to determine whether long term virologic, as well as immunologic responses to HAART are important and independent predictors of developing a new AE as well as to identify a level of persistent viremia that is predictive of disease progression. We chose patients with significant immune depletion because they are at highest risk for disease progression [3,5].
Methods
We identified HIV-infected patients treated at the University of Michigan HIV/AIDS Treatment Clinic (UMHATC) who had a baseline CD4 cell count < 200 cells/mm3 before initiating a HAART regimen. The study began on January 31, 1996; patients were enrolled into this prospective cohort from the time therapy was initiated if they received at least twelve months of HIV therapy and were followed to January 1, 2005. The definition of HAART for this study was guided by DHHS guidelines [18]. Patients were included if they were antiviral naïve before starting HAART or if they had previously received one or two nucleoside agents. Patient information was extracted from an electronic database (Solutions™ Patient Management System). The research was carried out according to the principles of the Declaration of Helsinki and the study was approved by the Institutional Review Board of the University of Michigan Health System.
Data on markers of response to HAART (i.e., CD4 and VL levels that were obtained during the course of routine UMHATC practice on an every three month basis), as well as occurrence of a new AE (1993 CDC AIDS Surveillance definition) [19], were extracted from the electronic database system. CD4 cell counts were determined using flow cytometry, and HIV RNA levels were performed using a reverse transcriptase polymerase chain reaction assay (Amplicor; Roche Diagnostics, Branchburg, New Jersey, USA). For purposes of consistency, a VL value of < 400 copies/mL was considered undetectable for the study period, as lower limit changed to < 50 c/mL in 1999 in the middle of our study period. We also recorded assessment of adherence that was entered into our electronic database by nurses, social workers, and physicians, as was our standardized practice during this study period.
To compare baseline characteristics between patients who developed an AE and those who did not, we used histograms to examine distributions of variables that were collected. Appropriate transformations were applied on variables found asymmetric in the following analysis. Baseline quantitative characteristics of patients (e.g. age, CD4, VL) were described using mean and standard deviation, and qualitative characteristics (e.g. gender, presence of AE at baseline) using proportion. A t test was used to compare quantitative characteristics between patients who developed an AE and those who did not. A Chi-square test was used to compare qualitative characteristics between groups.
To determine probabilities of occurrence of a new AE from HAART initiation, we measured time to first AE and used Kaplan-Meier curves. To determine averaged VL trends after initiating HAART we used lowess smooth curves fitted to all patients, as well as to those who developed an AE and those who did not. We then determined VL values for the entire study period, as well as for four specified periods: 1–12 months, 13–24 months, 25–36 months and 37–48 months after initiating HAART.
To determine risk factors for developing a new AE, we used a Cox proportional hazard multivariate regression model. The proportional hazard assumption was then examined by plotting the complementary log-log transformation of the survival functions separately within subgroups. Pre-treatment factors that were examined as predictors for a new AE included demographic factors (e.g. age and gender), prior NRTI use, and presence of AE at baseline. For a given VL measurement at a particular time point after initiation of HAART, persons were considered at risk for newly diagnosed AE for 1–6 months after the time the measurement was made. For each time period, we classified VL responses to HAART (adjusted for a CD4 response of <> 200 cells/mm3) into separate categories to determine whether a particular VL category was independently associated with presence of a new AE. The relative risks of new AE for patients in specified VL categories were based on the fitted proportional hazard model and 95% confidence intervals were estimated. All statistical tests were two-sided, and a p-value less than 0.05 was considered statistically significant. Statistical software R 1.7.0 was used for all statistical analysis.
Results
Patient Population and AE Description
Two hundred and fourteen patients (21% of 1,014 patients treated at UMHATC during the study period) were included in the analysis. Table 1 gives the demographic and clinical features of the population, including prior nucleoside use and presence of AE at baseline prior to initiating HAART. The mean duration of follow up for the entire population after starting HAART was 42.3 ± 37.4 (range 13–82 months), and a total of 9,288 person-months was included in the analysis. The incidence rate of a first new AE was 6.1/100 person years for the entire population. There was no significant difference in presence of baseline AE (P = 0.14), or prior nucleoside agent use (P = 0.54) between those who developed an AE and those who did not (P = 0.28). Similarly, there was no significant difference in adherence between the two groups; 48 of the 56 patients who developed a new AE were noted to be adherent with their prescribed HAART regimen on a regular basis (86%) as opposed to 143 of the 158 patients who did not (91%). Table 1 shows that patients without an AE had a more brisk mean CD4 increase, more substantial VL decline, and a greater percentage of undetectable VL than those who did develop an AE.
Table 1 Demographic, Clinical Features, Mean CD4* and Viral Load** According to Whether an AE Occurred.
Laboratory Value Patients with AE (n = 56) Patients without AE (n = 158)
Demographic
Age ± SD (range) 37.9 ± 8.28 (16 – 65.4) 39.4 ± 8.79 (15.2 – 67.8)
HIV risk (no.) MSM 27 MSM 82
HTS 7 HTS 34
IVDU 4 IVDU 12
Other 22 Other 36
Clinical
OI (no.) at Baseline (%) 7 (12%) 25 (15%)
Prior NRTI (no.) 12 (21%) 27 (17%)
Baseline Laboratory Values
CD4 62 107
VL 257,136 196,832
Follow Up Laboratory Values
CD4 (Change from Baseline) 109 (47) 369 (262)
VL (Change from Baseline) 159,517 (-97,619) 29,743 (167, 089)
Number (%) with Undetectable VL at any point during follow up 46 (82%) 142 (89%)
Number (%) with Undetectable VL 12 (23%) 95 (62%)
*CD4 in cells/mm3
**VL in c/mL
Abbreviations: HTS (Heterosexual), NRTI (Nucleoside Reverse Transcriptase Inhibitor).
Fifty-six of the 214 patients (26%) developed a new AE; the mean time from HAART initiation to AE was 26.5 ± 21.2 (range 1 – 84 months). Four patients, all of whom had experienced a new AE, died during the study period. Figure 1 shows that the probability of a patient remaining free of a new AE after 12 and 48 months of HAART was 96.1% (95% CI, 93.5,98.8), and 78% (95% CI, 71.9,86), respectively. There were 60 episodes of new AE that occurred in 56 patients (two patients had 2 new AE and one had 3), as well as AE that were present pre-treatment. Disseminated Mycobacterium avium complex (MAC) infection (13 patients), Pneumocystis jiraveci pneumonia (PcP, 8), Kaposi's sarcoma (6), candida esophagitis and non Hodgkin's lymphoma (5 each) were the most common AEs. Three of these AEs represented a recurrence of an AE that was present at baseline prior to initiating HAART: PcP (1 patient), MAC (1) and CE (1). 77% of patients in whom MAC was diagnosed were receiving prophylaxis with a macrolide agent, and 88% of those with PcP were receiving pneumocystis prophylaxis. Others included mycobacterial and cryptococcal infections (4 patients each), progressive multifocal leukoencephalopathy and cryptosporidiosis (3 each), HIV encephalopathy and recurrent bacterial infections (2 each), and chronic Herpes simplex infections and histoplasmosis (1 each).
Figure 1 Kaplan-Meier curve of First AE for the entire population. The percentage of patients without an AE is plotted according to time (months) after HAART initiation. The probability of a patient remaining free of a new AE after 48 months of HAART was 78% (95% CI, 71.9, 86).
On Treatment CD4 and VL Responses and Association with a New AIDS Event
Figure 2 plots the averaged CD4 responses to HAART throughout the entire study period for all patients, and according to whether a new AE occurred. Figure 2 shows that the CD4 response for patients in whom a new AE developed was much lower throughout the entire time period than for those in whom an AE did not occur. Throughout the study period, the averaged HAART-restored CD4 cell count failed to elevate > 200 cells/mm3 in a significantly higher proportion of patients with a new AE, 85% (48/56 patients), than those without an AE, 8% (13/158 patients) (p < 0.001). Table 2 shows that a CD4 cell count response to HAART that remained < 200/mm3 was a significant risk factor for developing new AE for each 12 month interval after initiating HAART (RR = 9.7–12.5; p < 0.001).
Figure 2 CD4 Response to HIV Therapy for Entire Population and Patients with or without AE. The averaged CD4 responses to HAART throughout the entire study period were determined by lowess smooth curves fitted to all patients for all patients, as well as for those in whom a new AE occurred or did not occur. At each time point during the study, the CD4 response for patients in whom a new AE developed was much lower than for those in whom an AE did not occur.
Table 2 Risk of Developing New AIDS Event according to CD4 and VL Response Categories.
CD4 Response Time After HAART Relative Risk and CI P Value
Absolute CD4 < 200 (1–12 months) RR = 4.45 (95% CI: 1.8, 10.9); p < 0.001
(13–24 months) RR = 3.95 (95% CI: 1.62, 9.63); p = 0.002
(25–36 months) RR = 5.19 (95% CI: 1.99, 13.5); p = 0.001
(37–48 months) RR = 3.21 (95% CI: 1.61, 10.8); p = 0.02
VL Response
Absolute VL > 1,000 (1–12 mos) RR = 2.34 (95% CI: 0.87, 6.26); p = 0.09
(13–24 mos) RR = 2.77 (95% CI: 0.81, 9.48); p = 0.10
(25–36 mos) RR = 4.41 (95% CI: 0.99, 19.7); p = 0.052
(37–48 mos) RR = 2.74 (95% CI: 0.71, 9.1); p = 0.14
Absolute VL > 5,000 (1–12 mos) RR = 4.66 (95% CI: 2.02, 10.7); p = 0.001
(13–24 mos) RR = 3.24 (95% CI: 1.06, 9.88); p = 0.03
(25–36 mos) RR = 6.41 (95% CI: 1.8, 22.8); p = 0.004
(37–48 mos) RR = 4.23 (95% CI: 1.05, 6.0); p = 0.05
Absolute VL > 50,000 (1–12 mos) RR = 3.77 (95% CI: 1.68, 8.43); p = 0.001
(13–24 mos) RR = 2.78 (95% CI: 1.04, 7.39); p = 0.021
(25–36 mos) RR = 2.89 (95% CI: 1.80, 5.51); p = 0.04
Figure 3 plots the averaged VL responses to HAART throughout the entire study period for all patients, and according to whether a new AE developed. Figure 3 shows that the virologic reduction for patients in whom a new AE developed was much lower throughout the entire time period than for those in whom an AE did not occur. Throughout the study period, the averaged HAART-restored VL remained > 5,000 c/mL in a significantly higher proportion of patients with a new AE; 55% (31/56 patients), than those without an AE; 24% (37/158 patients) (p = 0.02). In a multivariate analysis adjusting for CD4 counts of 200 cells/mm3, a VL response > 5,000 c/mL was a significant risk factor for developing a first new AE for each 12 month interval after initiating HAART (RR = 6.7–12.8; p = 0.001) (Table 2). Similarly, a VL response that remained > 50,000 c/mL was also a significant risk factor for developing new AE for each 12 month interval after initiating HAART (RR = 3.7–12.6; p = 0.007) (Table 2). However, VL responses that remained > 1,000 c/mL approached but did not reach statistical significance for developing a first new AE (RR 2.34–4.41; p = 0.05–0.14) for each individual 12 month interval after initiating therapy (Table 2).
Figure 3 Viral Load Response to HIV Therapy for Entire Population and Patients with or without AE. The averaged VL responses to HAART throughout the entire study period were determined by lowess smooth curves fitted to all patients for all patients, as well as for those in whom a new AE occurred or did not occur. At each time point during the study, the VL reduction for patients in whom a new AE developed was much lower throughout the entire time period than for those in whom an AE did not occur.
Discussion
The present study demonstrates that failure to maintain low levels of viremia is a significant risk factor for HIV disease progression that is independent of CD4 responses. Our results are consistent with studies showing the clinical importance of achieving a VL response at 6–12 months that is independent of CD4 responses [10-14], and stress the benefit of maintaining a durable VL response throughout a 48 month period. The VL level identified as carrying an increased risk for an AIDS event in our study, > 5,000 c/mL, is within the range that some reports identify as independently predictive of disease progression, i.e., > 1,000 c/mL [10], > 7,000 c/mL [13], but lower than another report (20,000 c/mL) [11]. Regardless of the precise value, a reduction in VL that cannot be maintained at low or undetectable levels, a predictor of new AIDS events in our study, is not uncommon for patients receiving HAART [20]. It is unlikely that the difference in VL response between the groups in our study can be attributed to different adherence rates between the two groups. However, because adherence measurement tools were not used as part of routine clinical care, we could not calculate the percentage of each patient's adherence over a long-term basis and then compare this variable between groups.
In fact, several published studies suggest that virologic failure rates are especially high, varying from 20% to 70%, in those who are either treatment-experienced or have low baseline pre-treatment CD4 cell counts and high VL values [21-23]. In one study, for example, lower CD4 counts and higher VL at baseline predicted virologic failure, but there was no clear cut-off value at which the risk started to increase [24]. In another study, baseline CD4 count < 25 cells/mm3 was associated with a significantly higher risk of virologic failure, as was a baseline VL ≥ 100,000 copies/mL [25].
Identifying that long-term virologic response is an independent predictor of new AE has relevance for the sizeable number of patients receiving HAART who cannot maintain prolonged virologic suppression [20]. Also, the meaning of persistent viremia is pertinent to those who have high-level resistance to antivirals, but nonetheless sustain an immunologic benefit while continuing their failing HIV regimen [26]. At present, the outcome and optimal management of these patients remains undefined [27]. Switching to an optimized regimen may not be an available option, since the durability of a salvage regimen selected for patients with high level resistance to many antivirals may be brief. Consequently, some physicians continue a failing regimen despite persistent viremia for patients who maintain an immune benefit until other active treatment options become available. By identifying VL as a risk factor for disease progression that is independent of CD4 response, our study suggests that discordant responders may be at risk for disease progression while being maintained on a virologically failing regimen.
It is not rare for patients, such as those followed in the present study, to first receive HAART after they develop an AIDS event or their CD4 cell count falls below 200/mm3. In one report, "late testers" – those whose diagnosis of HIV occurred within 12 months of their AIDS diagnosis, accounted for 39% of persons diagnosed with AIDS in San Francisco between 2001 and 2002 [28]. Also, in a New York study, 27% of persons with newly diagnosed HIV infection were concurrently diagnosed with AIDS [29]. Furthermore, in clinical practice, some patients, despite having been tested earlier, may not have received HAART until a later stage because they may not have wished to take HIV therapy until becoming symptomatic. These reasons, along with sole availability of mono or dual nucleoside agents prior to 1996, accounted for why 21% of patients followed in our practice did not receive HAART until their CD4 count fell below 200/mm3 and therefore met criteria for our study.
Our finding that PCP and MAC remain the most frequent AE in the HAART era is consistent with prior studies [30]. Similarly, our finding that chemoprophylaxis failure among those with the most advanced immunosuppression was the most significant source of new PCP cases is also consistent with a HOPS cohort study [31], along with studies showing that KS and NHL remain the major AIDS-associated malignancies in the HAART era [32-34]. In contrast, our study showed that CMV and cerebral toxoplasmosis occur less often than in the Pre-HAART era. One possible explanation for this observation is that memory CD4 cells, expanded by HAART, may recognize antigens on previously acquired pathogens that persist in a latent state, thereby preventing clinical reactivation. This possibility, if confirmed, suggests that there may be benefits of receiving HAART that are independent of immunologic and virologic responses currently used to monitor therapy.
Conclusion
In summary, our study emphasizes the clinical benefit of maintaining virologic responses as well as immunologic throughout long-term HAART treatment. Incomplete responses to HAART are not uncommon in clinical practice, and identifying them as risk factors for disease progression emphasizes several treatment options that are available. For example, changing regimens in patients who have persistent virologic replication who have other antiviral options, as recommended by treatment guidelines [35], is supported by identifying VL as an independent predictor of disease progression. Nevertheless, for those with no option other than to remain on their current regimen for prolonged periods of time because of high level drug resistant HIV, clinical progression may eventually occur, even in those who have a CD4 benefit despite persistent viremia. Our findings stress the importance of developing new potent antiretroviral agents in order to sustain the overall decline in AIDS-associated illnesses that has been witnessed since the introduction of HAART.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
PK was responsible for study design, data collection, data analysis, and preparation of manuscript.
KA participated in data collection and data analysis.
MB participated in statistical design and analysis
WW participated in statistical design, statistical analysis, and preparation of the manuscript.
Acknowledgements
Financial Support: There was no financial support.
==== Refs
Palella F Delaney K Moorman A Declining morbidity and mortality among patients with advanced HIV infection N Engl J Med 1998 338 853 60 9516219 10.1056/NEJM199803263381301
Ledergerber B Egger B Erard V AIDS-related opportunistic illnesses occurring after initiation of potent antiretroviral therapy JAMA 1999 282 2220 6 10605973 10.1001/jama.282.23.2220
Ferrer E Santamarina E Santin M Progression to AIDS/Death is higher in patients initiating ARV with CD4 counts < 350 [abstract] 10th Conference on Retroviruses and Opportunistic Infections; Boston, MA 2003 910
Van Sighem AI van de Wiel MA Ghani A on behalf of the ATHENA cohort study group Mortality and progression to AIDS after starting highly active antiretroviral therapy AIDS 2003 17 2227 36 14523280 10.1097/00002030-200310170-00011
Egger M May M Chene G ART Cohort Collaboration Prognosis of HIV-1 Infected patients starting Highly Active Antiretroviral Therapy: A Collaborative Analysis of Prospective Studies Lancet 2002 360 119 29 12126821 10.1016/S0140-6736(02)09411-4
Sterling TR Chaisson RE Keruly J Improved Outcomes with Earlier Initiation of Highly Active Antiretroviral Therapy among Human Immunodeficiency Virus-Infected Patients who Achieve Durable Virologic Suppression: Longer Follow-Up of an Observational Cohort Study J Infect Dis 2003 188 1659 65 14639536 10.1086/379741
Tierney C Lathey JL Christopherson C Prognostic value of baseline HIB type 1 DNA measurement for disease progression in patients receiving nucleoside therapy J Infect Dis 2003 187 144 8 12508159 10.1086/345870
Wood E Hogg R Yip B Higher baseline levels of plasma HIV type 1 RNA are associated with increased mortality after initiation of triple drug antiretroviral therapy J Infect Dis 2003 188 1421 5 14624366 10.1086/379201
de Wolfe R Spijkerman I Schellekens P AIDS prognosis based on HIV-1 RNA, CD4+ T cell count and function: markers with reciprocal predictive value over time after seroconversion AIDS 1997 11 1799 806 9412697
Tarwater PM Gallant JE Mellors JW Prognostic value of plasma HIV RNA among highly active antiretroviral users AIDS 2004 18 2419 23 15622318
Raffanti SP Fusco JS Sherrill BH Effect of Persistent Moderate Viremia on Disease Progression During HIV Therapy J Acquir Immune Defic Syndr 2004 37 1147 54 15319674
ART Cohort Collaboration Prognostic importance of initial response in HIV-1 infected patients starting patent antiretroviral therapy: analysis of prospective studies Lancet 2003 362 679 86 12957089 10.1016/S0140-6736(03)14229-8
Kaplan JE Hanson DL Jones JL Viral load as an independent risk factor for opportunistic infections in HIV-infected adults and adolescents AIDS 2001 15 1831 36 11579245 10.1097/00002030-200109280-00012
Olsen CH Gatel J Ledergerber B Risk of AIDS and death at given HIV-RNA and CD4 cell counts, in relation to specific antiretroviral drugs in the regimen AIDS 2005 19 319 30 15718843
Ledergerber B Egger M Opravil M Clinical progression and virological failure on highly active antiretroviral therapy in HIV-1 patients: a prospective cohort study Lancet 1999 353 863 68 10093977 10.1016/S0140-6736(99)01122-8
Belec L Piketty C Si-Mohammed A High levels of drug resistant HIV in patients exhibiting increasing CD4 T cell counts despite virologic failure of protease inhibitor-containing ARV J Inf Dis 2000 181 1808 12 10823790 10.1086/315429
Kaufmann G Furrer H Perrin L Low Baseline CD4 T-cell Count and Higher Age Predict Poor CD4 T-cell Recovery in Treated HIV-1 Infected Individuals Suppressing HIV-1 RNA to Levels <1000 copies/mL for 5 Years [abstract] 11th Conference on Retroviruses and Opportunistic Infections; SF, CA 2004 557
Department of Health and Human Services [DHHS] Guidelines
Centers for Disease Control and Prevention 1993 Revised classification system for HIV infection and expanded surveillance case definition for AIDS among adolescents and adults MMWR Morb Mortal Wkly Rep 1992 41 1 19
Deeks SG Hecht FM Swanson M HIV RNA and CD4 cell count response to protease inhibitor therapy in an urban AIDS clinic: response to both initial and salvage therapy AIDS 1999 13 F35 F43 10397555 10.1097/00002030-199904160-00001
King MS Bernstein BM Walmsley SL Baseline HIV-1 RNA level and CD4 count predict time to loss of virologic response to nelfinavir but not lopinavir/ritonavir in antiretroviral therapy-naïve patients J Inf Dis 2004 190 280 4 15216462 10.1086/422037
Ribaudo H Downey G Fischl M Factors Associated with Virologic Failure and Their Impact on Treatment Outcomes: An Analysis of Virologic Failure in ACTG 388 [abstract] 11th Conference on Retroviruses and Opportunistic Infections; SF, CA 2004 553
Paredes R Mocroft A Kirk O Predictors of virologic success and ensuing failure in HIV-positive patients starting highly active antiretroviral therapy in Europe Arch Intern Med 2000 160 1123 32 10789605 10.1001/archinte.160.8.1123
Mocroft A Ledergerber B Viard JP Time to Triple Drug Class Failure after Initiation of HAART [abstract] 11th Conference on Retroviruses and Opportunistic Infections; SF, CA 2004 554
van Leth F Andrews S Grinsztjen B Virologic Failure in Antiretroviral Therapy Naive Patients Is Only Determined by Extreme Low Values of CD4+ Cells or High Values of HIV-1 RNA Concentration, Not by Choice of Treatment with Nevirapine or Efavirenz [abstract] 11th Conference on Retroviruses and Opportunistic Infections; SF, CA 2004 550
PLATO Collaboration Predictors of trend in CD4-positive T cell count and mortality among HIV-1 infected individuals with biological failure to all three antiretroviral drug classes Lancet 2004 364 51 62 15234856 10.1016/S0140-6736(04)16589-6
Kuritzkes DR Effect of Antiretroviral Resistance on Response to Treatment-Experienced Patients The AIDS Reader 2003 13 S5 11
Schwarcz S Hsu L AIDS in the HAART Era: the Extent and Characteristics of Late HIV Testers [abstract] 11th Conference on Retroviruses and Opportunistic Infections; SF, CA 2004 864
Nash D Manning SE Ramaswamy C Descriptive Epidemiology of HIV/AIDS in New York City: Incorporation of Newly Available Population-based Surveillance Data on HIV (non-AIDS), 2001 [abstract] 11th Conference on Retroviruses and Opportunistic Infections; SF, CA 2004 87
Centers for Disease Control and Prevention HIV/AIDS Surveillance Report 1998 9 28
Moorman AC Von Bargen JC Palella FJ Pneumocystis carinii pneumonia incidence and chemoprophylaxis failure in ambulatory HIV-infected patients. HIV Outpatient Study (HOPS) Investigators J Acquir Immune Defic Syndr Hum Retrovirol 1998 19 182 8 9768629
Mocroft A Lederberger B Katlama C EuroSIDA study group Decline in AIDS and Death rates in the EuroSIDA study: an observational study Lancet 2003 362 22 39 12853195 10.1016/S0140-6736(03)13802-0
Ives N Gazzard BG Easterbrook PJ Changing pattern of AIDS defining illnesses with introduction of Highly active antiretroviral therapy in a London clinic J Infect 2001 42 134 9 11531320 10.1053/jinf.2001.0810
Sabin C D'Arminio Monforte A May M The Changing incidence of Clinical AIDS events in 12,574 treatment naive patients starting HAART [abstract] 10th Conference on Retroviruses and Opportunistic Infections; Boston, MA 2003 569
Yeni PG Hammer SM Hirsch MS Treatment for adult HIV infection: 2004 recommendations of the International AIDS Society-USA Panel JAMA 2004 292 251 65 15249575 10.1001/jama.292.2.251
|
16262894
|
PMC1298340
|
CC BY
|
2021-01-04 16:39:28
|
no
|
J Transl Med. 2005 Oct 31; 3:40
|
utf-8
|
J Transl Med
| 2,005 |
10.1186/1479-5876-3-40
|
oa_comm
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.